From the angle of a health policy enthusiast, one of the most interesting moments in this year’s inaugural IDWeek was the oral abstract session on the Impact of National Policy on HAI Reporting and Prevention. The topic is highly relevant because the start of the new fiscal year coincides with critical changes in how the Centers for Medicare & Medicaid Services (CMS) pays hospitals: the Hospital Value-Based Purchasing program and the Hospital Readmission Reduction Program now tie Medicare payments to quality of care metrics, incentivizing or penalizing deviations form baseline measures such as re-admission after pneumonia and collection of blood cultures prior to starting antibiotic therapy.
Both schemes fall under the umbrella term Pay for Performance (P4P), a key feature of the Affordable Care Act framed by policymakers as “a strategy to improve health care delivery that relies on the use of market or purchaser power.”
Bits of P4P “carrots and sticks” approach have been in effect for some time now: back in October 2008, CMS stopped reimbursing hospitals for additional complications associated with central-line associate bloodstream infections (CLABSIs), catheter-associated urinary tract infections (CAUTIs), surgical infections, and a list of other hospital-associated “never events” considered preventable. Additional requirements were put in place in January 2011, mandating the reporting of infection rates through the CDC’s National Healthcare Safety Network (NHSN) in order to collect of data on which to base further P4P policies.
Two recent studies by the Harvard School of Pubic Health look back at the effects of non-payment. An article in NEJM by Lee and colleagues use NHSN data to compare infection rates before and after the 2008 policy, reporting that non-payment had no measurable effect on CAUTI and CLABSI rates. In the same week, research published in JAMA by Joint et al. analyzed Medicare administrative data and found public reporting for infections after percutaneous coronary intervention – the most common treatment for ischemic heart disease – had no effect on mortality and led to decrease in the use of the procedure.
Several abstracts presented at IDWeek further these findings with numbers hot off the press. Katherine Ellingson of the CDC’s Division of Healthcare Quality Promotion gave two excellent presentations: one compared CLABSI rates between states that had mandatory vs voluntary reporting of rates in the 2009-2011 period. Both groups experienced a decline in infections, and although he decline was steeper for mandatory reporting group, adjusted regression analysis showed no statistically significant difference; The second study by the same group found a similar non-differentiation of trends when comparing rates for three infections across states that received government funding to set up HAI infection prevention collaboratives. The only outcome for which state funding played a detectable role were CAUTIs, likely because existing federal effort in that area was insufficient.
Another interesting question broached at the session was whether Medicare P4P policies have had unintended consequences on care delivery. Research presented by Dan Morgan and published in this month’s Clinical Infectious Diseases retrospectively compares the pre- and post-October 2008 use of urinary cultures and fluoroquinolone antibiotics for over 2 million admissions over 32 months at 39 hospitals. The authors hypothesized that the P4P scheme would lead to excessive admission screening for urinary infections in order to detect and document the presence of an existing infection and avoid the “punishment” of withheld CMS payments. In turn, the increased use of testing would lead to overuse of fluoroquinolone antibiotics in order to formally “treat” culture-positive patients, even when they show no symptoms of infection and would not benefit from antibiotics.
Interestingly, the analysis revealed a downward trend in the use of both urinary cultures and fluoroquinolones in the post-policy period (see the graphs above). However, although the large number of evaluated data points helped detect a statistically significant effect, the magnitude was very small (a 0.25% and 0.5% decrease per post-intervention month for each outcome, see figures above), leading the authors to conclude that the study “did not identify adverse consequences associated with the CMS policy on hospital-acquired CAUTI.” Note that, in Dr. Morgan’s own words, “this by no means indicates hospitals do not currently overtest and overuse antibiotics for asymptomatic UTIs,” but rather that the reporting rules did not change practices.
There are several potential explanations for the admittedly disappointing lack of effect. One is that these evaluations all relied on retrospective data available through Medicare administrative databases, surveys of convenience samples and NHSN data. A lot of information crucial to the internal and external validity of these comparisons may have been unobservable or miscoded. Second, the evaluation period of 12 – 16 months around the interventions may just be too short, given that healthcare practices can be notoriously sticky and slow to change. Finally, let’s not forget that thus far policies were intended to cut perverse incentives to bill for preventable complications, and to collect more data. P4P schemes that tie bigger payments to outcomes have yet to take effect. Thus, we can only remain hopeful that using bigger carrots and sticks will yield more encouraging results in the coming years.