Study: Roommate assignment can play important role in improving patient outcomes

Nearly 1.2 million Americans live in nursing homes across the country, and legislators often seek new ways to improve their care, from hiring more staff to increasing the number of private rooms. Most of these solutions would significantly add to the cost of nursing home care, which is already supported by $91 billion in annual Medicaid and Medicare expenditures. But a new study by Alden Cheng — a postdoctoral research associate in finance at Gies College of Business and the National Bureau of Economic Research — and Martin Hackmann at UCLA, suggests that one simple change could substantially reduce mortality rates in nursing homes. And the best news is, it’s absolutely free.

Cheng has long been fascinated with the study of peer effects, which explores how a person’s behavior, choices, or outcomes are influenced by the people around them. These effects play a significant role in every aspect of our lives, from the grades we achieve to the products we consume. Cheng suspected they might also play a role in the care patients receive in nursing homes. But even he was surprised by the results of the study, which is the first to show that who a person rooms with could matter just as much as where they receive care.

That’s the key finding of the large-scale study, which analyzed 2.6 million nursing home visits over the course of a decade. The paper, “Patient Peer Effects: Evidence from Nursing Home Room Assignments,” confirmed some of the authors’ prior intuitions — that room assignment matters for patient outcomes — but also produced some unexpected results: that these effects are large enough to influence mortality. For example, placing patients with roommates who have Alzheimer’s or Alzheimer’s-related diseases (AD/ADRD) leads to a 14% higher 90-day mortality rate overall compared with placing them in private rooms.

Interestingly, these effects vary substantially across patients. While it’s true that patients with normal cognitive functioning do better in private rooms, the type of roommate they’re assigned to in a multi-occupancy situation doesn’t seem to matter. For patients with AD/ADRD, however, it’s a much different story.

“What’s best for these patients tends to be being assigned to a roommate who doesn’t have AD/ADRD,” said Cheng, who found that this kind of arrangement can reduce the 90-day mortality by about 5%.

“We think this is because cognitively healthy roommates can potentially play a support role for these patients,” said Cheng. “Many nursing homes are understaffed, so perhaps a roommate with AD/ADRD gets into trouble, and their roommate can help them or alert staff to the problem.”

This interpretation is also backed by the study, which showed that peer effects were twice as large in facilities where staff levels were below the median.

“Some nursing homes have dedicated Alzheimer’s care units, and we find smaller effects there,” says Cheng. But in general nursing homes, where care is less specialized and patients are often grouped by their cognitive status, adopting a different roommate assignment policy could have a big impact.

“Policy makers have long been concerned with nursing home quality and have passed many regulations over the decades with mixed results,” says Cheng. “The fact that we can achieve improvements without additional costs is quite important.”