This cohort study evaluates the association between evacuation and health care outcomes among Florida assisted living residents exposed to Hurricane Irma.
Key Points
Question
What is the association between evacuation and health care outcomes among assisted living (AL) residents exposed to Hurricane Irma?
Findings
This cohort study of 25 130 Florida AL residents found that evacuation was associated with increased odds of emergency department use and nursing home placement within 30 days of the storm.
Meaning
These findings suggest evacuation may be associated with adverse outcomes after a hurricane among AL residents, which should be taken into consideration during emergency preparedness planning.
Abstract
Importance
Evacuation has been found to be associated with adverse outcomes among nursing home residents during hurricanes, but the outcomes for assisted living (AL) residents remain unknown.
Objective
To examine the association between evacuation and health care outcomes (ie, emergency department visits, hospitalizations, mortality, and nursing home visits) among Florida AL residents exposed to Hurricane Irma.
Design, Setting, and Participants
Retrospective cohort study using 2017 Medicare claims data. Participants were a cohort of Florida AL residents who were aged 65 years or older, enrolled in Medicare fee-for-service, and resided in 9-digit zip codes corresponding to US assisted living communities with 25 or more beds on September 10, 2017, the day of Hurricane Irma’s landfall. Propensity score matching was used to match evacuated residents to those that sheltered-in-place based on resident and AL characteristics. Data were analyzed from September 2022 to February 2024.
Exposure
Whether the AL community evacuated or sheltered-in-place before Hurricane Irma made landfall.
Main Outcomes and Measures
Thirty- and 90-day emergency department visits, hospitalizations, mortality, and nursing home admissions.
Results
The study cohort included 25 130 Florida AL residents (mean [SD] age 81 [9] years); 3402 (13.5%) evacuated and 21 728 (86.5%) did not evacuate. The evacuated group had 2223 women (65.3%), and the group that sheltered-in-place had 14 556 women (67.0%). In the evacuated group, 42 residents (1.2%) were Black, 93 (2.7%) were Hispanic, and 3225 (94.8%) were White. In the group that sheltered in place, 490 residents (2.3%) were Black, 707 (3.3%) were Hispanic, and 20 212 (93.0%) were White. After 1:4 propensity score matching, when compared with sheltering-in-place, evacuation was associated with a 16% greater odds of emergency department visits (adjusted odds ratio [AOR], 1.16; 95% CI, 1.01-1.33; P = .04) and 51% greater odds of nursing home visits (AOR, 1.51; 95% CI, 1.14-2.00; P = .01) within 30 days of Hurricane Irma’s landfall. Hospitalization and mortality did not vary significantly by evacuation status within 30 or 90 days after the landfall date.
Conclusions and Relevance
In this cohort study of Florida AL residents, there was an increased risk of nursing home and emergency department visits within 30 days of Hurricane Irma’s landfall among residents from communities that evacuated before the storm when compared with residents from communities that sheltered-in-place. The stress and disruption caused by evacuation may yield poorer immediate health outcomes after a major storm for AL residents. Therefore, the potential benefits and harms of evacuating vs sheltering-in-place must be carefully considered when developing emergency planning and response.
Introduction
With hurricanes increasing in frequency and intensity in recent years, the decisions of long-term care administrators whether to evacuate or shelter-in-place, deserve increased attention.1 Older adults, especially those who live in long-term care communities, are already at increased risk of morbidity and mortality during and after hurricanes.1,2,3,4,5,6,7,8 As many as 433 excess deaths were reported among nursing home residents that were attributable to exposure to Hurricane Irma.2 Previous research has found that evacuation further increases the risk of mortality among nursing home residents after a disaster,8,9,10,11,12,13,14,15 especially for residents with dementia11 and those with high levels of functional impairment.12 Functional and cognitive impairments can limit the capability for long-term care residents to respond appropriately during an evacuation. Furthermore, stress from relocation can exacerbate the effect of the storm on poor health.1
Although sheltering-in-place is associated with better outcomes than evacuating on average,8,9,10,11,12,13,14,15 it can present difficulties such as prolonged power loss, flooding, and exterior damage to the building that may pose serious threats to resident and staff safety and well-being.16 Rehabilitation Center at Hollywood Hills received national attention for failing to evacuate when 12 residents died due to heat exposure after Hurricane Irma.16 As the decision to evacuate or shelter-in-place is not always straightforward given differential resident, facility, and event characteristics,1,10 research is needed to better understand the association between evacuation decisions and resident outcomes in different contexts.
Importantly, little is known about the association between evacuation vs sheltering-in-place and health outcomes among assisted living (AL) residents. AL communities provide housing, at least 2 meals per day, and oversight 24 hours per day to a primarily older adult population.17 AL communities are home to over 800 000 older adults who need assistance with personal care nationwide. Over 30% of residents have dementia and over 50% need assistance with transferring.1 Previous work has found that AL residents experienced an increased risk of emergency department (ED) visits after Hurricane Irma.5 However, it is unclear whether ED visits were related directly to residents’ experiences while sheltering-in-place or evacuating.
Despite being home to a population with high levels of impairment,17 AL communities are regulated at the state level and tend to have less specific emergency preparedness regulations than nursing homes.18 Previous qualitative studies have reported that AL communities that evacuated during hurricanes experienced similar difficulties as nursing homes, such as maintaining adequate staffing19 and communicating with family members.20,21 A survey that combined Texas AL communities with nursing homes found higher rates of resident mortality after Hurricane Katrina among communities that evacuated compared with those that did not; however, AL communities were not examined separately from nursing homes in the analysis.22
Hurricane Irma made landfall as a category 3 storm in southwest Florida, exposing 85 000 residents in the 3112 Florida AL communities to storm-related issues, including strong winds and flooding.23 The objective of this study was to compare 30- and 90-day poststorm health outcomes in AL residents exposed to Hurricane Irma who evacuated compared with a matched sample of AL residents who sheltered-in-place. Based on previous research with nursing home residents, we expected to observe worse health outcomes among residents from AL communities that evacuated compared with those in ALs that made the decision to shelter in place.8,9,10,11,12,13,14,15
Methods
Data
The institutional review board at Brown University approved the research protocol for this cohort study. The need for informed consent was waived because data were deidentified. The study followed the Strengthening the Reporting of Observational Studies in Epidemiology (STROBE) reporting guidelines. Information about AL addresses, number of beds, license type, and payment type came from the Florida Agency for Healthcare Administration website. AL resident characteristics came from Medicare administrative data. Specifically, data on resident demographic characteristics came from the Medicare Master Beneficiary Summary File. The Chronic Conditions Warehouse segment contained beneficiaries’ chronic conditions and the Residential History File (a file that combines Medicare claims and assessment data) was used to identify hospitalizations and nursing home visits.24 Additionally, we used the area deprivation index (ADI) file, a resource created and made publicly available by researchers at the University of Wisconsin-Madison to assess geographic differences in socioeconomic status and relative deprivation.25
Cohort
We used a previously published method26 to identify AL residents living in Florida AL communities with at least 25 beds on September 10, 2017. We included all of the 25 or more bed AL communities in the entire state of Florida because of the large size and uncertain path of Hurricane Irma (945 AL communities).23 We excluded individuals aged less than 65 years (5178 individuals [11.1%]) and individuals residing in any other setting (ie, nursing home or hospital) on September 10, 2017 (4464 individuals [9.6%]). As claims data are not complete for individuals enrolled in Medicare Advantage, we excluded AL residents with any Medicare Advantage coverage within the 6 months before September 10, 2017 (10 163 individuals [21.8%]). Of the 26 831 residents that met our inclusion criteria, we excluded 1506 residents (5.6%) with missing ADI data due to data suppression for low population density or a high group quarters population and 195 (0.7%) with missing information on for-profit status; 131 AL communities were omitted from the analysis after applying the resident exclusion criteria. Additional information about the data and methods used for these analyses can be found in the Brown Digital Repository.27
Measures
For our outcome measures, we created 8 separate indicators of whether each resident visited the ED (outpatient only), was hospitalized, died, or had a nursing home stay within 30 or 90 days after the hurricane landfall on September 10. For demographic characteristics, we included age in years, female sex, race and ethnicity (Black, Hispanic, non-Hispanic White, and other races and ethnicities), and dual eligibility for Medicare and Medicaid on September 10, 2017. Race and ethnicity were assessed using the Research Triangle Institute variable in the Medicare Master Beneficiary Summary File. This variable imputes race based on surname, geographic location, and self-reported data.28 We included race and ethnicity because of the inequities in access to AL experienced by Black and Hispanic individuals that may impact whether they lived in an AL that evacuated. We created the following categories: Black or African American, Hispanic, White, and Other race. The category other race contains American Indian or Alaska Native, Asian or Pacific Islander, and unknown race.29
We also included indicators for whether the AL resident had been diagnosed with any of the following chronic conditions before September 10, 2017: Alzheimer disease and related dementias, anemia, atrial fibrillation, chronic kidney disease, chronic obstructive pulmonary disease, depression, diabetes, heart failure, hyperlipidemia, hypertension, ischemic heart disease, and stroke. At the AL community level, we incorporated the following AL characteristics that have been found to have a significant relationship with residents’ health outcomes, such as the number of beds, whether the AL was a for-profit entity, and the AL’s license (standard without a specialty license, limited mental health, extended congregate care/limited nursing service).23
Additionally, we incorporated distance from the storm path in the 24 hours before landfall in kilometers using 4 categories based on the quartiles of the distribution (0-24.01, 24.02-110.46, 110.46-182.87, and >182.87). We also included the area disadvantage of the neighborhood in which the AL community was located using the ADI. The ADI is an index that combines US Census measures pertaining to education, employment, housing, and poverty to create an index of neighborhood disadvantage at the Census block–group level. ADI scores range from 1 to 100. Higher scores on the ADI indicate higher levels of disadvantage.20 Areas with higher ADI often have worse health outcomes26; AL communities in these neighborhoods also may have fewer resources to evacuate.
Statistical Analysis
We used propensity scores from a logistic regression model to match individuals in an AL community that evacuated to individuals in an AL community that did not evacuate using a 4:1 optimal matching algorithm without replacement.30 A 4:1 matching ratio was chosen to maximize statistical power.30 We applied a caliper of 0.2 for the propensity score matching process, which allowed us to restrict the matching to pairs of AL residents with propensity scores that were within 0.2 SDs of each other. A standardized mean difference of less than 0.2 is considered a threshold indicating a satisfactory match.31 We attempted to match on all covariates; however, we excluded the ADI index and the distance measure from the propensity score because it resulted in a standardized mean difference of greater than 0.2. The remaining standardized mean differences, all below 0.2, indicated a satisfactory match (see eTable in Supplement 1).
After the matching process, we assessed whether there were statistically significant differences between the AL communities that evacuated and the AL communities that did not evacuate using t tests and χ2 tests. Next, using the matched sample, we examined the association between evacuation status and health outcomes within multilevel logistic regression models with a random intercept. These models adjusted for ADI and distance to the storm path. Odds ratios and 95% CIs (thus setting the P value at a 2-tailed .05 significance level) are reported. Odds above 1.00 indicate an increased risk of an outcome; odds ratios below 1.00 indicate a reduced risk of an outcome. Data were analyzed using SAS version 9.4 (SAS Institute) from September 2022 to February 2024.
Results
The study cohort included 25 130 Florida AL residents. The 3402 residents (13.5%) who evacuated were similar in age to the 21 728 (86.5%) who did not evacuate (mean [SD] age, 81 [9] years). The evacuated group had 2223 women (65.3%), and the group that sheltered-in-place had 14 556 women (67.0%) (Table 1). In the evacuated group, 42 residents (1.2%) were Black, 93 (2.7%) were Hispanic, and 3225 (94.8%) were White. In the group that sheltered in place, 490 residents (2.3%) were Black, 707 (3.3%) were Hispanic, and 20 212 (93.0%) were White. Residents who evacuated lived in larger AL communities, on average, than residents who were not evacuated (mean [SD] number of beds, 107 [50.8] vs 100 [51.6]). Evacuated residents were also more likely to live in for-profit AL communities than residents who sheltered in place (2706 residents [79.5%] vs 16 263 residents [74.9%]) and live in wealthier areas as measured by the ADI (mean [SD], 4.5 [2.9] vs 5.4 [2.7]). After 1:4 propensity score matching, our sample included 17 010 residents; 3402 of these residents evacuated, and 13 608 residents sheltered-in-place. Of the 789 AL communities in the matched sample, 108 AL communities evacuated, and 681 sheltered-in-place.
Table 1. Baseline Characteristics of Assisted Living Residents and Florida Assisted Living Communities by Evacuation Status, Before and After Propensity-Based Matching.
| Variable | Residents, No. (%) | P value | Propensity-matched, residents who sheltered-in- place, No. (%) | P value | |
|---|---|---|---|---|---|
| Residents who evacuated | Before matching, residents who sheltered-in-place | ||||
| Total residents | 3402 (13.5) | 21 728 (86.5) | <.001 | 13 608 (80.0) | <.001 |
| Age, mean (SD), y | 80.88 (8.9) | 81.04 (8.8) | .31 | 80.82 (8.9) | .73 |
| Sex | |||||
| Female | 2223 (65.3) | 14 556 (67.0) | .06 | 8901 (65.4) | .87 |
| Male | 1179 (34.7) | 7172 (33.0) | .06 | 4707 (34.6) | .94 |
| Race and ethnicity | |||||
| Black or African American | 42 (1.2) | 490 (2.3) | <.001 | 142 (1.0) | .34 |
| Hispanic | 93 (2.7) | 707 (3.3) | .11 | 377 (2.8) | .91 |
| White | 3225 (94.8) | 20 212 (93.0) | <.001 | 12 927 (95.0) | .64 |
| Othera | 42 (1.2) | 319 (1.5) | .29 | 162 (1.2) | .83 |
| Dual eligibility for Medicare and Medicaid | 325 (9.6) | 2474 (11.4) | .002 | 1342 (9.9) | .59 |
| Capacity, mean (SD) | 107.2 (50.8) | 99.92 (51.6) | <.001 | 106.65 (53.9) | .55 |
| Area disadvantage index, mean (SD)b | 4.5 (2.9) | 5.43 (2.7) | <.001 | 5.44 (2.7) | <.001 |
| For profit | 2706 (79.5) | 16 263 (74.9) | <.001 | 10 962 (80.6) | .18 |
| License types | |||||
| Limited mental health | 116 (3.4) | 881 (4.0) | .12 | 470 (3.5) | .90 |
| Extended congregate care or limited nursing services | 1470 (43.2) | 10 499 (48.3) | <.001 | 5893 (43.3) | .92 |
| Distance from storm path 24 h before landfall, mean (SD), kmb | 95.31 (83.8) | 106.16 (82.6) | <.001 | 104.72 (82.4) | <.001 |
| Chronic conditionsc | |||||
| Alzheimer disease and related dementias | 1397 (41.1) | 9388 (43.2) | .02 | 5593 (41.1) | .97 |
| Anemia | 2833 (83.3) | 18 194 (83.7) | .50 | 11 341 (83.3) | .93 |
| Atrial fibrillation | 1052 (30.9) | 6531 (30.1) | .31 | 4233 (31.1) | .84 |
| Chronic kidney disease | 1624 (47.7) | 10 711 (49.3) | .09 | 6519 (47.9) | .86 |
| Chronic obstructive pulmonary disease | 1418 (41.7) | 9234 (42.5) | .37 | 5698 (41.9) | .84 |
| Depression | 1760 (51.7) | 11 658 (53.7) | .04 | 7052 (51.8) | .93 |
| Diabetes | 1545 (45.4) | 10 308 (47.4) | .03 | 6176 (45.4) | .98 |
| Heart failure | 1497 (45.0) | 9727 (44.8) | .41 | 5996 (44.1) | .95 |
| Hyperlipidemia | 3074 (90.4) | 19 978 (92.0) | .002 | 12 331 (90.6) | .64 |
| Hypertension | 3088 (90.8) | 19 995 (92.0) | .01 | 12 376 (91.0) | .75 |
| Ischemic heart disease | 2542 (74.7) | 16 319 (75.1) | .63 | 10 208 (75.0) | .72 |
| Stroke/transient ischemic attack | 1010 (29.7) | 6815 (31.4) | .05 | 4028 (29.6) | .92 |
Includes American Indian or Alaska Native, Asian or Pacific Islander, and unknown race.
Not included in the propensity score.
First diagnosed before the landfall date.
Unadjusted outcomes of the propensity-matched sample are displayed in Table 2. In Table 3, we display the association between evacuation status and health care outcomes using the matched sample, adjusting for ADI and distance to the storm path. In the adjusted model, those who evacuated had a 16% greater odds of an ED visit when compared with those who sheltered-in-place (adjusted odds ratio [AOR], 1.16; 95% CI, 1.01-1.33; P = .04) at 30 days after the storm and an 11% greater odds of an ED visit within 90 days of the storm (AOR, 1.11; 95% CI, 1.01-1.23; P = .04). Risk of hospitalization and mortality did not differ by resident evacuation status at 30 days or 90 days. AL residents who evacuated had about 51% greater odds of experiencing a nursing home placement within 30 days of the storm (AOR, 1.51; 95% CI, 1.14-2.00; P = .01); this association was not statistically significant at 90 days.
Table 2. Outcome Variables by Evacuation Status Before and After Matching.
| Variable | Residents, No. (%) | P value | Propensity-matched, residents who sheltered-in- place, No. (%) (n = 13 608) | P value | |
|---|---|---|---|---|---|
| Residents who evacuated (n = 3402) | Before matching, residents who sheltered-in-place (n = 21 728) | ||||
| Emergency department visit | |||||
| Within 30 d of landfall | 218 (6.4) | 1258 (5.8) | .15 | 746 (5.5) | .04 |
| Within 90 d of landfall | 487 (14.3) | 3079 (14.2) | .82 | 1882 (13.8) | .46 |
| Hospitalization | |||||
| Within 30 d of landfall | 174 (5.1) | 1110 (5.1) | .99 | 672 (4.9) | .67 |
| Within 90 d of landfall | 407 (12.0) | 2729 (12.6) | .33 | 1669 (12.3) | .63 |
| Died | |||||
| Within 30 d of landfall | 34 (1.0) | 159 (0.7) | .10 | 92 (0.7) | .049 |
| Within 90 d of landfall | 105 (3.1) | 579 (2.7) | .16 | 356 (2.6) | .13 |
| Nursing home visit | |||||
| Within 30 d of landfall | 86 (2.5) | 435 (2.0) | .045 | 268 (2.0) | .04 |
| Within 90 d of landfall | 199 (5.9) | 1249 (5.8) | .81 | 748 (5.5) | .42 |
Table 3. Logistic Regression of the Association Between Evacuation Status and Outcomes in the Matched Sample.
| Outcome variable | AOR (95% CI)a | P value |
|---|---|---|
| Emergency department visit | ||
| Within 30 d of landfallb | 1.16 (1.01-1.33) | .04 |
| Within 90 d of landfallb | 1.11 (1.01-1.23) | .04 |
| Hospitalization | ||
| Within 30 d of landfall | 1.17 (0.97-1.42) | .11 |
| Within 90 d of landfall | 1.08 (0.95-1.24) | .23 |
| Died | ||
| Within 30 d of landfall | 1.34 (.85-2.10) | .21 |
| Within 90 d of landfall | 1.22 (0.94-1.56) | .13 |
| Nursing home visit | ||
| Within 30 d of landfall | 1.51 (1.14-2.00) | .01 |
| Within 90 d of landfall | 1.19 (1.00-1.43) | .57 |
Abbreviation: AOR, adjusted odds ratio.
Models adjusted for distance from the storm path in the 24 hours before landfall and area disadvantage index.
The emergency department visits are outpatient only.
Discussion
Using a propensity-matched sample design including Florida AL residents, this study examined the 30- and 90-day health care outcomes of the strategy to deal with an approaching hurricane by evacuation vs sheltering-in-place. With hurricanes increasing in frequency and intensity in recent years, the question of whether to evacuate or shelter-in-place may become even more critical. Although studies have examined the impact of evacuation among older adults in settings such as nursing homes and the community, this is the first study to examine the association between evacuation and outcomes in the AL setting.8,9,10,11,12,13,14,15
We found that 30 days after storm landfall, AL residents who evacuated experienced greater risk of ED visits and nursing home stays than their propensity score–matched counterparts who sheltered-in-place. This finding is largely consistent with research reported in nursing home settings, where adverse outcomes were also reported following evacuation, particularly among more frail residents.8,9,10,11,12,13,14,15 The exact causes that underlie the apparent adverse effects of evacuation cannot be discerned from our study, but are likely to include psychological stress associated with removing AL residents—who may have vision and hearing difficulties, dementia, and functional impairment—from their familiar environment.11 This agitation has been observed among AL residents in qualitative research.21 These combined factors may contribute to increased morbidity and mortality. This finding highlights the importance of comprehensive disaster planning for AL community administrators and the county emergency operations officials issuing evacuation orders.
Most associations between evacuation and health care outcomes were not statistically significant at 90 days following the storm.7 Furthermore, there was no association between evacuation and mortality among AL residents. These findings deviate from results with nursing home residents. The lack of association may reflect the possibility that AL residents more readily stabilize poststorm and/or may be a more resilient population than nursing home residents to the long-term consequences of evacuation. For example, an estimated 34% of AL residents have dementia, nationally, when compared with 49% of nursing home residents.17 Nursing home residents also have much higher levels of functional impairment.17
The decision whether to evacuate or shelter-in-place is undoubtedly nuanced and influenced by a full spectrum of factors. Some AL communities may need to evacuate to protect residents from wind damage and storm surge as they are often in close proximity to the ocean.10,23 However, previous work has found that 65 AL administrators during Hurricane Irma decided to evacuate even if they were not in mandatory evacuation zones, for reasons that remain unclear.23 In qualitative work, some AL clinicians expressed insecurity about the structural integrity of their buildings.21 Risk of flooding or risk of a loss of power without an adequate backup source may be other factors driving voluntary evacuation. Due to the adverse associations between evacuation and resident outcomes, there is a need for AL clinicians and emergency management officials to communicate and to devise strategies that do not default to evacuation.
Given the potential benefits of sheltering-in-place, a particular concern is whether AL communities are sufficiently prepared to shelter-in-place during a hurricane or other natural disaster. AL and nursing home operators in Florida are both required to prepare and submit disaster preparedness plans to their local emergency management offices annually. However, AL communities differ from nursing homes in terms of the health and nursing services available to residents. Despite the health care needs of AL residents, particularly with medication management, direct care staffing and training tends to be lower in AL communities than in nursing homes.32 The interest in protecting nursing home residents spurred by Hurricane Katrina and storms that followed should be extended to AL residents exposed to the same dangers. Unlike other states, Florida requires that AL communities have comprehensive disaster preparedness plans. Therefore, the association between evacuation and outcomes may be more pronounced in other states that also experience extreme weather events. Consideration should be given, however, to ensure that clinicians with fewer resources are not disproportionately affected by any future mandates.
Limitations
Our analysis was not without limitations. Our study did not include those enrolled in Medicare Advantage, which may impact findings given differences in the population of Medicare beneficiaries enrolled in Medicare Advantage vs traditional Medicare. In 2017, approximately 42% of Florida residents were enrolled in Medicare Advantage.33 Our method was dependent on identifying residents in larger AL residences (25 or more beds) in Florida, which limited its generalizability to smaller settings and other states. In Florida, smaller communities serve a higher proportion of Medicaid residents and individuals with mental illness.34 Additionally, we do not have data regarding the severity of the chronic conditions or functional limitations. We also did not incorporate data on where AL residents were evacuated to. AL communities that evacuated after the date were included in the comparison group, which could bias our findings toward the null. Propensity score analysis helps adjust for unmeasured confounding on observed variables; however, as the study is observational, the possibility for residual confounding cannot be ruled out.
Conclusions
This cohort study of Florida AL residents found an increased risk of ED use and nursing home visits within 30 days of Hurricane Irma’s landfall among residents who evacuated before the storm when compared with residents who remained in the AL community. The immediate health-related impacts of evacuation should be taken into consideration when designing emergency preparedness plans for AL communities.
eTable. Standardized Mean Differences in the Variables Between the Residents who Evacuated and Sheltered-in-Place, Before and After Propensity Score Matching
Data Sharing Statement
References
- 1.Dosa DM, Hyer K, Brown LM, Artenstein AW, Polivka-West L, Mor V. The controversy inherent in managing frail nursing home residents during complex hurricane emergencies. J Am Med Dir Assoc. 2008;9(8):599-604. doi: 10.1016/j.jamda.2008.05.007 [DOI] [PubMed] [Google Scholar]
- 2.Dosa DM, Skarha J, Peterson LJ, et al. Association between exposure to Hurricane Irma and mortality and hospitalization in Florida nursing home residents. JAMA Netw Open. 2020;3(10):e2019460. doi: 10.1001/jamanetworkopen.2020.19460 [DOI] [PMC free article] [PubMed] [Google Scholar]
- 3.Hua CL, Patel S, Thomas KS, et al. The relationship between exposure to Hurricane Harvey and mortality among nursing home residents. J Am Geriatr Soc. 2023;71(3):888-894. doi: 10.1111/jgs.18143 [DOI] [PMC free article] [PubMed] [Google Scholar]
- 4.Dosa D, Feng Z, Hyer K, Brown LM, Thomas K, Mor V. Effects of Hurricane Katrina on nursing facility resident mortality, hospitalization, and functional decline. Disaster Med Public Health Prep. 2010;4 Suppl 1(0 1):S28-S32. doi: 10.1001/dmp.2010.11 [DOI] [PMC free article] [PubMed] [Google Scholar]
- 5.Hua CL, Thomas KS, Peterson LJ, Hyer K, Dosa DM. Emergency department use among assisted living residents after Hurricane Irma. J Am Med Dir Assoc. 2021;22(4):918-922.e1. doi: 10.1016/j.jamda.2020.10.010 [DOI] [PMC free article] [PubMed] [Google Scholar]
- 6.Bell SA, Miranda ML, Bynum JPW, Davis MA. Mortality after exposure to a hurricane among older adults living with dementia. JAMA Netw Open. 2023;6(3):e232043. doi: 10.1001/jamanetworkopen.2023.2043 [DOI] [PMC free article] [PubMed] [Google Scholar]
- 7.Bell SA, Abir M, Choi H, Cooke C, Iwashyna T. All-cause hospital admissions among older adults after a natural disaster. Ann Emerg Med. 2018;71(6):746-754.e2. doi: 10.1016/j.annemergmed.2017.06.042 [DOI] [PMC free article] [PubMed] [Google Scholar]
- 8.Willoughby M, Kipsaina C, Ferrah N, et al. Mortality in nursing homes following emergency evacuation: a systematic review. J Am Med Dir Assoc. 2017;18(8):664-670. doi: 10.1016/j.jamda.2017.02.005 [DOI] [PubMed] [Google Scholar]
- 9.Blanchard G, Dosa D. A comparison of the nursing home evacuation experience between hurricanes katrina (2005) and gustav (2008). J Am Med Dir Assoc. 2009;10(9):639-643. doi: 10.1016/j.jamda.2009.06.010 [DOI] [PMC free article] [PubMed] [Google Scholar]
- 10.Dosa D, Hyer K, Thomas K, et al. To evacuate or shelter in place: implications of universal hurricane evacuation policies on nursing home residents. J Am Med Dir Assoc. 2012;13(2):190.e1-190.e1907. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 11.Brown LM, Dosa DM, Thomas K, Hyer K, Feng Z, Mor V. The effects of evacuation on nursing home residents with dementia. Am J Alzheimers Dis Other Demen. 2012;27(6):406-412. doi: 10.1177/1533317512454709 [DOI] [PMC free article] [PubMed] [Google Scholar]
- 12.Thomas KS, Dosa D, Hyer K, et al. Effect of forced transitions on the most functionally impaired nursing home residents. J Am Geriatr Soc. 2012;60(10):1895-1900. doi: 10.1111/j.1532-5415.2012.04146.x [DOI] [PMC free article] [PubMed] [Google Scholar]
- 13.Nomura S, Blangiardo M, Tsubokura M, et al. Post-nuclear disaster evacuation and survival amongst elderly people in Fukushima: a comparative analysis between evacuees and non-evacuees. Prev Med. 2016;82:77-82. doi: 10.1016/j.ypmed.2015.11.014 [DOI] [PubMed] [Google Scholar]
- 14.Nomura S, Gilmour S, Tsubokura M, et al. Mortality risk amongst nursing home residents evacuated after the Fukushima nuclear accident: a retrospective cohort study. PLoS One. 2013;8(3):e60192. doi: 10.1371/journal.pone.0060192 [DOI] [PMC free article] [PubMed] [Google Scholar]
- 15.Castle NG, Engberg JB. The health consequences of relocation for nursing home residents following Hurricane Katrina. Res Aging. 2011;33(6):661-687. doi: 10.1177/0164027511412197 [DOI] [Google Scholar]
- 16.Skarha J, Gordon L, Sakib N, et al. Association of power outage with mortality and hospitalizations among Florida nursing home residents after Hurricane Irma. JAMA Health Forum. 2021;2(11):e213900. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 17.Sengupta M, Lendon JP, Caffrey C, Melekin A, Singh P. Post-acute and long-term care providers and services users in the United States, 2017-2018. Vital Health Stat 3. 2022;(47):1-93. [PubMed] [Google Scholar]
- 18.Kennedy KA, Hua CL, Nelson I. A comparison of emergency preparedness policies by long-term care setting type in Ohio: assisted living settings lag behind nursing homes in key areas. J Appl Gerontol. 2021;40(4):377-386. doi: 10.1177/0733464820902010 [DOI] [PubMed] [Google Scholar]
- 19.Jester DJ, Thomas KS, Peterson LJ, Dosa DM, Andel R, Hyer K. Effect of Hurricane Irma on daily direct-care nurse staffing in nursing homes. J Am Geriatr Soc. 2021;69(8):2298-2305. doi: 10.1111/jgs.17220 [DOI] [PMC free article] [PubMed] [Google Scholar]
- 20.Brown LM, Christensen JJ, Ialynytchev A, Thomas KS, Frahm KA, Hyer K. Experiences of assisted living facility staff in evacuating and sheltering residents during hurricanes. Curr Psychol. 2015;34:506-514. doi: 10.1007/s12144-015-9361-7 [DOI] [Google Scholar]
- 21.Peterson LJ, Dobbs D, June J, Dosa DM, Hyer K. “You just forge ahead”: the continuing challenges of disaster preparedness and response in long-term care. Innov Aging. 2021;5(4):igab038. doi: 10.1093/geroni/igab038 [DOI] [PMC free article] [PubMed] [Google Scholar]
- 22.Castro C, Persson D, Bergstrom N, Cron S. Surviving the storms: emergency preparedness in Texas nursing facilities and assisted living facilities. J Gerontol Nurs. 2008;34(8):9-16. doi: 10.3928/00989134-20080801-01 [DOI] [PubMed] [Google Scholar]
- 23.Peterson LJ, June J, Sakib N, et al. Assisted living communities during Hurricane Irma: the decision to evacuate or shelter in place and resident acuity. J Am Med Dir Assoc. 2020;21(8):1148-1152.e3. doi: 10.1016/j.jamda.2020.01.104 [DOI] [PMC free article] [PubMed] [Google Scholar]
- 24.Intrator O, Hiris J, Berg K, Miller SC, Mor V. The residential history file: studying nursing home residents’ long-term care histories. Health Serv Res. 2011;46(1 Pt 1):120-137. doi: 10.1111/j.1475-6773.2010.01194.x [DOI] [PMC free article] [PubMed] [Google Scholar]
- 25.Kind AJH, Buckingham WR. Making neighborhood-disadvantage metrics accessible—The Neighborhood Atlas. N Engl J Med. 2018;378(26):2456-2458. doi: 10.1056/NEJMp1802313 [DOI] [PMC free article] [PubMed] [Google Scholar]
- 26.Thomas KS, Dosa D, Gozalo PL, et al. A methodology to identify a cohort of Medicare beneficiaries residing in large assisted living facilities using administrative data. Med Care. 2018;56(2):e10-e15. doi: 10.1097/MLR.0000000000000659 [DOI] [PMC free article] [PubMed] [Google Scholar]
- 27.Hua CL. Code for evacuation and healthcare outcomes after Hurricane Irma. Brown University Library Digital Repository . 2024. Accessed March 13, 2024. https://repository.library.brown.edu/studio/item/bdr:vjrd2sw9/
- 28.Jarrín OF, Nyandege AN, Grafova IB, Dong X, Lin H. Validity of race and ethnicity codes in medicare administrative data compared with gold-standard self-reported race collected during routine home health care visits. Med Care. 2020;58(1):e1-e8. doi: 10.1097/MLR.0000000000001216 [DOI] [PMC free article] [PubMed] [Google Scholar]
- 29.Cornell PY, Zhang W, Smith L, Fashaw S, Thomas KS. Developments in the market for assisted living: residential care availability in 2017. J Am Med Dir Assoc. 2020;21(11):1718-1723. doi: 10.1016/j.jamda.2020.08.011 [DOI] [PubMed] [Google Scholar]
- 30.Xu S, Shetterly S, Cook AJ, et al. Evaluation of propensity scores, disease risk scores, and regression in confounder adjustment for the safety of emerging treatment with group sequential monitoring. Pharmacoepidemiol Drug Saf. 2016;25(4):453-461. doi: 10.1002/pds.3983 [DOI] [PMC free article] [PubMed] [Google Scholar]
- 31.Stuart EA. Matching methods for causal inference: A review and a look forward. Stat Sci. 2010;25(1):1-21. doi: 10.1214/09-STS313 [DOI] [PMC free article] [PubMed] [Google Scholar]
- 32.Zimmerman S, Sloane PD, Katz PR, Kunze M, O’Neil K, Resnick B. The need to include assisted living in responding to the COVID-19 pandemic. J Am Med Dir Assoc. 2020;21(5):572-575. doi: 10.1016/j.jamda.2020.03.024 [DOI] [PMC free article] [PubMed] [Google Scholar]
- 33.Gretchen J, Damico A, Neuman T, Gold M. Medicare Advantage 2017 spotlight: enrollment market update. Kaiser Family Foundation . 2017. Accessed October 16, 2023. https://www.kff.org/medicare/issue-brief/medicare-advantage-2017-spotlight-enrollment-market-update/
- 34.June JW, Dobbs D, Molinari V, Meng H, Bowblis JR, Hyer K. Factors associated with assisted living facility closure. Gerontologist. 2022;62(2):181-189. doi: 10.1093/geront/gnab105 [DOI] [PubMed] [Google Scholar]
Associated Data
This section collects any data citations, data availability statements, or supplementary materials included in this article.
Supplementary Materials
eTable. Standardized Mean Differences in the Variables Between the Residents who Evacuated and Sheltered-in-Place, Before and After Propensity Score Matching
Data Sharing Statement