See discussions, stats, and author profiles for this publication at: https://www.researchgate.

net/publication/23156530

Discourse analysis

Article in The BMJ · February 2008

DOI: 10.1136/bmj.a879 · Source: PubMed

CITATIONS READS

160 49,114

3 authors, including:

Ayelet Kuper Scott Reeves

University of Toronto Kingston University & St George's, University of London
115 PUBLICATIONS 5,525 CITATIONS 386 PUBLICATIONS 25,460 CITATIONS

SEE PROFILE SEE PROFILE

All content following this page was uploaded by Scott Reeves on 22 May 2014.

The user has requested enhancement of the downloaded file.

PRACTICE For the full versions of these articles see bmj.com

QUALITATIVE RESEARCH
Discourse analysis
Brian David Hodges,1 Ayelet Kuper,2 Scott Reeves3

1
Department of Psychiatry, Wilson This articles explores how discourse What is empirical discourse analysis?
Centre for Research in Education, Researchers using empirical discourse analysis4 do not
University of Toronto, 200
analysis is useful for a wide range of use highly structured methods to code individual
Elizabeth Street, Eaton South 1-
565, Toronto, ON, Canada
research questions in health care and the words and utterances in detail. Rather, they look for
M5G 2C4 health professions broad themes and functions of language in action using
2
Department of Medicine, approaches called conversation analysis (the study of
Sunnybrook Health Sciences
Previous articles in this series discussed several “talk-in-interaction”)6 and genre analysis (the study of
Centre, and Wilson Centre for
Research in Education, University methodological approaches used by qualitative recurrent patterns, or genres of language that share
of Toronto, 2075 Bayview researchers in the health professions. This article similar structure and context—such as the case report,
Avenue, Room HG 08, Toronto, the scientific article).7
ON, Canada M4N 3M5
focuses on discourse analysis. It provides background
3 information for those who will encounter this approach Conversation analysis and genre analysis give more
Department of Psychiatry, Li Ka
Shing Knowledge Institute, Centre in their reading, rather than instructions for conducting prominence to sociological uses of language than to
for Faculty Development, and such research. grammatical or linguistic structures of words and
Wilson Centre for Research in sentences and are used to study human conversations
Education, University of Toronto,
200 Elizabeth Street, Eaton South What is discourse analysis? or other forms of communication in order to elucidate
1-565, Toronto, ON, Canada Discourse analysis is about studying and analysing the the ways in which meaning and action are created by
M5G 2C4 individuals producing the language.4 Lingard and
uses of language. Because the term is used in many
Correspondence to: B D Hodges
[email protected] different ways, we have simplified approaches to colleagues, for example, studied communication
discourse analysis into three clusters (table 1) and between nurses and surgeons during 128 hours of
Cite this as: BMJ 2008;337:a879 observing 35 different procedures in the operating
doi:10.1136/bmj.a879
illustrated how each of these approaches might be used
to study a single domain: doctor-patient communica- room and categorised recurrent patterns of commu-
tion about diabetes management (table 2). Regardless nication. They then used their findings to draw links
of approach, a vast array of data sources is available to between interpersonal tensions, the use of language,
the discourse analyst, including transcripts from inter- and the occurrence of errors in the operating room.8
views, focus groups, samples of conversations, pub- Genre analysis is presented in detail in box 1.
lished literature, media, and web based materials.
What is critical discourse analysis?
What is formal linguistic discourse analysis? Researchers in cultural studies, sociology, and phi-
The first approach, formal linguistic discourse analysis, losophy use the term critical discourse analysis to
involves a structured analysis of text in order to find encompass an even wider sphere that includes all of the
general underlying rules of linguistic or communica- social practices, individuals, and institutions that make
tive function behind the text.4 For example, Lacson and it possible or legitimate to understand phenomena in a
colleagues compared human-human and machine- particular way, and to make certain statements about
human dialogues in order to study the possibility of what is “true.” Critical discourse analysis is particularly
using computers to compress human conversations concerned with power and is rooted in “constructi-
about patients in a dialysis unit into a form that vism.” Thus the discourse analyses of Michel Foucault,
physicians could use to make clinical decisions.5 They for example, illustrated how particular discourses
transcribed phone conversations between nurses and “systematically construct versions of the social
25 adult dialysis patients over a three month period and world.”4 Discourse analysis at this level involves not
This is the fourth in a series of six
articles that aim to help readers to coded all 17 385 words by semantic type (categories of only the examination of text and the social uses of
critically appraise the increasing meaning) and structure (for example, sentence length, language but also the study of the ways in which the
number of qualitative research
articles in clinical journals. The word position). They presented their work as a “first very existence of specific institutions and of roles for
series editors are Ayelet Kuper and step towards an automatic analysis of spoken medical individuals to play are made possible by ways of
Scott Reeves. dialogue” that would allow physicians to “answer thinking and speaking.
For a definition of general terms
relating to qualitative research, questions related to patient care by looking at Foucault’s study of madness, for example, uncov-
see the first article in this series. [computer generated] summaries alone.”5 ered three distinct discourses that have constructed

570 BMJ | 6 SEPTEMBER 2008 | VOLUME 337

 PRACTICE

Table 1 | Three approaches to discourse analysis

Orientation to discourse Sources of data Analysis
Formal linguistic discourse analysis (such as Samples of written or oral language and texts Microanalysis of linguistic, grammatical, and
sociolinguistics)1 semantic uses and meanings of text
Empirical discourse analysis (such as Samples of written or oral language and texts; Microanalysis and macroanalysis of the ways in
conversation analysis, genre analysis)2 and data on the “uses” of the text in social which language and/or texts construct social
settings practices
Critical discourse analysis (such as Foucauldian Samples of written or oral language/texts; and Macroanalysis of how discourses (in many forms)
analysis)3 data on the “uses” of the text in social settings; construct what is possible for individuals and
and data on the institutions and individuals who institutions to think and to say
produce and are produced by the language texts

what madness is in different historical periods and in “consumers,” or “survivors” and are made possible by
different places: madness as spiritual possession, specific institutional practices and ways for individuals
madness as social deviancy, and madness as mental to “be.”11
illness.10 In a similarly oriented study, Speed showed In a different context, Stone contrasted the specific
how different discourses about mental health service in discourses used in the education literature for diabetes
use today construct individuals’ identities as “patients,” patients (“patient self care” and “autonomy”) with the
medical literature’s use of doctor centred discourses
(“compliance” and “adherence”). Stone related the
resulting tension (and the important implications for
Box 1 An empirical discourse analysis (genre analysis) of case presentations by medical patients’ behaviours) to the ways in which the roles that
students*
physicians and patients play are historically deter-
This study took place at a tertiary care teaching hospital in Canada. It was conducted in the mined by different and conflicting models of what
context of a medical student rotation in paediatrics. The aim of the study was to gain disease and healing are.12
understanding of how the formal linguistic structure of the case presentation is used in
Finally, Shaw and colleagues used a discourse
academic medical settings.
analysis to illustrate the many ways in which research
The researchers conducted 21 in-depth interviews with medical students and faculty
itself can be defined (for example, by a lay person, a
members. Pairs of researchers also observed 16 oral case presentations as well as the
medical editor, the World Medical Association, a
teaching exchanges that surrounded them. All of these encounters were tape recorded and
transcribed (for a total of 555 pages of text); the transcriptions were iteratively analysed. The
hospital, the taxman) and how these various definitions
analysis was structured to allow themes to emerge from the data (that is, as indicated by are linked to the power and objectives of particular
multiple examples of such themes throughout the data). However, it particularly focused on institutions.13
themes that helped to illuminate the rules around certain modes of case presentation and In these examples of critical discourse analysis, the
on the role of these rules in teaching and learning. language and practices of healthcare professionals and
The study showed a pronounced tension between the educational (“schooling”) uses and institutions are examined with the aim of under-
clinical (“workplace”) functions of case presentations. For example, students saw the case standing how these practices shape and limit the ways
presentation as a school mode and emphasised that they wanted to get through their that individuals and institutions can think, speak, and
presentations without being asked any questions. Faculty, on the other hand, understood conduct themselves. Table 2 illustrates how a critical
the case presentation as a way for professionals to jointly create shared knowledge. Their discourse approach to diabetes education would
cross-purposes affected the effectiveness of faculty feedback to the students about their compare with discourse analyses using other linguistic
case presentations. and empirical approaches to research.
*Description based on study by Lingard et al 9
Although our categorisation (tables 1 and 2) empha-
sises the distinctions between these approaches to
discourse analysis, in practice researchers often use
Box 2 Further reading more than one of the approaches together in a study.
For example, genre analysts may invoke critical
Books
theorists in order to study the origins of the sanctioned
 Fairclough N. Language and power. London: Longman, 1989.
methods of communication, asking, for example,
 Foucault. The archaeology of knowledge and the discourse on language. New York: “What historical and contextual factors led to the
Random House, 1972.
adoption of the scientific journal article as a legitimate
 Jaworski A, Coupland N, eds. The discourse reader. London: Routledge, 1999. form of expression of medical ‘truth’ rather than the
 Kendall G, Wickham G. Using Foucault’s method. London: Sage, 2003. adoption of another format?”
 Mills S. Discourse. London: Routledge, 2004.
Journal articles What should we be looking for in a discourse analysis?
 Barnes R. Conversation analysis: a practical resource in the health care setting. Med Given the wide variety of approaches to discourse
Educ 2005;39:113-5. analysis, the elements that constitute a high quality

study vary. Rogers has argued that some discourse
Ford-Sumner S. Genre analysis: a means of learning more about the language of health
care. Nurse Researcher 2006;14(1):7-17. analysis research suffers from scanty explanation of the
analytical method used.14 Thus one should expect clear
 Roberts C, Sarangi S. Theme-oriented discourse analysis of medical encounters. Med
documentation of the sources of information used and
Educ 2005;39:632-40.
delimitation of data sources3 (including a description of

BMJ | 6 SEPTEMBER 2008 | VOLUME 337 571

PRACTICE

Table 2 | Threeapproachestoaspecificresearchquestion:exampleofdoctor-patientcommunicationsaboutdiabetesmanagement
Orientation to
discourse Question Data collection and analysis Application
Formal linguistic What are the characteristics of linguistic Analyse sentences from a patient Could be used to restructure the linguistic
discourse analysis structures that doctors use to instruct education pamphlet such as “you must structure of patient education materials
patients on medication? control your blood sugar” in terms of rules
of linguistic function in general use
Empirical discourse What is the nature of conversations Record and analyse doctor-patient Could be used to create programmes for
analysis or between doctors and their patients about conversations about diabetes training of health professionals in
conversation diabetes management? management; analyse the kinds of effective doctor-patient communication
analysis utterances commonly used, their strategies
meanings, and their effects in the context
studied
Critical discourse What sociohistorical phenomena have Create an “archive” of data (for example, An analysis of relations of power and
analysis made it possible for individuals in society verbal, text, graphic) illustrating the constructions of “truths” could be used to
to have roles called “doctor” and “patient” nature of these roles and relationships, rethink or reconfigure the roles,
and to create an interaction called the how they came about, and how they have relationships, or institutional practices
“doctor-patient relationship,” in contrast shifted or changed over time such as teaching, work policies, codes of
to what might be possible in other places professional (or patient) behaviour
or times. What enables or constrains this
relationship (such as power, definitions
of “truth”)?

decisions made with regard to selection of groups or human communication to the inner workings of
individuals for interviews, focus groups, or observa- systems of power that construct what is “true” about
tion) and, importantly, a description of the context of health and health care. While these methods are
the study. The method of analysis should be clearly gaining popularity, much remains to be done to
explained, including assumptions made and methods develop a widespread appreciation for the use, funding,
used to code and synthesise data. Finally, given that the and publication of discourse analyses. As a start, we
goal of critical discourse analysis is to illuminate and hope this article will help readers who encounter these
critique structures of power, it is especially important approaches to understand the basic premises of
that researchers describe the ways in which their own discourse analysis. Box 2 offers further reading for
individual sociocultural roles may influence their those interested in learning more or undertaking
perspectives. discourse analytical research.
Contributors: All authors contributed to the conception and drafting of the
Conclusion article and its revisions, and all approved the final version.
Competing interests: None declared.
Discourse analysis is an effective method to approach a Provenance and peer review: Commissioned; externally peer reviewed.
wide range of research questions in health care and the
health professions. What underpins all variants of 1 Harris ZS. Methods in structural linguistics. Chicago: University Press,
discourse analysis is the idea of examining segments, or 1951.
2 Sacks H. Lectures on conversation. Jefferson G, ed. Cambridge, MA:
frames of communication, and using this to understand Blackwell, 1995.
meaning at a “meta” level, rather than simply at the 3 Foucault M. The archaeology of knowledge and the discourse on
language. New York: Random House, 1972.
level of actual semantic meaning. In this way, all of the 4 McHoul A, Grace W. A Foucault primer: discourse, power and the
various methods of discourse analysis provide rigorous subject. New York: New York University Press, 1993.
5 Lacson RC, Barzilay R, Long WJ. Automatic analysis of medical
and powerful approaches to understanding complex dialogue in the home hemodialysis domain: structure induction and
phenomena, ranging from the nature of on-the-ground summarization. J Biomed Informatics 2006;39:541-55.
6 Ten Have P. Medical ethnomethodology: an overview. Human Studies
1995;18:245-61.
7 Ford-Sumner S. Genre analysis: a means of learning more about the
SUMMARY POINTS language of health care. Nurse Researcher 2006;14(1):7-17.
8 Lingard L, Espin S, Whyte S, Regehr G, Baker GR, Reznick R, et al.
Discourse analysis is an effective method for approaching a Communication failures in the operating room: an observational
wide range of research questions in health care and the classification of recurrent types and effects. Qual Saf Health Care
health professions 2004;13:330-4.
9 Lingard L, Schryer C, Garwood K, Spafford M. “Talking the talk”: school
Discourse analysis is about studying and analysing the uses and workplace genre tensions in clerkship case presentations. Med
of language Educ 2003;37:612-20.
10 Foucault M. Madness and civilization; a history of insanity in the age of
A vast array of data sources is available to the discourse reason [Howard R, translation]. New York: Vintage Books, 1988.
(Original work published in 1961.)
analyst 11 Speed E. Patients, consumers and survivors: a case study of mental
The various methods of discourse analysis provide rigorous health service user discourses. Soc Sci Med 2006;62(1):28-38.
12 Stone MS. In search of patient agency in the rhetoric of diabetes care.
and powerful approaches to understanding complex Technical Communication Quarterly 1997:6:201-17.
phenomena, ranging from the nature of on-the-ground 13 Shaw S, Boynton PM, Greenhalgh T. Research governance: where did
human communication to the inner workings of systems of it come from, what does it mean? J R Soc Med 2005;98:496-502.
power that construct what is “true” about health and health 14 Rogers R, Malancharuvil-Berkes R, Mosley M, Hui D, O’Garro JG.
Critical discourse analysis in education: a review of the literature. Rev
care Educ Res 2005;75:365-416.

572 BMJ | 6 SEPTEMBER 2008 | VOLUME 337

 PRACTICE

LESSON OF THE WEEK

Extensive transmission of Mycobacterium tuberculosis
from 9 year old child with pulmonary tuberculosis and
negative sputum smear
S Paranjothy,1 M Eisenhut,2 M Lilley,3 S Bracebridge,4 I Abubakar,5,6 R Mulla,2 K Lack,3 D Chalkley,7 J Howard,4
S Thomas,4 M McEvoy3

CLINICAL REVIEW, p 564 A negative sputum smear does not exclude release of interferon γ by T cells in response to M
tuberculosis specific antigens and are commercially
1
substantial risk of infection from patients available in the United Kingdom. These tests correlate
Department of Primary Care and
Public Health, Clinical with pulmonary tuberculosis better with exposure to M tuberculosis than the Mantoux
Epidemiology Interdisciplinary test.3
Research Group, School of
Medicine, Cardiff University, Patients with pulmonary tuberculosis and either a The three siblings started chemoprophylaxis (three
Cardiff CF14 4YS positive sputum smear or cavitating pulmonary lesions month course of rifampicin and isoniazid). Three
2
Luton and Dunstable Hospitals have been considered to be infectious1 as these indicate children from further extended family had positive T-
NHS Foundation Trust, Luton higher bacterial load accumulating with longer dura-
LU4 0DZ
SPOT tests; two of these had normal chest radiographs
3 tion of infection. The source of infection in outbreaks and started chemoprophylaxis. The third, asympto-
Bedfordshire and Hertfordshire
Health Protection Unit, among children is usually an adult with these features.2 matic child had bilateral hilar lymphadenopathy on
Letchworth, Hertfordshire Children have been considered less likely to transmit chest radiography and started chemotherapy. All five
SG6 1BE infection because they were unlikely to expectorate
4
adults in the immediate and extended family screened
Health Protection Agency (East
infective droplet nuclei. In the absence of a positive positive and required chemoprophylaxis. In the
of England) Regional
Epidemiology Unit, Institute of sputum smear, guidelines do not recommend screen- absence of an apparent adult source of infection
Public Health, University Forvie ing of wider contacts in addition to household among household contacts, screening was extended
Site, Cambridge CB2 2SR
5
contacts,3 although in the United States, contact to include the index case’s school contacts.
Respiratory Diseases
Department, Centre for Infections,
screening is recommended if resources are sufficient.4
Health Protection Agency, London Here, we report extensive transmission of tuberculosis School contacts of the index case
NW9 5EQ in a junior school in Luton, England, from a child with The boy who was the index case attended a junior
6
School of Medicine, Health Policy pulmonary tuberculosis in whom a sputum smear was school with 200 other pupils aged 8 to 12 years.
and Practice, University of East
Anglia, Norwich NR4 7TJ negative. Initially, all contacts within his year group (classes 5
7
Luton Training PCT, Liverpool and 6) who shared lessons and class teachers were
Chest Clinic, Luton LU1 1HH The index case screened using Mantoux tests. In his year group 36% of
Correspondence to: S Paranjothy The index case was a 9 year old boy of Black African pupils were white; 23% were Bangladeshi, Indian, or
[email protected]
ethnicity born in the United Kingdom. He was Pakistani; 14% were Black African or Black Caribbean;
Cite this as: BMJ 2008;337:a1184 investigated by his family doctor because of a six year and 21% were of mixed ethnicity (white and either
doi:10.1136/bmj.a1184 history of recurrent cough evolving into a daily chronic Asian or Black Caribbean parents). Positive Mantoux
cough with night sweats and weight loss for eight weeks tests were confirmed using T-SPOT tests, in accor-
before referral to a paediatrician. The cough was dance with national guidance.3
occasionally productive with one episode of haemop- As the infection rate among these contacts was high
tysis. He continued to attend school while sympto- (30/43=69.7%), screening was extended to include all
matic. A chest x ray film showed a right upper lobe pupils. Owing to the numbers involved and because
consolidation and multiple poorly defined opacities in the school term was coming to an end, pupils were
the right lower lobe (figure). Three sputum collections screened using the T-SPOT test and chest radiography.
were smear negative, and culture showed Mycobacter- Each chest radiograph was read by a paediatrician with
ium tuberculosis, fully sensitive to all drugs tested expertise in tuberculosis and independently read again
(isoniazid, rifampicin, pyrazinamide, ethambutol). He and reported by a consultant radiologist. Children with
started chemotherapy with a two month course of pulmonary parenchymal changes on chest radiog-
pyrazinamide, rifampicin, and isoniazid, followed by raphy were regarded as potentially infectious and
four months of isoniazid and rifampicin. All household investigated by obtaining three sputum and three
contacts of the child were screened for M tuberculosis gastric lavage samples. As we were seeking an adult
infection. source of infection, adults with at least eight hours’
The boy’s three siblings had positive Mantoux cumulative contact with the school (including staff,
readings, positive T-SPOT tests (Oxford Immunotec), support workers, and adult visitors) during the
and normal chest x ray examinations. The T-SPOT test previous eight months were invited for chest
is one of two immunological tests that measure the radiography.

BMJ | 6 SEPTEMBER 2008 | VOLUME 337 573

PRACTICE

The decision to screen wider contacts of the index

case was not consistent with current national
guidelines3 but was driven by the high rate of infection
among the children who were household contacts of
the index case. We were unaware then of the extended
family contact who had previously had a diagnosis of
tuberculosis and had no direct link with the school. We
screened staff and pupils attending the school as we
were seeking an adult source of infection and any other
infected children.
Our results led us to hypothesise that the index case
was the source case for the infected school children.
Several findings supported this hypothesis. Firstly, the
infection rate in the class of the index case was higher
than the rates in the other classes. Secondly, the
identification of an epidemiological and DNA finger-
First chest radiograph of the index case, a 9 year old boy with printing link of the index case with an adult contact who
chronic cough and M tuberculosis grown from sputum samples had had no contact with the school of the index case
supports the hypothesis that the index case was the
source case of the school outbreak. The shared strain
type between the two children at the junior school
Altogether, 85 children (42% of all school pupils) had
provided good evidence that child to child transmis-
positive T-SPOT tests; of these, 16 had hilar lympha-
sion occurred.
denopathy but were asymptomatic and two had
Thirdly, the strain identified is relatively rare in
pulmonary parenchymal changes and were admitted
England as it was found in only two other patients in the
for gastric lavage and sputum collection. One of these
national strain typing database that currently holds
two, an 8 year old boy in year 3, had acid fast bacilli on
17 305 records. Both of these were from the same West
gastric lavage and a negative sputum smear (second
African country as the index case but had no
case).
epidemiological link with the index case or the school.
All adult chest x ray examinations were normal, and Finally, the area of Luton where the school is located
no additional cases of active or latent tuberculosis were has an annual incidence of tuberculosis of 45 per
detected among the household contacts of the children 100 000 people, meeting the definition of a high
diagnosed with active tuberculosis on screening. incidence area.3 Previous screening of school contacts
in this area following cases of sputum smear positive
Epidemiological findings pulmonary tuberculosis in teenage children in the past
Infection rates were highest in the class of the index case two years found rates of positive Mantoux tests of
(79%). The infection rate in this class was significantly pupils between 0% and 2.4%. This indicated that most
higher than that among the other pupils at the school of the high rate of infection in the school (42%) could
(35%) (P<0.01). Infection rates did not correlate with not be explained by the expected rate of infection
ethnicity. associated with the prevalence in the area. The high
rate of infection among school children may have been
Microbiological investigations and DNA fingerprinting facilitated by poor ventilation and all pupils sharing a
DNA fingerprinting—using a 15 loci based typing room for lunch.
scheme known as mycobacterium interspersed repeti- Our report, together with evidence from other
tive units variable number tandem repeats (MIRU- studies,4 6-8 highlights the need for increased awareness
VNTR)—showed that the M tuberculosis strain (42433 of pulmonary tuberculosis as a differential diagnosis in
23315 14321, S lineage) of the index case was children who present to general practitioners with a
genetically identical to a strain isolated six years long history of recurrent or chronic cough.
previously in an adult from the extended family of
the index case who was living in London at the time of Conclusion
his diagnosis but had since returned to Africa. M This report is evidence of extensive transmission of M
tuberculosis grown from the gastric lavage sample of the tuberculosis from a young child with smear negative
second case detected in this outbreak was found to be tuberculosis. National guidelines, which place empha-
genetically indistinguishable from the sample of the sis on sputum smear positivity as main indicator of
index case. infectivity for starting contact tracing outside the
household setting,3 may have to be revised, with
Discussion longstanding exposure to a coughing tuberculous
Smear negative pulmonary tuberculosis is significantly patient with extensive pulmonary lesions included in
less infectious than smear positive disease, and the criteria for contact screening.
although transmission has been documented,5 this is Contributors: All authors were involved in the outbreak investigation. SP
the first report of extensive transmission from a child. was responsible for data analysis and the first draft of this paper. All

574 BMJ | 6 SEPTEMBER 2008 | VOLUME 337

 PRACTICE

authors contributed to interpreting the data, critical revisions, and 4 Guidelines for the Investigation of Contacts of Persons with Infectious
approving the final version to be published. MMcE is the guarantor for this Tuberculosis. Recommendations from the National Tuberculosis
article. Controllers Association and Centers for Disease Control and
Funding: No special funding. Prevention. Morbidity and Mortality Weekly Report
Competing interests: None declared. 2005;54(RR15):1-37.
Provenance and peer review: Not commissioned; externally peer 5 Behr MA, Warren SA, Salamon H, Hopewell PC, Poce de Leon A,
reviewed. Daley CL, et al. Transmission of Mycobacterium tuberculosis from
Patient consent obtained. patients smear-negative for acid-fast bacilli. Lancet 1999;353:444-9.
6 Curtis AB, Ridzon R, Vogel R, Mc Donough S, Hargreaves J, Ferry J, et al.
Extensive transmission of Mycobacterium tuberculosis from a child. N
1 Hertzberg G. The infectiousness of human tuberculosis; an Engl J Med 1999;341:1491-5.
epidemiological investigation. Acta Tuberc Scand Suppl
1957;38:1-147. 7 Loudon RG, Spohn SK. Cough frequency and infectivity in patients
2 Singh M, Mynak M L, Kumar L, Matthew JL, Jindal SK. Prevalence and with pulmonary tuberculosis. Am Rev Resp Dis 1969;99:109-11.
risk factors for transmission of infection among children in household 8 Lienhardt C, Sillah J, Fielding K, Donkor S, Manneh K, Warndorff D,
contact with adults having pulmonary tuberculosis. Arch Dis Child et al. Risk factors for tuberculosis infection in children in contact with
2005;90:624-8. infectious tuberculosis cases in The Gambia, West Africa. Pediatrics
3 National Collaborating Centre for Chronic Conditions. Tuberculosis: 2003;111:608-14.
clinical diagnosis and management of tuberculosis, and measures for
its prevention and control. London: Royal College of Physicians, 2006. Accepted: 9 July 2008

UNCERTAINTIES PAGE
How can we best prevent new foot ulcers in people with
diabetes?
Fay Crawford

University of Edinburgh, Division Screening people with diabetes for peripheral neu- regrettably—almost a decade later—this message
of Community Health Sciences: ropathy and the presence of peripheral pulses every seems to have been overlooked.8
GP Section, Edinburgh EH3 8DE 15 months forms part of the quality and outcomes An RCT conducted after the systematic review
[email protected]
framework of the General Medical Services contract.1 compared the number of recurrent ulcers in patients
Cite this as: BMJ 2008;337: a1234 Our understanding about the best ways to identify receiving hospital based care who were randomised to
doi:10.1136/bmj.a1234 those at risk of foot disease, although not entirely receive routine chiropody free of charge or non-routine
definitive, has been greatly helped by epidemiological chiropody paid for by themselves.10 A statistically
research to evaluate the prognostic value of tests.2 significant effect was seen only when the analysis was
Unfortunately, the same progress has not been made based on the number of feet rather than the number of
with preventive strategies, and uncertainty exists about patients. This suggest that the study was underpowered
the best ways to prevent foot ulcers, especially new for an analysis at the patient level, and although
cases.3 specialist teams may improve patient outcomes, a
thorough evaluation of the cost effectiveness of this
What is the evidence of uncertainty? type of care pathway is needed.
Little evidence from randomised controlled trials
(RCTs) is available to inform clinical guidelines on Pressure deflecting insoles and footwear
preventing diabetic foot ulcers.4-6 Naturally, good High peak plantar pressure is known to be a risk factor
glycaemic control is highly desirable and has been for foot ulceration.3 RCTs provide some evidence that
shown to reduce the incidence of neuropathy.7 How- specially manufactured shoes and insoles reduce the
ever, the effects of interventions that explicitly focus on incidence and frequency of repeated ulcers in high risk
foot health, specialist foot care, pressure deflection, and patients, but the cost effectiveness of these inter-
patient education have not been rigorously evaluated. ventions remains unclear because of the small sample
sizes.8 11
Screening and specialist foot care Guidance issued by the National Institute for Health
A systematic review of interventions to prevent and Clinical Excellence (NICE) in 2004 lists the
diabetic foot ulcers found no evidence that foot evaluation of pressure relieving devices (shoes and
screening and care by multidisciplinary teams of orthoses or insoles) as a research priority in diabetes.6 A
This is a series of occasional
healthcare professionals prevents foot ulcers.8 The trial of callus reduction with the use of pressure
articles that highlights areas of reviewers found one RCT that showed the number of deflecting insoles, which includes an economic evalua-
practice where management lacks foot amputations was reduced when patients were tion, would be particularly useful for those who make
convincing supporting evidence.
The series advisers are David screened for neuropathy and pulses and, if the test decisions about cost effective care.
Tovey, editorial director, BMJ results were poor, were given support hosiery,
Knowledge, and Charles Young, protective shoes, and foot health education.9 Although Patient education
editor of BMJ Clinical Evidence,
and editor in chief, BMJ Point of the review concluded that RCTs urgently need to be Evidence about the value of patient education strate-
Care. conducted in much less select groups of patients, gies to reduce the incidence of foot ulcers is

BMJ | 6 SEPTEMBER 2008 | VOLUME 337 575

PRACTICE

inconsistent. A Cochrane systematic review found that Researchers who have access to large numbers of
an intense educational intervention prevented foot patients with diabetes need to conduct high quality
ulcers in high risk patients compared with a brief RCTs. Evidence is needed from all patients, treated in
educational intervention, but this effect was not seen in all healthcare settings where foot care is delivered.
a second RCT in the same review.12 People with diabetes should be encouraged to take part
The evidence for educational interventions is incon- in such trials.
clusive in the short term and unknown in the longer Competing interests: None declared.
term.8 Large RCTs are needed to find which education Provenance and peer review: Commissioned; externally peer reviewed.
formats are effective and enduring. Such trials of the
general diabetic population should use foot ulceration 1 Revised GMS contract 2007/8. www.dh.gov.uk/en/Healthcare/
Primarycare/Primarycarecontracting/GMS/DH_4125638.
as the primary outcome and have sample sizes of 430- 2 Crawford F, Inkster M, Kleijnen J, Fahey T. Predicting foot ulcers in
870 per treatment arm.12 people with diabetes: a systematic review and meta analysis. QJM
2007;100:65-86.
3 Boyko E, Arhoni JH, Nelson K. Prediction of diabetic foot ulcer
Is ongoing research likely to provide relevant evidence?
occurrence using commonly available clinical information. Diabetes
The update of the Cochrane systematic reviews of Care 2006;29:2562-3.
preventive interventions is ongoing and necessary 4 McIntosh A, Peters J, Young R, Hutchinson A, Chiverton R, Clarkson S,
et al. Prevention and management of foot problems in type 2
(personal communication, Cochrane Wounds Group, diabetes: clinical guidelines and evidence. 2003. National Institute
2007).11 12 However, randomised evaluations of spe- for Health and Clinical Excellence. www.nice.org.uk/guidance/index.
cialist foot care clinics, insoles or orthoses, footwear, jsp?action=byID&o=10934.
5 Scottish Intercollegiate Guideline Network. Management of diabetes.
callus reduction to reduce pressure, and patient SIGN 55. 2001. www.sign.ac.uk/guidelines/fulltext/55/index.html.
education are all urgently needed to inform effective 6 International Working Group on the Diabetic Foot. International
clinical practice. A search of the National Research consensus on the diabetic foot. Amsterdam: IWGDF, 1999.
7 Singh N, Armstrong DG, Lipsky B. Preventing foot ulcers in patients
Register Archive identified only one ongoing RCT, with diabetes. JAMA 2005;293:217-28.
which aims to evaluate the value of padded socks.13 8 Mason J, O’Keeffe C, McIntosh A, Hutchinson A, Booth A, Young RJ.
Although such an evaluation is welcome, it seems to A systematic review of foot ulcer in patients with type 2 diabetes
mellitus. Diabet Med 1999;16:801-12.
focus on people who already have or have had a foot
9 McCabe CJ, Stevenson RC, Dolan AM. Evaluation of a diabetic foot
ulcer and may contribute little to the prevention of this screening and protection programme. Diabet Med 1998;15:80-4.
serious consequence of diabetes. 10 Plank J, Haas W, Rakovac I, Gorzer E, Sommer R, Siebenhoper A, et al.
Evaluation of the impact of chiropodist care in the secondary
prevention of foot ulcerations in diabetic subjects. Diabetes Care
What should we do in the light of the uncertainty? 2003;26:1691-5.
Healthcare professionals should continue to conduct 11 Spencer S. Pressure relieving interventions for preventing and treating
annual foot screening for people with diabetes. Those diabetic foot ulcers. Cochrane Database Syst Rev 2000;(3):
CD002302.
with poor test results should be referred to their 12 Valk GD, Kriegsman DMW, Assendelft WJJ. Patient education for
podiatry department for specialist care, where patient preventing diabetic foot ulceration. Cochrane Database Syst Rev
education, callus reduction, and weight deflecting 2001;(4):CD001488.
13 Boulton AJ. Efficacy of padded hosiery to reduce diabetic foot ulcers
insoles can be provided in accordance with national and amputations. National Research Register Archive. https://portal.
guidelines. nihr.ac.uk/Profiles/NRR.aspx?Publication_ID=N0453145344.

Occluded circulation
Thirty years ago my boss had told me of an aortic valve The engineer we called out disagreed and said the
replacement he had performed. The heart came off bypass problem would lie in the outflow from the header tank.
well with a good output, but later it was apparent that, I was not convinced, but, like a registrar with a consultant, I
although the patient’s legs were warm to the touch, there followed him to the tank. The water was crystal clear. “Feel
was no pulsatile flow. “It was as if,” he said, “there was a the bottom for sludge,” he said. There was none. “Feel the
sponge in the aorta.” And then the penny dropped. During outflow pipe.” I put my arm in again and palpated the
the operation, it was customary to put a sponge in the left orifice of the pipe. It was occluded. A piece of insulating
ventricular cavity to prevent calcium from the valve getting foam had obviously fallen in a day earlier when I had taken
into the heart. This sponge in those days did not appear on the lid off the tank. Being sponge, it was not totally
the swab count, and it had not been removed after the valve occlusive but was allowing only a very slow flow. Sponge
replacement. At laparotomy the embolic sponge was removed, there was a whoosh of water and the problem
removed, and a healthy flow restored to the legs. was solved. The “consultant” was right, the incredulous
I am now a retired cardiothoracic surgeon and have thus trainee wrong.
spent most of my life dealing with human pumps, valves, Perhaps I should have learnt from that original episode,
tubes, leaks, and blockages, so that when we had a new it would have saved me many hours of fruitless activity
central heating boiler installed and there were difficulties trying to bleed radiators without an adequate fluid flow.
with water flow I expected to be able to solve the problem.
The filling pressure was fine in that the header tank was full,
the valves were fully open, and the pump was running, but Jules Dussek retired cardiothoracic surgeon, Sevenoaks
the flow to the radiators was so slow that I was unable to
bleed them. I unkindly thought some sealant had got into [email protected]
the pipes when the plumbers had fitted the new pump. Cite this as: BMJ 2008;337:a992

576 BMJ | 6 SEPTEMBER 2008 | VOLUME 337

View publication stats