Evaluation of the Reliability of Medicare Spending Per Beneficiary for Post-Acute Care.

Background: A measure of episode spending, such as Medicare Spending Per Beneficiary (MSPB) is increasingly used to evaluate provider performance. Yet if the measure is unreliable, as is often true for low-volume providers, it cannot distinguish "good" from "poor" performance.

Objective: The objective of this study was to evaluate the reliability of a uniformly calculated MSPB measure for post-acute care (PAC) and the tradeoffs involved in setting a minimum case count threshold.

Data: Medicare claims for 15 million PAC episodes from April 2013 to March 2015.

Research Design: Given the overlap in patients treated in PAC settings, we developed a uniformly calculated MSPB measure for PAC providers that measures spending during the PAC stay and the following 30 days. We examine variation in the MSPB-PAC measure and characterize the measure's reliability and its relationship to provider case counts.

Results: Applied to our MSPB-PAC measure, a minimum threshold of 20 Medicare episodes as currently used by the Centers for Medicare & Medicaid Services (CMS) would not establish reasonably reliable measures and could result in drawing unduly erroneous conclusions about provider performance. The measures for home health agencies were considerably less stable and reliable than for institutional PAC providers.

Conclusions: CMS should consider adopting a more stringent reliability standard for setting minimum case counts for MSPB-PAC and other measures. Its current threshold (R-statistic=0.4) reflects more random variation than differences in actual provider performance. To include as many providers as possible, CMS should consider pooling data over multiple years to avoid drawing incorrect conclusions about low-volume providers.