Nursing Home Quality Varies between and within Markets
Following a hospitalization, about one-fifth of Medicare beneficiaries are discharged to skilled nursing facilities (SNFs). These facilities provide short-term care for recovery from a procedure — such as a hip replacement — or a medical event — such as a stroke. The average length of stay is 45 days; at a cost to Medicare of about $470 per day, this care accounts for approximately 8 percent of Medicare’s spending.
In Producing Health: Measuring Value Added of Nursing Homes (NBER Working Paper 30228), Liran Einav, Amy Finkelstein, and Neale Mahoney develop a framework to assess the quality of these care providers while adjusting for the mix of patients who are present in each SNF at any given time. The premise of their measure of quality is that patients recover more quickly when treated by higher quality SNFs, leading to an earlier discharge to their homes.
The researchers measure the “value added” of a SNF as the average increase, over the first 30 days of care, in the probability that a typical patient is healthy enough for discharge to the community within the next week. The probability is calculated based on the results of detailed health assessments that all SNFs are required by Medicare to administer at the time of admission and at regular intervals afterwards.
The researchers use 109 components of these assessments, which are recorded in the Long-Term Care Minimum Data Set for Resident Assessment and Care Screening (MDS), to predict each patient’s probability of discharge within a week at the time of each assessment. The components include indicators of physical health, mental well-being, and pain; the patient’s use of treatments such as oxygen and equipment such as walkers; and the patient’s ability to interact with others. Among the most influential measures for predicting discharge are pain intensity, a mood score, and the level of independence with toileting and mobility.
Overall, the quality measure indicates that the average SNF increases this likelihood of community discharge by 4.5 percentage points (relative to a 13 percent probability at admission) over the first 30 days of care. But this value-added measure varies substantially across facilities, with higher-quality SNFs increasing the likelihood of discharge by more. The results suggest that a facility at the 90th percentile of value added can prepare a patient for discharge about a week sooner than a facility at the 10th percentile.
Average value added varies considerably across medical markets. It tends to be lower in markets across eastern Texas, the Deep South, and Appalachia; it tends to be higher in New England and in the Mountain states.
The figure shows that even within nursing home markets, sizeable differences in quality exist. That is, high-quality facilities are available in markets with low average value added, and low-quality facilities are present in markets with high average value added.
This within-market dispersion raises the possibility of improving patient outcomes by encouraging patient use of higher-quality SNFs within their local market. The researchers conclude that “if patients at the 10th-percentile SNF within a market could be moved to the 90th-percentile SNF within the same market, the gains would be on average equivalent to getting a patient home at the same health level 5.6 days sooner.” Such a transition to a higher-quality facility would be financially consequential, reducing Medicare costs by $2,600 per affected patient.
The researchers acknowledge financial support from the National Institute on Aging through grant R01-AG032449 and from the Laura and John Arnold Foundation.