PATTERN RECOGNITION IS A CLINICAL REASONING

PROCESS IN MUSCULOSKELETAL PHYSIOTHERAPY

Peter Andrew Miller

B Phty, Grad Cert Health Science (Education)

Thesis submission for Master of Medical Science

(Physiotherapy)
The University of Newcastle, Australia
Submitted March 2009
STATEMENT OF ORIGINALITY

The thesis contains no material which has been accepted for the award of
any other degree or diploma in any university or other tertiary institution and,
to the best of my knowledge and belief, contains no material previously
published or written by another person, except where due reference has
been made in the text. I give consent to this copy of my thesis, when
deposited in the University Library, being made available for loan and
photocopying subject to the provisions of the Copyright Act 1968.

ACKNOWLEDGEMENT OF AUTHORSHIP

I hereby certify that the work embodied in this thesis is the result of original
research, the greater part of which was completed subsequent to admission
to candidature for the degree.

Date __________________________

Signed __________________________
Peter Andrew Miller

ii
ACKNOWLEDGEMENTS

Considerable gratitude is extended to the participants of this study who gave

up their valuable time to put their problem solving skills on display for
analysis. The Capacity in Research & Evaluation (CAPRE) is acknowledged
for the Primary Health Care Small Research Grant that funded the study.
Many thanks are offered to the supervisors of the study, Professor Darren
Rivett and Ms Rosemary Isles, for their support and guidance. Finally
appreciation must be offered to Jessica, Josh and Max, for allowing me the
time to complete this career goal.

SUPERVISOR:
Prof Darren Rivett
BAppSc (Phty), MAppSc (ManipPhty), PhD

CO-SUPERVISOR:
Rosemary Isles
BPhty (Hons), M Ed, GradCertEduc (Tertiary)

iii
TABLE OF CONTENTS

DECLARATION ……………………………….…… ii
ACKNOWLEDGEMENTS ……………………………….…… iii
TABLE OF CONTENTS ……………………………….…… iv
LIST OF TABLES ……………………………….…… x
LIST OF FIGURES ……………………………….…… xii
LIST OF APPENDICES ……………………………….…… xiii
ABSTRACT ……………………………….…… xiv
ABBREVIATIONS ……………………………….…… xvi

CHAPTER 1. INTRODUCTION …………………………………… 1

1.1 RATIONALE OF THE STUDY …………………………………… 1
1.2 STUDY AIMS …………………………………… 2
1.3 THESIS OUTLINE …………………………………… 3

CHAPTER 2. LITERATURE REVIEW …………………………………… 4

2.1 BACKGROUND …………………………………… 4
2.1.1 Literature search strategy …………………………………… 5
2.1.2 Problem solving in physiotherapy clinical practice …………… 5
2.1.3 Integration of paradigms …………………………………… 6
2.2 CLINICAL REASONING RESEARCH METHODS …………… 10
2.2.1 Single case study design …………………………………… 10
2.2.2 Simulated case design …………………………………… 11
2.2.3 Simulated client assessments …………………………………… 12
2.2.4 Verbal reported data …………………………………… 13
2.2.5 Retrospective interview …………………………………… 14
2.2.6 Observational data …………………………………… 16
2.2.7 Mixed methods …………………………………… 16
Embedded design …………………………………… 17
Triangulation …………………………………… 18
2.3 HYPOTHESIS GENERATION …………………………………… 19
2.3.1 Hypothesis categories …………………………………… 20

iv
2.3.2 Diagnostic hypotheses …………………………………… 23
2.3.3 Basis for empirico-analytical reasoning models …………… 24
2.4 HYPOTHETICO-DEDUCTIVE REASONING …………………… 25
2.4.1 Defining hypothetico-deductive reasoning …………………… 25
2.4.2 Hypothetico-deductive reasoning in physiotherapy …………… 26
2.4.3 Characteristics of hypothetico-deductive reasoning …………… 27
2.4.4 Summary of hypothetico-deductive reasoning …………………… 29
2.5 PATTERN RECOGNITION …………………………………… 29
2.5.1 Defining pattern recognition …………………………………… 30
Single hypothesis …………………………………… 32
Significant case features …………………………………… 32
Information availability …………………………………… 33
2.5.2 Pattern recognition in physiotherapy …………………………… 34
2.5.3 Characteristics of pattern recognition …………………………… 35
Knowledge …………………………………… 35
Categorisation …………………………………… 36
Efficiency …………………………………… 37
Accuracy …………………………………… 38
2.5.4 Summary of pattern recognition …………………………………… 40
2.6 INTEGRATION OF CLINICAL REASONING MODELS …… 41
2.6.1 Further exploration of forwards reasoning …………………… 42
2.6.2 Separating forwards from backwards reasoning…………………… 42
2.6.3 Context specificity …………………………………… 45
2.7 KNOWLEDGE …………………………………… 46
2.7.1 Types of knowledge …………………………………… 46
2.7.2 Structured knowledge …………………………………… 47
2.7.3 The role of knowledge …………………………………… 49
2.8 EXPERTISE …………………………………… 49
2.8.1 Common attributes of an expert physiotherapist …………… 50
2.8.2 Clinical reasoning skill …………………………………… 51
2.8.3 Significant case features …………………………………… 52
2.9 REASONING IMPACT ON CURRENT HEALTHCARE ………… 53
2.9.1 Errors in clinical reasoning …………………………………… 54

v
2.9.2 Possible impact of accurate clinical patterns …………………… 56
2.9.3 Reasoning skill as a cost effective variable …………………… 57
2.10 SUMMARY …………………………………… 57

CHAPTER 3. METHODS …………………………………… 59

3.1 INTRODUCTION …………………………………… 59
3.2 PREPARATION …………………………………… 61
3.2.1 Ethical approval …………………………………… 61
3.2.2 Case development …………………………………… 61
3.2.3 Training of case actor …………………………………… 64
3.2.4 Pilot trial of method …………………………………… 66
3.3 SAMPLING …………………………………… 66
3.3.1 Sample size …………………………………… 66
3.3.2 Expert inclusion criteria …………………………………… 67
3.3.3 Novice inclusion criteria …………………………………… 69
3.3.4 Recruitment process …………………………………… 70
3.4 DATA COLLECTION …………………………………… 71
3.4.1 Location …………………………………… 71
3.4.2 Equipment …………………………………… 72
3.4.3 Participation process …………………………………… 74
3.4.4 Semi-structured interview technique …………………………… 74
3.4.5 Transcription …………………………………… 75
3.5 QUALITATIVE TRANSCRIPT ANALYSIS …………………… 76
3.5.1 Code development …………………………………… 76
3.5.2 Qualitative data analysis software …………………………………… 77
3.5.3 Data analysis process …………………………………… 78
Overview Reading …………………………………… 79
Coding …………………………………… 79
Coding accuracy and reliability …………………………… 81
3.5.4 Theme analysis …………………………………… 81
3.5.5 Interpretation …………………………………… 82
3.6 QUANTITATIVE TRANSCRIPT ANALYSIS …………………… 82
3.6.1 Identification of pattern recognition …………………………… 82

vi
3.6.2 Expert and novice group differences in pattern recognition use…. 85
3.6.3 Accuracy of pattern recognition …………………………………… 85
3.6.4 Hypothesis category utilisation …………………………………… 85
3.7 QUANTITATIVE VIDEO DATA ANALYSIS …………………… 86
3.8 TRIANGULATION …………………………………… 87
3.9 SUMMARY OF METHODOLOGY …………………………… 89

CHAPTER 4. RESULTS …………………………………… 90

4.1 INTRODUCTION TO FINDINGS …………………………………… 90
4.1.1 Examples of coded text provided as results …………………… 91
4.2 PARTICIPANT RECUITMENT …………………………………… 91
4.3 CODING AND THEMATIC ANALYSIS …………………………… 92
4.3.1 Tree codes …………………………………… 92
Data collection …………………………………… 93
Hypothesis related …………………………………… 94
Category 1 Activity & participation …………………………… 94
Category 2 Patients perspective & psychosocial factors …… 95
Category 3 Pathobiological mechanisms …………………… 95
Category 4 Physical impairments and structural sources …… 96
Category 5 Contributing factors …………………… 96
Category 6 Precautions and contraindications …………… 96
Category 7 Management and treatment …………………… 97
Category 8 Prognosis …………………………………… 97
Category 9 Non-specific …………………………………… 98
Planning …………………………………… 98
Examination …………………………………… 98
Management …………………………………… 99
Significant case feature …………………………………… 99
Knowledge …………………………………… 100
Propositional knowledge …………………………………… 101
Non-propositional knowledge …………………………………… 102
Professional non-propositional knowledge …………………… 102
Experiential knowledge …………………………………… 102

vii
 Structured knowledge …………………………………… 102
Personal non-propositional knowledge …………………… 103
Self awareness …………………………………… 103
4.3.2 Intra-coder reliability …………………………………… 105
4.3.3 Free codes …………………………………… 106
4.3.4 Thematic analysis …………………………………… 112
4.4 PATTERN RECOGNITION …………………………………… 115
4.4.1 Identification of pattern recognition …………………………… 115
4.4.2 Comparison of expert and novice use of pattern recognition …… 122
4.4.3 Accuracy …………………………………… 124
4.4.4 Efficiency …………………………………… 127
Pattern recognition efficiency …………………………………… 128
Comparison of efficiency between groups …………………… 129
4.4.5 Participant order of assessment questions …………………… 130
4.5 QUALITATIVE ANALYSIS FINDINGS …………………………… 132
4.5.1 Analytical process …………………………………… 132
4.5.2 Pattern related …………………………………… 133
4.5.3 Developing patterns …………………………………… 135
4.6 PARTICIPANT HYPOTHESES …………………………………… 139
4.6.1 Final hypothesis category utilisation …………………………… 139
4.6.2 Overall hypothesis category utilisation …………………………… 141
4.7 ADDITIONAL FINDINGS …………………………………… 146
4.7.1 Relative importance of data …………………………………… 146
4.7.2 Concurrent integration of data …………………………………… 148
4.7.3 Awareness of errors …………………………………… 152
4.7.4 Predictive reasoning …………………………………… 152
4.7.5 Person centred approach to assessment …………………………… 153
4.8 STUDY DESIGN RELATED RESULTS …………………………… 155
4.8.1 Case simulation data …………………………………… 155
4.8.2 Study context influences …………………………………… 159

CHAPTER 5. DISCUSSION …………………………………… 163

5.1 NON-ANALYTICAL REASONING …………………………… 164

viii
5.1.1 Accuracy of pattern recognition …………………………………… 167
5.1.2 Efficiency of pattern recognition …………………………………… 170
5.2 PARTICIPANT ORDER OF QUESTIONING …………………… 172
5.2.1 Triangulation …………………………………… 172
5.3 PATTERNS …………………………………… 174
5.3.1 Categorisation and patterns …………………………………… 174
5.3.2 Knowledge structure and pattern accuracy …………………… 176
5.3.3 Developing patterns …………………………………… 178
5.3.4 Specificity of patterns …………………………………… 180
5.3.5 Significant case features …………………………………… 182
5.3.6 Pattern elimination …………………………………… 183
5.4 HYPOTHESIS CATEGORY USE …………………………………… 184
5.5 OBSERVED APPROACHES TO REASONING …………………… 187
5.5.1 Predictive reasoning …………………………………… 187
5.5.2 Metacognition …………………………………… 188
5.5.3 Awareness of errors …………………………………… 189
5.5.4 Person centred approach …………………………………… 190
5.6 STUDY DESIGN …………………………………… 192
5.6.1 Participant recruitment …………………………………… 192
5.6.2 Experimental context influences …………………………………… 193
5.6.3 Simulated client …………………………………… 194
5.6.4 Coding process …………………………………… 195
5.7 STUDY LIMITATIONS …………………………………… 196
5.8 SUMMARY OF DISCUSSION …………………………………… 198

CHAPTER 6. CONCLUSIONS …………………………………… 199

6.1 KEY FINDINGS …………………………………… 199
6.2 IMPLICATIONS AND FURTHER RESEARCH AND EDUCATION 201

REFERENCE LIST …………………………………… 204

APPENDICES …………………………………… 218

ix
LIST OF TABLES

Table 2.1 Interpretive reasoning strategies …………………………… 8

Table 2.2 Hypothesis categories …………………………………… 22
Table 2.3 Common elements of pattern recognition described in the
literature …………………………………… 31
Table 2.4 Clinical reasoning errors in physiotherapy …………………… 55
Table 3.1 Qualitative analysis process …………………………………… 78
Table 3.2 Tree codes: predetermined code schema …………………… 80
Table 3.3 Identification features of pattern recognition …………… 83
Table 4.1 Self awareness sub-codes …………………………………… 105
Table 4.2 Intra-rater coding reliability …………………………………… 106
Table 4.3 Free nodes …………………………………… 107
Table 4.4 Themes identified …………………………………… 114
Table 4.5 Overall participant scoring tool identifying pattern recognition
in Expert D …………………………………… 116
Table 4.6 Overall participant scoring tool without evidence of pattern
recognition in Novice S …………………………………… 118
Table 4.7 Summary of pattern recognition related results …………… 121
Table 4.8 Comparison of pattern recognition use between groups …… 122
Table 4.9 Stated primary hypothesis relative to pattern recognition use 125
Table 4.10 Pattern recognition accuracy between groups …………… 126
Table 4.11 Pattern recognition accuracy overall …………………… 126
Table 4.12 Timing data relating to central hypothesis formation …… 128
Table 4.13 Total assessment time relative to reasoning method and
participant group …………………………………… 129
Table 4.14 Total assessment time relative to participant group only …… 130
Table 4.15 Final hypothesis extent of category use …………………… 140
Table 4.16 Extent of overall hypothesis category use …………………… 142
Table 4.17 Group comparisons for the use of each hypothesis category. 144
Table 4.18 Novice use of hypothesis categories …………………… 145
Table 4.19 Expert use of hypothesis categories …………………… 145

x
Table 4.20 Total assessment time relative to delayed versus concurrent
integration of data …………………………………… 151
Table 4.21 Total assessment time of participants using HDR relative to
delayed versus concurrent integration of data …………… 151
Table 4.22 Simulated client response data …………………………… 157

xi
LIST OF FIGURES

Figure 2.1 Relationship of clinical reasoning to decision-making …… 4

Figure 2.2 Reasoning strategies within diagnosis and management …. 9
Figure 2.3 Embedded design …………………………………… 17
Figure 2.4 Triangulation design …………………………………… 18
Figure 2.5 Example of diagnostic hypothesis variations based on the
range of categories …………………………………… 24
Figure 2.6 Available cues / data for developing hypotheses using PR… 34
Figure 2.7 A separate view of forwards and backwards reasoning
models …………………………………… 43
Figure 2.8 Deductive strategies following incorrect pattern recognition . 44
Figure 2.9 Pattern recognition following deductive strategies …… 44
Figure 2.10 Example of encapsulated concepts …………………… 48
Figure 3.1 Study method and chapter outline …………………………… 60
Figure 3.2 X-ray image of the grade 3 spondylolisthesis case …… 64
Figure 3.3 Research location and recording equipment …………… 72
Figure 3.4 Notebook computer recording setup in adjacent room …… 73
Figure 3.5 Embedded method design …………………………………… 88
Figure 3.6 Triangulated design …………………………………… 88
Figure 4.1 Knowledge sub-coding …………………………………… 100
Figure 4.2 Credible intervals for the identification of pattern recognition 124
Figure 4.3 Final hypothesis category use by experts and novices …… 141
Figure 4.4 Overall hypothesis category use by experts and novices …. 143
Figure 5.1 Study method triangulation …………………………………… 173
Figure 5.2 Knowledge structure and pattern types …………………… 175
Figure 5.3 Pattern accuracy and knowledge relationship …………… 177
Figure 5.4 Specificity of patterns …………………………………… 181
Figure 5.5 Input of clinical data into broad and specific patterns …… 182
Figure 5.6 Hypothesis categories and reasoning types …………… 185

xii
LIST OF APPENDICES

APPENDIX 1. ETHICS APPROVAL …………………………………… 219

APPENDIX 2. CASE SCENARIO …………………………………… 220
APPENDIX 3. ACTOR TRAINING INFORMATION ................. 226
APPENDIX 4. SIMULATED CLIENT RESPONSE CHECKLIST …… 235
APPENDIX 5. EXPERT PARTICIPANT RECRUITMENT LETTERS ... 237
APPENDIX 6. ADVERTISEMENT FOR NOVICE PARTICIPANTS....... 241
APPENDIX 7. INFORMATION STATEMENT & PARTICIPANT
CONSENT FORM …………………………………… 242
APPENDIX 8. ETHICS VARIATION …………………………………… 246
APPENDIX 9. STUDY SETUP PROCEDURES & EQUIPMENT ....... 247
APPENDIX 10. PARTICIPANT ORIENTATION SHEETS ................. 250
APPENDIX 11. INTERVIEW PROTOCOL SHEET ……………………… 251
APPENDIX 12. FINAL CODEBOOK …………………………………… 253
APPENDIX 13. PATTERN RECOGNITION IDENTIFICATION TOOLS . 255
APPENDIX 14. SELF-AWARENESS SUB-CODE EXAMPLES ………… 257
APPENDIX 15. FINAL PARTICIPANT HYPOTHESES …………… 258

xiii
ABSTRACT

Pattern recognition is a non-analytical clinical reasoning process which has

been reported in the medical and allied health literature for some time. At a
time when clinical problem solving was largely considered to consist of the
analytical process of hypothetico-deductive reasoning, pattern recognition
was introduced in the literature with observations of greater efficiency and
accuracy. The research that followed these apparent opposing models of
clinical reasoning resulted in significant growth in the understanding of
problem solving in healthcare. On commencing this thesis the knowledge
surrounding pattern recognition in physiotherapy was insufficient for its
inclusion in educational design. Consequently the aims of the study
described in this thesis were to clearly identify pattern recognition using high
fidelity case methods and observe its relationship with accuracy and
efficiency.

The study utilised a single case study with multiple participants. A real clinical
case with a diagnosis of high grade lumbar spine spondylolisthesis was
simulated using a trained actor. This provided a high fidelity case study
method allowing the observation of more realistic problem solving practices
as compared with the common low fidelity paper case approach.

Two participant groups were included in the study to investigate the common
belief that pattern recognition is an experience based reasoning process. The
expert group comprised ten titled musculoskeletal physiotherapists with a
minimum of ten years overall clinical experience and greater than two years
experience following the completion of postgraduate study. The novice group
included nine physiotherapists in their first year of clinical practice following
completion of an undergraduate degree.

Qualitative data collection methods included observation of the participant

taking a patient history of the simulated client and a stimulated retrospective
recall interview with the participant. The mixed method analysis used in the

xiv
study provided methodological triangulation of the results and supported the
presence of pattern recognition in musculoskeletal physiotherapy. The
quantitative research findings indicated that pattern recognition was
significantly more likely to produce an accurate diagnostic outcome than
analytical reasoning strategies during a physiotherapy history. However its
use was not a guarantee of success with only three of the four experts using
pattern recognition identifying the correct diagnosis. Although four experts
utilised pattern recognition as compared with only one novice, no significant
overall differences were found in the use of pattern recognition between the
expert and novice participant groups. The findings relating to time data found
that expert participants took longer to conduct the client history than novices.
Similarly those participants identified using pattern recognition also required
more time which seemingly contradicts the view of pattern recognition being
an efficient clinical reasoning process. This finding was limited by the
incomplete nature of the study which did not include a physical examination
or any client management.

xv
ABBREVIATIONS

APA Australian Physiotherapy Association

E Expert
HDR Hypothetico-deductive reasoning
LBP Low back pain
MPA Musculoskeletal Physiotherapy Australia
N Novice
PR Pattern recognition
SIJ Sacro-iliac joint

xvi
CHAPTER 1. INTRODUCTION

Clinical reasoning is a vital skill in physiotherapy. It requires critical thought,

needs to be practiced with reflection and is adapted to each unique situation.
The clinical reasoning process is the problem solving that occurs during
clinical encounters which integrate client specific information with individual
practitioner knowledge.

The clinical reasoning literature describes a few core models of reasoning

from the medical research and several other models based on qualitative
physiotherapy research over the past decade (Edwards & Jones, 2007).
These models complement each other and provide an overall view of
problem solving that adapts to the requirements of decision-making in
everyday clinical practice. The original reasoning models are associated
primarily with diagnostic reasoning and are commonly known as forwards
and backwards process models (Edwards & Jones, 2007; Higgs & Jones,
2000). Forwards and backwards refers to directional movement of the data
collection during a clinical encounter. The most common forwards and
backwards models are pattern recognition (PR) and hypothetico-deductive
reasoning (HDR) respectively.

The educational implications for PR as a forwards reasoning process model

in physiotherapy were initially considered as a topic of research. However a
review of the relevant literature found the level of understanding of PR in
musculoskeletal physiotherapy to be insufficient to enable such educational
research. PR as a phenomenon needed to be better understood first.

1.1 RATIONALE OF THE STUDY

The clinical reasoning model of PR and its attributes form the basis for the
research study reported in this thesis. Its existence as a diagnostic reasoning
model has acceptable evidence in medicine based on profession specific
research and psychology foundations (Norman et al, 2007). The research in

1
both of these domains underpins its current understanding in physiotherapy,
however questions relating to its existence and advantages remain within the
profession (Jones & Rivett, 2004).

PR has great potential as an efficient reasoning strategy (Arocha et al, 1993;

Higgs & Jones, 2000; Ridderkhoff, 1989). The modern healthcare climate has
increased scrutiny on the value of all services and challenged time efficiency
in physiotherapy clinical practice. The medical profession has also provided a
view that experts are able to generate accurate clinical outcomes via PR
(Coderre et al, 2003). Efficiency and accuracy are unquestionably desirable
in relation to clinical reasoning in modern healthcare.

Pressures for cost effectiveness in physiotherapy practice also have the

possibility to enforce potentially time saving clinical reasoning processes
upon clinicians without sufficient knowledge or experience. The consequence
of inappropriate use of PR in physiotherapy is increased risk of clinical
reasoning error (Jones, 1992). In particular, its use by novices without
sufficient knowledge or experience is commonly agreed to have negative
effects on clinical outcomes (Coderre et al, 2003; Norman et al, 2000;
Norman, 2005).

Despite the large amount of literature relating to PR, insufficient research is

available to provide sound conclusions regarding the value of PR as a
diagnostic reasoning process in physiotherapy. It is currently a commonly
used term in the physiotherapy literature without being well understood.

1.2 STUDY AIMS

This explorative study aimed to provide greater insight into PR use in

musculoskeletal physiotherapy practice. It utilised a high fidelity case study to
answer questions in relation to the existence and consequences of PR as a
reasoning strategy. The project specifically aimed to:

2
1. Determine whether PR is utilised by expert and novice clinical
physiotherapists in the musculoskeletal field
2. Relate the use of PR to efficiency within a physiotherapy assessment
3. Relate the use of PR to accuracy within a physiotherapy assessment

1.3 THESIS OUTLINE

This thesis is divided into 6 chapters including this introduction. Chapter 2

summarises the existing clinical reasoning literature in relation to PR as a
clinical reasoning process. It also provides the background in relation to
methodologies and findings of similar studies in the literature. The method of
the study is described in Chapter 3 with respect to the research questions.
Chapter 4 provides the results of the study including the supporting
qualitative data. The findings of the study are discussed in Chapter 5 and the
overall conclusions of the study form Chapter 6.

3
CHAPTER 2. REVIEW OF THE LITERATURE

2.1 BACKGROUND

Clinical reasoning is a problem solving process undertaken in healthcare. It

refers to a process that integrates cognition and knowledge in clinical
practice resulting in decision-making with a client (Higgs & Jones, 2000) for
the purpose of achieving realistic and meaningful goals. Clinical reasoning
research commenced in the medical profession more than 30 years ago and
has resulted in the development of various models that attempt to explain the
process (Norman, 2005). These empirical models have been adapted and
further developed in other health professions, whilst further models have
more recently been developed in the interpretive research paradigm (Higgs &
Jones, 2000). Clinical reasoning is undoubtedly complex and multifaceted in
nature and can be considered somewhat specific to each individual health
profession. The terms clinical reasoning and problem solving are
interchangeable terms and will be used as such within this thesis. Decision-
making is also used synonymously with clinical reasoning in the literature,
however this term has been differentiated from clinical reasoning, as
indicated in Figure 2.1.

Clinical reasoning / problem

solving

Cognitive process during

clinical encounters

Knowledge
Interpretation Decision-
of case making

Management

Figure 2.1 Relationship of clinical reasoning to decision-making

4
2.1.1 Literature search strategy

For the purpose of this literature review, searches were conducted in several
databases to maximise the capture of information relevant to the topic.
Medline, CINAHL, PsycINFO and AMED databases were searched using
both MeSH headings (if available) and key words. The MeSH headings
utilised included problem solving, decision-making and diagnosis
(differential). Key words searched in isolation and combination included:
clinical reasoning, pattern recognition, hypothetico-deductive reasoning,
cognition, metacognition, knowledge, expert / expertise, experience and
physiotherapy. These databases were searched for English papers only
throughout all available years. The ‘find citing articles’ command contained in
individual databases was used to search for more recent relevant articles.
Each paper’s reference list was scrutinised to identify further articles not
located in prior searches. Searches were repeated during the entire period of
the study to ensure that recently published papers were included.

2.1.2 Problem solving in physiotherapy clinical practice

The physiotherapy health profession has several proposed models that

describe and interpret the clinical reasoning process. These have developed
over several decades and across empirical and interpretive research
paradigms. The empirical research basis of physiotherapy clinical reasoning
lies predominantly within the medical literature, which has progressed
through the domains of hypothesis generation, memory performance and
knowledge organisation (Norman, 2005). Physiotherapy problem solving
research initially followed the medical profession within the empirical
paradigm, concluding that no single model or type of reasoning can be stated
as preferable. It was subsequently influenced by the interpretive paradigm
towards a complex expansion of reasoning models. All models of
physiotherapy clinical reasoning are now viewed as more closely related

5
based on observations of concurrent use within a clinical decision-making
process (Edwards & Jones, 2007).

2.1.3 Integration of paradigms

Jones and associates (Jones, 1995; Jones et al, 2000) describe a

collaborative hypothesis oriented model of clinical reasoning which is centred
on achieving diagnostic understanding and optimal decision-making in
collaboration with the client. Developed within the empirico-analytical
paradigm, the collaborative hypothesis oriented model has subsequently
been challenged by the interpretive research paradigm. It has been argued
that important elements of continual collaborative and interactive involvement
with the client during the problem solving process, and additional strategies
such as ethical, predictive and narrative reasoning (Edwards et al, 2004;
Higgs & Jones, 2000) provide a better overall understanding of physiotherapy
problem solving across the sub-disciplines.

The generation of a diagnostic understanding in musculoskeletal

physiotherapy, similar to medical reasoning, is an important element of
clinical practice and education. However clinical reasoning is known to
involve more than just diagnosis (Edwards et al, 2004). The complexity of
clinical reasoning has led to the expansion of models within the interpretive
research paradigm in an attempt to further understand clinical reasoning
based on phenomenology (Higgs & Titchen, 2000). Humanistic research has
allowed physiotherapy clinical reasoning to be understood via interpretive
models, such as collaborative, narrative, interactive, predictive and ethical
reasoning (Edwards et al 2004; Edwards et al, 2005; Higgs & Jones, 2000).
These models have provided a clearer picture of what occurs in
physiotherapy problem solving and also highlighted that although diagnostic
reasoning is an important component of problem solving, its role is limited
within the more complex clinical cases presenting to physiotherapists.

6
Integrating research paradigms is useful in providing a better understanding
of clinical reasoning. Edwards and Jones (2007) have comfortably integrated
both empirico-analytical and interpretive paradigms into their understanding
of clinical reasoning but without requiring a process model as previously
developed. In particular, they describe the use of interpretive reasoning skills
with more complex clinical encounters where management without diagnosis
is commonplace. The empirico-analytical reasoning models are
comparatively described in relation to more characteristic clinical
presentations with recognisable management strategies.

The development of reasoning strategies in the interpretive paradigm can be

viewed as observations from clinical practice. The interpretive paradigm does
not specifically look at the outcome resulting from the reasoning pathway or
process but rather a whole phenomenon in a real context (Higgs & Titchen,
2000). Interpretive reasoning strategies (Table 2.1) have been understood in
the physiotherapy profession via qualitative multiple case study research
(Edwards et al, 2004; Jensen et al, 2000). These strategies have been
observed to simultaneously exist in physiotherapy practice and are not
isolated strategies to facilitate decision-making. Additionally, these
interpretive strategies can co-exist alongside empirical reasoning models
such as those associated with diagnostic reasoning (Edwards & Jones,
2007).

7
Table 2.1 Interpretive reasoning strategies

(Edwards et al, 2005; Edwards & Jones, 2007; Higgs & Jones, 2000)

Reasoning strategy Brief description

Collaborative Cooperative goal setting and decision-making regarding

management

Ethical Consideration of ethical dilemmas within decision-making

and management

Interactive Social interaction as a means of developing rapport with

client and enhanced understanding of their perspective

Narrative Interprets the complexity of the client’s personal

perspective of their problems via story telling

Predictive Predicting implications of management options within

decision-making

Procedural Relating to treatment / management procedures

Teaching Client education towards further understanding of the

person and their problems

The empirico-analytical hypothesis models are based on process or pathway

throughout a clinical encounter and are particularly suited to diagnostic
reasoning. The collaborative hypothesis oriented model (Jones, 1995; Jones
et al, 2000) integrates backwards (deductive) reasoning and forwards
(inductive) reasoning and entwines knowledge and cognitive skills throughout
the process. Specific clinical reasoning models relating to backwards and
forwards reasoning types include HDR and PR respectively which are also
described alongside diagnostic reasoning. A basic view of the different
reasoning types commonly used in diagnosis and management has been
provided in Figure 2.2. This view is not to limit the scope of any single

8
reasoning type but rather to introduce PR and HDR as primarily diagnostic
reasoning strategies.

Clinical
Reasoning

Interactive reasoning
Teaching as reasoning
Procedural reasoning
Predictive reasoning
Diagnosis Management
Ethical reasoning
Collaborative reasoning

Narrative reasoning
Hypothetico-
Pattern
deductive
recognition
reasoning

Figure 2.2 Reasoning strategies within diagnosis and management (based

on Edwards & Jones, 2007)

HDR refers to the generation and testing of hypotheses based on clinical

information obtained during an entire patient assessment (Higgs & Jones,
2000). It moves in both forwards and backwards directions with several
hypotheses until sufficient evidence is gained to support a predominant
hypothesis. PR is a purely forwards reasoning process and refers to the
generation of a single hypothesis based on instant recognition of significant
case features (Coderre et al 2003; Ridderkhoff, 1991). The empirico-
analytical models of HDR (Section 2.4) and PR (Section 2.5) will be
examined in further detail with respect to the physiotherapy and medical
literature. Hypothesis generation and testing will also be explored as the
basis of both reasoning models.

9
More recently the terms analytical and non-analytical models of clinical
reasoning have been used to encompass HDR and PR respectively (Norman
et al, 2007). The term non-analytical reasoning has in essence been used as
another name for PR. It seeks to separate experience based models from
methodical problem solving processes associated with unfamiliar
circumstances. The present chapter has primarily made reference to the
original terms of HDR and PR, whereas Chapters 4 and 5 have also used the
more contemporary terms of analytical and non-analytical reasoning.

2.2 CLINICAL REASONING RESEARCH METHODS

An important pre-requisite to critique any study is an understanding of the

relevant research methods. This section provides a summary of the methods
suited to clinical reasoning research. Following this section, the clinical
reasoning literature on PR will be interpreted.

Pre-eminent among the plethora of qualitative research approaches is case

study methodology which may investigate one or more cases within a given
setting (Creswell, 2007). A case study is uniquely able to provide a view of a
specific situation in an authentic setting. Yin (2003) describes the single case
study as a useful means towards testing existing theory. This type of method
has been extensively used in medical problem solving research over the past
20 years and continues to be a popular choice amongst physiotherapy
researchers (Doody & McAteer, 2002; Edwards et al, 2004; Jensen et al,
2000; Noll et al, 2001; Smart & Doody, 2006).

2.2.1 Single case study design

A case study design is a suitable method to investigate clinical decision-

making and problem solving. A single case study is appropriate when the
phenomenon under investigation is known and a single ‘critical case’ allows it
to be further examined (Yin, 2003). A single case study in clinical reasoning

10
research can include a variety of methods from real to paper cases, with
each offering their own benefits.

2.2.2 Simulated case design

Case methodology in decision-making and problem solving research allows

for a more thorough understanding of a health professional’s thought
processes and actions. Case studies have the opportunity to be wide-ranging
in design to accommodate the needs of the researcher. The design may
incorporate the use of consistently reproducible and low cost paper or
electronic text cases (Arocha et al, 1993; Coderre et al, 2003; Coughlin &
Patel, 1987; Grant & Marsden, 1987, 1988; Hasnain et al, 2004; Joseph &
Patel, 1990; King & Bithell, 1998; Patel & Groen, 1986; Patel et al, 1990;
Patel et al, 1993; Ridderikhoff 1985), however these limit the uniqueness of a
clinician’s data collection search and offer no visual data during the process.
Video sources of client data (Rivett & Higgs, 1997) allow for visual
information to be incorporated into problem solving but also lack the
interaction necessary to effectively transfer results to everyday clinical
practice. At the opposite end of case methodology is the use of real client
cases (Doody & McAteer, 2002; Edwards et al, 2004; Embrey et al, 1996;
Gale & Marsden, 1982; Jensen et al, 2000; Noll et al, 2001; Payton, 1985)
which enable a mostly natural flowing process but are limited in their
reproducibility. A mid-point between the two ends of the spectrum is case
methodology involving simulated clients. Simulating a situation that can be
assessed in a real person allows for unique case data exploration but is
reproducible to allow for direct comparison. Such methods have been
employed in health based problem solving research (Elstein et al, 1978;
Ladyshewsky, 2002, 2004; Nendaz et al, 2004; Norman et al, 1985;
Ridderikhoff, 1991).

11
2.2.3 Simulated client assessments

Utilising simulated patients is quite common in education and research. A

simulated patient refers to a person who portrays the role of a patient for
specific educational or research purposes. They allow for more realistic
interaction with clinical reasoning tasks, including the integrated search for
data and cognitive processing during problem solving. The use of real or
simulated patients in clinical reasoning research has been referred to as high
fidelity methods, as compared with the low fidelity methods of paper or
electronic based cases (Doody & McAteer, 2002; Ladyshewsky et al, 2000).
High fidelity research methods are comparatively more costly and time
consuming however allow for more realistic and richer data.

Clinical reasoning research has extensively used low fidelity methods,

providing results that underpin the theoretical base of problem solving
including its inherent complexity. However these studies are largely contrived
in nature and lack face validity for understanding all aspects of clinical
reasoning in everyday clinical practice.

Further investigation of the use of PR within physiotherapy clinical reasoning

requires a valid method that enables all information to be immediately
available for a clinician. This enables independent choice of the pathway
through the data available. It is for this reason that low fidelity methods lack
the ability to clearly differentiate the use of PR from the more widely used
HDR strategy (Coderre et al 2003; Elstein et al, 1978). Section 2.5.1
discusses this in further detail.

A reported limitation of simulated client research relates to how accurately a

situation can be repeatedly portrayed. However the ability of a simulated
patient to provide consistent information in a physiotherapy context has been
investigated by Ladyshewsky et al (2000) who concluded that “researchers
interested in evaluating … the clinical reasoning process can use simulated

12
patient technology as a means of introducing much needed experimental
control in their studies” (p.24).

2.2.4 Verbal reported data

High fidelity methods examining clinical reasoning are reliant on verbal

reported data. Verbal reporting is a means of accessing the participant’s
thoughts when engaged in a cognitive task. Two methods utilised to obtain
this cognitive information in health based problem solving research are
simultaneous recall and retrospective recall (Patel & Arocha, 2000).
Simultaneous or immediate recall occurs during the collection of clinical
information, whereas retrospective recall follows and consequently does not
influence the normal process of clinical assessment.

Simultaneous recall occurs most commonly via ‘think-aloud’ protocols, which

require the participant to make explicit their understanding of a case either
verbally or in writing at the time of thinking. Ladyshewsky (2004) reports
these are “assumed to be direct representations of what is stored in short-
term memory” (p.18). Simultaneous think-aloud methods are considered an
excellent means of obtaining actual cognitive processes from the time of an
event, resulting in high correlation between data obtained and actual thought
processes. However think-aloud methodology also has the potential to
facilitate metacognition and create forced reflection-in-action within
participants which could significantly alter the behaviour of participants
compared with routine clinical practice. A clinician who is skilled may not be
influenced by this method, however it is likely to substantially alter the
behaviour of those not so competent.

Retrospective recall involves the verbalisation of thoughts after a cognitive

task, which has been reported as an appropriate method to obtain the
participant’s actual thoughts from the time of a cognitive event
(Ladyshewsky, 2004) but without influencing the clinical assessment process.

13
A particular method of retrospective recall involves a stimulus to trigger prior
cognitive thoughts (Elstein et al, 1978; Ladyshewsky, 2004). Observing a
video replay of an assessment is considered an ideal stimulus for recall and
has been utilised in several studies investigating clinical reasoning (Embrey
et al, 1996; Gale & Marsden, 1982; Jensen et al, 2000; Noll et al, 2001). It is
important that during the video replay the subject actually verbalises their
thoughts from the cognitive event without ‘theorizing’ at the time of recall
(Patel & Arocha, 2000). Timing of recall is important to ensure optimal
accuracy of verbal reporting. Although delayed thoughts are reported to
utilise a subject’s short and long term memory, the data obtained is
considered sufficiently similar to the thoughts during the actual event
(Ladyshewsky, 2004). Immediate review of the cognitive event allows for a
high level of accuracy of data obtained via stimulated recall. Any increase in
time delay following the event reduces the accuracy of recalled thoughts
(Barrows, 2000; Ladyshewsky, 2004; Patel & Arocha, 2000).

Verbal reported data has been criticised by some authors based on the
potential for participants to adapt to the study and report what they think the
researcher wants to hear (Elstein et al, 1990; Elstein & Schwartz, 2000). The
alternative research methods enabling investigation of clinical reasoning
have been those utilised in conventional psychology research which
emphasise the relationship between the observed responses to each
stimulus rather than participant’s verbalisations (Elstein et al, 1990; Elstein &
Schwartz, 2000). This study method also has its limitations, particularly with
respect to face validity when generalising results to real life clinical
reasoning. Collectively, the use of these different research methods should in
fact provide a better overall understanding of clinical reasoning.

2.2.5 Retrospective interview

A participant interview stimulated by video replay immediately following the

cognitive event provides a valuable source of verbal reported data. The use
of interviews to obtain research information is well recognised in many

14
professional fields (Britten, 1995) with those following an event referred to as
retrospective. Clinical reasoning research in physiotherapy has commonly
incorporated retrospective interviews with high fidelity studies (Doody &
McAteer, 2002; Edwards et al, 2004; Embrey et al, 1996; Jensen et al, 2000;
Noll et al, 2001). Careful consideration of interview method and structure is
necessary to obtain the required information that will answer the research
question.

Qualitative interviews can take a more or less structured form dependent on

the research aims. At one end of the spectrum ‘structured interviews’ are
based on standardised questionnaires which offer little flexibility to explore
the topic being considered. ‘Semi-structured interviews’ however allow for
more broad discussion via open ended questions configured to direct the
conversation yet allow the participant the liberty to converse freely (Britten,
1995). At the opposing end of the spectrum ‘in depth interviews’ have little
structure allowing the participant even more freedom in dialogue. These are
more appropriate to a grounded approach and suited to developing rather
than testing theory.

Skilled interview technique is imperative to obtain quality data and

meaningful research outcomes (Britten, 1995; Barrows, 2000). Regardless of
whether or not the interviewer is the researcher, rapport must be built with
the participant and an atmosphere of trust developed. This includes an
explanation of the purpose of the interview and clear instructions that there
are not right or wrong responses. Any verbal or non-verbal response should
not be perceived by the participant as judgmental or critical. The interviewer’s
verbal inquiry and listening skills should give control to the participant. Most
importantly it is recognised that a researcher conducting interviews should
not bring their own beliefs to a topic and should strive to keep an open mind
to new possibilities.

An interview must direct data collection towards useful information relative to

the research question. Interviews designed to validate physiotherapy clinical

15
reasoning theory require careful design of a semi-structured interview to
minimise reflection during stimulated recall but encourage recollection of
thought processes from the actual event. Audio or video technology captures
all information from an interview and allows for critical appraisal of interview
technique to ensure maximum value from the data.

2.2.6 Observational data

Observation of a participant’s behaviour is another data source that is

valuable in decision-making research. It is insufficient alone to provide a
good understanding of the cognitive actions of a health professional during
clinical practice but offers additional information that can support other data
sources such as verbal transcripts. Observations are a rich data source that
allow for behaviour to be observed in context. Videorecording during a
therapist assessment has a minimal effect on behaviour and allows for
subsequent qualitative or quantitative analysis.

2.2.7 Mixed Methods

Mixed method research refers to merging the desired aspects of qualitative

and quantitative research in single or multiple studies. Particularly within a
single study the use of both qualitative and quantitative data sources is
becoming more accepted as an approach to conducting research. Although it
has been previously reported as a means of achieving triangulation (Patton,
1990), the recent development of mixed method research as a stand alone
methodology recognises the value of integrating the traditionally separate
methods to provide greater strength in certain areas of research (Barbour,
2008; Creswell & Plano Clark, 2007). This type of methodology provides
great value in circumstances where either source of traditional data alone is
unable to provide a satisfactory understanding of the topic being investigated.

The identification of diagnostic reasoning process models requires careful

exploration of a clinician’s thoughts. To obtain such information requires a

16
study to utilise qualitative data sources such as observation or verbal
reporting, as has been extensively and appropriately used to date. However,
two commonly reported mixed method approaches suited to research in
clinical reasoning include embedded and triangulated designs (Creswell &
Plano Clark, 2007).

Embedded Design

The embedded design type generates either qualitative or quantitative data

from within the other. In clinical reasoning research the quantitative data is
generally obtained from within the qualitative sources and adds weight to the
overall interpretation (Figure 2.3). This type of embedding is similar to that
described by Yin (2003) where multiple levels or ‘subunits’ of qualitative data
are embedded within an overall single case study.

Qualitative Data

Quantitative
Data

Interpretation

Figure 2.3 Embedded design (Creswell & Plano Clark, 2007)

Qualitative data sources common in problem solving research include

verbally reported and observational data. The qualitative component usually
outweighs the quantitative element which provides a supporting role in the
overall interpretation (Cresswell & Plano Clark, 2007). Quantitative analysis

17
can vary from simple number counts to more complex statistical analysis
depending on the data generated. Embedded designs are useful in providing
greater strength to research studies of existing theory.

Triangulation

Triangulation is a well documented method of validation in qualitative

research, referring to the use of multiple perspectives on the same research
question within a single study (Patton, 1990; Richards & Morse, 2007).
Although most commonly recognised in pure qualitative research, the use of
both qualitative and quantitative data sources to achieve triangulation is
reported as a mixed method approach (Creswell & Plano-Clark, 2007).

Integrating data sets allows for a more detailed understanding of the entity
under investigation. A second data set in parallel with an embedded design
allows for a triangulation design (Cresswell & Plano Clark, 2007) as indicated
in Figure 2.4. This is reported as a type of methodological triangulation (Sim
& Wright, 2000) which may involve differing data sources within mixed
method research. Other types of pure qualitative research triangulation are
beyond the scope of this thesis.

Qualitative Data
Quantitative
Quantitative Data
Data

Interpretation

Figure 2.4 Triangulation design (Creswell & Plano Clark, 2007)

18
2.3 HYPOTHESIS GENERATION

Physiotherapy clinical reasoning results in the generation of hypotheses.

Musculoskeletal physiotherapy in particular relies on an initial assessment to
gain a hypothetical understanding of each client’s case and continual
reassessment of this hypothesis. Previous research has shown that
musculoskeletal physiotherapists do generate hypotheses during patient
assessments (Payton, 1985; Rivett & Higgs, 1997). The study conducted by
Rivett and Higgs (1997) involved a video-taped patient history of a lumbar
disorder. They found all nineteen physiotherapists, with varied levels of
experience, generated hypotheses. The development of hypotheses is
considered essential to interpret and manage the considerable amount of
data available in a clinical encounter.

The nature of a hypothesis should be considered when interpreting previous

problem solving research. A hypothesis in clinical reasoning should be
regarded as the understanding or interpretation of the presenting information
at any stage of the process (Rivett & Higgs, 1997). There may be a single
hypothesis or several competing hypotheses whilst undertaking clinical
reasoning. Groen and Patel (1985, p. 95) define a hypothesis as “a verbal
statement about a situation that may either be true or false”. It is essential
that a clinician undertakes an assessment to interpret and understand the
presenting problems. A clinician’s understanding can be represented as a
working hypothesis that can be tested to confirm or negate its accuracy. In
other words, the hypothesis becomes the basis for further reasoning and
testing (Anderson, 1989).

The use of the term hypothesis in clinical reasoning should be separated

somewhat from that utilised in scientific research. Ridderkhoff (1991)
discusses the less precise use of the term ‘hypothesis’ in clinical problem
solving as compared with the specific meaning of the word in scientific
research. In clinical practice, testing via questioning or physical examination

19
may provide information that rejects the primary hypothesis, however the
clinician rarely sets out to disprove their hypothesis (or prove the null
hypothesis) which is often the case in experimental research. It has been
suggested by prominent researchers in medical clinical reasoning that
confirmation strategies are more often utilised in clinical problem solving than
those that negate a hypothesis (Arocha, Patel & Patel, 1993). Physiotherapy
clinical reasoning includes testing that may confirm or reject a clinical
hypothesis, however it is unknown if either is more commonly utilised.

The complexity of a patient encounter requires hypothesis formation to

manage and organise the presenting information throughout the entire
process. Hypothesis generation has been stated as a “psychological
necessity” due to the potential complexity of a clinical case (Elstein, Shulman
& Sprafka, 1990, p. 9). The information available from any patient often
outweighs the capacity of the working memory and requires ‘chunking’
groups of information. One possible means of separation or chunking of
clinical data involves using groups of hypotheses known as categories.

2.3.1 Hypothesis categories

Clinical hypotheses may be developed under the guise of various categories.

Several versions of hypothesis categories have been reported in the medical
(Barrows & Feltovich, 1987) and physiotherapy literature (Jones, 1992; Jones
& Rivett, 2004; Payton, 1985; Rivett & Higgs, 1997). Categorising hypothesis
types is beneficial from a research and educational viewpoint. These
divisions in thought processing have evolved alongside the progressive
models of healthcare, such as the mature organism model (Gifford, 1998;
Jones & Rivett, 2004). Jones and Rivett (2004) have detailed several
categories (Table 2.2) that are suited to hypothesis generation in current
musculoskeletal physiotherapy practice and are influenced by progressive
healthcare models such as Gifford’s mature organism model (1998). These
categories allow for an overall clinical understanding within health and

20
disability models that provide a more holistic view of a person and their
problem rather than merely the pathology.

The hypothesis categories in Table 2.2 allow for a broad array of possible
clinical descriptions within the same clinical case. The utilisation of some of
the individual categories as reported by Jones and Rivett (2004) has not
been investigated in current clinical practice and as such the relative
frequency of their use remains somewhat unknown. Only a few studies of
physiotherapy clinical reasoning have recorded frequency of use in differing
hypothesis category classifications (Payton, 1985; Rivett & Higgs, 1997).
Payton (1985) utilised the following categories (observed occasions in
parentheses): pathological (3), pathokinesiological (18), pathophysiological
(8), and psychosocial (5). Rivett and Higgs (1997) reported on an earlier
version of those presented in Table 2.2, including: source of the symptoms
and / or dysfunction, contributing factors, precautions for and
contraindications to physical examination and treatment, management,
prognosis, mechanisms of signs and symptoms, and reassessment.
Consideration of hypothesis categories is useful reflection in preventing a
narrow view of musculoskeletal physiotherapy clinical reasoning, or in other
words expanding on basic impairment / structural problem solving.

21
Table 2.2 Hypothesis categories (Jones & Rivett, 2004, pp.13-20)

Category Definition
Activity and Concerns the capabilities or restrictions of an individual
participation during a specific activity or being involved in a life situation.
Patient’s perspective / An individual person’s perspective may be considered in
psychosocial factors terms of their understanding, feelings or beliefs related to
the presenting problems. The patient’s perspective may be a
contributing factor or a consequence of the pain or
restriction in activity / participation, but either way may be
relevant in the recovery process.
Pathobiological Consideration of tissue healing and pain mechanisms allows
mechanisms for reasoning related to initial onset or maintenance of signs
and symptoms by the nervous system. Normal tissue
healing can be a basis for hypothetical understanding of a
patient’s presentation. Pain mechanisms can help in
understanding the activity or participation levels, patient’s
perspectives and physical impairments. These refer to the
input mechanism of pain, the central nervous system
processing of the input, and the output mechanisms that
may result in distorted movement patterns and motor
activity.
Physical impairments & Impairments identified in the physical examination are
associated structure / atypical findings in the neuromusculoskeletal system.
tissue sources Associated pathological sources are the structures or
tissues hypothesised to be related to the client’s symptoms
and signs. This category alone is insufficient to understand a
problem, its effect on a patient or the reason for
management.
Contributing factors A causative factor may be “environmental, psychosocial,
behavioural, physical / biomechanical, and even hereditary”
(p. 17). This category refers to any aspect of a client’s case
that may contribute towards or is associated with the onset
or maintenance of the presenting problem(s).
Precautions & Hypotheses regarding precautions and contraindications to
contraindications examination and / or treatment. The type of pathology /
disorder, stage of healing, severity / irritability, and patient’s
perspectives must all be considered when hypothesising in
this category. Precautions and contraindications can be
viewed simply as safety related hypotheses.
Management & Any intervention to assist a patient towards recovery or
treatment achieving their stated goals, including specific treatment
techniques, is considered a hypothesis in management or
treatment.
Prognosis Predicting a possible response to treatment intervention or
an outcome for a particular problem / pathology is a
prognostic hypothesis. Features of a case that may
influence the outcome in a positive or negative way can be
labelled as prognostic features.

22
2.3.2 Diagnostic hypotheses

A clinical diagnosis can be considered as the final hypothesis of a case that

is most probable following examination and testing. Diagnostic reasoning in
physiotherapy has been referred to as the process undertaken whilst forming
a diagnosis within certain categories (Edwards & Jones, 2007), as listed in
Table 2.2. Edwards and Jones (2007) state a diagnosis relates primarily to
hypotheses of activity / participation and associated physical impairments
and structure / tissue sources. It should also consider the pathobiological
mechanisms and potential contributing factors. In other words, a
physiotherapy diagnosis is typically limited to four of the eight hypothesis
categories reported by Jones and Rivett (2004). Hypotheses relating to
management, prognosis and precautions / contraindications should be
considered equally important within problem solving although may not
contribute to diagnostic reasoning. The patient’s perspective / psychosocial
factors category is noted as an essential consideration in any clinical
encounter due to its impact on overall outcomes but this area is also arguably
outside the scope of usual diagnosis within physiotherapy.

A diagnostic hypothesis within clinical reasoning refers to a statement that

interprets a case (Joseph & Patel, 1990). Clinical reasoning research often
enters into the experimental domain of accuracy which relies on a
participant’s diagnostic statement for comparison with a known outcome.
This comparison is not without its own complications given the array of
different descriptions for any given case study.

The range of hypothesis categories makes it possible that the same

interpretation of a case could be labelled in several ways with respect to a
final hypothesis. An example of a hypothesis combining several categories
would be ‘the patient is unable to climb stairs (activity restriction) with the
mechanical limitation and pain resulting from excessive inflammation
(pathobiological mechanism) within the synovial joint capsule of the knee

23
(structure / source)’. Alternatively a diagnostic hypothesis may lie within only
one of the categories. Figure 2.5 displays the use of different hypothesis
categories to explain the variations in stated final hypotheses despite a
similar understanding of a case. This has implications for designing studies in
diagnostic accuracy.

Clinical case

Therapist Therapist
Therapist
A C
B

Pathobiological Activity & Contributing Physical impairments /

mechanisms participation factors structural sources

Similar overall case understanding described using different hypothesis categories

Figure 2.5 Example of diagnostic hypothesis variations based on the range

of categories

2.3.3 Basis for empirico-analytical reasoning models

Hypothesis generation is the basis for empirico-analytical process models of

clinical reasoning. Studies conducted by Elstein et al (1990) concluded that
subjects generated hypotheses whether instructed to or not and also did so
when specifically instructed not to. This clearly supports the view that
hypothesis generation is central to clinical reasoning. Hypothesis generation
is specifically a requirement for PR to be identified and is included as a stage
of HDR.

24
2.4 HYPOTHETICO-DEDUCTIVE REASONING

HDR was initially developed as a general model of the clinical reasoning

process. It allowed for the generation and evaluation of competing
hypotheses as a result of collecting and interpreting clinical data. It provided
an early understanding of clinical reasoning as a process or hypothesis
generating pathway through a clinical encounter.

2.4.1 Defining hypothetico-deductive reasoning

The HDR model was first developed as a general problem solving strategy by
Elstein and associates in the medical profession (Elstein, Shulman &
Sprafka, 1978). It provided a sequence of problem solving stages that could
be associated with any clinical encounter. As implied in the term, the
deductive process refers to a search for findings to support or validate a
limited number of previously generated hypotheses (Elstein & Schwartz,
2000). It has been the base of much clinical reasoning research and debate
over the last three decades.

HDR commences with the ‘acquisition of cues’, which refers to collecting or

gathering data from the many available sources throughout an organised
clinical assessment (Elstein, Shulman & Sprafka, 1978). This search aims to
identify cues that are relevant and meaningful to the clinician allowing for
subsequent ‘generation of hypotheses’.

The second stage of ‘generation of hypotheses’ based on the relevant data

acquired is necessary for problem solving in clinical practice (Elstein,
Shulman & Sprafka, 1978). Hypotheses form a framework to consider a
problem and guide further data collection. As previously discussed in section
2.3, this is a necessity when confronted with a substantial amount of clinical
information.

25
The third stage relates to the evaluation of cues relevant to each hypothesis
generated. ‘Cue interpretation’ requires awareness and knowledge of the
various data sources to understand the value of each cue in supporting the
hypotheses (Elstein, Shulman & Sprafka, 1978). The validity and reliability of
each question in the history or specific test during a physical assessment
must be considered fundamental for accurate cue interpretation.

The final stage of HDR is ‘hypothesis evaluation’ or judgement of competing

hypotheses (Elstein, Shulman & Sprafka, 1978). This stage of the process
attempts to achieve a final understanding of the case but can also trigger
continued deductive stages. Although stated in four stages, these are not
strictly linear in nature and overlap exists. The process undoubtedly starts
with acquiring data and ends with evaluating competing hypotheses, however
hypothesis generation and cue interpretation could occur concurrently in an
experienced clinician. Data collection then continues throughout the entire
process but it becomes more targeted in the later stages of the process. The
development of HDR was an important step in understanding the intrinsic
nature of medical problem solving.

2.4.2 Hypothetico-deductive reasoning in physiotherapy

The origins of clinical reasoning in physiotherapy relied heavily on the work of

medical problem solving research. The identification of both hypothesis
generation and HDR as a model of problem solving was initially found in
musculoskeletal physiotherapy. An early physiotherapy study replicated that
of Elstein and associates, which included physician assessment of simulated
patients followed by video stimulated retrospective recall methodology
(Elstein, Shulman & Sprafka, 1978). Payton (1985) investigated 10
physiotherapists assessing real patients and identified the application of all
four stages of the HDR process. He concluded that similar clinical reasoning

26
was used by physicians and musculoskeletal physiotherapists (Payton,
1985).

The belief that HDR existed in musculoskeletal physiotherapy was further

supported by Rivett and Higgs (1997) who utilised observation of a prior
recorded video case in a problem solving study. This research involved both
expert and novice physiotherapists who all generated hypotheses during
problem solving, consequently providing further evidence that HDR was
utilised in physiotherapy practice.

HDR has been incorporated in undergraduate and postgraduate

physiotherapy programs as a structural basis for clinical reasoning (Higgs,
1993). However given that HDR cannot adequately explain all problem
solving practices, its inclusion in teaching encouraged reflection on the
analytical thought processes occurring during physiotherapy clinical
reasoning. Professional education subsequently incorporated a profession
specific model based on HDR (Jones, 1995; Jones et al, 2000) with the
understanding that it is a process utilised within diagnostic reasoning,
especially that of novices (Edwards & Jones, 2007).

2.4.3 Characteristics of hypothetico-deductive reasoning

HDR is known to be a slow but thorough analytic process that when utilised
by a skilled clinician leads to effective management of more complex and / or
unfamiliar situations. It is a process recognised in physiotherapy to continue
beyond management strategies. Encouraging critical reflection on response
to treatment interventions enables further evaluation and refinement of the
final hypothesis.

An early criticism of the HDR model related to its generality, which refers to
its applicability to all problem solving situations. All analytical problem solving
models like HDR have generality limitations when attempting to understand
clinical reasoning. The problem solving approach varies depending on the

27
case and the clinician’s knowledge in the particular domain (Elstein et al,
1990; Groen & Patel,1985; Norman, 2005). The original misconception that
HDR is the sole explanation for clinical reasoning should not reduce the
value it has provided in understanding part of the problem solving process
within medicine and physiotherapy.

HDR has also been criticised for being a strategy utilised predominantly
when knowledge and experience is insufficient. Groen and Patel (1985) refer
to the cognitive psychology literature on differences between novice and
expert, and note that HDR is characteristic of novices. Expertise research in
medicine supports this view via a lack of association between HDR
processing and expert practice (Norman, 2005). This leaves us with the view
that an expert operating in their domain does not generally use HDR
processing, but novices who are attempting to develop a structured
knowledge base across many areas of practice do rely on analytical and
backwards problem solving strategies such as HDR.

The backwards nature of HDR and its relative use in less familiar situations is
thought to result in overall inefficiency when compared with other forwards
reasoning strategies. The formation and testing of several competing
hypotheses during a deductive assessment process is slowed particularly by
the need to gain information that supports or negates each hypothesis. HDR
has been reported as a slow process when compared with inductive type
processes or what is actually observed in experts within familiar situations
(Arocha et al, 1993; Higgs & Jones, 2000; Jones & Rivett, 2004; Patel &
Groen, 1986). Although relatively inefficient, HDR remains commonly utilised
by novices and a ‘fall back’ strategy for experienced clinicians encountering
complex or unfamiliar cases requiring a diagnostic approach.

28
2.4.4 Summary of hypothetico-deductive reasoning

HDR is a problem solving strategy utilised by physiotherapists (Payton, 1985;

Rivett & Higgs, 1997), however it is limited in providing a full understanding of
clinical reasoning (Norman, 2005). It provides a good structural basis to
analytical problem solving within clinical practice but falls short of explaining
all observations of clinical reasoning. HDR is considered applicable in
situations where experience and knowledge isn’t readily available with
respect to the case at hand.

Diagnostic reasoning in physiotherapy practice should not be considered as

either a backwards or a forwards process but rather an integration of process
models. An integrated view of HDR and PR may provide a better
understanding of diagnostic reasoning.

2.5 PATTERN RECOGNITION

Clinical reasoning in physiotherapy can at times be observed as recognition

based on previous experience. This refers to patterns derived from
experience with similar patients / conditions which form a prototype in a
clinician’s non-propositional knowledge base. The pattern is triggered when
similar case features are confronted and a hypothesis relating to the
presenting case is consequently formed. Like HDR, PR is a diagnostic
reasoning strategy (Edwards & Jones, 2007) that can be employed at an
appropriate time within problem solving. It has been associated with the
terms inductive and forwards reasoning (Higgs & Jones, 2000) which refer to
the movement from cues to hypothesis. In contrast, backwards or deductive
reasoning moves from hypotheses back to clinical data for further testing of
each hypothesis. Induction from cues and deduction from hypotheses
(Arocha et al, 1993; Coughlin & Patel, 1987) creates a simple separation of
forwards and backwards movement during clinical reasoning.

29
PR is notably only one type of several inductive models of clinical reasoning.
Ridderkhoff (1989) also describes inductive-heuristic and inductive-algorithm,
alongside PR, as inductive reasoning models. However it is PR that almost
exclusively exists as a forwards reasoning model in the physiotherapy
literature.

2.5.1 Defining pattern recognition

A barrier to understanding PR as an interpretation of the clinical reasoning

process lies with its varied descriptions in the literature. There is a clear need
for consistent terminology amongst researchers of clinical reasoning
(Barrows & Feltovich, 1987), however the term ‘pattern recognition’ has been
used with varied meanings (Gale & Marsden, 1983), a situation that
continues at present. Despite the lack of a universal definition of PR,
common elements have been identified from a review of the literature in
medicine and allied health. Table 2.3 identifies the common elements of PR
reported in the literature.

30
 Table 2.3 Common elements of pattern recognition described in the literature

Element Description Research articles Commentary

Timing Immediate / almost Arocha et al, 1993 Caputo & Mior, 1998
instantaneous Coderre et al, 2003 Higgs & Jones, 2000
Doody & McAteer, 2002
Groves et al, 2002
Result Hypothesis formation Coderre et al, 2003 Round, 2000
Doody & McAteer, 2002
Noll et al, 2001
Reliance Organised knowledge Gale & Marsden, 1982 Edwards & Jones, 2007
from prior experience Ridderikhoff, 1985 Norman et al, 2007
Rivett & Higgs, 1995
Utilises Significant case Coderre et al, 2003 Caputo & Mior, 1998
features Groves et al, 2002
Noll et al, 2001
Basis Highly organised Ridderikhoff, 1985 Caputo & Mior, 1998
knowledge Roberts, 1996 Jones & Rivett, 2004

Prototypes of single
patient or abstract
model (several
patients combined)
Direction Forwards reasoning Coughlin & Patel, 1987 Edwards & Jones, 2007
strategy Noll et al, 2001
Patel & Groen, 1986
Patel et al, 1990

31
The synthesis of these common elements found in the literature leads to an
overall understanding of PR as a process within clinical reasoning:

PR involves immediate hypothesis formation based on a pattern of

highly significant features / cues recognised in a case that are
matched with similar instances from prior experience. It is deemed to
have occurred when a single hypothesis is formed based on a
prototype from a similar case and context.

Single hypothesis

The stand out feature that clearly separates PR from other hypothesis
generating models is the single dominant hypothesis. A recent study
undertaken by Coderre et al (2003) attempted to determine the diagnostic
reasoning strategy used by novices and experts in medicine. They labelled
PR as one strategy and identified its use via “a single diagnosis with only
perfunctory attention to the alternatives” (Coderre et al, 2003, p.703).
Ridderkhoff (1989, 1991) also identified PR as a type of inductive reasoning
with a single hypothesis.

It has been argued however that more than one hypothesis can be
considered as a result of PR. Arocha et al (1993) refer to different data
sources (cues) triggering different hypotheses within a forwards reasoning
process. Although this is probably valid, it becomes increasingly difficult to
separate PR from hypothesis generation within a HDR process when more
than one unrelated hypothesis is present throughout a clinical assessment.

Significant case features

The utilisation of significant case features to move forwards within reasoning

is also a key component of PR. The cues or features of the case utilised for
hypothesis generation have been shown to influence the forwards reasoning

32
process. Patel et al (1990) demonstrated that comparison of irrelevant cues /
features against the primary hypothesis disrupted the forwards reasoning
process. On this basis, pure PR should be evident by the use of predominant
significant case features that relate directly to the primary hypothesis.
Contrary to this, following the generation of an initial hypothesis, any
unrelated data collection may indicate a backwards process.

Information availability

A research design consideration has been reported in relation to identifying

PR. Elstein et al (1978) contend HDR can only be utilised if all the clinical
information is not initially present. This raises a technical issue whereby
identifying PR using low fidelity methods (section 2.2.3) requires immediate
availability of all case data. A recent study of diagnostic reasoning strategy
utilised clinical case vignettes that had all the necessary information available
to the participants (Coderre et al, 2003). However the use of such case
vignettes with essential information available at the outset has also been
criticised as having “problems of temporal unfolding” (Barrows & Feltovich,
1987, p.89). It is clear that clinical practice involves progressive data
acquisition and a ‘temporal unfolding’ of information. These research design
issues are best avoided by using high fidelity methods such as real or
simulated clients.

A single hypothesis generated via PR can occur at any time throughout an

assessment. Noll et al (2001) demonstrated that recognition of similar cases
can occur during either the history or physical components of a
physiotherapy examination. This finding is again important with respect to
study design and would suggest PR research requires high fidelity methods.

In musculoskeletal physiotherapy a large number of visual cues such as non-

verbal communication, body type, posture, and functional movement are
quickly obtained via initial patient observation. Auditory data acquired by
verbal communication are obtained sequentially. Tactile data along with the

33
observations of specific movements do not usually add to the assessment
until a thorough understanding of the problem has already been formed (i.e.
after reaching the initial hypothesis). Norman et al (1992) highlighted in
radiology that a visual cue alone can trigger a diagnostic hypothesis. Figure
2.6 illustrates the range of cues within a physiotherapy assessment. This
suggests that PR is more likely to relate to visual or verbal cues during the
early stage of a clinical encounter but may possibly occur at any stage during
an assessment.

Patient

Visual Verbal Tactile

General availability to a therapist during an assessment

Availability during a physical assessment

Most likely stage for hypothesis

generation via pattern recognition

Figure 2.6 Available cues / data for developing hypotheses using PR

2.5.2 Pattern recognition in physiotherapy

Qualitative research in musculoskeletal physiotherapy has identified PR

(Doody & McAteer, 2002; King & Bithell, 1998; Noll et al, 2001). However

34
varied definitions of PR create discord amongst the findings of these studies.
Doody and McAteer (2002) described PR only by movement from hypothesis
generation to evaluation and bypassing the appraisal of data relative to the
hypothesis. Hypothesis evaluation in this study referred to assessing each
hypothesis for best fit with the case. Noll et al (2001) reported PR as a
qualitative code, labelled as a pattern of data (signs and symptoms)
recognisable from prior experience and used to develop a working
hypothesis. King and Bithell (1998) used only reported similarity to previous
clinical experience, during a retrospective interview, as identification of PR.
These studies provide some support for PR based on each researcher’s
understanding of the model, however the evidence supporting PR as a
diagnostic strategy in physiotherapy lacks strength.

2.5.3 Characteristics of pattern recognition

The synthesis of PR definitions from the literature may provide some clarity in
describing its elements as a model. Further insight can be provided with
respect to knowledge, categorisation, efficiency and accuracy. These
characteristics relate to or underpin PR as previously described and offer
potential benefits within musculoskeletal physiotherapy research.

Knowledge

Knowledge is undoubtedly a factor when considering PR or inductive

reasoning. The recognition of a unique clinical pattern is dependent upon an
individual’s propositional and non-propositional knowledge. A highly
organised knowledge structure is reported as a basis for PR (Edwards et al,
2004; Jones, 1992). Clinical patterns rely on an elaborate network of clinical
and biomedical / biopsychosocial knowledge structures. In other words, the
clinical case is recognised by certain features but linked with a complex
integrated understanding of the presentation. PR facilitates this knowledge
link between clinical patterns and their underlying (knowledge) structures.

35
Categorisation

The theory of categorisation has been utilised when dissecting PR. Several
authors advocate categorisation, which refers to the grouping of objects or
events, as a means of understanding PR (Brooks et al 1991; Hayes &
Adams, 2000), while others believe categorisation and PR utilise similar
concepts in achieving a diagnosis (Elstein & Shwartz, 2000 & 2002).
Categorisation specifically refers to the comparison of “two or more
distinguishable cases, objects or events” (Hayes & Adams, 2000, p. 45). The
recognition of a clinical pattern relies on finding a similarity between separate
but similar cases and could therefore be considered as a type of
categorisation.

Models of categorisation identified in the cognitive psychology literature are

able to provide further insight into clinical reasoning. Hayes and Adams
(2000) provide a summary of two models of categorisation relevant to
medical clinical reasoning. First, the prototype model has features such as
characteristic signs and symptoms of a condition, which have been
abstracted and stored as a summary. The prototypes develop over time as a
result of clinical exposure to similar cases. Cognitive psychology research
has provided considerable support for the prototype model of categorisation
(Hampton, 1998). Second, exemplar-based processing models involve each
individual instance being memorised. The exemplar-based models have also
been referred to as instance-based recognition where “a new instance is
classified by resemblance to memory of a past case” (Elstein & Shwartz,
2000). A pattern may therefore be viewed as a single case from experience
(i.e. exemplar) or more commonly one that is formed over time with multiple
clinical experiences and associated integration of knowledge (i.e. prototype).

Categorisation clearly resembles PR at a superficial level, depending on the

definitions applied. However one opposing view is that categorisation theory
should be regarded not just by the recognition of patterns but rather as the
grouping of knowledge structures with underlying meanings and associated

36
actions (Hayes & Adams, 2000). This view ultimately depends on the depth
of understanding of PR.

A pattern can be simplistic in nature like the recognition of an acromio-

clavicular joint injury from the description of a fall onto the point of the
shoulder and the observation of a joint deformity on visual inspection. Even
this simple pattern would likely have a knowledge base involving a network of
Patho physiological, anatomical, prognostic and management features that
are all enabled on recognition. At the other end of the spectrum, more
elaborate and intricate networks of knowledge can allow recognition of more
complex conditions such as cervicogenic headaches. These may be
triggered by a known pattern using a few features of the case and then
confirmed with further data collection. Regardless of the level of complexity,
the clinical patterns of a reflective practitioner will have an underlying basis
that is similar to that described in categorisation.

Efficiency
Efficiency in clinical practice has become a feature of modern healthcare
provision. Considerable external pressure is placed on clinicians from various
sources, including business management within both public and private
sector practice. The government currently demands evidence of value for
money in public sector health. Similarly the private sector has the internal
pressure of maintaining financial viability in an economic environment that is
becoming more client / user funded (Higgs & Jones, 2000). Clinical reasoning
remains a complex and potentially imprecise part of healthcare, however
some authors have argued that forwards reasoning strategies such as PR
are more efficient (Arocha et al, 1993; Higgs & Jones, 2000; Ridderkhoff,
1989) and potentially cost effective than backwards reasoning models such
as HDR. These claims are apparently based on plausible theory but as yet
are not supported by research evidence within physiotherapy.

37
Early research in medical reasoning found that experts generate a diagnostic
hypothesis earlier than novices during a patient’s history (Joseph & Patel,
1990). More recent qualitative studies of physiotherapy clinical reasoning
have also found differences in timing between novices and experts (King &
Bithell, 1998; Doody & McAteer, 2002). The experts in these studies were
however reported to spend more time taking a patient history and longer to
express their initial hypothesis. This view supports earlier cognitive
psychology research findings that experts are often slower during the initial
phases of problem solving but are faster overall (Glaser & Chi, 1988). Doody
and McAteer (2002) conducted an assessment of a real patient and
compared the time taken for all parts of each management session. The
experts spent less time on the physical examination, although this was not a
statistically significant finding. Review of the small amount of data available
relating to timing and efficiency does not produce any conclusive finding.
Nevertheless, Doody and McAteer (2002) contend that experts have refined
a more definitive hypothesis by the end of the patient history and use more
specific confirmation testing during the physical examination, thus requiring
less overall time compared with novices.

PR may indeed be associated with efficiency in clinical practice but only in

those familiar cases associated with a well developed structure of
knowledge. Any relationship between efficiency and PR is therefore reliant on
familiarity and knowledge of the case at hand. PR may be an efficient
process when used accurately by an expert in a given domain, however this
possibility at present lacks empirical evidence.

Accuracy
The ability of an expert clinician to develop an accurate diagnostic hypothesis
has been previously proposed to be influenced by the direction of reasoning.
Early research in medicine has found associations between forwards
reasoning and the accuracy of diagnostic performance (Patel & Groen,
1986). More recently, Coderre et al (2003) found similar results in that PR

38
had the greatest likelihood of producing an accurate diagnosis during their
study. These studies notably included the use of low fidelity methods
involving diagnosis of a written / paper case.

In professions such as radiology and dermatology that are largely based on

visual perception, diagnostic accuracy has been associated with case feature
recognition from previous cases (Norman et al, 1992). This provides some
support for visual cues facilitating accuracy in PR, but is not the same as the
more complex and multi-faceted interaction required in a physiotherapy
assessment.

These medical studies generally support the contention that forwards

reasoning strategies may be more accurate with experts, but all notably have
utilised either visual cues or low fidelity paper case methods. Indeed, no
studies have looked at diagnostic accuracy with high fidelity methods such as
real or simulated patients.

Accuracy can only be assessed by comparison of a clinician’s understanding

of a case with the actual case diagnosis. Diagnostic hypotheses have been
criticised by Gale and Marsden (1982, p.26) who believe that a diagnostic
hypothesis focus may “overshadow equally important prior stages in the
clinical problem solving process”. This is true if the participant’s focus is to
produce a diagnosis, however a carefully designed study should be able to
observe the reasoning processes undertaken and the evolving understanding
of a case. This may result in a final diagnostic understanding in the form of a
hypothesis without impacting on the processes utilised along the way.

There is only one study in the physiotherapy literature that sheds any light on
the accuracy of the clinical reasoning process. King and Bithell (1998)
conducted a study involving a segmentally reviewed paper case integrated
with a structured interview. All five participants in the ‘physiotherapy

39
specialist’ group provided an accurate diagnosis compared with only one of
five in the ‘generalist’ group. This study concluded that there is an
association between diagnostic accuracy and reported recognition of
previous cases. No studies investigating accuracy with respect to clinical
reasoning process have been found which use high fidelity methods such as
real or simulated patients.

Opposing views exist in the literature relating to the potential of PR to result

in errors. Indeed the use of this form of reasoning by novices has potentially
negative consequences due to the lack of consolidated clinical experience
and a sound knowledge base (Coderre et al, 2003; Norman et al, 2000;
Norman, 2005). Jones (1992) in particular has raised caution regarding the
use of PR within physiotherapy assessment due to the increased risk of
errors, referring to limited flexibility within hypothesis generation leading to
errors in problem solving. Jones (1992, p.882) specifically warns that
“anything that has any resemblance to a standard pattern will be seen as that
pattern”. This is more likely with respect to the clinician who is not a master of
their domain.

The association between PR and accurate outcomes in diagnostic reasoning

is not fully understood at present. The potential link between these problem
solving variables is undoubtedly complex and dependent on knowledge
structures and experience. No conclusive research evidence is currently
available in physiotherapy that demonstrates PR is more accurate or
alternatively leads to errors. The possibility that PR leads to more accurate
outcomes in expert clinicians is worthy of further exploration.

2.5.4 Summary of pattern recognition

The diagnostic reasoning strategy of PR has been primarily reported in the

medical and physiotherapy literature. Despite the inconsistencies relating to

40
its interpretation, PR is commonly reported as a forwards reasoning process
model. The majority of research evidence relating to PR lies in medicine and
this has predominantly used low fidelity methodology.

The recognition of a pattern may be simplistic in nature but the understanding

of patterns that lead to efficient and accurate outcomes is of greater
complexity. Consideration of PR requires an understanding of what actually
constitutes a pattern, which necessitates consideration of the knowledge
literature. A review of the integration of forwards and backwards models
within diagnostic reasoning is also important to better understand the clinical
reasoning process.

2.6 INTEGRATION OF DIAGNOSTIC REASONING MODELS

The clinical reasoning literature suggests that neither PR nor HDR alone are
sufficient to understand clinical problem solving. Elstein and Shwartz (2000)
refer to the difficulty of a specific case and the experience of the clinician as
predictors of the problem solving strategy utilised in any given clinical
situation. On this basis any single strategy on its own cannot be simply
considered as superior (Norman & Eva, 2003). Grant and Marsden (1987)
give support to the view that no single best way of thinking exists when
determining a medical diagnosis and that varied pathways (and thus models
of clinical reasoning) can exist to reach similar conclusions. In a study
identifying case interpretations and forceful (clinical) features, they
demonstrated that experts did not use the same thought processes nor did
they utilise the same important case features to reach their interpretations
(Grant & Marsden, 1987). Diagnostic reasoning can therefore be described
as complex and ever changing and include both the forwards and backwards
models of PR and HDR respectively.

Physiotherapy clinical reasoning has been explained using a model allowing

both forwards and backwards movement during problem solving (section
2.1.3). This collaborative clinical reasoning model (Jones et al, 2000) places

41
more emphasis on the backwards-reasoning loops and multiple hypotheses
in the clinical reasoning process, however also allows for forwards reasoning
or PR to occur with confirmation testing of a single hypothesis.

2.6.1 Further exploration of forwards reasoning

It is widely believed that forwards reasoning strategies are utilised within

clinical reasoning, however there is a paucity of evidence within the
physiotherapy literature supporting this presumption. The existence of PR as
a model of forwards reasoning has been observed in expert physiotherapists
(Noll et al, 2001; Doody & McAteer, 2002), but our understanding
surrounding these observations is limited. Despite the existing beliefs
regarding both the medical and physiotherapy professions, a full
understanding of PR as a model of forwards reasoning requires further
exploration. This understanding must lead further down the path of any
potential associated benefits. In particular, the diagnostic accuracy and time
efficiency should be investigated to help determine the overall benefit of PR
to clinicians of varying levels of experience.

Notably, the assessment of accuracy with respect to the direction of

reasoning has not been undertaken with high fidelity methods in
physiotherapy research. If a particular process alone or a specific
combination of processes were found to be consistently more accurate in
diagnosis, this could provide evidence to support these clinical reasoning
processes being utilised by clinicians.

2.6.2 Separating forwards from backwards reasoning

It is generally believed that both forwards and backwards reasoning will be

utilised during physiotherapy clinical practice. Unfamiliar or more complex
cases are believed to require a backwards approach alone (Jones, 1992;

42
Rivett & Higgs, 1997), whilst pure forwards reasoning is likely only in familiar
cases with little variance in differential diagnosis (Figure 2.7). It is plausible
that an integration of both directions of reasoning may well occur in many of
the semi complex cases presenting to clinicians.

Data collection

Forwards Backwards
reasoning reasoning

Single Multiple hypotheses

hypothesis
Working
Generated, rejected, refined, modified
hypothesis

Confirmation testing

Final hypothesis

Figure 2.7 A separate view of forwards and backwards reasoning models

Several integrated combinations of forwards and backwards processes are

probable within physiotherapy assessment. A condition may be sufficiently
recognised from highly significant case features using prior experience and
lead directly to confirmation testing. This may then confirm the understanding
or alternatively initiate a backwards process. Figure 2.8 depicts confirmation
testing following PR not supporting the hypothesis and therefore initiating a
deductive process. Alternatively, a clinical assessment may commence with
a hypothesis based deductive search of the clinical data and then be followed
by the recognition of a component of the condition, which in turn triggers a
forwards reasoning confirmation process (Figure 2.9).

43
 Data collection

Significant case
features recognised

Single
hypothesis Working
Multiple hypotheses
hypothesis

Confirmation testing

Confirmation testing

Final hypothesis

Figure 2.8 Deductive strategies following incorrect pattern recognition

Data collection

Significant case
features recognised

Multiple Single
hypotheses hypothesis

Confirmation testing

Final hypothesis

Figure 2.9 Pattern recognition following deductive strategies

The possible integration of reasoning processes presented in Figures 2.8 and

2.9 highlight the complexity of clinical reasoning research. For example, the
identification of a PR process resulting in a single hypothesis midway through
a backwards search for understanding in a clinical case becomes
increasingly difficult to separate from HDR. This is especially so if the single

44
hypothesis was already one of the multiple hypotheses previously generated.
This means that some episodes of PR use could be difficult to detect within
deductive problem solving strategies if recognition occurred later during a
clinical assessment and after a deductive data collection process had
commenced.

2.6.3 Context specificity

A high probability exists that both HDR and PR reasoning strategies

contribute to the understanding and decisions made in a large number of
clinical encounters (Eva, 2004). Indeed, Norman and Eva (2003) state that
various strategies are likely to be used by experts and novices when solving
problems. One of the key features relating to the type of reasoning utilised is
the case at hand. ‘Case specificity’ has long been identified as having a
significant impact on the clinical reasoning process (Elstein et al, 1978). The
clinician’s experience and organised knowledge base referred to as ‘content
specificity’, also significantly influences the clinical reasoning process (Elstein
et al, 1978; Eva et al, 1998). Thus there are multiple variables likely to
influence the type of strategy used in clinical reasoning at any stage of the
process. The context of clinical reasoning includes the clinical environment,
the case at hand, and the knowledge and experience of the therapist with
respect to a case (Eva, 2004). All of these contextual factors will have an
impact on the direction of reasoning utilised by clinicians, both experts and
novices alike.

45
2.7 KNOWLEDGE IN CLINICAL REASONING

Clinical reasoning research has taken several directions in the past three
decades. Research involving the exploration of memory followed the failure
of single process models to fully explain how a clinician problem solves.
Based on memory recall research with chess experts / masters, research in
medical reasoning investigated memory recall of experts but was unable to
find a clear association between expertise and memory performance
(Ericsson, 2004). This led to a belief that the structuring of knowledge leads
to improved performance in medical reasoning (Norman, 2005). A full
discourse on the complexity of knowledge is beyond the scope of this
literature review but an overview has been provided to facilitate an
understanding of the clinical reasoning process, particularly in respect to
methodology within problem solving research.

2.7.1 Types of knowledge

Knowledge refers to specific information or understanding on a subject,

unique to an individual or group of people (Wilkes & Krebs, 1989). It can be
considered from an individual or personal perspective, where what is known
by a person comprises their knowledge. However this type of knowledge may
not be valid to the next person. Higgs and Titchen (2000, p.24) refer to
“public validation” of knowledge to highlight that overall knowledge about a
topic at any point in time may rightly or wrongly differ from an individual’s
knowledge on a topic. Public or common knowledge on a topic is usually
based on the evidence provided by specific research in the area in
combination with expert views. Personal and public knowledge can be
likened to non-propositional and propositional knowledge types.

Several types of knowledge should be considered when attempting to

understand physiotherapy clinical reasoning. Propositional knowledge refers

46
to theories and concepts or objective views based on sound research (e.g.
biopsychosocial knowledge) (Higgs & Titchen, 2000), whilst non-propositional
knowledge includes experiential, personal and practical types of knowledge
(Higgs, 1992). Experiential knowledge encompasses the learning from both
personal and practical experience, whilst practical knowledge in particular
refers to that developed from professional practice in a specific domain.
Practical knowledge has also been labelled ‘professional craft knowledge’
within the physiotherapy literature (Higgs, 2004).

Consideration of the different types of knowledge is important when relating

clinical reasoning process back to expertise. No single type of knowledge is
sufficient when dealing with the complexities of everyday clinical practice,
which instead relies on a problem solving process that likely integrates all of
the various knowledge types at appropriate times.

2.7.2 Structured knowledge

Structuring knowledge has been considered by way of the theory of

‘encapsulation’. This refers to knowing a clinical case via associated
knowledge concepts that provide a greater depth of understanding. The
integration of knowledge concepts attached to any clinical case are unique to
each clinician. A mind map or concept map is a tool utilised to depict the
development of an integrated knowledge base. Figure 2.10 provides an
example of a concept map portraying encapsulation. The case may be
known by a clinician at a number of levels, for example from management
and prognosis to the underlying pathophysiology of normal tissue healing.
Boshuizen and Schmidt (1992) provided support for the encapsulation of
concepts into clinical knowledge during a think-aloud problem solving study.
They found that family medical physicians with four years experience utilised
far greater pathophysiological knowledge underlying the case when

47
compared with medical students. Encapsulation is a means to understand
organised knowledge structures which in turn relates to expertise.

Pain Illness
Tissue healing
mechanisms experience
Inflammatory
process Patho- Psychosocial
physiology

Impairment

Contributing
Biomedical
factors

Structural
Management
Prognosis
options

Clinical case

Figure 2.10 Example of encapsulated concepts (utilising some hypothesis

categories reported by Jones & Rivett, 2004)

PR incorporates the use of experiential knowledge (Gale & Marsden, 1982;

Ridderikhoff, 1985; Rivett & Higgs, 1995) and the presence of an organised
knowledge base (Gale & Marsden, 1982; Ridderikhoff, 1985; Jones, 1992;
Rivett & Higgs, 1995). Some patterns may be simple and utilise only
professional craft or experiential knowledge, however an expert’s pattern will
have superior knowledge structure with integrated knowledge types. The
model of PR presumably relies on far more than a single type of knowledge.

48
2.7.3 The role of knowledge

Knowledge is undoubtedly important to clinical performance. Some evidence

does exist for enhanced clinical diagnostic skill based on greater use of basic
science knowledge in novices (Woods et al, 2005). However knowledge
related research in the field of medical reasoning has concluded that no
distinct type of knowledge is alone able to explain expertise (Bordage &
Lemieux, 1991; Norman, 2005). Additionally, despite the belief that an expert
is likely to have more knowledge than a novice, the overall amount of
knowledge is not a sole predictor of expertise (Norman, 2005).

The development of a structured knowledge base is without doubt a

requirement for attaining expert status. The research literature supports the
understanding that a structured and refined individual knowledge base is
required for expertise in healthcare (Boshuizen & Schmidt, 1992; Gale &
Marsden, 1982; Ridderikhoff, 1985; Rivett & Higgs, 1995). With respect to
clinical reasoning, it is also likely that a sound structured knowledge base is
essential for superior clinical performance. The type of diagnostic reasoning
pathway employed will be partly based on a therapist’s knowledge related to
that particular case.

Despite the findings that clinical reasoning process and knowledge alone are
not fully able to explain expertise, both remain important elements of problem
solving in clinical practice. The interaction of many factors, including problem
solving skill and knowledge, is likely to be central to the development of
expert status.

2.8 EXPERTISE

Expert clinical practice is a sought after goal for any motivated

physiotherapist with a desire to provide the best possible care. Expertise
refers to extensive skill or knowledge in a particular field (King & Bithell,

49
1998; Wilkes & Krebs, 1989). The study of expertise crosses many
professional fields, including and beyond healthcare. A view of expert
characteristics based on cognitive psychology research includes:

 Efficient ability to solve problems in a specific domain with few errors

 Greater short and long term memory capacity
 Enhanced visual representation of a problem’s structure using concepts
and principles rather than superficial features
 Increased perceptual ability to see meaningful patterns
 Better ability to build a mental representation of a problem to enhance
understanding and further problem solving
 Exceptional ability to monitor self-performance during problem solving
(Glaser & Chi, 1988).

Most of Glaser and Chi’s (1988) findings on experts relate in part to a greater
and more structured knowledge base. These findings can be utilised to assist
with understanding expertise in modern healthcare.

2.8.1 Common attributes of an expert physiotherapist

The physiotherapy profession has developed a broad template to understand

the qualities and strengths displayed by an expert. Jensen et al (2000)
qualitatively assessed and described the observed qualities of physiotherapy
experts across several domains. These authors developed a theoretical
model outlining an expert’s common attributes, including:

 Have a patient centred knowledge base that evolves through reflection

 Skilled collaborative problem solving (clinical reasoning) with the patient
 Focus on the assessment of movement in relation to patient function
 Display caring and committed virtues towards clients
(Jensen et al, 2000).

50
The two primary expert attributes of interest to this thesis relate to knowledge
and clinical reasoning skill. Both are essentially linked when considering a
reasoning pathway such as PR and the associated underlying knowledge
structures. Knowledge has been discussed with respect to expertise in
section 2.7, whilst the clinical reasoning skill of experts will now be further
considered.

2.8.2 Clinical reasoning skill

Clinical reasoning skill includes the effective integration of diagnostic and

interpretive reasoning types reported in section 2.1.3. Clinical reasoning is
without doubt a critical component of expertise (Higgs & Jones, 2000; Jensen
et al, 2000) but its varied nature adds complexity when relating these topics.
A clearer view of expert clinical reasoning skill can be found when looking at
diagnostic reasoning alone.

There is general agreement that no single problem solving pathway allows

for the attainment of expertise. Nor does knowledge represent the sole
attribute of an expert. Skilled clinical reasoning should not be considered as a
single type or model of reasoning, but rather an adaptive approach
dependent on case complexity and familiarity. The most evident aspect of
expert clinical reasoning rests with an approach that involves the client in
problem solving and adopts a reasoning approach specific to a situation. As
per Glaser and Chi’s (1988) views, expert reasoning in physiotherapy is likely
to include recognition of familiar cases via prior categorised patterns.

The direction of diagnostic reasoning (forwards versus backwards) is

believed to be associated with expertise. The general understanding in the
literature is that forwards reasoning is utilised more commonly by experts,
whereas novices rely heavily on backwards reasoning (Norman et al 2000).
Importantly, Patel et al (1990) noted that the basis of this view was in

51
psychology research, which essentially only had a theoretical basis due to
the lack of strength of the original empirical studies. However several authors
have since demonstrated that medical expert problem solving does vary from
that of novices (Doody & McAteer, 2002, King & Bithell, 1998; Coughlin &
Patel, 1987; Joseph & Patel, 1990). In recalling a normally structured
problem, experts were significantly better able to recognise and interpret
cases than novices (Coughlin & Patel, 1987). The experts recognised
patterns of familiar problems via significant case features or critical cues.

Support for PR as a forwards reasoning strategy in physiotherapy has also

drawn on differences between expert and novice clinicians. Investigations
into clinical reasoning differences between experts and novices have shown
an association between expertise and the use of PR (Doody & McAteer,
2002; King & Bithell, 1998). Retrospective think-aloud verbal protocols were
utilised by Doody and McAteer (2002) following the assessment and
treatment of a real patient to investigate clinical reasoning in practice and
determine differences between novices and experts. They concluded that
novices and experts both utilised HDR, but only the experts demonstrated
use of PR.

2.8.3 Significant case features

Clinical problem solving involves the generation of hypotheses following the

collection of relevant case information. A significant case feature refers to
any part of a clinical case that is important in interpretation or problem
solving. It may be data from the history, an observation of posture or other
visual information, or information obtained from a specific physical test or
combination of tests. Significant case features have also been labelled as
critical cues or forceful features (May & Dennis, 1991). Each clinician is likely
to have their own unique critical cues that assist them in their interpretation
and decision-making.

52
Just as knowledge and experience are unique to an individual, the features of
the case deemed important are also expected to vary amongst expert
clinicians. Grant and Marsden (1987) found that significantly different forceful
features were utilised to arrive at the same end point by experts whilst
solving paper based clinical cases. However despite the differences in type
of case features utilised, expert clinicians more effectively identify and use
significant features in a case as compared with novices.

Clinical reasoning can be partly viewed via the use of significant case
features to promote the recognition of specific clinical patterns (May &
Dennis, 1991). As previously highlighted, this characterises a forwards
process which likely enhances efficiency, and one that is primarily observed
in experts.

2.9 REASONING IMPACT ON CURRENT HEALTHCARE

The global economy has highlighted the importance of cost effectiveness in

health service delivery. The cost effectiveness of a healthcare service can be
influenced by the efficiency of which an accurate understanding of the case is
obtained. Less time spent developing a working understanding of the client
as a person with associated problems, should in turn lead to improved
outcomes based on client goals and arguably less overall cost. Accuracy and
efficiency are therefore key elements for cost effective healthcare.

Cost effectiveness may also be considered indirectly. Firstly, overall

healthcare costs may be reduced by early and appropriate intervention. A
more accurate initial assessment of a problem may decrease the overall cost
via early provision of necessary healthcare services and a potential reduction
in unnecessary medical costs such as imaging services. A second indirect
view of cost effectiveness relates to lost work time. Employee time away from
work is one factor that has a considerable financial effect on individual

53
business and the overall economy (National Occupational Health & Safety
Commission, 2004). More effective and efficient health service provision
theoretically may reduce lost work time and enhance economic gain. Overall
better healthcare should have a positive effect on an economy in many ways.

2.9.1 Errors in clinical reasoning

Clinical reasoning is a complex skill in the best of circumstances. Problem

solving may be misdirected from its path at numerous stages. Scott (2000)
outlines three main causes of error during problem solving: poor elicitation of
key case features from data; incomplete knowledge; and not applying
relevant knowledge to a specific problem. An error in clinical reasoning will
most likely result in misdirected diagnosis and management strategies, which
in turn influences outcome and cost-effectiveness of care. Table 2.4 outlines
five common types of clinical reasoning error in physiotherapy (Rivett &
Jones, 2004).

54
Table 2.4 Clinical reasoning errors in physiotherapy (Rivett & Jones, 2004,
p.409)

Component Example
Information collection Neglecting or misinterpreting relevant information
Premature decision-making
Not recognising data inconsistencies
Hypothesis formation Confirmation bias – overemphasis on supporting
features and neglecting negating features of a
hypothesis
Limited hypothesis category use
Not testing hypotheses
Identifying flags Missing data indicative of red (serious pathology) or
yellow (psychosocial barriers) flags
Diagnosis Presumption that a relationship between symptoms
confirms cause and effect and thus diagnosis
Treatment Use of recipe treatments and not clinically reasoned
management strategies
Lack of involvement of client in decision-making

Certain process models have been linked with greater possibility of errors in
problem solving. As discussed in section 2.5.3, PR is possibly more prone to
error than backwards reasoning models, especially with novice clinicians.
Jones and Rivett (2004, p.8) go as far as stating, “pattern recognition …
represents perhaps the greatest source of errors in our thinking”. However
these authors also highlight that critical reflection on clinical patterns may
reduce inaccuracy in reasoning. This may occur over time as more
experience is gained. Errors from PR should be reduced by continual
refinement and development of each pattern through reflective practice.

The time available to a clinician during an encounter with a client can impact
on the clinical reasoning process and thus affect hypothesis accuracy. When
healthcare is placed under financial pressure from a governing body, it has
greater potential for inaccurate outcomes (Rivett & Jones, 2004). Clinical

55
reasoning errors may then directly impact on the effectiveness of
management, which will indirectly impact on efficiency in achieving outcomes
or client goals.

Little conclusive research has been conducted into clinical reasoning error.
There is no single process that has been clearly associated with greater error
in physiotherapy clinical practice. The causes of error outlined above indicate
that incomplete or inappropriate use of knowledge is linked to inaccurate
reasoning. Knowledge is a core element of any model or process of clinical
reasoning and thus could be theoretically identified as a primary associate to
error in clinical practice.

Current support for enhanced cost effectiveness related to experienced or

expert clinicians comes from the physiotherapy professional association.
Australian physiotherapy service provision is moving towards a specialisation
framework that relates service cost to the clinician’s level of expertise. The
service provider descriptors developed by the Australian Physiotherapy
Association (2001) support the notion that an expert clinician achieves more
efficient benefits and better outcomes compared with those of less
experience.

2.9.2 Possible impact of accurate clinical patterns

Modern healthcare needs to be cost effective to be sustainable. The clinician

is placed under pressure from various financial and management sources.
This pressure will influence the time available to assess and treat a client and
undoubtedly impact on the type of assessment and reasoning strategies
employed. The recognition of clinical patterns using accurate underlying
knowledge structures may offer value in improving the efficiency of the
assessment process, but is limited in its application with inexperienced
physiotherapists. Experts seemingly utilise forwards reasoning with positive
results, but when does a clinician have enough experience to be considered

56
an expert? In other words, at what stage during a physiotherapist’s career
development does accuracy outweigh inaccuracy when utilising PR or other
efficient forwards reasoning strategies? The answer to this question is
undoubtedly complex and currently unknown.

2.9.3 Reasoning skill as a cost effective variable

The current healthcare climate will continue to drive physiotherapy services

to be more efficient in achieving better outcomes. The profession should
therefore closely consider the many variables associated with cost
effectiveness. Achieving accurate and efficient clinical outcomes relies on
effective clinical reasoning skills integrated with an extensive knowledge
base in the relevant domain. The use of the ever growing knowledge base
within evidence based practice relies on the operator to effectively reason
and integrate the available evidence with the case at hand. Clinical
reasoning skill is arguably the ground substance beneath the surface of cost-
effectiveness and should be further explored in the field of musculoskeletal
physiotherapy.

2.10 SUMMARY

Expert status is regarded highly within the physiotherapy profession. An

expert must have extensive knowledge and be skilled in clinical reasoning
(Jensen et al, 2000). Clinical reasoning is therefore a desirable skill for
physiotherapy clinicians and an important area to consider in educational
curricula design. This should include all reasoning types, including but not
limited to process models of diagnostic reasoning.

The area of interest in this thesis is the diagnostic reasoning models that
relate to the pathway a clinician takes in current physiotherapy practice. The
present review of the literature provides a background and basis to the most

57
common forwards and backwards diagnostic reasoning models and the
potential impact of these process models on healthcare.

The empirico-analytical strategies of HDR and PR have been observed in

musculoskeletal physiotherapy practice but require further investigation. The
term PR continues to be utilised in professional education, yet it lacks a
comprehensive research basis to its use. Research into the frequency,
efficiency and accuracy of PR in physiotherapy practice potentially offers
great benefit for professional education in problem solving.

This literature review highlights a general lack of knowledge surrounding

accuracy in physiotherapy clinical reasoning. Empirical support for the
various problem solving and decision-making strategies lies within the
medical field and is not evident to the same degree in other health
professions such as physiotherapy. Forwards reasoning in particular has
limited available evidence in both the medical and allied health literature. This
thesis therefore seeks to determine the use and potential benefit of PR in
physiotherapy. The study aims were to:
1. Determine whether PR is utilised by expert and novice clinical
physiotherapists in the musculoskeletal field
2. Relate the use of PR to efficiency within a physiotherapy assessment
3. Relate the use of PR to accuracy within a physiotherapy assessment.

The current study is based on a pragmatic world view combining both

deductive and inductive processes (Cresswell & Plano Clark, 2007).
Inductive analysis in qualitative research refers to the discovery of categories
as they surface from the data, whereas a deductive process can involve an
existing theoretical model as a category that is deductively tested during
analysis (Pope et al, 2000). The primary research question involves
deductive analysis to test an existing theoretical model of PR.

58
CHAPTER 3. METHODS

3.1 INTRODUCTION

This chapter describes a case study used in a mixed method research design
to gain insight into the clinical reasoning model of PR in physiotherapy. Section
2.2 provides support for the mixed method design based on the research aims
and prior case study research in the literature. Albeit a non-traditional research
approach, both qualitative and quantitative methods have been commonly
used in combination in similar research. The aims of the study outlined in
sections 1.2 and 2.10, required a mixed method approach to ensure successful
deductive testing of PR.

The study used a carefully chosen single critical case to provide research data
relating to the phenomenon of PR. The case was a real life clinical situation
portrayed by a trained actor. Details of the case study and its simulation are
provided in section 3.2. The single case study was repeatedly assessed by
research participants made up of expert and novice physiotherapy clinicians
with varied clinical experience and qualifications. The chosen study sample
and rationale for their inclusion is outlined in section 3.3. Qualitative
observation and interview data collection methods form the foundation of the
study (section 3.4) with subsequent analysis incorporating qualitative (section
3.5) and quantitative (sections 3.6 & 3.7) methods. The result of mixed method
analysis provided a comparative view of the clinical reasoning process
employed by participants and inherent relationships. The overall study design
and chapter outline are depicted in Figure 3.1.

59
 Preparation (section 3.2)
Ethical
approval Case Actor Piloting
development training

Sampling (section 3.3)

Sample size
Recruitment
Inclusion /
exclusion criteria

Data collection (section 3.4)

Location Transcription
Equipment Interview
Participation
process

Analysis
Qualitative transcript Quantitative video
analysis analysis
(section 3.5) Quantitative transcript (section 3.7)
analysis
(section 3.6)

Figure 3.1 Study method and chapter outline

60
3.2 PREPARATION

3.2.1 Ethical approval

Prior to commencing recruitment or data collection the study was granted

approval from the Human Research Ethics Committee of The University of
Newcastle. The approval letter has been attached as Appendix 1 with approval
number H-149-1105.

3.2.2 Case development

A key element of any research study involving a simulated client is the actual
case itself. Developing a case study is complex and demands careful
consideration (Creswell, 2007). In this study, a suitable case was sought to
facilitate exploration of the primary research question. Firstly, the identification
of PR requires a clinical presentation that is known or familiar to the therapist.
Secondly, a real life case that was taken from a single physiotherapy clinical
experience was deemed appropriate as it allows for a documented outcome
including clear diagnosis of the primary pathology and problem identification.
This is necessary to ascertain the accuracy associated with PR.

The case utilised for the study involved a 20 year old female presenting to a
physiotherapist for the first time with lower back pain (LBP). It is known that
LBP is frequent in the community and a regular presenting problem to
musculoskeletal physiotherapists. A high percentage (85%) of LBP cases are
labelled non-specific as they cannot be sufficiently diagnosed from a patho-
anatomical perspective with available radiological and other investigations
(O’Sullivan, 2005; Waddell & van Tulder, 2004). The remaining 15% of specific
LBP cases are generally still common in presentation to physiotherapy
clinicians working in the musculoskeletal field. One such specific condition is
spondylolisthesis of the lower lumbar spine with established diagnostic criteria
via X-ray imaging. Spondylolisthesis refers to the forwards movement of one
vertebral body relative to the vertebra below. It is a type of mechanical

61
instability of the spine and most commonly occurs due to a bilateral pars
interarticularis defect or spondylolysis (Floman, 2000; Herman & Pizzutillo,
2005; Rossi & Dragoni, 2001).

The incidence of spondylolisthesis is reported as between two and six percent

of the general population (Herman & Pizzutillo, 2005; McNeely et al, 2003;
Stanitski, 2006; Treble et al, 2005). The distribution of spondylolisthesis
between males and females is two to one, however the frequency of this
condition presenting as symptomatic LBP is greater in females reportedly due
to activity predisposition (Earl, 2002). Supporting this observation, a study of
twenty-one consecutive clients presenting with high grade spondylolistheses
over a fifteen year period found that more than 60% were females (DeWald et
al, 2005). The Meyerding classification is a widely utilised grading system for
the extent of vertebral translation and labels a spondylolisthesis from grade 1
to grade 5. ‘High grade’ refers to the grades 3 or above which relates to an
anterolisthesis or forwards movement of greater than 50% of the vertebral
body below (DeWald et al, 2005; Haun & Kettner, 2005; Lim et al, 2004; Rossi
& Dragoni, 2001).

Higher grade spondylolisthetic conditions are commonly symptomatic. The

symptoms associated with higher grade spondylolistheses include bilateral
lower lumbar pain spreading to the gluteal regions and posterior thighs
(Hensinger & Michegan, 1989; Herman & Pizzutillo, 2005; McNeely et al, 2003;
Stanitski, 2006; Treble et al, 2005). Mechanical activity tends to provoke the
symptoms and rest or reduced activity levels will ease the reported symptoms
(Hensinger & Michegan, 1989; Lim et al, 2004). Extension motion of the
lumbar spine is consistently reported as provocative of pain (Lim et al, 2004;
McNeely et al, 2003; Stanitski, 2006; Treble et al, 2005).

The level most commonly observed with an anterolisthesis deformity is the 5th
lumbar segment (Beutler et al, 2003; Earl, 2002; Haun & Kettner, 2005; Lim et
al, 2004; Rossi & Dragoni, 2001; Stanitski, 2006; Treble et al, 2005). Although
the history of spondylolisthesis can involve insidious symptomatic onset,

62
Hensinger and Michegan (1989, p.1098) state that a “history of minor trauma is
common … and an episode of trauma often initiates the onset of symptoms”.

Given that clients with spondylolisthesis are more likely to have LBP than the
general population and that LBP is one of the most common problems
presenting to a musculoskeletal healthcare professional, the prevalence of
such conditions will likely be greater in a healthcare setting than in the general
population (Treble et al, 2005). Thus an experienced clinician should be
familiar with this condition enabling recognition and use of forwards reasoning
strategies. The details of the case utilised in this study are found in Appendix
2. The case was found to be consistent with the reported presentations in the
literature.

The real life case chosen for the study was considered from the perspective of
the hypothesis categories (Jones & Rivett, 2004, p.13-20) as summarised in
Table 2.2. This provided an actual case outcome for comparison with
hypotheses formulated by participants. The primary management of the case
included X-ray imaging undertaken after the first physiotherapy assessment
and demonstrated a grade 3 spondylolisthesis of the 5th lumbar on the first
sacral vertebra (Figure 3.2). An evidence based specific stabilisation exercise
approach to management was commenced (O'Sullivan, Twomey & Allison,
1997). Short term follow up demonstrated a reduction in symptom levels via
two main treatment interventions; activity modification and specific exercise.
The pain mechanism involved in reported symptoms was primarily mechanical
without any clear neuropathic or ongoing inflammatory components. Leg
symptoms were attributed to somatic referral from the lower lumbar spine (i.e.,
non-radicular pain). Precautions and contraindications identified for the case
included extension based manual therapy techniques or exercises, and spinal
manipulation of the unstable segment. No clear psycho-social features were
deemed relevant to the outcome or recovery of the client. Medical
management involved an orthopaedic specialist monitoring the degree of
anterolisthesis movement over a 12 month period. Evidence of continued
vertebral translation resulted in surgical fusion of the unstable segment.

63
Activity / participation restrictions continued beyond the post-operative
rehabilitation period however a return to full functional capacity was achieved
with respect to completion of studies, subsequent employment and sports
participation. With conservative management alone the prognosis towards
achieving full function in this case was predictably poor, however with surgical
intervention the prognosis can be stated in hindsight as very good.

Figure 3.2 X-ray image of the grade 3 spondylolisthesis case

3.2.3 Training of case actor

Preparing the actor for the role as a simulated patient involved a three part
process. Firstly, the case was converted to lay format (Appendix 3) and
presented to the actor who simultaneously took their own notes during
discussion to complement their understanding. Each aspect of the case was
individually discussed without offering the actor more than a lay understanding

64
of the data. Following a period of independent review of the case information, a
second training session was undertaken involving the researcher questioning
the actor on the case details and providing feedback regarding responses.
Questioning was undertaken as per a routine physiotherapy assessment but
was completed in sections allowing for feedback and further note taking by the
actor. A third phase of training involved a second researcher conducting an
uninterrupted videotaped subjective assessment (history) with the actor as the
simulated client. The second researcher provided feedback regarding the
realism and accuracy of the case and the videotape was then viewed by both
the actor and primary researcher. The final phase of training was incorporated
into a pilot trial as described in section 3.2.4. Once more the actor was
provided with feedback regarding accuracy and performance.

Consistent with previous research involving the training of simulated patients,

the total time spent training the actor was six hours without including the
actor’s independent study time. The training was predominantly associated
with the portrayal of subjective examination case information given that a
physical examination was not required. Training related to physical aspects of
the case was limited to basic observations of client entry to the assessment
location and sitting postures during the assessment. This was also monitored
during training and participation via the videorecorded assessments. To ensure
ongoing accuracy of the actor’s portrayal of the case, the simulated client
assessments were reviewed by the researcher throughout the data collection
phase of the study. Evaluation of the accuracy of case simulation was
undertaken during subsequent review of videorecorded assessments via a
simulated client response checklist (Appendix 4) that was transferred to
nominal data for each participant. Areas of inaccuracy could therefore be
identified immediately and feedback provided to the actor. Overall statistical
analysis of simulated client accuracy across all participant assessments could
then be calculated as discussed in section 3.7.

65
3.2.4 Pilot trial of method

A single participant with more than 10 years of clinical experience and a

postgraduate qualification in musculoskeletal physiotherapy from the United
Kingdom volunteered to participate in a pilot trial of the study design / method.
This participant was provided with an information statement and signed a study
consent form. Data collection was then undertaken including subsequent
transcription of the interview and coding process. Following this trial no
changes in study procedure were deemed necessary.

3.3 SAMPLING

Experienced musculoskeletal physiotherapists were chosen as the group of

interest due to their proposed use of PR strategies during problem solving and
probable previous exposure to spondylolisthesis as a condition. To help ensure
the findings could be related to experience / expertise a second group of
registered but novice physiotherapists were included in the study. Systematic
non-probabilistic sampling (Mays & Pope, 1995) was therefore used with two
specific groups that were predicted to have differing cognitive skill relevant to
the clinical reasoning process. The inclusion criteria for the expert and novice
sample populations have been detailed in sections 3.3.2 and 3.3.3
respectively, and the supporting physiotherapy and medical clinical reasoning
literature cited.

3.3.1 Sample size

On commencement of the study, it was anticipated that ten expert and ten
novice physiotherapists were required for data collection. These numbers were
based on previous comparable published research projects in physiotherapy
(Doody & McAteer, 2002; Rivett & Higgs, 1997) and predicted saturation rates.
Qualitative research refers to saturation when further data collection does not
reveal any new themes than previously identified.

66
3.3.2 Expert inclusion criteria

The physiotherapy and allied health literature has generally failed to agree on
consistent criteria for expertise. Problem solving research involving medical
experts has largely been based on specialisation groups as defined by the
various registration bodies (Grant & Marsden, 1987; Joseph & Patel, 1990;
Patel et al, 1990). At the time of designing the methodology of the study, the
process of attaining the title of ‘specialist musculoskeletal physiotherapist’ was
a work in progress by the Australian Physiotherapy Association (APA)
(Australian Physiotherapy Association, 2008). During the period of recruitment
for the study, the number of musculoskeletal physiotherapists in Australia, who
had attained the title of ‘specialist’ remained insufficient for the study sample.

Physiotherapy experts within the clinical reasoning literature have been

defined in different ways, from academic qualifications and positions (King &
Bithell, 1998) to peer nomination (Jensen et al, 2000). The criteria for this
study have been adapted from prior physiotherapy clinical reasoning research
(Doody & McAteer, 2002; Edwards et al, 2004; King & Bithell, 1998; Rivett &
Higgs, 1997) involving participants with a measurable level of experience and
knowledge to meet the requirements of an expert.

The amount of time in clinical practice alone is insufficient to define an expert.

It must be combined with domain specific knowledge and critical reflection.
Knowledge has traditionally been described as an essential requirement for
expertise (Simon, 1980), however medical research conducted regarding the
organisation and accessibility of knowledge within problem solving highlights
that knowledge alone is also insufficient for expertise (Norman, 2005). Thus
expertise in this study was based on both measured duration of clinical
experience and also extent of knowledge in the musculoskeletal domain.

67
When considering experience, a minimum of ten years exposure in any given
domain has been reported as being necessary to achieve expertise (Simon,
1980). This is supported in the physiotherapy clinical reasoning literature by
King and Bithell (1998) who utilised ten years experience as a minimum
timeframe in their research to predict the presence of expert clinical patterns. It
was deemed necessary in the present study that expert physiotherapists
remained currently practising due to the association between relevant and
accessible clinical experience and the use of forwards reasoning strategies
such as PR.

Identifying clinicians deemed more knowledgeable and worthy of recognition

as an expert was based on a recognised postgraduate degree in manipulative /
musculoskeletal physiotherapy. Such qualifications require comparable
standards to be met of both academic and professional bodies and involve
written and practical examinations to assess levels of knowledge and clinical
skills. The physiotherapy profession in Australia identifies titled
‘Musculoskeletal Physiotherapists’ as those who have completed recognised
postgraduate study and have extensive musculoskeletal physiotherapy clinical
experience (Australian Physiotherapy Association, 2005). Titled
Musculoskeletal Physiotherapy Australia (MPA, a sub-group of the APA)
members can therefore be considered to have a greater and more organised
propositional knowledge base. When integrated with experiential knowledge
resulting from continued clinical practice, it was considered that this should
result in critically refined clinical patterns.

The current study identified an expert musculoskeletal physiotherapy clinician

by the following inclusion criteria:

1. Titled membership of MPA based on completion of a recognised

postgraduate qualification in manual / manipulative therapy;
2. Greater than two years experience following postgraduate qualification
enabling titled membership;

68
3. More than ten years overall clinical experience in musculoskeletal
physiotherapy;
4. Currently involved in clinical practice.

3.3.3 Novice inclusion criteria

Given that PR has been strongly linked with clinical experience and domain
specific knowledge it is unlikely that a novice will display this strategy. This
view has been supported by previous research however the novice groups
investigated have often been undergraduate students without complete
education in their particular professions (Arocha et al, 1993; Doody & McAteer,
2002; Gale & Marsden, 1982; Grant & Marsden, 1987; Grant & Marsden, 1988;
Groves et al, 2003). Alternate research designs have used ‘sub-experts’ for the
comparison group, who are experts in their own domain but not with that of the
case type presented (Joseph & Patel 1990; Patel et al 1990) or generalists
who have considerable overall clinical experience but no specific postgraduate
qualifications (King & Bithell, 1998). The ‘sub-expert’ comparison group is
fraught with potential difficulties of the intermediate effect in which participants
with less domain specific knowledge or experience may outperform experts in
their domain (Patel & Arocha, 2000).

A more robust approach towards answering the research question and

exploring whether PR is a strategy of more experienced clinicians requires a
comparable group with minimal experience. Undergraduate physiotherapy
students were considered underdeveloped in their ability to conduct an
assessment and problem solve during the reasoning process. However
recently graduated physiotherapists in their first year of clinical practice,
generally having undertaken a musculoskeletal rotation (usually a minimum of
three months), were considered a more representative group but without
sufficient clinical experience to have developed clinical patterns to any great
degree.

69
The novice physiotherapy participants included in the study were classified by
the following inclusion criteria:
1. Completed a recognised physiotherapy qualification and be registered with
the NSW Physiotherapists Registration Board;
2. Less than one year of clinical experience as a physiotherapist;
3. Currently involved in clinical practice.
Any formal postgraduate study in physiotherapy excluded participation in the
novice group.

3.3.4 Recruitment process

Registered physiotherapists from the Hunter region of New South Wales,

Australia formed the source of potential participants. Expert physiotherapists
were identified via the Directory of APA Musculoskeletal Physiotherapists
(Australian Physiotherapy Association, 2005). Every MPA titled physiotherapist
located within a 90 minute radius of the research venue was sent an invitation
letter to participate (Appendix 5). If no response was received following initial
invitation, the follow-up procedure involved a single reminder letter (Appendix
5) after a two-week period of time followed by a single phone call to ensure the
letters of invitation had been received. New graduate physiotherapists (novice
physiotherapists) were recruited via advertisement (Appendix 6) in the
Australian Physiotherapy Association Hunter Regional Group newsletter.

A written information sheet and consent form (Appendix 7) was provided for
further consideration once potential subjects contacted the researcher.
Following an opportunity to ask additional questions and sign the consent form,
participants were included in the study and scheduled a time that suited their
work commitments. The first ten physiotherapists meeting the inclusion criteria
and consenting to participate formed each group.

70
3.4 DATA COLLECTION

Several aspects of data collection required careful consideration to minimise

influencing participant behaviour and to help ensure a realistic client
assessment. Similarly, accurate recollection of thought processes from the
actual client assessment relied on a skilled interview. The location for
participation and the equipment used in the study are two key elements that
will be discussed in this section, along with the participation process including
client assessment and stimulated recall interview. Lastly details regarding
interview transcribing are provided.

3.4.1 Location

The project’s data collection phase took place within the School of Health
Sciences Research Laboratory at The University of Newcastle. The
assessment was conducted at a small table with both the participant and
simulated patient seated on facing chairs. A room adjacent to the Research
Laboratory allowed for the setup of the notebook computer recording the video
to be out of participant view during the client assessment. The retrospective
interview was undertaken at the same table in the Research Laboratory with
the notebook computer located on the table. The videorecorded client
assessment was replayed on the notebook computer as the stimulus for
retrospective recall. The adjacent room was not required during the interview
process.

The location of the study became a barrier to recruitment of expert participants

during data collection. This resulted in an ethics variation regarding this aspect
of the method being submitted to the Human Research Ethics Committee of
The University of Newcastle (Appendix 8), providing approval for data
collection to be conducted at the workplace of some expert participants. This
required the simulated patient actor and the researcher to travel to the expert’s
practice location to complete data collection. A similar setup was utilised at the

71
alternate location except the notebook computer was located in the same
room.

3.4.2 Equipment

Conducting the data collection sessions required specific equipment to allow

for the audio / videorecorded client assessment. A standard one square metre
flat desk and two chairs were utilised for both the simulated client assessment
and interview. The participant and actor portraying the simulated client were
seated for the majority of the assessment but were free to move as required.
Similarly, the researcher and participant were seated for the entirety of the
subsequent interview. A microcassette audio tape recorder was located on the
desk during both client assessment and interview. Additionally the camcorder
utilising mini digital video cassettes was situated on a tripod two metres from
the desk with an external microphone placed one metre from the desk (Figure
3.3).

Figure 3.3 Research location and recording equipment

The notebook computer requisites included video recording software and a

serial data transfer protocol for high bandwidth applications (IEEE 1394 port)
for direct downloading from the camcorder. This equipment was located in an
adjacent room whilst recording the client assessment (Figure 3.4). A fire wire
cable connected the camcorder to the laptop for simultaneous downloading

72
during the simulated client assessment and retrospective downloading of the
interview data. The notebook computer was relocated to the interview table in
the Research Laboratory for the participant interview.

Figure 3.4 Notebook computer recording setup in adjacent room

The transcription data included the dialogue between participant and

researcher during observation of the simulated client assessment. To avoid
transcribing the replay of the client assessment, two headsets were utilised to
remove this auditory source during recording of the interview.

The client assessment and participant interview were recorded from two
sources to ensure a backup source of data. The separate sources included the
microcassette audio tape recorder and mini digital video cassette camcorder. A
full list of setup procedures and equipment utilised for data collection is
documented in Appendix 9. The study investigated only the subjective
examination (history) information of a physiotherapy assessment, which
reduced the complexity and extensiveness of equipment required, recording
methods and overall space required.

73
3.4.3 Participation process

On completion and return of the consent form a session time was scheduled
for the study. The study equipment was setup in advance (Appendix 9). On
arrival the participant was provided with an ‘orientation to the patient
assessment’ information sheet (Appendix 10). This did not state the client to be
simulated however the participant was made aware via the research project
information statement (Appendix 7) that the client was an actor simulating a
real clinical case. After reading the orientation sheet the audio and video
equipment commenced recording / downloading and the simulated patient was
introduced to the participant. The researcher was not present in the room for
the entirety of the assessment with the participant instructed to let the
simulated client know when the assessment was complete.

During the preparation for the stimulated interview, the participant was
provided with an ‘orientation to the interview’ information sheet (Appendix 10).
The interview was then commenced and conducted in the same location.

3.4.4 Semi-structured interview technique

The purpose of the retrospective interview was to obtain information relating to

the clinician’s problem solving strategies, in particularly the use of case data
and subsequent generation of hypotheses. A semi-structured interview design
(Appendix 11) utilising a combination of open-ended questions and more
directed questions was developed by the researchers. This type of interview
design is appropriate when the topic is sufficiently understood but the
responses to questioning are not (Richards & Morse, 2007). The interview was
divided into two parts.

The first group of questions were asked throughout the video-replayed portion
of the interview. These were open-ended enquiring questions allowing the
participant to discuss their problem solving from the client assessment without

74
creating bias or reflection in the participant’s responses. The timing of the
video stops throughout the interview was either following each group of similar
clinical questions or if the participant sought to comment on their thoughts from
that part of the assessment. A few examples of groups of clinical questions
commonly leading to video stops include location of symptoms, description or
severity of symptoms, current or past history, or investigation questions. Due to
the varied nature of individual physiotherapy assessments the questioning and
subsequently the time stops were never the same between participants.

The second more specific group of interview questions followed the completion
of the video replay. At this point the participant had described their choice of
pathway through the data, provided reasons for their chosen methods and
described their understanding of the case along the way. Once the video
replay had finished all the data was available for use in discussing the final
hypothesis. This second group of questions were designed to ensure the
necessary information had been gathered from the interview to address the
research questions. These related to the first and final hypotheses developed,
the physical examination plan, and the influence of the study method on
participants. If the information had already been obtained for any question it
was not repeated.

3.4.5 Transcription

The participant retrospective interviews were transcribed from the audio

recorded source of data. The transcript was reviewed for completeness in
comparison to the secondary source of interview data. This backup data was
obtained from the video recordings. Any erroneous or incomplete words or
section of a transcript was completed from the video data to ensure accurate
and complete coding and analysis.

75
3.5 QUALITATIVE TRANSCRIPT ANALYSIS

The predominant data source in the study was the verbal interview transcripts.
Qualitative analysis of these transcripts was similar to that described by
Creswell and Plano Clark (2007). Preparation of data for analysis first involved
the professional transcription of interview data from audio tape recordings.
Corrections were then made where required to complete the transcripts
(section 3.4.5). The participant interview transcripts were then imported into
NVivo 7 (QSR International, n.d.), a qualitative software program (section
3.5.2).

A broad review of transcripts enabled the completion of a codebook and note

taking regarding overall content and researcher thoughts. Finally, detailed
analysis of the data involving coding and categorisation allowed for an overall
qualitative understanding to be formed and comparison between groups.

3.5.1 Code development

Coding is a process that allows for data to be organised and categorised.

Grouping common ideas that develop during data analysis allows for easier
association and comparison. The development of codes prior to commencing
analysis facilitates the identification of the categories required to test existing
theory, however it is recognised that this can limit the depth of understanding
outside the walls of the phenomena being studied (Creswell, 2007). A thorough
review of the literature pertaining to PR has been detailed in Chapter 2. The
common elements of PR outlined in Table 2.2 and the resultant definition
enabled the development of an initial coding schema. The early codebook was
subsequently refined following initial data collection to ensure completeness of
the pre-determined codes.

To ensure agreement amongst researchers regarding the identification of pre-

determined codes, a process was undertaken to review these codes and their
definitions. This was not dissimilar to the process described by Creswell

76
(2007). The first two participant transcripts of each group were independently
coded by all three researchers. Following complete coding of each transcript,
the researchers met to discuss and reach consensus on coding data. This
process was repeated for all four transcripts following which the researchers
had a consistent understanding regarding the allocation of codes to the textual
data. At this coding agreement stage only the ‘hypothesis’ sub-codes were
reviewed.

The three researchers involved in the coding agreement process included the
student researcher who was currently practising in musculoskeletal
physiotherapy and had completed post-graduate study in clinical reasoning.
The other two researchers have a track record in clinical reasoning research
and related publications. All three researchers each had more than ten years
of physiotherapy clinical experience.

To minimise the potential limiting effect of a pre-determined codebook,

additional codes were added throughout the analysis (Creswell, 2007). These
new codes were placed separate to the prior developed codes but easily
compared and associated at subsequent stages by the qualitative software
program NVivo 7 (QSR International, n.d.).

3.5.2 Qualitative data analysis software

The computer software program NVivo 7 was used throughout data analysis.
Qualitative data analysis software has recently become commonplace
amongst academic research of a qualitative nature (Davidson & Jacobs,
2008). NVivo software is produced by QSR International Pty Ltd, originating
with NVivo 1 in 2001 and releasing NVivo 7 in February 2006 (QSR
International, n.d.).

The primary benefits of qualitative data analysis software in the study included
the organisation of files, notes, memos, codes and their descriptions. The
program allowed for efficient data retrieval and enhanced ability to compare or

77
relate transcript sections or themes. The transcripts from novice and expert
participants were placed in individual NVivo 7 files to allow for separate
thematic analysis and easier comparison.

The pre-determined coding schema was entered into the NVivo 7 software as
tree nodes, which were able to be organised in a hierarchical fashion with sub-
codes. A node is referred to as the location for a compilation of references
identified by the same code. The new codes identified during transcript
analysis were labelled as free nodes and documented with an associated
description. These free nodes contain codes that don’t necessarily relate to
others in a clear structure. The stand alone information in the free nodes was
useful for identifying potential emerging themes throughout the analysis.

3.5.3 Data analysis process

The qualitative analysis process involved a series of steps to allow for

inductive development of new ideas and deductive testing of the primary
research question. The steps have been outlined in Table 3.1 with further
detail provided in this section.

Table 3.1 Qualitative analysis process

Steps:
1. Initial reading of transcripts whilst making memo notes
2. Coding from pre-existing codebook (tree nodes) and identifying new
codes (free nodes)
3. Checking for coding accuracy and reliability
4. Creating categories or themes from codes and memo notes
5. Comparing themes across groups

78
Overview reading

An inductive review of each transcript was undertaken by the student

researcher to gain an overall feel for its content and noting potential new codes
and concepts. Subsequent coding from the existing codebook was then able to
be completed without limiting the analysis to existing theory. During this stage
a memo was created for each participant and notes from the overview reading
were taken. Any new code identified during the overview reading was entered
into NVivo 7 as a free node.

Coding

Each transcript was analysed in detail to identify data fitting the pre-determined
(tree) codes. The tree codes have been detailed in Table 3.2. Researcher
notes were made during this coding process and any specific notes relating to
each participant or code were recorded as memos. Any new codes revealed
from raw transcript data during this phase were labelled as free nodes.

Sub-coding the hypothesis codes based on previously reported hypothesis

categories was undertaken during the initial coding process as this sub-level
had been previously developed. Each section of text coded with a hypothesis
was simultaneously sub-coded into one of nine hypothesis categories. The first
eight of these had been sourced from Jones and Rivett (2004) with one
additional sub-code (non-specific) being added following the review of the
codebook. The hypothesis code and sub-codes, including descriptions and
examples, are provided in the final codebook (Appendix 12). A similar process
was undertaken for data collection and planning codes. Examples for each
pre-determined code and sub-code have been included in Appendix 12.

Sub-coding for knowledge and self-awareness codes was undertaken as

separate processes during the analysis stage following initial coding. All coded
data in each of these nodes were reviewed and divided into sub-groups with
meaning based on data collected and knowledge relating to the specific code.

79
The sub-coding of knowledge and self awareness and associated descriptions
have been outlined in section 4.3.1.

Table 3.2 Tree codes: predetermined code schema

Code Description
Data collection Therapist reports collecting information in a routine
manner without the data relating to a hypothesis.
Hypothesis related Therapist states an understanding of the case in
any of the hypothesis categories. This may be
several competing possibilities or a single clear
hypothesis. Any lay level of response, where no
interpretation has occurred, was not coded as a
hypothesis.
Knowledge Participant refers to their knowledge in an attempt to
apply it to the problem at hand.
Planning Therapist verbalises their use of data during the
history to plan physical examination or treatment
procedures.
Self awareness Therapist verbalises awareness of their own
thinking. These reflections may relate to data,
process, hypotheses, decisions or knowledge.
Significant case feature Therapist highlights the relevance / significance of
data obtained to case at hand.

Any new code identified during coding was reviewed for content and assigned
a description. These were assigned a free node with associated description. If
a free node was identified mid-way through transcript coding, each prior coded
transcript was re-read specifically for the identification of the newly developed
free node. This was a cyclical process until no further free nodes were
identified. Free nodes developed during analysis have been listed in section
4.3.3 of this thesis.

80
Coding accuracy and reliability

Of concern during coding was the possibility that inconsistent or inaccurate

grouping of text segments may occur and impact on analysis. To ensure
minimal effect on the overall results, coding accuracy and reliability were
further considered after the development and review of the codebook by all
researchers.

Accuracy during the coding process was ensured by node content review.
Each tree or free node and associated sub-nodes were opened and contents
reviewed relative to the description. Editing at this stage involved un-coding or
re-coding any data found to be inappropriately labelled. The description of
each node was also reconsidered with respect to its contents.

Following the coding review process involving all three researchers, the
remainder of the study involved the student researcher coding alone. Given the
sole coding nature of the study an intra-rater reliability evaluation was
undertaken. This involved the first two transcripts of each group being repeat
coded at a later stage following complete coding of all transcripts. This
occurred after more than a three month period after the initial coding of each
transcript to reduce the likelihood of memory recall during repeat coding. Intra-
rater reliability was undertaken only for tree nodes and analysed via
percentage agreement and kappa reliability coefficient (Domholdt, 2005).

3.5.4 Theme analysis

The reduction of data into themes occurred throughout the coding process.
Themes are patterns repeated throughout the transcripts. Notes and memos
relating to any aspect of a developing theme were documented by the
researcher during coding. These notes were then compared alongside the
codebook and coded data to further develop and encapsulate themes. The
final element of analysis involved displaying the data and themes allowing for

81
visual representation of the research findings. This included the comparison
between the novice and expert groups.

3.5.5 Interpretation

Qualitative analysis gradually develops over a prolonged period of continuous

analysis (Barbour, 2008). It requires a systematic and thorough approach to
cover the available data without losing track of the study goals. Throughout the
qualitative interpretation within this study it was imperative the student
researcher remained open to new possibilities to ensure comprehensive
qualitative results. The qualitative process outlined has been undertaken to
gather the necessary data and subsequently allow quantitative analysis to
follow. Generating an overall qualitative understanding of the primary research
question via themes and categories and also enabling a subsequent
quantitative interpretation, arguably provides greater strength to the study
findings. In addition, the qualitative process ensures that any new themes will
be identified in relation to the clinical reasoning process of physiotherapists.

3.6 QUANTITATIVE TRANSCRIPT ANALYSIS

Quantitative analysis of verbal transcripts involved reviewing the tree nodes for
coded portions of the transcripts that related to the identification of PR and
type of hypothesis category used. Completion of this aspect of analysis was
undertaken following the qualitative coding and thematic analysis and thus did
not influence the qualitative conclusions.

3.6.1 Identification of pattern recognition

Identifying PR within a clinical encounter is partly dependent on the complexity

of the case. In more complex clinical encounters differentiating PR from a
deductive reasoning process is likely to be difficult. For example, simply having
a participant refer to prior experience alone does not provide conclusive
evidence of PR use.

82
The formation of a synthesised understanding of PR from the prior literature
has been outlined in section 2.5.1. Based on this interpretation the features of
PR were refined and detailed to provide a consistent structure when reviewing
transcript data (Table 3.3). An associated identification tool for each interview
time stop and an overall scoring tool (Appendix 13) for each participant were
developed and utilised in conjunction with the features of PR. The application
of the identification and scoring tools to the coded transcript data provided a
structured method of identifying PR from each participant’s clinical encounter.

Table 3.3 Identification features of pattern recognition

Component Description
Central hypothesis A central hypothesis was developed at a distinct
point in time and maintained as a predominant
understanding throughout the assessment.
Significant case features Case features that were considered significant or
important to the central hypothesis are described.
Professional knowledge Professional knowledge relevant to the central
hypothesis was stated at any point in time during the
retrospective interview.
Clinical experience Prior clinical experience was referred to by the
participant in reference to the current case and the
central hypothesis.
Management plan A plan for management was evident and relevant to
the stated central hypothesis formed at the distinct
point in time. The plan doesn’t need to be stated at
the same distinct point in time but rather relates to
the hypothesis of this time.

Identifying PR relies on more than just using the word ‘pattern’ or stating that
this case had been recognised. In the current study, the identification of PR
required a central hypothesis to be stated at a distinct time related to the client

83
assessment. As discussed in section 2.5.1, separating PR from HDR requires
the hypothesis to be immediately formed and not developed gradually.

Further support for PR can subsequently be provided by identifying significant

case features described during the interview that were relevant to the central
hypothesis. Identifying professional knowledge and direct clinical experience
related to the central hypothesis were also included based on the commonly
agreed elements of PR in the literature. These provide additional support for
the presence of PR.

The final feature included in identifying PR was management. This component

provided insight into the presence of a central hypothesis and the diagnostic
reasoning process utilised.

Efficiency was not included as a feature in identifying PR but is rather a

consequence of its use. The commonly reported feature of immediacy when
forming a hypothesis through PR was separated from efficiency due to
inconsistencies in the literature (see section 2.5.3). However the timing data
obtained for the formation of a central hypothesis generated from an existing
pattern could provide some insight into efficiency as an outcome of PR.

The ‘time stop identification tool’ was applied to each participant transcript and
involved reviewing all coded text from each point in time where the client
assessment was paused to obtain the participant’s thought processes. This
involved transferring hypothesis coded text transcripts into the ‘Hypothesis
formed’ column and subsequently indicating the hypothesis category identified.
The primary benefit of this process was to observe the presenting case
hypothesis in sequential order throughout the assessment. Additional tree
codes were then reviewed for significant case features, knowledge, experience
and management. These were directly compared with the central hypothesis if
present. Relevant transcript text supporting the positive identification of any of
these codes was documented in the ‘comments / quotes’ column.

84
The data was then summarised from the ‘time stop identification tool’ into the
‘overall pattern recognition scoring tool’. This provided a score from 0 to 5 for
each participant relating to PR use. A central hypothesis was necessary
throughout the assessment for overall identification of PR. Higher overall
scores provided greater support for utilisation of PR.

3.6.2 Expert and novice group differences in pattern recognition use

The comparison between expert and novice use of PR followed its

investigation across all participants using the scoring tool (Appendix 13).
Comparison between groups was via simple number counts following transfer
of data to categorical form where 0 = not identified and 1 = identified.

3.6.3 Accuracy of pattern recognition

The stated final hypothesis for each participant was listed in summarised form
for comparison with identified PR. Subsequent participant numbers per group
utilising the PR strategy and the respective case accuracy relative to the
known case diagnosis was presented in a 2x2 format for visual comparison.

3.6.4 Hypothesis category utilisation

The literature indicates that hypothesis generation clearly occurs throughout

physiotherapy problem solving. Section 2.3.1 highlights one classification of
the various types of hypotheses that can be formed as reported by Jones and
Rivett (2004). The use of various hypothesis categories during a clinical
encounter is predictably unique to each therapist, however it is arguable that
an expert would have a greater depth of understanding of a case and that this
may be represented in the extent of hypothesis category use. In other words, a
higher level of integrated thought process should relate to a greater number of
interrelated hypothesis categories. However this has not been reported or
investigated in the literature to date.

85
Hypothesis sub-codes stored within NVivo 7 software allowed for simple
review of data relating to the range of categories used by each participant and
the percentage use of each category by the expert and novices groups.
Following this quantitative process the integration of various categories into
one overall case understanding can be reviewed qualitatively to add more
depth to the findings.

3.7 QUANTITATIVE VIDEO DATA ANALYSIS

The final stage of data analysis involved obtaining quantitative data from the
simulated client assessments. The videorecorded client assessments provided
observational data which included:
 Order of participant questions to the simulated client
 Timing data during the client assessment
 Simulated client response accuracy.

The participant questions to the simulated client were manually transcribed

then grouped into traditional categories of similar question types, such as area
of symptoms or past history questions. This allows for more detailed
investigation of the order of questioning undertaken by each participant and
has potential to shed more light on the pathway undertaken and possible use
of forwards reasoning strategies. Subsequently a comparison of pathways
taken between participants can be analysed for similarities and differences.

A second area of quantitative analysis involved obtaining and analysing the

total amount of time per client history, along with the time taken to develop any
central hypothesis that was immediately formed via PR. This time data was not
a key component of identifying PR given that physiotherapy experts who are
more likely to use PR have been noted to spend more time taking a client
history (Doody & McAteer, 2002; King & Bithell, 1998). However the reported
efficiency of PR (Arocha et al, 1993; Higgs & Jones, 2000; Ridderkhoff, 1989)
required observation of time during the client assessment component of the

86
study. The analysis of time data took place following identification of reasoning
processes.

The intra-coder and simulated client reliability were also analysed

quantitatively to evaluate the consistency of the methodology of the study.
The accuracy of the simulated client actor was obtained via video
observation data. The data obtained from the simulated client response
checklist (Appendix 4) were transferred to nominal categories for every group
of similar questions for each participant. This provided an indication of
accuracy via calculating proportions of correct responses for each question
group and overall. Validity was also evaluated using comments from
participants on the realism of the actor playing the simulated client. Training
towards achieving consistent accuracy of the simulated client across all
participants has been discussed in section 3.2.3.

3.8 TRIANGULATION

The current study incorporated triangulation involving a mixed methods

approach combining qualitative and quantitative data from participant
transcripts, in addition to a third source of video observation data. The
embedded and triangulation designs introduced in Figures 2.3 and 2.4 have
been further detailed in Figure 3.5 to assist understanding of their application
in this study.

The two forms of triangulation utilised in the current study include data
triangulation and methodological triangulation (Patton, 1990). Data
triangulation involves the use of more than one type of data to understand a
phenomenon, whilst methodological triangulation can refer to the inclusion of
mixed methods. The present research design has included a combination of
data and method triangulation to enhance the validity of the findings. This
triangulated design has been depicted in Figure 3.6.

87
 Qualitative transcript
analysis
Quantitative
Quantitative analysis video analysis
of transcript data

Study interpretation

Figure 3.5 Embedded method design

Qualitative data
obtained from
interview
transcripts

Quantitative data Quantitative data

obtained from obtained from video
interview observations
transcripts

Figure 3.6 Triangulated design

88
3.9 SUMMARY OF METHODOLOGY

This chapter details the inclusion of a mixed methods research approach with
a single case study design to investigate PR employed during clinical
reasoning. The use of multiple sources of data is aimed at enhancing the
overall quality of the findings and to allow more confidence in the results.
Integrating quantitative and qualitative sources of data is an increasingly
utilised method (Creswell & Plano-Clark, 2007) that can enhance the
interpretation of the evidence within clinical reasoning research.

The method outlined in Chapter 3 is considered the most suitable to

investigate PR as a forwards pathway during clinical reasoning. Physiotherapy
experts are predicted to utilise these strategies to a greater degree than clinical
novices. This carefully chosen methodology is aimed at providing a better
understanding of this clinical reasoning strategy in musculoskeletal
physiotherapy.

89
CHAPTER 4. RESULTS

4.1 INTRODUCTION TO FINDINGS

The clinical reasoning process represents an important component of

physiotherapy problem solving in practice. The two groups of participants
recruited for the present study have been chosen to provide more insight into
the problem solving process or pathways undertaken and particularly the use
of PR as a forwards reasoning pathway. The key study findings can be found
in section 4.4, including the identification of PR and its relationship with
diagnostic accuracy and efficiency.

This chapter commences with the details relating to participant recruitment

(section 4.2). Examples of coded data and the thematic analysis are then
presented in section 4.3, along with intra-coder reliability results. The
outcome of the thematic analysis relative to the reasoning process is
provided in section 4.5. These findings present a qualitative view of the
reasoning processes observed during the study and provide the researcher’s
view of clinical pattern use amongst participants.

Participant use of the hypothesis categories (section 2.3.1) during problem

solving has been detailed with respect to expert and novice differences in
section 4.6. This analysis has been separated into categories identified from
the final participant hypotheses (section 4.6.1) and those observed to be
used by participants ‘overall’ or at any stage during the interview (section
4.6.2).

The new codes labelled as free nodes in NVivo 7 (section 4.3.3) and
subsequent themes identified (section 4.3.4) were condensed into five
additional topics of interest. These have been reported as ‘additional findings’
in section 4.7, which provide some interesting qualitative observations related
to clinical reasoning. Lastly, analysis of study location, case simulation and

90
study equipment influences on participant behaviour are presented in section
4.8.

4.1.1 Examples of coded text provided as results

The presentation of specific examples of participant transcript text will be

included in subsequent results sections. As with the majority of qualitative
studies, only those statements that help give the reader an understanding of
the basis to the study findings will be presented rather than an exhaustive list
of quotations. Every attempt has been made to openly and thoroughly
present sufficient transcript examples that relate to the results. The aim was
to provide meaningful transcript quotations that allow the reader to develop
their own opinion of the presented material and be able to critically review the
study findings.

The transcript data includes the interviewer questions at points where it

allows for a more complete understanding of the responses. Throughout this
chapter, any reported text including the researcher’s questions or comments
have been labelled with the letter R and participant responses with the letter
P. If no label exists in association with transcript data it includes only those
responses of participants.

4.2 PARTICIPANT RECUITMENT

In all nineteen physiotherapists participated in the study. The targeted

number of ten experts was recruited over a one year period. The nine
novices were also recruited over one year with completion of data collection
occurring due to unavailability of the simulated patient actor. There was
overlap between data collection and data analysis of experts and novices.
The total timeframe for data collection was approximately fourteen months.

Throughout data collection each participant was labelled with either the letter
E for expert or N for novice followed by a consecutive number associated

91
with recruitment (e.g. E4, N6, E9, N2). In line with confidentiality
requirements of the ethics approval, the participants were randomly assigned
a letter which replaced the number. The expert participants were randomly
assigned one of the first ten letters of the alphabet and the novice
participants the next nine letters, for example ‘Expert G’ or ‘Novice P’. This
format will be used to identify a participant in the results and discussion
chapters of this thesis.

4.3 CODING AND THEMATIC ANALYSIS

Complete transcript coding and subsequent analysis of results was

undertaken by the student researcher. The study limitations associated with
sole coding were considered and managed by the coding review process
(section 3.5.1), an intra-coder reliability evaluation, and most importantly by
general agreement between all researchers on the key study findings.

The pre-determined (tree) codes listed in Table 3.2 have been included in
this section to offer real participant examples of each code to the reader. The
pre-determined codes of knowledge and self awareness were sub-coded
following completion of participant recruitment and transcript coding. This
subsequent process attempted to provide more depth to the understanding of
the clinical reasoning pathway and possible identification of PR. The free
nodes developed throughout the process of coding have been similarly
introduced, described and examples provided.

The overall results of thematic analysis are presented in section 4.3.4.

Subsequent sections then detail the process of grouping relevant qualitative
transcript data into each meaningful theme.

4.3.1 Tree codes

Throughout the coding process transcript data was not limited to any single
code within the pre-determined coding schema. Overlap between codes and

92
within the hypothesis categories did occur and was important for accurate
storage and easy retrieval of data during analysis.

Data collection
‘Data collection’ coding occurred when a participant reported routine data
collection during the assessment without relating the clinical information to
any hypothesis of the case. ‘Data collection’ was included within coding due
to its negative impact on determining whether forwards reasoning was
evident. ‘Data collection’ unrelated to a hypothesis tended to suggest that
forwards reasoning or PR was not being utilised.

The ‘data collection’ code also highlighted that an element of routine always
existed during the client assessment, regardless of the reasoning process
utilised. For example:
Expert F: R: So were there any unexpected findings out of all the general
health questions?
P: No, no that was all fine. I always go through those with them.
Expert A: R: Are these fairly standard questions that you would normally
ask at the end of an assessment?
P: Yeah I ask every one there.

During the early stages of each participant interview, data collection was
often prompted by the student researcher / interviewer to initiate discussion
and facilitate more open dialogue relating to the thought process during the
client assessment. Any ‘data collection’ coded text that was prompted by the
researcher was noted for subsequent review.

Examples of prompted ‘data collection’ text:

Novice R: R: Does that information lead where you go next, or do you just
continue on with data collection, routine data collection?
P: Um, from there I sort of thought that early information didn’t
really delve too deeply into it at that stage -- And sort of when
back to another routine run through.

93
Expert I: R: At this stage were you more collecting data or were you
actually forming an understanding that you –
P: No still at this stage collecting data.

Examples of unprompted ‘data collection’ coded text:

Novice P: Um just the filling out the body chart, trying to get a bit of an idea
of the location of, of the location of her symptoms and the quality
of her symptoms, which ah the moment probably just probably
not even thinking too much about, about what’s going on with her
and just trying to gather as much information at this point in time.
You know just sort of gathering information
Expert D: Just routine questions that I would a ordinarily ask, yeah, you
know occasionally someone would say oh yeah I’ve got
rheumatoid arthritis and well you’d be thinking OK so I just, I
always ask them, I’m sort of on autopilot going through that just
to make sure that I’ve covered all of those specific things

Hypothesis related
Hypothesis was coded when a participant stated an understanding of the
case in any of the hypothesis categories (Jones & Rivett, 2004). Any lay level
of response, where no interpretation was evident was not coded as a
hypothesis. For example, descriptive comments such as “the pain was fairly
strong” weren’t coded. Simultaneous sub-coding of hypotheses occurred into
the relevant category.

Category 1: Activity & participation

Any hypothesis relating to the capabilities or restrictions of the client during a
specific activity or life situation was sub-coded into this category. Examples
included:
Expert F: it didn’t ever stop her playing netball. So it couldn’t have been,
like it wasn’t a severe -- thing that interfered with the lifestyle.

94
Novice N: She hasn’t lost any time off work although she’s stopped playing
sport and it does hamper her ability to perform the normal things
that she needs to do.

Category 2: Patient’s perspective and psychosocial factors

Hypotheses relating to the client’s perspective in terms of their
understanding, feelings or beliefs related to the presenting problems was
placed within hypothesis category 2. The patient’s perspective may be a
contributing factor to or a consequence of the pain or restriction in activity /
participation, but in either case may be relevant to the recovery process. This
category directly relates to the illness experience that is unique to each
individual person. For example:
Expert G: looking at a goal of hers that you know she’s worried she’s got
another clinical placement.
Expert D: we know she’s got a poor understanding of what her problem is.
It just gives me a little bit more of a picture of who she is. What
she’s about. I’m thinking at that point in time she said I’m a little
bit worried about how this is going to affect me in the future. I’m
really looking for how it’s affecting her from a psychological or
emotional perspective. Because that will impact on how she
responds to any treatment or intervention that I might give her

Category 3: Pathobiological mechanisms

Hypotheses in this category may include stage of tissue healing and / or pain
mechanisms that relate to initial onset or maintenance of signs and
symptoms by the nervous system. Comments from participants related to the
underlying mechanism of symptom maintenance. For example:
Expert G: I was actually thinking maybe there was a, a sort of an
inflammatory wind up so I guess a, a secondary sensitisation to
the pain
Expert I: still behaved mechanically to a degree. Um at rest it still gives
her varied amounts of aches, um -- that makes it less, less likely
to be a significant inflammatory pathology or some kind of nasty

95
 metastatic thing or bony, bony thing that would give unrelenting
night pain.

Category 4: Physical impairments and structural sources

Any structures or tissues hypothesised to be a pathological source of the
client’s symptoms and signs were placed in category 4. Given the
assessment did not include any physical examination, the physical
impairments included in this category were predictions of abnormal findings
in the neuromusculoskeletal system:
Expert J: started to confirm that it wasn’t the disc that it was the facet joints
because most of the discs that I see don’t like to sit and often
they won’t, they can’t sit. They’ll stand in the waiting room, and
the fact that standing made her worse and she sat to relieve it
and sat slouched made me think it’s not a disc.
Novice S: I suppose you’d have to describe it as non specific low back
pain, but the structure I was particularly interested was a, a
lumbar disc.

Category 5: Contributing factors

Reference to any aspect of the client’s condition that may have contributed to
the onset or maintenance of the presenting problem(s) has been listed in
category 5. Any reference to instability without statement of any specific
structure was coded within contributing factors. Examples of category 5
included:
Expert A: lack of rest and the fact she’s doing the same thing on her feet,
bending, just her body can’t cope with constant demands
Expert E: I’m going to expect if it’s happened on a regular basis that she’s
going to be um, really maybe weak in her core. She’s she may
have some instability in her back.
Novice M: there’s a postural component, um to her pain and I guess I’m not
really thinking about so much about structurally what’s going on.

Category 6: Precautions and contraindications

96
Hypotheses that related to safety of the client relative to the type of pathology
/ disorder, stage of healing, irritability and patient’s perspectives are noted in
category 6. Precautions and contraindications were categorised as safety
related hypotheses with respect to both physical examination and treatment:
Expert H: she doesn’t like extension so I would need to be careful not to
leave her lying on her back or even leave her lying prone for
extended periods
Novice N: Just making sure that there’s, she doesn’t have any um --
symptoms like that which could indicate something more
insidious that was meant for further investigation by medical
officers rather than physio.

Category 7: Management and treatment

Any comment relating to therapeutic intervention, including specific treatment
techniques, was considered a hypothesis in management or treatment:
Expert I: led me to think that she was probably not really resting as much
as she thinks she is, she’s still got a significant aggravating
component to it and that, that may be something what we
needed to address as far as um -- time off when she needed to.
She may need time off uni. She may need to look into special
consideration and social factors that might need to be addressed
in relation to her pain.
Novice R: I’m starting to think well maybe some investigations, but you
have to wait till after the objective to say that definitively.

Category 8: Prognosis
Predicting a possible response to treatment intervention or an outcome for a
particular problem / pathology is a prognostic hypothesis. Any reference to a
feature of the case that may influence the outcome in a positive or negative
way was placed in this category:
Expert G: at that time I was um thinking this, this might not be someone I
am aiming to get pain free but maybe to get back to her, to her
preclinical levels perhaps.

97
Expert D: I guess probably maybe that makes me think about if from a
prognostic sort of factor -- With just that information alone the
only estimation that I would make would be perhaps it’s going to
be more difficult to help her than someone who has really
intermittent sort of pain.

Category 9: Non-specific
This sub-code was added to the eight previously reported hypothesis
categories to place any hypothesis that the student researcher could not
clearly place into any other category. These were mostly descriptive
comments without a clear case understanding:
Expert H: I’m trying to work out the relationship of the pains, to make sure
they’re all connected, so I don’t so I’m not moving to, following
individual pains. So I’ve got, I’ve worked out, I’m pretty
convinced now that they’re all -- the leg one is a progression of
the back one
Novice Q: the time frame that she’d had the back pain, like, leaning towards
more of an acute on chronic condition rather than just an acute

Planning
Interview transcript comments from participants relating to planning were
divided into those relating to examination and those of management. Sub-
coding in both divisions provided an indication of the direction of participant’s
thought processes.

Examination:
Any participant statement relating to the physical examination of the client’s
condition was coded within examination ‘planning’:
Expert G: that’s going to make me look in the assessment when I check
her postural control through movement. When she sits up
straight how it, how is she doing that, where is that coming from
Novice M: I’m going to do a McKenzie, a McKenzie um assessment now to
go through the symptoms that make it better or worse

98
Management:
Participant comments relating to management planning included any
comment relating to therapeutic intervention. The quotations found in this
sub-code included all of those in ‘Hypothesis category 7’ in addition to those
that did not have a clearly stated management hypothesis. Examples of the
quotations unique to this node only included:
Expert A: looking for interventions. Yeah like if she had some intervention
that had done something for her that would be a big clue for me
in which way to head treatment
Novice Q: I’m really exploring um treatment options, um pain management
options ah obviously she’s um reasonably in, in a lot of pain, so I
mean I guess the first priority trying to settle things down.

Significant case feature

The significant case features associated with identifying PR have been
introduced in section 2.5.1. This code was identified when the therapist
highlighted the importance of the data obtained to the case at hand. It may
be singular or grouped data that is considered significant to the participant.
This code was only identified in the transcripts of expert participants:
Expert I: When she said it started when she was 11, um I deliberately
went into that in a lot more detail and asked a question about did
it trouble you as a teenager because you occasionally get
patients who recall it to something that’s happened in their past,
something completely random that may be irrelevant or may not
be irrelevant and if having said that she recalls an incident as an
11 year old and then had problems with manageable back pain
right the way through the past 11 years, is more likely to mean
that that initial episode is significant.
Expert D: Right there from that perspective I’m really starting to think hard
about some sort of like structural dysfunction. Like a
spondylolisthesis or like a pars defect, some sort of, you know
just the way she’s describing landing on her bottom and then

99
 physically walking for a couple of days that’s when I’m really
starting, start to head off down that way.

Knowledge
Knowledge coding was necessarily based on the clinical reasoning literature.
In this study, any participant reference to their knowledge in an attempt to
apply it to the problem at hand was knowledge coded.

A sub-coding process took place following the completion of coding all

participant transcripts into knowledge and other pre-determined codes. The
knowledge sub-coding tree (Figure 4.1) depicts the final catgeorisation of
transcript text at three further levels. These sub-categories of knowledge
were developed based on the literature (section 2.7). The purpose of sub-
coding knowledge related transcript text into separate categories was for
easy retrieval and subsequent analysis.

Propositional

Knowledge Personal

Non-propositional

Professional

Experiential

Structured

Figure 4.1 Knowledge sub-coding

100
Separating propositional and non-propositional knowledge is reasonably
clear in theory, as outlined in section 2.7. However a practical separation was
more complex due to the overlap between knowledge types within a
clinician’s highly structured and integrated knowledge base. Importantly, the
sub-coding of knowledge into propositional and non-propositional knowledge
types was not necessary from a perspective of identifying PR as all
knowledge coded comments could offer insight into the reasoning process
when considered individually.

The following transcript quotes are all examples from the knowledge node.
They have been presented in their sub-codes to present the reader with a
view of the potential overlap between knowledge types. All knowledge coded
text was placed in either propositional or non-propositional sub-codes.
Similarly all non-propositional text was placed in either personal or
professional sub-sections. Lastly, all text placed in the professional sub-
section of non-propositional knowledge was then separated into either
experiential or structured. The definitions used for each sub-code are stated
prior to the examples.

Propositional knowledge
Identifying propositional knowledge required reference to research or
learning from textbooks, journal articles, conferences or specific structured
learning courses that would indicate use of public knowledge (section 2.7.1).
It was recognised that propositional knowledge most likely underlies a large
number of the coded transcript text segments but could not be clearly
identified. This code was only identified in expert transcripts:
Expert C: that’s based on the information we have collected over twelve
years on our back program … and those benchmarks have been
published
Expert I: there’s that extension aggravator, and flexion is a, a relieving um
factor. Um based on the last MPA conference in Brisbane where
they reported disc was aggravated by sitting -- that, that probably

101
 changed what I was thinking, traditionally thought about, discy
behaviour.

Non-propositional knowledge
Professional non-propositional knowledge
Non-propositional knowledge was reported in the literature to incorporate
several knowledge types including professional, personal and experiential
(section 2.7.1). The sub-coding of non-propositional knowledge depicted in
Figure 4.1 was created as a practical categorisation of such knowledge types
that often overlap or cannot clearly be separated. Sub-coding ‘experiential
knowledge’ was useful to monitor the use of participant’s own professional
experience, except those that were coded under propositional knowledge.
The ‘structured knowledge’ sub-code was developed as an alternate non-
propositional knowledge category to experiential.

Experiential knowledge
Experiential knowledge was coded with any reference to knowledge directly
attributed to clinical experience or clearly able to be inferred from experience:
Expert I: No I don’t see a lot of kids because kids typically don’t get, get
low back pain
Novice Q: you know usually when everything’s flared up, you know it mucks
up the tests and it cause everything hurts so I’m looking at some
treatment options to settle things down and then hopefully that
will present a clearer picture over the next few, few treatment
sessions

Structured knowledge
Any use of structured or a conceptual type of professional non-propositional
knowledge that could not be directly attributed to clinical experience was
coded within this category:
Expert J: if it’s equal left and right I’m usually thinking it’s a central problem
and therefore it could be the disc

102
Expert G: I do tend to think of it more as a, as a motion segment rather
than a disc and joint.
Novice R: it’s still going to be one of you’re balls that you’re juggling, but
the 24 hour pattern doesn’t really fit
Novice P: when she was talking about the nursing um thing, it could, given
her age could be a possible discy type irritation.

Personal non-propositional knowledge

Personal knowledge was placed within the non-propositional sub-code along
with professional knowledge. This division was to acknowledge during coding
that clinicians of all levels of professional experience utilise their own
personal experiences during practice. This sub-coding was only used during
one expert transcript:
Expert D: I guess probably the way she described or she said she’s a
nursing student and I know from personal experience, not that
I’ve been a nursing student or anything, but seeing nursing
students in hospitals and things like that, they spend a hell of a
lot of time standing around. Standing, listening to people talk to
them, being shown things.

Self-awareness
Self-awareness was created as a pre-determined code due to its importance
as an element of expert clinical practice. It was hypothesised that the expert
participants would display greater self-awareness during cognition than their
novice counterparts. The inclusion of the self-awareness code did not
however offer any direct evidence for the identification of PR.

Any occasion when the participant verbalised an awareness of their own

thinking was coded as self awareness. These were likened to metacognitive
reflections and could relate to any aspect of reasoning including data,
process, hypotheses, decisions or knowledge:
Expert H: Because that was going to be quite confusing, that would send
me down a completely different track … I thought I’ll just make a

103
 note of it. I put a star by it to make sure that I do chase it up … I
put it aside, I put in the slightly too hard at the moment basket
and then I’ll come back to it
Novice S: I don’t like to um to rush into decision-making at the end of my
subjective. Um so I, I try to, to piece together the information that
I’ve, I’ve collected and, and form an objective around that, and
sort of ah gradually um get to, get to an answer in my head.
Which is the way I normally go about it

The self-awareness coded text was reviewed following the complete coding
of all transcripts and separated into 11 sub-codes (Table 4.1). Although these
categories offered potential to provide insight into the cognitive process of
participants, these data were not seen by the researchers to be useful
towards the aims of the study. Examples of each of the self-awareness sub-
coded text have been provided in Appendix 14.

104
Table 4.1 Self awareness sub-codes

Concern Thoughts conveying concern related to the

assessment
Deferred integration Participant comments on the approach towards
understanding being one of collect data now and
think later
Diagnosis Thoughts relating to diagnosis
Direction Awareness of the direction of the thought
process in this case
Interesting The participant verbalises the awareness of the
data or the case as interesting
Prediction A thought process that is predicting a response
or outcome
Process General approach to an assessment is
commented on as a process with all cases, not
specific to this case alone
Significance Recognition of the importance of the data
collected
Surprise Unexpected response leading to thoughts of
surprise regarding the understanding of the case
Uncertainty Confusion or uncertainty acknowledged relating
to the case understanding at that point in time
Other Including: Distrust of client / Client’s perception /
Doesn't fit / Data not useful

4.3.2 Intra-coder reliability

Intra-rater reliability was considered given the sole coding process during the
majority of data analysis. The first two transcripts of each group were repeat
coded after complete coding of all transcripts. The duration between initial

105
and repeat coding of each transcript was greater than 3 months to reduce the
likelihood of memory recall during repeat coding. The intra-rater reliability
was analysed via percentage agreement and kappa reliability coefficient
(Domholdt, 2005). This process was undertaken only for tree nodes which
included the six codes outlined in Table 3.2.

The reliability analysis indicated a very high level of agreement between the
two episodes of coding. Individual agreement percentage and kappa
coefficient statistics have been provided in Table 4.2.

Table 4.2 Intra-rater coding reliability

Participant Percentage agreement Kappa coefficient

Expert J 97% 0.96
Novice R 96% 0.94
Expert I 96% 0.94
Novice S 98% 0.97

4.3.3 Free codes

New codes were created during the overview reading and coding stages of
analysis. These were entered into NVivo 7 software as free nodes and notes
taken by the student researcher as outlined in section 3.5.3. Each identified
free node has been documented below with definition and examples
provided. A summary of free nodes is listed in Table 4.3.

106
Table 4.3 Free nodes

 Analytical reasoning
 Data confirmation
 Direction of reasoning
 Goal setting
 Hypothesis confirmation
 Hypothesis elimination
 Negative predictive
 Pattern related
 Predictive reasoning
 Simulated assessment
 Thinking after the event

Analytical reasoning
Definition: Any comment that could directly be seen to support HDR (and
thus potentially negate PR)
Purpose: Subsequent review of this free node attempted to provide
qualitative support to the quantitative interpretations regarding reasoning
process.
Examples:
Novice S: at that stage I didn’t have a clear cut diagnosis in my head. Um, I
had some areas and some structures that I, I was interested in
that I would have um been testing
Expert J: Cause I was still confused whether I really thought it was a disc
or the facet joints which is what I’m really trying to work out, and
it didn’t quite fit to me

Data confirmation
Definition: The participant repeated information back to client for
confirmation of accuracy in understanding a single piece or collective group
of data

107
Purpose: Reiteration is potentially an interview method that increases
accuracy of data collection. Errors in reasoning may in part be related to data
collection. Identification of this free node was able to be compared with
overall accuracy.
Examples:
Novice M: I’m trying to make sure that I’ve got everything important and
haven’t missed anything and I’m also just trying to clarify in my
mind what’s going on and making sure that there’s no
discrepancies in the story
Expert I: trying to clarify that she’s telling me that it has been an 11 year
history of episodic low back pain. It’s always been there, um that
it’s been specifically this clinical placement, that, that’s brought it
to a head.

Direction of reasoning
Definition: During the retrospective recall interview the participant referred to
their direction of thought process from the client assessment.
Purpose: Direction of thoughts was directly related to the interpretation of
forwards or backwards reasoning process.
Examples:
Novice S: I think if I felt that I was thrown off a little bit, then I was just more
intent on getting more data, um to clear that up
Expert I: I generally tend to keep it as open as I can with the history so
that people could volunteer as much information and if they’re
not doing it, then I’ll prompt them a little bit more, but otherwise
let them go with it. Um when she said it started when she was
11, um I deliberately went into that in a lot more detail and asked
question about did it trouble you as a teenager because you
occasionally get patients who recall it to something that’s
happened in their birth or something completely random that
maybe irrelevant or may not be irrelevant

Goal setting

108
Definition: The participant refers to the client’s goals within the interview.
This was either goals obtained from the client or reference to a question that
was attempting to investigate the client’s goals.
Purpose: Goal setting clearly involves the client collaboratively in the
assessment process which is an important part of clinical reasoning. It is not
known whether collaboration is particularly associated with any one type of
reasoning strategy.
Examples:
Novice P: a few of the questions getting back to see what does she want
out of life, where are we going from here, what’s important to
her, and are we just looking to get, get rid of the pain or are we
looking to get rid of the pain and then getting to play netball for
Australia
Expert G: looking at a goal of hers that you know she’s worried she’s got
another clinical placement starting to think about again goals or
treatment where we, what sort of goals we’re going to have for
her from that point of view

Hypothesis confirmation
Definition: Any direct or indirect reference towards a prior hypothesis being
confirmed.
Purpose: The grouping of text segments highlighting confirmation of a
hypothesis potentially added to the identification of a forwards reasoning
process.
Examples:
Novice Q: feels pretty good in the morning it’s less likely to be disc. Um and
tends to you know um support more that lumbopelvic instability
Expert I: it’s further adding to my um thoughts that there’s something
structural underlying the, and structural and permanent
underlying the, the problem

Hypothesis elimination

109
Definition: Reference to removing a hypothesis from the overall
understanding following new information or data from the assessment.
Purpose: Hypothesis related data organised within the hypothesis tree node
were also placed in this free node if they related to eliminating any
hypotheses through ongoing data collection. This node was included to
monitor hypothesis elimination as a component of HDR.
Examples:
Novice R: Pattern really wasn’t that discy … you know most people say stiff
and sore, she said she’s usually pain free in the morning
Expert D: OK that starts to sort of in her case rule out some of the other
things that I was wondering about like you know, a lot of times
people with disc problems for example sitting might aggravate it

Negative predictive
Definition: Participant questions deliberately used during the assessment to
provide evidence that a clinical hypothesis is false.
Purpose: Process of elimination was observed in association with both
analytical and non-analytical reasoning types and monitored as a client
questioning strategy.
Examples:
Novice Q: I was just sort of you know chucking in a few questions to sort of
negate structures perhaps, and it sort of leads me towards more,
more that chronic um instability
Expert G: I was expecting that she would say um that she didn’t have any
pins and needles or numbness so definitely more to confirm what
I was thinking

Pattern related
Definition: Any data found during analysis that related to participant pattern
use, including recognition and elimination.
Purpose: Similarly to the analytical reasoning free node, the subsequent
review of this free node attempted to provide qualitative support to the
quantitative interpretations regarding reasoning process.

110
Examples:
Novice Q: Even though it doesn’t seem likely given the pattern and the
description and aggravating, um or something ah maybe
something like a um stenosis or foramina or something like that.
Um which may, may relate given the area and that and that
extension
Expert C: the pattern is one that I recognise and looks mechanical

Predictive reasoning
Definition: The participant predicts a response to confirm a picture or
comments on a prediction after data is gathered. Any comments on the
therapist’s predictive strategies in clinical practice were also coded.
Purpose: Predictive reasoning is a strategy reported in the literature that was
noted during coding. It was monitored during qualitative analysis for thematic
purposes and comparison with PR users.
Examples:
Expert G: And what muscles is she using … she said sitting up straight
makes it worse, I was then getting in my head I’m guessing that
when she sits up straight she does it with the wrong pattern so
she uses her superficial muscles
Expert F: it hasn’t fitted into what I thought. And I mean I guess as you, I
mean that’s what makes your practice interesting. I mean I do
play games with myself at guessing what I think the problem is.
I’ll guess, I’ll try and guess what their answers are going to be
before they give me answers. Yeah and then I see whether I’m
right or wrong.
Only one participant using a PR process was also observed to use predictive
reasoning during the study.

Simulated assessment
Definition: Any comments that relate to the simulated patient’s realism or
performance.

111
Purpose: Easy storage of participant comments relating to the research
method increases efficiency of analysing this information.
Examples:
Novice R: at one point I um I looked up at Michelle and I thought, gee she’s
a good actor (laugh) and that just sort of swayed me a bit but you
know other than that no not at all
Expert D: I started wondering I wonder if she’s actually got this problem.

Thinking after the event

Definition: Any direct or assumed reference that indicates the participant
was thinking after the event (i.e. problem solving the data whilst watching the
video) is placed in this node
Purpose: Any text coded in this free node was reviewed relative to study
findings. The possibility of thinking after the event is a potential limitation of
this methodology and was monitored for discussion purposes.
Examples:
Novice Q: in hindsight I probably didn’t explore that enough.
Expert J: Maybe I should have asked if it was the rotation opening but I
don’t know if she would have known that. Because then that
might have helped if it was the disc. But because it’s bilateral, I
don’t think, I wouldn’t manipulate it.
Review of data in this code found no occasions that influenced the study
findings relating to PR identification or its accuracy.

4.3.4 Thematic analysis

The primary research aim was to investigate the phenomenon of PR within

musculoskeletal physiotherapy clinical reasoning. Thematic analysis in the
study served two purposes. Firstly, to provide a qualitative source of data that
can support the identification of forwards and backwards reasoning
strategies. Secondly, to identify additional findings from a sole qualitative
theme development process that may inter-relate to the reasoning process.

112
In particular, noting differences between expert and novice physiotherapists
relating to clinical reasoning.

The themes were common observations identified during coding. These have
been documented wherever a topic was noted in two or more participant
transcripts. The frequency of underlying observations of a theme (Table 4.4)
highlights the weakness or strength of that theme relative to the participants
in the study. Those findings identified in the responses of only a small
number of participants have been presented to offer a comprehensive view of
the complexity and variability of problem solving in practice.

In respect of the first intention of thematic analysis, only two of the themes
directly assisted identification of directional reasoning process. Analytical
reasoning and pattern related themes were utilised to further support the
quantitative study findings. This provided strength to the study via
triangulation using different types of data sources and analysis methods.
Section 4.5 details the qualitative data from these two themes.

The qualitative aspect of the study allowed for the identification of new
themes as outlined in section 3.5.5. All themes other than ‘analytical
reasoning’ and ‘pattern related’ have been analysed and reported as
additional study findings in section 4.7.

113
 Table 4.4 Themes identified
Theme Description Experts Novices
Analytical reasoning When hypothetico-deductive reasoning was determined to E,F,G,H,J K,L,M,N,O,P,Q,R,S
have taken place
Collect data now and Participant refers to their approach in this manner A,J R,M,L
think later
Focused on diagnosis Predominant focus on identifying a diagnosis J S
Importance of history Stated emphasis on client history information F,J
versus physical Stated emphasis on physical examination information E
examination information Stated emphasis on integration of all data G,H R,S
Open minded approach Participant refers to their desire to be open to other possible D,H,I
to problem solving hypotheses whether having a primary understanding or not
Outcome data search The search for data that can be used later as an outcome A,C,B,G K,L,Q
measure. Including reference to client goal setting
Pattern related  It doesn’t ‘fit’ – pattern not recognised E,F
 Disbelieving approach E
 Differentiating from hypothetico-deductive reasoning G Q,R
 Recognition C,D,I
Person centred A search for an understanding of the person is apparent A,B,C,D,F,G,H,I K,N,O,P
approach along with a problem based understanding
Predictive reasoning Participant is predicting the outcome of a question prior to D,F,G N,Q
asking it
Reference to recent Participant refers to formal education during interview A,F K,M
professional education
Reiterating information Participant was observed to repeat data collected back to D,I M,P,Q
back to client client during the assessment
Search for symmetry in Reference to searching for symmetry or asymmetry during C,E,F,H L
symptoms symptom location

114
4.4 PATTERN RECOGNITION

The primary study aim was to determine whether PR is utilised by expert and
novice clinical physiotherapists in the musculoskeletal field. This section
provides quantitative results with respect to the research aims of identifying
PR and evaluating its relationship to accuracy and efficiency. Agreement
between all three researchers was achieved with respect to the reported
findings.

4.4.1 Identification of pattern recognition

The identification of PR incorporated the use of a ‘time stop identification tool’

and an ‘overall scoring tool’ (Appendix 13). These were applied to all
participant transcripts via the NVivo 7 software and coding retrieval options.
Section 3.6.1 and Table 3.3 have detailed the individual features of PR used
in this analysis.

An example of an ‘overall scoring tool’ finding positive identification of PR has

been provided in Table 4.5. Conversely an example of the same tool unable
to locate any evidence of PR is provided in Table 4.6. The percentage time
row of these tables is the actual time at which the central hypothesis was
formed as a percentage of the total assessment time. This relates to
efficiency data which has been detailed in section 4.4.4.

115
 Table 4.5 Overall participant scoring tool identifying pattern recognition for Expert D
Feature of pattern Present Evidence Comments
recognition
Central hypothesis Yes Time stop 3:50
formed “I’m really looking for is if she’s got any time signs of spinal cord Distinct point in time.
compression, cauda equina sort of issues”
Did you have those at this point in mind? “Yeah, yeah”
“one of the things that I sort of think about there would be say
spondylolisthesis”

Time stop 5:45

“Right there from that perspective I’m really starting to think hard about some
sort of like structural dysfunction. Like a spondylolisthesis disorder. Like a
pars defect, some sort of, you know just the way she’s describing landing on
her bottom and then difficulty walking for a couple of days that’s when I’m
really starting, start to head off down that way”

Actual time of central

hypothesis formation as 21% Central hypothesis formed at 3 minutes and 50 seconds, equating to 21% of
a percentage of the total Expert D’s total assessment (18 minutes)
assessment time

Significant case features Yes Time stop 3:50

“bilateral nature of the symptoms, the spread, the fact that she’s sort of 3:50 was the distinct
saying she’s, she’s up, she’s active that sort of stuff” point in time.
Time stop 5:45
“the way she’s describing landing on her bottom and then difficulty walking for
a couple of days”

Professional knowledge Yes Time stop 6:30

“you know sometimes some people will lay, they’ll say if I lay flat on my Repeatedly relates
stomach I’m better and if it was what I’m thinking spondylolisthesis or some knowledge to primary
sort of extension based sort of disorder, um that probably wouldn’t be the hypothesis.
case”

116
 Time stop 7:50
“I know from personal experience, not that I’ve been a nursing student or
anything, but seeing nursing students in hospitals and things like that, they
spend a hell of a lot of time standing around. Standing, listening to people talk
to them, being shown things, doing pretty crappy sort of jobs … or they’re sort
of leaning over making beds and things like that, and I’m thinking OK it’s
upright postures, um maybe sort of sustained semi flexion, that sort of stuff
seems to be the thing that’s made her worse. It’s a significant sort of change
in her normal activities which if she’s a student normally she’d be sitting down
… if she’s got a spondylolisthesis or some sort of posterior sort of structural
issue, but now she’s upright and on her feet a lot more and it’s you know it’s
made her condition feel worse”

Time stop 13:09

“Manipulation induced analgesia is pretty common in a lot of back problem …
and I’d also just think even with like some sort of posterior element to the
instability, the spondylolisthesis or you know some sort of hard defect or even
just an instability generally you can manipulate around that area and it will
give you some symptomatic relief”

Prior clinical experience Yes Time stop 15:41

of this case “So it kind of confirmed what I was thinking”. So does that fit with then what Directly referring to prior
you’ve seen before? “Yeah, that’s based on previous experience with people experience with the
who are describing a similar story to what she is” same type of case.

Time stop 18:00

Have you seen a case similar to this before?
“several times … several times previously I would have this story being
described”

Management stated Yes Time stop 15:41

“from a treatment perspective we’ll need to teach her how to control her
spinal position”

117
 Table 4.6 Overall participant scoring tool without evidence of pattern recognition for Novice S
Feature of pattern Present Evidence Comments
recognition
Central hypothesis No Time stop 5:30
formed “I was fairly certain at this point that um her, her pain level was very activity Initial hypothesis was
dependent. Um and that it was directly related to how much activity activity dependent and
specifically netball, she was doing as to how bad the pain got, and I also all pains were related to
wanted to know whether her thigh pains her knee pain, buttock pain and back the lower back.
pain were all related. And um from the way that she described that, each of
them came on fairly systematically with levels of activity. Um that made me
start to think that they were all related, to the one problem”

Time stop 7:35

“I was tossing up things like um ah, disc, ah pain that was originated from a L, Then structural
a lumber disc, um SIJ pain. Things like that um and I was, I was leading more hypotheses highlighted
towards the um the lumber spine um maybe disc related pain or the joint a deductive process.
related pain, um because of the movements and things that were stirring it
up”

Time stop 10:55

“I had narrowed down … my main hypotheses for her pain and that was that Disc became the
she had um -- I suppose you’d have to describe it as non specific low back developing hypothesis
pain, but the structure I was particularly interested was a, a lumbar disc. Um - but was considered
- and I was thinking that her clinical had stirred that up, um and made that, alongside other possible
made that more painful” hypotheses.

Time stop 14:20

“sort of chasing a lumbar disc as a possible structure. Um but also the, the
referral pain down both legs and into her knee I hadn’t um clearly determined
whether that was, was coming from a structure in her lower back, or not. So I
wanted to look at things like um her piriformis, her SIJ a little bit more with
some testing, um to try and determine whether I could figure out if those all
were referred pain into her buttocks and thighs and just below the knee where
it related to the one area”

118
 Time stop 14:20
“a lumbar disc is one that is standing out um, more than others. Although it’s -
- just the lumbar spine specifically that is standing out some more, so I
wouldn’t, um, I wouldn’t rule out ah some Z joint involvement um or some,
some muscular involvement either. Um -- but that, those are probably my, my
top ones”

Actual time of central Nil

hypothesis formation as
a percentage of the total
assessment time

Significant case features No

Professional knowledge No Time stop 8:45

“she was still standing when she did that, that should take some pressure off. Use of knowledge
Um, something like a disc or the low back when she’s in standing, so that related to the structural
refocused me a little, which was good” disc hypothesis but
within a reasoning
process with several
possible hypotheses.

Prior clinical experience No Time stop 14:20

of this case “I was thinking at, at the time going through the subjective that I didn’t have No similar clinical
something else to compare it to” experience was noted.

Management stated No

119
The results of the two analysis tools were then summarised in Table 4.7. This
tabled information identified each feature of PR and provided an overall view
of the presence of PR within each participant interview. A central hypothesis
was necessary throughout the assessment for overall identification of PR
(section 3.6.1). Any alternative hypotheses in conflict with another indicated a
deductive process and opposed the identification of PR.

In all cases where PR use was determined, both a ‘central hypothesis’ and
associated ‘significant case features’ where identified. The ‘number of yes
responses’ did not determine whether PR was used however higher scores
provided greater support for utilisation of PR. The ‘number of yes responses’
formed a score from 0 to 5 for each participant. Those participants using PR
were found to have a score of 3 or above.

The numbers of participants clearly utilising PR during the study can be seen
in the final column of Table 4.7. In all five participants, four out of ten experts
and one out of nine novices, incorporated PR into their reasoning of the
clinical case. The overall novice score was 3 out of 5, whilst three of the
experts scored 4 and one scored 5 out of 5.

120
 Table 4.7 Summary of pattern recognition related results

Central Prior Number of Pattern

hypothesis Time Significant Professional clinical Management yes recognition
formed formed case features knowledge experience stated responses identified
Expert B Y 23% Y Y N Y 4 Yes
Expert I Y 36% Y Y N Y 4 Yes
Expert A N . N N N N 0 No
Expert G N . N N Y N 1 No
Expert J N . N N N Y 1 No
Expert E N . N Y N N 1 No
Expert C Y 31% Y N Y Y 4 Yes
Expert H N . N Y N Y 2 No
Expert D Y 21% Y Y Y Y 5 Yes
Expert F N . N N N Y 1 No

Novice N N . N N N N 0 No
Novice R N . Y N N N 1 No
Novice L N . N N N N 0 No
Novice P Y 26% Y Y N N 3 Yes
Novice O N . N N N N 0 No
Novice S N . N N N N 0 No
Novice M N . N Y N N 1 No
Novice K N . N N N Y 1 No
Novice Q N . N N N N 0 No

Y = Yes; N = No
‘Time formed’ is the actual time at which the central hypothesis was formed as a percentage of the total assessment time
‘Number of yes responses’ is the total number of Y responses in each row (not including ‘Pattern recognition identified’)
‘Pattern recognition identified’ required the ‘central hypothesis’ and ‘significant case features’ to have a Y response
4.4.2 Comparison of expert and novice use of pattern recognition

A comparison of experts and novices was undertaken based on the model of

PR being associated with experience and expertise. That is, in order to
determine whether the findings were associated with expertise the novice
group was included in the study. Table 4.8 provides a direct comparison
between groups and includes the conversion of data to categorical form for
statistical analysis.

Table 4.8 Comparison of pattern recognition use between groups

Participant PR Identified Categorical data

conversion
Expert B Yes 1
Expert I Yes 1
Expert A No 0
Expert G No 0
Expert J No 0
Expert E No 0
Expert C Yes 1
Expert H No 0
Expert D Yes 1
Expert F No 0
Total / 10 4

Novice N No 0
Novice R No 0
Novice L No 0
Novice P Yes 1
Novice O No 0
Novice S No 0
Novice M No 0
Novice K No 0
Novice Q No 0
Total / 9 1
0 = No 1 = Yes

First, Fisher’s exact test was conducted using SPSS statistical program
(version 15) to determine group differences with respect to identifying PR.
This analysis produced a value of 0.303 when comparing experts to novices

122
indicating no significant difference between groups based on a significance
level of 0.05.

Secondly, the credible intervals for proportions were calculated. Credible

intervals are from frequentist statistical theory and refer to the Bayesian
equivalent of a confidence interval (Gelman et al, 2004). This approach to
analysis was used given the common approach, involving normal
approximation to the binomial distribution, being less reliable with small
samples. The 2.5 and 97.5 percentile points of the beta posterior distribution
were used to determine the limits, the posterior being based on a binomial
likelihood and conjugate beta prior, with both parameters being equal to 1 to
give a uniform prior distribution (Gelman et al, 2004). In cases when the
number of events observed was zero or equal to the numbers of trials the
interval was calculated as one sided as recommended by Carlin and Louis
(1996).

The credible intervals for the identification of PR amongst all participants and
each group separately are displayed in Figure 4.2. These findings suggest
no significant difference between groups, but this is particularly related to the
small sample size and lack of power. Additional sample size calculations
determined that 42 participants would be required in each group to
demonstrate a statistically significant difference between expert and novice
participants. This sample size calculation used a proportion derived from a
Po of 0.111 (based on 1 of 9 novices utilising PR equating to 11.1% of this
group) and Pi of 0.4 (based on 4 of 10 experts utilising PR equating to 40%
of this group) with a type I error rate of 0.05 and power of 0.80.

Closer analysis of the credible intervals produced by Bayesian analysis finds

PR to exist as a phenomenon. This is based on the view that the lower
margin of each interval was above zero for both groups and all participants
combined (Figure 4.2).

123
 100

80
% Chance of PR use

60

40

20

0
All participants Expert group Novice group

Figure 4.2 Credible intervals for the identification of pattern recognition (PR)

4.4.3 Accuracy

Accuracy of any case depends on comparison with a stated known

diagnosis. As described in section 3.2.2, the case utilised in the study was
known to be a lumbar spine spondylolisthesis in a 20 year old female. An
accurate outcome was therefore only considered if the participant labelled
spondylolisthesis as their primary or predominant hypothesis in the case. The
overall summary of participants’ stated primary hypotheses is outlined in
Table 4.9 which highlights that three of the experts correctly identified the
pathology. Details of transcript data leading to the primary hypothesis
summary are provided as Appendix 15.

124
Table 4.9 Stated primary hypothesis relative to pattern recognition use

Pattern
Participant recognition Stated primary hypothesis
identified
Expert B Yes Spondylolisthesis
Expert I Yes Spondylolisthesis
Expert A No Instability
Expert G No Motion segment dysfunction / Neurogenic / Instability
Expert J No Disc vs Joint
Expert E No Instability
Expert C Yes Mechanical pelvic asymmetry
Expert H No Instability / Joints
Expert D Yes Spondylolisthesis
Expert F No Nil clearly stated

Novice N No Lack of support / restriction

Novice R No Instability
Novice L No Disc
Novice P Yes Instability
Novice O No Sacroiliac joint
Novice S No Disc > Joint
Novice M No Postural
Novice K No Disc / Sacroiliac joint / Postural
Novice Q No Instability

Three of four expert participants using PR achieved an accurate diagnosis of

spondylolisthesis. The only novice participant using PR did not attain the
correct diagnosis. Conversely all the participants not using PR (six experts
and eight novices) did not achieve the correct stated diagnosis. Converting
the PR accuracy data into a two by two table for the separate groups (Table
4.10) allows accuracy relative to PR to be more easily viewed. If considering
PR accuracy across all participants regardless of experience, the number of
participants using PR accurately (3) was only just higher than those who did
not (2). Table 4.11 depicts overall PR accuracy across all participants.

125
Table 4.10 Pattern recognition accuracy between groups. Accuracy was
only considered as identification of spondylolisthesis

Pattern Pattern
recognition recognition
No Yes No Yes

No 6 1 No 8 1
Accuracy
Yes 0 3 Yes 0 0

Experts Novices

Table 4.11 Pattern recognition accuracy overall. Accuracy was only

considered as identification of spondylolisthesis

Pattern recognition

No Yes

No 14 2
Accuracy

Yes 0 3

Statistical analysis of the two by two tabled data was conducted using
Fishers’ exact test for all participants using PR regardless of group and for

126
the expert group alone. Analysis of PR users achieving a correct diagnosis
across all participants when compared with participants using analytical
reasoning strategies produced a significance value of 0.01. Relative to a
significance level of 0.05, this indicates that the use of the PR strategy was
significantly more likely to produce a correct diagnosis during a subjective
history than using analytical reasoning strategies.

Similar analysis of the expert participant group alone was considered

appropriate with respect to accuracy as the phenomenon is one that is
associated with experience. The Fishers’ exact test applied to experts using
PR as compared with those using analytical reasoning in achieving a correct
diagnosis produced a significance value of 0.033. Relative to a significance
level of 0.05, this calculation also indicated that the use of the PR strategy
was significantly more likely to produce a correct diagnosis during a
subjective history than the use of analytical reasoning strategies.

4.4.4 Efficiency

The ‘time stop identification tool’ allowed for timing data to be easily obtained
relative to the formation of a predominant hypothesis. The timing data
collected per participant included the total amount of time taken to conduct
the client history and the time taken to the formation of a predominant
hypothesis. Due to the varying lengths of each participant assessment the
time for formation of a predominant hypothesis was compared as a
percentage of each overall assessment time. Table 4.12 provides the overall
assessment times for each participant and the timing data relating to
hypothesis formation.

The actual time elapsed when the central hypothesis was first mentioned
(formed) by the participant was calculated as a percentage of the total
assessment time. In gradually developing hypotheses this stated time was
when the final predominant hypothesis was stated and held above other
possibilities. The timing data was able to be confidently and clearly stated in

127
those cases where PR was identified but less so in those without. Therefore
the time to initial predominant hypothesis was more of an estimate in non-PR
cases and not an exactly defined point in time. As such, these times were not
deemed useful in the interpretation of efficiency.

Table 4.12 Timing data relating to central hypothesis formation

Overall Time to initial Initial central
Participant assessment central hypothesis as %
time hypothesis of overall time
Expert B 20:10 3:35 23%
Expert I 19:00 6:51 36%
Expert A 13:45
Expert G 22:05
Expert J 14:00
Expert E 10:00
Expert C 9:55 3:03 31%
Expert H 18:50
Expert D 18:00 3:50 21%
Expert F 17:20

Novice N 8:10
Novice R 10:08
Novice L 12:00
Novice P 19.35 5:25 26%
Novice O 15:10
Novice S 14:20
Novice M 18:50
Novice K 16:00
Novice Q 15:20

Pattern recognition efficiency

The results of timing data highlighted that when PR use was identified the
predominant hypothesis that was maintained throughout the entire
assessment was formed in the first 36% (range 21% - 36%; median 26%;
mean 27%) of the subjective client assessment.

The overall assessment time for those using identified PR as compared with
those using analytical reasoning strategies have been analysed and
compared via simple statistics (Table 4.13).

128
Table 4.13 Total assessment time relative to reasoning method and
participant group
Both participant groups Number Range Median Mean
Pattern recognition identified 5 9:55- 19:00 17:21
20:10
Analytical strategies without 14 8:10- 14:45 14:42
pattern recognition 22:05

Experts Number Range Median Mean

Pattern recognition identified 4 9:55- 18:30 16:46
20:10
Analytical strategies without 6 10:00- 15:40 16:00
pattern recognition 22:05

Novices Number Range Median Mean

Pattern recognition identified 1 19:35 19:35 19:35

Analytical strategies without 8 8:10- 14:45 13:44

pattern recognition 18:50

Comparison of efficiency between groups

Comparison of overall time taken for the assessment between experts and
novices was also analysed independently from reasoning strategy and
compared via simple statistics as shown in Table 4.14.

129
Table 4.14 Total assessment time relative to participant group only
Participant Group Number Range Median Mean
Experts 10 9:55- 17:40 16:18
22:05
Novices 9 8:10- 15:10 14:25
19:35

4.4.5 Participant order of questioning

The physiotherapy questions observed in the study were obtained from

review of the videorecorded data of the simulated client assessment.
Transcription of this data was conducted by the student researcher.
Following this the participant questions that directed the problem solving
process were able to be analysed. The questions were subsequently
grouped based on standard areas within a musculoskeletal physiotherapy
history (Petty, 2006). Categories of questions included:
 Introductory question
 Body chart of symptoms
 Current and past history of symptoms
 Social history
 Previous treatment
 Aggravating and easing factors for symptoms
 Specific questions relating to condition irritability
 24 hour behaviour of symptoms
 Exercise related
 Goal setting
 Special questions relating to bladder / bowel dysfunction
 Special questions relating to balance
 Special questions relating to coughing and sneezing
 General health
 Medication
 Investigations

130
 Work related
 University participation
 Client beliefs
 Age

The categories of questions were similar to the areas used in the training and
assessment of simulated client response accuracy. The primary purpose was
to provide a comparison of the order of questions posed with respect to
analytical versus non-analytical reasoning and between groups.

The analysis involved placing the categorised questions into a Microsoft

Excel 2003 spreadsheet (Microsoft, 2003) based on the order of questions
irrespective of time. A second spreadsheet was then developed relating
when each question was asked to the minute of time during each participant
assessment. These data are not included in the thesis due to the shear size
of the spreadsheets and the lack of useful outcomes. However, the
observational data relating to grouped standard questions displayed some
similarities and differences between the expert and novice participants. The
entire expert participant group commenced the assessment with an
introduction question followed by body chart related questions. The novices
similarly use an introduction question but only three of the novices then
requesting body chart information, with the others asking about current or
past history information. Following this there are no obvious differences
between groups relative to the order of information requested. The data were
not analysed statistically based on the results of visual analysis. To analyse
such data requires a question of value and comparison of such data between
groups was not considered useful relative to the research question.

The primary benefit of collecting and analysing this observational data was
the potential for obtaining evidence for confirmation questions following the
identification of a pattern. Interestingly two of the experts using PR (Experts
D and I) formed their pattern immediately at the time of asking a special
question relating to bladder / bowel dysfunction. This question was

131
considered potentially related to a high grade spondylolisthesis condition.
The significance of this finding however was countered by the fact that two of
the experts using analytical reasoning and not considering spondylolisthesis
(Experts G and H) also asked the same question. It may therefore have been
an observation of question order based on undergraduate or postgraduate
training and has no relation to confirmation questioning post PR. No other
trends were observed in the data of those experts using PR or analytical
reasoning.

4.5 QUALITATIVE ANALYSIS FINDINGS

Transcript data from the free codes ‘analytical reasoning’ and ‘pattern related’
were utilised to provide qualitative support for the identification of a
diagnostic reasoning strategy. Clearly both analytical and non-analytical
strategies existed within quantitative data analysis (section 4.4). A qualitative
review of transcript quotations underlying these two themes offers an
element of triangulation to the prior reported study findings.

4.5.1 Analytical process

HDR is undoubtedly a strategy employed when unfamiliarity with a case is

present. Repeated quotations supporting an analytical reasoning process
were observed during analysis of all novice transcripts with the exception of
novice P who was previously identified as using PR. Examples of qualitative
data supporting the presence of an analytical strategy are provided.
Interestingly, these are all competing hypotheses within the ‘physical
impairments and associated structure / tissue sources’ hypothesis category
(Table 2.2) and are predominantly competing structural sources of the
symptoms:
Novice N: I was automatically thinking lumbar spine joint or SIJ (sacroiliac
joint) then with the aggravating factors ah I was thinking that it,
it’s in that area but I wasn’t yet sure which one of those it would
be more likely to be

132
Novice O: I was initially thinking discy kind of pain but with the um sitting as
an easing factor usually not so much but then again prolonged
sitting ah but then sitting in extension, so I don’t know whether
there’s ah like a, I guess it’s a facet joint kind of thing going on
with the extension kind of or still maybe an SIJ kind of thing with
the extension

Five of the ten experts were coded with ‘analytical reasoning’ and none of
these were identified as using PR (section 4.4). Only Expert A of the
participants not using PR was absent from this code. Examples of the
‘analytical reasoning’ code in expert participants are:
Expert J: Cause I was still confused whether I really thought it was a disc
or the facet joints which is what I’m really trying to work out, and
it didn’t quite fit to me
Expert G she landed on her buttocks again that made me think oh maybe
it is, is an SIJ component and um -- or is it just sort of still a
lumbar spine motion segment getting that, that compression over
ten years you know if, if she’s had fairly significant trauma then
what sort of processes have been going on.

4.5.2 Pattern related

The ‘pattern related’ free code was separated into four sub-codes as listed:
1. It doesn’t fit – pattern not recognised
2. Disbelieving approach
3. Differentiating from HDR
4. Recognition.

The first two of these sub-codes related to pattern elimination (It doesn’t fit –
pattern not recognised) or the case data not fitting a known pattern
(disbelieving approach). These sub-codes could not qualitatively support PR
but had relevance to experts attempting to fit the presenting case findings

133
with prior known experiences. Examples of the first sub-code ‘It doesn’t fit –
pattern not recognised’ include:
Expert E: I’m thinking I’m eliminating the disc as the cause. The early
morning first thing is to get out of bed. And coughing and
sneezing there’s sometimes the two of them can go together.
Um and um it doesn’t sort of fit so I’ve gone onto something else
then
Expert F: I think it’s surprising that someone her age can say that she’s
like basically got constant pain there all the time ... I mean that
surprises me. It doesn’t, it still doesn’t seem to fit into a proper
pattern, to me. Yeah, usually someone her age, I mean you can
have intermittent back pain over a long period of time, but you’ll
usually have periods where you have no pain

The second ‘pattern related’ sub-code of particular interest was a

‘disbelieving approach’ from Expert E. Several comments were made during
the retrospective interview in relation to questioning the client’s responses
because they did not fit a recognisable form:
Expert E: I wouldn’t have expected it … I may need to check her reliability
of her information because she might not either clearly hear me
or interpret me or give me the correct feedback OK so a couple
of times I’m gunna requestion her and just check her out
Expert E: I’m gathering information, um but already I didn’t like the fact that
it was symmetrical … I was just thinking that I don’t believe you
… There has to be an asymmetry

The third sub-code of ‘pattern related’ had elements of pattern use but also
clear identifiable analytical strategies. These formed a ‘differentiating from
HDR’ sub-code. Novices Q and R were identified from this sub-code and
qualitatively analysed as developing patterns (section 4.5.3). The only expert
identified from this sub-code was Expert G who made reference to patterns
but during analysis via the ‘time stop identification tool’ (Section 4.4.1) was
found to develop an understanding of the case gradually. Expert G stated a

134
broad overall understanding of “motion segment dysfunction” with “less
muscle control” and referred to their clinical experience in “hearing patients
say”:
Expert G: just from a -- hearing patients say that over and over again, so I
guess from a pattern, pattern point of view, um yeah time on her
feet, bending over, so that can still, to me still fits in with um with
the motion segment dysfunction probably less muscle control
sort of standing being perhaps a, a weight bearing um bit of in
extension but then having that control of support into flexion

The last ‘pattern related’ sub-code included quotations from transcripts that
individually supported pattern ‘recognition’ as a non-analytical reasoning
strategy:
Expert C: the pattern is one that I recognise and looks mechanical
Expert D: that’s based on previous experience with people who are
describing a similar story to what she is
Expert I: I don’t see a lot of kids because kids typically don’t get, get low
back pain … but when we do we, we seem to get them with long
term symptoms but a lot of the time there’s structural reasons why

4.5.3 Developing patterns

The identification of PR in the present study could only be assured with a

clear dominant hypothesis throughout the entire clinical assessment. This
allowed for clear separation and comparison between participants using PR
versus those incorporating more analytical methods of reasoning. Analysis
also found evidence of developing patterns in the transcripts of two of the
novice participants. Novices Q and R referred to an ‘instability’ hypothesis
early in the assessment but then continued with an analytical reasoning
process with multiple competing hypotheses. Developing hypotheses from
the transcripts of Novices Q and R have been included for visual review and
subsequent discussion in Chapter 5.

135
Novice R hypotheses:
Time stop 2:26 A previous injury and something reasonably serious and
the inability to walk for two days is indicative of
something fairly strong or fairly strong pain. Um so there
could be some previous instability or damage that’s
been re-aggravated
Time stop 3:34 the amount of referral is something that’s you know, a
fair degree of instability or something’s going on there.
She’s getting a fair bit of referral, so start to you know,
lean towards a, a more serious sort of thing
Time stop 3:34 Could be instability. Could be disc bulge, could be a few
things
Time stop 8:02 Pattern really wasn’t that discy
Time stop 10:55 that referral could be due to um sciatic impingement …
that was something that I was juggling. Um, but also
there could be you know she might have other things
going on. Um facet joint irritations or general instability,
global instability

The problem solving of Novice R commenced with what seemed to be

recognition of an instability pattern labelling the past history and strong pain
as the case features. However progression to multiple competing hypotheses
gave greater overall weight to evidence of a HDR process. The development
of the instability hypothesis is viewed as forwards movement within an
analytical process. Analytical evidence from the Novice R transcript was
provided in the following statements:
Novice R: I hadn’t really at this staged locked myself into any sort of
hypothesis. I was really just sort of pulling it all, just making a
stew, just getting all that information in there and some, seeing
what I thought in the end (Time stop 6:14)
Novice R: when you look at it, you know it’s (disc hypothesis) still going to
be one of you’re balls that you’re juggling, but the 24 hour
pattern doesn’t really fit, you know (Time stop 8:02)

136
Novice Q hypotheses:
Time stop 4:30 I’m starting to think disc, discogenic sort of given that it’s
radiating up into, into the glut’s. Um -- also you know I’m
looking at possibly from adverse neural tension
involvement as well. Um, but it, yeah likely you know
given the time frame and that I, I’d certainly start to think
discogenic, um overall and, and more, more likely like a
chronic ah a lumbopelvic instability
Time stop 6:19 It’s reasonably consistent um with what I’m thinking, with
the that lumbopelvic instability
Time stop 6:56 again it’s sort of, there is a, um there is some um some
support again for that, that lumbopelvic instability. But
then again it could be discogenic as well
Time stop 7:19 Um sort of confirmed it. There’s certainly an extension to
the aggravation um -- sitting with good posture is
obviously or reasonable posture is quite difficult for her
which would suggest instability. Um sort of takes away
that disc … I guess I mean there could be a more ah a
lesion anteriorly in the disc perhaps
Time stop 8:40 I think it does tend to support um (prior understanding of
instability). I guess the other, other structures particularly
with extension and that, you have to consider is a SIJ. But
not very likely given the area of pain um it’s my
understanding that SIJs rarely go beyond that you know,
that far around the groin and things like that
Time stop 10:10 tends to you know um support more that lumbopelvic
instability … there’s no, not a really inflammatory
response or anything like that with it. Um -- which there
tends to be at times you know with particularly with discs
you know they report stiff and sore in the mornings and
that sort of thing

137
Time stop 10:58 I’m still leaning towards um that, that instability um you
know lack of, lack of stability through that area
Time stop 15:25 I guess really there’s two possibilities. But I, that sort of sit
in my mind. Um that still that same, you know that
lumbopelvic instability in that area, um poor activation of
transversus, it certainly would be interested to see her
ability to activate um through there and see if there was
any alteration in her symptoms. Um, the other, other
possibility is ah like a stenosis compression of a nerve
root

Gradual development of the multiple hypotheses in the Novice Q transcript

indicates a predominant analytical process. However, some evidence of
clinical patterns within a hypothetico-deductive process is apparent. The
participant’s references to the use of patterns was not indicative of clear
recognition of a known pattern but may be considered as evidence of a
developing pattern triggered within an analytical process. The evidence for a
developing pattern was observed in the following statements:
Novice Q Looking at patterns um to see if there’s any. Um given that it
feels pretty good in the morning it’s less likely to be disc. Um and
tends to you know um support more that lumbopelvic instability
… there’s no, not a really inflammatory response or anything like
that with it. Um -- which there tends to be at times you know with
particularly with discs you know they report stiff and sore in the
mornings and that sort of thing (Time stop 10:10)
Novice Q it gives me an idea, I guess I’m trying to establish ah a pattern
recognition, you know, relating the pattern to, to the possible
problem (Time stop 10:10)
Forwards reasoning movement was evident at one stage in the form of
confirmatory questioning:
Novice Q I was just sort of you know chucking in a few questions (cough &
sneeze) to sort of negate structures perhaps, and it sort of leads

138
 me towards more, more that chronic um instability (Time stop
8:00)

4.6 PARTICIPANT HYPOTHESES

Participants were encouraged to verbalise their understanding of the case

during the stimulated recall interview. As outlined in the ‘orientation to the
interview’ sheet (Appendix 10) read by participants prior to the interview, the
term hypothesis was used to refer to any understanding or explanation of the
case. This was particularly emphasised to be in any form or any way that
suited each individual clinician. Analysis included coding and sub-coding
within the hypothesis node which enabled an overall view of the extent of
hypotheses formed during the clinical reasoning task.

A summary of each participant’s final hypothesis has been listed in Table 4.9.
The data associated with each summary has been provided as evidence
behind the student researcher’s interpretation of final hypothesis, along with
the breakdown of hypothesis categories (Appendix 15). This section provides
the analysis of participant’s hypothesis category use within the final stated
hypothesis (section 4.6.1) and during the entire participant interview (section
4.6.2).

4.6.1 Final hypothesis category utilisation

Final hypotheses described by the participants were also analysed with

respect to the hypothesis categories outlined in section 2.3.1. This involved
each final hypothesis being labelled with any number corresponding with the
category observed within the data. Table 4.15 presents the final hypotheses
for each participant by the type of categories used.

139
Table 4.15 Final hypothesis extent of category use
Participant Categories Participant Categories
Expert A 4,5 Novice K 4,5
Expert B 4,6 Novice L 4
Expert C 4,5 Novice M 5
Expert D 2,4,5 Novice N 5
Expert E 4,5 Novice O 4
Expert G 3,4,5 Novice P 5
Expert H 4,5,6,7 Novice Q 4,5
Expert I 3,4,5,7 Novice R 4,5
Expert J 4,5,6 Novice S 4

The total range of categories used in expert final hypotheses included 2, 3, 4,

5, 6 and 7; whereas comparative use by novices included only categories 4
and 5. Four of the experts utilised only category 4 and 5 or 4 and 6 in their
final hypothesis. The remaining five experts utilised a combination of three or
four categories in their final understanding. One expert did not state a final
understanding. The data from Table 4.15 have been used to display the
percentage of participants per group using each hypothesis category in the
final stated understanding of the case (Figure 4.3).

140
 100

90

80
Percentage of participants
70

60
Novice
50 Expert

40

30

20

10

0
n

s
s

s
ns
s

t
en
ce
or

si
m
io

or

tio

no
s
at

m
ct

ct
ur
ni
ip

ca

at
fa

fa

og
so
ha
c

tre
di
al

g
rti

Pr
ec

in
ra

in
ci
pa

nd
ut
m
so

a
tu

r
rib

ta
nt
&

c
al
o

ru
ch

co
ic
ity

en
nt
st
og
y

Co
iv

em
d
ps

an
ol
t
Ac

an

ag
bi
d

ns
an

an
ts
th

io
en

M
e

Pa

ut
iv

irm
ct

ca
e

pa

e
sp

Pr
m
r
pe

li
ca
t’s

i
ys
n
tie

Ph
Pa

Figure 4.3 Final hypothesis category use by experts and novices

4.6.2 Overall hypothesis category utilisation

The extent and frequency of hypothesis formation with respect to the eight
reported hypothesis categories (Jones & Rivett, 2004) outlined in Table 2.2
during a clinical assessment has not been previously reported. Analysis of
the frequency of hypothesis category use at any stage during the interview
was assessed to provide information pertaining to the extent of hypothesis
development by physiotherapy clinicians. This process notably differed from
the final hypothesis category use in Table 4.15 as it involved a review of any
hypothesis observed in the transcript data. Examples of each hypothesis
category observed in the transcripts have been provided in section 4.3.1.

141
The extent of use was assessed via review of the hypothesis coding and sub-
coding obtained during qualitative analysis. The overall extent of hypothesis
category use has been presented in Table 4.16, including the additional ‘non-
specific’ category for hypothesis types not indicated in the literature. These
data present the number of participants in each group and overall that utilised
each hypothesis category at any stage during their reasoning. The frequency
of use by each participant and the depth of content within each category
were not included in this analysis.

Table 4.16 Extent of overall hypothesis category use

Hypothesis Category Experts % Novices % Overall %
n=10 n=9 N=19
1 Activity / participation 6 60 4 44 10 53

2 Patient’s perspective / psychosocial 8 80 2 22 10 53

3 Pathobiological mechanisms 7 70 5 56 12 63
4 Physical impairments / structural sources 10 100 9 100 19 100
5 Contributing features 10 100 9 100 19 100
6 Precautions and contraindications 6 60 7 78 13 68
7 Management and treatment 9 90 5 56 14 74
8 Prognosis 6 60 3 33 9 47
9 Non-specific 7 70 5 56 12 63

The percentage use of each hypothesis category by the 19 participants

reflects the extent of hypothesis formation within a small sample of
physiotherapists assessing the same case. The results of hypothesis
categories use by all participants demonstrated a predominance of the
impairment / structure / source and contributing features categories which
were utilised by all participants during the problem solving task. Management
(74%) and precautions / contraindications (68%) were next in frequency of
use, followed by pathobiological mechanisms (63%). Prognosis (47%),
activity / participation (53%) and patient’s perspective / psychosocial (53%)
were the least frequently used hypotheses. Hypotheses that did not fit clearly

142
into any of the eight reported categories and were placed in the non-specific
group were used by 63% of all participants.

Table 4.16 also allows for comparison of each hypothesis category relative to
participant group. The stand out difference between groups occurred within
category 2 (patient’s perspective / psychosocial factors) where 80% of
experts described their understanding within this category as compared with
only 22% of novices. The percentage of participants per group (Table 4.16)
using each hypothesis category at any stage during problem solving is
displayed in Figure 4.4.

100

90

80
Percentage of participants

70

60 Novice
Expert
50

40

30

20

10

0
s
n

rs

s
ns
s

t
en
ce
sm

si
or
io

to

tio

no
at

m
ct
ur
ac

ni
ip

ca

at
fa

og
so
ha
lf
c

tre
di
g
rti

Pr
ia

ec

al

in

n
oc
pa

nd
ur

ai
ut
lm

r
os

ct

ta
nt
&

tri
ru
ch

co
ic
ity

en
n
st
g
y

Co
iv

em
d
lo
ps

an
t

io
Ac

an

ag
d

ob

ns
an

ts

an
th

tio
en

M
e

Pa
tiv

au
irm
c

ec
pe

pa

Pr
rs

m
pe

i
c al
t’s

si
n

y
tie

Ph
Pa

Figure 4.4 Overall hypothesis category use by experts and novices

143
The existence of any significant difference between groups was analysed
individually for each hypothesis category. Fisher’s exact test was utilised
given the small numbers in the groups and the results demonstrated that only
the second hypothesis category of ‘patient’s perspective / psychosocial
factors’ was found to be significantly different between the groups using a
significance level of 0.05 (Table 4.17). No calculation was possible for
categories 4 and 5 given no differences in frequency occurred between
groups.

Table 4.17 Group comparisons for the use of each hypothesis category
(using Fisher’s exact test)
Hypothesis Experts Novices p value
category n=10 n=9
1 6 4 0.656
2 8 2 0.023*
3 7 5 0.65
4 10 9 -
5 10 9 -
6 6 7 0.628
7 9 5 0.141
8 6 3 0.37
* ≤ 0.05 significance level

Table 4.18 highlights the extent of use of hypothesis categories by each

novice and their overall use by novices as a group. The overall median
number of categories used by novices was 5 (62.5%) and mean was 4.9
(61.1%).

144
Table 4.18 Novice use of hypothesis categories
Participant 1 2 3 4 5 6 7 8 Total %

Novice P 1 1 1 1 1 1 0 1 7 87.5
Novice Q 0 0 1 1 1 1 1 1 6 75
Novice M 0 0 1 1 1 1 1 0 5 62.5
Novice O 1 1 0 1 1 0 1 0 5 62.5
Novice L 0 0 1 1 1 1 0 1 5 62.5
Novice N 1 0 1 1 1 1 0 0 5 62.5
Novice S 1 0 0 1 1 1 0 0 4 50
Novice R 0 0 0 1 1 1 1 0 4 50
Novice K 0 0 0 1 1 0 1 0 3 37.5

Median 5 62.5

Mean 4.9 61.1

The experts were notably higher than novices with their use of hypothesis
categories within the single case. The overall median number of categories
used by experts was 6 (75%) and the mean was 6.2 (77.5%). Table 4.19
presents the expert data.

Table 4.19 Expert use of hypothesis categories

Participant 1 2 3 4 5 6 7 8 Total %

Expert G 1 1 1 1 1 1 1 1 8 100
Expert A 1 1 1 1 1 1 1 1 8 100
Expert B 1 1 1 1 1 1 1 0 7 87.5
Expert I 1 1 1 1 1 1 1 0 7 87.5
Expert F 1 1 0 1 1 0 1 1 6 75
Expert C 0 1 1 1 1 0 1 1 6 75
Expert H 1 0 1 1 1 1 1 0 6 75
Expert D 0 1 1 1 1 0 1 1 6 75
Expert J 0 0 0 1 1 1 1 1 5 62.5
Expert E 0 1 0 1 1 0 0 0 3 37.5

Median 6 75

Mean 6.2 77.5

145
4.7 ADDITIONAL FINDINGS

The primary study aims related to identifying PR and its relationship to

accuracy and efficiency, however the qualitative nature of data collection and
analysis provided additional findings worthy of reporting and discussion.
These results have been obtained from the thematic analysis process that
has been outlined in section 4.3.4 and Table 4.4 and are summarised into
five areas:
 Relative importance of data
 Concurrent integration of data
 Awareness of errors
 Predictive reasoning
 Person centred approach to assessment.

4.7.1 Relative importance of data

Musculoskeletal physiotherapy can be viewed as having two main sources of

information; data from questioning a client and that obtained from a physical
assessment. This study was based around the subjective data obtained from
questioning the simulated client, given that no physical examination was
conducted during this investigation. It was noted during free coding and
subsequent thematic analysis that some participants verbally commented on
the importance of a specific data source. These comments were considered
to fall into three areas:
 Emphasis on patient history information during problem solving
 Emphasis on physical examination information during problem solving
 Emphasis on integration of all data.

In total seven participant transcripts were considered by the student

researcher to have data relating to this additional finding, including five of ten
experts and two of nine novices. The data presented may have been directly

146
stated by the participant during the interview or the student researcher may
have inferred the finding from the transcript data.

Examples of statements inferred by the student researcher relating to the

relative importance of data:
Expert F: by this stage I normally have a very clear picture of what I think’s
wrong with someone whether it right or wrong
Expert H: P: Um, I guess I was trying to, to make sure that I wasn’t missing
a disc, a disc problem”
R: And is that something you would now be able to put aside?
P: Not, not entirely until I’d done the physical examination so if I
found things like SLR restricted flexion and obviously
neurological signs, I’d probably put it back. Yeah but at the
moment I’m not thinking that
Expert G: I was a bit surprised about given I thought there was, there is a
neural component to that, but again I, you know, I’ve just, I’ve
gathered that information, I’m going to store it but I will then just I
would probably probe it more in the, in the objective with her
(referring to further questioning at an appropriate stage during
the physical examination)

Examples of direct comment made by participants relating to the relative

importance of data:
Expert E: I would place more (emphasis) on the objective … Because I find
that’s more reliable
Expert J: P: the objective assessment I do, is pretty standard … I mean I
just do everything and then at the end, decide what, what the
diagnosis was
R: Do you find you get more information in terms of
understanding what might be going on from your subjective, or
more from your, your physical testing?
P: Subjective
R: More from your subjective?

147
 P: Mm.
R: And then you’ve just got a standard -- physical testing that
you do to confirm or further understand?
P: Yeah, yeah. Yeah I think I do pretty well a full assessment on
everyone.
Novice S: I don’t like to um to rush into decision-making at the end of my
subjective. Um so I, I try to, to piece together the information that
I’ve, I’ve collected and, and form an objective around that … I
think more often than not um I don’t, I have an answer to a
particular structure that might be causing pain at the end of the
subjective. Um I, I’m more often have an answer at the end of
my objective
Novice R: from pretty much the majority of the subjective I was just really
pooling information and formulating ideas and just waiting, I
usually just wait till after the objective before I, and even then I
still might have three different diagnoses. So yeah at this stage
it’s all pretty in the formative stage

The quality of the data are unable to lead to any specific conclusions relating
to a preference of one part of the clinical examination over another with
experts or novices, but rather highlights variability in each clinician’s
weighting of clinical findings. This area of findings was not requested of
participants but noted during coding of the interview transcripts.

4.7.2 Concurrent integration of data

Metacognition relates to the awareness of one’s thinking about data

collection, reasoning, hypotheses or knowledge at the time of cognition
(Jones & Rivett, 2004). It is considered an important expert characteristic and
supports the development of clinical reasoning skill via reflective practice.
The results of the present study highlight that several participants including
two of the experts adopted a delayed integration of data process, arguably at
odds with metacognition. Examples of the two experts include:

148
Expert J: I usually take the history and then think what -- I just go through
the sheet I’ve got, and then at the end I really start thinking,
putting it together. I usually get all the facts before I start putting
it together … the objective assessment I do, is pretty standard …
I mean I just do everything and then at the end, decide what,
what the diagnosis was
Expert A: I think I tend to data gather and stash it in a big heap and use it
later … I’m pretty much in automatic mode asking questions … I
don’t try and bundle it too much at this stage, although I guess I
must do in my head.
Expert A stated their delayed integration approach but makes comment
somewhat to the contrary in the statement “I don’t try and bundle it too much
at this stage, although I guess I must do in my head”. This could be
interpreted as reflective of an automated approach without good awareness
of the problem solving processes employed.

Both experts A and J demonstrated an ability to reason individual pieces of

information / data along the way but were not found to effectively integrate
clinical data during the assessment. The difference noted between these
experts was that expert J was focused on the search for a structural
pathology based diagnosis, whereas expert A stated that the clinical
diagnosis was the focus and not the pathology.

The novice participants found to use a delayed integration approach

included:
Novice R: at the time I was really just running through the process. I wasn’t
really nutting it out that, that much in detail
Novice L: I’m thinking at the time I was just sort of collecting information
and wasn’t really um contemplating what that was
Novice M: I’m more concentrating on collecting the data as accurately as I
can
Novice L stated this delayed integration related to one aspect of clinical data
collection. This quotation was not necessarily indicative of their approach to

149
the entire assessment. Novice M tended to revert to a delayed integration
approach later in the assessment, possibly when the clinical data was not
able to be interpreted as a clear case understanding. These results are not
unexpected for a novice physiotherapist with little exposure to such a case,
however only three of the nine novice participants referred to a delayed
integration approach. The remainder were clearly integrating the data during
the assessment even when their level of knowledge and clinical experience
arguably did not allow for effective assimilation.

The effect of delayed integration of data on assessment time was analysed

(Table 4.20). The timing data compares total length of the client assessments
for those participants using a delayed integration approach versus those
concurrently integrating clinical data.

150
Table 4.20 Total assessment time (minutes:seconds) relative to delayed
versus concurrent integration of data (all participants)
Number Range Median Mean
Delayed integration 5 10:08-18:50 13:45 13:44
(participants coded delayed
integration)
Concurrent integration 14 8:12-22:08 16:40 15:59
(participants not coded
delayed integration)

The five participants having some evidence of delayed integration of clinical

data were found to have a lower mean and median time for the total
assessment when compared with all other participants. All five of the
participants coded ‘delayed integration’ were not found to use PR and were
subsequently compared with all other participants using HDR (Table 4.21).

Table 4.21 Total assessment time (minutes:seconds) of participants using

hypothetico-deductive reasoning relative to delayed versus
concurrent integration of data (excluding participants using
pattern recognition)
Number Range Median Mean
Delayed integration 5 10:08-18:50 13:45 13:44
(participants coded delayed
integration and using
hypothetico-deductive
reasoning)
Concurrent integration 9 8:12-22:08 15:20 15:14
(participants not coded
delayed integration and
using hypothetico-
deductive reasoning)

The data were not analysed statistically due to the possibility that the
presence of this code did not reflect the participant’s approach to the whole
assessment. Section 5.5.2 discusses this limitation further.

151
4.7.3 Awareness of errors

The potential for error is an important element when considering clinical

reasoning process. The non-analytical strategy of PR has been described in
the literature as being associated with both error and accuracy in clinical
practice. The results show three experts made reference to a willingness to
remain open to other possibilities during the interview, arguably minimising
one potential source of error (bias):
Expert H: I do try and keep quite an open mind as far down the track as I
can because I know that you can get quite influenced and then,
and then find that it’s not really the case, so I try really to make
judgment ah, fairly far down the track
Expert I: I generally tend to keep it as open as I can with the history so
that people could volunteer as much information
Expert D: That would be my working hypothesis that I would be wanting to
test with my examination … but I would be completely prepared
to find something completely different

Experts D and I had attained a central hypothesis that was correct prior to
these statements, but both indicate they were still capable of discounting this
hypothesis if contradictory data were found.

4.7.4 Predictive reasoning

A predictive strategy during questioning was anticipated by the student

researcher to be potentially associated with a forwards reasoning process.
Evidence of this strategy was found in the transcripts of three experts and
two novice participants. Of these five participants, only one expert utilised PR
based on the results in section 4.4. Examples of predictive reasoning
quotations included:
Expert G: that probably wasn’t what I was expecting her to say. I probably
would have expected her to say when I slouch its worse and
when I sit up straight its better

152
Expert D: If my hypothesis of it being some sort of either structural
instability or like a dynamic instability or spondylolisthesis is
correct maybe she’ll be able to tell me that when she’s running
and then stops suddenly bang it will grab her
Novice Q: I was just sort of you know chucking in a few questions to sort of
negate structures perhaps, and it sort of leads me towards more,
more that chronic um instability
Novice N: based on the aggravating factors and the area of the pain I’d
expect it to be sore in to flexion … Well, in, in a way that if, if she
flexed all the way down to her toes I’d be surprised … Um, it
wouldn’t fit what I was going down

A unique observation of predictive reasoning involved Expert F who

described a game like approach to predicting client responses:
Expert F: P: it hasn’t fitted into what I thought. And I mean I guess as you,
I mean that’s what makes your practice interesting. I mean I do
play games with myself at guessing what I think is the problem
R: And you’re playing those sort of games mentally?
P: Oh yeah, yeah, yeah … I’ll try and guess what their answers
are going to be before they give me answers. Whether I’m right
or wrong
R: Yeah so you ask a question and try and have a prediction in
mind?
P: Yeah and then I see whether I’m right or wrong

Predictive reasoning was therefore a strategy used by some physiotherapists

but not found to be associated with PR.

4.7.5 Person centred approach to assessment

The current study method was targeted towards identifying a diagnostic

reasoning process and involved an actor, which may have biased
participants in their behaviour accordingly. Despite this some participants

153
attempted to understand the client as a person. Eight of the ten expert
transcripts and four of the nine novice transcripts were found to contain some
evidence of a ‘person centred’ approach to the assessment (Table 4.4). This
theme was based on the student researcher’s notes during analysis of the
hypothesis sub-codes such as category 1 (activity and participation) and 2
(patient’s perspective / psychosocial factors), along with the ‘management’
tree node and the ‘goal setting’ free node. Transcript examples supporting a
person centred approach include:
Expert G: looking at a goal of hers that you know she’s worried she’s got
another clinical placement
Expert I: she’s still got a significant aggravating component to it and that,
that may be something what we needed to address as far as um
-- time off when she needed to. She may need time off uni. She
may need to look into special consideration and social factors
that might need to be addressed in relation to her pain
Novice N: She hasn’t lost any time off work although she’s stopped playing
sport and it does hamper her ability to perform the normal things
that she needs to do
Novice O: I was just trying to get a, a kind of idea of where she’s at like
with, cause it’s been going on for so long I don’t exactly know her
age but um she, her ah pain behaviours and how she deals with
the pain like she says rest is the best thing but she still, she’s still
playing netball despite the pain and so just trying to get an idea
of how she copes with the pain and that sort of thing

These quotations provide evidence that the participants were able to think
about the person within the problem solving process, which therefore adds
support to the realism of the clinical encounter and suggests the study design
had minimal effect on their reasoning behaviours.

154
4.8 STUDY DESIGN RELATED RESULTS

Any research relies on sound methodology for the validity of its overall
conclusions. This is particularly the case with clinical reasoning research due
to the complexity of investigating cognitive processes. Several aspects of the
current study method that could impact on the outcomes were monitored
during data collection and subsequently analysed. The results of the study
design on participant behaviour are presented hereafter.

4.8.1 Case simulation data

A potential criticism of simulated case research involves realism and

accuracy of case portrayal. The videorecording of the client assessment was
primarily utilised for stimulated retrospective recall but was also subsequently
reviewed and analysed for accuracy of the case portrayal. A simulated case
response checklist (Appendix 4) was developed incorporating fifty-two
response areas which allowed the accuracy of the actor’s responses
compared with the case data to be determined. This was undertaken
immediately following data collection for the first four participants and
feedback provided to the actor. Continued data collection for all participants
occurred in relation to case accuracy and statistical analysis followed the
completion of participant recruitment.

The data obtained from the simulated client response checklist was
transferred to nominal data and analysed via proportions averaged over the
fifty-two response areas. A summary of the response data is provided in
Table 4.22. For each question response area the number of occasions it was
requested could be viewed in addition to the number of times it was
answered accurately and inaccurately. This resulted in a proportion of correct
responses out of the total number of times requested for each response area.
This basic method of analysing data accounted for the fact that not all
response areas were requested by every participant. The results found that
only two of the fifty two response areas had a proportion of below 1,

155
indicating fifty questions were answered correctly on 100% of occasions. The
average proportion of all fifty-two response areas was 0.99 equating to 99%
accuracy of overall responses to participant questions.

The breakdown of fifty-two response areas found the question on

‘unsteadiness of gait / giving way of the legs’ was answered inaccurately
during the first two participant assessments. Feedback provided to the actor
led to accurate responses in this area of questioning for the remaining
participants (if requested). The simulated patient’s age was the only other
area of questioning to be answered incorrectly on one occasion when the
date of birth was requested rather than current age.

Accurate portrayal of the case by the actor was a critical part in achieving
valid and meaningful results. The participants were aware of the case being
portrayed by an actor via the study information statement (Appendix 7) as
ethically required. They were subsequently requested to comment on their
overall experience compared with a real clinical situation, including the
realism of the case and simulated client. Eight of the nineteen participants
directly commented on the actor, while others remarked on the study setup
as a whole without specifically making comment on the actor.

156
Table 4.22 Simulated client response data
Experts Novices Total Incorrect
Response area requested requested (out of 19) responses
Location of pain types (Pain A) 10 9 19 0
Location of pain types (Pain B) 10 9 19 0
Association of pains A & B 10 9 19 0
Location of pain types (Pain C) 10 9 19 0
Location of pain types (Pain D) 10 9 19 0
Location of pain types (Pain E) 10 9 19 0
Association of pains C, D & E 10 9 19 0
Severity of pain types (Pain A) 6 8 14 0
Severity of pain types (Pain B) 6 7 13 0
Severity of pain types (Pain C) 6 5 11 0
Severity of pain types (Pain D) 6 5 11 0
Severity of pain types (Pain E) 6 5 11 0
Description of pain types (Pain A) 9 7 16 0
Description of pain types (Pain B) 9 7 16 0
Description of pain types (Pain C) 9 7 16 0
Description of pain types (Pain D) 9 7 16 0
Description of pain types (Pain E) 8 7 15 0
Constancy of pain (Pain A) 10 9 19 0
Constancy of pain (Pain B) 10 9 19 0
Constancy of pain (Pain C) 10 9 19 0
Constancy of pain (Pain D) 10 9 19 0
Constancy of pain (Pain E) 9 8 17 0
History of current episode 10 9 19 0
Past history 10 9 19 0
Mechanism of injury 10 8 18 0
Primary aggravating activities 10 9 19 0
Standing tolerance 8 8 16 0
Walking tolerance 6 2 8 0
Sitting tolerance 9 6 15 0
Primary easing factors 10 9 19 0
Previous physiotherapy treatment 9 5 14 0
Other previous treatment 10 9 19 0
Morning pain / stiffness 10 8 18 0
Night pain / ability to sleep 10 9 19 0
Pain behaviour through day 10 9 19 0
Primary patient goals 8 7 15 0
Activity – netball participation 10 7 17 0
Anterior knee pain 6 3 9 0
Unsteadiness / giving way of legs 4 2 6 2
General health 10 9 19 0
Paraesthesia / numbness 10 9 19 0
Cough / sneeze 7 2 9 0
X-rays 10 9 19 0
Prior surgery 9 6 15 0
Weight loss 9 7 16 0
Medications 10 8 18 0
Investigations 10 9 19 0
Social history 6 1 7 0
Age of patient 8 6 14 1
Cord / cauda equina questions 8 6 14 0
Current employment 3 4 7 0
Nursing student – full time 10 9 19 0

157
The participant comments provide a qualitative view of the realism of the
case presentation:
Expert I: She was really good and certainly not enough things that
changed um the way you were thinking.
Expert D: She’s realistic to the point where I’m, I started wondering I
wonder if she’s actually got this problem.
Novice R: It seemed like an actual patient.
Novice M: Yeah, I forgot I was not seeing a real person.

The influence of the simulated client on participant behaviour was

commented on during a few participant interviews. These quotations highlight
a brief focus on the study’s use of client simulation for these two participants:
Novice K: R: You mentioned halfway through one of the questions you
asked you got a response that you thought was um an
impromptu response as opposed to a real response.
P: Um from the patient? Ah yeah that was I don’t think though
that that had anything to do with the camera or the audiotape. I
think that was ah something which ah I guess I, it’s quite a, I
guess I’d say just from my limited experience that would be quite
a funny um ah symptom to get
R: So were you thinking about this person on the way being an
actor in that role the whole time or was it just at that sort of
moment.
P: No just for 15 seconds that she said it was an achy, crampy
(pain) but apart from that I was just thinking that she was a
patient.
R: So apart from that you thought she was fairly realistic?
P: Yeah I thought she was good. Did she actually have back
pain before?
Expert J: I thought she was very good. Um at first I was thinking oh she’s
an actor and she’s been told what to say. But she knew all the
ans, like there wasn’t any question that she, except when I said -
- ah with the leg pain was it superficial or deep she sort of didn’t

158
 really know -- And other than that I thought she knew it all well
and then I started to really believe that she had this problem.
When I asked her about what treatment – you know when she
was 11 and she went to the chir, going to the chiropractor when
he cracked her back. Oh when I come home, I’m at uni but when
I come home in the holidays I still go and see the chiropractor,
every two or three months and he cracks it. I started to think then
that she really had the problem.

4.8.2 Study context influences

The main location of the study and placement of equipment attempted to

minimise any effect on the behaviour of the participant during the study.
However the potential effect of the context on participant behaviour was
unavoidable. Consequently participant perception on the impact of study
context on participant behaviour was surveyed at the end of the interview.
The final questions outlined in the interview protocol (Appendix 11) were in
relation to:
1. Did the location of the assessment influence your assessment compared
to normal?
2. Did the presence of the video camera influence your assessment
compared to normal?
3. Was your assessment conducted more thoroughly than normal?

The majority of participants did not think the context or study equipment
altered the process of conducting a subjective assessment in this case:
Expert A: R: what you did today with this particular case… do you think
that was fairly standard of what you would have done for the
subjective with the next person that walks in the door?
P: Yeah pretty standard.
R: So the video and the audio equipment didn’t interfere too
much with what you did?
P: Oh no, not at all.

159
Expert D: R: This environment do you think it changed how you went
about that assessment, the video camera?
P: No, no I think probably that’s pretty much as I would
Expert J: R: did you think the location of the assessment, the video
camera influenced the way you went about your subjective
assessment?
P: No not at all.
R: And do you think you conducted your assessment any more
or less thoroughly than normal?
P: No that’s standard.
Expert E: R: Did you think the, the video camera made you do the
assessment differently than what you would do normally?
P: No, no.
Novice P: R: Did the location of this assessment … influence the way you
went about your assessment?
P: Um being a subjective ah, not greatly.
R: Did the video camera um and the audio recorder influence?
P: No it didn’t bother me at all.
Novice K: R: In terms of the set up here, do you think the camera or just
being audio recorded actually changed or altered or influenced
the way you went about your subjective?
P: Ah not really. No.
R: You mentioned halfway through one of the questions you
asked you got a response that you thought was um an
impromptu response as opposed to a real response.
P: Um from the patient? Ah yeah that was I don’t think though
that that had anything to do with the camera or the audiotape. I
think that was ah something which ah I guess I, it’s quite a, I
guess I’d say just from my limited experience that would be quite
a funny um ah symptom to get.
Novice O: R: Did the recording equipment -- Or the environment influence
the way you went about it?
P: No, no didn’t even notice.

160
 R: Did you think that was reasonably -- normal to what you
would normally do.
P: Yeah absolutely. Yeah, yeah, yep. I didn’t, I didn’t feel like it
was invasive or anything like that, so.

The occasions where contextual effects were noted by expert participants

have been reported to allow the reader to make judgement on this
component of the study. All comments were considered by the student
researcher to indicate minimal overall influence on the results of the study:
Expert H: R: do you think this location did in the end influence the way you
went about your assessment?
P: No probably not.
R: The video camera or the audio recorder didn’t --
P: No, I was able to forget about that -- I started, I started to find
myself thinking halfway through -- about the process but
managed to sort of but then I started losing the plot a bit.
R: So as an assessment it’s fairly standard as to what you would
have -- generally done in the clinic.
P: Yes, it’s, it’s not different I would always do that that sort of
way.
Expert B: P: I just think sometimes when I’m watching the video I probably
wouldn’t go into that amount of detail the questioning. I think I’d
probably make those jumps um ah you know clinically I probably
wouldn’t go through as much detail. I think that’s probably one of
the um -- the fact that it ‘s sort of like this video and that type of
thing.
Expert G: R: so the location of the assessment, do you think that influenced
how you went about your assessment today -- Not being in your
normal environment?
P: Um -- no I think that probably the only thing that, about the
environment might have been the presence of the, the video
camera.

161
 R: And so did that affect you much in, in the end in terms of how
you went about it, or changed anything?
P: No, no I don’t think so.
Novice S: R: did the video affect do you think, the way you assessed the
patient in this case. The fact that you have a video in the
background or being audio recorded?
P: No
R: Being in this environment as opposed to in a clinic with a
plinth beside you, do you think that changed the way you went
about things at all?
P: Not the way that I went about things in any way. Um, it was, it
did feel a bit unfamiliar though, so I don’t know whether that
would have affected me. I don’t feel like it has.

Participant comments regarding the research process highlighted an

awareness of the context mid-way through the client assessment, however
this was generally brief and did not alter the overall flow of the assessment
compared with a normal client history.

162
CHAPTER 5. DISCUSSION

The present study used a carefully designed high fidelity case study method
as outlined in Chapter 3. It is the first study of its type in physiotherapy that
has assessed diagnostic accuracy relative to reasoning process. In
particular, the study findings add to the physiotherapy clinical reasoning
literature with respect to our understanding of PR. The findings also
potentially impact on our understanding of pattern development from the
perspectives of both accuracy and education.

This chapter specifically addresses several key discussion points based on

the aims of the reported study:
 The presence of PR as a clinical reasoning process in musculoskeletal
physiotherapy
 Evidence for the diagnostic accuracy of PR, and
 Observations and limitations relating to the efficiency of PR.

Following the discussion relating to the primary study aims, this chapter also
addresses:
 The ‘makeup’ of a clinical pattern as the basis of recognition
 Significant case feature use in PR
 Observations of hypothesis category use in problem solving with respect
to differences between experts and novices
 Relevant comparisons between novice and expert groups
 Additional findings relating to different approaches to reasoning observed
during the study, and
 Limitations of the study based around the retrospective recall
methodology and context specificity.

163
5.1 NON-ANALYTICAL REASONING

The research literature supporting PR is largely based on studies of the

medical profession. Comparison studies exist between medicine and
physiotherapy in analytical reasoning processes (Payton, 1985), however
non-analytical reasoning has not been compared in the same way. PR has
been reported in qualitative physiotherapy research in two studies involving
high fidelity real cases (Doody & McAteer, 2002; Noll et al, 2001), however
these have not allowed for assessment of diagnostic accuracy. A more
recent medical study assessing the overall benefit of PR including outcome
was based on low fidelity paper case methods (Coderre et al, 2003). In
neither profession has a high fidelity case been utilised to undertake a
replicable study investigating the accuracy of non-analytical diagnostic
reasoning strategy. High fidelity case study methods allow participants to
obtain the clinical data that are suited to the case and their reasoning style
rather than be guided by what is presented to them.

A concern relating to the PR literature involves the varied reports of its

makeup. This has been extensively discussed in section 2.5.1, resulting in
an amalgamation of the commonly reported elements of PR. Subsequently
criteria for identifying PR were identified (Table 3.3), derived from published
authoritative commentaries and research articles. These criteria then
became the basis for the PR identification tools (Appendix 13) and are
worthy of further discussion.

The essential requirement in identifying PR during this study (and separating

it from an analytical strategy) was evidence of a predominant or central
hypothesis based on significant case features. This is similar to the low
fidelity study conducted by Coderre et al (2003), where the basis for
identifying PR involved “a single diagnosis with only perfunctory attention to
the alternatives” (Coderre et al, 2003, p. 703). These researchers also
described PR as being based on salient cues but this was not a component
they used in its identification.

164
The additional three components used in identifying PR in the present study
involved the use of professional knowledge, reference to prior clinical
experience, and a stated management plan. It was essential that these
components had to relate to the central hypothesis that was based on
significant case features. Including these additional criteria was intended to
strengthen the divide between analytical and non-analytical strategies, and
not rely just on a single feature of PR such as detection from prior clinical
experience.

The study has provided strong evidence for the existence of PR in current
musculoskeletal physiotherapy practice. The fact that four of ten expert
participants were clearly found to be using PR using the stringent
identification criteria indicates that it exists. This was supported by the
credible interval calculations (section 4.4.2; Figure 4.2) having a lowest
margin above zero.

The nature of PR as a reported strategy of experienced clinicians and

dependent on past experience led to expectations that differences would
exist between the expert and novice groups. The novices were found to only
have one participant using a form of non-analytical reasoning, whereas the
expert group had four participants. The statistical analysis of differences
between experts and novices in the use of PR found that no significant
difference existed (section 4.4.2). However the lack of power resulting from
the small sample size was likely contributory to the lack of significance.
Furthermore, analysis of the five criteria indicative of PR found the single
novice participant using PR only scored three out of five (including the
central hypothesis, related case features and associated knowledge). Of the
four experts identified to have used PR, three utilised four criteria and one
employed all five criteria. It is plausible that the novice (who identified a
neuromuscular instability pattern) was using a broad form of PR but one that
wasn’t necessarily closely associated with prior clinical experience. The
pattern could have been one learned via academic education, for example.

165
Prior clinical experience as a non-essential component of identification of
PR was supported in that it was present in the transcripts of only two of the
four experts utilising PR. Although it underpins the phenomenon, it was
considered insufficient alone as a feature for identifying PR and was not
stated at times when PR clearly existed. The two experts who did not refer
to their experience of a similar case may have utilised recognition based on
a conglomeration of propositional and non-propositional knowledge, rather
than a single prior experience. This will be further discussed in relation to
categorisation in section 5.3.1.

The relatively broad inclusion criteria for the expert group facilitated the
generalisability of the results to standard physiotherapy practice of
musculoskeletal therapists. In essence, it was considered that the
physiotherapists comprising the expert group had clinically practiced for a
sufficient time and completed a recognised postgraduate musculoskeletal or
manual therapy qualification that enabled the development of experience
based patterns. The pattern used in this simulated case (i.e.
spondylolisthesis) was considered likely to be familiar to at least some
expert participants in this group based on the prevalence of the condition.
This assumption was supported by the PR results.

The primary research aim was to determine whether PR is utilised by expert

and novice clinical physiotherapists in the musculoskeletal field. The results
demonstrate that it was used by four experts and one novice participant.
These findings therefore support a similarity in diagnostic process models
between the medical and physiotherapy professions. Non-analytical
diagnostic reasoning that is highly dependent on case and context specificity
has been observed in both professions.

Although not statistically significant, the comparison between groups in this

study also tends to support the view that PR is a strategy increasingly
utilised with greater knowledge and relevant clinical experience. The

166
question of what type of and how much experience is required to develop
the use of this strategy was beyond the scope of this study. Arguably the
answer will lie with the integration of knowledge and experience relevant to
any specific condition.

5.1.1 Accuracy of pattern recognition

Clinical reasoning research in medicine has utilised case study methodology

to investigate PR accuracy (Coderre et al, 2003; Norman et al, 1992; Patel &
Groen, 1986). These studies however have utilised either visual cues in
radiology or low fidelity paper cases to achieve their respective outcomes
supporting its accuracy in diagnostic reasoning. The transferability of these
results to present physiotherapy clinical practice is questionable.

Careful consideration was given to the study design relating to high fidelity
case methods. The study was developed with the purpose of not only being
able to identify PR, but also to shed light on the question of its accuracy. A
key element in interpreting diagnostic accuracy in clinical reasoning
research lies with the level of confidence in the clinical data leading to the
actual case diagnosis. The specific diagnosis for the real case used as the
basis for the simulated case study was that of a lumbar spine
spondylolisthesis pathology (section 3.2.2), a condition with a substantial
level of research evidence in the published literature. Thus the assessment
of accuracy was limited to the diagnosis of a structural pathology alone.

The study found positive results for the use of PR in identifying the case
diagnosis of spondylolisthesis. The 2X2 table relating to the expert group
(Table 4.10) demonstrated that PR appeared to facilitate accuracy but its
use did not guarantee accuracy. Three of the four experts adopting a PR
strategy identified the correct pattern based on case data from no more than
the initial 36% of the total client assessment time, which only included the
client history.

167
Statistical analyses (section 4.4.3) compared the accuracy of those
participants using PR to those using analytical reasoning. It was found that
the expert participants using PR were significantly more likely to achieve an
accurate outcome when compared with the experts using analytical
reasoning strategies (p=0.033). When the same statistical test was repeated
with both participant groups combined, PR remained significantly more likely
to achieve an accurate outcome (p=0.01). These findings support an
association between accuracy and the PR reasoning strategy. The statistical
assumption of these analyses is that all participants are aware of the
diagnostic condition of spondylolisthesis. Given the level of experience and
education of the expert group it is highly likely that this would be the case. It
is also likely that the novice physiotherapists had been exposed to this
condition during their education, although they may not have encountered it
clinically.

The presence of the spondylolisthesis pattern in only three of the ten expert
interview transcripts could be considered surprising. However PR relates to
experience with the case at hand and may have been identified in more
instances had more data from different cases been collected with the expert
group. At least two of the experts stated within the interview that they
predominantly practised in a sole body region other than the lumbar spine.
Obtaining further qualitative data relating to the participant’s awareness and
prior experience with high grade spondylolisthesis pathologies could have
provided more insight relating to the presence and makeup of such a
pattern.

The possibility of errors relating to PR has been stated as a concern

regarding its use in physiotherapy. The finding that three of the four experts
using PR identified the precise pathology based on just findings from the
patient history reduces this concern in an expert group of clinicians. The
only evidence from the study relating to the inaccuracy of PR use by experts
involved expert C who identified a mechanical pelvic asymmetry. It is
possible that the pelvic hypothesis could have been a contributing factor to

168
the clinical symptoms but pelvic assessment data were not available from
the original case. The additional difficulty in including such a hypothesis in
the research investigation relates to the confirmation of pelvic dysfunctions.
The assessment of positional variations in pelvic position usually relies on
surface palpation during the physical examination which is known to have
poor reliability (Holmgren & Waling, 2008). Thus the only conclusion that
could be made relating to this expert’s understanding was that it was
incorrect with respect to identifying a spondylolisthesis pattern.

The analysis of the expert C interview found evidence of a deductive ‘fall

back’ strategy that led (albeit inefficiently, based on a time perspective) back
to the correct diagnostic pathology. It could be stated that the identified
pattern was incorrect based on the participant’s statement that a
spondylolisthesis would be separately considered, along with more serious
disc pathologies, if no positive response occurred with treatment of the
pelvis. This fits with a deductive fall back strategy that follows when the
treatment associated with the predominant pattern didn’t result in expected
outcomes. The following data have been presented here in relation to this
discussion point:
Expert C: P: If we can’t keep the leg raise above fifty degrees day one … I
start to worry about um getting something structurally …
R: What do you mean by that?
P: If the disc if the annulus has sequestrated come adrift or just
dissect totally, if there’s an underlying structural instability with a
pars defect with a grade two or three spondylolisthesis.

The only occasion of novice PR use in the present study led to a diagnosis
of neuromuscular instability. This case understanding was similarly identified
by two other novices and four experts who used analytical reasoning. A
clinical link between neuromuscular instability and spondylolisthesis is
supported by research that found the presence of neuromuscular
dysfunction in a sample of patient’s with grade 1 or 2 spondylolistheses and
chronic LBP (O'Sullivan, Twomey, Allison, Sinclair et al, 1997).

169
Neuromuscular instability, or clinical instability as it was originally termed
(Panjabi, 1992), is considered a contributing factor to LBP, however it lacks
a clear clinical presentation. Age, bilateral symptoms and history of trauma
were the significant case features (section 2.8.3) used by the experts in
obtaining the hypothesis of spondylolisthesis (section 3.2.2), whereas
consistent symptoms and case features for neuromuscular instability are not
defined within the literature. Consequently it could be argued that
neuromuscular instability is a contributing component of spondylolisthesis,
but neuromuscular instability alone cannot be considered accurate as a
case diagnosis.

The concerns relating to PR accuracy may relate more to less experienced

clinicians who may lack the experience and ability to integrate all clinical
data effectively. This study provides minimal evidence of such ineffective
use of PR in first year practising physiotherapists who overwhelmingly were
found to use analytical reasoning. This does not allay the possibility of this
outcome in physiotherapists with intermediate levels of experience which is
an area deserved of further investigation.

5.1.2 Efficiency of pattern recognition

Does PR actually increase the efficiency of problem solving? This question

was considered from the perspective of time to reach an end point, which
was the central understanding of the case, and the overall time spent
conducting the client history.

The original inclusion of a novice group in the study primarily related to

exploring whether PR is a strategy of more experienced or expert
practitioners. However the use of two groups also allowed for the
comparison of assessment time between experts and novices, regardless of
reasoning strategy employed. The results provided in section 4.4.4 showed
a trend towards experts taking longer to conduct a client history when
compared with the novice group. This finding was similar to that of prior

170
physiotherapy studies (Doody & McAteer, 2002; King & Bithell, 1998). The
simple data analysis provided in Table 4.14 demonstrates the median and
mean time taken for novices was 15% and 12% respectively less than that
for the experts.

The identification of PR occurred in the first 36% of the client history (section
4.4.4). In relation to actual time, all the patterns were identified within the
first 7 minutes of the assessment. This gives an appearance of efficiency,
however the overall assessment times for expert participants relative to
reasoning strategy (Table 4.13) found that the identification of PR was
associated with a longer time taken to complete the client history. Similarly
the only novice to incorporate PR into their assessment took the greatest
amount of time out of the entire novice group. When placing all participants
into either PR or analytical reasoning groups irrespective of experience, the
outcome was the same.

This outcome is seemingly at odds with the assumption that PR is more

efficient. However it should be remembered that the study consisted of only
part of a complete physiotherapy assessment (the client history). If
participants were also required to complete a physical examination these
results may well have been different. Where a clear understanding and
correct pattern exists then arguably the physical examination should be
more specific and targeted in nature, as compared with a series of tests to
prove or disprove several hypotheses.

The completion of an entire assessment (history and physical examination)

might provide a better view of efficiency but has its own research limitations.
Training an actor to portray the physical findings from a simulated case in an
accurate way is difficult and complex. Additionally, research involving real
life high fidelity cases is much more suited to assessing the accuracy of PR
but is limited with respect to efficiency. The present study design had a
greater focus on the identification of PR and assessment of its accuracy
which consequently influenced the findings relating to efficiency.

171
A final consideration of efficiency relates to the outcome of the case used in
the present study (section 3.2.2). Best management of a high grade
spondylolisthesis case involves referral to medical specialist. Earlier
identification of the correct pathology would presumably improve efficiency
to the referral part of management. However this would not necessarily
equate to overall efficiency given the spondylolisthesis condition was
monitored by a medical specialist over a 12 month period. Identifying the
correct pathology could also increase the efficiency of appropriate
physiotherapy management but this would rely on knowing the appropriate
management for this client based on agreed best practice for this condition.
To consider efficiency based on the cost effectiveness of treatment services
is beyond the scope of this study but worthy of further research.

5.2 PARTICIPANT ORDER OF QUESTIONING

The nature of the high fidelity study allowed for unbiased collection of clinical
information at the discretion of each participant. It was anticipated that
retrospective inspection of the order of questions posed to the simulated
client would provide another means of gaining insight into the predominant
diagnostic reasoning strategy. A summary of the results relating to the
participant order of questions was provided in section 4.4.5.

During the design phase of the study, the order of questions were
considered a data source able to provide a form of data triangulation.
However, the results indicated that no firm conclusions could be made with
respect to identifying confirmation questions following the use of PR. No
evidence was found that opposed or supported the identification of PR from
the order of questioning. The only conclusion that could be taken from this
data source is that each participant took a unique pathway with respect to
the order of gathering clinical information.

5.2.1 Triangulation

172
The insufficient findings relating to the observational order of question data
precluded complete data triangulation as introduced in section 3.8. This type
of triangulation requires separate data sources all reaching the same
conclusions.

The methodology of this study did not use the student researcher’s own
interview notes as a data source to provide triangulation. Although this data
source is common with qualitative research, it has potential for introducing
personal bias to the results and consequently was not considered with
triangulation of this study.

Confidence in the primary study findings relating to the presence of PR can

be taken from the methodological triangulation of the mixed qualitative and
quantitative methods utilised. The use of the participant’s own words as
qualitative data relating to reasoning process, in addition to application of
the predetermined coding schema and identification tool relating to PR, were
complemented by the quantitative statistical analyses regarding its presence
and accuracy. The actual study method triangulation is depicted in Figure
5.1.
Quantitative data
obtained via
predetermined coding
schema and
identification tool

Pattern
recognition

Direct qualitative Statistical analysis

data relating to regarding presence
reasoning process and accuracy

Figure 5.1 Study method triangulation

173
5.3 PATTERNS

The phenomenon of PR should always be considered in relation to the

understanding of what comprises a pattern. The makeup and development
of a pattern has been included in this chapter as it forms a foundation for the
use of PR as a reasoning strategy. This section is based not on direct
evidence from the results, but rather the student researcher’s unfolding view
of what comprised the participants’ patterns and consideration of the
theoretical concepts existing in the literature. This multi-faceted presentation
of a pattern has not been previously reported within the medical and allied
health literature and helps provide a more detailed understanding of the
concept.

5.3.1 Categorisation and patterns

PR can be better understood if considered alongside categorisation theory

where similarities between objects or events have associated underlying
knowledge structures. The models of categorisation described from the
literature in section 2.5.3 provide a clear basis to analyse patterns. The
patterns found in the present study findings were viewed from the
perspective of either ‘prototype’ (abstracted patterns from several cases) or
‘exemplar’ (instance-based recognition).

In relation to ‘exemplar’ and ‘prototype’ patterns, the underlying knowledge

compositions were considered from a varying complexity perspective
(Figure 5.2). Arguably an increased complexity of knowledge structure will
associate with prototype or multiple abstracted cases and lesser complexity
with instance-based case recognition. However even single accurate
exemplar patterns may have an elaborate underlying knowledge structure
consistent with prototypes.

174
 Exemplar Prototype
Patterns Patterns

Single prior case Multiple prior cases

Simple knowledge Complex knowledge

structure structure

Figure 5.2 Knowledge structure and pattern types

Of the five participants employing PR during the study, the knowledge code /
experiential sub-code data was analysed in an attempt to provide insight into
the type of patterns triggered by the simulated case. The experts correctly
identifying spondylolisthesis were found to use a ‘prototype’ model of the
case at hand. Several observations provided insight to this effect:
Expert I: I don’t see a lot of kids because kids typically don’t get, get low
back pain … but when we do we, we seem to get them with long
term symptoms but a lot of the time there’s structural reasons
why
Expert D: R: So does that fit with then what you’ve seen before?
P: Yeah, that’s based on previous experience with people who
are describing a similar story to what she is
R: Have you seen a case similar to this before?
P: Several times
R: Recently or over the years?
P: not recently, not immediately in short term memory but
certainly, several, several times previously I would have this story
being described

175
Interestingly there was no evidence in the data that could support the use of
an ‘exemplar’ pattern in the participants employing PR. The only participant
employing PR and making reference to having seen this type of case before
was expert D who recognised the case on multiple occasions from prior
experience.

5.3.2 Knowledge structure and pattern accuracy

The study data relating to knowledge in the observed ‘prototype’ patterns

was unable to provide a good view of knowledge complexity. Therefore a
view of the type of knowledge (propositional and non-propositional
knowledge types as described in section 2.7.1) integrated within a pattern
formed the basis to discuss the structure and accuracy of a pattern.

The overall accuracy of any pattern is likely far greater when knowledge
from clinical experience exists in conjunction with structured knowledge from
a public perspective (Figure 5.3). The continual comparison of individual /
personal knowledge to that of public / common knowledge (Higgs and
Titchen, 2000) in relation to a familiar case is likely to result in a more
integrated knowledge structure for that pattern. This is effectively integrating
experiential and propositional knowledge types via active reflection.

176
 Knowledge

Agreed expert
views
Individual Public Research validated
personal common knowledge
perspective perspective

Pattern accuracy

Figure 5.3 Pattern accuracy and knowledge relationship

Reflection in clinical practice has long been discussed in relation to deeper

learning (Boud et al, 1985; Schon, 1987). The use of active reflection by a
clinician with respect to a given case, its familiar aspects and the level of
evidence supporting that case will likely be associated with a better
outcome. Critical reflection on the integration of knowledge types during
clinical experience is a method of education that may enhance the accuracy
of developing patterns.

Returning to the perspective of non-analytical reasoning, it is likely that

those experience based patterns refined with public / propositional
knowledge should be more accurate. This does not mean that personal
patterns are not useful in clinical practice, especially to the novice
practitioner. As long as these personal patterns are utilised with some
caution and awareness of their limitations, they form a basis for
development and refinement of more accurate patterns. The end point of
pattern development is that those patterns with better underpinning
evidence are likely to lead to more effective management decisions and

177
improved outcomes. To develop a pattern within non-analytical reasoning
requires reflection on the knowledge structure that triggers its activation.

The coded data relating to knowledge were separated into propositional and
the various non-propositional types as outlined in section 4.3.1. Only one of
the three experts identifying the correct pathology and two of the five
participants using PR were coded using propositional knowledge. Out of all
the participants using analytical reasoning only one was similarly sub-coded
as using propositional knowledge. This review of the data does not offer
much support for integration of knowledge types with respect to reasoning
pathway or level of clinical experience. There may be several reasons for
these findings.

Firstly, there was a dominance of non-propositional knowledge over

propositional knowledge coded from the transcripts. Although counting the
occurrences of a code or sub-code was not undertaken in this study, the
sub-coding of knowledge found that only three expert participants and no
novices referred to propositional knowledge during their interview. Secondly,
these findings are potentially a limitation of the semi-structured interview
utilised, which did not specifically seek out participants’ knowledge types or
structures being utilised. Returning to the study methods introduced in
Chapter 3, exploring the knowledge basis to reasoning would require a less
structured interview but with a focus on exploring participants’ knowledge
behind their primary hypothesis. Although some attempts were made to
obtain such data during the interview, exploring knowledge in greater depth
had considerable potential to influence the participants’ responses and
therefore bias results relating to the primary study aims.

5.3.3 Developing patterns

PR has been shown to be a characteristic of more experienced

physiotherapists. Therefore somewhere along the pathway of gaining clinical
experience, patterns are developed. Prototype patterns will presumably be

178
formed and continually refined with exposure to clinical cases and
information relevant to the specific pattern. An exemplar pattern may start as
a single case and remain so with frequent exposure to the same type of
case. Alternatively, exemplar patterns may develop into prototypes if
variations of the single case are encountered. From a research and
educational perspective it is interesting to ask whether we can actually
identify a developing pattern in a physiotherapist.

This study has introduced the notion of a developing pattern based on a

prototype model. As reported in section 4.5.3, two novices were considered
during analysis to have attempted to use patterns within their clinical
assessments but not found to employ PR as a predominant reasoning
approach. The reasons behind this interpretation were different for each
participant, and necessitate individual discussion in further detail. The
hypotheses developed by novice participants Q and R were listed in time
sequence relative to the clinical assessment to provide a view of potential
patterns within analytical reasoning.

The display of novice R data indicated possible PR use on one occasion

when judged by the criteria of a single hypothesis based on significant case
features. This ‘instability’ hypothesis however was immediately followed by
deductive reasoning with several hypotheses. Is this potentially a developing
‘instability’ pattern or in fact an occasion where a known pattern did not
clearly fit with the case and led back to deductive strategies? The ‘central
hypothesis’ requirement of PR clearly indicated novice R did not use this
strategy, however the data potentially highlights a developing pattern.

Listing the hypotheses for novice Q (section 4.5.3) resulted in a deductive

multiple hypothesis interpretation but interestingly the participant made
reference to a search for a pattern amongst the data. In addition to “Looking
at patterns um to see if there’s any”, novice Q used a forwards confirmation
questioning strategy during the latter stages of their assessment. Novice Q
stated they were “chucking in a few questions to sort of negate structures

179
perhaps, and it sort of leads me towards more, more that chronic um
instability”. This qualitative data potentially provides an alternate view of a
developing pattern and an attempt to use it in practice.

The present study aimed to identify clear use of PR strategies but as

reported in the literature (section 2.5.1), separating PR from a deductive
reasoning process is not without complexity. The chosen methodology was
designed to determine evidence of PR distinct from deductive strategies.
Although PR was identified, some occasions of pattern use may not have
been identified because of the central hypothesis requirement. The data
from these novices may indicate possible occasions of attempted pattern
use that cannot be clearly separated from deductive strategies. The
question is whether this is a premature use of patterns or just hypothesis
formation via deductive reasoning, or a mixture of both.

5.3.4 Specificity of patterns

The term ‘pattern specificity’ has been used in relation to the level of
intricacy or complexity of a clinical pattern. This is distinct from the term
‘specificity’ used within quantitative research. Considering the specific depth
of patterns may well be useful in understanding their development. For
example, a specific pattern may be a well known diagnosable condition such
as the case utilised in the present study.

Spondylolisthesis pathologies may be associated with more specific patterns

as compared with that of neuromuscular instability (Figure 5.4). Yet as
discussed in section 5.1.1, spondylolistheses are also known to sometimes
have neuromuscular instability as a contributing factor (O'Sullivan, Twomey,
Allison, Sinclair et al, 1997). A spondylolisthesis condition is therefore one of
several lumbar clinical presentations that can be associated with a
neuromuscular instability pattern. This view of ‘pattern specificity’ was
helpful in separating two common but similar patterns held by participants in
this study.

180
 Broad Specific
Pattern Pattern
“Neuromuscular “Spondylolisthesis”
instability”

Figure 5.4 Specificity of patterns

Pattern specificity may be worthy of consideration in further research

targeting the development of patterns. Based on the study findings, novice
patterns are likely to start more broadly. These early patterns are then likely
to become interrelated with other more specific patterns via clinical
experience, such as the example of spondylolisthesis to neuromuscular
instability. Developing specific patterns arguably depends on the extent of
reflection undertaken with common clinical presentations. A visual mind
mapping or concept mapping educational strategy (Beissner, 1991) might
assist reflection on such compositions.

Another factor influencing the specificity of patterns is research evidence

relating to the condition or case. Where clear evidence exists, there is more
opportunity for a pattern to be specific. This provides another link back to
prototype patterns with integrated propositional knowledge.

181
5.3.5 Significant case features

There has been sufficient support in the literature to introduce the concept of
significant case features within a definition of PR (section 2.5.1). This study
used reference to significant case features (Groves et al, 2002), similar to
other terms such as salient cues (Coderre et al, 2003) and key features
(Groves et al, 2003). Based on the specificity of patterns discussed in
section 5.3.4, Figure 5.4 has been modified to include the input of clinical
data that leads to an end point or diagnosis (Figure 5.5). The clinical data
considered significant may allow for recognition of broad or specific patterns.

Broad Pattern
Clinical Significant case features Specific Pattern
Data

Figure 5.5 Input of clinical data into broad and specific patterns

Recognition of significant case features were observed in the identification

of the specific spondylolisthesis case / pattern in this study. All three of the
experts identifying the precise diagnosis utilised similar features leading to
the recognised pattern. The significant case features included age, bilateral
nature of the symptoms, and the history of trauma as an 11 year old. These
common significant features were directly comparable with the case
literature in section 3.2.2, including bilateral lower lumbar pain spreading to
the gluteal regions and posterior thighs and either insidious onset or

182
associated with a history of minor trauma. The additional unmentioned
feature that could have influenced this pattern was potentially that of female
gender, however spondylolisthesis is not restricted to this gender (Earl,
2002).

The use of significant case features was an important component of the PR

identification tool. However, this does not mean that the use of significant
case features is exclusive to those using non-analytical reasoning
strategies, as identifying multiple deductive hypotheses may also include
key features to develop each hypothesis. Significant case features were
particularly interesting in their ability to trigger and support the patterns
observed in the study.

5.3.6 Pattern elimination

The pattern related free code (section 4.3.3) made reference to the
possibility of pattern elimination being used during reasoning in the study.
Two experts (E & F) indicated on several occasions that the clinical case
data ‘did not fit’, which was inferred by the student researcher to be not
‘fitting’ when compared to previous experience and possibly familiar
patterns. These experts were not found in the primary study results to have
utilised PR due to the lack of a central hypothesis.

Instead of this data providing evidence of pattern elimination, closer

inspection of the expert transcripts led more towards a picture of failed
pattern matching which subsequently influenced the reasoning strategy
employed. The comparison of the experts’ known pattern(s), which did not fit
with the presenting case, led to a deductive reasoning process. In other
words, these experts were observed to be possibly using a pattern matching
strategy during the assessment but the lack of familiarity or recognition led
them to revert to a different reasoning pathway.

183
On reflection of pattern elimination, it should be similar to the inclusion
features of PR. The key element to clearly identify PR was the reference to
a central hypothesis. Based on this perspective finding evidence of pattern
elimination would require participant statements that excluded a single case
hypothesis.

5.4 HYPOTHESIS CATEGORY USE

The predetermined coding schema of the study was such that it allowed for
observation of hypothesis category use. The two experience-separated
groups provided an interesting assessment of the extent of hypothesis use
learned via undergraduate physiotherapy programs as compared with
experienced clinicians. The obvious limitation of this section relates to the
numbers per group relative to the stated differences in observations.
However these observations may still be useful with respect to education
and further research relating to hypothesis development. This is in line with
previous research in earlier hypothesis classifications (Payton, 1985; Rivett
& Higgs, 1997). The classification of hypothesis types utilised in this study
follow the trend towards holistic healthcare of a person with their own
experiences and limitations (Jones & Rivett, 2004). No further reports of
hypothesis category use have been reported in the musculoskeletal
physiotherapy literature since 2004.

The hypothesis categories utilised in the study also allow for reporting on
use of hypotheses that are outside that of diagnostic reasoning. As
introduced in sections 2.1.3 and 2.3.2, the types of reasoning reported in the
literature can be separated into those associated predominantly with
diagnosis and those with management (Edwards & Jones, 2007). Although
this study was primarily focussed on identifying hypotheses developed via
diagnostic reasoning pathways, the extent of hypothesis categories utilised
by participants provides an indirect view of the other reasoning types in
action (Figure 5.6). The final understanding of a case, whether diagnostic or

184
otherwise, should be a composite of the various hypothesis types (Edwards
& Jones, 2007).

Diagnostic
reasoning

Narrative
reasoning Hypothesis types:
1. Activity / participation
Procedural 2. Patient’s perspective / psychosocial
Data reasoning 3. Pathobiological mechanisms
4. Physical impairments / structural sources
collection Predictive 5. Contributing features
reasoning 6. Precautions and contraindications
7. Management and treatment
Ethical 8. Prognosis
reasoning

Collaborative
reasoning

Case
understanding

Figure 5.6 Hypothesis categories and reasoning types (based on Edwards

& Jones, 2007)

The selected examples of coded hypotheses and associated categorisation

via sub-coding have been detailed in section 4.3.1. The simple analysis of
these data (sections 4.6.1 and 4.6.2) found a few interesting outcomes
worthy of discussion. Firstly, the clinical hypotheses developed by
participants were spread across all eight categories, which provide support
for the classification of hypothesis types described by Jones and Rivett
(2004). Secondly, a comparison of expert and novice groups with respect to
hypothesis category use demonstrates a trend for increased breadth of
hypothesis generation with more experienced or expert physiotherapists.

185
The review of all hypotheses developed at any stage during the client history
was used to compare the two groups with respect to the extent of
hypothesis formation (Figure 4.4). A greater degree of hypothesis formation
was observed in the expert group as compared with the novices. However
statistical analysis of the results found the only category to be significantly
different in its use was that of Category 2 (patient’s perspective /
psychosocial factors). Although not statistically different, the use of
‘management’ and ‘prognosis’ hypothesis categories suggested that experts
tend to think ahead more so than novices when problem solving.

The difference between the groups in the spread of hypotheses generated

throughout the history compared with the final hypothesis could be theorised
to be indicative of the expert clinician’s ability to integrate the various case
hypotheses. A review of the final stated hypotheses (Figure 4.3) found the
expert group used a greater number / broader range of hypothesis
categories in comparison to the novice participants. This could be
interpreted as a greater depth of understanding via integration of more
hypothesis categories. Figure 4.3 also highlights the limited ability of the
novice participants to integrate hypotheses outside of categories 4 (physical
impairments and associated structure / tissue sources) and 5 (contributing
factors) into their final understanding.

A final question that lay within the observation of hypothesis categories

related to the depth of cognition during clinical reasoning. Can we state that
integrating a greater number of hypothesis categories in reasoning indicates
more complex or deeper cognition? These data may provide some insight
into the potential effects of experience and / or postgraduate education
relative to hypothesis generation. Further research specific to the integration
of hypothesis types in problem solving is clearly required. Additionally, the
limited ability of novice participants to integrate the majority of hypothesis
categories into their understanding of a case is worthy of investigation.

186
5.5 OBSERVED APPROACHES TO REASONING

It appears that each physiotherapy clinician has a clinical reasoning style or

approach that varies with the clinical context and case at hand. The
physiotherapy literature has highlighted various interpretive reasoning
strategies (Table 2.1) that can be incorporated alone or in combination with
each other and alongside analytical or non-analytical reasoning processes.
This section utilises the additional study findings reported in section 4.7 to
support several aspects related to clinical reasoning in the physiotherapy
literature.

5.5.1 Predictive reasoning

One reasoning strategy reported from the interpretive paradigm was that of
predictive reasoning (section 2.1.3). The present study findings include the
identification of predictions within the problem solving of some clinicians
(section 4.7.4). Interestingly, none of the predictive reasoning data identified
from transcripts related to predictions of management as previously
reported. Rather all predictions were based around the interpretation of
clinical assessment findings. This finding could have related to the fact the
study requested a case understanding during the assessment but did not
specifically seek management information from participants.

The use of predictive reasoning in relation to assessment data was

theorised to be more a strategy utilised following the recognition of a pattern.
This hypothesis was effectively linking assessment predictions to
confirmation questioning. For example, a triggered clinical pattern such as
an achilles tendinopathy should be followed by targeted questions and
specific physical tests to confirm the pattern. The location and description of
pain may trigger the pattern which is tested by search questions relating to
morning stiffness or the 24 hour symptom behaviour, and physical testing
via direct palpation of the tendon. The pattern would most likely have links
with the potential effectiveness of various treatments and consequently

187
more accurate predictions of treatment outcomes or prognosis. This
relationship however was not found during analysis in this study. The direct
comparison of the data relating to predictive reasoning and PR found only
one participant to be utilising both within their assessment. Despite the small
numbers of participants, the study results do not support the hypothesis that
PR and predictive reasoning would be observed in conjunction.

5.5.2 Metacognition

An important element of the collaborative hypothesis oriented model of

clinical reasoning in physiotherapy (Jones, 1995; Jones et al, 2000) is the
theoretical notion of metacognition. Arguably metacognition should be
observable, partly via concurrent integration of data throughout a client
history. Logically this would suggest that any clinician using a delayed
integrative approach is not metacognitively skilled. However the use of this
approach could also be a deliberate strategy employed with the awareness
that a more comprehensive and possibly unbiased interpretation is able to
be gained once all the clinical information is available to the practitioner.

The study findings identified the delayed integrative approach within both
participant groups but did not explore the reasons associated with its use.
The timing data related to the free code of ‘delayed versus concurrent data
integration’ (section 4.7.2) found those participants who stated their
approach to involve collecting the data then subsequently reasoning, took
less time to complete their client history (Table 4.21). This time data was
misleading with respect to efficiency because the time was stopped on
completion of client questions. Any potential reasoning time following data
collection was not taken into account.

The limitation of associating the total assessment time data to the

participants coded with delayed integration lies with the assumption that
coding is indicative of one predominant approach. Novices L and M were
coded with delayed integration at one instance but this approach was not

188
applicable to the entire client assessment. As such, it was not appropriate to
conduct a statistical analysis of the timing data. The primary insight
obtainable from this aspect of analysis was that the ‘collect now, think later’
approach did not appear associated with PR use and did not result in the
generation of the correct spondylolisthesis hypothesis. A more explorative
unstructured interview would be required to fully understand the potential
interaction between delayed integration of clinical data and metacognition.

5.5.3 Awareness of errors

A common theme amongst a small number of expert participants involved

an open minded approach during data collection (Table 4.4). This is viewed
as an important mechanism to ensure that potential errors such as
hypothesis bias are avoided throughout the interpretative phase of data
collection. This theme was introduced in section 4.7.3 and only identified in
expert participants who displayed concurrent integration of data. Types of
errors have been reported in section 2.9.1. Several of the errors reported in
the literature (Rivett & Jones, 2004) may have been potentially avoided in
the study via a consciously open minded method of assessment. These
could include:
 Neglecting or misinterpreting relevant information
 Premature decision-making
 Not recognizing data inconsistencies
 Confirmation bias – overemphasis on supporting features and neglecting
negating features of a hypothesis
 Presumption that a relationship between symptoms confirms cause and
effect and thus diagnosis.

Particularly insightful was the finding that two of the three participants who
used PR with a correct diagnostic outcome, also showed an open minded
approach to other possibilities. This suggests that it is possible to use non-
analytical reasoning strategies and avoid common errors made during the
data collection and interpretative phases of clinical assessment. These two

189
participants confidently held their predominant case hypothesis throughout
the assessment but were willing to alter it if non-supportive information
became available. This open mindset that arguably is a desirable
accompaniment to PR is clearly displayed in the following transcript
quotation:
Expert D: That would be my working hypothesis that I would be wanting to
test with my examination … but I would be completely prepared
to find something completely different

The third participant showing a willingness to remain open minded did not
use a delayed integrative approach and was not found to have utilised PR.
They displayed an ‘error prevention’ approach within analytical reasoning
particularly associated with the case not fitting a known pattern:
Expert H: I do try and keep quite an open mind as far down the track as I
can because I know that you can get quite influenced and then,
and then find that it’s not really the case, so I try really to make
judgment ah, fairly far down the track

The ability to remain open minded could in itself be interpreted as indicative

of skilled metacognition, especially in the presence of PR. A balance
between non-analytical reasoning and an open minded approach is likely to
help minimise errors of data collection and interpretation.

5.5.4 Person centred approach

This final approach identified in the study findings involved a non-diagnostic

approach. Although the study aimed to explore the accuracy of a diagnostic
non-analytical reasoning strategy, it also allowed for interpretation of the
case using non-diagnostic reasoning strategies. It has been stated that a
case understanding solely based on the hypothesis category of ‘physical
impairments and associated structure / tissue sources’ is insufficient in
current physiotherapy practice (Edwards & Jones, 2007). A more global
understanding includes hypotheses relating to a person’s activity /

190
participation capacity or limitations and the patient’s perspective and / or
psychosocial factors that may present. This creates a holistic understanding
of a person with a diagnosable pathology rather than just a medical label or
diagnosis. The person centred approach to assessment has been
introduced in section 4.7.5 of the study findings.

The view of a person centred approach within physiotherapy is consistent

with the direction of recent research relating to diagnosis and outcomes. A
specific diagnosis for LBP is only possible in 15% of all lumbar spine cases.
Thus a more holistic and not a purely diagnostic approach may lead to a
better understanding of a larger number of LBP clients and their problems.
Even in specific LBP cases it can be argued that management decisions
should include the person and not just be based on diagnosis (Edwards &
Jones, 2007).

Considering the restrictions of a person’s desired activity levels or ability to

participate in life is an example of taking a person centred approach to
clinical reasoning. Outcome based research indirectly supports the view that
impairment based clinical findings are insufficient to justify a diagnosis or
evaluate the benefit of an intervention. Including functional activity /
participation findings within clinical practice has support via
recommendations to use functional disability instruments such as the
Quebec and Roland Morris questionnaires (Maher et al, 1999).

The analysis of hypothesis category use (section 4.6.2) supported the

findings relating to a person centred approach to problem solving. The depth
of thought processing (range of all hypothesis categories sub-coded) was
observed during the entire retrospective interview. Use of hypothesis
categories 1 (activity and participation) and 2 (patient’s perspective /
psychosocial factors) are particularly indicative of this approach.

Consideration of the client as a person with their own perceived problems

varied within the study sample. Both expert and novice groups had

191
participants who were holistic in their approach and others who were
narrowly focussed on a specific impairment based understanding. Not
surprisingly the novices were more likely to lack the holistic approach (as
indicated by use of the various hypothesis categories), which has potential
educational implications at an undergraduate level. Interestingly one
participant from the expert group also adopted a slightly narrower approach
to problem solving during the study, not utilising either activity and
participation or patient’s perspective / psychosocial categories of
hypotheses. This participant utilised the five other hypothesis categories and
adopted the delayed integration approach as discussed in section 5.5.2.

The simulated nature of the study may well have impacted on the
participants and their reasoning. However given that 80% of the experts
considered ‘patient’s perspective / psychosocial’ hypotheses and 60% the
‘activity and participation’ hypothesis, it would suggest that the results are
reasonably valid. The participant reports of the realism of the client
encounter were also of a satisfactory level despite knowing the client was an
actor.

5.6 STUDY DESIGN

Several key elements of the study design were monitored due to their
potential impact on the results. This section considers participant
recruitment with respect to study numbers attained, and the influence of the
study conditions on participant behaviour. The accuracy and realism of the
actor role playing the simulated case is also discussed, and lastly the coding
process is examined in further detail.

5.6.1 Participant recruitment

The outcome of recruiting ten expert and nine novice participants was
acceptable to the study aims. The findings relative to the primary research
question and qualitative analysis methods were considered adequately

192
supported by the sample size. Although a greater number of participants in
each group could have potentially altered the results comparing PR use in
experts to novices (section 4.4.2), this did not detract from the primary study
finding that PR was evident.

The sample population was chosen with consideration of the feasibility of

recruitment and those who potentially had sufficient experience in such a
case to demonstrate PR. The recruitment of the first ten participants in each
group meeting the inclusion criteria and consenting to participate provided
an unbiased sample that was not pre-selected. Had the study pre-selected
musculoskeletal physiotherapy clinical specialists (Australian Physiotherapy
Association, 2008) with significant experience in the lumbar spine as the
expert group, the proportion of observations of PR may have been greater.
This is nevertheless dependent on experience with the specific condition
chosen for this study. The inclusion criteria used for expert selection in the
present study greatly increases the generalisability of the study findings.

The novice recruitment target number of ten participants was not quite
achieved but this is unlikely to have impacted on the overall results of the
study. The original number of ten was based on previous clinical reasoning
research as discussed in Chapter 3, however saturation of data was
achieved in the novice group with respect to PR as the primary focus of the
study. A final participant was not able to be recruited via the method as
outlined for this group in Chapter 4. Potential reasons included the method of
advertising, time available to participate and novice self-confidence relating
to the research task.

5.6.2 Experimental context influences

The influence of the location of the study (section 3.4.1) and the video /
audio recording equipment (section 3.4.2) were evaluated by questioning at
the end of the interview. The qualitative interview data reported in section

193
4.8.2 provided confidence that the behaviour of participants was minimally
affected by the study setting.

It is possible that participants could have undertaken the clinical assessment

more thoroughly than normal due to the observational nature of the study.
The ‘orientation to the patient assessment’ information sheet (Appendix 10)
was designed to minimise this. Qualitative results found an early participant
focus on recording equipment was quickly transferred to a focus on
assessing the client (section 4.8.2). No comments were made by
participants that indicated the equipment altered their client assessment
from normal practice.

The effect of location could have been lessened by undertaking the data
collection process in each participant’s clinical practice. This was considered
during the study development phase but would have increased the overall
cost of conducting the study via actor employment and associated travel
costs. The recording equipment also required a closed and relatively quiet
room for effective recording and subsequent transcription accuracy. This
was considered difficult to ensure in some participants’ clinical
environments. Nevertheless, difficulty in recruiting experts was subsequently
managed by conducting the study in their usual clinical setting following
ethics variation approval (Appendix 8).

5.6.3 Simulated client

Critical to the study was the need for an actor who could roleplay a realistic
version of the case on repeated occasions. The time invested in this facet of
the study was described in section 3.2.3 and was consistent with prior
studies using simulated clients (Ladyshewsky et al, 2000). The results
relating to client simulation accuracy have been outlined in section 4.8.1 and
indicate a high level of response precision. This is consistent with prior
published results relating to case simulation in physiotherapy (Ladyshewsky
et al, 2000).

194
The qualitative responses from participants also indicated a high level of
case realism. The only two occasions where participants commented on the
case being simulated by an actor suggested the influence on participant
responses was brief and inconsequential (section 4.8.1). The ethical
requirement of making the participants aware of the simulated nature of the
case made it impossible to fully control this aspect of the study. Use of a real
client has significant ethical implications and raises other methodological
issues relating to diagnostic accuracy.

5.6.4 Coding process

Coding was used to organise textual data and allow for meaningful analysis.
Given the research question aimed to investigate an existing phenomenon,
codes were developed to identify PR rather than emerging from the data.
These codes were developed into the predetermined codebook (section
3.5.1). It is generally quite difficult to ensure validity in qualitative coding,
however it is important that the codes have face validity (Sim & Wright,
2000). General agreement related to interpretations of the data and their
coding was obtained between researchers (section 3.5.1) as a means of
ensuring face validity of the codes.

The reliability of coding was another factor considered with respect to the
rigour of the data collection. Coding reliability is generally optimised if more
than one coder is involved and results compared. This process was
undertaken qualitatively between researchers (section 3.5.1) for the first two
interview transcripts of each group which provided general agreement on
the predetermined tree codes and the hypothesis sub-codes. Subsequently
the coding reliability of the student researcher was considered with respect
to repeated coding as outlined in section 3.5.3. Intra-coder reliability was
found to be very high (section 4.3.2) in relation to the predetermined codes.

195
It is possible that such an intra-coder reliability process could be biased by
the sole coder’s recall of the first occasion of coding analysis. Every attempt
was made to minimise this by the time frame between initial and repeat
coding being greater than 3 months. Despite this possible limitation the
reliability evaluation indicated that the majority of data available for coding
from each transcript had been obtained.

5.7 STUDY LIMITATIONS

Several potential limitations of the study have been considered with regards
to the results. These can be separated into limitations of:
 Retrospective recall data accuracy
 Participant voice
 Semi-structured interview, and
 Case / context specificity.

It has been well reported that retrospective recall data may be limited in its
accuracy relating to actual cognition at the time of problem solving (Elstein
et al, 1990; Elstein & Schwartz, 2000). The basis for the chosen
methodology was reported in section 2.2.4, however it is recognised that this
remains an unavoidable limitation of the study method given the use of a
high fidelity case. This limitation was managed via the immediacy of the
retrospective recall and the stimulated form of recall using the videotaped
observation data.

During the participant interview, instructions were repeated to provide only

thoughts from the time of the actual assessment. Clear instructions were
also provided prior to the interview (Appendix 8): “Try to describe what was
going on in your mind at the actual time and not thoughts or decisions from
afterwards. In other words, it is important that you try to recall your thinking
at each step of the assessment and not to be influenced by information you
may have obtained later”. Despite these efforts, problem solving during the

196
retrospective observation of the clinical assessment remains a potential
limiting factor of this type of methodology.

One qualitative coding approach to monitor the limitation of the retrospective

recall method was via a free code labelled ‘thinking after the event’. During
the coding stage of analysis this code flagged attention to whether the
retrospective method was leading to a bias of ‘problem solving after the
event’. This free code (section 4.3.3) was used to monitor occasions where
problem solving using clinical data was noted to occur during the review of
the videotape recorded assessment. Thus, the use of this free code was a
means to identify occasions where problem solving after the event had
occurred. This was observed during analysis, however none of these
occasions were considered likely to influence the primary study findings.

The potential for participants to have unstated thoughts from the client
assessment (during the retrospective recall interview) is a possible limitation
of the study. The skill of the interviewer during the semi-structured interview
was essential in obtaining the necessary data but care was needed to not
influence the participant with leading prompts. The possibility existed though
that some participants were not able to articulate their thoughts or were less
willing to do so due to the nature of the study design. Similarly those with
less reflective ability may not have been able to fully describe their thought
processes. This potential limitation of participant voice relates back to the
chosen retrospective recall methodology. The alternative is to employ a
cognitive psychology approach of obtaining evidence via determining
relationships between observed responses and cognitive stimulus (Elstein et
al, 1990; Elstein & Schwartz, 2000). Such an approach relies on low fidelity
case types which were deemed not suited to this study. Further reading from
the literature relating to this consideration is summarised in section 2.2.

It is possible that the type of semi-structured interview used in the study did
not always elicit a participant’s comments relating to knowledge use during
problem solving. An example of this was introduced in section 5.1.1, where

197
additional data relating to the observed occasions of PR use could have
provided more insight into the presence and makeup of the patterns. This
potential limitation may have been managed by questions relating to the
participant’s awareness of spondylolisthesis pathology and their associated
prior clinical experience. Although such a discussion could have been
included following complete data collection relating to problem solving, this
was not included to minimise study bias due to participant contamination.
Even though participants were requested not to discuss the case study with
fellow professionals, the disclosure of the diagnosis would have increased
the chances of contamination and thus bias.

The last potential limitation of the study method relates to generalisability. A

case / context specificity effect is possible given that only one case was
utilised and mostly done so out of the clinical environment. This potential
limitation does not lessen the evidence presented that PR was found to exist
and appears to be more accurate than HDR. However the fact that only 15%
of LBP cases can be diagnosed (Waddell & van Tulder, 2004) potentially
limits the frequency of PR use in diagnostic reasoning of clients with lumbar
spine complaints. A further consideration in the diagnostic accuracy of PR is
the current lack of a positive relationship between the specificity of
management and the clinical outcome in chronic non-specific LBP (Critchley
et al, 2007; Kent et al, 2005). In other words, the cost-effective benefit of an
accurate diagnosis can only be stated if the subsequent related treatment
provides a superior outcome. Any benefits with respect to PR in non-specific
LBP are arguably potentially linked to the ability to sub-classify LBP.

5.8 SUMMARY OF DISCUSSION

Chapter 5 has provided discussion on the study results with respect to the
primary research aims and several related clinical reasoning findings. Along
with considering several design features of the study and potential
limitations, this chapter leads to conclusions (Chapter 6) relating to the key
findings and possible future implications for education.

198
CHAPTER 6. CONCLUSION

Research in clinical reasoning can adopt high or low fidelity study design
methods. Variations of both have been used extensively in the medical and
physiotherapy literature when attempting to answer questions associated with
problem solving in clinical practice. This study aimed to identify PR in
musculoskeletal physiotherapy using high fidelity research methods and
subsequently investigate its relationship to accuracy and efficiency.

6.1 KEY FINDINGS

This study provides supportive evidence for PR as a form of diagnostic

reasoning in musculoskeletal physiotherapy. A set of strict assessment
criterion for PR were developed to allow for definitive identification and clear
separation from HDR. This method found a number of participants used PR
during the clinical assessment of the lumbar spine case. Although there are
some potential limitations in relation to the type of methodology used in this
study, the findings support the conclusion that PR is utilised as a reasoning
process by musculoskeletal physiotherapists.

Two participant groups with a large experience and knowledge divide were
included in the study. This was to evaluate whether PR use is associated with
greater domain specific and experiential knowledge. The results found that
four of ten experts used PR, as did one of nine novices. Closer inspection of
the PR identification data in all five participants using PR suggested that its
use by the single novice was weak when compared with the four experts.
Despite the lack of statistically significant difference between the groups, the
presence of PR as a predominant reasoning process in five of nineteen
participants supports its existence in musculoskeletal physiotherapy.

Accuracy of PR has not previously been assessed in physiotherapy and rarely

has it been evaluated in medicine using high fidelity case simulation. The
present study utilised a real life spondylolisthesis case consistent with the

199
reported literature, to facilitate the examination of PR accuracy. It
demonstrated that PR was linked with accuracy but its use was not a
guarantee of success with only three of the four experts using this strategy
identifying the correct diagnosis. Nevertheless, statistical analysis found a
significantly greater likelihood of PR achieving an accurate diagnosis when
compared with the analytical process model of HDR. The results also suggest
that incorrect use of PR initially is not fatal and may still lead to a correct
diagnostic outcome if the clinician reverts to an analytical reasoning process.

PR use has been traditionally viewed as being a more efficient process when
compared with analytical reasoning. The present study monitored time as a
measure of efficiency during the clinical assessment (client history) but was
not able to determine whether PR is more or less efficient than analytical
reasoning strategies in physiotherapy. It was found that PR produced a
predominant hypothesis early in the clinical assessment (within the first 36% of
the client history time). However in these cases the total assessment time was
actually longer than for the participants not using PR. The present study also
demonstrated that experts took more time to conduct their client history than
novices which is in line with prior physiotherapy research (King & Bithell, 1998;
Doody & McAteer, 2002). The limitation of these results relates to the study
incorporating only one component of a physiotherapy assessment (i.e. client
history) and not the entire first clinical session which includes a physical
examination and management.

The final key area worth considering from the study findings relates to the use
of hypothesis categories amongst participants during problem solving.
Although this was not a primary study aim, monitoring the use of hypothesis
categories was particularly insightful in understanding differences in reasoning
between participant groups. The experts had a significantly greater use of the
‘patient’s perspective / psychosocial’ hypothesis category than novices during
problem solving. The expert group also utilised a larger number of hypothesis
categories in their descriptions of the final hypothesis. In this regard, all of the

200
novices were limited to hypotheses in the ‘physical impairments & associated
structure / tissue sources’ and / or ‘contributing factors’ categories.

6.2 IMPLICATIONS FOR FURTHER RESEARCH AND EDUCATION

The results of this study add to the increasing evidence for non-analytical
reasoning within healthcare. The inclusion of PR as a type of clinical reasoning
process in musculoskeletal physiotherapy is justified, however its use remains
highly dependent on the clinician’s knowledge and experience of similar cases.
This study is the first to investigate diagnostic accuracy of PR using a high
fidelity case method. The positive association between PR and accuracy
provides further support to the findings of recent low fidelity medical research
studies (Coderre et al, 2003) and the similarity of results between studies in
musculoskeletal physiotherapy and medicine.

Future research involving varied clinical cases would provide further insight
into the accuracy of PR in musculoskeletal physiotherapy. Additional data
collection regarding the expert participant’s level of knowledge and experience
with respect to the presenting case would add to the understanding of
developing patterns.

The educational implications for PR in physiotherapy were an important

rationale for the study. However at the time of developing the study, the level
of understanding relating to PR was insufficient for any meaningful educational
research question. Given the present study has provided new insights
regarding PR and its accuracy, the educational implications can now be
considered.

An area of future research lies with educational design to enhance diagnostic

reasoning in physiotherapy. Would a carefully designed case based approach
to education facilitate the use of PR? Would it produce more accurate
outcomes? If so, the time and cost associated with this type of educational
strategy would then need to be examined.

201
Whilst raising educational questions it should be considered that authors
advise against the use of PR by novices (Coderre et al, 2003; Norman et al,
2000; Norman, 2005). Certainly teaching the use of PR as a problem solving
strategy in isolation is not the answer, but facilitating physiotherapists to
recognise common conditions based on significant case features would
potentially assist pattern use and possibly improve diagnostic accuracy. This
notion is effectively stating that increased exposure to common clinical
presentations and their variations increases the clinician’s experiential
knowledge and based on the findings of this study, may potentially increase
diagnostic accuracy.

This educational theory is presumably reliant on developing sound reflective

practice amongst novices to help develop patterns and to avoid inappropriate
use of PR. Yet the interactions between the reflective abilities of
physiotherapists and pattern development and accuracy are poorly
understood.

Clinicians with several years or more of experience but not enough to enable
expert practice are referred to as ‘intermediates’ in clinical reasoning research.
Physiotherapists at this level were not included in this study to clearly separate
groups from an experience and expertise perspective, however this is an area
with potential in PR research. Do intermediate physiotherapists accurately use
PR? The present study has introduced the notion of developing patterns
amongst novice physiotherapists. Continual development and refinement of
recognisable clinical patterns is arguably a characteristic of intermediates on
the road to expertise. Yet there is a lack of research based understanding in
this area.

The pressure on clinical education placements in physiotherapy presently

requires more innovative thought to maximise the available clinical experience
throughout undergraduate and postgraduate programs. Given the relationship
between PR and accuracy in this study, the question of whether recognition of

202
clinical patterns can be facilitated through other educational activities is one
worth considering.

The finding of limited hypothesis category use by novice participants in the

study also merits consideration in educational design. Research into the
extent of hypothesis category use by undergraduate physiotherapy students
during clinical reasoning would complement the findings of this study.
Subsequent research could then trial and evaluate methods to enhance
novice capacity to think more broadly in clinical practice.

Research into the efficiency of clinical reasoning processes cannot be

reduced to just time spent undertaking a client assessment. It should
encompass the time taken to produce an accurate understanding of a case
sufficient to implement appropriate management strategies. Future
research therefore should strive to enhance our understanding of the
complex interaction of clinical reasoning with accuracy, efficiency and cost
effectiveness of treatment services.

203
REFERENCE LIST

Anderson, D. M. (1989). Dorlands Medical Dictionary. Philadelphia: WB

Saunders Company.

Australian Physiotherapy Association. (2001). National physiotherapy

service descriptors. Retrieved October 18, 2008, from
http://physiotherapy.asn.au/images/Document_Library/Service_Descriptors/
servicedescriptors.pdf

Australian Physiotherapy Association. (2005). Directory of APA

Musculoskeletal Physiotherapists 2005-2006. APA: Victoria.

Australian Physiotherapy Association. (2008). Specialisation. Retrieved

October 18, 2008, from
http://physiotherapy.asn.au/index.php/physiotherapy-a-
you/specialisation/specialisation

Arocha, J. F., Patel, V. L., & Patel, Y. (1993). Hypothesis generation and
the coordination of theory and evidence in novice diagnostic reasoning.
Medical Decision Making, 13(3), 198-211.

Barbour, R. S. (2008). Introducing qualitative research: A student guide to

the craft of doing qualitative research. London: Sage Publications.

Barrows, H. S., & Feltovick, P. J. (1987). The clinical reasoning process.

Medical Education, 21, 86-91.

Barrows, H. S. (2000). Stimulated Recall (Personalized assessment of

clinical reasoning). Springfield: Southern Illinois University School of
Medicine.

204
Beissner, K. L. (1991). The effectiveness of concept mapping for improving
problem solving. Michegan: UMI Dissertation services.

Beutler, W. J., Fredrickson, B. E., Murtland, A., Sweeney, C. A., Grant, W.

D., & Baker, D. (2003). The natural history of spondylolysis and
spondylolisthesis. Spine, 28(10), 1027-1035.

Bordage, G., & Lemieux, M. (1991). Semantic structures and diagnostic

thinking of experts and novices. Academic Medicine, 66(9), S70-S72.

Boshuizen, H. P. A., & Schmidt, H. G. (1992). On the role of biomedical

knowledge in clinical reasoning by experts, intermediates and novices.
Cognitive Science, 16, 153-184.

Boud, D., Keogh, R., & Walker, D. (Eds.). (1985). Reflection: Turning
Experience into Learning. London: Kogan Page.

Britten, N. (1995). Qualitative research: qualitative interview in medical

research. British Medical Journal, 311, 251-253.

Brooks, L. R., Norman, G. R., & Allen, S. W. (1991). Role of specific

similarity in a medical diagnostic task. Journal of Experimental Psychology:
General, 120(3), 278-287.

Caputo, L. A., & Mior, S. A. (1998). The role of clinical experience and
knowledge in clinical decision making. Topics in Clinical Chiropractic, 5(2),
10-18.

Carlin, B. P., & Louis, T.A. (1996). Bayes and empirical bayes methods for
data analysis. London: Chapman & Hall.

205
Coderre, S., Mandin, H., Harasym, P. H., & Fick, G. H. (2003). Diagnostic
reasoning strategies and diagnostic success. Medical Education, 37, 695-
703.

Coughlin, L. D., & Patel, V. L. (1987). Processing of critical information by

physicians and medical students. Journal of Medical Education, 62, 818-
828.

Creswell, J. W. (2007). Qualitative inquiry & research design. Choosing

among five approaches. (Second ed.). California: Sage Publications.

Creswell, J. W., & Plano Clark, V. L. (2007). Designing and conducting

mixed method research. California: Sage Publications.

Critchley, D. J., Ratcliffe, J., Noonan, S., Jones, R. H., & Hurley, M. V.
(2007). Effectiveness and cost-effectiveness of three types of
physiotherapy used to reduce chronic low back pain disability. Spine,
32(14), 1474-1481.

Davidson, J., & Jacobs, C. (2008). The implications of qualitative research

software for doctoral work. Qualitative Research Journal, 8(2), 72-80.

DeWald, C. J., Vartabedian, J. E., Rodts, M. F., & Hammerberg, K. W.

(2005). Evaluation and management of high-grade spondylolysthesis in
adults. Spine, 30(6S), S49-S59.

Domholdt, E. (2005). Rehabilitation research. Principles and applications.

(Third ed.). St Louis: Elsevier Saunders.

Doody, C., & McAteer, M. (2002). Clinical reasoning of expert and novice
physiotherapists in an outpatient orthopaedic setting. Physiotherapy, 88(5),
258-268.

206
Earl, J. E. (2002). Mechanical aetiology, recognition, and treatment of
spondylolisthesis. Physical Therapy in Sport, 3, 79-87.

Edwards, I., Jones, M., Carr, J., Braunack-Mayer, A., & Jensen, G. (2004).
Clinical reasoning strategies in physical therapy. Physical Therapy, 84(4),
312-330.

Edwards, I., Braunack-Mayer, A., & Jones, M. (2005). Ethical reasoning as

a clinical-reasoning strategy in physiotherapy. Physiotherapy, 91, 229-236.

Edwards, I., & Jones, M. (2007). Clinical reasoning and expert practice. In
G. Jensen, J. Gwyer, L. M. Hack & K. F. Shepard (Eds.), Expertise in
physical therapy practice (Second ed., pp. 192-213). St Louis: Saunders
Elsevier.

Elstein, A. S., Shulman, L. S., & Sprafka, S. A. (1978). Medical Problem

Solving. An analysis of clinical reasoning. Cambridge: Harvard University
Press.

Elstein, A. S., Shulman, L. S., & Sprafka, S. A. (1990). Medical problem

solving. A ten-year retrospective. Evaluation & The Health Professions,
13(1), 5-36.

Elstein, A. S., & Schwartz, A. (2000). Clinical reasoning in medicine. In J.

Higgs & M. Jones (Eds.), Clinical Reasoning in the Health Professions (2nd
ed., pp. 95-106). Oxford: Butterworth-Heinemann.

Elstein, A. S., & Schwartz, A. (2002). Clinical problem solving and

diagnostic decision making: selective review of the cognitive literature.
British Medical Journal, 324, 729-732.

207
Embrey, D. G., Guthrie, M. R., White, O. R., & Dietz, J. (1996). Clinical
Decision Making by Experienced and Inexperienced Pediatric Physical
Therapists for Children With Diplegic Cerebral Palsy. Physical Therapy,
76(1), 20-33.

Ericsson, K. A. (2004). Deliberate practice and the acquisition and

maintenance of expert performance in medicine and related domains.
Academic Medicine, 79(10), S70-S81.

Eva, K. W., Neville, A. J., & Norman, G. R. (1998). Exploring the etiology of
content specificity: factors influencing analogic transfer and problem
solving. Academic Medicine, 73(10), S1-S5.

Eva, K. W. (2004). What every teacher needs to know about clinical

reasoning. Medical Education, 39, 98-106.

Floman, Y. (2000). Progression of lumbosacral isthmic spondylolisthesis in

adults. Spine, 25(3), 342-347.

Gale, J., & Marsden, P. (1982). Clinical problem solving: the beginning of
the process. Medical Education, 16, 22-26.

Gale, J., & Marsden, P. (1983). Medical diagnosis: From student to

clinician. Oxford, Oxford University Press.

Gelman, A., Carlin, J. B., Stern, H. S., & Rubin, D. B. (2004). Bayesian data
analysis. (Second ed.). Boca Raton, FL: Chapman & Hall/CRC.

Gifford, L. (1998). Pain, the tissues and the nervous system: A conceptual
model. Physiotherapy, 84(1), 27-34.

208
Glaser, R., & Chi, M. T. H. (1988). Overview. In M. Chi, R. Glaser & M. Farr
(Eds.), The Nature of Expertise (pp xv-xxviii). Hillsdale: Lawrence Erlbaum
Associates.

Grant, J., & Marsden, P. (1987). The structure of memorised knowledge in

students and clinicians: an explanation for diagnostic expertise. Medical
Education, 21, 92-98.

Grant, J., & Marsden, P. (1988). Primary knowledge, medical education

and consultant expertise. Medical Education, 22, 173-179.

Groen, G. J., & Patel, V. L. (1985). Medical problem-solving: some

questionable assumptions. Medical Education, 19, 95-100.

Groves, M., Scott, I., & Alexander, H. (2002). Assessing clinical reasoning:
a method to monitor its development in a PBL curriculum. Medical Teacher,
24(5), 507-515.

Groves, M., O'Rourke, P., & Alexander, H. (2003). The clinical reasoning
characteristics of diagnostic experts. Medical Teacher, 25(3), 308-313.

Hampton, J. A. (1998). Similarity-based categorization and fuzziness of

natural categories. Cognition, 65, 137-165.

Hasnain, M., Onishi, H., & Elstein, A. S. (2004). Inter-rater agreement in

judging errors in diagnostic reasoning. Medical Education, 38, 609-616.

Haun, D. W., & Kettner, N. W. (2005). Spondylolysis and spondylolisthesis:

A narrative review of etiology, diagnosis, and conservative management.
Journal of Chiropractic Medicine, 4, 206-217.

209
Hayes, B., & Adams, R. (2000). Parallels between clinical reasoning and
categorization. In J. Higgs & M. Jones (Eds.), Clinical Reasoning in the
Health Professions (2nd ed., pp. 45-53). Oxford: Butterworth-Heinemann.

Hensinger, R. N., & Michigan, A. A. (1989). Spondylolysis and

spondylolisthesis in children and adolescents. The Journal of Bone and
Joint Surgery, 71(7), 1098-1107.

Herman, M. J., & Pizzutillo, P. D. (2005). Spondylolysis and

spondylolisthesis in the child and adolescent: A new classification. Clinical
Orthopaedics and Related Research, 434, 46-54.

Higgs, J. (1992). Developing knowledge: a process of construction,

mapping and review. New Zealand Journal of Physiotherapy, 20(2), 23-30.

Higgs, J. (1993). A programme for developing clinical reasoning skills in

graduate physiotherapists. Medical Teacher, 15(2/3), 195-206.

Higgs, J., & Jones, M. (2000). Clinical reasoning in the health professions.
In J. Higgs & M. Jones (Eds.), Clinical Reasoning in the Health Professions
(2nd ed., pp 3-14). Oxford: Butterworth-Heinemann.

Higgs, J., & Titchen, A. (2000). Knowledge & reasoning. In J. Higgs & M.
Jones (Eds.), Clinical Reasoning in the Health Professions (2nd ed., pp. 23-
32). Oxford: Butterworth-Heinemann.

Higgs, J. (2004). Educational theory and principles related to learning

clinical reasoning. In M. A. Jones & D. A. Rivett (Eds.), Clinical Reasoning
for Manual Therapists (pp. 379-402). Edinburgh: Butterworth-Heinemann.

Holmgren, U., & Waling, K. (2008). Inter-examiner reliability of four static

palpation tests used for assessing pelvic dysfunction. Manual Therapy, 13,
50-56.

210
Jensen, G., Gwyer, J., Shepard, K. F., & Hack, L. M. (2000). Expert
practice in physical therapy. Physical Therapy, 80(1), 28-43.

Jones, M. A. (1992). Clinical reasoning in manual therapy. Physical

Therapy, 72(12), 875-884.

Jones, M. (1995). Clinical reasoning and pain. Manual Therapy, 1, 17-24.

Jones, M., Jensen, G., & Edwards, I. (2000). Clinical reasoning in

physiotherapy. In J Higgs & M Jones (Eds.), Clinical Reasoning in the
Health Professions (second edition, pp. 117-127). Oxford: Butterworth-
Heinemann.

Jones, M. A., & Rivett, D. A. (2004). Introduction to clinical reasoning. In M.

A. Jones & D. A. Rivett (Eds.), Clinical Reasoning for Manual Therapists
(pp. 3-24). Edinburgh: Butterworth-Heinemann.

Joseph, G. M., & Patel, V. L. (1990). Domain knowledge & hypothesis

generation in diagnostic reasoning. Medical Decision Making, 10(1), 31-46.

Kent, P., Marks, D., Pearson, W., & Keating, J. (2005). Does clinician
treatment choice improve the outcomes of manual therapy for nonspecific
low back pain? A metaanalysis. Journal of Manipulative and Physiological
Therapeutics, 28(5), 312-322.

King, C. A., & Bithell, C. (1998). Expertise in diagnostic reasoning: a

comparative study. British Journal of Therapy & Rehabilitation, 5(2), 78-87.

Ladyshewsky, R. K., Baker, R., Jones, M., & Nelson, L. (2000). Reliability
and validity of an extended simulated patient case: A tool for evaluation and
research in physiotherapy. Physiotherapy Theory and Practice 16, 15-25.

211
Ladyshewsky, R. K. (2002). A quasi-experimental study of the differences
in performance and clinical reasoning using individual learning versus
reciprocal peer coaching. Physiotherapy Theory and Practice, 18, 17-31.

Ladyshewsky, R. K. (2004). Impact of Peer-Coaching on the Clinical

Reasoning of the Novice Practitioner. Physiotherapy Canada, 56(1), 15-25.

Lim, M. R., Yoon, S. C., & Green, D. W. (2004). Symptomatic

spondylolysis: diagnosis and treatment. Current Opinion in Pediatrics, 16,
37-46.

Maher, C., Latimer, J., & Refshauge, K. (1999). Atlas of clinical tests and
measures for low back pain: Australian Physiotherapy Association.
http://physiotherapy.asn.au/index.php/groups/musculoskeletal-
physiotherapy/mpa-resources-clinical

May, B. J., & Dennis, J. K. (1991). Expert Decision Making in Physical

Therapy - A Survey of Practitioners. Physical Therapy, 71(3), 190-206.

Mays, N., & Pope, C. (1995). Rigour and qualitative research. British
Medical Journal, 311, 109-112.

McNeely, M. L., Torrance, G., & Magee, D. J. (2003). A systematic review

of physiotherapy for spondylolysis and spondylolisthesis. Manual Therapy,
8(2), 80-91.

Microsoft. (2003). Microsoft office excel 2003.

http://www.microsoft.com/products/info/product.aspx?view=22&type=ovr&p
cid=ee959bd8-14b4-4601-91a7-eea1043249fa

National Occupational Health & Safety Commission. (2004). The cost of

work-related injury and illness for Australian employers, workers and the
community. Canberra: Australian Government Publishing Service.

212
Nendaz, M. R., Gut, A. M., Perrier, A., Reuille, O., Louis-Simonet, M.,
Junod, A. F., et al. (2004). Degree of concurrency among experts in data
collection and diagnostic hypothesis generation during clinical encounters.
Medical Education, 38, 25-31.

Noll, E., Key, A., & Jensen, G. (2001). Clinical reasoning of an experienced
physiotherapist: insight into clinician decision-making regarding low back
pain. Physiotherapy Research International, 6(1), 40-51.

Norman, G. R., Tugwell, P., Feightner, J. W., Muzzin, L. J., & Jacoby, L. L.
(1985). Knowledge and clinical problem-solving. Medical Education, 19,
344-356.

Norman, G. R., Coblentz, C. L., Brooks, L. R., & Babcook, C. J. (1992).

Expertise in visual diagnosis: A review of the literature. Academic Medicine,
67(10), S78-S83.

Norman, G. R., Brooks, L. R., Colle, C. L., & Hatala, R. M. (2000). The
benefit of diagnostic hypotheses in clinical reasoning: Experimental study of
an instructional intervention for forward and backward reasoning. Cognition
& Instruction, 17(4), 433-448.

Norman, G. R., & Eva, K. W. (2003). Doggie diagnosis, diagnostic success

and diagnostic reasoning strategies: an alternative view. Medical Education
37, 676-677.

Norman, G. (2005). Research in clinical reasoning: past history and current

trends. Medical Education, 39, 418-427.

Norman, G., Young, M., & Brooks, L. (2007). Non-analytical models of clinical
reasoning: the role of experience. Medical Education, 41, 1140-1145.

213
O'Sullivan, P., Twomey, L., & Allison, G. (1997). Evaluation of specific
stabilisation exercise in the treatment of chronic low back pain with
radiologic diagnosis of spondylolysis or spondylolisthesis. Spine, 22(24),
2959-2967.

O'Sullivan, P., Twomey, L., Allison, G., Sinclair, J., Miller, K., & Knox, J.
(1997). Altered patterns of abdominal muscle activation in patients with
chronic low back pain. Australian Journal of Physiotherapy, 43(2), 91-98.

O'Sullivan, P. (2005). Diagnosis and classification of chronic low back pain

disorders: Maladaptive movement and motor control impairments as
underlying mechanism. Manual Therapy, 10, 242-255.

Panjabi, M. M. (1992). The stabilizing system of the spine. Part II. Neutral
zone and instability hypothesis. Journal of Spinal Disorders, 5(4), 390-397.

Patel, V. L., & Groen, G. J. (1986). Knowledge Based Solution Strategies in

Medical Reasoning. Cognitive Science, 10, 91-116.

Patel, V. L., Groen, G. J., & Arocha, J. F. (1990). Medical expertise as a

function of task difficulty. Memory & Cognition, 18(4), 394-406.

Patel, V. L., Groen, G. J., & Norman, G. R. (1993). Reasoning and

Instruction in Medical Curricula. Cognition & Instruction, 10(4), 335-378.

Patel, V. L., & Arocha, J. F. (2000). Methods in the study of clinical

reasoning. In J. Higgs & M. Jones (Eds.), Clinical Reasoning in the Health
Professions (Second ed., pp. 78-91). Oxford: Butterworth-Heinemann.

Patton, M. Q. (1990). Qualitative research and evaluation methods (Second

ed.). Thousand Oaks: Sage Publications.

214
Payton, O. D. (1985). Clinical reasoning process in physical therapy.
Physical Therapy, 65(6), 924-928.

Petty, N. J. (2006). Neuromusculoskeletal examination and assessment: a

handbook for therapists (3rd ed.). New York: Churchill Livingstone.

Pope, C., Ziebland, S., & Mays, N. (2000). Qualitative research in health
care. Analysing qualitative data. British Medical Journal, 320, 114-116.

QSR International (n.d.). NVivo 7. Retrieved January 19, 2009, from

http://www.qsrinternational.com/

Richards, L., & Morse, J. M. (2007). Readme first for a users guide to
qualitative methods (Second ed.). Thousand Oaks: Sage Publications.

Ridderikhoff, J. (1985). Models of decision-making in the general practice: a

design for a descriptive research. Medical Informatics, 10(4), 323-337.

Ridderikhoff, J. (1989). Methods in medicine. Dordrecht: Kluwer Academic

Publishers.

Ridderikhoff, J. (1991). Medical problem-solving: an explanation of

strategies. Medical Education, 25, 196-207.

Rivett, D., & Higgs, J. (1995). Experience and expertise in clinical

reasoning. New Zealand Journal of Physiotherapy, 23, 16-21.

Rivett, D. A., & Higgs, J. (1997). Hypothesis generation in the clinical

reasoning behaviour of manual therapists. Journal of Physical Therapy
Education, 11(1), 40-45.

215
Rivett, D. A., & Jones, M. A. (2004). Improving clinical reasoning in manual
therapy. In M. A. Jones & D. A. Rivett (Eds.), Clinical Reasoning for Manual
Therapists, (pp. 403-419). Edinburgh: Butterworth-Heinemann.

Roberts, A. E. (1996). Clinical reasoning in occupational therapy:

Idiosyncrasies in content and process. British Journal of Occupational
Therapy, 59(8), 372-376.

Rossi, F., & Dragoni, S. (2001). The prevalence of spondylolysis and

spondylolisthesis in symptomatic elite athletes: radiographic findings. The
College of Radiographers, 7, 37-42.

Round, A. P. (2000). Introduction to clinical reasoning. British Medical

Journal (student), 8, 15-17.

Schon, D. A. (1987). Educating the Reflective Practitioner. Toward a New

Design for Teaching and Learning in the Profession. San Francisco: Jossey-
Bass.

Scott, I. (2000). Teaching clinical reasoning: A case-based approach. In J

Higgs & M Jones (Eds.), Clinical Reasoning in the Health Professions
(second edition, pp. 290-297). Oxford: Butterworth-Heinemann.

Sim, J., & Wright, C. (2000). Validity, reliability and allied concepts.
Research in Health Care: Concepts, designs & methods (pp. 123-139).
Cheltenham: Stanley Thornes Ltd.

Simon, H. A. (1980). Problem solving and education. In D. T. Tuma & F. Reif

(Eds.), Problem solving and education: Issues in teaching and research (pp.
81-96). Hillsdale, N.J.: Erlbaum Associates.

Smart, K., & Doody, C. (2006). The clinical reasoning of pain by

experienced musculoskeletal physiotherapists. Manual Therapy, 12, 40-49.

216
Stanitski, C. L. (2006). Spondylolysis and spondylolisthesis in athletes.
Operative Techniques in Sports Medicine, 14, 141-146.

Treble, S. L., Niezabitowski, B. A., & Bull, P. W. (2005). Spondylolisthesis

and associated clinical features: A retrospective study from a chiropractic
student outpatient clinic. Chiropractic Journal of Australia, 35(3), 112-116.

Waddell, G., & van Tulder, M. (2004). Clinical guidelines. In G Waddell

(Eds.), The back pain revolution (second edition, pp. 283). London:
Churchill Livingston.

Wilkes, G.A., & Krebs, W. A. (1989). The Collins Australian pocket

dictionary (2nd Edition). Sydney: HarperCollins.

Woods, N. N., Brooks, L. R., & Norman, G. R. (2005). The value of basic
science in clinical diagnosis: creating coherence among signs and
symptoms. Medical Education, 39, 107-112.

Yin, R. K. (2003). Case study research (Third ed.). California: Sage

Publications.

217
PATTERN RECOGNITION IS A CLINICAL REASONING
PROCESS IN MUSCULOSKELETAL PHYSIOTHERAPY

APPENDICES

Peter Andrew Miller

B Phty, Grad Cert Health Science (Education)

Thesis submission for Master of Medical Science

(Physiotherapy)
The University of Newcastle, Australia
Submitted March 2009

218
APPENDIX 1. ETHICS APPROVAL

219
APPENDIX 2. CASE SCENARIO

20 year old female 2nd year nursing student presented to physiotherapy

with primary complaint of low back pain.

Pain A I/M deep

ache 2-8/10

Pain B
I/M
ache
5/10

R L
Pain E I/M
Pain C I/M
ache / Pain D I/M
cramp like
cramp pain cramp like
pain 5/10
4 - 5 /10 pain 5/10

Primary Symptoms

 Pain as per body chart

Pain Relationship

 Pain B (both buttocks) occurs when Pain A (lower back) increases

 Pain C (right leg) & Pain D (left leg) can occur together or separately and
always with Pain A / B
 Pain C more frequent than Pain D
 Pain E (posterior thigh) only presents after onset of Pain C & D
 Overall pain tends to start in lower back then spread to buttocks before
either leg symptoms commence (Pain C – E)

220
Other Symptoms

 Intermittent giving way of legs with changing direction in walking –

unrelated to pain
 Bilateral anterior knee pain with stairs or sport – unrelated to pains A – E
 The knee pains have not changed with recent increased back pain
 Feels generally tired from pain with standing

Pins and Needles / Numbness

 Nil

Current Goals

 Aim of going for treatment was to decrease pain to continue nursing

placement

24-Hour Pattern

 Morning is better – usually no pain on waking then low back pain comes
on within 10-15 minutes of standing / getting up (but currently wakes with
2/10 pain following clinical placement)
 No stiffness in mornings
 Through day pain levels are activity dependant but generally worsen
 In afternoons back and leg pains pain will come on more easily than
mornings with same precipitating activities
 Night OK once got to sleep (not long to get to sleep – 15 min’s on
average) – mattress is OK (college mattress currently). No difference
between home and college mattresses.

Current Precipitating Factors

 Flat Walking > 30-40mins (Uphill and downhill walking are both about the
same as each other but both more difficult than walking on flat ground)
 Standing > 10mins (6-7/10 low back pain; worse the longer the period of
time standing)
 Carrying loads out in front or load off to side (e.g. one shopping bag to
side)

221
 Lying face down (especially with knees bent up behind)
 Clinical placement – bending over patients (5mins tolerance)
 Sitting > 30-40mins (worse if sitting upright – 10mins tolerance)
 Netball increases low back pain and mild leg pain (either side but R > L;
commonly both together and with posterior thigh pains)
 Worst aggravating activity (Lower back pain 8-/10) is standing for
prolonged periods (e.g. shopping for few hours). Leg pain would be onset
after about half an hour of shopping

Irritability of Symptoms

 Onset of symptoms and associated severity in precipitating factors

 If standing / walking, gets immediate reduction of severity of pain with
sitting (gets 3-4 point reduction of pain immediately i.e. 6-7/10 becomes
3-4/10). Mostly doesn’t have to sit due to pain but habitually does so to
control level of symptoms
 Pa/b settles quickly (10mins) with lying down (preferably side-ly), but if leg
pains are present (Pain A – E) takes longer to settle. If all pains are
present, can still reduce low back and buttock pain with lying down but
only slight reduction of leg pains. Leg pains generally stay for rest of day
and go away overnight.

Easing Factors

 Eases over 10mins on lying down (side-ly) after sport but doesn’t go away
completely
 Lying on back with knees bent (Lying with legs out straight can increase
pain)
 Left side-ly with top leg (right) bent and bottom leg (left) out straight
 Standing with hands on knees – bent forwards
 Slouch sitting relative to sitting straight
 Sitting eased pain compared with standing
 Occasionally used heat for pain relief in past – not currently. Minimal
relief.

222
General Functional Activity

 Sit to stand OK
 Generally prefers to keep moving
 Sits down whenever possible due to pains
 Lifting was generally OK when done correctly and light to moderate loads
(avoided lifting heavy loads) – except as indicated in precipitating factors
 Hanging washing – OK for limited time

Current History

 Nursing student – pain has notably increased associated with first clinical
placement (6 weeks - ended last week). Therefore pain has gradually
increased over the last 7 weeks and has continued at same level since
the placement ended last week. Thus worsened over the first 4 weeks of
placement and been the same since then
 Especially worsened (increased LBP) with bending over patients – as little
as 30 seconds endurance by end of shift; would have to stop assisting
patient after that time. Generally tried to avoid bending over patients.
 Recently pain has increased in severity related to more time on feet and
moving related to patient care
 More easily aggravated and more frequent pain. Harder to ease.
 Since clinical placement, waking with 2/10 pain in lower back (never
wakes with leg pain). Prior to placement, can be pain free completely
after sleeping overnight
 Legs give way occasionally (maximum 5 times in the last 12 months).
Never falls just feels like legs buckle under – into bending direction. Not
related to pain
 Previous exercises – 100 sit ups 2-3 times / week on bed (unable to on
floor as direct pressure on lower back causes pain); self-initiated related
to netball competition; generally didn’t help reduce back pains

Past History

 Fell backwards as an 11 year old & landed on bottom

 Difficulty walking immediately after incident

223
 Intermittent pain since injury – has lived with pain associated with activity
and aggravating activities. Flare-ups are usually not this severe (as rest
can be utilised to reduce symptoms). Flare-ups are mostly related to
netball carnivals (lots of games over a weekend). This episode is the
worst it has been and is now interfering with potential career
 Assessed by GP and sports medicine doctor in home town as 12 year old
– no Xrays taken; advised hamstring stretches (some improvement over
few months of stretches. Continued with routine of stretching with netball
training & games). Prescribed orthotics via podiatrist (no change in low
back pain

Social History

 Plays competitive netball with 4hrs training / week & 1-2 games / week.
Played from age 10yrs to currently (8-9 yrs)
 Swimming competitive for 7 years until 16 years old (no aggravation of
pain; no significant changes in pain levels)
 School soccer
 School golf (generally made low back pain worse – max 20 games)
 Tennis – 2 seasons of once / week at age 16 –17 years

Current Work

 Full time nursing student

 Works as casual sales assistant (newsagents) one shift of 4 hrs / wk -
generally could only stand at sales counter for 40 minutes before doing
another activity where sitting or bending is allowed (e.g. tidying or packing
shelves)

Previous Treatment

 Chiropractic at 14 years (self referral) 2-3 times / week for 8-12 wks
(improved)
 Chiropractic every few months at home when on holidays (temporary
relief)
 Orthotics as 12 year old from podiatrist

224
 No previous physiotherapy

Medications

 Nil currently
 No prior cortisone / steroid medications
 No NSAID’s (neurofen, voltaren etc) in past
 Rarely would take paracetamol / panadol if pain at its worst (e.g. after
netball carnival)

Other Special Questions

 X-rays with chiropractic but unsure of results

 Bladder / bowel function normal
 No paraesthesia (pins & needles / tingling) in saddle area
 General health – good
 Not aware of any diabetes, lung disorders (e.g. asthma) or heart
conditions
 Nil surgery
 Nil weight loss
 Cough / sneeze – nil problems associated
 Nil blood tests or other tests
 No unsteadiness with gait (only after a few drinks)
 No clumsiness in hands

Visual Cues

 Nil Obvious
 Walks in normally
 Finished clinical placement last week – dressed in casual clothes
 Not overweight

225
APPENDIX 3. ACTOR TRAINING INFORMATION

20 year old female 2nd year nursing student presented to physiotherapy

with primary complaint of low back pain.

Deep ache – comes

and goes (2-8/10)

Occasional R L
ache / cramp Occasional
pain (4-5/10) cramp pain Occasional
(5/10) cramp pain
Primary Complaints (5/10)

 Pain as per body chart above

Pain Relationships

 Pain starts in lower back then spreads to both buttocks (as low back pain
worsens)
 Outside leg pains (left and right legs) can occur together or separately but
always occur with lower back and buttock pains.
 Right leg pain is generally more frequent than left leg
 Posterior thigh pain only presents after onset of lateral (outside) leg pains
 Overall pain tends to start in lower back then spread to buttocks before
commencing in either leg

226
Other Symptoms

 Occasional giving way of legs with changing direction in walking –

unrelated to pain
 Pain in front of both knees with stairs or sport – unrelated to back pain or
other leg pains. The knee pains have not changed with recent increased
back pain.
 Feels generally tired from pain with standing

Pins and Needles / Numbness

 Nil

Current Goals

 Aim of going for treatment was to decrease pain to be able to continue

nursing placements (next placement in a few months)

24-Hour Pattern

 Morning is better – usually no pain on waking then low back pain comes
on within 10-15 minutes of standing / getting up (but currently wakes with
2/10 pain following clinical placement).
 No lower back or leg stiffness in mornings
 Through day pain levels are dependant on amount of activity but
generally worsen
 In afternoons back and leg pains pain will come on more easily than
mornings with same precipitating activities (see next section)
 Sleeping at night is OK once asleep (doesn’t take long to get to sleep –
15 min’s on average). Mattress is OK (college mattress currently). No
difference between home and college mattresses.

Current Precipitating Activities

 Flat Walking > 30-40mins (Uphill and downhill walking are both about the
same as each other but both more difficult than walking on flat ground)
 Standing > 10mins (low back pain score 6-7/10; severity of pain gets
worse the longer in standing)

227
 Carrying loads out in front or load off to side (e.g. one shopping bag to
side)
 Lying face down (especially with knees bent up behind)
 Being on nursing clinical placement – bending over patients (5mins
tolerance)
 Sitting > 30-40mins (worse if sitting upright – 10 mins tolerance)
 Netball increases low back pain and mild leg pain (either side but R > L,
commonly both together and with posterior thigh pains
 Worst aggravating activity (Lower back pain 8/10) is standing for
prolonged periods (e.g. shopping for few hours). Leg pain would be onset
after about half an hour of shopping.

Irritability of Symptoms

 Details of onset of symptoms and associated severity is found in

precipitating factors
 If standing / walking, gets immediate reduction of severity of pain with
sitting (gets 3-4 point reduction of pain immediately i.e. 6-7/10 becomes
3-4/10). Mostly doesn’t have to sit due to pain but habitually does so to
control level of symptoms
 Lower back and buttock pain settles quickly (10mins) with lying down
(preferably side-ly), but if leg pains are present takes longer to settle. If all
pains are present, can still reduce low back and buttock pain with lying
down but only slight reduction of leg pains. Leg pains generally stay for
rest of day and go away overnight.

Easing Factors

 Pain eases over a 10-minute period of lying down (side-ly) but doesn’t go
away completely (e.g. after sport)
 Lying on back with knees bent up (lying with legs out straight can
increase pain)
 Lying on left side with top leg (right) bent and bottom leg (left) out straight
 Standing with hands on knees – bent forwards
 Slouch sitting relative to sitting straight

228
 Sitting eases pain compared with standing
 Occasionally used heat for pain relief in past – not using currently
(generally gives minimal relief).

General Functional Activity

 No problems with getting up from or down into a chair

 Generally prefers to keep moving when on feet
 Sits down whenever possible due to pains
 Lifting is generally OK when done correctly and light to moderate loads
(avoids lifting heavy loads) Note the precipitating factors related to lifting
 Hanging washing – OK for limited time (within standing period 10-
minutes)

Current History

 Nursing student – pain has notably increased associated with first clinical
placement (6 weeks - ended last week). Therefore pain has gradually
increased over the last 7 weeks and has continued at same level since
the placement ended last week. Thus worsened over the first 4 weeks of
placement and been the same since then
 Especially worsened (increased low back pain) with bending over patients
– as little as 30 seconds endurance by end of shift; would have to stop
assisting patient after that time. Generally tried to avoid bending over
patients
 Recently pain has increased in severity related to more time on feet and
moving related to patient care
 More easily aggravated and more frequent pain. Harder to ease.
 Since clinical placement, waking with 2/10 pain in lower back (never
wakes with leg pain). Prior to placement, can be pain free completely
after sleeping overnight.
 Legs give way occasionally (maximum 5 times in the last 12 months).
Never falls, just feels like legs buckle under – into forwards bending
direction. Not related to pain.

229
 Previous exercises – 100 sit ups 2-3 times / week on bed (unable to on
floor as direct pressure on lower back causes pain); self-initiated related
to netball competition; generally didn’t help reduce back pains.

Past History

 Fell backwards as 11 year old & landed on bottom

 Difficulty walking immediately after incident due to pain
 Intermittent pain since injury – has lived with pain associated with activity
and aggravating activities. Flare-ups are usually not this severe (as rest
can be utilised to reduce symptoms). Flare-ups are mostly related to
netball carnivals (lots of games over a weekend). This episode is the
worst it has been and is now interfering with potential career.
 Assessed by general practitioner and sports medicine doctor in home
town as a 12 year old – no X-rays taken; advised hamstring stretches
(some improvement over few months of stretches. Continued with routine
of stretching with netball training & games); prescribed orthotics via
podiatrist (no change in low back pain).

Social History

 Plays competitive netball with 4hrs training / week & 1-2 games / week.
Played from age 10yrs to currently (8-9 yrs)
 Swimming competitive 7 years till 16 years old (no aggravation of pain; no
significant changes in pain levels)
 School soccer
 School golf (generally made low back pain worse – max 20 games)
 Tennis – 2 seasons of once / week at age 16 –17 years

Current work

 Full time nursing student

 Works as casual sales assistant (newsagency) one shift of 4 hrs / wk -
generally could only stand at sales counter for 40 minutes before doing
another activity where sitting or bending is allowed (e.g. tidying or packing
shelves)

230
Previous Treatment

 Chiropractic treatment when aged 14yrs (self referral) 2-3 times / week for
8-12 weeks (helped)
 Now gets chiropractic every few months at home when on holidays
(temporary relief)
 Orthotics as 12 year old from a podiatrist
 No previous physiotherapy

Medications

 Nil currently
 No prior cortisone / steroid medications
 No anti-inflammatory medications (neurofen, voltaren etc) in past
 Rarely would take paracetamol / panadol if pain at its worst (e.g. after
netball carnival)

Other Special Questions

 X-rays with chiropractic but unsure of results

 Bladder / bowel function normal
 No pins & needles / tingling in saddle area
 General health – good
 Not aware of any diabetes, lung disorders (e.g. asthma) or heart
conditions
 Nil surgery
 Nil weight loss
 Cough / sneeze – nil problems associated
 Nil blood tests or other tests
 No unsteadiness with gait (only after a few drinks)
 No clumsiness in hands

Visual Cues

 Nil Obvious
 Walks in normally
 Finished clinical placement last week – dressed in casual clothes

231
 Not overweight

Common Questions from Physiotherapists

Opening questions:

 As far as you are concerned what do you feel is your main problem?
 What is the problem today?
 What brings you here today?
 How can I help you today?

Pain / symptom related questions:

 Where exactly is the pain? (e.g. lower back both sides)

 What type of pain is that? How would you describe that pain?
 How severe would you rate the pain? If 0 was no pain and 10 was the
worst pain imaginable, what score would you give that pain?
 Are the pains constantly there or do they come and go?
 Do the pains feel deep inside or close to the skin?
 Do you have any pain in the legs?
 Do you get any pins and needles? If so where? When do you get these
sensations?
 Do you have any pins and needles in the pelvic or saddle area?
 Do your legs ever give way?

History questions:

 When did it start? When did your lower back pain start? How long have
you had low back pain for?
 How did it start?
 When do you remember the leg pains starting? Were they at the same
time as the low back pain?
 When did the pins and needles commence?
 Has the pain been the same since it first began?
 What has been happening with the pain more recently?
 Has the pain / problem changed at all? (Over the prior years)

232
Questions related to what makes pain better or worse:

 Does your pain vary in intensity?

 What activities make your pain worse?
 How long does it take for these activities to make the pain worse? How
bad does it get with these activities?
 Is there anything you can do to ease the pain?
 How much can you ease the pain? (i.e. fully, partially etc)
 How long does the pain last for after it comes on?
 Can you be completely free of pain?

24-hour questions:

 Does the pain vary through the day or night?

 Is one period of the day worse?
 Does the pain affect your sleep? Getting to sleep?
 Does the pain wake you up at night? How many times do you wake due
to the pain?
 How do you feel first thing in the morning?

Other questions:

 How is your general health? Do you have any medical conditions we

should be aware of?
 Have you lost any weight recently? Is there any reason for this?
 Do you take any medications?
 How is your heart / lungs / etc?
 Do you have diabetes / epilepsy / etc?
 Have you had any changes to your bladder or bowel function recently?
Do you have any problems with your bladder or bowel function?
 Do you ever feel clumsy when you walk? Do your legs ever give way?
 Have you ever taken steroid medications?
 Have you had any surgery in the past?

General points for training:

 Broad opening questions – provide a standard amount of information

233
 Pain scores are rated on a 0-10 scale where 0 is no pain and 10 is the
worst imaginable pain. Only provide the score out of 10 if requested by
the participant.
 Primary or main problem is low back pain
 Recent history relates to a nursing clinical placement

234
APPENDIX 4. SIMULATED CASE RESPONSE CHECKLIST

Participant number ____________ Date _____________

Case feature Accurate response by Information not

simulated patient requested
Location of pain types (Pain A) * Yes No
Location of pain types (Pain B) * Yes No
Association of pain A & B * Yes No
Location of pain types (Pain C) * Yes No
Location of pain types (Pain D) * Yes No
Location of pain types (Pain E) * Yes No
Association of pain C, D, E * Yes No

Severity of pain types (Pain A) * Yes No

Severity of pain types (Pain B) * Yes No
Severity of pain types (Pain C) * Yes No
Severity of pain types (Pain D) * Yes No
Severity of pain types (Pain E) * Yes No

Description of pain types (Pain A)* Yes No

Description of pain types (Pain B)* Yes No
Description of pain types (Pain C)* Yes No
Description of pain types (Pain D)* Yes No
Description of pain types (Pain E)* Yes No

Constancy of pain (Pain A) * Yes No

Constancy of pain (Pain B) * Yes No
Constancy of pain (Pain C) * Yes No
Constancy of pain (Pain D) * Yes No
Constancy of pain (Pain E) * Yes No

History of current episode Yes No

Past history Yes No
Mechanism of injury Yes No

Primary aggravating activities Yes No

Standing tolerance Yes No
Walking tolerance Yes No
Sitting tolerance Yes No
Primary easing factors Yes No

* Pain A – E relate to pain location indicated on the body chart in Appendix 2

235
Previous physiotherapy treatment Yes No
Other previous treatment Yes No

Morning pain / stiffness Yes No

Night pain / ability to sleep Yes No
Pain behaviour through day Yes No

Primary patient goals Yes No

Activity – netball participation Yes No
Anterior knee pain Yes No
Unsteadiness / giving way of legs Yes No
General health Yes No
Paraesthesia / numbness Yes No
Cough / sneeze Yes No
X-Rays Yes No
Prior surgery Yes No
Weight loss Yes No
Medications Yes No
Investigations Yes No
Social history Yes No
Age of patient Yes No
Cord / cauda equina questions Yes No
Current employment Yes No
Nursing student full time Yes No

Other Questions:

Additional Notes:

236
APPENDIX 5. EXPERT PARTICIPANT RECRUITMENT LETTERS

Associate Professor Darren A. Rivett

Head, Discipline of Physiotherapy

School of Health Sciences

Faculty of Health
University Drive, Callaghan
NSW 2308 Australia
Phone: +61 2 4921 7821
Fax: +61 2 4921 7902
Email: [email protected]

Date _________

Dear _______________,

A research project is currently being undertaken at The University of

Newcastle titled “The exploration of the physiotherapy clinical reasoning
process”. We require participants who are experts in the musculoskeletal
physiotherapy field. The inclusion criteria for expert requires participants to:

 Have more than ten years physiotherapy clinical experience in the

musculoskeletal field
 Be a titled member of Musculoskeletal Physiotherapy Australia (MPA)
 Have at least two years clinical experience following the postgraduate
qualification enabling titled membership
 Be clinically practicing in musculoskeletal physiotherapy at present

Your details have been obtained from a search of local physiotherapists in

the MPA Directory of Titled Members Handbook (2004-2005).

I would like to invite you to participate in this research project. The project
requires two hours participation time on a single occasion. The study will be
located at the School of Health Sciences, The University of Newcastle, at the
Callaghan campus in Newcastle. The time available for participation is
flexible to fit with your current work schedule.

237
If you meet the above criteria and would like further information about this
study, please contact research higher degree student Peter Miller on
[email protected] or 02 4921 6879. Your assistance will be
gratefully received and will hopefully lead to a better understanding of the
physiotherapy clinical reasoning process and improved professional
physiotherapy education. 1

Yours Sincerely

Peter Miller
BPhty, GC HS (Educ)
Research Student

A/Prof Darren Rivett

BAppSc(Phty), MAppSc(ManipPhty) PhD
Project Supervisor

Rosemary Isles
BPhty(Hons); GradCertEduc(Tertiary)
Project Co-supervisor

1
Complaints about this research:
This project has been approved by the University’s Human Research Ethics
Committee (Approval No. H-149-1105). Should you have concerns about
your rights as a participant in this research, or you have a complaint about
the manner in which the research is conducted, it may be given to the
researcher, or, if an independent person is preferred, to the Human Research
Ethics Officer, Research Office, The Chancellery, The University of
Newcastle, University Drive, Callaghan NSW 2308, telephone (02)
49216333, email [email protected]

238
 Associate Professor Darren A. Rivett
Head, Discipline of Physiotherapy

School of Health Sciences

Faculty of Health
University Drive, Callaghan
NSW 2308 Australia
Phone: +61 2 4921 7821
Fax: +61 2 4921 7902
Email: [email protected]
Date _________

Dear _______________,

This is a follow up letter relating to the research project at The University of

Newcastle titled “The exploration of the physiotherapy clinical reasoning
process”. We still require expert physiotherapists to complete the project and
note that you have not responded to date. The inclusion criteria for expert
requires participants to:

 Have more than ten years physiotherapy clinical experience in the

musculoskeletal field
 Be a titled member of Musculoskeletal Physiotherapy Australia (MPA)
 Have at least two years clinical experience following the postgraduate
qualification enabling titled membership
 Be clinically practicing in musculoskeletal physiotherapy at present

Your details have been obtained from a search of local physiotherapists in

the MPA Directory of Titled Members Handbook (2004-2005).

I would like to invite you to participate in this research project. The project
requires two hours participation time on a single occasion. The study will be
located at the School of Health Sciences, The University of Newcastle, at the
Callaghan campus in Newcastle. The time available for participation is
flexible to fit with your current work schedule.

If you meet the above criteria and would like further information about this
study, please contact research higher degree student Peter Miller on

239
[email protected] or 02 4921 6879. Your assistance will be
gratefully received and will hopefully lead to a better understanding of the
physiotherapy clinical reasoning process and improved professional
physiotherapy education.

If we have not received a response from you regarding this project, a follow
up phone call will be made two weeks after the above date to ensure you
have received this letter. If you do not wish to be contacted please ring
Shirley Parker (Discipline of Physiotherapy Administration) on 02 4921 7904
to prevent the call. 1

Yours Sincerely

Peter Miller
BPhty, GC HS (Educ)
Research Student

A/Prof Darren Rivett

BAppSc(Phty), MAppSc(ManipPhty) PhD
Project Supervisor

Rosemary Isles
BPhty(Hons), GradCertEduc(Tertiary)
Project Co-supervisor

1
Complaints about this research:
This project has been approved by the University’s Human Research Ethics
Committee (Approval No. H-149-1105). Should you have concerns about
your rights as a participant in this research, or you have a complaint about
the manner in which the research is conducted, it may be given to the
researcher, or, if an independent person is preferred, to the Human Research
Ethics Officer, Research Office, The Chancellery, The University of
Newcastle, University Drive, Callaghan NSW 2308, telephone (02)
49216333, email [email protected]

240
APPENDIX 6. ADVERTISEMENT FOR NOVICE PARTICIPANTS

The exploration of the physiotherapy clinical reasoning process

(Version 2, 1.2.06)

Research study investigating clinical reasoning

Recently graduated physiotherapists are invited to participate in a research

study conducted by researchers from The University of Newcastle. If you
are within your first year of clinical practice after finishing an undergraduate
or entry-level masters physiotherapy program, you are eligible to participate
in the project titled The exploration of the physiotherapy clinical reasoning
process. This project offers you exposure to unique clinical reasoning
research and the opportunity to reflect on your own clinical reasoning during
a subjective assessment (history). Your assistance will be gratefully
received and will hopefully lead to a better understanding of the
physiotherapy clinical reasoning process and improved professional
physiotherapy education.

The study requires two hours participation time on a single occasion. The
study will be conducted within the Discipline of Physiotherapy at The
University of Newcastle, Callaghan Campus. The times available to
participate are flexible to suit your schedule.

If you would like to know more about this study, please contact research
higher degree student Peter Miller on [email protected] or 02
4921 6879.

241
 APPENDIX 7. INFORMATION STATEMENT & PARTICIPANT CONSENT FORM

Associate Professor Darren A. Rivett

Head, Discipline of Physiotherapy

School of Health Sciences

Faculty of Health
University Drive, Callaghan
NSW 2308 Australia
Phone: +61 2 4921 7821
Fax: +61 2 4921 7902
Email: [email protected]

Information Statement for the Research Project:

The exploration of the clinical reasoning process
(Version 2, 10/12/05)

You are invited to take part in the research project identified above which is being conducted
by Peter Miller, as part of his Master of Medical Science (Physiotherapy) under the
supervision of A/Prof Darren Rivett and Rosemary Isles from the Discipline of Physiotherapy
at The University of Newcastle.

What is the purpose of the study?

The purpose of this research is to explore the clinical reasoning processes used by
physiotherapists in clinical practice. It aims to better understand the methods of decision-
making used by musculoskeletal physiotherapy clinicians with differing levels of experience.
The results could further enhance educational design within undergraduate and postgraduate
physiotherapy programs, in addition to further refining methods of self-directed learning and
professional development for clinicians.

Who can participate?

Physiotherapists who are currently working in clinical practice are being recruited for this
study. Potential expert participants have been identified via the published directory of titled
members handbook for Musculoskeletal Physiotherapy Australia (2004-2005). To be eligible
to participate you must meet the criteria for one of the following groups.

Expert Physiotherapists are required to:

 Have more than ten years physiotherapy clinical experience in the musculoskeletal field
 Be a titled member of Musculoskeletal Physiotherapy Australia (MPA)
 Have at least two years clinical experience following the postgraduate qualification
enabling titled membership
 Be clinically practicing in musculoskeletal physiotherapy at present.

Novice Physiotherapists are required to:

 Have completed a recognised physiotherapy qualification and be registered with the NSW
Physiotherapists Registration Board
 Have less than one year of clinical experience as a physiotherapist.

What choice do you have?

242
Participation in this research is voluntary. There is no obligation for you to participate in this
research study. Only those people who give their informed consent will be included in the
project. Whether or not you decide to participate, your decision will not disadvantage you in
any way. If you decide to participate, you may withdraw from the study at any time. You are
not required to give any reasons for withdrawal.

What will you be asked to do?

Physiotherapists agreeing to participate in this study will be asked to perform a subjective

examination / history as they normally would do in clinical practice, of a trained actor
simulating a real patients history. This will be videorecorded and used to facilitate your recall
during a subsequent interview with the student researcher. This interview will involve watching
the video of the assessment and reporting your thoughts about the case to explore the clinical
reasoning processes used during the assessment. The interview will be audio recorded and
transcribed without any identifying information. Your involvement in the study would take
approximately 2 hours on a single occasion. This will take place at the School of Health
Sciences Research Laboratory (HC35) within the Hunter Building of The University of
Newcastle, Callaghan Campus.

You will be able to review the video and audio recording and / or interview transcripts from
your participation. You may edit or erase your contribution and withdraw from the study at any
stage.

What are the risks and benefits of participating?

Completing the study offers you the opportunity to reflect on your clinical reasoning process
within a physiotherapy assessment. The results will be available on completion of the study
via professional seminar locally. This will take place approximately 12 months from the
commencement of data collection. You may also request a written summary of the study
results.

Risks to participating in this research are minimal. The assessment involves a routine
subjective examination (history) consistent with your current clinical practice. The interview
process requires recall of your thoughts from the assessment and may take up to an hour.
This will be conducted whilst seated at a table, however you may stand at any time during the
interview.

How will your privacy be protected?

The information collected during participation will be strictly confidential. Only the researchers
named on this information statement will have access to identifiable data during analysis. The
interview will be professionally transcribed in its de-identified form. Following analysis the
written data will be identifiable only by a study number. All data, including video and audio
recordings, will be securely stored during the project. The information is required to be kept
for a period of 5 years following the completion of the study, and will be destroyed after this
period.

How will the information collected be used?

The results of this study will form part of the thesis of the student researcher and will be
submitted for publication in scientific journals and presentation at professional conferences.
Individual participants will not be identified in any reports or presentations arising from the
project. Feedback on results of the study will also be presented locally at professional
seminars on completion of the final data analysis.

243
What do you need to do to participate?

Please read this Information Statement and be sure you understand its contents before you
consent to participate. If there is anything you do not understand, or you have questions
regarding the project, please contact the researchers directly:

Peter Miller: Tel. 4921 6879, email [email protected]

A/Prof Darren Rivett: Tel. 4921 7821, email [email protected]
Rosemary Isles: Tel. 4921 2041, email [email protected]

If you are willing to participate in this research, please complete the attached consent form
and return it to Peter Miller at The Discipline of Physiotherapy, School of Health Sciences,
The Faculty of Health, University Drive, Callaghan NSW 2308. I will then contact you to
arrange a convenient time for you to participate in the study.

Thankyou for considering this invitation. 1

Sincerely,

Peter Miller
BPhty, GC HS (Educ)
Research Student

A/Prof Darren Rivett

BAppSc(Phty), MAppSc(ManipPhty) PhD
Project Supervisor

Rosemary Isles
BPhty(Hons); GradCertEduc(Tertiary)
Project Co-supervisor

1
Complaints about this research:
This project has been approved by the University’s Human Research Ethics Committee
(Approval No. H-149-1105). Should you have concerns about your rights as a participant in
this research, or you have a complaint about the manner in which the research is conducted,
it may be given to the researcher, or, if an independent person is preferred, to the Human
Research Ethics Officer, Research Office, The Chancellery, The University of Newcastle,
University Drive, Callaghan NSW 2308, telephone (02) 49216333, email Human-
[email protected]

244
 Associate Professor Darren A. Rivett
Head, Discipline of Physiotherapy

School of Health Sciences

Faculty of Health
University Drive, Callaghan
NSW 2308 Australia
Phone: +61 2 4921 7821
Fax: +61 2 4921 7902
Email: [email protected]

Consent Form for the Research Project:

The exploration of the clinical reasoning process
(Version 1, 21/10/05)

I have read the information on the research project “The exploration of the
physiotherapy clinical reasoning process”, a study that involves taking the history of a
patient followed by an audio taped interview to recall my thought processes from the
assessment. Peter Miller, Master of Medical Science (Physiotherapy) Research
Candidate, is conducting this project under the supervision of Associate Professor
Darren Rivett (Principal Supervisor) and Rosemary Isles (Co-supervisor) from The
University of Newcastle.

All questions have been answered to my satisfaction. I understand that the study will
be carried out as described in the information statement, a copy of which I have
retained.

 I agree to participate in this investigation.

 I understand that I can access my video and audio-recorded information and
transcripts at any time.
 I understand that I can withdraw from this research at any time without reason or
penalty.
 All information from the project will remain confidential.
 I understand that data obtained from the study is required to be kept for 5 years
following the completion of the project, and will be destroyed after this period.
 My identity will not be revealed to anyone other than the researchers named on
this sheet.

Print Name: ____________________

Signature: ____________________
Date: ____________________

Contact Address: _________________________________________

_________________________________________
Contact Telephone: _________________________________________

245
APPENDIX 8. ETHICS VARIATION

246
APPENDIX 9. STUDY SETUP PROCEDURES & EQUIPMENT

Audio Equipment:
 Sony M727V microcassette recorder
 Olympus XD60 and Sanyo MC-60 microcassettes (60 minutes)
 Sanyo memo-scriber TRC-6030 (transcribing system)
Video Equipment:
 Samsung digital video recorder VP-D21i
 Canon MVX330i digital video camcorder
 Video tripod
 5 metre fire wire cable compatible with IEEE1394 Port
 Sony & JVC mini digital video cassettes
 Verbatim external microphone
Hardware:
 Compaq 800 notebook
 Dell inspiron 6400 notebook computer
 Notebook computer compatible audio headsets (2 sets) with dual connector
Software:
 Microsoft Windows XP
 Windows Movie Maker version 5.1
Furniture:
 One metre square table available for participant to make assessment notes
 Two chairs

247
Simulated Patient Assessment

Procedure
1. Preparation
 Assessment room setup with 2 chairs and desk space if required
 Video camera setup on tripod 2 metres from chairs
 Video camera DC power supply attached
 Video camera tape check
 Microphone connected to video camera and placed 1 metre from chairs
 Connecting cable attached from notebook computer to video camera
 Laptop setup in adjacent room with DC power supply
 USB mouse attached to laptop
 Windows Movie Maker program open
 Audio recorder with tape set to 1.2cm speed and placed on desk

2. Recording process (once participant and simulated patient are seated)

 Audio recorder switched on to record
 Video camera set to ‘recorder’
 Record button pressed to record onto video camera cassette
 Recording commenced in Movie Maker – recording directly onto laptop
computer

3. On completion of recording
 On completion of assessment, the laptop recording is stopped then saved
as file in ‘video files’ (participant 1, 2, 3, 4, etc)
 The video camera is then stopped
 The audio recorder is then stopped
 Remove recording cassettes and label with participant number and date

248
Audio Taped Interview

Procedure
1. Preparation
 Interview room setup with two chairs and desk with notebook computer
 New micro-cassettes placed in audio and video recorders
 Audio tape recorder positioned in front of participant and researcher
 Windows Movie Maker program open
 Video clip of patient assessment opened and paused
 Video camera setup on tripod 2 metres from chairs
 Video camera DC power supply attached
 Microphone connected to video camera and placed 1 metre from chairs
 Headsets connected to notebook computer via dual adapter

2. Interview recording process

 Commence recording on the video recorder
 Commence recording on the audio tape recorder
 Commence the interview by stating the participant number on recording
then commence play of the video clip of the recorded assessment
 Mouse control used to pause the recording at any stage
 Any stop of the video requires a time stop to be stated

3. On completion of recording
 The video and audio tape recording devices are stopped
 Remove recording cassettes and label with participant number and date
 The video tape of the interview downloaded onto the notebook computer
 The audio tape recording of the interview sent to the transcriptionist

249
APPENDIX 10. PARTICIPANT ORIENTATION SHEETS (provided separately)

Orientation to the patient assessment (Version 1, 21.10.05)

I am investigating the clinical reasoning process used in physiotherapy. I

would like you to conduct a subjective assessment (or history) with this
patient. You will not be required to undertake a physical examination or treat
this patient. A video camera will record the assessment and the videotape
will be used to help you recall your thoughts at the time of the assessment
during the following discussion. Please assess the patient as normally would
in your clinical practice.

Orientation to the interview (Version 1, 21.10.05)

To further explore your clinical reasoning process, I will now ask you some
questions relating to the assessment you have undertaken. It is important to
understand that there is no correct answer to these questions. The
questions will explore your thinking at the time of the assessment.

You will be observing the video of your patient assessment to prompt recall
of your thoughts at different times during the assessment. Try to describe
what was going on in your mind at the actual time and not thoughts or
decisions from afterwards. In other words, it is important that you try to recall
your thinking at each step of the assessment and not to be influenced by
information you may have obtained later.

The term hypothesis is used to refer to your understanding or explanation of

the case. You may describe your hypothesis or understanding in any way
that suits you.

I can pause the video whenever you would like to discuss your thoughts or
observations from the assessment. There will also be times where I will
pause the video to ask you what you were thinking at that point in the
assessment.

250
APPENDIX 11. INTERVIEW PROTOCOL SHEET

The exploration of the physiotherapy clinical reasoning process

Version 1, 21.10.05

This semi-structured interview protocol utilises the following questions to

prompt discussion during the interview. The video will be paused during any
interview discussion.

 What were you thinking at this stage?

 Were you thinking anything else at this stage? What were you thinking?
 Did that information help you at all? How did it help?
 Did that information assist your understanding of the case at the time? How
did it help?
 Why did you ask that? What information did you gain from that line of
questions?

Timing of Video Stops:

The video is to be paused when requested by the participant, after each group
of similar questions, or after a maximum period of 1 minute continuous play.
On completion of the video observation, the following structured questions will
be asked unless the information has already been provided earlier in the
interview.

First Hypothesis Questions:

1. When did you first generate an understanding of the case that you would
call a hypothesis?
2. How would you describe this first hypothesis?
3. What features of the case led to this first hypothesis?
4. Did you have any other hypotheses about the case at the same time? If so,
what were they?

251
Final Hypothesis and Wind-up Questions:
5. What is your final understanding or hypothesis of the presentation based
on all the information?
6. Were you thinking anything else during the assessment that we haven’t
discussed yet?
7. What physical examination tests are you planning to do at this stage?

Study and Method Related Questions:

8. How often on average do you see this type of patient?
9. Did the location of the assessment influence your assessment compared
to normal?
10. Did the presence of the video camera influence your assessment
compared to normal?
11. Was your assessment conducted more thoroughly than normal?

252
APPENDIX 12. FINAL CODEBOOK

Code name Code Sub-code Code description

Data Collection DC Therapist reports collecting information in a routine manner
without the data relating to a hypothesis (coded even if
prompted in interview)
Hypothesis Related H Hypothesis categories: Therapist states an understanding of the case in any of the
1. Activity & participation hypothesis categories. This may be several competing
2. Patients perspective / psychosocial possibilities or a single clear hypothesis.
factors
3. Pathobiological mechanisms Any lay level of response, where no interpretation has occurred
4. Physical impairments & associated should not be coded as a hypothesis e.g. descriptive comments
structure / tissue sources such as “the pain was fairly strong”
5. Contributing factors
6. Precautions & contraindications
7. Management & treatment
8. Prognosis
9. Non-specific

Knowledge K Participant refers to their knowledge in an attempt to apply it to

the problem at hand.

Self Awareness SA Therapist verbalises awareness of their own thinking. These

metacognitive reflections may relate to data, process,
hypotheses, decisions or knowledge.
Planning P 1. Examination Therapist verbalises their use of subjective data during the
2. Management history to plan further examination or Rx procedures
Significant Case SCF Therapist highlights the relevance / significance of data
Feature obtained to case at hand

253
 Code name Code Sub-code Examples of codes
Data Collection DC “just trying to gather information at this point”

Hypothesis H Hypothesis categories:

Related 1. Activity & participation 1. “she is still able to do a fair amount of activity despite being in pain”
2. Patients perspective / 2. “she seemed to be coping OK with the pain but is concerned about her next
psychosocial factors clinical”
3. Pathobiological 3. “I was thinking along the lines of an inflammatory wind up of the neural system”
mechanisms 4. “the problem could be a lumbar disc or the SIJ”
4. Physical impairments & 5. “there is a muscle control issue”
structure / tissue sources 6. “I would be careful not to leave her lying prone for extended periods”
5. Contributing factors 7. “active exercise targeting the stabilising muscles will be useful”
6. Precautions & 8. “I was thinking that we weren’t aiming to get pain free but back to preclinical
contraindications levels of pain”
7. Management & treatment 9. “I think this is an acute-on-chronic problem rather than an acute problem”
8. Prognosis
9. Non-specific
Knowledge K  “it’s so unlikely for kids to have long standing non-specific back pain and
unlikely to have a disc injury … I see enough kids to make that history unusual”
 “based on the last (MPA) conference … believe disc is aggravated by sitting”
 “from my experience at uni … she may need some time off ... special
consideration”
Self Awareness SA “I was aware at that stage that I needed more information to confirm …”
“that info didn’t fit and could have sent me down a completely different track”
“that wasn’t what I was expecting to hear”
“”I like to keep (the history) as open as possible”
Planning P 1. Examination “I’d be testing those flexion vs extension movements”
2. Management “I was using it to formulate what I’m going to do in the physical examination”
“I might need to use a flexion technique”
Significant Case SCF “that initial episode (of pain) is significant”
Feature

254
APPENDIX 13. PATTERN RECOGNITION IDENTIFICATION TOOLS

Time stop identification tool for pattern recognition

Participant number ________ Researcher initials _________
Time 1. Hypothesis formed Ho. 2. 3. 4. 5. Comments / quotes
Cat. SCF Kn Exp Mx

Key for use: Interview time stops are labelled in the time column as the exact assessment time stated in transcript data.
Hypotheses formed during each time stop of the participant interview are documented and hypothesis
category (Ho. Cat.) labelled relative to each hypothesis (numbered 1-9).
Information on features numbered 2 to 5 obtained via each specific code and quotations listed along with
researcher comments in ‘comments / quotes’ column. Yes or no placed in each column (2 to 5) for each
time stop.
SCF = Significant case feature Kn = Knowledge Exp = Experience Mx = Management

255
Overall pattern recognition scoring tool
Participant number ________ Researcher initials _________
Feature of PR Yes / No Evidence Comments
1. Central hypothesis
formed

Actual time of central ___ min’s

hypothesis formation as ___ sec’s
a percentage of the total
assessment time ____ % of
overall time
2. Significant case
features

3. Professional craft
knowledge

4. Prior clinical
experience of this
case

5. Management stated

Key for use: A clear predominant and central hypothesis must be formed to state pattern recognition has been used.
For each ‘yes’ response beside numbered (1 – 5) items a score of one is provided.
Total score is the number of ‘yes’ responses out of a maximum total of five.

256
APPENDIX 14. SELF-AWARENESS SUB-CODE EXAMPLES
These examples of the self-awareness coded text identify the sub-code name in bold
text within parentheses at the end of each quotation:
Expert F: I would have expected someone with this history at that stage to say they
had some stiffness in the morning (Prediction)
Expert B: there are some pieces of information which are just sort of like um which
are there sitting in the background and others which are really channelling
the flow through (Process)
Expert F: I just think it’s interesting to how people describe pain … I don’t really
know whether it makes any difference but I just think its interesting
(Interesting)
Expert J: So that doesn’t really tell me what the, what the cause is, what the
diagnosis is (Diagnosis)
Expert G: I’m perhaps a little bit surprised that she’s only taking Panadol every now
and then (Surprise)
Expert F: it hadn’t really fitted and I wasn’t really quite sure what I thought was
wrong with her and by this stage I normally have a very clear picture of
what I think’s wrong with someone whether it right or wrong (Uncertainty)
Expert J: I usually take the history and then … at the end I really start thinking,
putting it together. I usually get all the facts before I start putting it
together (Deferred integration)
Novice N: the way that it kind of panned out made me focus a lot more on the, the
structures of it as a source of the pain rather than anything else
(Direction)
Novice S: That threw me a bit. Um with the sitting um easing it. Um but with a, a
little bit more questioning I was able to sort it out a bit more.
(Uncertainty)
Novice M: when we started going into how long standing, she can stand for and then
when the pain comes on. I started thinking I really need to clarify this
cause it’s quite important. (Significance)
Novice R: I hadn’t really at this staged locked myself into any sort of hypothesis.
I was really just sort of pulling it all, just making a stew, just getting all
that information in there and some, seeing what I thought in the end.
(Deferred integration)
APPENDIX 15. FINAL PARTICIPANT HYPOTHESES

Participant Summary Evidence Categories

Expert A Instability I’d expect all directions probably would be painful. With none outstanding 4 Physical impairments &
more than other … I’m thinking instability … there’s no major pathology associated structure /
you know untoward going on … My picture is that her disc is probably tissue sources
extra squashy. This is a very crude you know analogy sort of model rather 5 Contributing features
than a reality. You know the disc is probably very squashy and they’re
sort of trying to bulge out all around and the supporting muscles just can’t
hold them

Expert B Spondylolisthesis I would still be concerned about a spondylolisthesis … it’s not a sort of 4 Physical impairments &
like a um ah a really irritable type problem associated structure /
tissue sources
6 Precautions and
contraindications

Expert C Mechanical pelvic I think there’s an underlying sort of asymmetry. An insidious history of the 4 Physical impairments &
asymmetry right pelvis I would expect to find on the assessment … a secondary associated structure /
pelvic asymmetry which is probably the underlying cause of the problem tissue sources
which makes it look like the pattern you’re seeing here … there’s a 5 Contributing features
mechanical aspect to it.

Expert D Spondylolisthesis she’s got a poor understanding of what her problem is, in fact probably 2 Patients perspective /
virtually no understanding of her problem … my hypothesis of it being psychosocial
some sort of either structural instability … spondylolisthesis 4 Physical impairments &
associated structure /
tissue sources
5 Contributing features

258
Participant Summary Evidence Categories
Expert E Instability there’s certainly a few directions that I’d head into … I think I’d probably 4 Physical impairments &
go towards something sort of instability or canal stenosis or something associated structure /
that’s central … she could have scoliosis or leg length inequality tissue sources
5 Contributing features

Expert F Nil clear stated no I didn’t (have an understanding of the case), I’m still quite confused Nil
about her. I don’t thing she’s typical

Expert G Motion segment I think it’s um -- probably low lumbar spine, motion segment sort of I 3 Pathobiological
dysfunction / guess pathology. Um dysfunction um with a neurogenic component … mechanisms
neurogenic / with um a poor muscle, I guess poor stability system that’s contributing to 4 Physical impairments &
instability that associated structure /
tissue sources
5 Contributing features

Expert H Instability / joints I think probably ah more moderate to low irritability um so I can examine it 4 Physical impairments &
fairly fully um I think there’s prob um -- there’s probably an instability associated structure /
component that I need to address in terms of checking out her core tissue sources
stability um muscles. So that’s, yes I’m, I’m expecting that I’ll find that 5 Contributing features
there’s some um stiffness and painful joints at the back of her spin which I 6 Precautions and
suspect I would be able to treat. Um and then maybe give her stability contraindications
type exercises. 7 Management

259
Participant Summary Evidence Categories
Expert I Spondylolisthesis Mechanically behaving back pain that has gone -- stemmed from an 3 Pathobiological
original injury as an 11 year old um that’s been symptomatically managed mechanisms
and in a period of a lot of a lot of overload with this recent clinical 4 Physical impairments &
placement … And structurally I’m thinking that there are enough, enough associated structure /
reasons to go and have it investigated, further. Um particularly looking for tissue sources
bony, bony changes where there was a bone injury originally … she’s not 5 Contributing features
getting discrete dermatomal symptoms and her aggravating factors aren’t 7 Management
consistent with nerve root type um problems. Um certainly in the back of
my mind I’m leaning strongly toward the possibility there may be some
kind of bony pathology there as well (Bony pathology was stated earlier in
the interview as spondylolisthesis)

Expert J Disc or joint I couldn’t say at this stage whether it’s more discy or facet, and I would 4 Physical impairments &
get back from my examination I would get more information from the associated structure /
examination about that. Whether it was extension, with her active tissue sources
movements, if it was extension that mainly brought on a pain. And if she 5 Contributing features
wasn’t very irritable, you know I could put her back in the quadrant 6 Precautions and
position and see if it’s really closing down the facet joints. contraindications

Novice K Disc / sacro-iliac I’m either thinking ah discogenic, or possibly SIJ … and ah postural as 4 Physical impairments &
joint / postural well associated structure /
tissue sources
5 Contributing features

Novice L Disc my yeah hypothesis is like a disc type injury which is aggravated by 4 Physical impairments &
loading … in a like standing position … Thinking a disc injury is impacting associated structure /
on the nerves and therefore causing some referring type things tissue sources

260
Participant Summary Evidence Categories
Novice M Postural I think I’ve started to develop an understanding. I’m definitely thinking that 5 Contributing features
there’s a postural component, um to her pain and I guess I’m not really
thinking about so much about structurally what’s going on

Novice N Lack of support / I’d say that she’s ah she’s had a history, a long history of back pain, um 5 Contributing features
restriction possibly begun with a, ah with a, a fall down the stairs. Um but I’d say
she’s definitely got a -- ah lack of support around the area. Um there
could be some actual derangement or dysfunction in the actual structures
but um I’d be thinking that there’s just a lack of support for the areas that
is worse when she is weight bearing. I would be expecting restriction um
into flexion. Um -- and possibly … extension. I guess I would have
thought that there would be some restriction … if she flexed all the way
down to her toes I’d be surprised … it wouldn’t fit what I was going down

Novice O Sacro-iliac joint a few ideas, nothing really specific. Um probably looking at like a, an SIJ 4 Physical impairments &
(SIJ) kind of ah especially if standing for prolonged periods of time um and associated structure /
with the nature of having a fall as well, um with activities like netball um tissue sources
repeated jarring that sort of thing

Novice P Instability I still think it’s a, a instability of the lumbar spine … there wasn’t kind of 5 Contributing features
any particular movement or loading strategy in a particular direction
which, which um tend to flare it up. It was, there was multiple … well
basically what I’m kind of getting at there is just the inability of the
muscular control system to hold the, the lumbar spine within it’s neutral
position

261
Participant Summary Evidence Categories
Novice Q Instability I guess really there’s two possibilities … that sort of sit in my mind … that 4 Physical impairments &
lumbopelvic instability in that area, um poor activation transversus associated structure /
abdominus, it certainly would be interested to see her ability to activate tissue sources
um to there and see if there was any alteration in her symptoms. Um, the 5 Contributing features
other, other possibility is ah like a stenosis compression of a nerve root.
Um given that her extension moments … especially with the sitting, sitting
in the slouched position, the opening up of the facet joints. Um, maybe a
little bit unusual in both sides at the same time

Novice R Instability you know she’s getting this hip and lumbar spine pain and that referral 4 Physical impairments &
could be due to um sciatic impingement through over performance or associated structure /
underperformance depending on her makeup. Um, so that was something tissue sources
that I was juggling. Um, but also there could be you know she might have 5 Contributing features
other things going on. Um facet joint irritations or general instability, global
instability, um that she can’t control along her spine and hence is getting
gross movements with her um increased load that she is putting under it

Novice S Disc > joint sort of chasing a lumbar disc as a possible structure. Um but also the, the 4 Physical impairments &
referral pain down both legs and into her knee I hadn’t um clearly associated structure /
determined whether that was, was coming from a structure in her lower tissue sources
back, or not. So I wanted to look at things like um her piriformis, her SIJ a
little bit more with some testing, um to try and determine whether I could
figure out if those all were referred pain into her buttocks and thighs and
just below the knee where it related to the one area … a lumbar disc is
one that is standing out um, more than others. Although it’s -- just the
lumbar spine specifically that is standing out some more, so I wouldn’t,
um, I wouldn’t rule out ah some Z joint involvement um or some, some
muscular involvement either

262