We established population-based surveillance with repeat cross-sectional surveys to document awareness, treatment, and control trends over time. We conducted a baseline survey during 2018–19 and plan to do a follow-up resurvey in 2023 in the same districts. The survey included nine out of 25 districts where phase-1 of the India Hypertension Control Initiative program was implemented. Two districts in each of the four states, namely Punjab, Madhya Pradesh, Telangana, and Kerala, and one district in Maharashtra, were included in the survey. The study population included adults aged 18–69 with raised blood pressure or already diagnosed hypertension and currently treated for hypertension. We selected non-contiguous districts with a mix of the urban-rural population. Predominantly urban districts were excluded.

Eligible households are structures with a shared kitchen, where family members with at least one individual aged 18–69 years have been residing for more than six months. Eligible individuals included those aged 18–69 years living in the selected household at the survey time. Household-level exclusion criteria included the inability to enumerate the members due to lack of availability, refusal, lack of competent or appropriate respondents to give information, or lack of any eligible members in the household. Individual-level exclusion criteria included lack of availability for the interview after three attempts, refusal to participate, reclassification as ineligible (based on age, period of stay, pregnancy status), and inability to provide consent at the time of the survey.

The outcome was controlled blood pressure among the adults aged 18–69 years who had hypertension. Assuming blood pressure control of 20% at baseline and 30% at follow-up, intra-cluster correlation coefficient as 0.04, design effect of 1.6, with 95% confidence and a power of 90%, we required sample size of 624 adults with hypertension at baseline as well as for the follow-up resurvey. We computed a sample size of 624 adults aged 18–69 years with raised blood pressure (BP)/diagnosed Hypertension (HT) (Fig. 1). We assumed a 25% prevalence of hypertension; hence we aimed to survey four times the number of expected individuals with hypertension. We developed the sampling design to survey people with hypertension to reduce the cost and time per district. A multistage cluster sampling design was adopted for each study district (Fig. 2).

Sample size for households and individuals for community-based survey in nine districts in India, 2018–19.

Stages of Sampling for community survey in nine districts in India, 2018–19.

At the first stage, 39 clusters (villages/wards) were selected by probability proportional to size (household as size) systematic sampling from each district. One Census Enumeration Block (CEB) was selected from each of the selected clusters by probability proportional to size (household as size) sampling at the second stage. In the third stage, all the households in the CEB that met the eligibility criteria were enumerated. A systematic sampling selected a hundred households based on the total eligible households. We enumerated the ages 18–69 years with gender in the selected household at the fourth stage. We selected one adult aged 18–69 years by simple random sampling. Once an adult aged 18–69 years was randomly selected, we measured the blood pressure twice at five-minute intervals. We recorded the history of hypertension and treatment to determine the eligibility for the survey questionnaire and anthropometric measurements.

Overall data collection required approximately 90 days in each district with a team of a supervisor and six data collectors. Each cluster was completed in 2–3 days. The first day was to meet the village/ward leaders, mapping and enumerating the selected primary sampling unit. Most of the individual data collection was completed on the second day, and those unavailable on the second day were surveyed on the third day. The approximate survey cost per district was 22,000 USD.

Data was collected using Open Data Kit (ODK) based forms in Android Tablet. First, we visited the selected household based on the unique household ID of the CEB. The second step was the line listing of adults aged 18–69 years in the selected household with age and gender and then a random selection of one adult aged 18–69 years. The third step was to measure two blood pressure using a digital professional BP monitor (Omron 1300) [8]. We collected information regarding prior diagnosis and treatment of hypertension for the selected adult aged 18–69 years. If both readings were normal, no detailed data were collected; hence data collection for the adult aged 18–69 years with normal blood pressure took less than 10 min.

We collected a detailed questionnaire only among adults aged 18–69 years with raised systolic or diastolic blood pressure in the first or second reading or prior diagnosis and treatment of hypertension. The duration of data collection was approximately 25–30 min. The key variables included socio-demographic characteristics, behavioral risk factors, salt intake, treatment-related information, and counseling. Height and weight were measured using a stadiometer (SECA 213) and a weighing scale (SECA 803).

Current Smoker/smokeless tobacco user was defined as those who smoked/used smokeless tobacco either daily or occasionally at the time of the survey. Alcohol users included respondents who reported using alcohol 30 days prior to the survey. People who added salt before eating or reported eating salty foods were categorized as always/often, occasional, and never used. Eating meals outside the home was classified as people who ate at quick-serve locations always or often versus occasionally or never. The individuals were classified as aware of salt reduction if they considered reducing salt important or had an awareness that high salt cause health problems. Adequate physical activity included either vigorous activities at least 75 min per week or moderate activities at least 150 min per week. We classified the body mass index as per WHO recommendations [9].

Hypertension was defined based on the average of second and third readings. The criteria included systolic blood pressure (SBP) > = 140 or diastolic blood pressure (DBP) > = 90 mmHg or treatment with antihypertensive medications in the previous two weeks. Awareness of Hypertension was defined as individuals who reported being diagnosed by a health provider and satisfied the definition of hypertension mentioned above. The treatment category included people with hypertension who had taken medications in the previous two weeks. Control was defined as SBP < 140 and DBP < 90 mmHg and taking the medication in the previous two weeks. Treatment providers were classified as informal, public, private, and AYUSH (Ayurveda, Yoga, Naturopathy, Unani, Siddha, and Homoeopathy) providers.

We cleaned the data immediately following data collection and provided prompt feedback to the field teams. After adjusting for household non-response at cluster level and individual non-response at district level for each district, the appropriate sampling weights were calculated based on multistage sampling design. We projected the 18–69 years population for 2019 based on the 2011 census using the average exponential growth rate. Using the 2019 projected population of 18–69 years, calibrated sampling weight was calculated for each district’s adults aged 18–69 years. Complex sample weighted analysis was used to generate weighted frequencies and percentages and 95% confidence intervals for each district’s hypertension, awareness, treatment, and control outcome variables. We pooled data from all districts to analyse the factors associated with control among all and only among those aware of the hypertension status. We computed the mean (standard deviation) and median (Inter-quartile range) for age, body mass index, systolic and diastolic blood pressure. We calculated frequencies and proportions for all the categorical socio-demographic, behavioral, and treatment-related variables overall and by gender. We did a chi-square analysis to test the association between BP under control and all the key categorical variables.

We computed unadjusted Prevalence Ratio (PR) and 95% CI for each covariate for BP under control as an outcome using the Log-Binomial model. We used the hierarchical well-formulated model by including all the covariates for BP under control as a complete multivariate Log-Binomial model. Minus two log-likelihood ratio criteria were used to eliminate the covariates one by one based on non-significant highest p-value (least significance) and covariates with p-value > 0.20. However, the variables place of stay (rural/urban), gender and age group, and other covariates with p-value < = 0.20 were retained until the final reduced model. We presented the adjusted prevalence Ratio (APR) with 95% CI for each covariate for the full model and the final reduced model. All analyses were two-tailed, and a P-value of < 0.05 was considered statistically significant. We analyzed the data using the software STATA SE (version 17.0) (StataCorp LLC, Texas, USA).

We obtained written informed consent from the respondents. The Institutional Ethics Committee approved the study. We used unique identifiers for the data collection and analysis.