Association of socio-demographic characteristics with hypertension awareness, treatment, and control in Bangladesh

Hypertension has been rapidly growing in Bangladesh. However, there has been limited analysis of differences in the hypertension cascade across socio-demographic groups. This study was a secondary analysis of the 2017–18 Bangladesh Demographic and Health Survey. Four dichotomous outcome variables – hypertension prevalence, awareness among those with hypertension, treatment among those aware, and control among those treated – were analyzed. The variation of each outcome was assessed across socio-demographic factors. The association between socio-demographic characteristics and outcomes was analyzed using logistic regression. Less than half of the hypertensive individuals were aware of their hypertension (42.5%), and awareness was higher among those who were older, female, of higher household wealth, and living in urban areas. Among those aware, most were receiving treatment (87.4%), and this proportion was higher in older individuals (89.2% among 65 + , 70.4% among 18–24; p < 0.001). One-third of those treated (33.8%) had their blood pressure controlled, and this was higher among younger and more educated individuals. In multivariable models stratified by rural/urban community, most of the aforementioned trends remained with additional differences between communities. Notably, the association of higher education level with treatment odds differed in rural and urban communities (OR 0.34 [95%CI 0.16, 0.75] in rural; OR 2.83 [95%CI 1.04, 7.73] in urban). Efforts to improve hypertension awareness among individuals who are younger, male, of lower household wealth, and in rural areas are required to address disparities in care. Socio-demographic variations in hypertension awareness, treatment, and control must be considered to design targeted interventions for each step of the cascade.


INTRODUCTION
Hypertension, a major metabolic risk factor for cardiovascular disease and other non-communicable diseases (NCDs), is responsible for more deaths globally (10.8 million, 19.2% of total deaths) than any other risk factor [1,2].The global distribution of hypertension has gradually shifted, due to economic development, urbanization, and shifting patterns of diet and physical activity, to low-and middle-income countries (LMICs) [3].The age-standardized prevalence of hypertension increased by 7.7% in LMICs but decreased by 2.6% in high-income countries between 2000 and 2010 [4].
Hypertension has been growing rapidly particularly in South Asia.The annual growth rate between 2000 and 2019 for population-adjusted disability adjusted life years attributed to hypertension was 0.14% in Sri Lanka, 0.67% in Pakistan, 0.80% in India, 1.39% in Bhutan, 1.42% in Nepal, and 3.03% in Bangladesh [5].This rapid growth of hypertension in Bangladesh is part of the country's greater epidemiological transition.According to the INTERHEART study, an international case-control study spanning 52 countries and consisting of 15,152 cases of acute myocardial infarction and 14,820 age-and sex-matched controls, controls in Bangladesh had the greatest prevalence of multiple harmful risk factors for cardiovascular disease and the lowest prevalence of protective risk factors of all South Asian countries [6].Bangladesh's NCD risk factor profile and hypertension trends are concerning and require efforts to address.
A cascade of care for hypertension can be conceptualized, and such a framework has been recently used to analyze gaps in the treatment of hypertension [7][8][9].The first step in the cascade is for individuals with hypertension to be screened and diagnosed so that they are aware of their condition.Subsequently, those who are aware of their hypertension should be treated, which can involve lifestyle modifications (regarding diet, physical activity, or smoking) and/or anti-hypertension medication.Finally, the goal for individuals on treatment is to lower their blood pressure below hypertension thresholds.Reduction of blood pressure, regardless of the intervention used, has been shown to reduce the risk of major cardiovascular events, including stroke [10,11].
While NCDs started to be included in Bangladesh's Health Nutrition and Population Sector Programme in 2007, Bangladesh has not fully integrated NCD prevention and treatment into its health system, and the majority of NCD care is financed through private out-of-pocket spending [12][13][14].However, the Bangladesh Ministry of Health's most recent "NCD Control Operational Plan 2017-22" aims to implement a community-based primary healthcare model by establishing community level screening and a referral network for hypertension [15].In the context of this goal, it is crucial to understand disparities in hypertension awareness, treatment, and control within Bangladesh to identify vulnerable populations and design targeted policies and interventions that address those disparities in the hypertension cascade of care.This study assesses hypertension awareness, treatment, and control in Bangladesh across socio-demographic groups (by age, sex, household wealth, education, and community type) and analyzes the impact of socio-demographic characteristics on the hypertension cascade separately in both urban and rural areas of the country.

Parent study design
This study was a secondary analysis of data from the 2017-18 Bangladesh Demographic and Health Survey (2017-18 BDHS) as part of the Demographic and Health Surveys (DHS) Program [16].The 2017-18 BDHS was carried out by the Bangladesh National Institute of Population Research and Training with funding from the United States Agency for International Development.The 2017-18 BDHS was a cross-sectional nationally representative survey with data collected from October 2017 to March 2018 and covering household and community characteristics, marriage and sexual activity, fertility, family planning, infant and child mortality, nutrition, women's empowerment, NCDs, and certain biomarkers such as blood pressure and blood glucose.
The target population of the 2017-18 BDHS was all individuals in Bangladesh living in non-institutional dwellings.Two-stage stratified sampling was carried out to have a nationally representative survey population that included individuals from both urban and rural areas in all eight administrative divisions.In the first stage of sampling, 675 enumeration areas/communities, each with an average of 120 households, were randomly selected with probability proportional to the size of the enumeration area.Subsequently, a complete sampling frame of households was generated for each enumeration area, and an average of 30 households were randomly selected from each enumeration area for a total of 20,160 households, of which 19,457 completed interviews and relevant questionnaires (96.5% completion rate) [16].
Questionnaires were developed in English and translated to Bangla following pre-testing of survey instruments in a sample of 100 households.The field staff who conducted the listing and mapping of all households within the sample enumeration areas as well as the field workers who interviewed survey participants were all trained on data collection using the survey tool, and data collection was monitored [16].Following listing of the members of all selected households, one-quarter of the households (seven or eight households per enumeration area; 5063 households total) were randomly selected for blood pressure and blood glucose measurements and completion of additional biomarker questions.In the selected households, all men and women over 18 years of age completed the additional biomarker measurements and questions following consent acquisition.Out of 8013 women and 6691 men who were selected for biomarker measurements, 7428 women (93%) and 5701 men (85%) had their blood pressures measured.Blood pressure was measured using the LIFE SOURCE® UA-767 Plus BP monitor, which has been validated [17].Three separate blood pressure measurements were taken in a single sitting with at least five minutes separating each measurement.The second and third measurements were averaged to obtain each respondent's final blood pressure.Those missing the third blood pressure measurement had the second measurement used as the average (0.65%), and those missing the second and third blood pressure measurements had the first measurement used as the average (0.97%) [16].
All questionnaire data for the 2017-18 BDHS was requested from the DHS Program, which provides this data to individuals for academic purposes upon request and registration.The dataset can be found at https://dhsprogram.com/methodology/survey/survey-display-536.cfm.

Variable definitions
The four separate dichotomous outcome variables analyzed were hypertension, hypertension awareness, hypertension treatment, and hypertension control.These variables were defined for each individual using blood pressure measurement data as well as questions administered as part of the Biomarker Questionnaire.First, consistent with the 2017-18 BDHS and 2020 International Society of Hypertension Global Hypertension Practice Guidelines, individuals were categorized as having hypertension if average systolic blood pressure was ≥ 140 mmHg, average diastolic blood pressure was ≥ 90 mmHg, and/or they were currently taking antihypertension medication [16,18].Awareness was defined as a subset of individuals with hypertension who indicated that they had been told previously by any healthcare worker of their high blood pressure.Treatment was defined as a subset of individuals aware of their hypertension who were currently prescribed anti-hypertension medication.Lastly, control was defined as a subset of individuals receiving treatment who had their systolic and diastolic blood pressure below 140/90 mmHg [8,16,19,20].
Individual-level, household-level, and community-level explanatory variables were considered.Individual-level variables considered in the multivariable model were age (18-24, 25-34, 35-44, 45-54, 55-64, 65 + ), sex (male, female), and education (no education, primary, secondary, college or higher).Household-level wealth index (quintiles) was included as a categorical variable; BDHS calculates this index using scores derived from principal component analysis based on household ownership of various consumer goods and housing characteristics [16].Lastly, community type (urban, rural) was treated as an effect modifier in the multivariable model, and stratified analyses were conducted.

Data analysis
The proportion of individuals with hypertension, proportion aware, proportion receiving treatment, and proportion with controlled blood pressure were calculated nationally and within each administrative division of Bangladesh.For each of the four outcomes of hypertension, awareness, treatment, and control, the prevalences across age, sex, household wealth, education level, and community type were calculated.Each outcome was compared using a bivariate Chi-squared test.
Next, multivariable logistic regression models were used to test the association between the four hypertension outcomes and the explanatory variables.Analyses were stratified based on community type (rural/urban) to analyze the potentially differential impact of individual-and householdlevel socio-demographics on hypertension outcomes.
All bivariate and multivariate analyses were conducted specifying appropriate sampling design and survey weights provided by the 2017-18 BDHS.DHS survey weights were used to ensure nationally representative estimates by adjusting for the oversampling of specific divisions in Bangladesh.Analyses were conducted using the survey package in R Studio 1.3.1073.The code for the analyses can be requested by written request to the corresponding author.

Human subjects research considerations
This study utilized BDHS data provided by the DHS program.These data are de-identified, and any geo-spatial data are displaced to protect the confidentiality of survey respondents.The BDHS methodology and all instruments were approved by institutional review boards at both ICF and the Bangladesh Medical Research Council.All survey respondents provided full written consent.
Across the eight administrative divisions in Bangladesh, Barisal had the highest prevalence of hypertension (32.6%), while Dhaka and Mymensingh had the lowest prevalence (23.6%) (Table 1).Among those with hypertension, the highest prevalence of awareness was in Sylhet (48.9%) and the lowest prevalence was in Rangpur (33.2%).Among those who were aware of their hypertension, Sylhet had the highest proportion of individuals receiving treatment (94.4%), and Rajshahi had the lowest proportion (79.0%).Lastly, among those receiving hypertension treatment, the highest proportion of individuals had reached blood pressure control in Mymensingh (48.7%), while the lowest proportion had done so in Rajshahi (22.1%).

Variation of hypertension cascade across socio-demographic characteristics
Hypertension prevalence was greater among females (28.4%) than males (26.2%) (Table 1; p = 0.010).Furthermore, there was a statistically significant positive association between age and hypertension prevalence (18-24 years old: 8.6% vs 65 + years old: 54.5%; p < 0.001) as well as between household wealth and hypertension prevalence (poorest: 23.7% vs richest: 32.9%; p < 0.001).In contrast, there was an inverse association between education level and hypertension prevalence (no education: 35.2% vs college or higher: 22.5%; p < 0.001); those with higher levels of education had lower prevalence of hypertension.Comparing urban and rural communities, there was no significant difference in hypertension prevalence.
Among individuals with hypertension, those of older age and higher household wealth had higher levels of awareness (Table 1).Likewise, a higher proportion of females (48.8%) were aware of their hypertension compared to the proportion of males (33.5%) (p < 0.001), and a higher proportion of individuals in urban communities (46.8%) were aware of their hypertension compared to those in rural communities (40.8%) (p = 0.005).However, there was no statistically significant difference in levels of awareness across individuals of different education levels.
Restricting to individuals with hypertension who were aware of their status, those of older age had higher levels of treatment (p < 0.001); specifically, 89.2% of individuals 65 + years old were treated compared to 70.4% of individuals 18-24 years old (Table 1).Meanwhile, the proportion of individuals treated did not differ across sex, household wealth, education level, or community type.
Lastly, among those with hypertension who were aware and receiving treatment, older age was associated with lower levels of blood pressure control (Table 1; 18-24 years old: 83.6% vs 65+ years old: 21.7%; p < 0.001).In contrast, higher education level was associated with higher levels of blood pressure control (no education: 27.7% vs college or higher: 42.1%; p = 0.007).There was no statistically significant difference in the proportion of individuals with blood pressure control across sex, household wealth, or community type.

Analysis of hypertension cascade in multivariable models
In multivariable models stratified by community type, those of older age had higher odds of hypertension compared to those 18-24 years old in both rural and urban communities (Table 2).Additionally, females had higher odds of hypertension compared to males in both rural (OR 1.57 [95% CI 1.39, 1.76]) and urban (OR 1.46 [95% CI 1.23, 1.72]) communities.Lastly, in contrast to bivariate associations, individuals with higher levels of education had greater odds of hypertension compared to individual with no education in both rural and urban communities.
Restricting to individuals with hypertension, those of older age had higher odds of awareness compared to those 18-24 years old in both rural and urban communities in multivariable models (Table 2).Similarly, odds of awareness were greater among females compared to males.Regarding household wealth, those in the highest three wealth quintiles in rural communities had higher odds of awareness compared to those in the lowest wealth quintile in rural communities.In contrast in urban communities, odds of awareness were greater only among the highest wealth quintile compared to the lowest wealth quintile.Lastly, only those in rural communities with primary education had higher odds of awareness compared to those with no education (OR 1.381 [95% CI 1.094,1.744]),whereas in urban communities those with at least secondary education had higher odds of awareness compared to those with no education (secondary education: OR 2.27 [95% CI 1.45,3.55];college or higher: OR 2.14 [95% CI 1.25,3.66]).
Considering individuals with hypertension who were aware of their status, the impact of age on odds of treatment in multivariable models differed by community type (Table 2).In rural communities, those over 45 years old had higher odds of treatment compared to those 18-24 years old; meanwhile, odds of treatment did not differ significantly across age groups in urban communities.Conversely, females had higher odds of treatment in urban communities (OR 2.01 [95% CI 1.08,3.75]),whereas there was no statistically significant difference in odds of treatment comparing females to males in rural communities (OR 1.03 [95% CI 0.65, 1.61]).Finally, those with college or higher education in rural communities had lower odds of treatment compared to those with no education (OR 0.34 [95% CI 0.16, 0.75]), whereas individuals with college or higher education in urban communities had higher odds of treatment compared to those with no education (OR 2.83 [95% CI 1.04, 7.73]).
Among individuals with hypertension who were aware and receiving treatment, those of older age had lower odds of blood pressure control compared to those 18-24 years old; this pattern was present in both rural and urban communities (Table 2).Meanwhile, odds of blood pressure control did not differ significantly across sex, household wealth, or education level.

DISCUSSION
While most people with hypertension who were aware of their status reported receiving treatment (87.4%), there were low levels of awareness among those with hypertension (42.5%) and low levels of blood pressure control among those receiving hypertension treatment (33.8%).Overall, only 12.5% of all individuals with hypertension in Bangladesh had adequate blood pressure control.This finding of poor blood pressure control in Bangladesh is a global challenge that is consistent with other multi-national studies [2].For instance, the Prospective Urban Rural Epidemiology studya cohort study involving 17 high, middle, and low-income countriesreported that 46.5% of those with elevated blood pressure were aware of their status, 87.5% of those aware of their status were being treated, and 32.5% of those being treated had their blood pressure controlled.Therefore, out of those with hypertension, only 13.2% had their blood pressure controlled across all countries, and 9.9% had their blood pressure controlled among lower-middle income countriesthe World Bank income classification of Bangladesh [8].
Within Bangladesh, hypertension awareness, treatment, and control varied significantly across socio-demographic characteristics.First, we found age to be positively associated with prevalence of awareness among those with hypertension and prevalence of treatment among those aware but negatively associated with prevalence of control among those treated.This latter finding appears to contrast with previous studies which report either no difference in control across age groups in Bangladesh or even a positive association between age and control across multiple lower middle-income countries [8,21].However, those studies analyze the prevalence of blood pressure control among all individuals with hypertension, not restricted to individuals already receiving treatment.Given that older individuals have higher proportions of awareness among those with hypertension and higher proportions of treatment among those aware, the lower proportion of control among those treated counteracts the previous two trends to result in a similar proportion of control across age groups among all individuals with hypertensionas reported in Iqbal et al. [21].Our   [22].Additionally, a study of hypertensive individuals receiving treatment in Bangladesh, Pakistan, and Sri Lanka which adjusted for medication adherence, diet, physical activity, comorbidities, and several biomarkers found no significant association between age and blood pressure control; this suggests that one of the aforementioned variables helps explain the lower blood pressure control in older individuals [20].However, further analysis is required to identify the exact reasons for this trend in Bangladesh.
Multiple previous studies from South Asia which analyzed awareness, treatment, and control among all individuals with hypertension reported higher proportions of all three when comparing females to males and comparing individuals of higher wealth to those of lower wealth [8,19,[21][22][23].While we found that females had a higher prevalence of awareness compared to males, we found no difference in prevalence of treatment among those aware or control among those treated when comparing males and females.Analogously, we observed a positive association between household wealth and awareness, but there was no association between household wealth and treatment among those aware or control among those treated.These findings show that the difference in hypertension outcomes between males and females and the difference in hypertension outcomes across household wealth primarily come from disparities in awareness.Therefore, future interventions should focus specifically on improving blood pressure screening and hypertension knowledge among males and those of lower household wealth.Once those individuals are made aware of their hypertension status, a prior trial in Bangladesh suggests that individuals will seek care and improve their blood pressure [24].
Notably, the impact of household wealth on odds of awareness differed by community type.After adjusting for all other sociodemographic variables, those of the top three wealth quintiles in rural communities had higher odds of awareness than those in the lowest quintile.Meanwhile in urban communities, only those of the highest wealth quintile had higher odds of awareness than those in the lowest quintile.This differential trend likely results from generally greater accessibility of blood pressure screening for individuals across socio-economic status levels in urban areas compared to those in rural areas.In rural areas, hypertension care is mainly available at sub-district and district level hospitals, which may not be close to all communities, and some private facilities, which require fee for service [25].
Finally, the association of education level with odds of treatment differed by community type.After adjusting for all other socio-demographic variables, individuals with college or higher education in rural communities had lower odds of receiving treatment compared to those with no education, whereas individuals with college or higher education in urban communities had higher odds of receiving treatment compared to those with no education.This opposing trend requires further analysis to understand the factors that limit treatment among more educated individuals in rural areas.

Strengths
Our study uses a nationally representative dataset to quantify the hypertension cascade across multiple socio-demographic characteristics, thus allowing identification of particularly vulnerable groups within Bangladesh that require improved hypertension care.Each step of the cascade was defined as a subset of the previous step, thus reflecting the impact of each socio-demographic characteristic on, for example, receiving hypertension treatment among those already aware of their hypertension.Such analysis avoids the propagation of disparities in awareness to the second and third step in the cascade and clarifies which step of the cascade requires further attention.Finally, our analysis investigates effect modification by community type, clarifying differential patterns across socio-demographic groups in rural and urban areas.

Limitations
While international guidelines recommend that a hypertension diagnosis is confirmed after measuring blood pressure over two or three office visits or conducting home or ambulatory blood pressure monitoring, our analysis based definitions of hypertension on three blood pressure measurements in a single time assessment that were available from a single household visit by survey field staff [18].Thus, this study does not aim to make individual diagnoses of hypertension but rather use the available blood pressure measurements to estimate the sample of individuals with hypertension.Conducting repeat visits for extensive blood pressure monitoring is impractical for population-level surveys.
Additionally, the 2020 International Society of Hypertension Global Hypertension Practice Guidelines' recommendations include immediate drug treatment for those with Grade 2 hypertension (≥ 160/100 mmHg).However, for those with Grade 1 hypertension (140-159/90-99 mmHg), drug treatment is recommended for non-high-risk patients only if elevated blood pressure persists after 3-6 months of lifestyle interventions [18].Despite this caveat, our analysis only regarded those prescribed medication as "treated" due to a lack of data on time since hypertension diagnosis and prior interventions.Nevertheless, the proportion of individuals receiving treatment out of those aware was very high.

CONCLUSION
More than one out of four adults have hypertension in Bangladesh, but only one out of eight individuals with hypertension in Bangladesh had their blood pressure controlled.Interventions are needed to improve hypertension screening among the general population and blood pressure control among those receiving treatment.Among other variations across sociodemographic groups, males and those of lower household wealth had lower hypertension awareness but not treatment or control.Comparison with previous studies reveals the significance of analyzing each step of the cascade independently to elucidate vulnerable groups that need to be targeted and design targeted interventions for each step of the cascade.

What is known about the topic
• There has been rapid growth of the hypertension burden in Bangladesh over the past two decades.Compared to other South Asian countries, Bangladesh has the highest prevalence of risk factors for cardiovascular disease.
• Hypertension awareness, treatment, and control levels are lower in low-income and lower-middle-income countries and, within such countries, lower in rural populations compared to urban populations.

What this study adds
• Among individuals with hypertension in Bangladesh, there were low levels of awareness and low levels of blood pressure control among those receiving hypertension treatment.
However, most individuals who were aware of hypertension were receiving treatment.
• Analysis of each step of the cascade as a proportion of individuals in the previous step revealed a more nuanced picture of socio-demographic groups that need to be targeted by interventions at each step of the cascade.

•
The difference in hypertension outcomes between sexes and across household wealth primarily stemmed from disparities at the level of hypertension awareness.Future policies and interventions should target improving blood pressure screening among males and those of lower household wealth.

Fig. 1
Fig. 1 Proportion of hypertension awareness, treatment, and control out of all individuals with hypertension in Bangladesh.Error bars indicate 95% confidence intervals.

Table 1 .
Prevalence (proportion and 95% confidence interval) of the four outcome variableshypertension prevalence, awareness among those with hypertension, treatment among those aware, and control among those treatedby socio-demographic characteristics and administrative division.Reported p-values are based on bivariate Chi-squared tests.
of each step of the cascade as a proportion of individuals in the previous step presents a more nuanced picture, indicating that interventions should focus on addressing lower hypertension awareness and treatment in younger individuals while addressing lower blood pressure control in older individuals on treatment.Comorbidities, such as diabetes and obesity, are associated with hypertension in Bangladesh and are more prevalent in older age groups; therefore, they may contribute to the trend of lower blood pressure control in older individuals analysis