Alternatives for using multivariate regression to adjust prospective payment rates.

Multivariate regression analysis has been used in structuring three of the adjustments to Medicare's prospective payment rates. Because the indirect-teaching adjustment, the disproportionate-share adjustment, and the adjustment for large cities are responsible for distributing approximately $3 billion in payments each year, the specification of regression models for these adjustments is of critical importance. In this article, the application of regression for adjusting Medicare's prospective rates is discussed, and the implications that differing specifications could have for these adjustments are demonstrated.