Healthy Skepticism Library item: 10145

Warning: This library includes all items relevant to health product marketing that we are aware of regardless of quality. Often we do not agree with all or part of the contents.

Publication type: Journal Article

Cates CJ.
Misuse of confidence intervals threatens conclusions
BMJ 2007 May 19; 334:(7602):1020
http://www.bmj.com/cgi/content/short/334/7602/1020-a?etoc

Abstract:

The data presented in table 1 of Ong’s review leave considerable uncertainty about whether atenolol is better or worse than other blockers.1 The confidence intervals for the results on other blockers are wide (as fewer patients have been studied), and the test for interaction shows that the relative risk for atenolol, compared with other blockers, for stroke is 1.05 (95% confidence interval 0.26 to 4.17), for myocardial infarction 1.22 (0.91 to 1.63), and for total mortality 1.21 (0.95 to 1.14). All of these confidence intervals include the possibility of no difference, and for stroke the results are compatible with atenolol being four times better or four times worse than other blockers. It is very misleading to draw conclusions based on whether significance is achieved with either treatment alone.2