Clinical reasoning skills not only help physicians in reaching an appropriate diagnosis, but are also the key to preventing diagnostic errors. Diagnostic errors in medicine may occur in 5-15% of cases [8]; two-thirds of these are cognitive errors which include flaws in diagnostic reasoning, and decision making. Deeper understanding of the processes involved in acquisition of clinical reasoning skills will thus help in reducing cognitive errors.

Educational psychologists have explored the process of medical decision making extensively, and these perspectives have changed over time [6, 9]. It is not within the scope of this article to discuss these theories, their evolution or evidence. We shall restrict ourselves to merely mentioning one of the models of clinical reasoning- namely Croskerry’s dual processing theory [10, 11].

Croskerry’s model proposes that clinicians use two types of cognitive processes to arrive at a diagnosis: System 1 approaches and System 2 approaches. System 1 approaches are intuitive, draw on past experience, and may be based entirely on pattern recognition or heuristics [10, 11]. On the other hand, System 2 approaches are slower, analytical, more deliberate and involve deeper thinking (Table I). The two processes are not mutually exclusive.

TABLE I Dual Processing Theory: System 1 and System 2 Approaches

Clinicians tend to switch between both approaches depending on the complexity of the case [12]. In the learning phase, one tends to use more of System 2 or analytical approaches, and as expertise sets in there is a tendency to use more of System 1 or pattern recognition approaches. However even when experts encounter challenging cases, they tend to revert to System 2 approaches. During teaching, more emphasis is laid on System 2 processes, but in practice, physicians use System 1 processes more often. Hence there is need to train learners in System 1 approaches right from the beginning.

Clinical reasoning is perhaps best taught during the course of a clinical encounter either conducted by the physician-teacher (for demonstration), or preferably during observation of a clinical encounter being carried out by the student. Clinical case presentations, case based discussions/ chart stimulated recall, clinical problem solving exercises and structured case presentation models like SNAPPS (refer Box 1) and One Minute Preceptor (all discussed in detail later) are good settings for teaching clinical reasoning skills.

We enumerate some strategies which when used consciously by educators will encourage students to learn clinical reasoning approaches (Table II) [13,14].

TABLE II Educational Strategies to Encourage Clinical Reasoning

Exposure to a wide variety of clinical conditions: Learning is contextual. Success in solving one kind of clinical problem is a poor predictor of success when faced with another clinical situation. A student thus needs exposure to a wide variety and large numbers of clinical cases during training [1, 6]. Students can learn prototypes of different diseases by this approach [15].

Activation of prior knowledge: Asking students for their reasoning based on pathophysiologic knowledge also enables them to recall and contextualize relevant basic sciences concepts [16, 17].

Emphasize forceful features: Many clinical conditions may be recognized by certain ‘key features’ or ‘forceful features’ or ‘anchor points’. Students must be asked to observe these forceful features that serve as trigger in the memory for recognition of the condition when encountered again [13]. Pattern recognition based on above helps the physicians not only in identification of clinical condition but also in discriminatory thinking processes [9].

Categorization and illness scripts: It is impossible to learn the frequency of every sign and symptom of each disease. One of the basic differences between the approaches of experts and novices is that experts are able to mentally categorize diseases in a logical manner [18]. The expert’s mind stores the information pertaining to clinical conditions or diseases in the form of ‘illness scripts’– or the predictable details of the condition such as predisposing factors, clinical presentation, complications, etc. [1, 19]. They learn to retrieve and apply this information reflexively.

Students need to be taught or exposed to clinical cases in a manner that they gradually develop these mental prototypes of disease or ‘illness scripts’. This can be done by guiding learners’ thought processes by way of asking relevant questions. These questions should encourage them to (a) propose differential diagnoses based on minimal clinical details; (b) modify diagnostic hypothesis as more information is available; and (c) justify or refute hypotheses based on their background knowledge or by asking them to compare and contrast most likely differential diagnoses [13,19]. Another way of teaching students to build on their illness scripts is to ask them to reflect on a previous patient with similar findings and compare presentations.

Two models of structured case presentations which encourage building of illness scripts or mental schemata are discussed here –the SNAPPS model, and the One Minute Preceptor (OMP) model. These can be utilized for teaching as well for formative assessment and are designed for use by the physician-teacher in a busy office or out-patient setting [20].

Use of checklists has also been mooted to help in avoiding errors of omission [24].

Formative feedback: The role of providing effective formative feedback to the learner is the single most important feature which affects learning. Both SNAPPS and OMP models have an inherent component of providing feedback to the students.

Encourage learners to use both System 1 and System 2 approaches: There is often a mismatch between what we know about diagnostic reasoning and the way we teach our students. We have already discussed that expert clinicians first make use of intuitive processes, and go to analytical processes only later when the case does not fit our illness scripts. We should make a deliberate effort to promote intuitive thinking by asking the students at periodic intervals during the presentation (i.e. when he still does not have the complete clinical history or physical findings), the various possibilities that can be considered. As each new piece of information is presented as the history or physical finding, this list should change. The knowledge of epidemiology further contributes to speed and accuracy of diagnosis. Students should be trained to utilize the epidemiological data such as seasonal, geographical or demographic variation of disease conditions for narrowing down the diagnosis [25, 26]. Students must learn to prioritize in a list of differential diagnoses in a given context, enumerating points in favor and against each diagnosis. They should be encouraged to explain the reason why any further particular information is being sought, and how they arrived at a particular conclusion [27].

Another method, the Clinical Problem Solving (CPS) exercise serves as a good setting to demonstrate clinical reasoning. An expert physician is presented an unknown case in a stepwise fashion. During the course of an interactive discussion between the expert and the audience, the diagnostic process and its nuances are demonstrated [28].

Reflection and metacognition: Students must be encouraged and provided an opportunity to reflect on their diagnostic approach, and think about what they could be missing. Morbidity and mortality conferences are a good place to do this. Residents can be asked to reflect on their delivery of patient care based on actual case records (case based discussions) or on the discharge papers of patients (chart stimulated recall) [29]. Portfolios may also be useful in achieving this purpose. Such a metacognitive approach may help learners to recognize the need to slow down and avoid errors that occur due to premature closure of reasoning [30].

Deliberate practice: Just like a musician needs to practise again and again to play well, clinicians too need to hone their skills through training activities which are designed to maximize improvement. Ericsson called these activities ‘deliberate practice’ [31]. Deliberate practice includes finding opportunities for repeated practice, requesting honest feedback on performance at frequent intervals, maximizing learning from each case, reflecting on feedback and errors to improve performance and using mental practice to support clinical experiences. This can be done during regular clinical activities such as, asking students to report back during the morning rounds or after an emergency floor/ call duty.

The assessment of clinical reasoning is challenging as these skills are not measurable and must be inferred from behavior. As clinical reasoning is context specific, it has to be assessed across multiple domains, on multiple clinical scenarios, using multiple assessment methods to draw meaningful and valid interpretations. Moreover, these skills should be assessed throughout the course. Learning strategies for developing clinical reasoning rely heavily on feedback and reflection, and this is possible only when continuous ongoing formative assessment is in place.

The assessment of clinical reasoning can be carried out either in an authentic workplace-based clinical setting or outside it. The standardization of assessment, reliability, feasibility and resource efficacy may appear higher for assessments carried out in formal examination settings. However, assessment carried out in authentic clinical settings inherently scores higher in terms of validity and educational impact. Reliability of these assessments can be improved by increasing the number of clinical encounters. The issue of feasibility can also be addressed by structured assessment tools such as the mini clinical evaluation exercise (mini-CEX). Usefulness of some methods in assessment of clinical reasoning ability is discussed below:

1.　Multiple Choice Questions (MCQ)

A well blueprinted MCQ-based examination has the potential of assessing wide content areas across different contexts in a short time. Simple recall type MCQs contribute little to assessment of medical decision making. However they can be improved to explore clinical problem solving ability by making them contextual. This is done by inserting clinical scenarios (Box 2).

Contributors: TS: conceptualized the draft; JNM, Anshu, TS: wrote the manuscript; PG: provided critical comments and edited the manuscript.

Funding: None; Competing interests: None stated.