Study design and data sources
This study is based on three South Asian countries, namely Bangladesh, India, and Nepal. Recent Demographic and Health Survey (DHS) data for these countries had information on both blood pressure and anthropometry for adult population.
DHS are periodical nationally-representative household surveys which provide data for a wide range of variables on population, health, and nutrition. These surveys usually are conducted by a national implementing agency with technical assistances provided by the DHS program. Surveys are based on two-stage stratified sampling of households – firstly, sampling census enumeration areas are selected using probability proportional to size (PPS) sampling technique through statistics provided by the respective national statistical office, and secondly, households are selected through systematic random sampling from the complete listing of households within a selected enumeration area. From these selected households, subsamples of eligible participants are additionally selected for biomarker testing, which includes height, weight, and blood pressure [14].
DHS surveys receive ethical approval both from the ICF Institutional Review Board and from a country-specific review board. Informed consent is taken from each participant for their participation in the survey and for anthropometric and blood pressure measurements. The DHS program authorises researchers to use relevant datasets for analysis upon submission of a brief research proposal. The data we received for this study were anonymized for protection of privacy, anonymity and confidentiality. More details on survey design, ethical approval, data availability can be found in the DHS program website [https://dhsprogram.com/].
We included those who had consented for measurement of blood pressure, height, and weight, as well as had valid information for those variables. DHS surveys have very high response rate, usually more than 90%. We used the household member record dataset which has one record for every household member, and includes variables for sociodemographic, height, weight and blood pressure measurement.
Anthropometric measurement and BMI classification
In the included DHS surveys, height and weight of the participants were measured by trained personnel using standardized instruments and procedures. BMI was then calculated by dividing body weight (kg) by squared height (m2). We classified participants based into four groups according to the conventional World Health Organization (WHO) classification system [15]: underweight (< 18.5 kg/m2), normal weight (18.5–24.9 kg/m2), overweight (25.0–29.9 kg/m2), and obese (≥30.0 kg/m2). We also classified them according to the proposed cut-offs for South Asian population: underweight (< 18.0 kg/m2), normal weight (18.0–22.9 kg/m2), overweight (23.0–27.4 kg/m2) and obese (≥27.5 kg/m2) [16].
Blood pressure measurement and hypertension
Blood pressure was measured for participants using a standard protocol [17]. In brief, three measurements were taken by trained health technicians, at seating position, at about 10 min intervals. The mean of the second and third measurement was used to record systolic blood pressure and diastolic blood pressure.
We defined hypertension based on the cut-offs provided by the Seventh Report of Joint National Committee on Prevention, Detection, Evaluation, and Treatment of High Blood Pressure (JNC7) guideline 2003 [18] and also the 2017 American College of Cardiology/American Heart Association (2017 ACC/AHA) Guideline for the Prevention, Detection, Evaluation, and Management of High Blood Pressure in Adults [19]. According to the JNC7, an individual was categorised as hypertensive if they had systolic blood pressure ≥ 140 mmHg or diastolic blood pressure ≥ 90 mmHg or reported about antihypertensive medication use during the survey. According to the 2017 ACC/AHA, an individual was categorised as hypertensive if they had systolic blood pressure ≥ 130 mmHg or diastolic blood pressure ≥ 80 mmHg or reported about antihypertensive medication use during the survey.
Other covariates
DHS collected information on wide range variables from the selected households and the respondents from those households using face-to-face interview conducted by trained personnel. DHS collected information on socioeconomic factors like area of residence and household’s wealth index. Place of residence (rural and urban) was defined according to country-specific definitions. For household’s wealth index, each national implementing agency constructed a country-specific index using principal components analysis from data on household assets including durable goods (i.e. bicycles, televisions etc.) and dwelling characteristics (i.e. sanitation, source of drinking water and construction material of house etc.) [14]. This wealth index was then categorized into five groups (i.e. poorest, poorer, middle, richer, and richest) based on the quintile distribution of the sample.
Statistical analyses
All analyses were conducted following the instructions in the DHS guide to analysis [20]. All analyses were performed using Stata v15.1 (Statacorp, College Station, TX, USA). Considering the two-stage stratified cluster sampling in DHS, we applied Stata’s survey estimation procedures (“svy” command) [21].
We looked at the descriptive statistics by sex on sociodemographic, anthropometric, and blood pressure variables using proportions for categorical variables and mean and standard deviation (SD) for continuous variables. We used sampling weights given in each DHS dataset in order to get nationally-representative estimates. 95% confidence intervals (CIs) for prevalence estimates were calculated using a logit transform of the estimate.
To examine the association between BMI and hypertension, we used multiple logistic regressions, separately for each included country. We also estimated the trend by estimating the odds ratios (ORs) with 95% confidence intervals (CIs) of hypertension for each 5 kg/m2 increase in BMI. All these analyses were adjusted for age, sex, are of residence, household’s highest education level, and household’s wealth index, as appropriate. We then examined the trend in subgroups of individuals defined by various characteristics.