Population-Based Epidemiology of Heart Failure in a Low-Income Country: the Haiti Cardiovascular Disease Cohort

Abstract

Background:

Cardiovascular disease disproportionately affects persons living in low- and middle-income countries (LMICs) and heart failure (HF) is thought to be a leading cause. Population-based studies characterizing the epidemiology of HF in these settings are lacking. We describe the age-standardized prevalence, survival, subtypes, risk factors and one-year mortality of HF in the population-based Haiti Cardiovascular Disease Cohort.

Methods:

Participants were recruited using multistage cluster-area random sampling in Port-au-Prince, Haiti. A total of 2,981 completed standardized history and exam, laboratory measures, and cardiac imaging. Clinical HF was defined by Framingham criteria. Kaplan-Meier and Cox proportional hazard regression assessed mortality among participants with and without HF; logistic regression identified associated factors.

Results:

Among all participants, the median age was 40 years (IQR 27–55) and 58.2% were female. Median follow-up was 15.4 months (IQR 9–22). The age-standardized HF prevalence was 3.2% (93/2,981 [95% CI:2.6–3.9]). The average age of participants with HF was 57 years (IQR 45–65) and 67.7% were female. The first significant increase in HF prevalence occurred between 30–39 and 40–49-years (1.1% vs 3.7%, p=0.003). HFpEF was the most common HF subtype (71.0%). Age (aOR: 1.36 [1.12–1.66] per 10-year increase), hypertension (2.14 [1.26–3.66]), obesity (3.35 [95% CI: 1.99–5.62]), poverty (2.10 [1.18–3.72]) and renal dysfunction (5.42 [2.94–9.98]) were associated with HF. One-year HF mortality was 6.6% versus 0.8% (HR: 7.7 [95% CI 2.9–20.6], p<0.0001).

Conclusions:

The age-standardized prevalence of HF in this low-income setting was alarmingly high at 3.2%--5-fold higher than modeling estimates for LMICs. Adults with HF were two decades younger and 7.7-times more likely to die at one year compared to those in the community without HF. Further research characterizing the population burden of HF in LMICs can guide resource allocation and development of pragmatic HF prevention and treatment interventions, ultimately reducing global CVD health disparities.

Registration:

This study is registered at https://clinicaltrials.gov/ct2/show/NCT03892265

INTRODUCTION

The global burden of heart failure (HF) is substantial, affecting over 64 million people worldwide with a global annual cost over $100 billion USD.^1,2 HF disproportionately affects individuals living in low- and middle-income countries (LMICs), who face growing health disparities due to the confluence of fragile health systems, extreme poverty, difficult access to healthcare, and potential differences in genetics and risk factor distribution. The burden of HF, defined as related morbidity and mortality, in LMICs is greater than in higher income countries due to limited diagnostic capacity, access to therapeutics and advanced care, and provider awareness and prescribing patterns.³ For example, recent studies indicate that patients hospitalized with acute heart failure in LMICs were less likely to receive goal directed medical therapy and were approximately 50% more likely to die in the first year after hospital discharge as compared to those in high-income settings.⁴ While the prevalence and outcomes of HF in high-income countries (HICs) have improved over the past three decades, the burden in LMICs is likely increasing.^1,5 HF case-fatality rates in hospital-based studies from LMICs are high, with six-month mortality estimates ranging from 18% to 58%.^6–8 Clear understanding of the epidemiology of HF in these settings is critical for targeted allocation of financial resources, country-level ministry of health guidelines and to inform interventions which may be unique to LMICs.

Haiti is the poorest country in the Western Hemisphere and experiences alarming rates of health disparities. Cardiovascular disease (CVD) is now estimated to be the leading cause of death in Haiti, having surpassed HIV and infectious diseases in the last decade.⁹ To date, data on the majority of CVD, including HF, in Haiti and many LMICs are derived from modeling estimates or hospital- and clinic-based samples, which suggest that the HF prevalence in Haiti is among the lowest globally. Conversely, HF has been shown to be the leading cause of hospital admission in both rural and urban hospitals in Haiti.^10,11 This disconnect between modeling estimates and hospital-based observations may be explained by the underlying data. Modeling studies used by the World Health Organization and others report data on individuals presenting with symptomatic, often end-stage, disease and may introduce ascertainment bias based on accessibility of care, income, and other individual and structural factors in these communities. Population-based studies are needed to identify the clinical epidemiology including the risk factors, onset and progression of HF that can be targeted to improve HF-related health outcomes in both Haiti and similar LMICs, under-resourced and under-represented settings where four billion persons live.^12–14

We evaluated the clinical epidemiology and characteristics of HF within the ongoing Haiti Cardiovascular Disease Cohort, which is Haiti’s first population-based longitudinal cohort study evaluating cardiovascular risk factors and diseases (Trial #NCT03892265).¹⁵ The primary objective of our analysis was to determine the age-standardized prevalence of clinical HF in Haitian adults. We also determined prevalence of HF subtypes and associated risk factors, characterized diagnostic and treatment patterns, and estimated one-year mortality.

METHODS

Data Access

The data supporting the findings of this study are available from the corresponding author after approval by the GHESKIO ethics board with a data sharing agreement.

Study setting

This study was conducted in Port-au-Prince, Haiti, which has an estimated adult population of 1.4 million. The population is young with 90% <55 years and mean life expectancy is 63 years. Approximately 75% of Haitians live on <$2/day, and Haiti is one of the most food insecure countries globally.^16–18 This study was based at the Groupe Haitien d’Etude de Sarcome et Kaposi ed de Infections Opportunistes (GHESKIO) center, which is a research, treatment and training facility dedicated to the care of Haitian residents.¹⁹ GHESKIO is located in downtown Port-au-Prince and provides free clinical services to the surrounding community.

Study design and population

This analysis included participants enrolled in the Haiti Cardiovascular Disease Cohort, a previously-described, longitudinal population-based cohort of adults ≥18 years in Port-au-Prince, Haiti (Trial #NCT03892265).¹⁵ Households were selected by multistage random sampling using global positioning system (GPS) waypoints across census blocks in Port-au-Prince, with the number of waypoints per block proportional to the estimated population size.²⁰ All adult household members were then invited for study participation including written informed consent and study enrollment at GHESKIO. Inclusion criteria included adults ≥ 18 years with a primary residence in Port-au-Prince without plans to leave in the next 24 months and absence of any serious medical condition or cognitive impairment preventing participation, as previously described.¹⁵ Enrollment took place from March 2019 to August 2021.

Study procedures

Study procedures of the Haiti Cardiovascular Disease Cohort have been previously described.¹⁵ Briefly, participants at time of study enrollment completed a clinical exam with medical history, blood pressure (BP) measurements, echocardiogram, study questionnaires on socio-demographics and health behaviors, and laboratory testing.¹⁵ BP was measured using an automated oscillometric sphygmomanometer (Omron HEM 907) following World Health Organization (WHO) and American Heart Association (AHA) guidelines.^21,22 Participants were seated for five minutes prior to measurement, with both feet on the ground and arms at heart level. Three measurements were taken in the left arm one minute apart. For this analysis, BP was determined by averaging the last two of three BP measurements. Clinical exam included questions regarding the history of hospitalization, past medical history of CVD including HF, treatment for chronic diseases, and current signs and symptoms of CVD, including the Framingham HF diagnostic criteria.²³ Cardiomegaly was determined by chest x-ray. HF medication use was also assessed through the questionnaire, specifically for the use of loop diuretics, angiotensin-converting enzyme (ACE) inhibitors, beta blockers, digoxin, spironolactone, and angiotensin receptor blockers. The study questionnaire quantified risk factors including diet, physical activity, tobacco use and alcohol consumption using a modified WHO STEPS questionnaire.²² Blood specimens were assessed for hematology, serum creatinine, glucose, HDL- and LDL-Cholesterol. Given the 2% population of HIV in Haiti and known associations between HIV and cardiomyopathy, HIV serostatus was also obtained via the Alere Determine HIV-1/2, Colloidal Gold HIV-1/2 antibody test. Follow-up study visits were conducted six- and 12-months after enrollment. Participants were additionally contacted every three months to encourage study engagement and assess for CVD events, hospitalizations and mortality.

Mortality

Death certificates and hospital/clinic records were reviewed to determine cause of death for all participants who died during follow-up. The research staff followed WHO verbal autopsy guidelines to determine the cause of death for those without these documents. Cause of death was adjudicated by two expert reviewers and disagreements were resolved by study adjudication committee review. CVD death was determined per standard guidelines and included deaths due to MI, HF or stroke.²⁴

Echocardiography

Echocardiograms were recorded for all participants with a past medical history or signs and symptoms of heart failure, stroke, myocardial infarction, or hypertension. Echocardiography was performed using a Sonosite M-turbo ultrasound machine using a P21x (5–1MHz) probe. Study staff were credentialed in echocardiography, and examinations and measurements were based on American Society of Echocardiography (ASE) guidelines.^25,26 Structural and functional disease such as systolic function and valvular disease was classified according to ASE disease-specific guidelines. More specifically, systolic dysfunction was defined as ejection fraction <50%. Diastolic function was determined per ASE guidelines with available data. HF etiologies were confirmed by two blinded, independent physician reviewers following standard criteria. In the case of disagreement between these two physician reviewers, a cardiologist was consulted for final diagnosis.

CVD risk factor definitions

Hypertension was defined as mean systolic BP (SBP) ≥140 mmHg and/or diastolic BP (DBP) ≥90 mmHg, or a self-report of anti-hypertensive medication use prescribed by a medical professional in the past two weeks, as per WHO guidelines.²² Dyslipidemia was defined as having at least one of the following: LDL >130, HDL <40 mg/dL if male and HDL <50 mg/dL if female, or triglycerides >150 mg/dL in serum blood count.²⁷ Obesity was defined as a BMI ≥30 kg/m² and overweight as BMI ≥25 kg/m² but <30 kg/m².²² Diabetes was defined as random blood glucose ≥200 mg/dl, fasting blood glucose ≥126 mg/dl or a self-reported previous diagnosis of diabetes.²⁸ An estimated glomerular filtration rate (eGFR) was calculated using the 2021 Chronic Kidney Disease Epidemiology Collaboration (CKD-EPI) creatinine-based equation without ethnic factor, as recommended by international guidelines.²⁹ Renal dysfunction was defined as eGFR <60 mL/min/1.73m². Extreme poverty was defined as income of <1 USD per day. Depression was assessed using the Patient Health Questionnaire (PHQ-9), with moderate, moderately severe and severe depression defined as PHQ-9 scores of 10–14, 15–19, and >=20, respectively.³⁰ For analyses, smoking was dichotomized as current versus never and former. Current alcohol use was defined as >1 drink per week.

Heart failure definition, subtypes and etiology classification

Clinical HF was defined per Framingham criteria (presence of at least two major Framingham HF criteria or one major criterion in conjunction with two minor criteria) without evidence of another primary etiology and/or a previous medical diagnosis of HF.²³ Available cardiac imaging included chest x-ray, ECG and echocardiogram. Natriuretic peptide testing was not available in the study setting. All HF cases including clinical criteria and supporting cardiac imaging were adjudicated by three independent reviewers and subsequently confirmed by a three-member expert adjudication panel following standard guidelines.³¹ Supplemental Table 1 describes adjudication methods.

Participants with an ejection fraction (EF) <40% were categorized as reduced EF (HFrEF), those with EF 40–49% as mid-range EF (HFmrEF) and those with EF ≥50% as preserved EF (HFpEF) per the universal definition of heart failure.³² The primary etiology of HF was determined for each participant based on echocardiographic and clinical history. Ischemic cardiomyopathy was defined as depressed left ventricular ejection fraction (LVEF) <50% with supportive ECG findings and/or presence of regional wall motion abnormality on echocardiography. Hypertensive heart disease (HHD) was defined as clinical HF and diagnosis of hypertension. Valvular and rheumatic heart disease was based on ACC and AHA guidelines.³² Idiopathic systolic cardiomyopathy was defined as clinical HF with an LVEF <50% without evidence of another etiological cause. Idiopathic HFpEF was defined as clinical HF with an LVEF ≥50% with no other underlying etiology.

Statistical analysis

Descriptive statistics were calculated to characterize study participants and the prevalence of clinical HF. Differences in means of continuous variables were assessed using Student’s two-sided t-tests. Logistic regression was used to assess the association of risk factors with clinical HF. First, univariate logistic regression models were constructed to assess each variable’s association with HF. All variables with a univariate p-value <0.20 were included in an initial multivariable model and eliminated by backward subset selection until all remaining variables had p-values <0.05. Dropped variables were individually added back to assess possible confounding. Multiple imputation with chained equations (MICE) was used prior to regression modeling for variables with over 2% missingness, which included eGFR (2.7%), diagnosis of diabetes (2.6%), and dyslipidemia (2.8%). HF prevalence was age-standardized using WHO standard population data.³³ Study participants who were lost-to-follow-up were censored at the last contact date. Kaplan-Meier survival analysis was used to estimate mortality and compared by log-rank test. We also estimated Hazard Ratios (HRs) using Cox proportional hazards regression. Given the rarity of mortality in the study population, there was insufficient power to conduct multivariable analysis. All analyses were conducted using R statistical software (Version 4.0.2).

Ethics approval and consent to participate

This study complies with the Declaration of Helsinki. The study protocol and ethical consent forms were approved by Weill Cornell Medicine (IRB#1803019037–20) and GHESKIO Ethics Board. Individuals agreeing to participate in the study provided written informed consent prior to enrollment.

RESULTS

Participant characteristics

Among 3,005 participants enrolled in the Haiti Cardiovascular Disease Cohort, 24 (0.8%) were missing information from the Framingham HF questionnaire; therefore, 2,981 participants were analyzed (Supplemental Figure 1). Table 1 describes the characteristics of the cohort. The median participant age was 40 years (IQR 27–55), 58.2% were female and all reported Black race. The majority had received secondary education or above (64.1%) and reported income ≤1 USD per day (70.3%). The crude prevalence of hypertension was 32.6%, while 17.2% had obesity and 2.9% had renal dysfunction. Dyslipidemia was present in 43.8% of the cohort, while diabetes and HIV were present in 5.6% and 1.6%, respectively. PHQ-9 depression screening reveled moderate-to-severe depression in 12.6% of participants. Nearly one in ten (8.3%) and 16.6% of participants reported current tobacco and alcohol use >1 drink per week, respectively.

Table 1:

Baseline characteristics of the 2,981 adults in Haiti CVD Cohort Study

Status	All Participants	No Heart Failure	Heart Failure	p-value*
n (%)	2,981	2,888 (96.9%)	93 (3.1%)
Age				<0.001
Median (IQR)	40 [27, 55]	40 [27, 54]	57 [45, 65]
18–29	882 (29.6)	875 (30.3)	7 (7.5)
30–39	564 (18.9)	558 (19.3)	6 (6.5)
40–49	527 (17.7)	507 (17.6)	20 (21.5)
50–59	496 (16.6)	475 (16.4)	21 (22.6)
60–69	387 (13.0)	362 (12.5)	25 (26.9)
70 and over	125 (4.2)	111 (3.8)	14 (15.1)
Sex
Female	1735 (58.2)	1672 (57.9)	63 (67.7)	0.074
Education level				<0.001
None	426 (14.3)	400 (13.9)	26 (28.0)
Primary	640 (21.5)	615 (21.4)	25 (26.9)
Secondary	1469 (49.4)	1434 (49.8)	35 (37.6)
Higher than secondary	438 (14.7)	431 (15.0)	7 (7.5)
Income
≤1 USD per day	2089 (70.3)	2011 (69.8)	78 (83.9)	0.005
Habits
Smoking, current or past	245 (8.3)	238 (8.3)	7 (7.5)	0.940
Alcohol, >1x per week	493 (16.6)	487 (17.0)	6 (6.5)	0.011
BMI				<0.001
Median (IQR)	24 (21–28)	24 (21–28)	28 (23–32)
Normal or underweight (BMI ≤25)	1689 (56.7)	1658 (57.4)	31 (33.3)
Overweight (25 ≤ BMI <30)	777 (26.1)	750 (26.0)	27 (29.0)
Obese (BMI ≥30)	513 (17.2)	478 (16.6)	35 (37.6)
Vital signs
Heart rate, bpm, median (IQR)	75 [67, 83]	75 [67, 83]	78 [70, 88]	0.035
Systolic blood pressure, mmHg, median (IQR)	119 [107, 138]	118 [107, 137]	133 [116, 159]	<0.001
Diastolic blood pressure, mmHg, median (IQR)	72 (63–84)	72 (63–83)	79 (68–94)	<0.001
CVD Risk factors
Hypertension	971 (32.6)	906 (31.4)	65 (69.9)	<0.001
Renal dysfunction†	84 (2.9)	65 (2.3)	19 (21.1)	<0.001
Dyslipidemia	1268 (43.8)	1215 (43.3)	53 (58.2)	0.006
Diabetes†	162 (5.6)	152 (5.4)	10 (11.1)	0.037
HIV	48 (1.6)	45 (1.6)	3 (3.2)	0.544
Moderate-to-severe depression (PHQ9 ≥10)	375 (12.6)	355 (12.4)	20 (21.5)	0.038

^*Heart Failure vs No Heart Failure

^†Missing more than 2%

Abbreviations: IQR: Inter-quartile range; USD: US Dollar; BMI: Body mass index; mmHg: millimeters mercury; eGFR: estimated glomerular filtration rate; PHQ-9: Patient-Health Questionnaire 9; LDL: Low-density lipoprotein; HDL: High-density lipoprotein; bpm: beats per minute

Heart failure prevalence and symptoms

The overall prevalence of clinical HF was 3.1% (93/2,981), with an age-standardized prevalence of 3.2% (95% CI: 2.6–3.9%) (Table 1). Among Framingham criteria in HF patients, dyspnea on exertion (80/93, 86.0%), orthopnea (67, 72.0%), lower extremity edema (63/93, 67.7%) and cardiomegaly (23/93, 24.7%) were most frequently reported (Table 2). Mean age for participants with HF was 57 years (IQR 45–65) and increased with age (age 18–29 years: 0.8%; 30–39: 1.1%; 40–49: 3.8%; 50–59: 4.2%; 60–69: 6.5%; ≥70: 11.2%, (p<0.001 for linear trend across groups). The first statistically significant increase in HF prevalence as compared to the previous age group (from youngest to oldest) occurred between the 30–39-year and 40–49-year groups (1.1% vs 3.7%, p=0.003) (Figure 1). HF prevalence was greater among women (3.6%) than men (2.4%) (p=0.050).

Table 2:

Framingham criteria of heart failure and non-heart failure participants

Characteristic	No Heart Failure	Heart Failure	p-value
n (%)	2,888 (96.9%)	93 (3.1%)
Major criteria
Pulmonary edema	0 (0.0)	0 (0.0)	-
Cardiomegaly	29 (1.0)	23 (24.7)	<0.001
Jugular venous distension	3 (0.1)	7 (7.5)	<0.001
Orthopnea/Paroxysmal nocturnal dyspnea	70 (2.4)	67 (72.0)	<0.001
Pulmonary rales	12 (0.4)	7 (7.5)	<0.001
S3 heart sound	0 (0.0)	1 (1.1)	-
Minor criteria
Ankle edema	155 (5.4)	63 (67.7)	<0.001
Dyspnea on exertion	455 (15.8)	80 (86.0)	<0.001
Hepatomegaly	0 (0.0)	3 (3.2)	<0.001
Pleural effusion	0 (0.0)	0 (0.0)	-
Tachycardia (HR > 120)	5 (0.2)	3 (3.2)	0.537

Figure 1:

Heart failure prevalence stratified by age and subtype (HFrEF, HFmrEF and HFpEF). Aggregate HF prevalence per age group represented by line.

Abbreviations: HF: Heart Failure; HFrEF: Heart Failure with Reduced Ejection Fraction; HFmrEF: Heart Failure with Mid-Range Ejection Fraction; HFpEF: Heart Failure with Preserved Ejection Fraction

One-year mortality

At the time of analysis, the median follow-up time for participants was 15.4 months (IQR: 9–22). Per Kaplan-Meier survival analysis, one-year mortality of persons with HF was 6.6% (95% CI: 2.8–15.4%) as compared to 0.8% (95% CI: 0.5–1.3%) among those without HF, with a hazard ratio of 7.7 (95% CI 2.9–20.6, p<0.0001) (Figure 2). After adjustment for age, mortality among participants with HF remained significantly higher as compared to those without HF (aHR: 4.54 [95% CI: 1.64–12.60]). Among HF participants, five died within one-year and 80% (4/5) were classified as cardiovascular deaths per the study adjudication committee. Overall mortality was higher among participants with HFrEF (14.2%, 3/21) as compared to those with HFpEF (3.0%, 2/66). Among those without HF, 20 died within one-year and 50% (10/20) were classified as cardiovascular deaths (Supplemental Table 2).

Figure 2:

Kaplan-Meier survival analysis comparing participants with and without heart failure at one-year.

Abbreviations: HF: Heart Failure; HR: Hazard Ratio; Cl: Confidence Interval

Heart failure subtype and etiology

Among participants with HF, 71.0% (66/93) had preserved (HFpEF), 6.5% (6/93) had mid-range (HFmrEF) and 22.6% (21/93) had reduced ejection fraction (HFrEF) (p<0.001) (Figure 1). Hypertensive heart disease was the most common underlying etiology (49.5%, 46/93), followed by idiopathic HFpEF (16.1%, 15/93), idiopathic HFrEF (15.1%, 14/93) and valvular heart disease 15.1% (14/93). Three participants with clinical HF (3.2%) had diagnostic evidence of rheumatic valvular disease on echocardiography. One patient (1.1%) with ischemic disease was identified.

Factors associated with heart failure

In univariate analysis, renal dysfunction (OR: 11.35 [95% CI: 6.46–19.93]), hypertension (5.07 [3.23–7.96]), obesity (3.92 [2.39–6.46]), dyslipidemia (2.93 [1.78–4.83]), diabetes (2.21 [1.13–4.33]), age (OR: 1.77 [1.51–2.08] for each 10-year increase in age) and moderate or severe depression (1.73 [1.05–2.75]) were significantly associated with a higher odds of HF (Table 3). Secondary education or greater (0.45 [0.30–0.68]) and alcohol use >1 drink per week (0.34 [0.13–0.71]) were associated with a significantly lower odds of HF.

Table 3:

Crude and multivariate regression models of factors associated with heart failure in the study cohort.

Age Categories	Crude Odds Ratio	p-value	Adjusted Odds Ratio	p-value
Per 10-year increase	1.77 [1.51–2.08]	<0.001	1.36 (1.12–1.66)	0.002
Sex
Male	1 (ref)	-
Female	1.53 [0.99–2.40]	0.060
Education
Primary or lower	1 (ref)	-
Secondary or higher	0.45 [0.30–0.68]	<0.001
Income (Daily)
>1 USD	1 (ref)
≤1 USD	2.25 [1.33–4.09]	0.004	2.10 [1.18–3.72]	0.011
Salt use
Never/Rarely/Sometimes	1 (ref)	-
Usually/Often	0.72 [0.39–1.21]	0.240
Alcohol use
≤1x per week	1 (ref)	-
>1x per week	0.34 [0.13–0.71]	0.011
Smoking Status
Never	1 (ref)
Current/Former	0.90 [0.37–1.83]	0.792
BMI category (kg/m2)
Normal Weight	1 (ref)
Overweight (25≤ BMI<30)	1.93 [1.14–3.25]	0.014	1.70 [0.99–2.92]	0.055
Obese (BMI>30)	3.92 [2.39–6.46]	<0.001	3.35 [1.99–5.62]	<0.001
Hypertension
No	1 (ref)
Yes	5.07 [3.23–7.96]	<0.001	2.14 [1.26–3.66]	0.005
Renal dysfunction
No	1 (ref)
Yes	11.35 [6.46–19.93]	<0.001	5.42 [2.94–9.98]	<0.001
Diabetes *
No	1 (ref)
Yes	2.21 [1.13–4.33]	0.021
Dyslipidemia
No	1 (ref)
Yes	2.93 [1.78–4.83]	<0.001
Depression
None or mild (PHQ9 <10)	1 (ref)	-
Moderate or severe (PHQ9 ≥10)	1.73 [1.05–2.75]	0.026

^*Missing >2%. Missing data imputed using MICE. Reported values are estimations based on 20 multiple imputations.

In multivariable analysis, renal dysfunction (aOR: 5.42 [2.94–9.98]), obesity (3.35 [95% CI: 1.99–5.62]), hypertension (2.14 [1.26–3.66]), poverty (2.10 [1.18–3.72]) and age (1.36 [1.12–1.66] per 10-year increase) were significantly associated with prevalent HF (Table 3).

Heart failure awareness and treatment

Among those with HF, 44.1% (41/93) were aware of the diagnosis and 11.8% (11/93) had been previously hospitalized with HF (Table 4). More than one-half of those aware of their diagnosis (63.4%, 26/41) were taking heart failure appropriate medications. Among those on heart failure medications, 38.5% (10/26) had been prescribed one and 30.8% (8/26) two, 19.2% (5/26) three and 11.5% (3/26) four medications. The most common medications previously prescribed were ACE inhibitors or angiotensin receptor blockers (ARBs) (65.4%; 17/26), followed by loop diuretics (61.5%, 16/26), beta blockers (50.0%, 13/26), digoxin (15.4%, 4/26) and spironolactone (11.5%, 3/26). Nephrilysin inhibitors were not available. Data on medication stock and adherence was unavailable. Among HF subtypes, participants with HFrEF, as compared to those with HFpEF, were significantly more likely to have been previously diagnosed with HF (77.8% (21/26) vs. 29.9% (20/67), p<0.001) and hospitalized (33.3% vs. 3.0%, p=0.003) (Table 4).

Table 4:

Heart failure clinical management stratified by subtype, including rate of hospitalization, diagnosis, and treatment by medication.

Status	All HF N (%)	HFpEF N (%)	HFmrEF N (%)	HFrEF N (%)
Number of HF cases	93	66 (71.0)	6 (6.5)	21 (22.6)
Previously diagnosed	41 (44.1)	20 (30.3)	3 (50.0)	18 (85.7)*
Previously hospitalized with HF	11 (11.8)	2 (3.0)	2 (33.3)	7 (33.3)*
On HF medications	26 (28.0)	11 (16.7)	2 (33.3)	13 (61.9)*

^*p<0.05 as compared to HFpEF

Abbreviations: HF: heart failure; HFpEF: heart failure with preserved ejection fraction; HFmrEF; heart failure with mid-range ejection fraction; HFrEF: heart failure with reduced ejection fraction

DISCUSSION

In this large population-based cohort of Haitian adults, age-standardized HF prevalence was 3.2%, which is alarmingly 50% higher than in high-income countries and 5-fold higher than modeling estimates in Haiti and other LMICs.¹ The average age of adults with HF was notably young at 57 years, with the first significant increase in HF prevalence between ages 30–39 and 40–49 years, and HF was associated with a significantly higher one-year mortality than in the general population. Additionally, HFpEF accounted for the vast majority (71%) of all HF cases, and hypertensive heart disease was the most common underlying etiology. Our data underscores that HF is a likely an under-recognized cardiovascular disease and health disparity given the difference in modeling estimates as compared to our population-based data and given the difference in prevalence and age of onset among this Haitian cohort as compared to higher-income settings. Further research and health policy prioritization for prevention and treatment are urgently needed.

The age-standardized prevalence of HF in this population-based study was 3.2%. This finding is striking for two reasons. First, modeling studies, which are the only source of HF epidemiology estimates currently available in most LMICs such as Haiti, approximate that the prevalence of HF is lower in LMICs than HICs.¹ Conversely, hospital-based studies indicate that the prevalence of HF may be dramatically higher, with numerous studies reporting HF among the most common causes of admission.^10,11 Recently, the Global Burden of Disease (GBD) study estimation of HF prevalence in Haiti was 0.68%—which is approximately one-fifth our finding of 3.2%.¹ The same study estimated HF prevalence across all low-income countries to be approximately 0.77%, again highlighting a possible discordance between modeling studies, where locally derived model inputs are sparse, and population-based estimates. Second, the prevalence of HF in the United States ranges from 0.96% according to the GBD to 2.2% as reported in the most recent NHANES population-based data.^1,34 Both are dramatically lower than the 3.2% reported herein. Of note, Black patients have the highest prevalence of HF in the US among all races, with estimates of 3.5% in 2016, which is more similar to our estimates.⁵ A similar, and high, prevalence of HF among this Black Haitian cohort and Black Americans underscores a known cardiovascular health disparity and raises opportunities for further investigation of individual-, neighborhood-, and environmental-level influences contributing to the disparity gap.³⁵

Taken together, these population-based data combined with global estimates and hospital-based studies indicate that HF prevalence may be dramatically higher than previously reported, and HF may represent an impending crisis in already fragile healthcare systems in these under-resourced settings most subject to diagnostic, economic and environmental constraints. While many have advocated for the need for population-based CVD cohorts in LMICs, they require resources and research infrastructure that often do not exist in under-resourced settings.¹³ As the prevalence of HF is expected to rise given the burgeoning CVD epidemic across LMICs, coordinated efforts are needed to describe the burden and characteristics of HF globally and inform both policy makers and interventions for prevention and treatment.

We also report that HF was present at younger ages in our study as compared to high-income settings. While increasing age was independently associated with HF, the median age of adults with HF in this population-based sample was 57 years compared to 72 and 70 in the U.S. and Europe, respectively.^36,37 Among those with HF in our study, 35.5% were under 50 years of age; a substantially higher proportion as compared to 8.4% reported in a MAGGIC group meta-analysis comparing 41,926 adults with HF in high-income countries.³⁸ Additionally, NHANES data from the U.S. recently indicated the prevalence of HF to be 0.3% in adults aged 20–39 and 1.5% among those 40–59, dramatically lower than the 2.3% and 4.0% in our study.⁵ Our findings also support hospital-based studies from Sub-Saharan Africa showing HF at younger ages and confirm a similar trend in a population-based cohort.^6,39,40 The underlying etiology of this shift is likely multifactorial including earlier onset of risk factors such as elevated blood pressure, higher burden of infectious diseases with resulting myocardial dysfunction (HIV, rheumatic, etc.) and increased socio-economic factors linked to HF such as extreme poverty, stress and nutritional deficiencies. For example, studies indicate that hypertension occurs at younger ages in LMICs including Haiti, is rapidly increasing in prevalence, and is associated with incident HF.^41,42 Black populations are also at higher risk of uncontrolled hypertension and resulting complications, such as hypertensive HF, due to biologic differences in vascular function.⁴³ In our study, hypertension was associated with a 2.1-fold increase in HF, independent of age, making it the leading modifiable risk factor in the cohort.

In our study HFpEF was the largest HF subtype, constituting 71% of HF cases. While studies in HICs have reported a shift in HF subtype from HFrEF to HFpEF, this trend has been underrecognized in LMICs.⁴⁴ Among known HFpEF risk factors, hypertension is the most strongly associated, and the burden of hypertension in Haiti and other LMICs is significantly higher as compared to high-income settings.⁴⁵ Additionally, given increasing understanding of the genetics of cardiovascular diseases, other diseases such as amyloidogenic transthyretin cardiac amyloidosis (ATTR-CA), which is associated with pathologic genetic polymorphisms within the TTR gene in populations of African descent, may have both a higher prevalence in Haiti and present with HFpEF phenotypes. Importantly, participants in our cohort with HFpEF were more than two-fold less likely to have been diagnosed prior to the study compared to those with HFrEF. This may be due to decreased clinician awareness of diastolic disease and/or increasing availability of imaging modalities such as echocardiography in these under-resourced settings, similar to the increasing awareness and availabilities in the United States from 1970–80.⁴⁴ These data, taken together, indicate that among the large burden of HF in Haiti, HFpEF is the most frequent, yet also most underdiagnosed and treated. In order to prevent the growing impact of HF in this and similar settings, studies investigating the care cascade of HFpEF are urgently needed.

Surprisingly, less than half of participants with HF in our cohort had been previously diagnosed, yet the majority (63.4%) of those previously diagnosed had been prescribed appropriate medical therapy. The proportion HF patients aware of their diagnosis in LICs remains unknown; however, recent studies in South Africa and Nigeria indicate that as many as 92% of hospital-presenting patients had not been previously diagnosed.^46,47 Encouragingly, treatment rates among those aware of their HF diagnosis in our study compare favorably to other, high-resource settings. A recent study including 3,518 HFrEF patients from 150 clinics in the United States reported that approximately 75% of HF patients were prescribed at least one appropriate HF medication, similar to the 63.4% of overall HF patients and 72.2% of HFrEF patients in our study.⁴⁸ These data illustrate that even in a fragile health system such as Haiti where access to care and treatment is extremely challenging given severe poverty, civil unrest, and limited public awareness of CVD, many patients were able to initiate treatment. To improve the proportion of patients with HF who receive optimal care, interventions targeting diagnosis are essential to ultimately improve the HF care continuum and avert HF-related morbidity and mortality.

Strengths of this study include that it is a population-based cohort of Black adults living in one of the poorest and most under-resourced countries in the world that faces a growing CVD epidemic with previously no population-based data on HF. Second, the study protocol systematically screened and diagnosed HF using Framingham clinical heart failure diagnostic criteria, clinical and echocardiography data for adjudicated HF classification comparable to HF outcomes in other international cohorts. Of note, estimates of HF prevalence defined by epidemiologic criteria using probable as well as definitive diagnosis of HF may be higher than our estimates based strictly on Framingham criteria.³¹ Limitations of our data included our inability to identify ischemic heart disease (IHD) given diagnostic capacity (lack of cardiac catheterization and stress testing); therefore, we may have underestimated the role of IHD as an etiology of HF, a limitation common in other studies in LMICs.^10,12,39 We also lacked advanced laboratory testing for biomarkers such as BNP and MR-proANP. The average life expectancy in Haiti may influence the relatively low mean age of participants with HF in this study cohort; however, our data still suggest that HF is much more prevalent among young age groups (e.g. < 50 years) in Haiti as compared to high-income settings and this age stratified comparison should not be influenced by life expectancy. Additionally, it is possible that our estimations of HF may be conservatively low if the distribution of education level in other parts of Haiti or LMICs is less educated than our study population. Finally, our survey did not contain sufficient information about participants’ pregnancy history to accurately diagnose peripartum cardiomyopathy.

In conclusion, among a population-based sample of 2,981 Haitian adults, the age-standardized prevalence of clinical HF was 3.2% which is alarmingly higher than both high-income settings and estimates for Haiti and LMICs. Additionally, HF occurs nearly two decades earlier than in HICs and HFpEF is the most common HF subtype, with hypertension the most common underlying etiology. Participants with HF had a significantly higher one-year mortality as compared to those without HF. Taken together, these data provide insight into the under-recognized and growing epidemic of heart failure that is emerging across LMICs, which bear 80% of the global burden of CVD and associated poor health outcomes. Research is urgently needed to further characterize HF at the population level in other LMICs, where four billion persons live, in order to guide accurate international resource allocation and develop pragmatic interventions to improve HF prevention, diagnosis and treatment to ultimately reduce global CVD health disparities.

What is Known?

• Cardiovascular disease is the leading cause of death globally, with 80% of the burden in low- and middle-income countries (LMICs).

• Heart failure prevalence in LMICs is estimated based primarily on modeling studies; therefore, population-based estimates are needed to provide accurate data on prevalence, age of onset, subtypes, and risk factors.

What the Study Adds?

• Among a population-based cohort of 2,981 adults in Haiti, the age-standardized prevalence of clinical HF was 3.2%; 5-fold higher than previous modeling estimates for LMICs and 50% higher than high-income settings.

• HF occurred two decades earlier in Haiti than high-income countries (57 years vs. 77 years), HF with preserved ejection fraction was the most common subtype (71% of cases), and one-year mortality was 6.6% among those with HF versus 0.8% among community members without HF (HR: 7.7 [95% CI 2.9–20.6], p<0.0001).

• These data underscore that HF is an under-recognized cardiovascular disease in LMICs.

NON-STANDARD ABBREVIATIONS AND ACRONYMS

ACE

Angiotensin-Converting Enzyme

AHA

American Heart Association

ASE

American Society of Echocardiography

ATTR-CA

Amyloidogenic Transthyretin Cardiac Amyloidosis

BMI

Body Mass Index

BP

Blood Pressure

CKD-EPI

Chronic Kidney Disease Epidemiology Collaboration

CVD

Cardiovascular Disease

DBP

Diastolic Blood Pressure

EF

Ejection Fraction

EGFR

Estimated Glomerular Filtration Rate

GBD

Global Burden of Disease Study

GHESKIO

Groupe Haitien d’Etude de Sarcome et Kaposi ed de Infections Opportunistes

GPS

Global Positioning System

HDL

High Density Lipoprotein

HF

Heart Failure

HHD

Hypertensive Heart Disease

HFmrEF

Heart Failure with Mid-Range Ejection Fraction

HFpEF

Heart Failure with Preserved Ejection Fraction

HFrEF

Heart Failure with Reduced Ejection Fraction

HIC

High-Income Countries

HR

Hazard Ratio

IHD

Ischemic Heart Disease

LDL

Low Density Lipoprotein

LMIC

Low- and Middle-Income Countries

LVEF

Left Ventricular Ejection Fraction

MAGGIC

Meta-Analysis Global Group in Chronic Heart Failure

MHz

Megahertz

MICE

Multiple Imputation by Chained Equations

NHANES

National Health and Nutrition Examination Survey

OR

Odds Ratio

PHQ-9

Patient Health Questionnaire

SBP

Systolic Blood Pressure