Study design
This descriptive multi-site cross-sectional study was conducted among 3675 adults aged 18 years and above from seven communities in four countries in Tanzania, Uganda, Kenya, and Nigeria. The study settings comprised rural areas in Nigeria (Olorunda Abaa in Oyo state, Ogane-Uge in Kogi state, and Okpok Ikpa in Cross River State); semi-urban (Ikire town in Osun state Nigeria and Ukonga ward in Dar es Salam in Tanzania); and urban communities (Soroti municipality in Uganda and Viwandani slum of Nairobi in Kenya).
Study populations and sampling procedures
Participants were recruited using a representative sample from each community. In Kenya, participants were randomly selected from the Nairobi Urban Health and Demographic Surveillance System (NUHDSS) registry. A list of potential participants was collected from the NUHDSS and the inclusion and exclusion criteria were applied. Finally, we randomly selected 300 participants from the list of potential participants.
Similarly, participants from Nigerian sites were selected using random sampling techniques. In Okpok Ikpa site, a house-to-house survey of adults was performed in the rural areas of Okpok Ikpa, Odukpani LGA, Cross River State, south-south region of Nigeria. In Ogane Uge site in Nigeria, we selected a random sample of households from rural areas of Ogane-Uge, Oganenigwu, Dekina L.G.A, all in Kogi State. In Olorunda Abaa of Oyo State, participants were selected from a random sample of households. In Ikire site, we conducted a household survey among adults in Ikire, Irewole LGA, Osun state, a semi-urban community in South West Nigeria.
In Uganda, participants were sampled from all divisions of Soroti municipality. Starting at a landmark such as church/mosque or school, and selected every third household to the right of the main entrance to the landmark. The first sampled household was the initiator of the sample in that area and sequentially sampled every third household on the right of the main entrance of the previous household until the sample was achieved.
In Tanzania, participants were selected by simple random sampling from a list of households of Ukonga ward, Ilala municipal area, Dar es Salaam region in the Dar es Salaam Health and Demographic Surveillance System (HDSS).
Participants were adults aged 18 years or greater residing in the area of study. Pregnant women and individuals with physical impairments preventing measurement of blood pressure or body weight and height were excluded. A resident was defined as someone who has stayed within the area for at least 3 months and is expecting to stay for another 3 months. If there were more than one eligible participant in a household, we used the Kish method[10] to select one of them. In the event that a selected individual was not home at the time of the visit, 3 attempts on separate days, including evenings on week days and weekends were made before sampling another eligible household member. If a selected household had no eligible individual, we visited the immediate neighboring household until an eligible participant was found.
Data collection procedures
Trained research assistants conducted data collection using a structured standardized questionnaire to collect information on socio-demographic and economic (asset ownership) characteristics of the participants. We also collected information on common risk factors for non-communicable diseases (NCDs) including tobacco and alcohol use, history of diagnosis and/or management of cardiovascular disease and its risk factors (hypertension, diabetes mellitus, dyslipidemia), and a list of current medications.
Measurements
Blood Pressure
Blood pressure was measured on the left upper arm using a digital blood pressure machine, with patient in a seated position after 3-5 minutes of rest. Three systolic blood pressure (SBP) and diastolic blood pressure (DBP) measurements were taken at least five minutes apart using portable sphygmomanometers (OMRON-Healthcare-Co HEM-7211-E-Model-M6; Kyoto, Japan). The mean of the second and third readings was used in this analysis. Hypertension was defined as average SBP ≥140mmHg and/or DBP ≥90mmHg and/or self-report of previous diagnosis with or without current treatment with antihypertensive medications in accordance with the Seventh Report of the Joint National Committee on Prevention, Detection, Evaluation and Treatment of High Blood Pressure.[11] Treatment of hypertension was defined as current or prior (those whose medication ran-out) use of antihypertensive medication. Among those treated, control was defined as having systolic blood pressure below 140 mmHg and diastolic blood pressure below 90 mmHg. We intentionally avoided using the 2017 American Heart Association (AHA) and American College of Cardiology (ACC) definition of hypertension as it would significantly increase the number considered hypertensive and the current national guidelines in these countries have not yet incorporated these new lower thresholds. We defined hypertension awareness as a self-report of ever diagnosis of hypertension by a healthcare provider.
Anthropometric measurements
Weight and height were taken with the participant wearing light clothing and with no shoes using the standardized scales (seca 762, Hanover, USA) and height using a roll-up measuring stadiometers (seca 206, Hanover, USA). Body weight was measured and recorded to the nearest 0·1kg and height was measured and recorded to the nearest 0·1 cm. Body mass index (BMI) was then calculated as body weight per height squared (kg/m2). Overweight was defined as BMI ≥25kg/m2 but <30kg/m2 and obesity as BMI ≥30kg/m2.[12] Waist and hip circumferences were measured to the nearest 0.1 cm (using seca tape measure) using the standard methods.[13]
Socioeconomic Characteristics
Data on ownership of household items such as radio, television, telephone, sofa, refrigerator, bicycle, car, and having working electricity; house ownership, construction materials (floor, walls and roofing materials); source of fuel for cooking and lighting; source of water supply for home use and drinking; and sanitation facility were also collected.
Other covariates
Sociodemographic information including age, gender, marital status, education level, and occupation were collected. Marital status was grouped into never married, married or living together, divorced or separated, and widowed. Educational level attainment was categorized according to the highest level reached in primary school, secondary school, or tertiary education (including vocational training). We collected occupation data in pre-coded categories: self-employed, government employee, private employer, and unemployed.
Statistical analyses
We estimated the prevalence of hypertension for all participants and by site, and hypertension awareness, treatment, and control of hypertension among those with a prior diagnosis of hypertension. We used principle component analysis to generate an assets ownership index score based on household utilities and assets to derive composite measures with highest discriminatory capabilities.[14] Participants were divided into quintiles of these scores (poorest, poor, fair, rich, and richest). [15]
We examined association between prevalence, awareness, treatment, and control of hypertension with a-prior set of covariates: age (continuous), gender (men and women), employment (unemployed, government, and private), health insurance (yes or no), education (primary school and below, secondary school, and tertiary education), alcohol use (yes or no), current smoker (yes or no), and diabetes (yes or no).
We used hierarchical models with a logit link function and communities (sites) as random intercepts, to identify both individual and community characteristics independently associated with mean systolic blood pressure after adjusting for age, marital status, highest level of education attained, smoking, alcohol use (Model 1); employment status, body mass index (Model 2), and additionally adjusted for health insurance (Model 3). The models with prevalence as outcome are for all participants; those of awareness are among those with hypertension; those for treatment are among those who were aware; and those for control are for those on treatment.
We computed standardized rates by employing direct standardization to the World Health Organization Standard Population age-structure for the period 2000-2025[16] using 10-year age bands. These allows for the calculation of standardized rates that are comparable across regions and time.[16] The overall rates by site indicate the rate that would result if all populations had the same age distribution.[17]
We used the Globorisk score[18] to predict the 10-year risk of a first fatal and non-fatal cardiovascular disease (CVD) (stroke and coronary heart disease) for adults aged 40 or greater for each site. The office-based Globorisk score is a country-specific CVD risk prediction model that estimates the 10-year risk of a first fatal and non-fatal stroke and ischemic heart disease, based on age (years), gender, systolic blood pressure (mm Hg), body mass index (BMI), and smoking status (yes/no).[19] We considered two different thresholds to define high risk for future cardiovascular disease: >20% risk scores on the basis of the WHO guidelines[20] and 30% as the threshold on the basis of the global NCD target.[21] Participants with a score <7·5% were considered low-risk. We used boxplots to compare predicted CVD risks for each site for men and women who were categorized as low-risk or high-risk. All analyses were complete case analyses performed using Stata version 15·1 (Stata Corp., TX, USA).