Pooled analysis of trials may, in the presence of heterogeneity inadvertently lead to fragile conclusions due to the importance of clinically relevant variables being either hidden or lost when the findings are pooled.