Creating Composite Indices From Continuous Variables for Research: The Geometric Mean

A large body of diabetes and cardiovascular research has reported the effect of various interventions or risk factors on a composite categorical outcome such as the first major adverse cardiovascular event (MACE). Defined as either a nonfatal stroke, nonfatal MI, or cardiovascular death, this outcome better reflects the underlying construct of vascular disease than any of its individual components, and has been a very useful diabetes research tool. In contrast, a simple approach for combining two or more continuous measurements into a composite continuous outcome has not been described. Such an approach could provide a better reflection of constructs such as glucose control than any one measurement alone (e.g., HbA1c, fasting plasma glucose, time-in-target). This paper suggests a simple and convenient approach to creating composite continuous indices that can be used by clinical researchers as either independent variables or outcomes, and provides worked examples illustrating its utility.