Selective-referral and unobserved patient heterogeneity - Bias in the volume-outcome relationship

This paper examines the causal effect of volume on outcome on the example of patients with a hip fracture. We use an instrumental variable approach and consider both the practice-makes-perfect and selective-referral hypothesis as well as unobserved patient heterogeneity. Our results indicate that unobserved severity drives the results in the volume-outcome relationship for hip fracture patients and with this the practice-makes-perfect hypothesis has an even stronger effect on hospital quality than expected so far.