The study uses comprehensive data on 1.4 million skilled nursing-home stays between 2000 and 2005 in California, New Jersey, Ohio and Pennsylvania. It takes advantage of two facts. First, Medicaid payment rates to nursing homes tend to be lower, by about 20 percent in the sample used, than private payers'. Second, Medicaid patients generally have minimal or no cost-sharing for nursing-home stays, whereas private payers typically have significant cost-sharing, so the incentives for consumers are much different depending on whether Medicaid is paying or not.

The study reaches three core conclusions. First, nursing homes are more likely to discharge Medicaid patients (for whom they receive lower payments) when they are at or near capacity than when they are not. The financial incentives thus seem to influence provider behavior: Nursing homes discharge Medicaid patients to make room for better-paying private patients, unless they have enough room for both.

Second, earlier discharges of Medicaid patients when nursing homes are full don't seem to harm the patients — in terms of mortality, rehospitalization or other measures. The implication is that many nursing-home stays are unnecessarily long, since earlier discharge doesn't carry negative consequences and since Medicaid patients are held longer when other beds are available at the nursing home. (One could perhaps argue that the evidence just shows that nursing homes are properly discharging patients only when they're ready, but the authors try to adjust for this possibility in various ways. Fundamentally, Medicaid patients are unlikely to become magically healthier just as other available beds in the facility disappear.)

Third, the authors extrapolate from their statistical analysis to assess changes in both provider incentives and consumer cost-sharing. In particular, they examine increasing costs for patients. They then compare the impact to a change in how the nursing homes are paid, moving partially away from the current structure of a payment per day (which encourages longer stays) and instead making part of the payment a lump sum per admission. As they conclude: “providers react more elastically to financial incentives than patients, so moving to episode-based provider reimbursement is more effective in shortening Medicaid stays than increasing resident cost-sharing.”

The numbers are striking: Estimates suggest that taking 11 percent of the Medicaid payment and giving it to nursing homes up front (which then attenuates the incentive to extend stays) would reduce the average Medicaid stay by about a quarter, from 32 weeks to 24 weeks, and save more than $1,000 per stay.

We are in the early days of changing how Americans pay for health care, moving away from fee-for-service payments and toward paying for value. But the results of studies like this strongly suggest we should keep at it: Changing how doctors, hospitals, nursing homes, and other providers are paid is central to saving money and improving quality in health care.

Read more: 

• Consumer research, focus groups part of health systems' improvement plan
• Use of alternative payment models continues to increase
• Changing the basis of payments: Driving sustainable innovation in health care

This column does not necessarily reflect the opinion of the editorial board or Bloomberg LP and its owners.