The odds ratio is an often used outcome measure in randomised trials with binary endpoints. The odds ratio overestimates, however, the relative risk. The degree of the bias depends on the baseline risk or incidence of the outcome. In some cases, the overestimation can be substantial.
Knol et al. (1) reviewed 288 randomised trials published in 2008 in Annals of Internal Medicine, British Medical Journal, Journal of the American Medical Association, Lancet, and New England Journal of Medicine.
Of 193 trials with a binary primary endpoint, a majority presented the outcomes as hazard ratios as the focus was on time-to-event, and 24 trials presented the outcomes in terms of odds ratios. In these, 5 of the trials had odds ratios that differed more than 100% from their corresponding relative risk. Of the 41 trials presenting binary secondary endpoints, 19 presented at least one odds ratio that differed more than 100% from its corresponding relative risks. None of the trials that presented outcomes in terms of odds ratios warned the reader about the risk of misinterpreting the results.
The authors conclude that the misinterpretation of odds ratios can seriously affect treatment decisions and policy making.
- Knol MJ, Duijnhoven RG, Grobbee DE, Moons KGM, Groenwold RHH. Potential Misinterpretation of Treatment Effects Due to Use of Odds Ratios and Logistic Regression in Randomized Controlled Trials. PLoS One. 2011; 6(6): e21248.