Study design and data

We designed LASI as a nationally representative longitudinal study to provide detailed, comprehensive longitudinal data on the key health, economic, and social characteristics of India’s older adults. Using the 2011 census as the sampling frame, LASI represents the nation as a whole, as well as each state and union territory, with interviews of adults aged 45 years and older and their spouses of all ages. The prospective protocol of the project can be found in S1 Appendix. For the current analysis, we used the 2017–2019 baseline wave (N = 72,262) and the 2010 pilot wave (N = 1,683). Although much smaller in sample size, the 2010 pilot sample was recruited from 4 states, Karnataka, Kerala, Punjab, and Rajasthan, to represent the diversity of the country. Sampling weights were constructed and used in all descriptive analyses. The details of the study design have been described elsewhere [17,18]. The response rates for the 2017–2019 and 2010 waves were 87.3% and 90.9%, respectively.

For this analysis, we excluded participants younger than age 45 years (N = 6,687 in 2017–2019; N = 225 in 2010), those who did not have at least 1 valid systolic BP (SBP) and diastolic BP (DBP) measurement (N = 5,962 in 2017–2019; N = 169 in 2010), and those missing information on education, household consumption, or health insurance (N = 129 for 2017–2019; N = 3 for 2010). For 2017–2019, we also excluded those with missing information on availability of healthcare facilities in the community (N = 51). After these exclusions, the sample sizes were 59,433 for the 2017–2019 baseline wave and 1,286 for the 2010 pilot wave.

Data were collected via in-home interviews. Relevant to the current analysis, interviewers asked each study participant about age, sex, education, and health insurance. In the analyses, we divided age groups into 45–49 years, 50–54 years, 55–59 years, 60–64 years, 65–69 years, 70–74 years, and 75 years and older. We also conducted the analyses using 10-year age groups and found qualitatively consistent results, suggesting the same direction of age gradients and statistical significance. Education was categorized as no education, completed primary school, and secondary school or more. The binary variable indicating health insurance coverage was based on the question “Are you covered by health insurance?”

The community survey identified access to public and private healthcare facilities. This survey was administered to village/neighborhood leaders and government officials, who listed all available public and private healthcare facilities in a rural village or in an urban ward where the respondents resided. Public healthcare facilities, run mainly by state governments, provide free or very low cost medical services, which would be particularly beneficial for the poorest subpopulation. An extensive private healthcare sector also covers the entire spectrum from outpatient visits to general and specialized hospitals.

Finally, the economic status of households was measured through per capita consumption, which is considered a better measure than income or wealth in developing countries [19]. Interviewers asked the individual most knowledgeable about household finances a detailed set of household consumption questions. Missing values for consumption were imputed (see Lee and colleagues [20] for details). Per capita total household consumption was constructed by summing the itemized responses and dividing them by household size. We then categorized these values into tertiles.

BP measurement and definition of hypertension

The 2017–2019 baseline and the 2010 pilot waves used an identical protocol to measure SBP and DBP levels [17,21]. Trained interviewers measured BP levels 3 times using an Omron automatic BP monitor. We used the mean of the second and third readings. If fewer than 3 readings were available, we used the mean of 2 readings or a single reading if only 1 reading was available. Hypertension was defined as an SBP level of 140 mm Hg or higher, a DBP level of 90 mm Hg or higher, or self-report of hypertension based on the question “Has any health professional ever told you that you have high blood pressure or hypertension?” The aforementioned BP thresholds were chosen based on the Indian guidelines on hypertension [5,7].

We defined awareness of hypertension as the self-report of diagnosed hypertension among all those classified as hypertensive. Participants who reported diagnosed hypertension were asked a follow-up question: “To control your blood pressure or hypertension, are you currently taking any medication?” Those who answered “yes” were categorized as taking antihypertensive medication (treatment). By definition, those who reported taking BP medications were all persons with hypertension. Participants who reported diagnosed hypertension were also asked about dietary restrictions: “In order to control your blood pressure, are you under salt or other diet restrictions?” Among those diagnosed, BP control was defined as an SBP level below 140 mm Hg and a DBP level below 90 mm Hg. We constructed the hypertension cascade [9] only for those with hypertension (as per the aforementioned definition), wherein the denominator was the same for each step.

Analysis

The survey provides sampling weights that match demographic distributions within each state. We used these weights for all descriptive analyses using the 2017–2019 data. For comparisons between the 2010 and 2017–2019 data, we restricted the latter to the same 4 states as the former, and in both cases rescaled the weights such that within each state they add up to the state’s population aged 45 years and over for the closest year available from India’s Registrar General and Census Commissioner and Ministry of Health and Family Welfare (2011 census for the 2010 data [22] and 2016 projections for the 2017–2019 data [23]). India is aging rapidly, and the changes in population age structure might contribute to changes in our analytic variables, in particular BP control. Therefore, we also calculated age-standardized estimates following the World Health Organization standard population, used in conjunction with the survey weights [24].

We first estimated hypertension prevalence among adults aged 45 years and older, using data from the 2017–2019 baseline wave, and then among hypertensive individuals, we estimated hypertension awareness, treatment, and control rates. As large health disparities were previously reported across socioeconomic groups and geographic locations [11,12], we also examined the hypertension cascade by subpopulations.

We then investigated the association between access to healthcare and hypertension awareness, treatment, and control among hypertensive adults. We used unweighted multivariable logistic regression with the following covariates: age, sex, education, per capita consumption, urban/rural residence, having health insurance, and access to public and private healthcare facilities. (The regressors include the demographics that were used in constructing the weights, and therefore the main effect of weighting would be to increase variability. Thus, it is often preferred to not weight regression models [19,25].) As healthcare facilities vary considerably across states [10], we also controlled for state in our logistic regression models.

We estimated 3 separate logistic regression models for awareness, treatment, and control of hypertension. Because only those with diagnosed hypertension were asked about antihypertensive medications, we first investigated how the aforementioned factors were associated with hypertension awareness, and subsequently investigated factors associated with taking medication among those who were aware of their diagnosis. Among treated individuals, we investigated factors associated with BP control. As we hypothesized that access to public healthcare facilities would be particularly important for those with low economic status, we introduced an interaction term between access to healthcare and economic status (i.e., per capita consumption tertile). We also included an interaction between health insurance and economic status to investigate potential moderating effects. To interpret the results from the logistic regressions, we present several statistics. For some relations, we graph the predictive margins, also known as average-adjusted predictions (AAPs) [26], which are the average predicted outcomes with all covariates as observed except the one of interest, which is set to a specific value. Our main tables present the average marginal effects (AMEs), which are the differences between the AAPs of the categories of interest and the reference category, for example, the difference in average predicted awareness between men and women, controlling for the other covariates. The corresponding odds ratios (ORs) are presented in S5 Table. Because the AMEs average interaction terms out, we discuss the ORs and show AAPs for subgroups when discussing interaction effects. All analyses were conducted using Stata statistical software version 14 [27].

Finally, we studied hypertension prevalence using data from the 2017–2019 baseline wave and the 2010 pilot wave, including an investigation of mean levels of SBP and DBP. We pooled the data from the 4 pilot states, Karnataka, Kerala, Punjab, and Rajasthan, for this comparison. Among hypertensive individuals, we compared hypertension awareness, treatment, and control in the 2 time periods of 2010 and 2017–2019, and report t-statistics. We compared the hypertension cascades of the 2 time periods for urban and rural locations separately and by subpopulations.

This study is reported as per the REporting of studies Conducted using Observational Routinely-collected Data (RECORD) guideline (S1 RECORD Checklist).

Ethics statement

LASI obtained approval from the University of Southern California Institutional Review Board (IRB) (UP-CG-14_00005), the Harvard University IRB (CR-16715-10), and the International Institute for Population Sciences IRB (Sr. No. 12/1054), and Health Ministry’s Screening Committee clearance from the Indian Council of Medical Research (F.No.T.21012/07/2012-NCD). Each LASI participant provided written consent for participation.

Patient and public involvement

No patients were involved in setting the research question or the outcome measures, nor were they involved in developing plans for study design or implementation. No patients were asked to advise on the interpretation or writing up of results.

Principal findings

Data from this nationally representative sample of adults aged 45 years and older in India indicate that slightly more than half (55.7%) of hypertensive individuals had been diagnosed, less than 2 in 5 (38.9%) took medication, and less than one-third (31.7%) achieved BP control. Therefore, the highest absolute losses to care occurred at the awareness stage (45.3 percentage points) and treatment stage (16.8 percentage points). This highlights a particular need for interventions that focus on awareness and treatment. Consistent with recent findings [28], we found substantial variation in hypertension management across socioeconomic groups, sexes, and geographic areas, calling for targeting of interventions for those currently lagging behind.

We found that access to healthcare was associated with hypertension awareness and management. Having health insurance was positively associated with hypertension awareness, but this positive relationship did not hold for the poorest tertile of per capita consumption. Health insurance in India typically pays for only inpatient hospitalization and treatment at hospitals [29,30], and its impact seems to remain modest in improving hypertension diagnosis, whereas residing in a community with a public health center increased the probability of hypertensive adults taking antihypertension medication across the economic spectrum.

Moreover, access to a public health center was particularly important for those with low economic status, highlighting the importance of free or very low cost medication provided by a public health center in achieving BP control. Pooled data from 4 states (Karnataka, Kerala, Punjab, and Rajasthan) in 2010 (1,286) and 2017–2019 (N = 7,804) suggested large improvements in hypertension awareness, treatment, and control, except for Kerala, which already had much better hypertension management in 2010. There were secular trends in urbanicity and economic development, which might have contributed to improvements in overall hypertension care, but our subpopulation analysis further suggests that hypertension care has improved among each subpopulation across urbanicity and socioeconomic status categories. Among subpopulations, the biggest improvement was observed among those with low economic status, whereas improvement was modest among those with high economic status.

Strengths and limitations of this study

To our knowledge, this is the largest nationally representative study of the middle-aged and older Indian population (N = 72,262) to date to assess hypertension prevalence, diagnosis, treatment, and control. As our sample represents adults aged 45 years and older both in the country overall and in each state and union territory, we were able to examine the hypertension care cascade at the national and state levels. The large sample size also allowed us to investigate subpopulation-level differences. Evidence showing where in the hypertension care process individuals were lost to care and how this was associated with access to healthcare provides important insights for designing appropriate interventions to improve BP control.

Study limitations include the mostly cross-sectional nature of our data. Although the 2010 pilot sample was drawn from the 2001 census to be representative of the country, its sample size was relatively small, and community-level data on access to public health facilities were not collected for the 2010 pilot study. We were only able to report changes in the hypertension care cascade comparing 4 pilot states, and therefore the observed changes may not be generalizable to other states.

Comparison with other studies

A previous systematic review of studies from 1950 to 2013 suggested that overall estimates of BP control among hypertensive individuals aged 18 years and older were 10.7% for rural Indians and 20.2% for urban Indians [6]. Our results showed major improvements in BP control rates among both rural and urban Indians (30.5% and 33.7% in 2017–2019). Our current rate of BP control, particularly among the economically well-to-do group (36.0%), is approaching that of more developed countries. In a recent study in the United States, the estimated age-adjusted proportion of hypertensive individuals with controlled BP was 43.7% in 2017–2018 [31].

The country as a whole, however, still lags behind more developed countries. For example, Marshall and colleagues [32] reported that hypertension awareness among American and English adults aged 50 years and older was 84% and 76%, respectively, more than 20 percentage points higher than in India. However, the rates in our data resembled those of Chinese adults aged 45 years and older with hypertension, which were 57% for men and 59% for women [33].

Research implications

Over the past decade, health insurance coverage in India has expanded substantially. In 2018, the Government of India launched the world’s largest health insurance program, Pradhan Mantri Jan Arogya Yojana (PM-JAY) (often referred to as Modicare) [34], promising free coverage for half a billion of India’s poorest citizens, which should help to continue the positive trajectory in BP control among hypertensive patients. However, we found that having health insurance was not associated with hypertension diagnosis for the poor. As noted earlier, health insurance in India typically pays for only inpatient hospitalization [29,30], and our finding supports the argument Baru et al. [35] make that even small healthcare expenses exclude the poorest individuals from utilizing healthcare. Thus, barriers to receiving care continue to exist even with health insurance, and further promotion of, and lower barriers to, preventative healthcare will need to be prioritized to reach these individuals.

Access to public health facilities was positively associated with taking antihypertensive medication. Strengthening primary healthcare is another important pillar of PM-JAY [36]. Until recently, almost all government health spending was on building public health facilities on the supply side. In 2005, the largest centrally funded supply-side program, the National Rural Health Mission, was launched, focusing on strengthening public health facilities in rural areas. A companion initiative, the National Urban Health Mission, was launched in 2014 to support urban areas [37]. Growth in public health facilities has been steady, with annual growth rates of 6.7% during 2012–2017 [38], and this expansion seems to have contributed to increased antihypertensive medication usage, leading to improved BP control.

Conclusions

In this study, we observed that BP control has substantially improved in India during the past decade. India now has many fewer undiagnosed hypertension cases and better BP control among hypertensive patients, outcomes that are associated with increased healthcare availability. Especially encouraging is that the largest improvements in BP control occurred among those most disadvantaged in education and economic standing. The Indian government’s launch of the world’s largest health insurance program and the current challenges of the COVID-19 pandemic have sharply delineated India’s longstanding need for universal health coverage. Improving healthcare access should help to continue the positive trajectory in BP control. LASI will be able to follow its baseline wave respondents over time and evaluate how the new insurance program may increase healthcare utilization and affect hypertension awareness and management in the coming years.