Clinical Decision Making in Health-Illness Management
In the realm of biomedical knowledge, diagnosis and management of health-illness places greater emphasis on probabilistic thinking and inferential or inductive reasoning, with greater attention to the specificity of the data and the precision of decisions. Rational justification, confirmation and elimination strategies, and judgment of value are critical reasoning skills within this domain. As outlined by Kassirer and Kopel-man (1991), the first step in the diagnostic process is hypothesis activation or the identification of diagnostic possibilities. Hypothesis activation is based on preliminary information such as the patient's age, medical history, clinical appearance, and presenting concerns. The next step is information gathering and interpretation. This step is strongly influenced by probabilistic thinking and inductive reasoning. The likelihood of various diagnostic hypotheses is carefully considered, with new data used to assist with confirming, eliminating, or discriminating between diagnoses. The working diagnosis is then selected based on causal attribution (i.e., whether all physiologic features are consistent with the favored diagnosis and underlying cause).
This hypotheses generation and revision occur in both novices and experienced clinicians, although the experienced clinicians' hypotheses are of higher quality. It is also noted that some of diagnostic reasoning variation between novices and experts does not appear to be problem-solving variations but instead dependent on the experts' increasing use of pattern recognition based on their knowledge organization and experiences. Retrieval of those patterns can be based on previous experienced exemplars or more abstract prototypes. Thus, expert-novice differences can be somewhat explained in terms of the volume of experts' experienced exemplars available for pattern recognition (Schwartz & Elstein, 2008).
The working diagnosis becomes the basis for therapeutic action, prognostic assessment, or further diagnostic testing. Final verification of the diagnostic hypothesis is determined through tests of adequacy and coherence. Adequacy ascertains whether the suspected disease process encompasses all of the patient's findings. Coherence determines whether all the patient's illness manifestations are appropriate for the suspected health concern. The final diagnostic hypothesis then becomes the basis for treatment decisions, in combination with evidence-based analysis of treatment options and patient-specific cost-benefit analyses for each of the treatment options.
Probability decision making is used to narrow the hypotheses. As new data are obtained, each diagnostic probability is recalculated. The posttest probability is then used to guide additional data collection and to generate the pretest probability of the usefulness of that data. The formal mathematical rule for this process is Bayes' theorem. Inaccurate application of Bayes' theorem explains some of the errors that occur in the diagnostic reasoning process, such as overestimation of pretest probability. The clinician tends to overemphasize rare conditions with inflation of pretest probabilities because those are the cases most memorable. In general, small probabilities are overestimated and large probabilities are underestimated by clinicians. Experience and mentoring are clearly necessary to learn biomedical decision making. Kassirer and Kopelman (1991) also advise parsimony in medical reasoning (i.e., seeking a simple, direct, and clear explanation for the patient's health-illness findings whenever possible). As commonly stated, if you hear hoof beats, look for horses, not zebras.
