Drugs are two-sided swords and statins are no exception. Schooling et al. demonstrate that, on average, statins produce small, but statistically significant, decreases in testosterone. They appropriately emphasize that the clinical significance of their observations is unclear but suggest that changes in testosterone might be related to the benefits of therapy as well as the risks, such as the increased chance of diabetes mellitus. Their findings and hypotheses are noteworthy. However, we believe this represents another example of the limitations in the published summaries of drug effects. How do we know all changes induced by drugs are normally distributed? Some may be affected much more than others. Moreover, the confidence intervals of a meta-analysis describe the variance of the mean effect, not the range of effects, and while the mean change characterizes the impact of a drug on a group of patients, the range more fully characterizes its effects on individuals. We treat individuals not groups. Averages do not disclose enough about the risks and benefits of drugs. See related research article here http://www.biomedcentral.com/1741-7015/11/57.