Background: Generalized, biomarker-based metrics of health status have numerous applications in fields ranging from sociology and economics to clinical research. We recently proposed a novel metric of health status based on physiological dysregulation measured as a Mahalanobis distance (DM) among clinical biomarkers. While DM was not particularly sensitive to the choice of biomarkers, it required calibration when used in different populations, making it difficult to compare findings across studies. To facilitate its use, here we aimed to identify and validate a standard version of DM that would be highly stable across populations, while using fewer biomarkers drawn exclusively from common blood panels.
Methods: Using three datasets, we identified nine-biomarker (DM9) and seventeen-biomarker (DM17) versions of DM, choosing biomarkers based on their consistent levels across populations. We validated them in a fourth dataset. We assessed DM stability within and across populations by looking at correlations of DM versions calibrated using different populations or their demographic subsets. We used regression models to compare these standard DM versions to allostatic load and self-assessed health in their association with diverse health outcomes.
Results: DM9 and DM17 were highly stable across population subsets (mean r = 0.96 and 0.95, respectively) and across populations (mean r = 0.94 for both). Performance predicting health outcomes was competitive with allostatic load and self-assessed health, though performance of these markers were somewhat variable for different health outcomes.
Conclusions: Both DM9 and DM17 are highly stable within and across populations, supporting their use as objective metrics of health status. DM17 performs slightly better than DM9 and at least as well as other comparable metrics, but requires more biomarkers. The metrics we propose here are easy to measure with data that are available in a wide array of panel, cohort, and clinical studies.