Lower Expected Post-Acute Length of Stay Predicts Referral to Higher Quality Nursing Homes

While Medicaid remains the primary payer of nursing home (NH) services, NHs have become increasingly dependent on Medicare dollars spent on post-acute care (PAC) for financial solvency. NHs endure financial risk by servicing individuals financed by Medicaid since its reimbursements are a fraction of the private-pay price. Medicaid also may not pay the cost sharing requirements of the Medicare skilled nursing facility (SNF) benefit. Consequently, NHs may be motivated to admit for PAC differentially according to how likely the admission results in either cost sharing with Medicaid, which occurs after 20 days of care, or leads to acquiring a new long-term resident. Assuming that individuals prefer the highest quality NH care, high-quality NHs may more easily select patients and minimize their Medicaid risk. A direct way to do this would be to service patients with conditions requiring shorter SNF stays. In our study we exploit information available to NHs about potential residents requiring SNF care before admission which may then be used to form an expectation of the number needed days of care. Specifically, using a rich set of covariates we estimate length of NH stay among Medicare fee-for-service beneficiaries admitted for SNF care between 2008 and 2014 and then predict the expected length of NH stay (eLOS) of beneficiaries admitted for SNF care in 2015. We then follow McFadden’s random utility maximization model to specify a NH discharge function that describes the relationship between NH choice and NH quality accounting for other NH characteristics among the set of NHs available to each patient. The NH characteristics that were controlled for included occupancy rates, share of Medicaid residents, and the proximity of the NH to the individual’s residence and discharging hospital, among others. We estimated the discharge function using conditional logit and tested whether the coefficient on NH quality varied by eLOS. We found that residents with a lower eLOS in the NH were much more likely to be admitted to a high-quality facility than residents with a high eLOS. Dual-recipients of Medicare and Medicaid were less likely than Medicare-only beneficiaries to be admitted to a high-quality facility, however, the payer-based disparity disappeared among individuals with a low eLOS. We observed beneficiaries with low eLOS who lived in low-income counties traveling greater distances to receive SNF care than their counterparts residing in higher-income counties, further suggesting sorting.