Using a longitudinal k-means clustering method to explore nursing home factors associated with SARS-CoV-2 infection peak and resilience to a COVID-19 surge

OBJECTIVE: Nursing home residents have been disproportionately impacted by respiratory virus-related morbidity and mortality due to inherent vulnerability and communal living environments. This study aims to identify nursing homes with higher infection rates during a period of intense SARS-CoV-2 transmission and explore facility-level characteristics potentially associated with infection surges. DESIGN: A longitudinal k-means clustering approach followed by exploratory regression analyses. SETTING: U.S. Nursing homes reporting to the Centers for Disease Control and Prevention's National Healthcare Safety Network (NHSN). METHODS: A longitudinal k-means method (kmlShape) classified the facilities based on their weekly SARS-CoV-2 incidence rate epidemic curve, identifying two categories (low vs high infection peak) based on the magnitude of infection peaks. A logistic regression model with bootstrapping was developed to assess facility characteristics associated with higher SARS-CoV-2 infection surges. RESULTS: Among 11,990 nursing homes analyzed, 9,058 were classified as having a low infection peak, while 2,932 had a high infection peak. Nursing homes that are for-profit (OR = 1.570, 95% bootstrap confidence interval [BCI] 1.441-1.807), with high staff turnover (OR = 1.292, 95% BCI 1.154-1.451), or located in areas with higher social vulnerability (OR = 1.457, 95% BCI 1.239-1.880) were more likely to experience high infection peaks. Nursing homes with higher residents' vaccination coverage (OR = .321, 95% BCI .248-.380) and located in urban areas were less likely to experience high infection peaks. CONCLUSIONS: The facility-level characteristics associated with lower SARS-CoV-2 infection peaks may indicate resiliency and help evaluate the capacity of nursing homes to endure stressors such as respiratory viruses and other communicable illnesses.