Patient Peer Effects: Evidence from Nursing Home Room Assignments

We provide causal evidence that patient peer effects generate mortality impacts comparable to provider quality differences. Drawing on administrative records covering 2.6 million stays (2000–2010) across 7,200 U.S. nursing homes, we exploit plausibly exogenous roommate assignments identified through unique room identifiers. We estimate that assignment to a roommate diagnosed with Alzheimer’s disease (AD) or Alzheimer’s disease related dementias (ADRD), relative to placement in a private room, increases 90-day mortality by 2.1 percentage points (14% of baseline)—equivalent to receiving care at a nursing home one full standard deviation worse in quality. Effects differ sharply by patient type: patients with AD/ADRD benefit substantially from cognitively healthy roommates but not from private rooms, suggesting important peer monitoring and support roles. In contrast, mortality of patients without AD/ADRD does not depend on roommate cognitive health but is reduced in private rooms. A simple assignment rule exploiting this heterogeneity could reduce overall mortality by 0.8 percentage points without additional resources.