CMStatistics 2021: Start Registration
View Submission - CMStatistics
Title: Individualized empirical null for profiling healthcare providers Authors:  Kevin He - University of Michigan (United States) [presenting]
Abstract: Existing methods for healthcare provider profiling typically assume that the risk adjustment is perfect and the between-provider variation is entirely due to the quality of care. However, in practice, even with very good models for risk adjustment, there will be characteristics of patients and perhaps providers that are not completely accounted for (e.g. unobserved socio-economic factors and comorbidities), and many of these characteristics will be related to the outcome and vary across providers. Thus, some of the between-provider variation in a quality measure will typically be due to this incomplete risk adjustment (or unmeasured confounders), which should be recognized in assessing and monitoring providers. Otherwise, conventional methods disproportionately identify larger providers, although they need not be ``extreme''. To fairly assess providers, we propose an individualized empirical null method that accounts for the unexplained variation between providers. The proposed method robustly models the between-provider variance as a function of effective provider size and avoids bias against large providers.