A hierarchical clustering procedure applied to nine million Austrian patients and forty-five million hospital stays identifies 132 recurring combinations of chronic diseases, each defined by which diseases co-occur and which do not.

In this talk, Katharina Ledebur will use these clusters to answer two questions. First, is it biology or is it billing? Running the same procedure on Danish registry data, where the healthcare system is organized along very different lines and the average patient receives half as many diagnoses, returns essentially the same 132 patterns. The clusters track something underneath the health system rather than something produced by it.

Second, what happens if we treat the clusters as compartments and let people flow between them? Calibrating age- and sex-stratified transitions on Austrian data and simulating to 2030 yields projections of multimorbidity growth, a ranking of prevention targets by downstream mortality, and a stress test for the long shadow of SARS-CoV-2.

Please note that this is an internal talk. If you want to participate, please write an email to .