Impact of Hurricanes and Associated Extreme Weather Events on Cardiovascular Health: A Scoping Review

Abstract

Background:

The frequency and destructiveness of hurricanes and related extreme weather events (e.g., cyclones, severe storms) have been increasing due to climate change. A growing body of evidence suggests that victims of hurricanes have increased incidence of cardiovascular disease (CVD), likely due to increased stressors around time of the hurricane and in their aftermath.

Objectives:

The objective was to systematically examine the evidence of the association between hurricanes (and related extreme weather events) and adverse CVD outcomes with the goal of understanding the gaps in the literature.

Methods:

A comprehensive literature search of population-level and cohort studies focused on CVD outcomes (i.e., myocardial infarction, stroke, and heart failure) related to hurricanes, cyclones, and severe storms was performed in the following databases from inception to December 2021: Ovid MEDLINE, Ovid EMBASE, Web of Science, and The Cochrane Library. Studies retrieved were then screened for eligibility against predefined inclusion/exclusion criteria. Studies were then qualitatively synthesized based on the time frame of the CVD outcomes studied and special populations that were studied. Gaps in the literature were identified based on this synthesis.

Results:

Of the 1,103 citations identified, 48 met our overall inclusion criteria. We identified articles describing the relationship between CVD and extreme weather, primarily hurricanes, based on data from the United States (42), Taiwan (3), Japan (2), and France (1). Outcomes included CVD and myocardial infarction–related hospitalizations (30 studies) and CVVD-related mortality (7 studies). Most studies used a retrospective study design, including one case–control study, 39 cohort studies, and 4 time-series studies.

Discussion:

Although we identified a number of papers that reported evaluations of extreme weather events and short-term adverse CVD outcomes, there were important gaps in the literature. These gaps included a) a lack of rigorous long-term evaluation of hurricane exposure, b) lack of investigation of hurricane exposure on vulnerable populations regarding issues related to environmental justice, c) absence of research on the exposure of multiple hurricanes on populations, and d) absence of an exploration of mechanisms leading to worsened CVD outcomes. Future research should attempt to fill these gaps, thus providing an important evidence base for future disaster-related policy. https://doi.org/10.1289/EHP11252

Introduction

Hurricanes and related extreme weather events (e.g., cyclones and severe storms) are destructive events and are likely to increase in both number and destructiveness driven by climate change. In the United States alone, severe storms such as hurricanes have increased by 37% between 1990 and 2020,¹ costing an estimated USD $\$1.8\text{\hspace{0.17em}trillion}$.¹ Substantial populations across the globe, including the United States, reside in coastal regions, often in large, diverse, urban environments, making them vulnerable to the adverse effects of hurricanes,² including loss of life and other morbidity, property damage, and homelessness.

There is growing evidence that stress induced by hurricanes increases the risk of adverse cardiovascular disease (CVD) events, including increased risk of CVD-related mortality and hospitalization from coronary heart disease (CHD)³ and stroke.³ Additionally, studies have suggested an increased incidence and worsened control of chronic CVD risk factors such as diabetes^4,5 and hypertension.⁶ These effects have been documented immediately after such events, but also in the months to years following, suggesting that the consequences of hurricanes extend beyond their immediate effects.^5,7

Concerns exists that certain racial/ethnic groups,⁸ older individuals,^9,10 and residents of socioeconomically disadvantaged neighborhoods¹¹ are at increased risk after hurricanes of CVD-related morbidity¹² and death^3,13,14 in comparison with other segments of the population. Why this occurs is unclear but may be driven by predisaster factors such as insurance status¹⁵ and postdisaster factors such as access to health care and financial resources that would otherwise insulate vulnerable individuals from short- and long-term harm.^15–17 Furthermore, evidence suggests these vulnerable subgroups may perceive threats from hurricanes differently¹⁸ and lack the means to leave,^19–22 forcing them to endure the hurricanes’ effects while sheltering in place.²³

Hurricanes and related extreme weather events are likely to worsen into the future.²⁴ Currently no systematic reviews of the literature exist that focus on the relationship between hurricanes (and associated extreme weather events) and CVD outcomes. Therefore, to better understand the association between hurricanes and CVD outcomes, our objective was to systematically examine the peer-reviewed evidence that describes the association between hurricanes and adverse CVD outcomes. A specific goal of this study was to understand the gaps in the scientific literature to develop a research agenda that ultimately will provide the scientific community with the means to mitigate the adverse effects of hurricanes.

Methods

We performed a scoping review to qualitatively describe the evidence from studies that examined the exposure of hurricanes and related extreme weather events on CVD. This study was performed following the Preferred Reporting Items for Systematic Reviews and Meta-Analyses extension for scoping reviews (PRISMA-ScR).²⁵ In accordance with these standards, a protocol was submitted and preregistered by the International Prospective Register of Systematic Reviews (PROSPERO; CRD42022299034). The PRISMA flow diagram is described in Figure S1.

Search Strategy

A medical librarian (MRD) performed comprehensive searches to identify studies that addressed the effects of hurricanes or extreme weather events on CVD outcomes, including a range of search terms related to myocardial ischemia and/or myocardial infarction (MI), heart failure, and stroke. Full description of the search terms is described in the Supplemental Material, “Search Terms by Database.”

Searches were run on 27 December 2021 in the following databases: Ovid MEDLINE (ALL - 1946 to present), Ovid EMBASE (1974 to present), The Cochrane Library (Wiley, 1996 to present), and Web of Science (Clarivate, 1900 to present). The search strategy included all appropriate controlled vocabulary and keywords for the concepts of extreme weather events (e.g., cyclone, hurricane, tropical storm, monsoon, tropical storm, tropical depression) and the cardiovascular outcomes of interest (e.g., MI, heart attack, heart failure, stroke, brain infarct). To limit publication bias, there were no language, publication date, or article type restrictions on the search strategy. However, due to the limitations of language competency of the study team, we subsequently excluded non-English publications.

Study Selection

Retrieved studies were screened (AKG, BPG, MRD, SNS, TY) for inclusion using Covidence systematic review software. Titles and abstracts were reviewed against predefined inclusion/exclusion criteria by two independent reviewers. Discrepancies were resolved by consensus. For final inclusion, full texts were retrieved, and then they were screened by two independent reviewers.

Inclusion and Exclusion Criteria

Our inclusion criteria were: a) Population: adult participants $\ge 18\mathrm{y}$ old; b) Exposure: extreme weather events involving wet precipitation, including heavy rainfall and flooding (i.e., not snow), such as hurricanes, cyclones, tropical storms, typhoons, or storm surge; and c) Outcomes studied: reported CVD-related outcomes, such as myocardial ischemia and/or MI, heart failure, stroke, and death relate to CVD, collectively defined as major adverse cardiovascular or cerebrovascular events (MACE).

Studies were excluded according to the following criteria: a) They were non-English; b) They were review articles, commentaries, editorials; c) Articles were case reports (i.e., $n=1$); and d) Articles did not provide enough data to determine whether CVD outcomes were related to MI or ischemia, heart failure, or stroke. The fourth exclusion criterion was included to ensure consistency in the CVD outcomes under study and because these CVD-related diagnoses are prevalent.

Data Extraction

Data extraction was performed independently in duplicate with predefined, standardized templates. Each extraction was reviewed independently by a secondary reviewer after extraction by the primary reviewer. Data points defined for extraction were study location, study setting, number of participants, exposure type, name and year, study objective, study design, data sources, CVD outcome, and key findings.

Data Synthesis

Following data extraction, CVD outcomes were synthesized qualitatively, assessed for the quality of evidence based on study design and number of participants and then categorized by the time frame of associated CVD outcomes, where short-term was defined as $<1\mathrm{y}$ and long-term as $\ge 1\mathrm{y}$. To reflect the importance of stress and broader social determinants of health as possible mechanisms moderating or mediating CVD outcomes with hurricane exposure, articles were grouped (AKG, SNS) by vulnerable populations, such as racial/ethnic groups; those with existing CVD-related comorbidities, including end-stage renal disease; diabetic patients; and patients with psychiatric comorbidities. These groups were not mutually exclusive and reflect our choices in highlighting important findings from our scoping review. Gaps in the literature were then identified based on this qualitative synthesis. No quantitative assessment of the literature (i.e., meta-analysis) was performed because of the heterogeneity in CVD outcomes.

Results

Summary of Articles

Figure 1 summarizes the full PRISMA flow diagram outlining the study selection process. Of the 1,103 citations identified after deduplicating, 48 met our inclusion criteria (Table 1).

Table 1.

Table of studies.

First author, year	City, country	Study setting	Study design	Study number	Exposure type and name, year	CVD outcome	Key findings
Becquart 201826	New Orleans, Jefferson, United States of American (USA)	Countywide analysis	Retrospective cohort study	$n=\mathrm{383,552}$	Hurricane Katrina, 2005	CVD hospitalization (including CHD, stroke)	Of individuals with CHD, hospitalizations for CVD were two times higher for Black patients than for White patients (9.8% vs. 4.7%, respectively).
Chang 202154	Taiwan	Nationwide, longitudinal using National Registry	Retrospective cohort study	$n=\mathrm{715,244}$	Typhoon Morakot, 2009	AMI and stroke-related incidence	For patients on dialysis, higher incidence of AMI events and stroke events in areas severely affected by hurricane than in those moderately affected.
Cifelli 201549	New York, USA	Emergency department	Retrospective cohort study	$n=\mathrm{23,776}$ (before hurricane); $n=\mathrm{24,815}$ (after hurricane)	Hurricane Sandy, 2012	AMI-related hospitalization	There was a statistically significant increase of 39% in daily AMI visits after the storm, and non-statistically significant increase of 14% in daily tachyarrhythmia visits after the storm.
Cruz-Cano 201965	Puerto Rico, USA	Statewide analysis	Retrospective time-series analysis	$n=\mathrm{3,493,593}$ (September); $n=\mathrm{3,489,119}$ (October)	Hurricane Maria, 2017	CVD-related mortality	Estimated 253 excess deaths (1 in 5) from heart disease associated with Hurricane Maria.
Deere 201827	New Orleans, USA	Single center hospital	Retrospective cohort study	$n=150$ (before hurricane); $n=\mathrm{2,724}$ (after hurricane)	Hurricane Katrina, 2005	MI-related hospitalization	3.0% of patients presented with AMI 11 years after vs. 0.7% before the Hurricane, a 4-fold increase; prevalent psychosocial, behavioral, and traditional CHD risk factors were significantly higher among the post-Katrina group.
Edmondson 201328	New Orleans, USA	Community-based	Prospective cohort study	$n=391$	Hurricane Katrina, 2005	CVD-related hospitalization and CVD-related mortality	Positive depression screening was significantly associated with increased risk of CVD-related hospitalization and mortality HR 1.33) but not PTSD
Gautam 20097	New Orleans, USA	Single center hospital	Retrospective cohort study	$n=150$ (before hurricane); $n=246$ (after hurricane)	Hurricane Katrina, 2005	AMI-related hospitalization	2.18% of the 2-y post-Katrina cohort were admitted for AMI vs. 0.7% before the hurricane, an almost 3-fold increase.
Gonzales 201531	New Orleans, USA	Single center hospital	Retrospective cohort study	$n=299$ (before hurricane); $n=\mathrm{1,479}$ (after hurricane)	Hurricane Katrina, 2005	AMI-related hospitalization	Compared to pre-Katrina data, 8 y post-Katrina AMI incidence decreased on Mondays, weekdays, and mornings. Post-Katrina AMI incidence increased on nights and weekends.
Gonzales 201629	New Orleans, USA	Single center hospital	Retrospective cohort study	$n=150$ (before hurricane); $n=\mathrm{1,982}$ (after hurricane)	Hurricane Katrina, 2005	AMI-related hospitalization	2.5% of the combined 9-year post-Katrina cohort were admitted for AMI vs. 0.7% before the hurricane - a 4-fold increase.
Gonzales 201630	New Orleans, USA	Single center hospital	Retrospective cohort study	$n=299$ (before hurricane); $n=\mathrm{1,606}$ (after hurricane)	Hurricane Katrina, 2005	AMI-related hospitalization	Compared to pre-Katrina data, 9 y post-Katrina incidence of AMI decreased on Mondays, weekdays, and mornings. Further, the post-Katrina cohort had increased AMI incidence at night and on weekends.
Gonzalez 202066	Puerto Rico, USA	Single center hospital	Retrospective cohort study	$n=235$ admitted 2017; $n=373$ admitted 2016	Hurricane Maria, 2017	CHD-related hospitalization (all ACSs)	AMI incidence of 9.79% (23 of 235) post hurricane, in comparison prehurricane MI incidence of 2.95% (11 of 373) [$p=0.004$]; Posthurricane CHD-related admissions was 27.17% (2017) vs. 33.04% prehurricane.
Hameed 201232	New Orleans, USA	Single center hospital	Retrospective cohort study	$n=\mathrm{21,079}$ (before hurricane); $n=\mathrm{28,597}$ (after hurricane)	Hurricane Katrina, 2005	AMI-related hospitalization	2.2% of patients presented with AMI after vs. 0.7% before the Hurricane - a 3-fold increase; post-Katrina group had a higher prevalence of unemployment, lack of medical insurance, smokers, medical noncompliance, substance abuse, psychiatric comorbidities, history of coronary artery disease, and percutaneous coronary interventions.
Harrison 202133	New Orleans, USA	Single center hospital	Retrospective cohort study	$n=\mathrm{3,619}$	Hurricane Katrina, 2005	AMI-related hospitalization	3.0% of patients presented with AMI after vs. 0.7% before the Hurricane - 3-fold increase; the posthurricane group had higher prescription rates but also larger nonadherence and was more likely to be unemployed and unmarried.
Hendrickson 199757	Hawaii, USA	Outpatient and emergency department chart review	Retrospective cohort study	$n=\mathrm{1,584}$	Hurricane Iniki, 1992	CHD, hypertension, and stroke presentation	Physician visits for CVD complaints were significantly increased in the post-Iniki period (RR 2.73, 95% CI: 1.51, 4.94); individuals aged 65 years and over experienced significant increases in CVD complaints (RR 2.67).
Howe 200834	New Orleans, USA	Outpatient clinic chart review	Retrospective cohort study	$n=465$	Hurricane Katrina, 2005	Presentation with hypertension, hypertensive urgency/emergency, MI, HF, arrhythmia, stable/unstable angina	Among 410 illness complaints, 11% were CVD-related diagnoses.
Hua 202158	Florida, USA	Emergency department	Retrospective time-series analysis	$n=\mathrm{30,358}$	Hurricane Irma, 2017	CHD, stroke, or HF emergency presentation	No statistically significant differences pre- vs. posthurricane in CVD-related emergency department outcomes: HF (25.5.% vs. 25.4%); CHD (47.5% vs. 47.0%); stroke (7.0% vs. 7.1%).
Huang 201752	Taiwan	Major medical centers and regional hospitals in Taiwan	Retrospective cohort study	Not reported	Typhoon Morakot, 2009	CHD-related incidence	During the week of Typhoon Morakot, there was a significantly higher incidence (9.14 cases per day) of ACS than the study average (6.48 cases per day).
Jiao 201235	New Orleans, USA	Single center hospital	Retrospective cohort study	$n=\mathrm{21,093}$ (before hurricane); $n=150$ (after hurricane)	Hurricane Katrina, 2005	AMI-related hospitalization	2.0% of the combined 3-year post-Katrina cohort were admitted for AMI vs. 0.7% before the hurricane - an almost 3-fold increase.
Kanaoka 202064	Japan	Nationwide analysis using registry	Retrospective cohort study	$n=\mathrm{10,782}$	Six typhoon landfalls between April 2013–March 2015	HF diagnosis, specifically TTS	Admissions for TTS were significantly higher on the day and immediately after the typhoon landfalls with IRR 2.84 (95% CI: 1.53, 5.30) on the day of the landfall.
Kim 201760	New Jersey, USA	Statewide analysis	Retrospective time-series analysis	Not reported	Hurricane Sandy, 2012	CVD-related mortality	CVD mortality during the Sandy quarter showed significant 6% increased risk in mortality compared to same time period in previous 3 years, but during month of Sandy, CVD-related mortality for elderly significantly higher by 10% and persisted over the quarter.
Kim 202014	Florida, USA	Hospital	Retrospective time-series analysis	$n=\mathrm{3,372,993}$	Tropical storms before 2007	HF-related hospitalization	No significant correlation was found between frequency of storms in each quarter with HF discharges.
Kostis 201450	New Jersey, USA	County-wide analysis	Retrospective cohort study	Not reported	Hurricane Sandy, 2012	AMI-related hospitalization and CVD-related death	23% increase in the number of AMI hospitalizations; the total number of AMI-related deaths (in-hospital plus out-of-hospital) increased by 28% all compared to same period in 5 prior years).
Lawrence 20195	New York, USA	Outpatient, hospitalization, and emergency department visits	Retrospective cohort study	$n=\mathrm{217,873}$	Hurricane Sandy, 2012	CVD-related hospitalizations	In the 4-month period following the storm, CVD risk (RR: 2.13) was significantly higher than corresponding period in other years [2 control groups of 5 y prior (2007–2011) and 1 y after (2013–2014)]. Comparing affected areas versus nonaffected areas, CVD had the greatest risk immediately (RR: 2.65, 4 months (RR: 2.62), and 12 months (RR: 2.64) after the hurricane period (all significantly different).
Lee 201651	New York, NY, USA	Emergency Department	Retrospective cohort study	Not reported	Hurricane Sandy, 2012	AMI-related ED presentation	Statistically significant increase MI presentations as a primary diagnosis in the week after Sandy’s landfall compared to the weeks preceding landfall.
Lenane 201936	New Orleans, USA	Community-based	Prospective cohort study	$n=\mathrm{2,194}$	Hurricane Katrina, 2005	CVD-related hospitalization and CVD-related mortality (i.e., stroke, AMI, HF)	The proportion of the sample experiencing a CVD event during follow-up was 18.0% and 11.0% among those with and without PTSD symptoms, respectively.
Matthew 201337	New Orleans, USA	Hospital	Retrospective cohort study	$n=707$	Hurricane Katrina, 2005	MI-related hospitalization	Post-Katrina AMI presentations increased on nights and weekends and decreased on Mondays, weekdays, and mornings in comparison with pre-Katrina AMI presentations.
McKinney 201167	Florida, USA	County-wide analysis	Retrospective cohort study	$n=624$	Hurricanes Charley, Frances, Jeanne, 2004	All-cause mortality and also CVD mortality as subgroup	CVD-related deaths significantly higher up to 1 month after Hurricane Ivan, 2 months after Hurricane Frances and Charley’s landfall in comparison with average mortality in the 3 y preceding.
Moscona 201239	New Orleans, LA, USA	Single center hospital	Retrospective cohort study	$n=999$	Hurricane Katrina, 2005	AMI-related hospitalization	2.2% of the 5-y post-Katrina cohort were admitted for AMI vs. 0.7% before the hurricane, an almost 3-fold increase.
Moscona 201338	New Orleans, USA	Single center hospital	Retrospective cohort study	$n=\mathrm{1,177}$	Hurricane Katrina, 2005	AMI-related hospitalization	2.8% of patients presented with AMI 6 y after vs. 0.7% before the Hurricane, a 4-fold increase.
Moscona 201940	New Orleans, USA	Single center hospital	Retrospective cohort study	$n=\mathrm{2,491}$	Hurricane Katrina, 2005	AMI-related hospitalization	2.8% of the combined 3-y post-Katrina cohort were admitted for MI vs. 0.7% before the hurricane, a 4-fold increase.
Nakhle 202041	New Orleans, USA	Single center hospital	Retrospective cohort study	$n=\mathrm{3,067}$	Hurricane Katrina, 2005	AMI-related hospitalization	3.4% of patients presented with AMI after vs. 0.7% before the Hurricane - an almost 5-fold increase; Patients suffered higher rates of cardiovascular risk factors and psychiatric risk factors of CHD.
Nakhle 202042	New Orleans, USA	Single center hospital	Retrospective cohort study	$n=\mathrm{3,278}$ (before hurricane); $n=150$ (after hurricane)	Hurricane Katrina, 2005	AMI-related hospitalization	3.3% of the 13-y post-Katrina cohort were admitted for AMI vs. 0.7% before the hurricane, an almost 5-fold increase.
Parks 202163	USA	Nationwide analysis, multiple hospitals	Retrospective cohort quasi-experimental study	$n=\mathrm{69,682,674}$	Multiple tropical cyclones from 1999 to 2014	Hospitalization (general); all CVD-related hospitalizations	CVD was leading cause of hospitalizations (30%); CVD-related hospitalizations (from MI, pulmonary heart disease, general vascular disease in particularly) decreased on day of tropical cyclone, increased then peaked from day 1 to 3, and then dropped to baseline rate of cause-specific hospitalization (defined as average hospitalization rate between 1999–2014) by day 7.
Peters 201443	New Orleans, USA	Single center hospital	Retrospective cohort study	$n=\mathrm{1,528}$	Hurricane Katrina, 2005	AMI-related hospitalization	2.24% of patients presented with AMI after vs. 0.7% before the Hurricane, a 3-fold increase.
Peters 201444	New Orleans, USA	Single center hospital	Retrospective cohort study	$n=\mathrm{1,595}$ (before hurricane); $n=\mathrm{1,296}$ (after hurricane)	Hurricane Katrina, 2005	AMI-related hospitalization	Post-Katrina AMI presentations increased on nights and weekends, and decreased on Mondays, weekdays, and mornings compared to pre-Katrina AMI presentations.
Peters 201446	New Orleans, USA	Single center hospital	Retrospective cohort study	$n=\mathrm{1,476}$	Hurricane Katrina, 2005	AMI-related hospitalization	2.4% of the 6-y post-Katrina cohort were admitted for AMI vs. 0.7% before the hurricane, a 3-fold increase.
Peters, 201345	New Orleans, USA	Single center hospital	Retrospective cohort study	$n=707$	Hurricane Katrina, 2005	AMI-related hospitalization	Post-Katrina AMI presentations increased on nights and weekends, and decreased on Mondays, weekdays, and mornings compared to pre-Katrina AMI presentations.
Quast 201962	4 states (Louisiana, Mississippi, Texas, or Alabama)	County-wide analysis	Retrospective cohort study with propensity score matching	$n=\mathrm{170,328}$	Hurricane Katrina and Hurricane Rita, 2005	All-cause and CVD-related mortality	Affected cohort had 34% higher risk of CVD-related mortality at 1 month, 17.4% at 6 months, and 14.6% at 1 year in comparison with unaffected matched controls.
Sen 201855	Houston, USA	Single center hospital	Retrospective cohort study	$n=88$	Hurricane Harvey, 2017	CVD-related ICU admission	There were 40 flood-related admissions, of which 6 were CHD- or HF-related (15%).
Sharma 200847	New Orleans, USA	Hospital and community-based	Retrospective cohort study	Not reported	Hurricane Katrina, 2005	Presenting with cardiovascular disease (CVD)	CVD diagnoses presented the most common condition (32.8%); the proportion of men presenting with CVD diagnoses increased with age from 14.0% among those ages 0–19 y to 54.9% among those ages 80 y or older. Among all women, the most common diagnosis was CVD (29.2%).
Shih 202053	Taiwan	Nationwide analysis using registry	Case–control study with propensity score matching	$n=\mathrm{715,244}$	Typhoon Morakot, 2009	CVD-related (i.e., stroke and CHD) hospitalizations	Compared to matched individuals in moderately affected areas, individuals in severely affected area had 9.6% higher odds of MI, 12.9% higher risk of stroke, and 15.2% higher risk of HF.
Singh 201548	New Orleans, USA	Single center hospital	Retrospective cohort study	$n=\mathrm{1,851}$	Hurricane Katrina, 2005	AMI-related hospitalization	2.57% of the 8-y post-Katrina cohort were admitted for AMI vs. 0.7% before the hurricane, an almost 4-fold increase.
Sunohara 202169	Nagano, Japan	Multiple hospitals	Retrospective cohort study	$n=\mathrm{2,426}$	Reiwa First Year East Japan Typhoon, 2019	CVD-related hospitalization (including ACS, stroke)	CVD hospitalizations increased 2 weeks after flooding, compared to same periods 2 y prior; unstable angina cases increased up to 2 months after flooding, and stroke increased 2 wk after flooding; also reported an increased incidence of missed medications and dyslipidemia.
Swerdel 20143	New Jersey, USA	Statewide analysis	Retrospective cohort study	Not reported	Hurricane Sandy, 2012	AMI and stroke-related hospitalization, and CVD-related mortality	In high-impact areas, AMI incidence increased by 22% in comparison with previous years, and 30-d AMI-related mortality increased by 31%. Stroke incidence increased by 7% and no significant change was observed in 30-d stroke-related mortality.
Tarnoki 201761	Florida, USA	Multiple hospitals	Retrospective cohort study	$n=\mathrm{1,410}$	Hurricanes Karl (2010), Irene (2011), Cina (2011), Isaac (2012), Sandy (2012), Barbara (2013) and Andrea (2013)	Ischemic stroke- and subarachnoid hemorrhages-related hospitalizations	Non-significant increased incidence rate of ischemic stroke was consistent with the daily lowest and highest air pressure, highest air temperature compared to periods of normal air pressure and temperature, and also not significantly associated with presence of hurricanes or storms in comparison with time periods where there were no hurricanes or storms.
Trebouet 201259	Vendée district, France	Single center hospital	Retrospective cohort study	$n=6$	Cyclone Xynthia, 2010	TTS	In 50% of patients, TTS was diagnosed.
Yan 202168	USA	Nationwide analysis, multiple hospitals	Retrospective cohort study	Not reported	Multiple tropical cyclones between 1990–2010	CVD hospitalizations, defined as HF, stroke, arrythmia, CHD	There was a significant 3% higher risk of CVD-related hospitalization during the storm period, associated with heart failure (8% increased risk) and AMI (5% increased risk) and CHD (3% increased risk) in comparison with counties not exposed to storms.
Zane 201156	Texas, USA	Statewide analysis	Retrospective cohort study	$n=74$	Hurricane Ike, 2008	MACE defined as cardiovascular failure	MACE ($n=\mathrm{12,\; 16}\%$) was the leading cause of illness-related deaths. All of these were classified as indirectly ($n=8$) or possibly ($n=4$) caused by Hurricane Ike.

Note: ACS, acute coronary syndrome; AMI, acute myocardial infarction; CHD, coronary heart disease; CI, confidence interval; CVD, cardiovascular disease; HF, heart failure; HR, hazard ratio; IRR, incidence rate ratio; MACE, major adverse cardiovascular event; MI, myocardial infarction; PTSD, posttraumatic stress disorder; RR, risk ratio; TTS, Takotsubo syndrome.

Only 6 out of 48 (12.5%) of the articles described the relationship between CVD and hurricanes outside the United States and were based on data from Taiwan, Japan, and France. Of those articles from the United States, the majority (25/42; 54.3%) described exposures in New Orleans after 2005 Hurricane Katrina.^7,26–48 Of particular note, 19 of these 25 articles were authored by the same research group that replicated the same pre- and posthurricane analysis to compare a) MI-related admission rates and b) MI admission chronobiology from the period before Hurricane Katrina to the period after Hurricane Katrina over ever-increasing periods of time (i.e., 2, 3 y after Hurricane Katrina, up to 14 y after Hurricane Katrina).

Other hurricanes and related weather events that were described were 2012 Hurricane Sandy (six studies),^{3,14,5,49–51} 2009 Typhoon Morakot that struck Taiwan (three studies),^52–54 2017 Hurricane Harvey (one study),⁵⁵ 2008 Hurricane Ike (one study),⁵⁶ 1992 Hurricane Iniki (one study),⁵⁷ 2017 Hurricane Irma (one study),⁵⁸ and 2010 Cyclone Xynthia (one study).⁵⁹ Six of the 48 studies described population-level effects of multiple hurricanes/cyclones/typhoons over an extended period of time.

Thirty-two of the 48 articles (66.7%) focused on the individual patient-level CVD outcomes in the hospitalized setting by analyzing hospital data from single centers. CVD- and MI-related hospitalizations were the most common CVD outcome studied, representing 30 of the 48 studies (62.5%). The second most common outcome (7 studies; 14.5%) studied was CVD-related mortality.

Three studies reported no association between hurricanes and CVD outcomes⁶⁰; the first focused on heart failure exacerbations in Florida after tropical storms, the second focused on strokes in southern Florida,⁶¹ and the third focused on emergency room presentations for residents in assisted living facilities.⁵⁸

Sources and Directions of Bias

The majority of the studies (46/48; 95%) employed a retrospective study design using secondary data, reflecting the observational nature of hurricane-related studies in general. Of these, one used a case–control design, 39 employed a retrospective cohort study design, and 4 employed a retrospective time-series analysis approach. As a result of the retrospective nature of these studies, unobserved confounders including posthurricane migration and displacement between hurricane exposure and CVD outcomes are likely a source of bias, particularly in the Hurricane Katrina studies that examined this relationship at a single institution authored by the same research group.

Several studies may have limited these sources of bias, owing to their large study population sizes over broad geographies (five used nationwide data, four used statewide data, four used countywide data) and through the use of causal methods, including propensity score matching^53,62 and quasi-experimental design⁶³ (Table 1). The largest of these studies⁶³ used Medicare fee-for-service beneficiary-level data between 1999 and 2014, yielding $>69\text{\hspace{0.17em}million}$ individual observations in a quasi-experimental design (pre- and posthurricane observation over a short time frame), thus improving the quality of the study findings through large numbers and limiting unobserved exogenous confounding as a natural experiment.

Hereafter, we provide a synthesis of findings from the scoping review that describe the CVD outcomes by short- and long-term outcomes, followed by studies that focused on special populations. A summary of final list of studies is described in the Table 1.

Short-term ($\mathbf{<}1\mathrm{y}$) CVD-related outcomes related to hurricanes.

Three studies^26,63,64 examined CVD-related outcomes exclusively in the immediate aftermath (within 1 wk) of hurricane/cyclone exposure. Based on findings from single institution study after Hurricane Katrina²⁶ and a larger study⁶³ that examined acute posthurricane hospitalizations from multiple hurricane exposures during the period 1999–2014, CVD hospitalizations peaked between days 2 and 6, peaking at an approximate 3%–4% increase in comparison with prehurricane levels.

Six studies^{14,5,49–51,3} met criteria for inclusion that focused on the short-term CVD outcomes related to Hurricane Sandy. Three of these studies used population-level measures to examine the effect of the hurricane on CVD events and CVD mortality in the weeks to months after landfall. Using New Jersey vital statistics, the first study found a 6% increase [95% confidence interval (CI): 2%, 11%) in CVD-related mortality overall the month after landfall compared to before,¹⁴ whereas two others, demonstrated a 22%–23% increased risk (95% CI: 1.16, 1.28) of MI-related hospitalization^50,3 [including a 31% increased risk (95% CI: 1.22, 1.41) of 30-d mortality], and a 7% increased risk (95% CI: 1.03, 1.11) of stroke-related hospitalization compared to the same periods in the years preceding.³

Two studies^65,66 examined short-term CVD outcomes in aftermath of Hurricane Maria in 2017 in Puerto Rico. The first, which examined CVD-related mortality in the month following landfall,⁶⁵ calculated 253 excess deaths from CVD using a time-series analytical approach, which represented overall 1 in 5 excess deaths. Separately, results from a single site observation study⁶⁶ suggested a significant increase in MI hospitalizations after landfall in comparison with MI hospitalizations before landfall [post: 23/235 (9.75%) vs. pre: 11/373 (2.95%)].

Four studies focused on the exposures of multiple hurricanes either across a hurricane season,⁶⁷ or multiple hurricanes over many years.^61,63,68 Similar to previous studies examining CVD-related mortality, authors⁶⁷ performed a countywide analysis of the 2004 hurricane season in Florida and showed increased mortality 1–2 months after landfall in comparison with weeks before landfall. In another,⁶³ CVD was the leading cause of posthurricane hospitalization (30% of total hospitalizations from the period were CVD-related) across multiple hurricane seasons.

Two studies described higher rates of short-term ($>1\text{\hspace{0.17em}month}$) CVD outcomes after hurricanes in comparison with outcomes before hurricanes. CVD reports increased in primary care and emergency departments at various locations in Hawaii after Hurricane Iniki in 1992 in comparison with prehurricane primary care visits [relative risk (RR) 2.73; 95% CI: 1.51, 4.94] prior to landfall.⁵⁷ After the Reiwa East Japan Typhoon,⁶⁹ MI-related hospitalizations significantly increased up to 2 months after landfall, and stroke hospitalizations increased 2 wk after landfall in comparison with stroke hospitalizations in prior years. After Hurricane Harvey, intensive care admissions related to worsened heart failure and CHD represented 15% of intensive care unit (ICU) cases related to hurricane flooding, although it was unclear whether this represented a deviation from prehurricane norms.⁵⁵

Two studies^59,64 reported the increased incidence of Takotsubo’s cardiomyopathy (a rare form of heart failure often related to severe acute stresses) in the aftermath of hurricanes in comparison with incidences in prehurricane timeframes. The larger of these two studies⁶⁴ showed, using a Japanese CVD registry, that Takotsubo’s cardiomyopathy admissions were significantly higher on the day of and immediately after typhoon landfalls [incidence rate ratio (IRR) 2.84; 95% CI: 1.53, 5.30] but not subsequent days.

Long-term ($\ge 1\mathrm{y}$) CVD-related outcomes related to hurricanes.

Of the 25 Hurricane Katrina studies, 19^{27,7,29–33,37–46,48,66,36} describe statistically significant 3- to 5-fold increases in MI-related hospitalization rates at a single institutional site from 2 y before Hurricane Katrina’s landfall to up to 14 y after landfall.

In patient populations under investigation, these same 19 studies also reported significant differences in rates of baseline CHD and other CVD risk factors, such as smoking, hypertension, diabetes, substance abuse, and psychiatric disease, and lower socioeconomic factors (including higher unemployment rates), with higher rates found in the post-Katrina patient population. Furthermore, a subset of these studies^{30,31,37,40,44,45} further report a consistent increase in MI-related hospital admissions over nights and weekends but decreased MI-related hospitalizations on weekdays and mornings both relative to prehurricane times.

Vulnerable populations.

Racial disparities in CVD-related outcomes.

In a post-Katrina study²⁶ examining racial differences in CVD-hospitalization based on Medicare claims, the authors reported that within the first month, Black patients were almost twice as likely White patients to be admitted within the first month after landfall (Black patients: $26.3\pm 23.7$ cases per day vs. White patients: $16.6\pm 11.7$ cases per day).

Patients with end-stage renal disease.

In a study⁶⁴ that used Taiwan’s National Health Insurance Registry to understand the relationship between 2009 Typhoon Morakot and associated CVD outcomes for end-stage renal disease (ESRD) patients, authors found that ESRD patients had lower rates of MI- and stroke-related hospitalizations after landfall compared to such hospitalizations before landfall (111 vs. 127 cases, respectively, per 1,000 population).

Patients with psychiatric comorbidities.

Two studies, both of prospective cohort design created for other purposes, have highlighted the link between depression and posttraumatic stress disorder (PTSD) and their association with CVD outcomes in hurricane-affected communities. In the first study,²⁸ which focused on patients with ESRD, positive depression screening (but not positive PTSD screening) was significantly associated with an increased risk of CVD-related hospitalization and mortality [hazard ratio (HR) 1.33; 95% CI: 1.06, 1.76]. In the second study³⁶ that evaluated CVD event incidence (for stroke, MI, and heart failure) in community-based older patients post-Katrina with hypertension, CVD event incidence was overall high (11.6%) and higher still among those patients with evidence of PTSD (18%).

Diabetic patients.

Two studies specifically focused on diabetic patients. The first⁷⁰ used Medicare claims to examine the association between CVD-related mortality in diabetic patients and the exposure of Hurricanes Katrina and Maria across multiple affected counties, finding that those affected had a 35% increased odds [odds ratio (OR) 1.35; 95% CI: 1.21, 1.51] of CVD-related mortality at 1 month, 17.4% (OR 1.17; 95% CI: 1.12, 1.23) at 6 months, and 14.6% (OR 1.15; 95% CI: 1.11, 1.19) at 1 y in comparison with unaffected matched beneficiaries.

Examining the effects of Hurricane Sandy in New York City, one study⁵¹ focused on diabetic patients reported a significant increase in MI-related hospitalizations in the week after landfall, whereas the other examined older patients, using Medicare and Medicaid claims data to highlight that CVD-related hospitalizations continued to be higher than prehurricane time periods even 1 y after landfall ($\text{RR}=2.64$; 95% CI: 2.64, 2.65).

Discussion

In this scoping review, we identified 48 English-language, peer-reviewed, original research articles that described the association between CVD and extreme weather events, including hurricanes. Of these, the majority focused on events in the United States and in particular the effects of Hurricane Katrina in 2005. This review provides an overview of publications that focus on a range of CVD outcomes and their association with hurricanes and other extreme weather events. Particularly, our scoping review summarizes the current literature on the short-term associations between hurricanes and CVD-related hospitalization and CVD-related mortality—evidence that has been described in the aftermath of Hurricanes Katrina,^{26–28,7,29–48,62} Sandy,^{14,3,5,49–51} Maria,⁶⁵ Irma,⁵⁸ Iniki,⁵⁷ and Harvey⁵⁵ and Typhoon Morakot,^57,58 as well as broadly across multiple hurricanes.^{52,60,61,63,64,67,68}

Furthermore, the scoping review sheds light on certain vulnerable populations at increased risk of adverse CVD outcomes. Several studies described the impact on older patients (including those in the Medicare^58,63 and dually eligible Medicare–Medicaid population),⁵ patients with underlying CVD risk factors such as diabetes^51,70 and ESRD,⁵⁴ and people suffering from psychiatric conditions such as depression and PTSD.^28,36

Most important, this scoping review also highlights several important gaps in knowledge. As hurricanes increase in frequency and destructiveness, developing a research agenda to fill these gaps will become increasingly important in the future to mitigate the adverse CVD-related effects of hurricanes. This significant need to increase knowledge is particularly the case as the importance of environmental justice becomes more prominent in the era of climate change,⁷¹ with the increasing acknowledgment of the role that structural factors play in worsening health outcomes for vulnerable populations affected by hurricanes.

Rigorous Long-Term Evaluation of Hurricane Exposure on CVD Outcomes, Including outside the United States

It remains unclear but important to understand whether the CVD-related associations of hurricanes persist over longer durations. In the United States⁷² and broadly,⁷³ disaster-related frameworks focused on health outcomes often concentrate resources in the immediate postdisaster period (often the first few months). If evidence suggests that the effects of hurricane exposure linger and potentially accelerate CVD progression, then such frameworks should be reconceptualized over a longer time frame.

Our scoping review did find a number of studies that examined longitudinal outcomes for hurricane victims. However, in the studies that did so, findings were limited by the potential confounding inherent in a single institution focused pre- and posthurricane design, the lack of adequate controls as a comparison group to define differences in outcomes. Furthermore, these findings lack generalizability outside the United States, given the high prevalence of studies performed in the United States. Therefore, to fill this important gap in the literature, more rigorous, well-defined cohorts to follow over time are required to quantify the long-term CVD risk resulting from hurricane exposures, particularly in non-U.S. settings.

Vulnerable Populations with Focus on Environmental Justice

Our scoping review reported one study that examined racial differences in MI-related hospitalization after Hurricane Katrina.²⁶ Although the relationship between hurricanes such as Hurricane Katrina and vulnerable populations has been clearly described in the general literature,^10,74 our scoping review highlighted the dearth of peer-reviewed literature focused on the differential role hurricanes play in affecting CVD outcomes by race/ethnicity, gender, socioeconomic status, older age, and children’s and adolescents’ health and development, as well as other important structural factors such as home ownership and social connectedness. This finding reflected, in our views, the unclear nature of how vulnerable populations were defined (other than Black vs. White individuals, as described in Becquart et al.).⁶¹ Given the prevalence of CVD and the clear socioeconomic gradient linking CVD outcomes to lower socioeconomic status,⁷⁵ we believe this to be a surprising finding stemming from the scoping review. Further research is required to cast an important light on the health-related disparities created by climate change-driven weather phenomenon.

Association between Multiple and Repeated Hurricanes Strikes and CVD Outcomes

Hurricanes and related weather phenomena have historically occurred in specific geographies during certain times of the year. Therefore, at-risk populations will continue to be exposed to hurricanes. In our scoping review, the majority of studies were limited to the exposure to one hurricane, not repeated or cumulative exposures over time. Nonetheless, the cumulative impact of multiple hurricanes or other extreme weather events over time is likely to have deleterious effects on CVD outcomes for affected individuals. The need for peer-reviewed evidence to describe and quantify the repeated and cumulative impacts of hurricanes on these CVD outcomes will be even more important as hurricanes become more frequent and more destructive.

Exploration of the Mechanisms for Increased CVD Outcomes

Two studies found in the scoping review describe the role depression and PTSD play on CVD outcomes in relation to hurricanes.^28,36 Although stress has been commonly attributed to the causal pathway from hurricane impacts to CVD outcomes, this has not been rigorously studied, based on our findings. One challenge in the exploration of mechanisms of CVD outcomes related to hurricanes is the lack of prospective epidemiological cohorts to follow over time. This challenge is likely the reason why most studies used a retrospective study design involving secondary data sources. Although this is a limitation in future work focused on understanding the mechanistic pathway linking CVD and hurricanes, novel, validated causal inference methods that allow identification are now available, making this work possible.⁷⁶ Future research should seek to use these tools and large data sources to develop an understanding of potential mechanisms.

A major limitation of this study is that it excluded non-English texts. Hurricanes and related events happen all across the world. It is highly likely that published non-English peer-reviewed literature exists that focuses on CVD outcomes and that we could not access this research because of the language barrier. In an attempt to offset this risk, our initial search did not exclude non-English language texts. However, subsequently these were excluded from further review because we lacked the resources to confidently review and appraise non-English texts.

In conclusion, our scoping review demonstrated a clear link between hurricanes and related extreme weather events exposure and CVD outcomes in the prevailing peer-reviewed, published, original research studies in English. Most studies linked hurricanes to CVD outcomes in the short-term, and also highlighted special populations, including those with other medical comorbidities. An important result is that our review revealed a number of gaps in the knowledge, which includes a lack of rigorous long-term evaluation of hurricane impacts, their effects on vulnerable populations including structural drivers (e.g., access to care, systematic inequalities in health care delivery, neighborhood-level social determinants of health) that place such individuals at risk predisaster and exacerbate adverse health outcomes postdisaster, the role that multiple hurricanes play on population health, and exploration of the mechanisms that lead to worsened CVD outcomes. Future research is currently underway to generalize these findings across multiple hurricanes and extreme weather events, and such research should attempt to fill these gaps. This enhancement will provide an important evidence base for future disaster-related policy that seeks to address environmental injustices as well as the overall adverse effects of these extreme weather events.