We thank Dr Messerli and colleagues for emphasizing the take-home message of our study, that is, the type of evidence trials provide is limited with regard to generalizability (ie, guidelines) to the general population. As illustrated in their Table 1, randomized controlled trials, even when targeted to the elderly, have tended to deal with highly selected populations.
These limits are not newly recognized. Some 38 years ago, questions arose regarding the direct application of results coming from trials as a successful strategy to advance the overall health of the individual or the population.1 In fact, while it has been highlighted that the relative benefit of interventions is constant across defined subgroups within trial populations,2 the heterogeneity of effect within the general population may well be substantial. The problem is that randomized trials are by design not able to address this heterogeneity. Trials examine primarily medical interventions; randomization (if successful) permits comparisons of interventions not confounded by the "individuality" of patients.3 Here is the paradox of the trial: it is the best way to assess whether an intervention works but is arguably the worst way to evaluate who will benefit from it. This contradiction leaves a gap in the evidence available for clinicians.