Association of Power Outage With Mortality and Hospitalizations Among Florida Nursing Home Residents After Hurricane Irma
Julianne Skarha,Lily Gordon,Nazmus Sakib,Joseph June,Dylan J Jester,Lindsay J Peterson,Ross Andel,David M Dosa
DOI: https://doi.org/10.1001/jamahealthforum.2021.3900
2021-11-24
JAMA Health Forum
Abstract:Importance: Exposure to hurricanes is associated with increased mortality and morbidity in nursing home (NH) residents, but the factors contributing to these outcomes are less understood. One hypothesized pathway could be power outages from hurricanes that expose NH residents to excess ambient heat. Objective: To determine the association of power loss from Hurricane Irma with hospitalization and mortality in NH residents in Florida. Design setting and participants: This retrospective cohort study of NH residents residing in Florida when Hurricane Irma landed on September 10, 2017, assessed mortality at 7 and 30 days after the storm and hospitalization at 30 days after the storm. The analysis was conducted from May 2, 2021, to June 28, 2021. All NH residents residing in Florida at landfall were eligible (N = 67 273). We excluded those younger than 65 years, missing power status information, or who were evacuated (13 178 [19.6%]). Exposure: We used state-administered surveys to determine NH power outage status. Exposed residents experienced a power outage poststorm, whereas unexposed residents did not experience a power outage poststorm. Main outcomes and measures: We used Medicare claims to assess mortality and hospitalization after Hurricane Irma landfall using generalized linear models with robust standard errors. Results: In the aftermath of Hurricane Irma, 27 892 residents (18 510 women [66.4%]; 3906 [14.0%] Black, 1651 [5.9%] Hispanic, and 21 756 [78.0%] White individuals) in 299 NHs were exposed to power loss and 26 203 residents (17 620 women [67.2%]; 4175 [15.9%] Black, 1030 [3.9%] Hispanic, and 20 477 [78.1%] White individuals) in 292 NHs were unexposed. Nursing homes that lost power were similar in size, quality star rating, and type of ownership compared with NHs that did not lose power. Power loss was associated with an increased adjusted odds of mortality among all residents within 7 days (odds ratio [OR],1.25; 95% CI,1.05-1.48) and 30 days (OR, 1.12; 95% CI,1.02-1.23) poststorm and hospitalization within 30 days, although only among residents aged 65 to 74 years (OR, 1.16; 95% CI, 1.03-1.33). Conclusions and relevance: In this cohort study, power loss was associated with higher odds of mortality in all affected NH residents and hospitalization in some residents. The benefits and costs of policies that require NHs to have emergency alternate power sources should be assessed.