Combining Composite Endpoints: Counterintuitive or a Mathematical Impossibility?
Kenneth M. Kessler · Circulation · 2003
Endpoints: Counterintuitive or a Mathematical Impossibility?To the Editor:In their excellent article, DeMets and Califf 1 discuss composite endpoints.I would appreciate their further thoughts.When the use of a treatment "X" reduces the composite endpoint of death, non-fatal myocardial infarction, and a softer endpoint such as revascularization or hospitalization rates, we are led to the illusion that all 3 endpoints are reduced.Often only the most frequent, ie, the softest endpoint is reduced.Because an increase in mortality would be detected by safety monitoring, isn't the inclusion of death in composite endpoints when the study is underpowered to show a decrease in mortality simply misleading?Aside from perhaps the calculation of event-free survival that uses time as a metric, is the combination of events of different metric "counterintuitive" as suggested in the article, or mathematically impossible?Weighted schemes may or may not be a viable approach to combining endpoints of different metric because, arguably, endpoints may be simply incommensurable.Can anything realistically be compared with death?Is revascularization that causes transient disability and may lead to improved quality, if not quantity, of life comparable to death or myocardial infarction?Until these issues are satisfactorily addressed, is there any mathematical or statistical methodology that validates the current practice of combining endpoints of different metric?