Clinicians Who Practice Primarily in Nursing Homes and the Quality of Care for Residents With Alzheimer Disease and Related Dementias

Abstract

Importance: The number of physicians and advanced practitioners (APs) whose care is concentrated in nursing homes (often referred to as nursing home or skilled nursing facility specialists [SNFists]) has increased rapidly. Therefore, whether these clinicians provide better care is important.

Objective: To examine the association between SNFist care and outcomes of long-stay nursing home (NH) residents with Alzheimer disease and related dementias (ADRD).

Design, setting, and participants: In this retrospective cohort study of 417 378 residents with ADRD in US NHs, claims for a 20% national sample of Medicare fee-for-service beneficiaries between 2013 and 2019 were analyzed. Adjusted estimates were based on a machine learning approach that incorporated a doubly robust procedure using a generalized estimating equation with inverse probability treatment weighting. Three secondary analyses were conducted: (1) stratified analyses for physicians and APs, (2) inclusion of physicians of any specialty and APs, and (3) use of proxy outcomes for in-place deaths. Data were analyzed from June 1, 2024, to May 3, 2025.

Intervention: Receipt of care from a SNFist; SNFists included generalist physicians and APs.

Main outcomes and measures: Hospitalizations and emergency department (ED) visits for ambulatory care-sensitive (ACS) conditions. Death without an ACS hospitalization and death without any hospitalization were used in secondary analyses.

Results: Of the total 417 378 residents, 242 540 received care from SNFists (mean [SD] age, 83.5 [8.7] years), and 174 838 never received care from SNFists (mean [SD] age, 84.8 [8.5] years). Compared with the residents who never received care from SNFists, the residents who received care from SNFists were more likely to be Black (12.6% vs 9.4%; P < .001), dually eligible (77.5% vs 73.1%; P < .001), and have more chronic conditions (eg, anemia, 60.9% vs 57.6%). Compared with non-SNFist clinicians, the SNFist clinicians were more likely to be female (physicians, 37.1% vs 23.3%; APs, 88.1% vs 85.1%), practice at more facilities (mean [SD] number of facilities, 9.4 [8.7] for SNFist physicians vs 6.4 [6.1] for non-SNFist physicians; 8.6 [8.1] for SNFist APs vs 7.1 [6.8] for non-SNFist APs), and less likely to practice in rural areas (physicians, 9.3% vs 25.4%; APs, 8.1% vs 20.2%). In adjusted analyses, receiving care from a SNFist vs non-SNFist was associated with 7% lower odds of an ACS hospitalization (odds ratio [OR], 0.93; 95% CI, 0.90-0.96) and 7% lower odds of an ACS ED visit (OR, 0.93; 95% CI, 0.90-0.96). In stratified analyses, receiving care from a SNFist physician vs a non-SNFist physician was associated with 13% lower odds (OR, 0.87; 95% CI, 0.83-0.90) of an ACS hospitalization and 7% lowers odds of an ACS ED visit (OR, 0.93, 95% CI, 0.88-0.97); comparisons of SNFist APs vs non-SNFist APs were not statistically significant. Estimates from the analysis including physicians of any specialty and APs were consistent with the primary results. SNFist care was associated with increased odds of in-place death.

Conclusions and relevance: Findings of this cohort study suggest that the use of SNFists by NHs may enhance the quality of care for residents with ADRD.

Figure.. Adjusted Odds of Experiencing an Ambulatory Care–Sensitive (ACS) Hospitalization or an ACS Emergency Department (ED) Visit Among Nursing Home Residents With Alzheimer Disease and Related Dementias Who Received Care From Nursing Home or Skilled Nursing Facility Specialists (SNFists) vs Non-SNFists

The estimates were obtained using a generalized estimating equation incorporating inverse probability treatment weights. The propensity scores were computed using machine learning for variable selection that accounts for nonlinearities and interactions, using the Super Learning algorithm, which builds predictions as a weighted average of predictions provided by a user given list of candidate algorithms. A detailed description of the propensity score calculation and the machine learning algorithm is provided in the eMethods in Supplement 1. AP indicates advanced practitioners; OR, odds ratio.