James Wright1 asks why we do not do more large simple randomized controlled trials (RCTs) in Canada. To support his point, Wright alludes to the differing results in observational studies on hormone replacement therapy and the results obtained in the Women's Health Initiative (WHI) clinical trial.2 However, as pointed out in a recent article by Garbe and Suissa,3 there were some serious methodological concerns with the WHI trial. In particular, the high rate of unblinding of gynecologists in the study introduced the potential for detection bias.
Clinical trials are important and have their place. However, we should not neglect the power of observational studies in determining drug outcomes. There is longstanding evidence that the results of careful observational research are very close to those obtained in clinical trials.4 The power of a clinical trial is its ability to control for unknown confounders through randomization. But randomization is not a guarantee — it merely means that on average the unknown confounders will be balanced.
In an era of limited resources for health research, we must realize that not every study can be a clinical trial and that observational studies can provide accurate answers to questions much faster than RCTs. This can be important for conditions that require lengthy periods of follow-up. The key is to ask the right question and then use the appropriate type of study to answer it.
J.A. Chris Delaney Statistician Division of Clinical Epidemiology Royal Victoria Hospital Montréal, Que.
Footnotes
-
Competing interests: None declared.