Inequality in prevalence, awareness, treatment, and control of hypertension in Iran: the analysis of national households’ data

Hypertension (HTN) is a leading noncommunicable disease affecting health and wellbeing of a large population of the world. It is a major risk factor of cardiovascular diseases globally [1]. In a rising trend, the prevalence of HTN will mount to 29% by 2025 globally [2]. The prevalence of HTN in the region of the Middle East and North Africa was 26.3% in 2014 among adults aged ≥ 18 years. Of Iranians aged ≥ 25 years, 30% were living with HTN in 2016 [2].

Universal Health Coverage (UHC) is a key tracer for monitoring countries’ achievements in sustainable development goals [3]. It is at the core of the “Sixth Five-Year Economic, Social, and Cultural Development Plan” of Iran too [4]. UHC prescribes that all individuals should receive health services they need without falling into poverty and services are of high quality. UHC is measured through four indicators consisting of prevalence, awareness, treatment, and control of health condition [5]. A country’s progress towards UHC is not only measured in terms of the level of UHC, but also in terms of equitable distribution of the UHC indicators among socioeconomic groups. Without adequate measurement of inequality, disadvantage populations will continue to be underutilized or receive no treatment at all and consequently remain uncontrolled. As a result, a country may enhance UHC indicators in a long run, yet improvements may remain pro-rich. In this research, we therefore examined whether and to what extent the distribution of UHC indicators prevalence, awareness, treatment, and control of hypertension systematically vary by socioeconomic status and which socioeconomic factors explain the distribution of the outcomes.

Inequality manifests a disparity in the health outcome that can be explained by the socioeconomic factors. It represents the degree of association between rates for a health indicator and the distribution of population among ordered groups based on socioeconomic status (SES) [6]. O’Donnell and colleagues used term ‘living standard’ to denote SES and provide direct, e.g., income and consumption, and indirect proxies, e.g., wealth index developed through a principal component analysis or factor analysis, to measure living standard [7]. Education, occupational status, neighborhood, and health insurance have also been used as SES proxies. At household level, SES includes demographic data such as place of residence to account for rural/urban socioeconomic differences [8, 9]. However, the choice of SES measures is not straightforward as it depends on various factors [8]. Each proxy displays different relationships with various health outcomes and would be addressed by different policies [10].

Hypertension has multiple dimensions that are relevant for examining universal health coverage. Hypertension outcomes are among the most common indicators for measuring health inequality in the world. Hypertension coupled with “life expectancy”, “infant mortality”, “obesity and overweight (body mass index)” and “mortality rate” have been most common indicators of health inequity [11]. As shown in Fig. 1, a direct link can be made between hypertension outcomes and SES.

Inequality driven by income, education, employment, and residence area is vastly reported in the literature. Strong evidence from 283 studies shows that low income, low SES, or low level of education are positively associated with the presence of noncommunicable diseases [12]. Income made the greatest contribution to inequality of late-life health [13]. The inequality of hypertension between poor and rich is increasing in the United States and the United Kingdom [14]. In Iran, a significant socioeconomic inequality in hypertension prevalence was reported disfavoring the poorest group and the distribution of hypertension at the disadvantage of women [15]. However, there is a paucity of evidence regarding the magnitude, direction, and contributing factors of inequality of awareness, treatment, and control in Iran. We measured inequality in the prevalence, awareness, treatment, and control of hypertension in Iran. Aiming to foster progress towards universal coverage of hypertension services, this study aspires to provide evidence for health system improvement in Iran. Our findings are expected to help policymakers take initiatives to improve health system for making hypertension services equitable with the existing (limited) resources.

This research was a cross-sectional study of inequality of PATC in Iran. The study population consisted of a representative sample of Iranians aged ≥ 25 years from urban and rural areas (27,738 participants) across the country in 2016.

In the previous step, C was used to determine the magnitude of inequality in hypertension outcomes. We took a further step to explain inequality by using a decomposition method. The analysis of decomposition relied on explaining distribution of hypertension outcomes by the explanatory factors that may systematically vary with wealth status. The decomposition method demonstrated the contribution of the determinants of wealth to health inequality. The contribution is a product of the sensitivity of hypertension outcomes with respect to an explanatory factor (elasticity) and the degree of wealth-related inequality in that factor (concentration index). We decomposed inequality by age, sex, marital status, area of residence, education, basic health insurance, complementary health insurance, and wealth index to determine contributing factors of PATC of hypertension [7]. Each hypertension outcome (\(y\)) can be described as a linear regression model as below:

where \({x}_{k}\) is a regressor i.e., explanatory factor and \({\beta }_{k}\) is the coefficient of that regressor. Then the concentration index for \(y\) can be written as follows:

where \(\mu\) is the mean of \(y\), \(\overline{x}\) is the mean of \({x}_{k}\), \({C}_{k}\) is the concentration index of \({x}_{k}\), and \(G{C}_{\varepsilon }\) is the generalized concentration index of the error term (\(\varepsilon\)). Eq. 4 shows that \(C\) is equal to a weighted sum of the concentration indices of the \(k\) regressors, where the weight for \({x}_{k}\) is the elasticity of \(y\) with respect to \({x}_{k}\left({\mu }_{k}={\beta }_{k} \frac{{\overline{x} }_{k}}{\mu }\right)\). The residual component, the last term in equations showed the socioeconomic inequality in hypertension outcomes that was not explained by systematic variation in the regressors by socioeconomic status. In a well-specified model this should approach zero. For a bounded binary outcome like the hypertension outcomes, \(C\) can be normalized through Eq. 5 as [25]:

For interpreting results of decomposition analysis, we reported three pieces of information; the elasticities of the hypertension outcomes with respect to covariates, concentration index for each covariate, and total contribution of each covariate to the outcome’s concentration index. Above information derives from Eq. 5. Total contribution of each covariate was a product of elasticity of outcome with respect to a covariate and a concentration index attributed to that covariate. Residual term determined the unexplained variation in socioeconomic-related inequality in the hypertension outcomes.

We tested three statistical models: generalized linear models (GLM), probit models, and logit models (Eq. 3). GLMs refer to models with a response variable that follows exponential family distribution and is a nonlinear function of the covariates. To test if variable selection (specifying covariates for a regression model) was appropriate, we used Linktest. This test if passed, i.e., non-significant, shows that covariates were adequate to explain the outcomes. This test revealed if covariates are misspecified [26].

All analysis was conducted in STATA version 14.1. We used codes and analysis packages developed by O’Donnell et al. that are made publicly available by the World Bank [7].

After excluding 573 (2%) out of 27,738 participants, we considered 27,165 participants for analysis, of whom about 70% were between 25 and 54 years old. We found that the prevalence of hypertension was 30% (95% confidence interval (CI): 29.2–30.6). Fifty-nine percent (58.0–60.3) of hypertensive individuals were aware and 80% (78.9–81.4) of hypertensive & aware individuals were receiving treatment. The control rate of HTN was 39% (37.4–40.7).

Using cross sectional data from the Iran 2016 STEPS study, this research aimed at providing a comprehensive analysis of wealth-related inequality of the UHC indicators for patients with hypertension in Iran. At the country level analysis, hypertension prevalence and control were slightly distributed at the disadvantage of poorest groups. Our study confirmed the minor pro-rich inequality of hypertension prevalence in Iran [15]. Studies report the pro-rich inequality of hypertension prevalence in the low-and middle-income countries such as Kenya [27] and China [28]. Yet, the magnitude of wealth-related inequalities in hypertension tends to be higher in poorer than in richer countries. In the lower income countries, hypertension awareness and treatment rates were observed to be higher among well-off households, while a hypertension control rate tended to be pro-rich regardless of the economic development level of those countries [29].

In a head-to-dead comparison between female and male, we found that the extent of inequity in the hypertension prevalence of women was statistically higher than the extent of inequality among men. One previous study also reported unequal distribution of hypertension at the disadvantage of women in Iran [15].

Decomposition analysis of prevalence showed that education, wealth status, and complementary insurance explained a large share of explained variation in the C of prevalence. Education and wealth status explained 35% and 9% of inequality in prevalence. In a study conducted in Kenya, about 10% of inequity in hypertension prevalence was explained by wealth index, 9% by education and 7% by paid employment. Sociodemographic factors explained 18% of inequality [27]. In total, our model explained 33% of variation in C of hypertension prevalence, in contrast, Gatimu et al. explained 99.7% of inequality in hypertension. However, they included body mass index that explained 47% of variation in the inequality of hypertension.

Most of the pro-rich inequality in hypertension control in Iran was explained by living in the urban area, complementary insurance, wealth index, and education, which confirms the established associations between health state and wealth status and education [27, 30, 31]. The underlying argument that explains such association is that wealth or income is materialized in resources such as nutrition and housing that affects health and wellbeing [32].

The main socioeconomic factors that explain most inequality in hypertension prevalence and control consist of the education, wealth status, and complementary insurance. Among these factors, basic health insurance and complementary insurance are of greater policy relevance as being related to the depth of coverage and degree of risk protection by insurance schemes. Almost the entire population of Iran has access to a basic insurance coverage, this explains why the basic insurance made no contribution to the explained variation in hypertension control. In a sharp contrast, complementary insurance made significant contributions to explaining variation in inequality of hypertension prevalence and control. We therefore call for extending complementary insurance coverage, as a means to extend the coverage of services, to reduce the inequality of hypertension prevalence and control in the country.

Hypertension treatment is a measure of crude coverage. This indicator was measured with a single question assessing if patient takes antihypertension medications. While it is an indicator of general access to hypertension medication, it does not measure if patients receive right treatment or doses. In this case, the inequality in treatment could have been different and factors like area of residence may affect it differently.

Despite hypertension treatment, the control of hypertension is an indicator of much complex outcome and goes far enough to embrace the quality of care as translated to improved access to medical care, continued utilization of services, and accessibility of multiple treatment lines for hypertension, and multimorbidity management. Living in urban areas is much likely to provide quality care and to improve hypertension control given the availability of advanced medical technologies and exhaustive list of medical specialists and care facilities that would improve the likelihood of hypertension control for patients living in these areas. As a result, the area of residence made a significant contribution to inequality of hypertension control. In contrast, the area of residence made no contribution to hypertension prevalence as the prevalence is dependent on healthy lifestyle that can be kept everywhere.

We showed that there was no wealth-related inequality in the awareness of hypertension between the poorest and richest group of Iranian population. However, this finding should be interpreted with caution as the question to measure awareness is prone to a measurement limitation. Multiple questions to measure awareness of hypertension are required to get an accurate picture of patient awareness of hypertension. Some other countries break down awareness to the knowledge of individuals about the impact of hypertension on heart, vessels, and body organs and awareness of healthy lifestyle for hypertension. For example, in Canada surveys with multiple questions have been used to measure each of these conditions [33]. With a such measurement, analysis may show different picture regarding the distribution of outcome between the poorest and richest groups.

This research faced the challenge of causal inferences from cross-sectional survey data. Causal relationship is paramount for developing valid evidence for policymakers to understand what factors need to be considered for reducing inequality. While we claim no causal relationship, we draw attention to some criteria that help maintain causal relationships; we relied on a compelling theoretical model with regard to examining preceding factors for the prevalence and control of hypertension [34].

We considered wealth index as the main living standard measure. However, a key question is what happens to inequality when living standard varies. Since we have no access to income data, we were unable to conduct the sensitivity analysis that tests living standard over income while other variables were held constant [35]. One province did not participate in the STEPS study. Still, we believe that the external validity of our findings is reasonably maintained by the multistage random sampling methods with proportional to size samples. Given this, the validity of inferences might be maintained over variations in persons or times [36].

While the regression model that was used for decomposing inequity in hypertension prevalence passed a Linktest (that examines specification of variables to explain the outcome), the specification of variables can be markedly improved. This residual tends to be close to zero in well-specified models. Other studies have involved lifestyle behaviors to improve the model fit [27], though it might be irrelevant for analysis of inequality as our interest was to measure effects of SES on the health outcomes rather than the effect of behaviors that derive from SES.

This study showed a pro-rich inequality in the prevalence and control of hypertension in Iran. We did not however find significant inequality in the awareness and treatment of hypertension in Iran. We found pro-rich inequality between men and women and between provinces. However, inequality at disadvantage of better-off was also found, for example, inequality in treatment that disfavors richest groups in some provinces such as Tehran. The main socioeconomic factors that explained most inequality in hypertension prevalence and control consist of the education, wealth status, and complementary insurance. Among these factors, basic health insurance and complementary insurance are of greater policy relevance as being related to the depth of service coverage and financial risk protection. Almost the entire population has access to a basic insurance coverage, subsequently this factor has made no contribution to explained variation in hypertension control which is in sharp contrast with pro-rich complementary insurance. We call for expanding complementary insurance coverage to reduce inequality of hypertension prevalence and control in the country. We also call for improved health system’s operations to reduce pro-urban inequality of hypertension control.