Teaching medical decision modeling: a qualitative description of student errors and curriculum responses

Novice medical decision modelers are prone to errors. The purpose of this article is to describe the common errors committed by decision modeling students and the changes made to a structured decision modeling project course in response.

The decision modeling curriculum includes month-long segments in decision analysis, cost-effectiveness analyses, and a project course. In the project course, students solve a decision problem with prespecified assumptions and input variable values, with the expectation that all reach the same answer. At first, students worked in groups of 2 or 3 but more recently work individually; over time, there have been other changes in the course. Originally, the only assignment was an abstract and 10-minute presentation describing their solution, but now periodic homework monitors progress.

Students' results frequently differed significantly from the expected answers. Errors were common in sensitivity analyses and model construction, among other areas. Defensible differences in structural programming decisions were rare. More recently, result ranges have narrowed as stepwise homework ensures progress and areas prone to error are emphasized. Student communication is timelier, facilitated by homework-related questions. Individual, rather than group, work avoids potential problems with weaker students being carried by the more advanced. Student satisfaction has increased, with more beginning their own projects afterward.

Solving a common decision problem is an efficient teaching tool. Closer supervision leads to increased satisfaction and decreased frustration, facilitated by a more structured approach and the anticipation of common errors.