Clinical Problem Solving and the Clinical Presentation Curriculum

In reply: We greatly appreciate the response of Mandin, Harasym, and Woloschuk to our article,1 because it furthers the discussion of a topic that is (in our view) seriously under-addressed in medical education: how medical students learn to cope with complex clinical cases, and, more particularly, how they construct differential diagnoses for such cases. However, we want to start by clarifying some critical aspects of the method we advocate, which is called clinical problem analysis (CPA), and the article we published. First, we do not refer to the article by Mandin, Harasym, Eagle, and Watanabe2 as a “representative presentation of standard schemas for constructing a differential diagnosis”—at least, we did not intend to. In fact, we referred to this article as one of the few in the medical education literature that actually addresses (albeit tangentially) the construction of a differential diagnosis. We just wanted to emphasize that while schemes can be helpful in constructing differential diagnoses, there is also a risk associated with their use, namely that students learn them by heart and use them rather mindlessly, without checking whether the alternatives the schemes provide are correct and exhaustive. Furthermore, in our view, students should be trained to generate possible causes for diseases based on their knowledge of pathology, pathophysiology, nosology, and other basic sciences, rather than to instantiate a standard scheme. Nevertheless, we definitely do not think that authors who advocate such schemes would endorse students' using them mindlessly. Second, in their letter, Mandin et al. appear to interpret CPA as a completely content-independent method for solving clinical problems. However, when we claim that “CPA … makes optimal use of content-independent (methodical) aspects of problem solving,” we do not mean to imply that CPA is completely content-independent! Therefore, we fully agree with Mandin et al.'s statement that “diagnostic accuracy depends more on mastery of knowledge domain than problem solving strategy,” but, on the other hand, we do not believe that medical problem solving is completely content-dependent. What we do claim is that CPA makes optimal use of that part of problem solving that is not content-dependent. Even if one does not support CPA, one might still agree that a systematic approach to analyzing clinical problems might be advantageous. Moreover, a method can be called systematic only if it is to some extent content-independent (even though it is always applied to specific content). Part of CPA's systematicness is that it explicitly forces the diagnostician to take into consideration every symptom, sign, and patient complaint, and use his or her knowledge to account for them. We are convinced that novice diagnosticians often miss relevant diagnostic hypotheses because they either neglect particular features or misinterpret them. Thus, though CPA may be problem-independent, it is not, as Mandin et al. appear to think, knowledge-independent. In addition, we also emphasize that CPA aims not only at developing students' diagnostic competency, also at identifying treatment. Although we think that the clinical presentation curriculum, as presented in the literature by Mandin et al.,2,3 is an interesting innovation in medical education, we do have some questions concerning the approach. In their letter to Academic Medicine, the authors state that “In the clinical presentation model, all possible patient presentations are identified and listed beforehand.” We doubt whether this is possible. For example, in practice, one should always be aware of comorbidity (which theoretically increases the number of clinical presentations from N to N2, even if the comorbidity of only two diseases is taken into account). Even more importantly, the medication a patient currently takes (e.g., for a chronic disease) often interacts with a new, acute illness, not to mention all kinds of drug—drug interactions. Are they all covered in a clinical presentation curriculum? In addition, it is also not quite clear what Mandin et al.2,3 actually mean by a scheme. On the one hand, the term apparently refers to a procedure to be followed for solving a clinical problem (e.g., “students are given opportunity to practice utilizing schemes in problem solving issues…,”2,p. 189), whereas on the other hand, it appears to refer to a compiled knowledge structure in the mind of the student or physician (“a mental categorization of knowledge”3,p. 173; it is also described this way in the letter to Academic Medicine). Apart from lack of theoretical clarity, a problematic consequence of this view is that it is all too easy to assume that students learn to solve clinical problems by emulating experts' behavior: apparently, the scheme an expert uses reflects the knowledge structure in his or her mind. In most areas of expertise development, however, watching and mimicking experts is not the optimal way to acquire expertise, for much of what experts actually do has become implicit. Therefore, in our view, students should initially learn clinical problem solving by following CPA's guidelines, that is, by explicitly working through and elaborating upon a complex case. By repeated application of the method, as well as by accumulation of clinical experience, students gradually develop finely tuned knowledge structures (e.g., illness scripts) that they learn to apply when appropriate—that is, when a clinical case is relatively simple, with one disease or condition responsible for all signs, symptoms, and complaints. Only in such cases is it appropriate to take shortcuts; in more complex cases, an explicit approach will be necessary to fully resolve a clinical problem, including deciding on treatment. The fact that experts are able to use illness scripts (or compiled knowledge structures, or prototypes, or “patterns”) in routine cases is, in our view, a consequence—may be even an epiphenomenon—of their acquired expertise; thus, encouraging inexperienced students to use such knowledge structures, which they possess only in a very rudimentary form, is not a fruitful way to develop clinical expertise. In sum, we think that CPA and the clinical presentation curriculum basically are not incompatible, for the clinical presentations Mandin et al. (1995, 1997) describe can be used by students to practice applying CPA. As the clinical presentations are probably representative of the full domain of clinical problems, repeated application of CPA to these presentations may foster students' development of illness scripts for a broad and representative sample of patients. Thus, CPA and the use of clinical presentations do not by necessity lead to completely different curricular structures, as Mandin et al. state in their letter. In our view, CPA definitely has some advantages: it explicitly addresses comorbidity and drug—drug interactions, it is geared to treatment, it takes into account not only the causes but also the consequences of the current condition of the patient, and it can be used in all cases, including those yet unsolved. Presently, it is not clear to us whether the clinical presentation curriculum has the same potential. We look forward to future contributions by Mandin and colleagues in which these issues are addressed.