Data sources
NCDRI study
The NCDRI Study was a population-based, cross-sectional household survey conducted in 2019. For administrative purposes, each state in India is divided into districts and blocks. Each block consists of groups of five to six villages known as gram panchayats. The study took place in two similar taluka/blocks among specified villages in the Patna district of Bihar. One block was used to pilot questionnaires and the second block was used for the full study.
The household questionnaire was adapted from questions previously validated in the World Health Organization (WHO) “STEPS methodology” [19], the NFHS-4 in India in 2015 [11] and the 2011 Census of India [20]. The questionnaires were developed and validated during a pilot study conducted in June 2019 using the SurveyCTO platform (Dobility, Inc; Cambridge, MA, USA). During the pilot study, a series of community engagement exercises were conducted to introduce the study to villages and to seek local permissions from community and village leaders, as well as representatives from the Ministry of Health. Feedback from these community engagement activities and from the interviewers involved in data collection in the pilot study, were used to inform the content and acceptability of the questionnaires, obtaining physical measurements from participants and electronic data collection using mobile tablets. Although measurement of blood pressure, weight and height were acceptable to community leaders and participants, the measurement of waist circumference was not and so this was not recorded.
Interviewers were recruited from selected local villages and were provided with two days of dedicated training in the local language by the study team. The training included instruction on the steps needed to take physical measurements of blood pressure, height and weight and the importance of using the same methods for each participant. Each interviewer was provided with a mobile tablet for data collection, a measuring tape, portable weighing scales and a sphygometer. Arrangements were made for the tablets to be charged on a daily basis by the study team so that no costs for electricity were transferred to the interviewers. All interviews were conducted in local dialects and all participants provided informed consent which was recorded electronically.
Participants were recruited into the main study during July 2019. In the villages selected for participation, community leaders provided maps of the village layout and routes for the interviewers to follow were agreed. The first household to be approached was identified and then every tenth household in the village was systematically approached for participation in the study. One eligible household member was invited to participate and this alternated between women and men to ensure deliberate equal representation. If a male member of the household was identified to be working in the fields near the village, interviewers were permitted to recruit and conduct interviews in the fields to avoid gender being a barrier to participation. Participants were eligible if they were a permanent resident of the household sampled (i.e. not a visitor/guest of the household from another locality), were aged 35–70 years inclusively and were able to provide proof of age (acceptable proofs of age included national identity cards, government issued ration cards, passport or driving licence). National identity cards were widely introduced in India in 2009, and as they are required to access Government benefits and services, > 95% of participants had one and this was the most common proof of age provided.
Information was collected on sociodemographic variables including age, sex, religion (Hindu, Muslim, other), education (no formal schooling, 1–4 years of schooling, 5–8 years or 9 years or more), occupation (housewife/hold, agricultural labour, self-employed/own business, other), household conditions (including access to running water, access to an indoor toilet), lifestyle factors (history of tobacco and alcohol use), and self-reported medical history for common NCDs, including a history of hypertension, and detail of any treatment.
Physical measurements were taken once at the time of the interview and included recorded measurements of height (cm), weight (kg) and blood pressure (mmHg). Every evening the study team collected the tablets from the interviewers so they could be charged and for the completed surveys to be uploaded onto the SurveyCTO platform. To ensure data quality, a random sample of 5% of participating households were re-surveyed within 24 h of interview (the timeframe was stipulated by the local ethics committee) using fifteen key pre-selected survey questions. The level of agreement for the responses obtained in the re-survey was very high and no concerns were identified about the quality of the initial data collection.
Ethical approval for the NCDRI study was obtained from the Local Institutional Ethics Review Board in India (Sigma Research and Consulting) and the Oxford Tropical Research Ethics Committee (OxTREC) at the University of Oxford.
National family health survey
The National Family Health Survey 2019–2020 (NFHS-5) is the fifth national, cross-sectional household survey conducted in India, and is co-ordinated by the International Institute for Population Sciences, Mumbai and data are publically available at state and district level [15]. Details of the sampling frame of all iterations of the NFHS are provided in Additional file: Supplementary Table 1. Full details of the study methodologies employed for the NFHS-5 are provided elsewhere [15]. Briefly, the NFHS-5 Survey employs a two-stage stratified sampling design and information on health and nutrition indicators and measurements of random blood glucose and standardized blood pressure are collected among urban and rural areas within each state. The NFHS-5 was conducted in 38 districts of Bihar between July 9th 2019 to February 2nd 2020 and information was collected from 35,834 households including 42,483 women aged 15–49 and 4897 men aged 15–54 years [21]. The state level reports are more granular than the district reports, with data reported by age and sex.
The 2011 census of India
In India, the national census is conducted every ten years. The 2011 Census of India has been used as baseline data to project the age compositions of the Indian population to the year 2036 and this data was published in the Population Projections for India and States 2011–2036 Report [10]. The population projections for 21 States and one Union Territory in India were calculated using the component method, which applies the assumptions for fertility, mortality, life expectancy and sex ratio at birth.
Statistical analysis
Hypertension was defined as a systolic blood pressure (SBP) ≥ 140 mmHg or a diastolic blood pressure (DBP) ≥ 90 mmHg at baseline or participants reported receiving blood pressure-lowering medication. Body mass index (BMI) was calculated as weight in kilograms divided by the square of height in meters, and was categorised as: underweight (< 18.5 kg/m2), normal weight (18.5–25.0 kg/m2) and overweight (≥ 25.0 kg/m2). These definitions for hypertension and assessment of BMI are the same as those used in the NFHS-5 Survey [21].
Categorical variables were reported as frequencies and proportions. Continuous variables were reported as means and standard deviations (SD). The prevalence estimates of hypertension and BMI in the NCDRI Study were calculated overall and by age (adjusted for sex) and sex (adjusted for age). The age-standardized prevalence estimates from the NCDRI Study and the Bihar State NFHS-5 Survey were only estimated for overlapping age groups; hypertension estimates were calculated for women aged 35–49 and men aged 35–54 years and BMI estimates were calculated for women and men aged 40–49 years. The age-standardized prevalence estimates were generated by weighting these estimates to the age distribution of the WHO standard population [22]. To estimate the potential future population burden of NCDs in Bihar, the age and sex-specific estimates of hypertension and BMI were applied to projected population estimates for 2021 and 2036 using data from the 2011 Census of India [10]. All analyses were performed using R (version 4.1.1).