Inequalities and risk factors analysis in prevalence and management of hypertension in India and Nepal: a national and subnational study

We used recently available data from the National Family Health Survey 2015–16 (NFHS-4) in India and Nepal Demographic and Health Survey 2015–16 (NDHS) in Nepal to analyze the risk factors and the inequalities in prevalence and management of hypertension. Both surveys were nationally representative population-based cross-sectional household surveys that included participants aged 15–49 years. Due to limited data availability population aged 50 years and above were not included in this study. However, inclusion of population aged 15–49 years emphasizes the burden of hypertension in younger age group in both countries. Data were collected from January 2015 to December 2016 in India and June 2016 to January 2017 in Nepal. NFHS-4 selected 616,346 occupied households and interviewed 601,509 households with response rate of 98.0%. NDHS interviewed 11,203 households out of 11,472 occupied households with response rate of 99.0%. Both surveys used multistage stratified cluster sampling design method to select sample. Out of total population of 1.3 billion in India [16], 687,230 women and 100,483 men aged 15–49 years were included in the study. In Nepal, out of total population of 29.3 million [17], 8488 women and 5966 men aged 15–49 years were included. Further details for data collection are provided in Supplementary Method 1&2, p68. Detailed methodology for each survey is available elsewhere [18, 19].

Prevalence, awareness, treatment, and control of hypertension were the main outcome variables in this study. Respondents were classified as being hypertensive if their systolic blood pressure (SBP) readings were at least 140 mm Hg (mmHg) or diastolic blood pressure (DBP) readings were at least 90 mmHg, or if they reported currently taking anti-hypertensive medication [20]. SBP and DBP readings were measured three times within five-minute intervals. Management of hypertension includes awareness, treatment, and control of hypertension. Awareness of hypertension was defined as self-reported case of hypertension previously diagnosed by a doctor or other health professionals [20]. Treatment of hypertension was defined as self-reported use of prescribed antihypertensive medication [20], and control of hypertension was defined as receiving antihypertensive medication and having an average SBP below 140 mmHg or DBP below 90 mmHg [20]. All the measurements were taken by the trained field staffs using standardized blood pressure monitors. Details are provided in Supplementary Method 3 p73.

We selected predictor variables for the analyses based on previous literatures on hypertension and its management [7, 21]. Covariates were stratified into three levels: individual-, household-, and community-levels. Individual-level variables included were; age, sex, educational status, body mass index (BMI), alcohol and tobacco consumption, and marital status. In both countries, participants aged 15–49 years of age were included in the study. Household-level variables included were availability of iodized salt and household wealth quintile. Community-level variables were region of residence, place of residence, and further stratified by province and states, respectively. Details of covariates are provided in Supplementary Method 3 p73.

We used descriptive statistics and frequency distributions to describe participants’ demographic and socio-economic characteristics. Prevalence, awareness, treatment, and control of hypertension were calculated by measuring proportion. To investigate the relationship between covariates and outcome variables, we employed multilevel logistic regression models with random intercept at household-and community-levels [22, 23]. Since NFHS-4 and NDHS used a multi-stage cluster sampling method (described above), the obtained data were nested in multiple categories (for example primary sampling units, states/provinces, and households). Thus, to account for the cluster level effect (within-cluster correlation), multilevel logistic regression was performed in this study [23]. Further, both descriptive and multilevel regression analyses were adjusted to complex survey design using sample weight.

To assess the inequalities in prevalence and management of hypertension, we estimated slope index of inequality (SII), relative index of inequality (RII), concentration index (CIX), and decomposition of concentration index at national and sub-national levels. Inequality analyses were performed based on household wealth index and education level. SII is a weighted measure of inequality that represents the absolute difference in estimated values of a health indicator between the most advantaged and disadvantaged, while taking into consideration all other subgroups using an appropriate regression model. Whereas, RII is a weighted measure of inequality that shows health gradient across multiple subgroups with natural ordering, on a relative scale. It is a ratio of estimated value of a health indicator in the most-advantaged to the most-disadvantaged (or vice-versa) while taking into consideration all other subgroups [24, 25]. Similarly, CIX is a measure of inequality in a health-related variable over distribution of another variable of interest such as household wealth quintile or education. CIX is defined in reference to the concentration curve (CC) which graphs the cumulative proportion of health-related variable against the cumulative proportion of the population ranked by variable of interest. It quantifies the degree of inequality in a specific health variable [9, 26‐28]. Further, we decomposed concentration index to see the contribution of other socio-economic and demographic factors towards the concentration of prevalence and management of hypertension [26, 27, 29]. Detail of statistical analyses is provided in Supplementary Methods 4 p76. Statistical analyses were performed using Stata version 15.1/SE.

Table 1 shows the sample characteristics of the selected population in India and Nepal. The overall prevalence of hypertension in India and Nepal was 11.4% (95% confidence interval [CI] 11.4–11.5) and 19.6% (18.9–20.2) respectively. Among the hypertensive population, 30.3% (30.0–30.6) in India and 40.0% (38.2–41.8) in Nepal were aware of their condition, 26.4% (26.2–26.7) and 20.2% (18.8–21.7), respectively, were receiving treatment. Only 17.8% (17.5–18.0) in India and 10.4% (9.3–11.6) in Nepal under the treatment had their blood pressure controlled. Detailed statistics are provided in Figs. 1, 2, 3, Supplementary Tables 1–2 p1, and Supplementary Figs. 1–10 p35.

Supplementary Table 3 and Supplementary Table 4 shows the risk factors associated with prevalence, awareness, treatment, and control of hypertension in India and Nepal. After adjusting for covariates, the odds of being hypertensive was higher in men compared to women in both countries (Odds ratio [OR] 1.47, 95% CI, 1.41–1.52 for India and 1.96 (1.59–2.44) for Nepal). However, the odds for awareness, treatment, and control were lower in men compared to women in both countries. In India, the odds of being hypertensive in the wealthy and poorest population were similar, however in case of Nepal these results were not statistically significant. But, the OR for treatment of hypertension were 1.58 (1.41–1.77) and 7.86 (1.92–32.11) and for control were 1.30 (1.14–1.48) and 9.48 (1.11–80.81) in the richest quintile in reference to the poorest quintile in India and Nepal, respectively. In both countries, urban residents had higher odds of being aware and having treatment. In India, odds of having the blood pressure controlled was higher in urban residents compared to the rural residents, however in Nepal this result was not statistically significant.

Table 2 provides the summary results of wealth-based inequality in prevalence and management of hypertension in India and Nepal. Prevalence of hypertension was 5.2 (95% CI 4.7–5.7) and 7.1 (2.9–11.4) percentage points higher in affluent populations compared to the disadvantaged ones in India and Nepal. Similarly, inequalities in awareness, treatment, and control were substantially high and concentrated towards the wealthy populations in both countries (Fig. 3). Both in urban and rural residences in India, affluent groups had higher percentage points of hypertension prevalence, awareness, treatment, and control than the disadvantaged groups. Unlike in India, absolute inequalities in prevalence, treatment, and control in Nepal were higher in urban residence compared to rural. Wealthy urban population in Nepal had higher prevalence, awareness, treatment, and control than the poorer and poorest population. However, in rural residence, poorest population had higher prevalence of hypertension than the affluent ones.

At sub-national levels, India’s North region had highest wealth-based absolute inequality in prevalence of hypertension, where hypertension was more prevalent in richer and richest population than their poorer counterparts. The Northeast and Central part of India had highest absolute inequalities in awareness and treatment. Wealth-based inequality in control of hypertension was highest in the South and Northeast regions. (Fig. 3 and Supplementary Table 8 p18). In Nepal, compared to other provinces, Province 3 had substantially high wealth-based inequalities in prevalence 19.4% (95% CI 9.1–29.9), awareness 43.4% (26.1–60.8), treatment 43.4% (27.7–59.1), and control 29.6% (15.9–43.2) of hypertension followed by Province 6 and 7. Absolute inequalities in prevalence and management of hypertension at sub-national levels were significantly higher in Nepal compared to India. Details are provided in Fig. 3, Supplementary Tables 5–12 p11, and Supplementary Fig. 11,12,15 p45 & 17–24 p51).

Prevalence of hypertension in India and Nepal was higher among the population with no formal education than their higher educated counterparts. However, awareness, treatment, and control were higher in educated populations than the uneducated ones in India. In Nepal, awareness and treatment were concentrated in the uneducated population (Supplementary Tables 13–20 p23 and Supplementary Figs. 13–15 p47 & 25–32 p57).

Figure 1 shows the concentration indices for prevalence, awareness, treatment, and control of hypertension in India and Nepal, decomposed by age group, household wealth quintile, body mass index, and education. In both countries, BMI was the major factor that accounted for the pro-rich concentration in prevalence of hypertension. Education didn’t show notable contribution towards the pro-rich concentration in treatment and control of hypertension in both countries.