A study in this month’s New England Journal of Medicine will not end debate on the topic of performance incentives for healthcare. The Centers for Medicare and Medicaid Services conducted a pilot program (administered by nonprofit hospital consortium Premier, Inc.), and will shortly issue $8.7 million in incentive payments.
When correcting for all the relevant variables, hospitals enrolled in the incentive program only performed 2.6% to 4.1% better than nonparticipants, depending on the condition being evaluated.
(“Better” is defined, in this case, as closer adherence to a predefined set of treatment guidelines. This study did not address medical outcomes. In the old medical data analysis lingo, it addressed “process” rather than “structure” or “outcome.”)
The Wall Street Journal quotes one professor of medicine as saying the impact of the program was “very modest.” That may be generous. The New England Journal’s editor, Arnold Epstein M.D., suggests that the difference may be entirely attributable to the “selection effect” – that is, that the hospitals that enrolled did so because they believed they were already achieving the program’s goals. That’s a plausible explanation, and the topic deserves further study.
The HMO/managed care industry is heavily invested in the pay-for-performance concept, and they won’t find much to cheer about here. The Federal government, via CMMS, has a lot more clout than any private payer – and its results are certainly less than conclusive.
Pay-for-performance has also been discussed in the workers’ compensation managed care arena. The ratio of comp to other health spending is 1:10, however, which creates yet another hurdle for adapting this model to workers’ comp medical reimbursements.
The study didn’t document the cost of the pay-for-performance program. At a minimum, that cost was ($8.7 million in incentive payments + Premier’s fee + cost of analysis + CMMS internal cost of administering program). Did it save money and/or create better health outcomes? We don’t know.
Another potential problem with this incentive plan is that institutions are being incentivized, but individuals (i.e., doctors) are still making the treatment decisions. That leaves extra layers of human interaction to go through before behavior is actually changed, which diffuses the impact considerably.
Still, it’s possible that the effect was more than simply selection. What’s more, once hospitals develop institutional procedures for changing care (in order to serve their largest purchaser, the Feds) those procedures can then be used for private payers. It still may be possible to build a working incentive model for private pay health and workers’ comp care.
So the pay-for-performance discussion isn’t over. It just didn’t get the boost from this study that managed care executives might have been hoping for.