Association for Public Policy Analysis & Management

Due to the complexities of medical care and information asymmetries between health care providers and patients, ensuring adequate quality is a persistent challenge for regulators and policymakers. A potential solution (growing in popularity) is public reporting of provider performance. Such information campaigns aim to encourage greater consumer search and discernment when selecting a health care provider and to simultaneously incentivize quality-promoting provider competition. Leveraging market forces in this manner has intuitive appeal, but it is a priori unclear if and how it would affect providers within the public sector.

In 2008, the Department of Veterans Affairs (VA) was required by Congress to produce and publicly release a hospital quality report, as part of new efforts to improve transparency and demonstrate patient safety—a recurring concern for this specific agency. This quality information initiative creates a rare opportunity to study how publicly owned and operated hospitals respond to learning about their relative performance. On the one hand, hospitals and their local managers may anticipate greater scrutiny from veterans’ advocacy groups as well as the executive and legislative branches of government. On the other hand, the salience and ease of interpretation for the information may be low, and importantly, VA facilities lack direct competitors.

We leverage the near-universe of VA health care provider employment data from 2004-2012 and match it to the corresponding hospital-level quality and performance metrics (139 VA medical centers in total). We then aim to demonstrate the heterogeneity of measures across the VA system, to investigate pre-2008 provider staffing and skill mix as predictors of low performance, and finally, to explore the staffing patterns and compensation levels of the lowest performers relative to the highest performers before and after the reports are released. The latter analyses are accomplished via a difference-in-differences research design adapted to an event study empirical approach. We analyze nurses and physician separately, but examine key measures common to both types of health care professionals (e.g., recruitment and retention rates and part-time employment status).

Early evidence shows meaningful heterogeneity in VA hospital health outcomes (e.g., 30-day mortality rates) as well as performance measures (e.g., inpatient length of stay and surgical productivity). We also observe increased clinician employment across the VA starting in 2008, especially among nurses (which we benchmark against private sector). There are only limited indications that VA hospitals respond to broad, composite scores; however, some specific metrics appear to drive a labor demand response. Hospitals with disproportionately high heart failure mortality rates reduce their reliance on part-time nurses, and likewise, those with unusually long lengths of stay for mentally ill veterans decrease their use of part-time psychiatrists. Such changes can increase provider availability without requiring the hiring of new staff. Interestingly, rather than invest in surgical output enhancements, the data suggest that low productivity facilities may cutback on related staffing—consistent with anticipating worse outcomes due to lower procedure output. Ongoing analyses are expanding and refining these approaches as well as exploring the effects of alternative (more aggregate) quality reporting introduced in later years.