CDM Guidelines
CDM Guidelines
CDM Guidelines
Table of Contents
Table of Contents .............................................................................................................................. 2
PREFACE
I.
INTRODUCTION ....................................................................................................................... 4
A.
II.
III.
IV.
V.
VI.
VII.
Page | 2
PREFACE
Dear reader-user:
As a member of one of the Medical Council of Canada (MCC) Test Committees or as a participant of
a workshop, one of the tasks assigned to you will be the development or review of key features
problems in your general area of expertise for the clinical decision making (CDM) section of the MCC
Qualifying Examination (MCCQE) Part I. Although this may appear as a daunting task, it is a skill for
which the MCC provides general guidelines to create well-constructed, reliable and valid items for
the CDM component of the MCCQE Part I.
This booklet will help guide you towards developing good CDM problems and test cases by
reviewing basic rules pertaining to key features, clinical scenarios, write-in and short-menu
questions, and scoring keys.
I would like to gratefully express my appreciation for the thorough review and editorial comments
provided by Drs. Gordon Page and Georges Bordage.
I trust that you will find this guide useful. If you have any comments or suggestions for improving this
set of guidelines, please forward them to the MCC CDM Test Committee.
Page | 3
I.
INTRODUCTION
The key features (KF) test format, first discussed at the Cambridge Conference in 1984, was
originally included in the MCC's Qualifying Examination (MCCQE) Part I in 1992 to replace Patient
Management Problems (PMPs) and to avoid a sole reliance on multiple-choice questions as the
basis for judging qualification for licensure (Bordage & Page, 1987; Page & Bordage, 1995; Page,
Bordage & Allen, 1995). Key features problems have subsequently been adopted by other groups,
including medical schools for in-house examinations and colleges of physicians for use in
certification and fellowship examinations.
The key features approach is based on the notion of "case specificity", namely that clinical
performance on one problem is not a good predictor of performance on other problems (Elstein,
Shulman & Sprafka, 1978). Typically, less than 10 to 15% of the variance in performance in one
problem is shared with any other problem. Each problem contains unique challenges.
Consequently, assessments of clinical performance need to sample broadly because skills do not
generalize across problems. In order to sample broadly in a fixed amount of time (3-4 hours),
assessment is best served by focusing exclusively on the unique challenges, that is, the key
features, in the resolution of each problem, be they essential issues or specific difficulties; the test
then contains many short, focused clinical cases.
When the key features approach was implemented in 1992, four main benefits were envisioned:
1.
2.
3.
4.
The development of key features cases for the Clinical Decision Making (CDM) part of the MCCQE
Part I has been guided by psychometric considerations of content validity and test score reliability,
and by sound principles of test development. Key features cases provide a great deal of flexibility on
issues of question format, multiple responses to questions, and scoring criteria. Key features
problems have been found to be useful in the assessment of clinical decision-making skills that
require not only medical knowledge but the ability to apply that knowledge within clinical scenarios
requiring critical decisions at specific decision points during the assessment and management of the
problem. These specific critical decision points are what constitute the key features of the problem.
Since November 2000, the MCCQE Part I is a computer-based examination. In the change from a
paper-and-pencil examination, the clinical decision-making component was transformed into a multiform examination whereby examinees are administered one of several parallel forms, each form
containing a certain number of cases. A statistical equating design was chosen to construct the
groupings of clinical cases based on content as well as psychometric criteria. Clinical cases are
grouped in sets of cases called caselets. Caselets are comprised of 6 clinical cases covering the
following major specialty areas of medicine: medicine, obstetrics-gynecology, pediatrics, psychiatry,
surgery, and population health, ethical, legal, and organizational aspects of medicine.
In developing key features problems for the MCCQE Part I, the following considerations must be
kept in mind when preparing test items:
The Medical Council of Canada Qualifying Examination Part I assesses the competency of
candidates after obtaining the MD degree for entry into supervised clinical practice in postgraduate
training programs with respect to their knowledge, clinical decision-making skills and attitudes as
defined by the MCC Objectives.
Page | 4
On an examination composed of key features cases, the following is an example of a typical test
case and questions that a candidate would see.
A.
At 20:00h, the nurse in the emergency department asks you to see a 24-year-old woman, who has 2
children and is now 31 weeks pregnant, reporting of bright red bleeding coming from her vagina that
began about two hours ago. The three sanitary napkins that she used were completely soaked. Her
pregnancy has been normal, as were the two previous ones. She has not had any pain in her belly
and she says that the fetus is moving as usual.
Her BP is 110/70 mm Hg and her pulse is 92/min. Your examination of the abdomen reveals a
uterine height of 31 cm with a soft, non-tender uterus. The fetus is in a breech position and has a
heart rate of 150/min. No bleeding has occurred since she arrived 25 minutes ago.
Question 1. What is your leading diagnosis at this time? List only one or write normal if you judge
this situation to be within normal limits.
1. _________________________________________________________
Question 2. What steps will you take in your immediate assessment and management of this
patient? Select as many as are appropriate.
1.
2.
3.
4.
5.
6.
7.
8.
9.
10.
11.
12.
13.
14.
15.
16.
The template used to prepare the test case and questions for the example above is illustrated below.
Page | 5
Example of a Key Features Template for the Third Trimester Bleeding Problem.
Clinical problem
MCC objective
Age group
Clinical situation
Site/setting
Key features: Given a pregnant woman experiencing third-trimester vaginal bleeding with no
abdominal pain, the candidate will:
1.
2.
3.
4.
Scoring key: Each key feature receives 1 point; key feature points are averaged for each case.
Question 1
KF-1
Score
Criteria
Synonyms
1.0
Placenta previa
0
Question 2
Score
Criteria
1.0
Question 2
Score
Criteria
1.0
Question 2
#6. Discharge home with instructions to come back if bleeding starts again
Selecting more than six items
KF-4
Score
Criteria
1.0
II.
Key features cases are designed to specifically assess decision-making skills rather than simple
recall of factual knowledge. While knowledge is obviously a very important requisite for effective
problem solving, the challenge posed by key features cases is the application of knowledge to the
resolution of a problem that is, the use of knowledge to guide decisions to elicit clinical clues, to
formulate diagnostic impressions, to order investigative or follow-up procedures, to acquire data to
monitor a course of action or evaluate the severity or probability of an outcome, or to select a
management course. For example, key features cases should not assess examinees ability "to
describe features of delirium tremens" (a knowledge issue); rather, they should assess their ability
"to recognize delirium tremens in a specific patient" (a clinical reasoning issue) and "to prescribe
appropriate therapeutic measures (a clinical decision issue). The distinction is not just semantic but
focuses on different skills. While it may be easier to list the cardinal features of delirium tremens, it
may be altogether different and more challenging to recognize delirium tremens within a clinical
scenario and to prescribe an appropriate course of action.
The line between testing knowledge and testing clinical decision making may become thin when a
particular key feature hinges on the possession of a single piece of factual knowledge. For example,
the key feature "Prescribe the appropriate dose of diazepam for a prolonged convulsion (status
epilepticus)", is an example of this situation. If this clinical action is recognized as a critical step in
the resolution of the problem of prolonged convulsion, that is, a key feature for this problem, then it
may be tested using a key features format.
The following example provides an illustration of questions that test only recall of knowledge versus
clinical decisions and actions. Such recall of knowledge questions should not be used in key
features cases.
A 20-year-old nulliparous type I diabetic woman presents to your office to discuss the complications
of diabetes in pregnancy.
1.
What are the maternal complications of type I diabetes in pregnancy? List up to four.
1.
2.
3.
4.
2.
______________________________________________________________
______________________________________________________________
______________________________________________________________
______________________________________________________________
What are the fetal complications of type I diabetes in pregnancy? List up to two.
1.
2.
______________________________________________________________
______________________________________________________________
The concerns with such questions are that they only test the recall of factual information and that
they could in fact be presented independently of the case. That is, the clinical scenario is not
necessary to answer the questions. To improve this situation, the case could be re-written to present
a poorly controlled diabetic pregnancy and the first question could then ask, "Which investigations
would you order at this time?". A general rule to keep in mind when developing key features cases
is that if the question asked can be answered without reference to the attached clinical scenario,
then it is not a good question and is likely not measuring clinical decision making.
Page | 7
III.
The term key features was introduced by Bordage and Page following a review and analysis by
Norman et al (1985) of the research on the nature and assessment of CDM skills. A key feature is
defined as:
1.
2.
3.
4.
The definition of a key feature highlights the fact that not all steps in the resolution of a problem are
equally important and that testing time is better spent by focusing on evaluating the critical or
challenging steps, the key features. The key features represent discriminating features between
successful and unsuccessful candidates. Consequently, the use of key features to develop test
cases will result in more discriminating measures of competence.
The concept of key features represents two important shifts from traditional positions in the
assessment of clinical competence. First, it shifts emphasis from the methods of assessment to the
object of assessment. The first question to be addressed in developing a key features problem is,
What is the problem to be assessed?; selection of the problem must be guided by and directly
linked to the MCCQE objectives. Once the MCCQE objective-linked problem has been selected, the
next question to tackle is, What are the critical, essential, or challenging elements in the resolution
of this problem?, that is, What are this problems key features that should be assessed?. Then,
and only then, the subsequent question is, Which method or methods are best suited to measure
the key features for this particular problem? Thus, the object of assessment is clearly linked to the
MCCQE objectives and the methods are adapted to the object.
Second, key features shift the emphasis from assessing all aspects of solving a problem to
assessing only the essential elements relative to each problem. This shift recognizes that the
essential elements in resolving a problem are unique or specific to each problem (referred to as
case specificity). For some problems, the key features may pertain to data gathering or data
interpretation while for others they may focus on choosing an appropriate therapeutic or follow-up
plan. As a newcomer to preparing key feature cases and questions, the shift from testing
thoroughness to only testing selected aspects of the clinical decision making process, the key
features, is one of the most challenging aspects of the test development task. For example, the key
feature may only focus on one diagnosis within a differential diagnosis rather than the entire set of
diagnoses (e.g., in a case of knee pain, the key feature may focus only on including septic arthritis in
the differential diagnosis because that is the diagnosis most likely to be missed by incoming
unsupervised residents). From a key features perspective, only septic arthritis will be given points;
the other elements of the differential diagnosis will not be credited with any points. If one views
septic arthritis as the signal, that is, the crucial element of the differential diagnosis to assess, it will
stand out better by only rewarding its inclusion in the differential diagnosis, rather than drowning it in
noise (the other elements of the differential diagnosis). By focusing exclusively on critical decisions,
in this case thinking about a septic arthritis, the discriminating power of the question, and the exam,
is greatly increased, that is, the question (exam) better discriminates between strong and weak
candidates. The signal-to-noise ratio is maximized.
Key features are often unique to different cases or presentations of a clinical problem, as they may
vary relative to the clinical presentation of the problem and relative to other issues such as the
patient's age and gender. It is therefore unusual to have a generic set of key features for a specific
clinical problem. For example, the critical elements in the resolution of a diabetic problem as a lifePage | 8
threatening event (e.g., coma) are quite different from those of a diabetic problem presenting as an
undifferentiated complaint (e.g., fatigue in an adolescent). Similarly, the critical elements in resolving
a problem of respiratory failure in an 18-year-old man may differ from those for a similar problem in
a 70-year-old woman.
The key features approach is appealing for assessing CDM skills. The clinical problems are
relatively short and focus only on a few critical elements in the resolution of the problem (i.e., the key
features), permitting more problems to be included on an examination within a fixed period of time
(e.g., 30-40 key feature cases compared to 10-12 PMPs in a half-day exam). This broader sampling
of problems has direct implications for improving the reliability of examination scores. The key
features problem format also permits a flexible approach to question format, the number of options to
choose from, and instructions regarding the number of allowable responses. More specifically, the
short-answer question format is available for situations where cueing from a list of options would
compromise measurement of the decision-making process of the examinee, or where listing options
would provide too great a crutch to weaker examinees, the individuals to whom the entire
examination process is most often directed. Finally, the scoring keys are flexible in terms of the
number and configuration of correct responses to a question and can accommodate the complexity
and configurations of actions often required in the resolution of clinical problems.
A focus on critical steps and broad sampling of problems (36 cases with respect to the MCCQE Part
I) provide a foundation for the content validity of a key features examination format. That is, if such
an examination is constructed from a carefully developed blueprint, it will consist of a representative
and adequate sample of clinical problems from the domain of problems for which examinees are
responsible and the questions within each problem will test only the important steps in its
resolutioneach problem's key features. Flexibility in formats and scoring keys emphasizes the
strengths of this case-based examination format in accommodating the realities and complexities of
clinical medicine.
IV.
As indicated previously, the preparation of key features test material occurs in three steps:
1.
2.
3.
Selecting problems from the domain, in this case, the MCC Objectives,
Defining key features for each problem, and
Writing test cases and questions as well as setting scoring keys for each key feature.
Steps 1 and 2 are done first (often during one test committee meeting), followed later by Step-3
(during a subsequent test committee meeting), thus keeping separate the process of defining key
features and preparing the test material per se.
4.1
Clinical situations and site of care. The clinical problems on a CDM examination are described in
the way in which the problem would be presented by a patient to a physician and can be classified in
one or more of the following clinical situations:
an undifferentiated complaint;
a single typical problem;
a multiple or multi-system problem;
a life-threatening event; and
preventive care and health promotion.
Only one, sometimes two, of these clinical situations are chosen for each problem to be assessed.
The one(s) selected will influence the definition of the key features for that problem. At the level of a
graduating medical student, it is advisable to select primarily typical presentations of clinical
problems. When atypical presentations are used, almost all the candidates will fail to effectively
resolve them and the problem loses its discrimination capability.
Finally, the site of care can also influence clinical actions and decisions because different resources
are available in different venues. For example, in a case of acute chest pain, an immediate
cardiology consultation and specialized investigations may not be available in an office setting
compared to the emergency room.
Hence, the problem selection and context definition process begins with (i) the selection of a problem
from the list of MCCQE objectives (e.g., abdominal pain, acute), followed by (ii) the selection of a
patients age group (e.g., adult), (iii) a clinical situation (e.g., life-threatening event), and (iv) the site
of care (e.g., outpatient department). Only once these elements have been selected and defined
can the development of key features proceed.
4.2
The process of defining key features for a given problem can be approached in several ways. One
straightforward approach is to select a clinical problem (e.g., shortness of breath on exertion), an age
group (e.g., adult), a clinical situation (e.g., an undifferentiated complaint), and a site of care (e.g.,
Page | 10
office) and then address the question, What are the critical steps in the resolution of this problem for
patients of this age group presenting in this way?" Here is an example of a set of key features for a
problem of 3rd trimester bleeding presenting in the emergency room.
________________________________________________________________________________
Each key feature has three basic elements:
1.
2.
3.
While every key feature must have the first two elements, the third element (qualifiers) is not always
present, as illustrated in key features 2 to 4 in the example above. The elements in parentheses in
the set of key features in the example above are not tasks per se but are included as part of the key
feature to remind the test developers of the reason or rationale for the task.
This approach to defining key features can often lead to a narrow statement of a problem's key
features, as it is specific to only one presentation of the problem. A more general approach to
defining key features can be used by addressing the following question for a given problem and age
group, "What are the critical steps in the resolution of this problem for patients in this age
group?" In answering this question, it is extremely useful to think of a variety of instances of
patients in which the clinical problem can present itself (e.g., as an undifferentiated complaint, as a
life threatening event, as a multiple problem, etc). This second approach to defining key features
may be more productive in defining a broader set of key features, some of which are common to
different clinical presentations of the problem. Furthermore, this approach can be further generalized
by not specifying an age group and then including age as part of the "given clinical information" in
the statement of the key features. When key features are more broadly defined in these ways, and a
case is developed to portray this problem on the examination, it may turn out that not all key features
defined are relevant to the case, and thus cannot be tested. Experience has shown that two to three
key features can typically be tested by one case - that is, there are two to three critical steps in its
resolution. Some cases have been found to possess as few as one key feature and others may
have as many as five. From a psychometric perspective, 2 to 3 key features (or questions) per case
optimizes reliability and efficient use of limited examination time (Norman et al, 2006).
The second more general approach to defining key features outlined above is preferable. In
summary, it entails the following steps:
1.
2.
Select a clinical problem from the list of MCCQE objectives and an age group.
Pose the question, "What are the critical steps in the resolution of this problem for patients in
this age group?" Wheres the challenge for a graduating medical student?
Page | 11
3.
4.
5.
Think of many instances or patients who present in different clinical situations (i.e., an
undifferentiated complaint; a single typical problem; a multiple or multi-system problem; a
life-threatening event; or preventive care and health promotion) with this problem.
List the key features for these patients, some that will be unique to specific patients, others
that will be common across patients.
Select the 2 to 3 most critical key features.
The development of good key features can be challenging and the opportunity to present key
features for review and refinement to colleagues or to other test committee members is a necessary
step in the process, especially for authors inexperienced in defining key features. Novice authors
often possess an initial reluctance to give up testing all aspects of the problem, from presenting
complaint to follow-up, in contrast to selectively focusing on only the few critical decisions and
actions and consequently giving credit only for those decisions or actions. Discussions of key
features with colleagues should focus on such questions as, "Is the critical issue in this problem one
of interpreting the data, or is it one of eliciting the data?" This distinction will result in asking very
different questions on the examination. If interpretation is the issue, then clinical data are given and
the examinee is asked questions aimed at interpretation (see Example 5.1 Clinical Problem #1). If
data acquisition is the issue, then a complaint is given and the examinee is asked to identify the
specific data that would be elicited from the patient (see Example 5.2 Clinical Problem #2). The
clarity of a key feature is as important as the clarity of a research question in a research study or an
instructional objective in a teaching session. If the key feature is clear, all other aspects of case
development will readily unfold.
The same issues come up when asking for a diagnosis. Is the key feature focusing on (a) eliciting a
probable diagnosis, that is, one diagnosis in a differential (as in an undifferentiated complaint or
typical case presentation) or (b) eliciting THE working diagnosis or leading diagnosis at this time (as
in a life-threatening situation)? The difference between the two key features is that in the first
instance (a) the test question will ask for a differential diagnosis whereas in (b) the question will ask
for one leading diagnosis at that point in time.
The process of defining key features can be frustrating if problems are defined too broadly. For
example, it would be difficult to define key features for the general problem areas of "cardiovascular
ischemic disorder" and "alcohol abuse," but not for problems of "stable and unstable angina" and
"alcohol withdrawal." Alter-natively, selecting a single patient as the basis for defining key features
provides too narrow a view of the problem and may inappropriately restrict the definition of key
features.
Each key feature statement contains two or three basic elements:
Initial clinical information (conditions) that casts the problem, patients age, and clinical
situation (i.e., Given an adult patient presenting with the candidate will). The initial clinical
information can consist of signs or symptoms, diagnoses, or laboratory test results (e.g., "Given
a patient presenting with a swollen and/or painful leg ") depending on the nature of the key
features tested.
A clinical task (e.g., order a venogram).The clinical task is a critical step, decision, or
action, that is, the key feature, in the diagnosis or management of the problem (e.g., " include
DVT in a differential diagnosis" or " elicit risk factors" or " order a venogram").
if necessary, a qualifier(s) may be added (e.g., order an initial work up, "order an
immediate antibiotic," or "list the most likely diagnosis(es)").
Key feature steps or actions need to be stated as specifically as possible (e.g., prescribe morphine
or order an arterial blood gas). One should avoid non-specific steps or actions, such as manage
appropriately or order appropriate investigations. Similarly replace exclude ongoing hypoxia, a
statement of goal, by order blood gases, that is, a specific clinical action. In addition, avoid
Page | 12
combining different actions or decisions into one key feature statement, such as include DVT in a
differential diagnosis and order a venogram. This statement should be separated into two distinct
key features.
Sometimes a key feature can focus on actions to avoid; for example, Avoid performing a pelvic
examination (because it may cause fatal bleeding) or Avoid discharging home from the emergency
department.
Finally keep in mind the level of experience or training of the candidates being assessed. The very
same problem and conditions can yield different key features depending on whether one is
assessing graduating students versus senior residents, or fellows.
4.3
CASE SCENARIO
After defining the problem's key features, a clinical case scenario must be developed to represent
the problem and its key features on the examination. Case scenarios can vary in length depending
on how much information is needed to get to the question testing the first key feature. For problems
whose key features emphasize the diagnostic work-up of the patient, the scenario is often very brief.
For problems in which the critical steps focus on laboratory or management issues, the scenario will
typically be longer and contain clinical data derived from history taking, physical examination, and/or
initial laboratory data. The usual elements common to any case scenario include the patient's age
and gender, the setting (site) of the encounter, and the patient's presenting condition or reason for
seeking care, followed by whatever clinical details are required prior to the presentation of the first
question. When writing clinical scenarios, think of real patients you have seen in practice, rather
than inventing them de novo. Using real patients or actual real-life situations will make the scenario
more authentic (less contrived).
Use of lay language descriptions is also preferred over use of medical terminology or jargon
(medicalese). In real life the information available to the candidates are lay terms (e.g., coughing
up blood), not medicalese (i.e., hemoptysis). When clinical scenarios are presented using
medicalese (such as Homans sign), candidates, especially the weaker ones, are more likely to
accurately identify the correct responses (Eva et al, 2010). The use of medicalese has the effect of
cueing and decreases the discriminating ability of the clinical decision-making examination format.
The use of lay language descriptions (such as painful calf muscle) is more challenging for weaker
candidates and increases the discrimination factor of the key feature being measured. The following
is an example of a scenario using lay terminology.
Also, in the initial scenario, it is important to put the candidates in a realistic situation and role. For
example, instead of saying, A 29-year-old woman with a red eye that is getting worse for two
days, write, You are asked to see a 29-year-old woman in the walk-in clinic. She tells you that
her right eye has become more and more red over the past two days. When asking questions,
Page | 13
instead of asking, Which diagnosis(es) should you consider?, write What is (are) your
diagnosis(es) at this time?. By going from should to What is(are), you are putting the candidates
in their clinical role with a patient rather than asking them to come up with the correct answer as
implied in the should. Finally, be concise in writing the scenario, while at the same time providing
all the necessary information leading up to the first question (KF); in other words, avoid useless
information in the scenario. Unnecessary information takes time to read and is a poor use of testing
time. However if the problem calls for extraneous yet pertinent information, make sure that that
information is included.
Experience has shown that committee members who are familiar with the level of training of the
candidates being tested can define very accurate and reliable key features. Clerkship directors from
across Canada in a 1995 study corroborated 92% of the key features generated by CDM test
committee members and generated de novo 94% of the committees key features when given the
problem, patient age, and clinical situation (Bordage, Brailovsky, Carretier, Page, 1995).
4.4
Generally, one question tests one key feature. Occasionally a question may be used to test more
than one key feature (see Example 5.1 Clinical Problem #1, Question #2). The question stem
consists of a direct formulation; for example, "What is your leading diagnosis at this point in
time?, compared to incomplete sentences that would be completed by an examinee's responses
(e.g., Your leading diagnosis at this point in time is __________________). The questions derive
directly from the key features and focus exclusively on clinical actions or decisions (as opposed to
the reasoning behind the action or decision); for example, instead of saying Consider contributing
factors write Order thyroid function tests. A second sentence provides instructions regarding the
number of permissible responses, for example, "List up to four. If the question is not the first in a
series of questions pertaining to the clinical scenario, it may also contain additional information, for
example, "The patient has now been in the hospital for two days, and his condition ... What treatment
would you now recommend?".
Two response formats are recommended for clinical decision-making cases: short-answer "write-in"
(WI) responses and "short-menu" (SM) responses. In the WI format, examinees supply their
responses. In the SM format, examinees select their responses from prepared lists of options
accompanying the question. The number of options in these lists varies depending on what the
question is testing. In general there are typically 15-20 options, but the numbers can range from as
few as 5 or 6 to as many as 30. A typical list of options for lab investigations would contain the list of
lab tests usually found on a lab order sheet. To avoid cueing, the lists of options should be
presented in ascending alphabetical order. The lists of options must, of course, contain all keyed
correct responses plus incorrect responses and common misconceptions (i.e., distractors). The
number of options should be sufficient to reduce guessing effects. The lists may also contain correct
options that are not scored as they are not integral to the key feature(s) being assessed. Such
correct non-scored options serve as neutral distractors and must be taken into consideration when
determining the total allowable number of responses. There is no steadfast rule about the number of
options to include. Try to keep the number within the range of what would be plausible in real life,
including plausible but incorrect options (e.g., example, include blood glucose in the list of options
because a blood glucose test is a correct action but is NOT part of the key feature, and
consequently will not be worth any points), especially options that would be attractive to weaker
candidates. An excellent way to generate and refine option lists for SM questions is to pilot test the
question as a WI question and then use the examinees' incorrect responses as a source of
distractor options for the short menu.
When listing each option, include only one action (decision) per option. For example, instead of
Administer one dose of ampicillin immediately in the ER and discharge patient with a prescription for
Page | 14
oral ampicillin, split into two options: Administer one dose of ampicillin immediately and Discharge
patient with a prescription for oral ampicillin.
Almost always, the last option on the list will be None, meaning that no action is needed at that
particular time in the case. This None option is especially useful when testing over-investigation or
over-treatment. By having the None option always present at the end of each list, the candidates
are not overly cued when the option is present. Examples of none statements:
Select (write) none (or Do nothing) if no action [investigation, treatment] is needed at this point.
Select (write) none (or Do nothing) if you choose no action at this point.
Select (write) none if none are warranted.
It is generally recommended to use the WI format for questions that test diagnoses, management,
and prescriptions (see Example 5.3 Clinical Problem #3, Questions #1 and #2) while the SM format
is generally used for history and physical exam findings or investigations (see Example 5.3 Clinical
Problem #3, Questions #3 and #4). Questions testing for diagnoses and treatments are also most
readily marked in the WI format, which is not the case for questions testing issues of history taking
and physical examination because examinees can phrase responses in many equivalent ways.
Questions that test issues of laboratory investigation are naturally presented in a SM "laboratory
sheet" format (see Example 5.3 Clinical Problem #3, Question #4). Studies of the MCCQE Part I
CDM examination data have shown that WI questions are more discriminating than SM questions,
especially around the pass-fail cut-score, and therefore are more effective at identifying weaker
candidates (Page, Boulais, Blackmore, Dauphinee, 2000).
While some problems will only have one key feature, the number of single key feature cases should
not exceed 30% of the cases on an examination. The optimal number of key features per problem is
2 to 3 (Norman, Bordage, Page, Keane, 2006). Single-KF cases tend to be less reliable. On the
other hand, asking 4 or more questions per case tends to be redundant and does not improve test
score reliability.
The sample problems in Section 5 of these guidelines illustrate both WI and SM formats.
Having standard questions on the exam will help reduce biases that might come from different
formulations of the same question. Thus uniform formulation of questions is desirable. See
examples below; statements in brackets represent alternatives.
Candidates are warned about these different types of limits in the general instructions to the exam
(see Section 4.8).
- List (select) only one.
This type of limit is suitable for requesting a single definitive answer, such as a leading
diagnosis or a most important management step.
- List (select) up to x.
This type of limit is suitable for questions in which one or more answers are sought and for
which the number of opportunities for examinees to provide these answers should be
capped (i.e., forcing the issue). The x (number of allowable responses) takes into account
the number of correct keyed responses sought, together with other factors such as
responses that would be reasonable but not in the keyed responses as well as incorrect
responses that might be appealing to weaker candidates.
Page | 16
This type of limit is used in situations where it is useful to determine how many actions an
examinee might take, as in the case of limiting the investigation or treatment (i.e., not over
investigating or treating). The maximum number of responses is specified in the scoring
key, but the examinees are blinded to this limit.
Finally, it is very useful to include a summary table illustrating the relationship between test questions
and key features. At a glance the test developer can see the layout of the case and its questions
and key features. For example in the table below, the case contains 2 questions that test 3 key
features.
KF-1
Q1
KF-2
KF-3
KF-4
Q2
Q3
Q4
It is usually possible to test all key features using one case scenario, but again there are exceptions.
Issues of excessive cueing across questions make it necessary, at times, to select a second case
and develop a second case scenario to test the remaining key features. The second case would be
separated from the first during the delivery of the examination, and presented in such a way that it
would not be readily recognized as related to the first.
4.5
SCORING KEYS
The scoring key consists of the list of the correct responses to a question and a system of assigning
numerical scores to the keyed responses. The correct scored responses should directly reflect the
key feature(s) being assessed. All the correct answers appearing in the scoring key should be part
of the key feature(s), no more no less. Some scoring keys contain only a single response, others
contain several responses and, in addition, the responses may be clustered on the basis of logical
considerations regarding the correct clinical actions to be taken. See examples below.
Examples of Scoring Keys
Key Feature 1. Given an adult patient presenting with a painful, swollen calf, the candidate
will include deep venous thrombosis in the differential diagnosis.
Score
1.0
0
Criteria
Synonyms
Page | 17
Key Feature 3. Given an adult brought to the emergency room with multiple seizures and
without having regained consciousness, the candidate will elicit a history to identify possible
causes of the seizures (including alcohol use, medications, illicit drugs, and diabetes
history).
Score
Criteria
0.25
0.25
0.25
0.25
#33.
or
Scoring keys for each key feature define the responses needed to produce a score of 1
for the key feature. Scoring keys assign a weight to each correct response such as the same
weight of 0.25 to each of four correct responses. Alternately, some responses could be considered
more important than others and assigned larger weights. As a general rule, use of equal weighing is
preferred (see Examples 5.3 Clinical Problem #3, Questions #3 and #4), since research has
shown that differential weighting (see Example 5.3 Clinical Problem #3, Question #2) does not
improve score reliability and is therefore an inefficient use of test development time determining the
weighted scores. Some scoring keys also list responses that result in a score of "0" no matter what
other responses an examinee makes (see Examples 5.1 Clinical Problem #1, Question #2 and
5.4 Clinical Problem #4, Question #2). These so-called killer responses are used for selecting or
listing too many responses or because they represent actions that are dangerous or harmful to the
patient (e.g., unnecessary invasive investigation or harmful treatment). Consequently, not all
dangerous or harmful actions will result in a score of 0. For example, if the candidate (e.g., a
graduating medical student) orders an unnecessary angiography on the examination, in real life, the
radiologist or the cardiologist would review and cancel the order before it would be executed. The
inclusion of too many killer options will have a negative effect on the discrimination of the question.
On the other hand, some harmful actions or decisions can be executed without any safety net
present, like discharging a patient from the emergency room in the middle of the night, or prescribing
a contra-indicated medication, or executing potentially harmful maneuvers (like doing a vaginal
rd
examination for a women with 3 trimester bleeding), and should result in a score of 0 for the key
feature.
Whatever weights are assigned to keyed correct responses, they should sum to "1" so that each
question within a problem is given a maximum score of "1" and a minimum score of 0. When an
examination question tests more than one key feature, each key feature is given a score of 1. Thus
key features, not questions, are assigned points. See example below.
Page | 18
Criteria
1.0
#11. None
Criteria
1.0
Key feature scores within problems are then averaged so that each clinical case on the examination
has a maximum score of "1". Examination scores are then the sum of case scores divided by the
total number of cases on the examination. This summing and averaging procedure ensures that key
features have equal weighting within a case and that cases have equal weighting across the entire
examination.
The passing score for a test is based on a content, criterion-based standard setting procedure, using
a modified Angoff procedure that takes about 30 minutes per problem (see Norcini (2003) for a
review of standard setting procedures).
Page | 19
4.6
DOCUMENTATION
For each problem and corresponding set of key features, include one or more references to support
the key features. For example, for the Seizures problem (Epilepsy - 92) (see Example 5.3 Clinical
Problem #3), the following reference was used, especially regarding the use of thiamine:
Sirven, JI & Waterhouse, E. Management of status epilepticus. Am Fam Phys 2003; 68:469-76.
4.7
On average it takes candidates 3 minutes to respond to each question and 6.5 minutes per case.
For a 3.5-hour exam, 32 cases can be tested (i.e., about 71 key features with 2-3 key features per
case) (Bordage, Carretier, Bertrand & Page, 1995).
On average, the reliability of test scores for a 3.5-hour exam (containing 32 cases) is about 0.70
(Cronbachs alpha). This represents a huge increase in reliability compared to Patient Management
Problems where about 10 cases were tested for the same amount of time, each case testing
everything from chief complaint to history & physical examination, investigation, management, and
follow-up. Using the Spearman-Brown Prophecy Formula, it would take about 45 to 50 cases (a
one-day exam) to reach a 0.80 level of reliability. It is important to note that test score reliability is
calculated using the case as the unit of measurement (e.g., 32 cases) and not the individual key
features (e.g., 71 key features).
4.8
INSTRUCTIONS TO CANDIDATES
Given the peculiarities of the key features approach, especially regarding scoring and the fact that
once a case is completed the candidate cant go back to change responses, candidates are provided
with clear instructions and practice cases on how to take this type of examination. For the complete
set of instructions, see: www.mcc.ca/english/examinations/qualifying_e1.html
On the left-hand side of every examination page the following information and links are available to
the candidates:
1.
2.
3.
4.
5.
link to the French version for English candidates (or the opposite if the candidate is taking
the examination in French),
a clock showing the amount of time remaining for the examination,
link to normal values (candidates are expected to know the normal values for common tests
but all the normal values are available within this link),
link to a calculator, and
a submit button (after which the candidate cannot go back to review or change any
response).
It is essential for candidates taking this type of examination to read the instructions carefully and
practice before showing up for the examination. Overall candidates do well in navigating this type of
examination.
Page | 20
4.9
DISSEMINATION
The Key Features approach to assessment has been widely discussed (Bowles, 1995; Norman,
Swanson, Case, 1996) and provides the following psychometric advantages:
Better reliability because of adequate sampling of the domain, that is, 30-35 short problems
for a 3.5-hour examination;
Better content validity because of more precise assessment of the important clinical
decisions for a representative sample of the domain, especially in assessing bottom-up
reasoning, that is, assessing knowledge application, what clinicians do;
Use of varied response formats to best assess particular decision skills, and focused and
more precise scoring;
Justifiable pass-fail decisions; and
Best predictor of future complaints (Tamblyn et al, 2007).
Beyond all Canadian medical schools, the key features approach is used worldwide to assess
clinical decision making. For example, the following institution and testing agencies use the key
features approach:
Collge des Mdecins du Qubec - Structured Oral Interviews (Jacques et al, 1995; Miller et al,
1997)
College of Physicians & Surgeons of Pakistan (Ali, Bordage, 1995)
American College of Physicians (MKSAP recertification)
American College of Colon & Rectal Surgeons - CARSEP recertification (Trudel, Bordage &
Downing,
-rectal surgeons)
Royal Australian College General Practitioners (Farmer & Page, 2005; Farmer, Hinchy, 2005)
Swiss National Examination Board
German medical schools (Fischer, Kopp, Holzer, Ruderich, Junger, 2005).
The approach has also been used for curriculum and instruction purposes (Doucet, Purdy, Kaufman,
Langille, 1998; Sturmberg, Crowe, Hughes, 2003).
V.
CLINICAL PROBLEM #1
Problem
Author
Dr. X
Clinical Problem
Page | 21
MCC Objective
Clinical Situation
_
undifferentiated complaint
life-threatening event
Location/Setting
Emergency Department
Patient's Age
20-30 years
Patient's Gender
Female
Key Features
Given a pregnant woman experiencing third-trimester vaginal bleeding with no abdominal pain, the
candidate will:
1.
2.
3.
4.
KF-1
Q1
Q2
KF-2
KF-3
KF-4
X
X
Case Scenario
At 20:00h, the nurse in the emergency department asks you to see a 24-year-old woman, who has 2
children and is now 31 weeks pregnant, presents with bright red bleeding coming from her vagina
that began about two hours ago. The three sanitary napkins that she used were completely soaked.
Page | 22
Her pregnancy has been normal, as were the two previous ones. She has not had any pain in her
belly and she says that the fetus is moving as usual.
Her BP is 110/70 mm Hg and her pulse is 92/min. Your examination of the abdomen reveals a
uterine height of 31 cm with a soft, non-tender uterus. The fetus is in a breech position and has a
heart rate of 150/min. No bleeding has occurred since she arrived 25 minutes ago.
Question 1
Format WI
What is your leading diagnosis at this time? List only one or write normal if you judge this
situation to be within normal limits.
1.
____________________________________________________________
Key Feature
1. Given a pregnant woman experiencing third trimester vaginal bleeding with no abdominal
pain, the candidate will consider placenta previa as a leading diagnosis.
Scoring Key
Score
Criteria
Synonyms
1.0
Placenta previa
Question 2
Format SM
What steps will you take in your immediate assessment and management of this patient?
Select as many as are appropriate.
1.
2.
3.
4.
5.
6.
7.
8.
9.
10.
11.
12.
13.
14.
15.
16.
Key Feature 2.
Given a pregnant woman with third trimester vaginal bleeding with no abdominal pain, the
candidate will avoid performing a pelvic examination because it may cause fatal bleeding.
Score
Criteria
1.0
#1.
#2.
#7.
Criteria
1.0
Did NOT select #6. Discharge home with instructions to come back if
bleeding starts again
Key Feature 4
Given a pregnant woman with third trimester vaginal bleeding with no abdominal pain, the
candidate will order pelvic ultrasound (to confirm placenta location).
Score
Criteria
1.0
Page | 24
5.2
CLINICAL PROBLEM #2
Problem
Author
Dr. Y
Clinical Problem
MCC Objective
Outpatient Clinic
Patient's Age
Patient's Gender
N/A
Key Features
Given an adult patient presenting with of a painful, swollen calf, the candidate will:
1.
2.
KF-1
Q1
Q2
KF-2
KF-3
KF-4
X
X
Page | 25
Case Scenario
A 56-year-old man sees you in the outpatient clinic because of pain in his left lower leg that began
two days ago and has been getting worse since. His ankle is also swollen. He has never had a
similar problem. His other leg is fine.
Question 1
Format WI
What diagnosis(es) are you considering at this time? You may list up to two.
1.
___________________________________________________________
2.
___________________________________________________________
Key Feature 1.
Given an adult patient presenting with a painful, swollen calf, the candidate will include deep
venous thrombosis in the differential diagnosis.
Scoring Key
Score
Criteria
Synonyms
1.0
Question 2
Format SM
With respect to your diagnosis(es), what elements of his history would you particularly want
to elicit? You may select up to seven. Select option #27 if no other information is needed at
this time.
1.
2.
3.
4.
5.
6.
7.
8.
9.
10.
11.
12.
13.
14.
15.
16.
17.
18.
19.
20.
Alcohol intake
Allergies
Angina pectoris
Anti-inflammatory medication
Cigarette smoking
Colour of stools
Cough
Headache
Hematemesis
Impotence
Intermittent claudication
Low back pain
Nocturia
Palpitations
Paresthesias
Paroxysmal nocturnal dyspnea
Polydipsia
Previous back problems
Previous knee problems
Previous neoplasia
Page | 26
21.
22.
23.
24.
25.
26.
27.
Key Feature 2.
Given an adult patient presenting with a painful, swollen calf, the candidate will elicit risk factors for
deep venous thrombosis through the patients history
Scoring Key
Score
Criteria
1.0
5.3
CLINICAL PROBLEM #3
Problem
Author
Clinical Problem
Seizures
MCC Objective
Clinical Situation
_
undifferentiated complaint
life-threatening event
Location/Setting
Emergency Department
Patient's Age
N/A
Patient's Gender
N/A
Key Features
Given an adult brought to the emergency room with multiple seizures and without having
regained consciousness, the candidate will:
1.
2.
3.
4.
KF-2
KF-3
KF-4
X
X
X
X
Case Scenario
A 36-year-old man is brought to the emergency room by ambulance because he fell onto a sidewalk
unconscious while waiting for the bus. A witness immediately called an ambulance and reported to
Page | 28
the ambulance crew that before falling to the ground, he seemed confused, agitated, and was
arguing with some invisible person. After falling, he began to twitch for a short while, his face
became blue, and then he began to have jerky movements all over his body for about a minute. He
did not recover consciousness after the episode. During the 10-minute ambulance trip, he presented
two other similar episodes, without recovering consciousness, and a third episode that you
witnessed on arrival in the emergency room.
His temperature is 37.8 degrees C. He looks neglected and is unconscious. No relatives or friends
accompanied the patient.
Question 1
Format WI
What is (are) your leading working diagnosis(es) at this point in time? You may list up to two.
1.
___________________________________________________________
2.
___________________________________________________________
Key Feature 1.
Given an adult brought to the emergency room with multiple seizures and
without having regained consciousness, the candidate will generate a provisional diagnosis of status
epilepticus
Scoring Key
Score
1.0
0
Question 2
Criteria
Status epilepticus (Note: both elements required)
Listing more than two responses
Format WI
What is your immediate management at this point in time (be specific)? List as many as are
appropriate.
1.
___________________________________________________________
2.
___________________________________________________________
3.
___________________________________________________________
4.
___________________________________________________________
5.
___________________________________________________________
6.
___________________________________________________________
7.
___________________________________________________________
8.
___________________________________________________________
9.
___________________________________________________________
10.
___________________________________________________________
Page | 29
Key Features 2.
Given an adult brought to the emergency room with multiple seizures and without having
regained consciousness, the candidate will begin initial therapy including administration of
intravenous vitamin B complex (thiamine), hypertonic glucose bolus (50% dextrose), and
lorazepam or diazepam or clonazepam or phenobarbital AND phenytoin
Scoring Key
Score
Criteria
Synonyms
0.25
IV Thiamine
IV Vitamin B complex
IV Hypertonic glucose
50% dextrose IV
IV Lorazepam
IV Diazepam, IV clonazepam,
IV Ativan, IV Valium, or
IV phenobarbital
AND
IV Phenytoin
IV Dilantin
Ten minutes after arrival, the patient is still unconscious. The nurse found a telephone
number in his wallet. What questions will you ask the person answering the phone
assuming that he/she knows the patient? Select up to six questions or select option 33 if you
think that it is not appropriate to call at this point in time.
1.
2.
3.
4.
5.
6.
7.
8.
9.
10.
11.
12.
13.
14.
15.
Abdominal pain
Alcohol history
Back pain history
Cancer history
Cocaine abuse
Coronary bypass history
Diabetes history
Diarrhea
Dizziness
Drug allergy
Family history
Food allergy
Headache
Hearing disability
Heroin abuse
Page | 30
16.
17.
18.
19.
20.
21.
22.
23.
24.
25.
26.
27.
28.
29.
30.
31.
32.
33.
Joint pain
Lung infection
Medication history
Muscular disease
Nausea
Palpitation history
Pet in household
Previous similar problem
Profession
Sexual history
Smoking history
Social integration difficulties
Surgery
Travel history
Viral infection
Visual impairment
Vomiting
Not appropriate to call at this point in time
Key Feature 3.
Given an adult brought to the emergency room with multiple seizures and without having
regained consciousness, the candidate will elicit a history to identify possible causes of the
seizures (including alcohol use, medications, illicit drugs, and diabetes history)
Scoring Key
Score
Criteria
0.25
#2.
Alcohol history
0.25
#5.
Cocaine abuse
OR
0.25
#7.
0.25
Diabetes history
Question 4
Format SM
It has now been 15 minutes since the patients arrival. What investigations would you order
at this point? Select as many as are appropriate or select option 34 if you think that
investigations are not needed at this point in time.
1.
2.
3.
4.
5.
6.
7.
8.
9.
10.
11.
12.
13.
14.
15.
16.
17.
18.
19.
20.
21.
22.
23.
24.
25.
26.
27.
28.
29.
30.
31.
32.
33.
34.
Amylase, serum
Arterial blood gases (ABG)
Aspartate aminotransferase (AST)
Brain computerized tomographic (CT)-scan
Brain magnetic resonance imaging (MRI)
Calcium, serum
Carotid ultrasound-doppler
Cerebral angiography
Cerebrospinal fluid examination
Complete blood count (CBC)
C-reactive protein
Creatine phosphokinase, serum
Creatinine, serum
Drug screening, serum
Drug screening, urine
Echovirus, serology
Electroencephalographic (EEG) recording
Electrolytes (Na, K, Cl)
gamma-Glutamyl transferase (GGT)
Glucose, serum
Lactate dehydrogenase, serum (LDH)
Lyme disease, serology
Protein electrophoresis, plasma
Syphilis, serology
T4, free
Temporal artery biopsy
Thyroid stimulating hormone (TSH)
Total protein, plasma
Urea, serum
No investigations needed at this point in time
Key Feature 4.
Given an adult brought to the emergency room with multiple seizures and without having
regained consciousness, the candidate will order immediate investigations to identify
potentially treatable causes of the seizures (including electrolytes, glucose, calcium, arterial
blood gases, drug screen, and CT or MRI of brain)
Scoring Key
Score
Criteria
0.17
0.17
#6.
#8.
or
#9.
#10.
0.17
0.17
#18.
or
#19.
#22.
#24.
0.17
0.17
0
5.4
CLINICAL PROBLEM #4
Problem
Author
Dr. A
Clinical Problem
Diarrhea
MCC Objective
Physicians Office
Patient's Age
N/A
Patient's Gender
N/A
Key Features
Given an adult patient with gastroenteritis and dehydration, the candidate will:
1.
2.
3.
KF-1
Q1
KF-2
KF-3
KF-4
Q2
Case Scenario
A 35-year-old mother of three sees you in the office at 1700h for watery diarrhea that started
yesterday morning. She has had about 15 watery bowel movements in the past 24 hours, has felt
sick to her stomach, but no vomiting. She works during the day as a cook in a restaurant but left
work to come to your office. On her chart, your office nurse notes a resting blood pressure of 105/50
mm Hg lying down (a pulse of 110/minute), 90/40 mm Hg standing, and an oral temperature of 36.8
degrees C. On physical examination, you find she has a dry mouth and active bowel sounds. A
urinalysis shows a normal microscopy, with a specific gravity of 1.030.
Question 1
Format SM
How will you manage this patient at this point in time? Select up to three.
1.
2.
3.
4.
5.
6.
7.
8.
9.
10.
11.
12.
Antidiarrheal medication
Antiemetic medication
Intravenous antibiotics
Intravenous fluids
Nasogastric tube and suction
Nothing by mouth
Oral antibiotics
Oral fluids
Rectal tube
Send home with close follow up
Surgical consultation
Transfer to hospital
Key Feature 1.
Given an adult patient with gastroenteritis and dehydration, the candidate will initiate oral
rehydration
Scoring Key
Score
Criteria
1.0
#8.
Oral fluids
Question 2
Format SM
After management of the patient's acute condition, what additional measures, if any, would
you recommend? Select up to four or select None if none is appropriate.
Page | 34
1.
2.
3.
4.
5.
6.
7.
8.
9.
10.
11.
Key Feature 2.
Given an adult patient with gastroenteritis and dehydration, the candidate will collect clinical
specimens
Scoring Key
Score
Criteria
1.0
#8.
Stool cultures
#11. None
Selecting more than four items
Key Feature 3.
Given an adult patient with gastroenteritis and dehydration, the candidate will
recommend exclusion of food handler from the workplace until asymptomatic
Scoring Key
Score
Criteria
1.0
VI.
#5.
Give immune serum globulin to close family contacts
#11. None
Selecting more than four items
Guidelines for Selecting Clinical Problem, Defining Key Features, and Preparing Test
Materials (test scenario, test questions and scoring keys)
Select the clinical problem or presenting complaint from the MCCQE list of objectives.
Page | 35
Select gender and age range of the patient; only specify a precise age (e.g., 27) if the problem
specifically calls for a precise age.
Select a clinical situation(s) best suited to the chosen clinical problem:
1.
2.
3.
4.
5.
The initial clinical information (i.e., Given an adult patient presenting with , the
examination candidate will);
The clinical task (e.g.., Order investigations including creatinine ); and
If necessary, a qualifier(s) may be added (e.g., Order immediate creatinine)
Avoid trying to assess all clinical decision making steps in the resolution of a clinical problem.
Focus only on the more challenging or discriminating essential steps, the difficult steps in
practice where clinicians tend to go wrong.
Discuss and review key features with colleagues and fellow committee members before
developing a clinical scenario(s) for the key feature(s).
Guidelines for Developing Clinical Case Scenarios
Write short clinical scenarios if the key feature(s) is(are) primarily eliciting data; write longer
clinical scenarios if key feature(s) is(are) requiring data interpretation or management.
Use real patient data as much as possible; using factitious data can risk introducing
inadvertent inconsistencies (i.e., incompatible laboratory values).
Avoid the use of medicalese and technical jargon when writing the clinical scenario; use the
words that the patient would actually say or describe the physical findings rather than giving the
medical term (e.g., finger nails and nail beds are normal instead of no clubbing.).
Use of photos (i.e., skin rashes, radiographs, etc) is encouraged in the clinical decision making
examination format.
Avoid the use of humor as this can potentially distract examinees.
Page | 36
are usually homogeneous in content (e.g., all are investigations or all are therapies)
may include correct answers that are not integral to the key feature(s) (i.e., neutral
non-scoring distractors)
would seem plausible and attractive to the minimally competent medical graduate or
one selecting an inappropriate course of action
In the SM format, the list of choices should appear in alphanumeric ascending order, ending
most of the time with a None option.
In the WI format, the total number of allowable answers should be limited to the expectation of
how many reasonable opportunities a medical graduate should have to include the correct
response(s) (for example, if the most likely diagnosis should be included within the top 3
options, then the total number of allowable write-in answers should be no more than 3).
Avoid questions that can be answered without reference to the clinical scenario; such questions
are likely not measuring clinical decision making.
Guidelines for Preparing the Scoring Key
All keyed correct scored responses must directly match the key feature(s).
Use equal weighting if there is more than one correct response. Avoid use of differential
weighting because it does not improve the reliability of the scores.
Each key feature receives a total score of 1; hence, if more than one correct response within a
key feature, weighted scores should add up to 1.
If more than one key feature is being assessed by a question, the scoring key should give a
score of 1 for each key feature that is scored in the answer key.
Use of incorrect dangerous responses (killers) that automatically receive a score of 0
regardless of other responses for that particular question should be used judiciously, as would
be plausible in the given clinical situation (with no safety net) and selected by weaker
candidates.
Include acceptable synonyms (equivalences) for keyed correct responses in WI questions, to
facilitate scoring of such questions.
Page | 37
Problem Number:
2.
Author(s):
3.
4.
5.
6.
7.
8.
Patient's Age (specify only if essential to the problem; otherwise say N/A)
9.
Patient's Gender (specify only if essential to the problem; otherwise say N/A)
10.
Key Features (each key feature should be linked to the MCCQE objective(s) identified in
above):
11.
Dates of Revisions (list all dates, including initial date of problem development and any
subsequent revision dates by month/year):
Page | 38
12.
13.
Case Scenario (clinical case stem required before asking the first question)
14.
15.
Scoring Key following each question with accompanying key feature(s) being
assessed
VIII. REFERENCES
Ali SK, Bordage G. Validity of Key Features for a Family Medicine Pilot Exam at the College of
Physicians and Surgeons Pakistan. J Coll Phys Surg Pakistan 1995, 5(6):256-60 .
Bordage, G., Page, G. An Alternative to PMPs: The Key Features Concept. Further
Developments in Assessing Clinical Competence, 2nd Ottawa Conference, 1987, 59-75.
Bordage G, Brailovsky C, Carretier H, Page G. Content Validation of Key Features on a
National Examination of Clinical Decision-making Skills. Ac Med 1995; 70:276-81.
Bordage G, Carretier H, Bertrand R, Page G. Comparing Times and Performances of French
and Englishspeaking Candidates Taking a National Examination of Clinical Decisionmaking
Skills. Acad Med 1995; 70:359365.
Bowles, T.L. A Worthy Search: The Development of the Key-Features Concept. Acad Med
1995;70:89-90.
Doucet MD, Purdy RA, Kaufman DM, Langille DB. Comparison of problem-based learning and
lecture format in continuing medical education on headache diagnosis and management. Med
Educ 1998; 32:590-96.
Elstein AS, Shulman LS, Sprafka SA. Medical Problem Solving. Cambridge, Mass: Harvard
University Press; 1978.
Eva, KW, Wood, TJ, Riddle, J, Touchie, C, Bordage, G. How clinical features are presented
matters to weaker diagnosticians. Med Educ. 2010; 44: 775-85.
Farmer EA, Hinchy J. Assessing general practice clinical decision-making skills: the key
features approach. Aust Fam Physician 2005; 34:1059-61.
Farmer, E.A., Page, G. A Practical Guide to Assessing Clinical Decision-Making Skills using the
Key Features approach. Medical Education 2005; 39: 1188-1194.
Fischer MR, Kopp V, Holzer M, Ruderich F, Junger J. A modified electronic key feature
examination for undergraduate medical students: validation threats and opportunities. Med
Teach 2005; 27:45055.
Hatala R, Norman GR. Adapting the key feature examination for a clinical clerkship. Med Educ.
2002; 36: 16065.
Jacques A, Sindon A, Bourque A, Bordage G, Ferland JJ. Structured Oral Interview. One Way
to Identify Family Physicians' Educational Needs. Can Fam Phys 1995; 41:1346-1352.
Page | 39
Page | 40