Introduction
Individuals with numerous chronic diseases and functional limitations face an increased risk of mortality during major disasters.Reference Mace and Doyle 1 -Reference Mace, Doyle and Askew 3 After Hurricane Katrina (2005), the Louisiana Department of Health invested heavily in the development of an MNS system to facilitate the triage and care of individuals with medical and functional needs that general population shelters could not effectively accommodate. Reference Shao, Jackson and Ha4,Reference Greenough, Lappi and Hsu5 The system includes pre-specified locations, supply and pharmaceutical caches, and staffing augmented by state employees, public-private partnerships, and occasionally, support via the National Disaster Medical System (NDMS). This report describes the clinical attributes of individuals served within all activations of Louisiana’s MNS following 3 major hurricanes from 2020 to 2021.
During emergencies, the MNS system identifies qualifying individuals via real-time triage or pre-existing registries of at-risk persons. Those with minor mobility limitations or the ability to self-manage chronic conditions are directed toward general population shelters. Rare situations (e.g., ventilator-dependence) require hospital admission as a sheltering mechanism. The MNS system aims to provide care for individuals between these two extremes. Louisiana’s nursing homes are legally required to pre-identify evacuation processes should they experience critical damage during a hurricane. 6 However, during Hurricane Ida, several facilities failed to safely execute evacuation plans, requiring state-led evacuation and temporary care of many patients within the MNS system. 7,Reference Gallo and Russell8 Their records are intentionally included in this review given the realities faced by the shelter system during these events.
Within the shelters, nurses/nursing assistants, APPs, and physicians provide 24/7 care. Patients are permitted one personal caregiver (i.e., family member) who may stay alongside them. Nurse case managers, licensed clinical social workers/behavioral health specialists, and pharmacists are also present daily. Paramedics are located onsite to respond to medical emergencies. Durable medical equipment, disposable supplies, and extensive pharmaceutical caches are deployed to each site. There are no laboratory or imaging capabilities onsite. Examples of shelter care include tube feeds, respiratory (including tracheostomy) care, wound care, medication management (including insulin, IV fluids, some IV antibiotics), and dialysis coordination. Severe psychiatric concerns require stabilization via local emergency departments, as there is no capability for psychiatric consultation on site. Shelters also facilitate discharge planning of evacuees to their home (if habitable post-disaster) or transition to subsidized housing or skilled nursing facility (SNF).
Methodology
A retrospective chart review was performed on all patients admitted to MNSs after Hurricanes Laura (2020), Delta (2020), and Ida (2021). Shelters utilize standardized paper medical records, which include demographic information, medical history, and home medications. Clinician notes/orders, vital signs, and a medication administration record (MAR) are included in this record. If available, existing medical records (i.e., printed discharge papers from a transferring facility) are collated into the chart. Data was extracted by a single reviewer into an electronic database for further review.
Sixteen of the 696 records reviewed were excluded due to a patient’s failure to arrive or near-complete absence of clinical information. Absence of one or more clinical or demographic variables (e.g., undocumented race) prompted exclusion from analysis of that individual variable but was otherwise included. Data was analyzed using R and Microsoft Excel. This study was considered exempt from oversight by the Louisiana Department of Health Institutional Review Board.
Results
Among all patients, average length of stay (LOS) within a shelter was 5.2 days (SD ± 4.0 d, range 1-29 d). A shorter average LOS was witnessed during the Laura response (3.6 d), compared to Delta (5.2 d) and Ida (5.2 d). The average delay from storm landfall to shelter arrival (per person) was 6.0 days; patients had notably longer delays after Ida (6.6 d), compared to Laura and Delta (−0.8 d and 3.8 d, respectively). Specific reasons for a patient’s delay in presentation are indeterminable from available records, as was the presence or absence of intermediate care (e.g., hospitalization) during their delay. Patients had a mean age of 66 years; 51.5% identified as male and 48.5% as female. Further descriptive statistics are contained within Table 1A.
Demographic characteristics of shelter patients by storm
Table 1B describes the prevalence of disease states and medical needs among shelter patients. Patients had significant resource needs including supplemental oxygen (21.0%), hemodialysis (7.1%), tube feeds (5.3%), wound care (23.5%), and chronic urinary catheterization (8.7%). Regarding mobility, 29.3% of patients were bedridden, 37.7% were wheelchair-dependent, 18.4% could ambulate with assistance or walker, while only 14.6% were independently ambulatory. Nonverbal individuals comprised 7.7% of all patients.
Frequency of comorbid conditions among all shelter patients (n = 680)
Notes. *ESRD on HD patients not included in above category of CKD (not ESRD).
**DNR/DNI proportion calculated of those with documented code status (47.8%, n = 325).
Comorbid conditions were considered independent of one another. Pre-existing psychiatric disease was prevalent in 37.4% of patients, with 15.4% having two or more comorbid psychiatric conditions. An additional 6.2% of patients had active substance use disorder. Neurocognitive dysfunction (including unspecified dementia, Alzheimer’s, Parkinson’s, etc.) existed in 24.6% of patients, prior cerebrovascular accident in 24.3%, and a seizure disorder in 15.2%. Cardiovascular disease was the most prevalent disease category in the shelters, with 16.8% of patients carrying a diagnosis of congestive heart failure and 13.7% coronary artery disease. Among pulmonary diseases, chronic obstructive pulmonary disease (COPD) was the most prevalent diagnosis (23.5%). Patients had a high burden of comorbid cardiopulmonary disease: over half of patients (51.1%) had two or more cardiopulmonary diagnoses, 27.9% had 3 or more, and 12.8% had 4 or more distinct diagnoses. Insulin-dependent diabetes affected 18.1% of all patients. Active cancer was present in 4.0% of patients. Of all patients, 5.7% were actively receiving hospice care at the time of admission to a shelter. Table 1B contains additional findings related to patient comorbidities and clinical needs.
COVID-19 testing was administered upon admission and as clinically appropriate. A positive test occurred in 5.6% of patients, who were cohorted into isolated areas.
During sheltering, 20.6% of patients experienced a documented clinically significant event, of which 77.9% involved transport for evaluation at a local emergency department. Such events were defined as occurrences with an unexpected clinical change in the patient’s status requiring immediate evaluation by shelter medical staff and potential transportation. Examples included falls, mental status changes, physical exam or vital sign abnormalities, or patient complaints of concerning symptoms such as chest pain, new shortness of breath, or severe abdominal pain. Respiratory (17.9%), cardiac (e.g., chest pain, dysrhythmia [16.4%]), wound care or soft tissue infection (15.0%), psychiatric (e.g., suicidal ideation, agitation [10.7%]), non-soft tissue infection (7.1%), altered mental status (6.4%), and gastrointestinal (6.4%) concerns were the most common categories of event, with falls accounting for an additional 5.0% of events. One patient died while in the shelter.
Discussion
This retrospective analysis of 680 patients admitted to medical shelters successfully describes the varied and significant clinical needs of this subset of displaced individuals following major hurricanes. Among the study population, age, sex, race/ethnicity, and English proficiency were consistent across sites. However, the sheltered population did not resemble Louisiana’s population demographics. 9 Certain demographic groups may avoid shelters due to government mistrust, unawareness, or social support networks that preclude sheltering needs. Thus, there may be limited generalizability to regions with variable socioeconomic dynamics.
In two storms (Delta and Ida), a multi-day delay separated storm landfall and patients’ presentation to a medical shelter. This delay is likely multifactorial and attributable to characteristics of the individual storm (e.g., forecast accuracy), community vulnerability, and triage or transportation delays. During this gap, some patients may receive medical care (e.g., hospitalization); however, others presumably experience a lack of utilities, safe housing, and general medical care prior to shelter arrival. Those with severe decompensation are more likely to present to emergency departments (and thus absent from this dataset); however, those with more subtle exacerbations may present to shelters. Shelters caring for individuals with significant gaps in care may encounter a higher burden of clinical needs among patients. Efforts to reduce this delay where possible (e.g., expedited evacuation and transportation) may improve patient outcomes and lessen shelter workloads.
During Ida, patients presenting from SNF or assisted living facility (ALF) were overrepresented compared to other storms, a direct result of multiple unplanned facility evacuations into the shelters. Contrary to MNS design, though necessary due to extreme need at the time, state shelters pivoted to accommodate patients from uninhabitable SNFs/ALFs. This patient population likely contributes to the high rate of discharge to SNF/ALF (45.7% during all storms), as patients already in a nursing facility are likely to return after emergency sheltering.
Overall, disease states were likely underreported, given that few patients (31.9%) arrive with written medical records, thereby increasing reliance on self-reported medical history. Nonetheless, overall medical complexity suggests that shelters must ensure the presence of the appropriate skilled personnel (e.g., behavioral health, wound care), resources (e.g., oxygen concentrators, tube feeds), and partnerships (e.g., offsite hemodialysis). This population’s mobility limitations (and the implications on other ADLs) suggest an even greater need for nursing staff than anticipated. 10
Clinically significant events occurred frequently in the shelters. Minor events (e.g., falls without injury) are likely underreported. It is unclear which events might be related to shelter care versus the anticipated incidence of complications for this population at baseline (such as a report of chest pain from an individual with multiple underlying cardiopulmonary conditions). Exacerbations of psychiatric disease may be inevitable in a shelter, given the psychological stress of displacement, disruption of normative social support networks, and especially during the COVID-19 pandemic, self-perceived risk of disease transmission due to congregate sheltering. Other diseases may face hazard-specific insults (e.g., COPD exacerbations in wildfires, wound infections in flooding) in disasters, regardless of sheltered status.Reference Dayrit, Bintanjoyo and Andersen 11 –Reference Ismail, Warner and Chaudhry 13 However, appropriate capacity for onsite stabilization and urgent transportation should be considered essential within medical shelters. Local hospitals may also consider how nearby medical shelter operations may impact ED utilization and prepare accordingly.
Conclusions
In this large retrospective analysis of 680 patients cared for within Louisiana’s MNS system following three major hurricanes, patients were found to have high degrees of medical complexity and reliance on medical shelters for management of chronic disease. Sites supported the temporary care of individuals with significant medical needs previously living at home or in licensed long-term care facilities and directly accepted many patients from local acute-care hospitals. Clinically significant events were common among the shelter population, frequently requiring transport to hospital emergency departments. Major limitations of this study include suspected underreporting of medical needs, discordance of shelter population compared to statewide demographics, and a retrospective nature that precludes a causal understanding of patient outcomes. Emergency managers may consider these findings in the design of mass care systems for vulnerable populations, and future research may explore relationships among disease states and health outcomes within a shelter setting.
Author contribution
JAH contributed to the scope of the project, performed primary data extraction, analysis, and drafted the manuscript. RP contributed to the scope of the project, provided technical assistance, and drafted the manuscript.
Funding statement
This study was granted exemption by the Louisiana Department of Health Institutional Review Board. Funding for the initial data extraction was provided by the Louisiana Hospital Association under the federal Hospital Preparedness Program grant.
Competing interests
None.
