Purpose of review: Mortality/morbidity-based end points have been useful in evaluating treatments that modulate 'mediator variables' with a large effect size. Ventilation is usually a supportive measure, and hence is best seen as a 'moderator variable'. It can, therefore, have only a modest impact on disease-specific mortality. In this context, over reliance on final outcome-based end points (mortality, length of stay, etc.) risks the abandonment of several potentially useful developments. These concepts are important in considering how future developments should be evaluated.
Main findings: A modest effect size implies that large sample sizes will be necessary to demonstrate mortality/morbidity benefits. Recruiting large numbers over geographically/culturally/economically heterogeneous areas over long periods (during which clinical practice is unlikely to remain constant) has several limitations. Furthermore, manifestations of critical illness are based on nonlinear interactions between insult, host responses and other moderator variables. In such nonlinear systems the final outcome is unpredictable and does not follow simple linear assumptions. Such 'unexpected' events occurring in clinical trials involving moderator variables, may potentially lead to erroneous conclusions.
Summary: It is crucial that a more dynamic approach, not based on final outcome alone, is considered in designing new clinical trials involving new ventilation strategies.