Pay-for-performance (P4P) may be better at improving documentation of outcomes rather than actually improving outcomes. Farmer, Black and Bonow give the following example:
Beginning in the fourth quarter of 2008, [CMS] ceased to reimburse for costs due to selected preventable adverse events, including patient safety indicator 5 (PSI-5) (leaving a foreign object in the body during surgery) and PSI-7 (central line–associated bloodstream infection [CLABSI], unless present on admission). Data from the National Inpatient Sample indicate no change in the PSI-5 rate over the years 1998 through most of 2008 with an increase and plateau of the PSI-7 rate over the same period. However, when reimbursement for these preventable adverse effects stopped in 2008, the reported PSI-5 and PSI-7 events decreased by 50% in a single quarter. By contrast, a study based on clinical laboratory data finds no evidence that the nonpayment policy affected the true CLABSI rate. Accordingly, the true rate of these important, preventable adverse events can no longer be reliably monitored using administrative data.
Other Medicare policy decisions may also affect the quality of administrative data. If Medicare institutes more bundled payments, for instance, then hospitals and other providers will likely decrease the amount of detail they include on claims for items that do not affect their reimbursement. Thus, increased bundling will likely omit clinically relevant information.
- Steven A. Farmer, Bernard Black, Robert O. Bonow. Tension Between Quality Measurement, Public Quality Reporting, and Pay for Performance. JAMA, January 23/30, 2013—Vol 309, No. 4