In Reply: In response to Dr Dören, our study focused on changes in the RSMRs for nonfederal hospitals caring for Medicare patients in the United States. We were specifically addressing whether hospital performance has improved over time. The assessment of patient-level changes in risk and whether specific populations experienced more or less decrease in risk over this period is beyond the scope of our study.
Dr Blumberg raises 2 separate hypotheses for the reduction in between-hospital variation in RSMRs we observed. The first hypothesis is that the reduction is attributable to a decrease in AMI hospital volume because of the approach we took to estimation. The method we used to estimate RSMR is via hierarchical modeling, which will shrink “outlying” observations. While the amount of shrinkage does relate to the AMI hospital volume, it also relates to case-mix, overall mean performance, deviations from mean performance, and the relative AMI volumes of other hospitals. Only under very strict circumstances (eg, the relative reduction of all the hospital AMI volumes would need to be the same for all hospitals, the outliers would need to be the same from year to year, and the performance of each hospital would need to be the same from year to year) could the reduction in between-hospital variation in RSMRs be attributable only to a reduction in hospital volume. Moreover, the national average mortality following AMI decreased.