Clinical Reasoning Skills
Problems in the Clinical Reasoning Process
Mistakes made by learners lacking expertise and those made by experts differ. The three most common errors involved in developing a set of working hypotheses upon presentation of a clinical case for the first time include the following errors, each present in about half of clinical reasoning cases:
Identification: failure to identify relevant clinical information in the case.
Interpretation: failure to properly interpret the clinical information obtained.
Hypotheses: failure to generate an appropriate hypothesis or hypotheses from the information in the case.
Less experienced learners tend to have problems identifying and interpreting information, but this improves with experience. Although students at more advanced levels may have acquired more biomedical knowledge to correctly identify and interpret the relevant clinical data, they have not yet developed memory schemes for synthesis of the information into a diagnostic hypothesis.
Experienced physicians make fewer errors in hypothesis generation, but they tend to use less data to arrive at hypotheses, and may fail to identify all relevant clinical information, which may cause problems when their initial hypotheses require revision. They use pattern recognition and do not consciously incorporate biomedical knowledge into hypothesis generation.
Failure of identification and interpretation of findings increases with the level of difficulty of the clinical case. Failure of hypothesis generation is proportional to case difficulty as well.
Availability bias affects learners at all levels. The availability heuristic is based upon the tendency to weigh likelihood of a diagnosis by how easily it can be recalled. This bias helps to make common diagnoses, but obscures less frequent diagnoses. The classic medical student bias is giving the next case seen the same diagnosis as the last case seen. Experts are subject to this bias when they fail to incorporate analytical reasoning into the diagnostic process.
Bias in the form of conjunction fallacy occurs when one incorrectly assumes that the probability of multiple events that occur together is greater than the probability of any one of the events alone. Since multiple findings can occur in conjunction with a clinical presentation, the findings together may be judged to support a diagnosis more strongly than they should separately.
Detail bias occurs when a finding or a diagnosis is accompanied by a more detailed description, or 'he who talks loudest and longest' is given more credence.
Conservatism, or anchoring, describes the bias associated with the order of presentation of data. The initial data (the anchor point) tends to get the most attention, and subsequent data is less likely to lead to revision in hierarchy of diagnostic probability.
Learners misjudge likelihoods of events based on exposure to prior examples of cases:
they note the resemblance of the present case to prior cases they have seen, but they cannot move from one case context to another (generalize)
they have a limited exposure to prior cases
they fail to recall prior cases
they recall some cases more vividly than others
Learners misjudge the likelihood of diseases, suspecting rare diseases more often than is appropriate.
Learners overemphasize the significance of a single finding, such as a positive test result, fail to generate an appropriate representation of the problem, and produce random hypotheses.
Learners are distracted. Multitasking is actually multisequencing, and the sequences can become so short that nothing is accomplished or remembered.
Learners are too hurried. 'There's never time to do it right, but there's always time to do it over.' Suboptimal outcomes preclude doing it over.
Learners jump to conclusions with little information.
Learners presume prematurely that they have a working (or final) diagnosis.
Role of Biomedical Knowledge
Generating hypotheses through a chain of explicit, causal reasoning requires an elaborate, time consuming process that is prone to generate errors. It is more efficient to use known associations between clinical features and illnesses (called scripts, or pattern recognition). Each further encounter with a patient with a specific disease will add more information to the related illness script. However, it takes time and practice to build up long-term memories with scripts.
Biomedical factual knowledge can be utilized in forming the framework for an illness script. It places constraints on the acceptable values for the different attributes of scripts and on their relationships. It also alerts clinicians when they encounter abnormal findings or events that violate physiological expectations that are normally found in a specific type of disease, serving as a coherence criterion for hypotheses about the patient.
Biomedical knowledge can also be used in situations where no available scripts are adequate. In such cases, clinicians use their biomedical knowledge to understand the situation and to find pertinent hypotheses through a chain of causal reasoning. However, as above, the application of causal reasoning is a difficult and time-consuming process.
Bowen JL. Educational strategies to promote clinical diagnostic reasoning. N Engl J Med 2006;355:2217-2225.
Elstein AS, Schwartz A. Clinical problem solving and diagnostic decision making: selective review of the cognitive literature. BMJ. 2002;324(7339):729-32.
Kassirer JP. Teaching clinical reasoning: case-based and coached. Acad Med. 2010; 85:1118-1124.
Groves M, O'Rourke P, Alexander H. Clinical reasoning: the relative contribution of identification, interpretation and hypothesis errors to misdiagnosis. Med Teach. 2003;25(6):621-5.
Mamede S, van Gog T, van den Berge K, Rikers RM, van Saase JL, van Guldener C, Schmidt HG. Effect of availability bias and reflective reasoning on diagnostic accuracy among internal medicine residents. JAMA. 2010;304(11):1198-203.