Charles M. Grossman, MD
To the Editors: Oxman and colleagues (1) make some good arguments about the limitations of subgroup analysis and conclude that clinical practice should be based on "the average estimate of effect from the overall analysis (the more robust finding) or on a subgroup analysis (which more closely reflects the specific situation at hand)."
The latter has too often been ignored, and generalizations have been broader than warranted by the data. The ISIS-1 report (2), for example, showed an impressive and statistically significant difference after the use of intravenous atenolol in acute myocardial infarction. Subgroup analyses showed large differences in mortality
Grossman CM. Subgroup Analyses. Ann Intern Med. ;117:93–94. doi: 10.7326/0003-4819-117-1-93_2
Download citation file:
Published: Ann Intern Med. 1992;117(1):93-94.
Copyright © 2018 American College of Physicians. All Rights Reserved.
Print ISSN: 0003-4819 | Online ISSN: 1539-3704
Conditions of Use