Concerns about using the patient safety indicator-90 composite in pay-for-performance programs.

In 2003, the Agency for Healthcare Research and Quality (AHRQ) released 20 patient safety indicators (PSI) to facilitate measurement of adverse events. Though intended for internal quality measurement and improvement , several PSIs are now being widely publicly reported , including on the Centers for Medicare & Medicaid's (CMS's) Hospital Compare website. Additionally , on October 1, 2014 (fiscal year 2015), CMS began using AHRQ's Patient Safety for Selected Indicators (PSI-90) as a core metric in 2 of its pay-for-performance programs: the Hospital-Acquired Condition (HAC) Reduction program and the Hospital Value-Based Purchasing (VBP) program. PSI-90 is a composite measure consisting of 8 weighted component PSI measures (Table). 1 In the HAC Reduction program, PSI-90 is responsible for 35% of the overall score, and the poorest-performing hospital quartile will have their CMS payments reduced by up to 1% (~$330 million). In the Hospital VBP Program , CMS will reallocate 1.5% (~$1.4 billion) of its diagnosis related group payments to hospitals according to their overall score, 30% of which is composed of PSI-90 and 4 other outcome measures. Thus, the PSI-90 composite measure has been given substantial weight in attempting to align the financial interests of hospitals with the quality of care they provide. As evidenced by a lack of continued maintenance endorsement from the National Quality Forum in 2014, numerous problems exist with the current PSI-90 composite measure: (1) flawed component measures; (2) clinical areas targeted; (3) accuracy of adverse events identified; (4) adequacy of the risk adjustment; and (5) formulation of the composite measure. These flaws may incorrectly identify problem areas for hospitals to address, unfairly penalize hospitals financially, and adversely influence clinician engagement in quality improvement. However, there are opportunities to improve this measure in hopes of more accurately monitoring hospital performance for potentially preventable complications. First, the PSI-90 measure includes components for which the measurement is flawed. For example, post-operative venous thromboembolism (VTE; PSI-12) measurement is vulnerable to surveillance bias (ie, more testing leads to more detection). In a recent study, 2 as hospital-level rates of VTE diagnostic imaging increased (ie, " looking more for VTE "), there was a step-wise increase in PSI-12 event rates (ie, " finding more VTE "). Moreover, hospitals with higher VTE prophy-laxis rates paradoxically had worse risk-adjusted performance on the PSI-12 VTE outcome measure when compared with hospitals with lower rates of VTE prophylaxis. 2 Thus, the inclusion of the VTE outcome …