Substantial evidence suggests an increasing burden of hypertension (HTN) in urban sub-Saharan Africa (SSA). However, data on HTN prevalence in rural SSA are sparse. In a cross-sectional study, we investigated magnitude and correlates of HTN in rural SSA. Study participants (N=1485), 18 years and above, were selected using a stratified random sampling technique from three villages (in Malawi, Rwanda and Tanzania) that participated in the Millennium Villages Project. Information on socio-demographic characteristics, risk factors and blood pressure measures was collected using standardized protocols. Prevalence of HTN and pre-HTN were 22 and 44%, respectively. Older age (P<0.001), higher body mass index (BMI) (P=0.07), television ownership (P=0.01) and less work-related vigorous physical activity (P=0.02) were associated with higher prevalence of HTN and higher blood pressure measures (all P<0.05). Frequent meat and fat intake were associated with higher HTN prevalence (trend P=0.02 and 0.07, respectively). Frequent fruit and vegetable intake was significantly associated with lower blood pressure measures (all P<0.05). HTN and pre-HTN are common in rural SSA. Modifiable risk factors (such as BMI, dietary intake and physical activity) are associated with HTN prevalence in this population, indicating potential opportunities for prevention measures.
By the year 2020, non-communicable diseases such as cardiovascular diseases (CVD) will be the major causes of morbidity and mortality in developing countries, accounting for almost four times as many deaths as from communicable diseases.1 This shift potentially coincides with socio-economic changes and the ‘nutrition transition’ associated with poverty alleviation.2, 3, 4, 5, 6
The increasing prevalence of hypertension (HTN) and associated burden on the population and health systems in urban sub-Saharan Africa (SSA) has been well documented.7, 8, 9 However, the few studies of HTN conducted in remote settings of rural SSA have reported inconsistent findings. In a recent review, the prevalence of HTN in SSA ranged from 2.3 to 41.1% among rural populations.7, 10, 11, 12, 13 Further, risk factors for HTN, such as dietary habits, are not clearly defined in these populations. This is especially important because of the global effort to improve the socio-economic status of this region and associated changes, which could potentially have both beneficial and adverse consequences.
We conducted a cross-sectional study to evaluate the prevalence of HTN and assess its correlates in three poor rural villages in east and southeast Africa as part of the Millennium Villages Project. Findings from this study may help to inform needs for public health interventions/recommendations, to identify potential risk factors and guide prevention strategies, and to set a baseline for monitoring the changing pattern of disease in an area that is hopefully experiencing a socio-economic transition associated with poverty alleviation.3, 4, 5, 6
Materials and methods
The study setting, the Millennium Villages Project,14, 15 is a proof of concept to demonstrate achievability of the Millennium Development Goals and of potential to eliminate extreme poverty in rural Africa. Following UN Millennium Project recommendations, Millennium Villages Project empowers local communities by facilitating practical, proven, low-cost investments leading to increased food production, improved access to health services, to safe water and infrastructure such as transportation, communication and connectivity. Experience gained during this community-led, integrated approach to rural development is anticipated to provide critical insights that can be scaled up in national development strategies. This project is led by the Earth Institute at Columbia University, in partnership with UNDP, Millennium Promise and national as well as regional governments. Since August 2004, Millennium Villages (MV) operate in 10 countries and 14 geographic sites (comprising over 400 000 people) selected to represent most of the agro-ecological zones in SSA. Each MV consists of ∼5000 people (∼1000 households) and each site is a cluster of up to 11 Millennium Villages. In the late 2006, surveys (demographic, socio-economic, health, nutrition, anthropometry) and blood testing were administered to all households to collect baseline data for project implementation and impact evaluation. This study on hypertension was conducted in the MVs of Mwandama (Malawi), Mayange (Rwanda) and Mbola (Tanzania) MVs using these baseline data and blood pressure measurements conducted from January to March 2007.
Study participants were recruited from residents of 300 households in each village. These households were selected using a stratified random sampling procedure that takes into account the wealth status of each household and gender of the household head using information obtained using the baseline demographic and socio-economic surveys. The sampling ensured accurate representation of the makeup of the larger population. All residents of selected households aged 18 and over were recruited for the study. Among 2091 eligible adults, 1485 (71%) were available for participation during the study period and all agreed to participate. The institutional review board of the participating universities as well as local health ministries reviewed and approved the project study protocol. Informed consent was obtained from each village headman and each study participant. All information obtained from surveys was kept confidential and all MV site and research team members received institutional review board training.
Information was collected by data collectors (enumerators) who received a 3-day training and certification by master trainers. Data on socio-demographic characteristics and smoking habits of study participants were collected using standardized questionnaires. Work-related vigorous physical activity was measured using the Global Physical Activity Questionnaire, an instrument developed by the World Health Organization (WHO) for physical activity surveillance in developing countries.16 Information on alcohol intake and dietary habits was collected using a detailed 140-item food frequency questionnaire for a sub-population of study participants (N=665) who were part of a survey conducted to assess the nutritional status of residents of the three MV. Foods were then categorized into food groups (meat, fish, fruits/vegetables, carbohydrates and fat) based on the primary component of the mixed dishes (see Appendix A).
Anthropometric and blood pressure measurements were conducted by health-care professionals who received additional training using validated field protocols. Body mass index (BMI) was calculated as weight in kilograms over height in meters squared (weight in kilograms/height in meters squared). An automated measuring device, OMRON 907 (OMRON, Hoofdorrp, The Netherlands), was used to measure blood pressure. Cuff size was chosen based on mid-upper arm circumference measures, and blood pressure was measured in a sitting position after the participant rested for at least 5 min. Three measurements were taken with intervals of 3 min between consecutive measurements. The average systolic and diastolic blood pressure measurements (SBP and DBP) and the Seventh Report of the Joint National Committee on Prevention, Detection, Evaluation and Treatment of High Blood Pressure criteria (JNC 7 criteria) were used to define HTN status.17 On the basis of this criteria, HTN was present if SBP140 mm Hg and/or DBP90 mm Hg, or if there was a history of HTN diagnosed by a care professional or use of anti-hypertensive medication(s). Pre-HTN was defined by SBP between 120 and 139 mm Hg and/or DBP between 80 and 89 mm Hg. Hypertensive status was further characterized into Stage I (SBP between 140 and 159 mm Hg and/or DBP between 90 and 99 mm Hg) and Stage II (SBP160 mm Hg and/or DBP100 mm Hg).3, 17 Participants with HTN or pre-HTN were referred to local health-care workers for further assessment and treatment.
Surveys and field protocols used in this study, modified from previous surveys, and protocols were designed by social, health and biophysical scientists based at the Earth Institute at Columbia University in New York.
Continuous variables were expressed as mean (s.d.), whereas categorical variables were expressed as percentages. Crude and age-standardized HTN prevalence rates were estimated.18 Logistic regression was used to examine associations of HTN with risk factors. Associations of SBP, DBP and mean arterial blood pressure (MABP=DBP+1/3 (SBP−DBP)) with HTN risk factors, as continuous or categorical variables were evaluated in regression models. Dietary intake frequencies were skewed and thus were log-transformed. Participants were grouped into quartiles based on frequencies of intake of food items in each food group. Logistic regression was used to compare the odds of HTN among participants in each of the upper three quartiles with that of participants in the lowest quartile. In addition, associations of frequencies of intake of food items with SBP, DBP and MABP were evaluated.
Unadjusted and adjusted models were fit to evaluate associations with or without potential confounders (including age, sex, BMI, village, sampling variables as well as dietary variables in dietary intake analysis). We also calculated 95% confidence intervals. Linear tests for trend were used to evaluate dose–response relationships. A P-value cut off of <0.05 was used to evaluate significance. Statistical analysis was conducted using SAS (Cary, NC, USA) and STATA (College Station, TX, USA) software.
The average age of study participants was 41. 4 years, and a comparable number of males (48%) and females (52%) were enroled (Table 1). Overall, 22% of study participants had HTN. The prevalence was comparable to the age-standardized rate of 22.8% (Table 2). Male participants had higher HTN prevalence compared with female participants (24% vs 20%). Most participants with HTN had Stage I HTN (70%). Moreover, 44% of study participants, 49% of males and 40% of females, had pre-HTN. Apparent variation in HTN prevalence (16–27%), but not pre-HTN prevalence, was observed across villages in the different countries.
Several traditional risk factors were also associated with HTN in this population (Table 3). Older age was associated with increased prevalence of HTN (P-value for trend <0.001). Study participants in age ranges 40–59 (1.8-fold), 60–79 (4.5-fold) and 80 (2.3-fold) had greater odds of HTN compared with participants 18–39-years old (all P-values <0.05). Higher BMI was marginally associated with increased prevalence of HTN (P-value for trend=0.07). Although the mean BMI in this population was low (20.9 kg m−2 for males and 21.6 kg m−2 for females), participants with BMI 25 kg m−2 had a 1.6-fold greater odds of HTN prevalence compared with participants with BMI <18 kg m−2 (OR 1.64: 95%CI 0.95–2.85, P-value=0.08). Vigorous work-related physical activity and television ownership were associated with lower and higher prevalence of HTN, respectively (OR 0.66; 95%CI 0.47–0.93 for physical activity and OR 2.99; 95%CI 1.26–7.19 for television ownership). Smoking status, alcohol intake and radio ownership were not associated with HTN prevalence. Study participants for whom information on dietary intake was available were similar to other study participants in terms of HTN prevalence and distribution of risk factors (data not shown).
Dietary intake patterns were associated with prevalence of HTN in this population (Table 4). Higher frequencies of eating high-fat foods and meat were associated with greater odds of HTN (P-value for trend for high-fat foods and meat 0.02 and 0.07, respectively). Participants in the highest quartile of frequency of eating high-fat foods had a twofold higher odds of HTN compared with participants in the first quartile (OR 2.08; 95%CI 1.10–3.92). Participants in the top three quartiles for frequency of consumption of meat had approximately twofold higher odds of HTN compared with participants in the lowest quartile (P-values range from 0.03 to 0.06). Higher frequency of eating fruits/vegetables was associated with lesser odds of HTN (P-value as a continuous variable=0.02 and P-value for trend in quartile regression models=0.10). Participants in the highest quartile for frequency of consuming fruit/vegetables had 54% lower odds of HTN (OR 0.47; 95%CI 0.22–1.00). Frequencies of consuming fish, high-carbohydrate foods and dairy products were not associated with prevalence of HTN in this population. Associations with dietary habit and HTN persisted after adjustment for potential confounding variables (age, sex, BMI, work-related vigorous activity and television ownership).
In analysis evaluating associations of risk factors with components of blood pressure measurements (Table 5), age, BMI and television ownership were positively associated with SBP, DBP and MABP (all P-values <0.05). Work-related vigorous physical activity was inversely associated with SBP, DBP and MABP (all P-values <0.05). Gender was only associated with SBP, whereby males had about 4 mm Hg higher SBP compared with females (P-value <0.05). Among dietary patterns, higher frequency of fruit/vegetables was associated with lower SBP, DBP and MABP (all P-value <0.05).
In this study, we found that both HTN and pre-HTN were highly prevalent in rural SSA (prevalences greater than 20 and 40%, respectively). Male participants had higher HTN and pre-HTN prevalence compared with female participants. Older age, higher BMI, television ownership and less work-related vigorous physical activity were associated with higher prevalence of HTN and higher blood pressure measures. Frequent consumption of meat and high-fat foods was associated with higher HTN prevalence. Frequent intake of fruit and vegetables was associated with lower HTN prevalence and lower blood pressure measures.
Several researchers have investigated the prevalence of HTN in various parts of Africa, including SSA.7, 10, 11, 12, 13, 19, 20, 21 These studies conducted in both rural and urban areas used different criteria for classification of HTN, surveyed relatively older adults compared to those in our study and sometimes focused on specific sub-groups. Although these inconsistencies make comparisons difficult, reported prevalence of HTN by most studies was similar to ours, ranging from 3 to 27% in peri-urban and rural Africa.7 Two studies in rural Cameroon and one in a mixed rural and urban population in Nigeria, both published in the late 1990s, reported ∼15% prevalence of HTN.3, 12, 22 A more recent study conducted among male and female residents of rural and semi-urban villages in Ghana (mean age 55 years) found an overall prevalence of HTN of 29%.3, 4 On the other hand, higher prevalences of HTN have also been reported among rural SSA. For instance, Thorgood et al.3 reported prevalence of 44 and 42% among men and women, respectively, in rural South Africa. Further, a study of a socio-economically advantaged rural area of Tanzania reported prevalence of 41 and 39% among males and females aged 35–54 years and 54 and 61% among males and females aged 55 and older.3, 21, 23 To our knowledge, no study reports HTN prevalence exclusively among poor rural populations in SSA. Our study settings of Mwandama (Malawi), Myange (Rwanda) and Mbola (Tanzania) have been documented to have chronically undernourished populations based on higher than average rates of stunting among children under age 5 (>20%).24 Reports from previous studies in other parts of SSA, together with our study findings, strengthen the evidence for high prevalence rates of HTN and pre-HTN and their importance as potential causes of significant morbidity and mortality in this population.25 Investigations of previously reported relationships between childhood undernutrition and adulthood HTN, and potential mechanisms related to impaired energy expenditure regulation, fat deposition and susceptibility to insulin resistance are needed in this population.26
Associations of traditional risk factors (such as BMI, age and physical activity) with HTN in SSA have also been areas of previous investigations. Our findings of higher rates of HTN and pre-HTN among males compared with females are similar to other previous reports from the region and could potentially be related to differential distribution of risk factors.3, 21, 23 Higher BMI was associated with higher blood pressure and greater prevalence of HTN in rural and urban Africa, similar to our findings.19, 27, 28, 29, 30, 31 In our survey, the prevalence of HTN increased consistently with increasing age in all population sub-groups, similar to the pattern reported in other surveys in Africa.9 Physical activity characterized by work-related vigorous physical activity or physical inactivity estimated indirectly by television ownership, was associated with HTN prevalence and blood pressure measurements in our study. Similar trends have been shown throughout urban SSA, where inactivity is more prevalent than in rural SSA.20, 32, 33 We are not aware of any published study that has demonstrated either of these relationships in rural SSA.
Associations of dietary patterns with risk of HTN have been well documented.34, 35, 36, 37, 38, 39, 40, 41 Although high-fat intake is sometimes associated with higher risk of HTN, substantial evidence suggests that this association is related to uncontrolled confounding by salt intake or other aspects of diet (carbohydrate intake).38, 39 We did not measure urinary sodium so we could not directly evaluate effect of salt intake in our study. Meat intake (particularly red meat) has also been associated with increased blood pressure and risk of HTN in most,34, 35, 41 but not all studies,37 but again residual confounding has not been excluded. Evidence supporting role of fruits and vegetables in lowering blood pressure and risk of HTN is strong and consistent.36, 37, 40, 41 Few studies have investigated associations of dietary factors with HTN in rural SSA. Most of these studies evaluated relationships between salt intake and blood pressure using urinary measurements, instead of dietary-intake surveys.42 Although evidence from these studies supports associations of salt intake with HTN in this population, our understanding of dietary patterns and their relationships with HTN is limited, hampering identification of potential interventions to reduce the burden of HTN in this population. Our findings of an association between lower intake of fruits and vegetables and prevalence of HTN support a role for dietary modification in the prevention of CVD in rural SSA. Further evaluation of the role of dietary habits (both macro- and micronutrient intakes) in the etiology of HTN in SSA using standardized9 tools in other populations is needed.
Prevention of CVD and reduction of its risk factors through lifestyle changes including nutritional and other health interventions in developing countries, particularly in SSA, are receiving timely consideration.23, 43, 44, 45, 46 Murray and colleagues have argued for a ‘frameshift in thinking about priorities and public health strategies for less-developed regions’, calling for a new emphasis on addressing the burden of vascular disease in low-income as well as high-income countries.3, 47, 48 Although infectious/communicable diseases (for example, intestinal parasitosis 20–60% in our study populations, unpublished) as well as related conditions (for example, anaemia prevalence 15–40% in our study populations, unpublished) account for substantial health burden in these populations, our findings of a high prevalence of HTN and risk factor burden in this population, and the anticipated socio-economic change related to poverty alleviation, emphasize the need for CVD prevention research and activities in this setting. Evidence of associations of modifiable risk factors such as BMI, physical activity and dietary habit with HTN in this population presents an opportunity for primary prevention of HTN. Further, the need for active surveillance, and the identification and treatment of hypertensive patients are demonstrated by the decreased awareness of HTN status observed in this study.
Our study has several strengths. It was conducted in sites in rural Africa selected to represent remote and impoverished areas of SSA with chronically undernourished populations. Few studies before have managed to collect information to characterize prevalence and risk factors of HTN in such settings. Data were collected with the same automated devices and standardized protocols across all settings in an effort to minimize measurement errors and bias associated with training multiple research teams in the use of with manual sphygmomanometers. Study participants were selected in a systematic manner to achieve adequate representation of the rural population. We had information on multiple risk factors, including dietary habits. The study was conducted in settings that are parts of a long-term integrated effort to improve socio-economic and health conditions of the rural population. This will provide an opportunity to evaluate the impact of these changes on CVD in general and the prevalence of HTN in particular, and may help bring attention to the need these issues while promoting economic development.
Some limitations of our study deserve mention. In this cross-sectional study, we measured blood pressure among participants at one point in time. This may lead to misclassification of HTN status. Dietary frequencies of fruits and vegetables are higher than previous reports in other populations. This may be related to the interpretation of questions by responders. However, because the dietary data were collected before the blood pressure measurements, we do not expect differential misclassification by categories of blood pressure. In addition, misclassification of other characteristics may bias our findings of associations of risk factors with HTN. However, these misclassifications are likely to be non-differential such that the true associations are even stronger. Residual confounding by unmeasured variables and misclassification of measured confounders is a possibility. We had crude measures of dietary intake (frequency) and physical activity (work related) determined from participants using questionnaires. Finally, our study evaluated southeastern rural Africa and study findings may not be generalizable to other parts of rural Africa where populations may have different disease and risk factor distribution.
In conclusion, we found that HTN and pre-HTN are common in rural poor SSA. We also observed associations of non-modifiable (age) and modifiable (BMI, physical activity and dietary intake) risk factors with HTN in this population. Future studies that evaluate trends in risk factors and CVD morbidity and mortality (longitudinal studies), associations of risk factors with disease in similar populations and barriers to health care for CVD are needed. Information from such studies may enhance the primordial, primary and secondary prevention of HTN and CVD in rural Africa.
We thank Drs Daniel Levy and Frank Sacks for their helpful insights in the design and analysis of this study. We are also indebted to Dr Ranvir Dhillon and to the MVP field teams of Tanzania, Rwanda and Malawi for their help with data collection. We thank Dr Freeman Changamire for additional statistical help. We are grateful for the wealth index and nutrition guidance we received from Drs Jyotsna Puri and Jessica Fanzo, respectively. Finally, we are indebted to the Sarnoff Cardiovascular Research Foundation and the Harvard Medical Office of Enrichment Programs for funding Sarah Stewart de Ramirez's time to carry out this research.
Food group assignments
Fruits and vegetables
Sukuma Wiki (kale)
Sweet potato leaves
Carrots as a whole
Avocado (part of dish)
Ripe banana (large)
Ripe banana (small)
Fish as main dish
Fish as part of mixed dish
Other vegetable oil
Beef as main dish
Beef as part of a mixed dish
Pork as main dish
Pork as part of a mixed dish
Goat as a main dish
Goat as part of a mixed dish
Sheep as a main dish
Sheep as part of a mixed dish
Veal as a main dish
Veal as part of a mixed dish
Chicken large piece
Chicken small piece
Bird (pigeon, quail)
Wild game (rabbit, and so on)
Maize meal ugali
Mixed grain ugali
Maize (boiled or roasted)
Maize meal porridge
Mixed grain porridge
Mixed grain porridge (with milk)
Glass of milk
Milk in food or drink