Cardiovascular Health in a Single Community in Rural Haiti: A Cross-sectional Study

Introduction There is a growing burden of cardiovascular disease in low- and middle-income countries and assessment of cardiovascular health (CVH) may identify populations at risk for poor CVH. Methods Between July 2014 and August 2014, we performed a household survey from a convenience sample among adult community members in rural northern Haiti. We used a modified World Health Organization STEPwise approach to chronic disease questionnaire to capture self-reported data on tobacco, diet, physical activity, and diabetes, and measured blood pressure and body mass index. We used an adapted American Heart Association definition and thresholds for determining ideal, intermediate, and poor cardiovascular health. We used linear and logistic regression to examine associations between socio-demographic characteristics with CVH score and ideal CVH. Results Among 540 participants (mean [SD] age = 40.3 [17.1] years, 67% women), there was a high prevalence of poor CVH (n=476, 88.1%) compared with intermediate (n=56, 10.4%) and ideal (n=41, 7.6%) CVH. Ideal metrics for blood pressure (47%) and diet (26%) were least often met, while body weight (84%), physical activity (83%), and smoking (90%) were most often met. Men were associated with better CVH score (0.31, [0.04–0.59]; P=0.03), and being a farmer was associated with ideal CVH (P=0.006). Conclusion In this community-based sample of a farming community in rural Haiti, very few adults had ideal CVH. Higher CVH score was associated with male sex, and farming as a primary occupation. Women and non-farmers may represent at-risk subgroups within this population. Blood pressure and diet may represent possible areas for improvement.


INTRODUCTION
Haiti is the most impoverished nation in the Americas and has a high burden of cardiovascular disease (CVD) including hypertension, diabetes, and heart failure. [1][2][3][4][5] In Haiti, ischemic heart disease and cerebrovascular disease are estimated to be the top two causes of death. 6 One strategy to reduce CVD is primary prevention by improving cardiovascular health (CVH). CVD incidence and mortality is lower when a set of health behaviors (smoking, healthy diet, body mass index [BMI], and physical activity) and health factors (blood pressure, cholesterol, and glucose) are at ideal levels. 7,8 Identifying areas of improvement in CVH may inform policymakers and guide interventions to promote CVH through screening and public health initiatives. The Haiti Mortality, Morbidity and Service Use Survey (EMMUS-VI 2016-2017) reported a high prevalence of CVD risk factors like hypertension and diabetes, with low health service utilization. Studies comparing CVH in low-and middleincome countries may have region specific needs to reduce CVD, for example although in much of the world men are at greater risk of ischemic heart disease than women, there is no sex difference observed in sub-Saharan Africa, underscoring the need for regional data. 9 Therefore, obtaining data at a regional level is extremely important for public health officials to make appropriate guidelines and recommendations. The objective of this manuscript is to report the CVH of a community sample of rural Haitians to identify the prevalence of ideal CVH and subpopulations at greatest risk.

METHODS
From July 2014 to August 2014, we obtained a convenience sample of 572 community-based participants at their homes in Fontaine, a town of approximately 10,000 people in northern Haiti. A private hospital and non-profit clinic is available approximately 8km away, and the primary economic activity is farming. Research assistants trained by one of the investigators (V. Polsinelli) conducted interviews in Haitian Creole, the locally spoken language. Every household within a 3-5 square kilometer radius of the town center was offered participation in the study. Subjects at least 18 years of age were selected through a non-random voluntary basis in each household, members of the household who were not present were re-visited at a later time. If household members were living and working away from home, they were not sampled.
We used a modified step 1 and 2 of WHO STEPwise approach to chronic disease questionnaire. 10 Routine STEPwise questions were used to obtain information pertinent to CVH including daily fruit and vegetable intake, physical activity by Global Physical Activity Questionnaire, tobacco use, and self-reported diagnosis of hypertension, diabetes, and cardiovascular diseases. Anthropometric measurements were obtained in each participant's home. Height was measured on a flat surface with the participant standing upright without shoes using a tape measure and clipboard. Weight was measured with electronic scales (Seca 803, Chino, CA) on a flat, hard surface; participants were permitted to wear light clothing but no shoes. Brachial oscillatory blood pressure and pulse were measured three times using an automated device (Omron BP785 10 Series, Lake Forest, Illinois) after the participant was sitting at rest for at least 5 minutes with arm, back, and feet supported. The mean of the three measurements was used. Scales and blood pressure devices were calibrated by the manufacturer but not by the study team. Blood glucose samples were not taken.
We adapted the AHA definition for ideal, intermediate, and poor CVH to categorize the study population. 11 Definitions of each category are outlined in Table S1. Each of the six available metrics were allocated a score of 0 (poor), 1 (intermediate), or 2 (ideal) based on convention. 11 Three categories, smoking, diet, and diabetes, were allocated a score of 0 or 2 because data on recent tobacco cessation and impaired fasting glucose were not collected, diabetes was self-reported. Ideal CVH was defined as full achievement of all factors (CVH score = 12 out of 12). Intermediate CVH is defined as a participant having at least one intermediate metric, and no poor metrics. Poor CVH is defined as having at least one poor metric.
To determine differences in CVH score between sexes, we used X 2 for categorical variables, Student's t-tests for continuous variables and Wilcox rank sum for nonparametric data. Univariate (model 1) and multivariate (model 2) linear regression were used to examine associations between demographic characteristics and CVH score, and ideal CVH. Model 1 is age adjusted linear regression of the independent covariates, and model 2 is age adjusted and adjusted for all other listed covariates.
A two-sided P value < 0.05 defined statistical significance. All statistical analyses were performed using Stata version 12 (StataCorp, LLC, College Station, Texas, USA). The Comité National de Bioéthique, Haiti and the Health Sciences Institutional Review Board of the State University of New York at Buffalo approved the study. We obtained informed consent from all participants.

RESULTS
Among the 572 participants interviewed, complete information for CVH determination was available in 540 participants. The number of individuals screened was not recorded, but participation in study was high (estimated >95% of those invited participated). Participants' mean (SD) age was 40.2 (17.1) years, and 66% were women. Only a minority completed secondary school (11% The prevalence of ideal, intermediate, and poor CVH was 7.2%, 9.8%, and 83.1% respectively and there were no differences by sex (P=0.6). Prevalence by each metric is shown in the Figure 1. Overall, ideal CVH metrics were more often observed among men compared to women. A greater proportion of men had ideal CVH metrics for blood pressure, body weight, and physical activity more often (P<0.02). One exception is smoking, as fewer women reported smoking tobacco compared to men (P=0.005). There were no differences for diabetes or diet.
Prevalence of ideal (green), intermediate (yellow), and poor (red) cardiovascular health metrics overall, and by category is separated by sex. Percentage of the total men or women is on the x-axis. Significant sex differences were observed in the categories of blood pressure, body weight, and smoking (P<0.05) †.
The mean ± SD CVH score was 9.1 ± 1.6 overall in the sample. Within the total cohort we observed a difference in CVH score by sex, 9.3 ± 1.5 among men and 9.0 ± 1.7 among women (P = 0.004). We evaluated the association between social and demographic determinants and total CVH score using multivariable regression-shown in the Table 1. After adjustments for age, the female sex was associated with worse CVH score. After adjustments for educational achievement and occupation, there were no observed associations with social and demographic characteristics. After adjustments for age, sex, and educational achievement, being a farmer was associated with ideal CVH (ß =2.38 (0.69 -4.08; P=0.006).

DISCUSSION
In this convenience sample of a rural agrarian community in northern Haiti, we described the distribution of CVH among Haitian adults, and several population characteristics associated with ideal CVH. Among all surveyed adults, we observed a very low prevalence of ideal CVH of 7.2%. Several ideal CVH metrics were more common among men compared to women, and an occupation of farmer was associated with ideal CVH. These data may help inform public health officials in rural Haiti to develop programs aimed at improvement of CVH.
Among our study participants, low smoking prevalence and smoking differences between men and women were similar to other studies in Haiti. 2,6 Comparing our data to a 2018 study from rural Haiti, the current study's prevalence of overweight or obese was close to the reported estimate of 18%. 2 Thus, these measurements of body weight appear consistent with rural lifestyle. Very few participants, (3% women; 2% men) had selfreported diabetes. This prevalence is likely an underestimate due to ascertainment bias as other studies utilizing fingerstick random glucose or hemoglobin A1c measurements have reported diabetes prevalence ranging from 5% to 20%. 2, 5, 6 EMMUS-VI observed a relationship in diabetes prevalence and socioeconomic position. They observed diabetes prevalence to be lower for people in the lowest vs highest socioeconomic quintile for both women (11% vs. 18%) and men (5% vs. 10%), though the data from EMMUS-VI was based on an age range of 35 to 64, which is different than our studied sample. Other studies have observed high prevalence of food insecurity, and associations of food insecurity with illiteracy, poverty, less diverse diet, and death from cholera. 12,13 Thus, it is possible that regional food insecurity may explain our findings of lower BMI and lower prevalence of diabetes.
Our reported prevalence of ideal CVH of 7.2% is severalfold higher than other countries, including the United States (0.1%) and Ghana (0.3%), although we used a different scale in the current study. 7,14,15 By each category, achievement of ideal CVH metrics was most often met within smoking, physical activity, diabetes, or BMI, and least often met within the hypertension and diet categories. These trends are similar to what has been Figure 1. Distribution of cardiovascular health by sex observed in rural Ghana. 14 Many lifestyle characteristics are optimized for ideal BP (low BMI, high level of physical activity); however, BP control levels are low. 1 Thus improving access to medications to lower BP treatment may contribute to improved CVH in this population. Our findings of low ideal CVH are consistent with the high observed burden of heart failure and estimated burden of ischemic heart disease and stroke in Haiti. 4,16,17 This study observed relevant trends with social and demographic characteristics. After adjustment for all other covariates, farmers were more likely to have ideal   CVH; we speculate this is likely due to a physically active lifestyle necessary for farm work. Several metrics of CVH were observed to be worse in women, including our report of physical activity and weight. Culturally, women typically sell goods, or work around the home which are activities that may provide less physical activity and may explain why CVH metrics were worse in women. Furthermore, there exists in Haiti a high prevalence of pre-eclampsia which may explain poor CVH metrics, particularly BP, among women. These data suggest a need for CVH promotion particularly among women. Further investigation may be integrated at surveillance data including why women may be at higher risk, and linking those data with health system data to trial resource-effective ways to improve CVH.
This study has several limitations. Random sampling methods were not used but convenience sample chosen to optimize sample size, however, may biased towards participants who were wealthier and closer to the urban center. Re-sampling participants who were working during initial was inconsistently performed, and could bias the population toward less healthy individuals not able to work. Our diet metric was adapted as a simple and convenient measure, and the adaptation of the WHO STEPS instrument for development of the CVH score has not yet been validated. Diabetes was self-reported, however, we used a standardized instrument to minimize reporting bias and to facilitate comparisons to other populations.

CONCLUSION
In conclusion, this study among this rural, Haitian population demonstrates a low prevalence of ideal CVH overall. Farmers are most likely to have ideal CVH, and women had worse CVH compared with men. We identified blood pressure and healthy diet as targets for interventions to improve CVH at a regional level. These data may have important implications for CVH health promotion and cardiovascular disease prevention and control.