In an era of cost containment, payers and policy makers have championed the use of pay-for-performance.
Recently, Medicare has initiated a pay-for-performance (P4P) policy of nonpayment for hospital-acquired infections (HAIs). One recent New England Journal of Medicine article examined the impact of this policy on rates of catheter-associated blood stream infections (CA-BSIs), catheter-associated urinary tract infections (CA-UTIs), and ventilator-associated pneumonia (VAP) in the intensive care units and step-down units of 1,166 hospitals.
To adjust for hospital and unit level variation in device usage (central line, urinary catheter, or ventilator), the authors used quarterly infection rates per 1,000 device-days per hospital unit as their primary outcome measure. Since the policy affected all hospitals at the same point in time, the authors appropriately chose an interrupted time-series design. To add robustness, VAP rates served as a comparison series since this measure was not specifically targeted by Medicare. Finally, the statistical analysis accounted for secular trends and clustering at both the hospital and unit levels.
There was no statistically significant decrease in infection rates after the implementation of nonpayment for hospital-acquired infections. The authors hypothesized that hospitals located in states which were newly subject to reporting would be more impacted by the policy since they were not previously under public scrutiny. However, secondary analyses showed that the Medicare policy actually slowed the rate of improvement in these hospitals.
In their discussion, the authors asserted three possible explanations for their findings. First, hospitals may be increasing the rate of “present on admission” coding thus making billing data unreliable. Additionally, the policy may have yielded diminishing returns from prior measures aimed at reducing HAIs over the past decade. Third, given the average hospital was subject to a payment reduction of merely 0.6%, the policy provided an insufficient penalty to change provider behavior.
With the Affordable Care Act surviving its recent Supreme Court challenge, pay-for-performance (P4P) and other variations of value-based purchasing (VBP) are here to stay. This study adds to the growing body of literature evaluating the influence of P4P measures on patient outcomes. Unfortunately, the null finding significantly complicates future policy options.
One possible interpretation is that hospital-acquired infections are poor targets for P4P programs. As articulated by the authors, a second and perhaps more likely explanation is that the amount of financial reimbursement at jeopardy is not large enough for providers to incentivize success. Perhaps the next iteration of Medicare’s P4P program should include a sliding scale of both incentives and disincentives.
Teasing out just what magnitude of incentives and disincentives will change behavior is of paramount importance for physicians, policy makers, and health care payers as P4P and VBP strategies evolve.
Andrew Gonzalez, MD, JD, MPH