Shifting Medicaid Nursing Home Payment Toward Upfront, Episode-Based Reimbursement

We study 551,000 nursing home stays in California, New Jersey, Ohio, and Pennsylvania from 2000 to 2005, focusing on residents who began as private payers, many of whom later qualified for Medicaid (Hackmann et al. 2024). Because we follow the same resident as their payment source changes, we can see how patient and provider incentives shift the timing of discharge to the community.

Nursing Home vs. Patient Responsiveness to Payment Incentives

Facilities play a large role in when a resident is discharged, and they respond far more strongly to financial incentives. When a nursing home is nearly full (above 95% of capacity), it has reason to discharge a Medicaid resident and admit a higher-paying private-pay patient. In the states we studied, Medicaid rates were about 15 to 18% lower than private pay rates, and Medicaid discharge rates did rise when beds were scarce. Overall, facilities are about six times as sensitive to payment as residents are to their own cost-sharing. Even large increases in what patients pay do little to move them home sooner.

What an Upfront Payment Would Do

Because facilities are much more price-sensitive, changing how Medicaid pays them is a more powerful lever than changing patients’ out-of-pocket costs. We tested three approaches. One approach—making residents pay the full private rate, offset by an upfront Medicaid voucher—shortened stays the most but raised Medicaid spending, since the voucher cost more than the shorter stays saved. A second approach—paying facilities a bonus for each community discharge—barely changed spending.

The third approach delivered the largest savings while also shortening stays: shifting part of a resident’s expected cost into a lump sum paid at admission and paying a reduced rate for the rest of the stay. Paying 10% of a stay’s expected cost upfront and the remaining 90% as a reduced daily rate cut the average Medicaid stay from about 30 to 25 weeks and lowered Medicaid spending by about $3,829 per stay, or 8.4%.

Would Earlier Discharge Harm Patients?

Our model predicts discharge timing and spending, not health, so we turn to the underlying 2000-2005 data to see whether earlier discharge would harm patients. Only 4% of Medicaid residents discharged to the community were readmitted to a nursing home within 30 days—in line with readmission rates under existing policy. Residents discharged from nearly full facilities were no more likely to be readmitted than others. We also find no evidence that shorter stays increased hospitalizations or deaths.

Together, these findings suggest that encouraging earlier discharge is unlikely to harm patients’ health. We did not, however, measure broader effects of this policy on well-being. Moving care out of a facility could still disrupt some residents’ medical care, daily living support, or housing—questions beyond the scope of our paper.

Medicaid programs should shift 10% of a resident’s expected total cost into a one-time payment at admission. This keeps Medicaid residents attractive to admit even as the daily rate falls. Paying the remaining 90% as a lower daily rate would deter unnecessarily long stays without pushing discharge so early that it harms health. Policymakers would need a standard way to calculate expected costs, or the risk-adjusted daily rate times the expected length of stay—likely drawn from existing case mix and risk adjustment data.

The two parts do different work. The lower daily rate weakens the incentive to extend stays, addressing the first distortion described above—overly long stays when beds are empty. The upfront payment preserves the incentive to admit Medicaid residents in the first place, so that a lower daily rate does not cost them access. Neither part restrains discharge once a resident is admitted: the upfront payment is already made, and a lower daily rate makes a scarce bed more tempting to free for a private-pay admission. The second distortion therefore needs a separate safeguard. To guard against premature discharge, policymakers should consider explicit discharge protocols or making part of the payment contingent on the resident remaining in the community. The San Diego incentive experiment paid facilities in full only if a discharged resident stayed in the lower-level setting for 90 days (Jones and Meiners 1986; Norton 1992). Because such a contingency is not settled at admission, it reaches the discharge margin an upfront payment cannot, and it would also deter discharging and readmitting a resident to collect a second upfront payment. By making Medicaid residents more attractive to admit—and by freeing beds sooner as stays shorten—the reform could also improve Medicaid enrollees’ access to nursing home care where beds are scarce, though our analysis does not measure this directly.

The model should target residents who could reasonably return to the community. Those with very low odds of community discharge—about 10% of stays in our data, typically residents with the most severe cognitive and physical disabilities—should be excluded. In practice, implementation should combine risk adjustment based on administrative assessment data with clinical discharge criteria and safeguards. The payment model should also be limited to areas with adequate home- and  community-based services (HCBS); where HCBS is weak, discharged residents are more likely to return to a nursing home for the daily support they need (Wang et al. 2024).

For eligible residents, shifting 10% of their expected total costs to an upfront amount would shorten the average stay by about 5 weeks and save about $3,829 per stay, or 8.4%, in 2022 dollars. In our study sample, 39% of all nursing home stays ended with community discharge. Conservatively applied to 30% of Medicaid nursing home stays—those of residents who could plausibly move to the community—this could save Medicaid roughly $1.5 billion a year in 2022 dollars (CMS 2024). That amounts to roughly 0.2% of Medicaid spending, 0.35% of national long-term care spending, or 2.5% of Medicaid’s spending on nursing homes (CMS 2023; CMS 2024; KFF 2024).

These changes do not cut the total amount facilities are paid; they change how the payment is structured. A facility that keeps its current discharge practices would see no change in revenue or profit. One that shortens stays earns more by freeing beds for private-pay admissions; in turn, Medicaid saves because it is billed for fewer days. This depends on enough demand to fill the vacated bed. Nursing homes are about 90% full on average in our data, which suggests demand is generally sufficient, but the reform would do less to shorten stays in markets with persistent vacancies.

Medicaid’s Money Follows the Person (MFP) demonstration lets funding follow a resident from the facility to the community, paying for transition assistance and other home-based supports. But it leaves the daily payment to nursing homes unchanged—the incentive at the heart of this proposal.