[Skip to Content]
[Skip to Content Landing]
June 16, 1989


JAMA. 1989;261(23):3409. doi:10.1001/jama.1989.03420230060017

In Reply. —  The letter by Dr Karlin provides useful insight into the early development of the MedisGroups severity measure. In the small study by physician-consultants at Boston University to which he refers, we asked four internists who were unfamiliar with MedisGroups to list those clinical variables that they thought indicated severity of disease for patients with stroke, lung cancer, chronic obstructive pulmonary disease, acute myocardial infarction, congestive heart failure, gastrointestinal tract hemorrhage, and diabetes mellitus.1 We then compared these lists with the key clinical findings (KCFs) contained in an earlier version of MedisGroups than the version reported in our article (a version containing almost 550 KCFs rather than the current total of approximately 250). More than 80% of the variables identified by our consultants were included in the previous MedisGroups algorithm; many of the remaining 20% of variables either were often measured unreliably (such as pulsus paradoxus) or