Hierarchical causal variance decomposition for institution and provider comparisons in healthcare

Disease-specific quality indicators are used to compare institutions and health care providers in terms of processes or outcomes relevant treatment a particular condition. In the context surgical cancer treatments, performance variations can be due hospital and/or surgeon level differences, creating hierarchical clustering. We consider how observed variation received at patient decomposed into that causally explained by performance, within hospital, case-mix, unexplained (residual) variation. For this purpose, we derive four-way variance decomposition, with attention causal interpretation components. estimation, use inputs from mixed-effect model nested random hospital/surgeon-specific effects, multinomial logistic for populations. investigate our methods simulation study demonstrate them through analysis administrative data on kidney Ontario.