|
|
| |||||||||||||||||||||||||||||||||||||||||
In general, such an approach can be an important aid for clinical practice, so that for any specific patient, one can judge the absolute benefit of a new treatment (set against the absolute risk of side effects, as well as costs if relevant) and decide whether the treatment is warranted for that patient. This approach is also applicable when, with regard to relative risk, there is no evidence of subgroup interactions with treatment. The question of whether prasugrel is preferable to clopidogrel in acute coronary syndromes4 is an interesting example; it may be that the absolute benefit of the reduced risk of ischemic events outweighs the increased risk of bleeding only for high-risk patients.
Similarly, if the absolute risk of myocardial infarction due to rofecoxib (Vioxx) had been properly quantified in appropriate risk-stratified subgroup analyses, the benefits of pain relief and reduced gastrointestinal bleeding might have been more clearly seen to outweigh the low absolute risk of myocardial infarction in a substantial proportion of patients, thus permitting the continued prescription of rofecoxib with appropriately restrictive labeling.
Stuart J. Pocock, Ph.D.
London School of Hygiene and Tropical Medicine
London WC1E 7HT, United Kingdom
stuart.pocock{at}lshtm.ac.uk
Jacobus Lubsen, Ph.D.
SOCAR Research SA
CH-1260 Nyon, Switzerland
Dr. Pocock reports serving on paid advisory boards for Eli Lilly and Merck. No other potential conflict of interest relevant to this letter was reported.
References
It is also important to consider the robustness of the conclusions of such analyses with regard to different choices of risk categories. For example, patients in the RITA-3 trial were initially classified into four categories (1, 2, 3, and 4) and then further classified into five categories by splitting category 4 into two groups (4a and 4b).2 The five-category classification suggests relatively small absolute treatment differences in all groups except 4b (19%), whereas the four-level categorization suggests a large effect only in category 4 (13%), which might lead to a different treatment strategy for patients in category 4a. Furthermore, the uncertainty in the estimated risk reduction within each risk group needs to be considered. For example, in the RITA-3 trial, the estimated risk reduction for category 4a is 4.1%, but the corresponding 95% confidence interval ranges from a 16.7% reduction to an 8.5% increase. Finally, within-level comparisons should be preceded by evidence of the heterogeneity of treatment effect based on an interaction test oriented to differences in absolute risk.
Rui Wang, M.S.
Stephen W. Lagakos, Ph.D.
Harvard University
Boston, MA 02115
References
| |||||||||||||||||||||||||||||||||||||||||
HOME | SUBSCRIBE | SEARCH | CURRENT ISSUE | PAST ISSUES | COLLECTIONS | PRIVACY | HELP | beta.nejm.org Comments and questions? Please contact us. The New England Journal of Medicine is owned, published, and copyrighted © 2008 Massachusetts Medical Society. All rights reserved. |
Previous
