Considerations on the strengths and limitations of using disease-related mortality as an outcome in clinical research.

Abstract

Disease-related mortality (eg, cardiovascular mortality or breast-cancer mortality) is often used as an outcome in randomised clinical trials and systematic reviews. The rationale why disease-related mortality might be used in addition to, or instead of, all-cause mortality seems to be that disease-related mortality may more readily detect the experimental intervention effects. Disease-related mortality is theoretically what most interventions aim at influencing; disease-related intervention effects are not 'diluted' by events unrelated to the disease that may be occurring in both the experimental group and the control group (eg, traffic accidents). Intervention-effect estimates are indeed theoretically diluted and affected if events unrelated to the disease or the trial interventions are occurring. Although sounding attractive, we will in the present paper consider the several methodological limitations of using disease-related mortality instead of all-cause mortality as an outcome. When mortality is a relevant outcome, we recommend using all-cause mortality as a primary outcome and disease-specific mortality as a secondary or exploratory outcome depending on power.