Aging and hypertension among the global poor—Panel data evidence from Malawi

Hypertension is a rapidly growing disease burden among older persons in low-income countries (LICs) that is often inadequately diagnosed and treated. Yet, most LIC research on hypertension is based on cross-sectional data that does not allow inferences about the onset or persistence of hypertension, its correlates, and changes in hypertension as individuals become older. The Mature Adults Cohort of the Malawi Longitudinal Study of Families and Health (MLSFH-MAC) is used to provide among the first panel analyses of hypertension for older individuals in a sub-Saharan LIC using blood pressure measurements obtained in 2013 and 2017. We find that high blood pressure is very common among mature adults aged 45+, and hypertension is more prevalent among older as compared to middle-aged respondents. Yet, in panel analyses for 2013–17, we find no increase in the prevalence of hypertension as individuals become older. Hypertension often persists over time, and the onset of hypertension is predicted by factors such as being overweight/obese, or being in poor physical health. Otherwise, however, hypertension has few socioeconomic predictors. There is also no gender differences in the level, onset or persistence in hypertension. While hypertension is associated with several negative health or socioeconomic consequences in longitudinal analyses, cascade-of-care analyses document significant gaps in the diagnosis and treatment of hypertension. Overall, our findings indicate that hypertension and related high cardiovascular risks are widespread, persistent, and often not diagnosed or treated in this rural sub-Saharan population of older individuals. Prevalence, onset and persistence of hypertension are common across all subgroups—including, importantly, both women and men. While age is an important predictor of hypertension risk, even in middle ages 45–55 years, hypertension is already widespread. Hypertension among adults aged 45+ in Malawi is thus more similar to a “generalized epidemic” than in high-income countries where cardiovascular risk has strong socioeconomic gradients.


Introduction
Hypertension is not only widespread in higher-income countries. To the contrary, levels of hypertension are actually often higher in sub-Saharan African (SSA) and low-and middle- observed in cross-sectional data can arise from cohort differences, and such age gradients thus may not reflect changes in hypertension prevalence as individuals become older. Hence, if agebased increases in hypertension are used to create thresholds for targeting care efforts, crosssectional data may misguide clinical decisions. Cross-sectional studies are also limited in their ability to identify the extent to which enhancing awareness and knowledge about blood pressure will improve blood pressure levels over time [12,13]. Ultimately, only analyses of longitudinal data can provide a better understanding of hypertension as a dynamic process that is interrelated with individual aging and socioeconomic changes. Yet, very few such studies have been conducted in SSA, or more broadly in low-income countries (LICs). The purpose of this paper is to provide among the first longitudinal analyses of widespread hypertension among aging global poor using panel data from a low-income country. The individuals that we study had lifecourse experiences that are typical for the aging global poor: They lived most of their lives with per capita incomes of less than $1/day [14]. Fertility was high in these cohorts [15], the status of women was generally low, and income generation through subsistence agriculture and small scale agriculture entailed relatively high levels of physical activity [16]. Older study participants were born when under-5 mortality was almost one out of three [5], and period life expectancies at birth were around 35 years in the 1950s, and remained below 46 years until the early 2000s [5]. All members of our study population have survived sustained poverty, frequent undernutrition, and epidemics. Health at older ages is often poor, and individuals experience accelerated aging with an early onset of morbidity across multiple domains [8,[17][18][19].
Due to the lack of longitudinal studies of hypertension in such low income populations, large gaps remain the literature about some basic aspects of hypertension risk during the lifecourse. Key research questions guiding our analyses for instance include: In this LIC population, do blood pressure and the prevalence of hypertension increases as individuals get older over time? What are socioeconomic predictors of the high blood pressure, or the onset of hypertension? Do patterns of hypertension and their predictors differ between men and women? What are possibly life-course consequences in terms of mortality or physical/mental health of hypertension? Do individuals in our LIC study population have knowledge about the health risk associated with hypertension? Finally, what can analyses of the cascades of care tell us about the diagnosis and treatment (or lack thereof) for individuals affected by hypertension?
While our paper does not directly test hypotheses related to these research questions, our results are useful for targeting hypertension prevention and care efforts to the populations with the most need and towards risk factors with the greatest potential benefit. For example, our results inform decisions about whether hypertension care should be targeted by age and sex, and they help clarify the degree to which intervention efforts should focus on addressing behavioral risk factors or improved medical treatment of hypertension.

Data
Data for this study come from the Mature Adults Cohort of the Malawi Longitudinal Study of Families and Health (MLSFH-MAC), which to our best knowledge is currently the only population-based cohort study of adults in a SSA LICs that collects longitudinal information on blood pressure along with data on hypertension prevalence, incidence, treatment, awareness and knowledge [16,20]. The MLSFH is based in Malawi, one of the poorest countries in the world that is ranked 174 out of 189 countries in terms of the Human Development Index in 2019 (HDI 2019 = .485) [21] and has a per-capita GDP equal to about 5% of the global average [22]. In rural areas, where our study is based and most Malawians (85%) live, about 60% of the rural population was considered poor in 2016/17, thus having a total consumption that does not provide 2,400 calories per day per person [23].
The MLSFH-MAC study population consists of adults aged 45+ years living in rural areas in three districts in Malawi (Mchinji, Balaka, Rumphi). Baseline enrollment in 2012 was 1,266 individuals, and the cohort was followed up in 2013, 2017, and 2018, when each time additional eligible MLSFH respondents reaching age 45 years were enrolled in MLSFH-MAC. An MLSFH-MAC Cohort Profile provides detailed information about sampling procedures, study instruments, attrition/follow-up rates [16], and summary statistics for the study population are provided in S1 and S2 Tables. Life expectancy at birth is currently �63 years, and life expectancy at age 45 is currently �28 years [5]. Healthy life expectancy at age 45 is estimated around 22 years [8], and older adults can expect to live a large proportion of their later years subject to physical limitations on their activities [24].
We focus on MLSFH-MAC respondents with longitudinal measurements of blood pressure in 2013 (baseline) and the 2017 follow-up. The mean age is about 60 years in 2017, with 37% of respondents being age 45-54 years and 14% age 75+ years. 60% of the study population is female, and one quarter is Muslim. The study population is characterized by low levels of formal schooling (30% of respondents have no formal education, and 64% have only primary education). 58% of female and 94% of male respondents were married in 2017, and about half of the respondents resided in a house with a metal/tiled roof. The HIV prevalence for the MLSFH-MAC cohort was 8% in 2017, with HIV+ individuals concentrated at ages 45-49 years. The majority of respondents (66%) have body mass index (BMI) that is within the normal range (18 � BMI � 25) or are underweight (17% have BMI below 18). While only a small fraction of the study population is overweight (12%) or obese (4.7%), women are more likely to fall into these high risk BMI categories.

Measurement of blood pressure in MLSFH-MAC
Blood pressure (BP) in MLSFH-MAC was measured in 2013 and 2017 following the protocol developed by the Health and Retirement Study (HRS) [25]. Three BP measurements on the left arm were taken about 1 minute apart in a sitting position in a chair, using an upper arm automated BP monitor (Omron HEM-780N). BP data is available for 1,229 respondents aged 45+ in 2013 and 1,516 in 2017. Longitudinal BP measurements are available for 1,065 MLSFH-MAC participants. Respondents who measured BP above 160 mmHg systolic and/or 110 mmHg diastolic were given a referral for follow up with a doctor or medical professional, but were not provided any other information about blood pressure. Virtually all respondents (e.g., 99% of participants in 2017) agreed to the BP measurement. Prior to the BP measurement, respondents were asked if they have been diagnosed with hypertension by a doctor or medical personnel in the last two years before the survey, and if they are currently taking medication for reducing BP. Respondents' knowledge about the risks associated with high BP was elicited by asking why it is important for individuals to know their blood pressure, and the importance that respondents attributed to knowing their BP was assessed through a question asking how far (long) they would be willing to walk to a clinic to obtain a (free) BP measurement. Additionally, the 2017 survey included questions on knowledge about BP and associated risk factors, and a set of questions focused on the ability of respondents to recognize symptoms associated with high BP.

Approach
Individuals are defined as pre-hypertensive if 120 � BP sys < 140 or 80 � BP dia < 90, as hypertensive stage 1 if 140 � BP sys < 160 or 90 � BP dia < 100 and hypertensive stage 2 if BP sys � 160 or BP dia � 100, where BP sys denotes systolic and BP dia diastolic blood pressure. Onset of hypertension is defined as having normal blood pressure or pre-hypertension at baseline (2013), and having Stage 1 or 2 hypertension at follow-up in 2017. In 85% of cases, systolic BP determines the hypertension classification. Persistent hypertension is defined as being Stage 1 or Stage 2 hypertensive at both baseline (2013) and follow-up (2017). Persistent Stage 2 hypertension is defined as being Stage 2 hypertensive in both 2013 and 2017.
The "cascade of care" concept [26,27] is used to measure linkage to care, health seeking behavior, and effective treatment for individuals with hypertension. The cascade represents the proportion of individuals who reach each separate stage of care, conditional on being included in the previous stage, and it is useful to identify the stages of the disease care cascade where the management of a disease fails. For hypertension care, the following cascade stages are defined: (1) being hypertensive (47% and 46% of all respondents in 2013 and 2017 respectively), either based on measured BP (90% and 84% of hypertensive cases in 2013 and 2017 respectively) and/or a reported diagnosis (22% or 39% of hypertensive cases in 2013 and 2017 respectively); (2) diagnosis of hypertension by a doctor or medical personnel within 2 years prior to the survey; (3) treatment of high BP with medication, and (4) treatment for high BP and controlled BP (= measured BP does not indicate hypertension at time of survey).
We employ standard regression analysis and focus on the results from the longitudinal analyses of hypertension during 2013-17, which are enabled by the longitudinal cohort information available in MLSFH-MAC. Related cross-sectional or complementary results are reported in the Supplemental Materials.

Ethics approval
This research based on the MLSFH-MAC was approved by the IRB at the University of Protocol #815016) and the College of Medicine (now Kamuzu University of Health Sciences) in Malawi (COMREC Protocol #P.01/12/1165).

Prevalence of hypertension, hypertension incidence and persistent hypertension
At baseline in 2013, 42.5% of mature adults were hypertensive (23.5% Stage 1, 19% Stage 2; (Table 1, Column 1). In multivariate analyses, prevalence of hypertension is higher at older ages, is similar between women and men, varies across MLSFH study regions and is not associated with having obtained formal schooling (S3 Table). Hypertension prevalence is similar between the longitudinal sample ( Table 1, Column 2), which is used for our cohort analyses of changes in hypertension 2013-17, and the cross-sectional sample at baseline in 2013 (Column 1), except for a lower prevalence of Stage 1 and Stage 2 hypertension that is due to selective survival (hypertension predicts mortality, but not other attrition, see below).
Cohort analyses (Table 1, Columns 3-5) show that 10.5% respondents with normal BP at baseline were hypertensive (Stage 1 or 2) by 2017; this proportion increases to 28.5% among those who were pre-hypertensive at baseline. 59% and 83% of respondents with Stage 1 or Stage 2 hypertension at baseline respectively were still hypertensive (Stage 1 or 2) by 2017. 58% of those with Stage 2 hypertension were still Stage 2 hypertensive by 2017. These patterns of prevalence, onset and persistence of hypertension among adults aged 45+ are very similar for men and women, and with one exception (onset of hypertension among respondents classified as pre-hypertensive at baseline), not statistically different (S4 Table).
Several factors are associated with the onset or persistence of hypertension among adults aged 45+ during 2013-17 (Table 2). There is a weak indication that women are more likely to have an onset of hypertension, but otherwise no clear gender difference in the persistence of hypertension over time. Being overweight or obese increases the odds of hypertension onset and the persistence of hypertension, and poor physical health (low SF12 physical score) and nutritional stress (no consumption of fish or meat in last seven days) predict the onset of hypertension. The same factors also predict changes in systolic BP among respondents not hypertensive at baseline, and in some cases also for the overall study population or other subgroups. Being HIV-positive or Muslim protects against increases in systolic BP. Being married is associated with increases in blood pressure and the persistence of hypertension, contrary to common findings that marriage often found to be protective for cardiovascular health [28]. Notably wealth (various indicators), schooling and grip strength are not predictors of the onset/persistence of hypertension or changes in blood pressure (S5 Table). Table 3 shows only modest within-person increases in systolic blood pressure during 2013-17: For women aged 45-54 years there is an annual increase in systolic BP of.94 base points per year of age during 2013-17 (Column 1). This increase is similar to the age gradient in Western, but more than that in some other low-income populations [29]. There is no systematic agerelated increase for men, or women aged 55+ (Columns 2-4). Diastolic blood pressure declines on average as respondents age. Identical conclusions are obtained comparing the adjacent box plots in Fig 2 that show 2013 and 2017 BP among longitudinally followed individuals: there is only minimal change in the distribution of systolic BP as cohort members aged during 2013-17, while diastolic BP declined as the cohort got older.

Blood pressure and aging
In combination, Table 3 and Fig 2 indicate that there is no increase in average hypertension risk in the MLSFH-MAC cohort as cohort members aged four years during 2013-17. While hypertension is highly prevalent and often persistent among MLSFH mature adults (Table 1), overall hypertension was not worsening in the MLSFH-MAC cohort: The distribution of hypertension categories in 2017 is indistinguishable from that in 2013 (p(χ 2 ) = .94) for cohort members who survived and had BP measured at both waves (S6 Table).

Contrasting cohort and cross-sectional patterns in hypertension
In contrast to the relatively "flat" hypertension profile as the cohort got older during 2013-17, there is a strong cross-sectional increase in systolic BP and hypertension with age (Fig 2). For instance, 54.3% adults aged 45-54 had systolic BP of 120 or higher in 2013, and only 22% had systolic BP of 140 or higher. At ages 75+, the respective factions are 78.1% (p < .01) and 52.5% (p < .01). The cross-sectional age gradient for systolic BP is.73 (CI: .95-.87) per year of age in 2013, and.077 (CI: .0062-.13) for diastolic BP (S3 Table). There is also a corresponding crosssectional increase in hypertension with age: 30% of adults aged 45-54 were hypertensive (Stage 1 or 2) in 2013, increasing to 54% (p < .01) at ages 75+. A 10 year higher age increases the odds of being hypertensive in the cross-section by about 42% (CI: 32-54%), with similar increases also pertaining to of stage 2 hypertension (S3 Table). This cross-sectional patterns remains stable in 2017 (S3 Table).
The strong cross-sectional age gradient in BP is in contrast to the flat BP profile documented in our earlier longitudinal analyses (Table 3). Specifically, this divergence of the BP age-gradient between cohort and cross-sectional analyses is suggestive of potential cohort influences on BP and hypertension, that is, the possibility that the higher BP and hypertension prevalence among older adults is due to higher levels of life-course adversity experience by older persons. Such cohort effects are plausible because worse early-life conditions among  older cohorts has been linked to worse health outcomes at older ages (including worse cardiovascular health) [30,31]. This aspect is especially important in countries having experienced socioeconomic development or improvements in health (and thus more pronounced cohort differences in early-life environments) during the 20th century [32,33]. For example, height, a common indicator of early-life nutritional status that was measured in 2017, declines strongly in our study population with age (age-gradient = -.15cm, p < .01, with controls for region and gender).

Blood pressure, hypertension prevalence and body mass index
Our earlier analyses have shown that being overweight/obese or having high BMI is a risk factor factor for the onset and persistence of hypertension during 2013-17 (Table 2), as well as for having elevated blood pressure or being hypertensive (S2 Table). Yet, the overall high prevalence of hypertension among MLSFH Mature Adults contrasts sharply with an overall low prevalence of adiposity (Fig 3). Only a small fraction of respondents are overweight (12% of the total sample in 2017), and an even smaller fraction (4.7%) is obese. Above age 75, where the hypertension prevalence peaks, more than 25% of the respondents are underweight, and only few are obese (2%). There is neither a cross-sectional increase of BMI with age, nor a substantive within-person change in BMI during 2012-17. Age-related increases in BMI are not a factor driving the higher prevalence of hypertension at older ages, further confirming that other factors-such as worse early-life conditions or higher levels of life-course adversity-are contributing to the higher levels of hypertension among older persons.

Consequences of being hypertensive
Hypertension predicts mortality in the MLSFH-MAC cohort, with hypertensive individuals facing about 60% higher odds of dying during 2013-18 (Table 4, Column 1). Importantly for our cohort analyses, hypertension does not predict survey attrition for reasons other than mortality (Column 2). Among survivors, hypertension at baseline is also associated with several adverse outcomes. For example, hypertension predicts declines in subjective health and SF12 physical health score relative to non-hypertensive cohort members during 2013-17, and it predicts (relative) increases in PHQ9 depression score and GAD7 anxiety scores (Columns 3-6). There is also evidence that hypertension reduces effort in common moderately-intensive work activities (e.g., time spent in animal care, firewood preparation, planting, etc.) (

Awareness of hypertension and knowledge of associated health risks
Respondents attribute significant value to knowing their BP, and around 60% report in 2013 a willingness to walk up to 1 hour to obtain a BP measurement, and 1/3 reports a willingness to walk more than 2 hours (Table 5). These reports are consistent with the almost universal consent to BP measurement as part of our study. This willingness does not increase among respondents who are hypertensive, but it is consistent with respondents associating BP with several dimensions of their overall health. However, not all of these associations are accurate.
For instance, about 37% of respondents in 2013 (33% in 2017) think that knowing one's BP may indicate being HIV-positive (Table 5), an association that is not generally established [34,35]. More than 70% of baseline respondent correctly state that BP knowledge can be informative about the with risk of stroke, heart disease, and diabetes. About 80% of respondents think BP is informative about their ability to work and being productive in daily life, in sharp contrast and the mostly asymptomatic nature of high BP (albeit there is some evidence in our study population that stage 2 hypertension leads to reductions in work effort; Table 4). About 46% of baseline respondents associate BP with malaria, an association that is empirically tenuous [36]. Notably, these assessments about the importance of knowing ones BP do not differ by respondent's own hypertension status: hypertensive individuals do not state more accurate reasons of why it is important to know BP, nor do they give more importance to BP knowledge. At follow-up in 2017, respondents report similar associations of blood pressure with other health outcomes as in 2013, albeit overall there is a declining perception that knowledge about BP can be indicative of other health issues.

Cascade of care for hypertension
There remain substantial gaps in the diagnosis and treatment of hypertension among adults aged 45+ in rural Malawi (Fig 4A). Among all respondents who were identified as hypertensive in 2013, only 22% have been diagnosed in the last 2 years by a doctor or medical professional as hypertensive (red bar). Among those who were diagnosed, less then 9% reported in 2013 to be taking treatment for hypertension (green bar); that is for about 61% of respondents their medical needs for disease treatment were not met in 2013. Among those who were on treatment, only 11% had BP levels "under control," i.e., within the normal range. The cascade for 2017 among cohort respondents shows an overall improved cascade (Fig 4B), but there remain substantial gaps in care: only 58% of respondents with measured high levels of BP were not diagnosed in the past as hypertensive, and 54% of the diagnosed do not report taking treatment. However, only 17% of hypertensive respondents have attained a measured BP in the normal range. In a related study we have has documented that prior testing-and specifically the receipt of referral letters to health care providers-helped reduce blood pressure, increase diagnoses and update of medication [12] (similar findings have also been shown in South Africa [13]). The blood pressure screening that was conducted as part of the MLSFH-MAC may therefore have been instrumental in achieving improvements in case between 2013 and 2017 ( Fig 4A+4B).

Summary and discussion
Cardiovascular diseases (CVDs) are rapidly becoming a major source of morbidity and mortality in many Sub-Saharan African (SSA) countries, while disease burdens attributable to communicable diseases are decreasing [8]. Addressing the implications of this shifting disease burden, which occurs together with a rapid aging of SSA populations, is critical for achieving the "grand convergence" in health that has been proposed as an achievable global health goal by 2035 [37]. Importantly, a renewed focus on NCDs-including hypertension as a top priority-will be critical as the Covid-19 pandemic has further revealed the incredible challenges of addressing a dual burden of communicable and non-communicable diseases in LICs [38], as well as the health and social vulnerabilities in LIC population that are caused by an often inadequately treated burden of noncommunicable diseases [39]. The rural subsistence agriculture population that is the focus of this study has low levels of modernization, very limited exposure to "classic" hypertension risk factors such as fast food consumption/unhealthy life styles, and is characterized by high levels of physical activity. Hypertension in such low-income populations has rarely been investigated using longitudinal data, and this is especially the case for older persons who are most at risk of cardiovascular diseases. Drawing on MLSFH-MAC panel data, this study is able to fill an important niche in the literature by highlighting cohort patterns of hypertension and highlighting the dynamics of onset and persistence of hypertension as individuals age during 2013-17.
Key findings of our analyses of blood pressure in this aging poor SSA population pertain to several domains of hypertension research: (1) Prevalence of hypertension and its predictors: Hypertension is highly prevalent among individuals aged 45+, with more than 40% being hypertensive and more than 30% being pre-hypertensive. These high rates of hypertension affect both men and women age 45+, in contrast to strong gender differences that have been documented in higher-income populations (Fig 1A). In our study population, the prevalence of hypertension is much higher than that of being overweight or obese, and while being overweight/obese is a predictor of hypertension, the higher prevalence of hypertension at older ages is not associated with higher levels of adiposity at older ages.
(2) Hypertension and aging-longitudinal changes in blood pressure: (a) While there is a strong cross-sectional age gradients in blood pressure and hypertension, with older individuals having worse cardiovascular health, our longitudinal analyses using MLSFH-MAC data 2013-17 that on average a relatively flat profile of blood pressure or hypertension. As cohort members aged during 2012-17, there is no discernible increase in average blood pressure or the prevalence of hypertension: among longitudinally followed MLSFH-MAC respondents, the distribution of hypertension categories in 2017 is indistinguishable from that in 2013. The aging of the cohort during 2013-17 did therefore not increase average blood-pressure-related cardiovascular risk in this poor study population. (b) This divergence of cross-sectional and longitudinal age-gradient in BP and hypertension is likely due to cohort influences: earlier cohorts had worse early-life environments that, consistent with the Barker hypotheses [40], translated into worse cardiovascular health at older ages. (c) Despite the relatively stable distribution of blood pressure as the MLSFH-MAC cohort aged during 2021-17, blood-pressurerelated cardiovascular risk is dynamic on the individual level. Among respondents with normal BP in 2013, about 10% experienced an onset of hypertension by 2017; among pre-hypertensive respondents in 2013, this rate almost triples and 28.5% experienced an onset of hypertension. (d) Stage 1 or 2 Hypertension, however, are often persistent: 59% of respondents with Stage 1 hypertension in 2013 were still hypertensive in 2017, and this is the case for 84% of respondents with Stage 2 hypertension at baseline in 2013. (e) The overall stable hypertension profile in the MLSFH-MAC cohort during 2013-17 arises because the onset of hypertension is compensated by who experience a decline in blood pressure or a transition from being hypertensive to normal blood pressure or pre-hypertension.
(3) Predictors of onset of hypertension over time: (a) Being overweight/obese is a strong predictor of the onset and persistence of hypertension during 2012-17, and poor physical health (low S12 physical score) and facing nutritional stress (not having eaten fish or meat in last 7 days at baseline) predicts the onset of hypertension. Being HIV-positive or Muslim implies lower odds of experiencing increases in systolic BP. Interestingly, marriage may not be protective for cardiovascular health, in contrast to a common finding in other populations [28]. Being currently married in our study population is associated with increasing blood pressure over time and the persistence of hypertension. (b) Nevertheless, there are few systematic socioeconomic predictors of the onset or persistence of hypertension in this study population aged 45+ during 2013-17; notably neither schooling, wealth, nor grip strength (all measured in 2013) predict the onset/persistence of hypertension during 2013-17 or changes in systolic/ diastolic blood pressure (albeit these factors predict cross-sectional level differences in BP or hypertension prevalence in the expected direction).
(4) Life-course consequences of hypertension: Importantly, hypertension is has measurable consequences among adults aged 45+ in Malawi. Hypertension for example predicts mortality in the MLSFH-MAC population during 2013-17, with hypertensive individuals facing about 60% higher odds of dying. Yet, hypertension is not associated with non-mortality attrition in the MLSFH-MAC. Hypertension at baseline also associated with declines in subjective health, SF12 physical health scores and work efforts during 2013-17 and it is associated with increases in depression and anxiety. These findings thus reaffirm that hypertension is an important public health concern among adults aged 45+ in rural Malawi that is associated with more rapid declines in health as individuals get older.
(5) Cascade of hypertension care: While diagnoses and treatment of hypertension are improving, our analyses of the cascade of care document that there remain substantial gaps in the diagnosis and treatment of hypertension among adults aged 45+ in rural Malawi. In 2013, for example, only 22% have been diagnosed in the last two years by a doctor or medical professional as hypertensive, and among those who were diagnosed, less then 9% reported in 2013 to be taking treatment for hypertension. Yet, this cascade of care substantially improved by 2017. 43% of hypertensive individuals were diagnosed, and 46% of the diagnosed were taking treatments. Yet, despite these improvements in the cascade of care, only 17% of hypertensive respondents have attained a measured BP in the normal range. These gaps in the cascade of care are disconcerting as high blood pressure is one of the main modifiable causes of cardiovascular disease risk with a large body of evidence showing that lowering blood pressure through low-cost and widely available medications significantly reduces CVD mortality [41]. The findings of our cascade-of-care analyses are also important as they help inform possible primary health care interventions that can increase management and treatment of hypertension and related chronic non-communicable diseases. Importantly, prior research [42,43] indicates that the chronicity of these diseases necessitates a restructuring of healthcare to address the multidisciplinary, sustained care including psychosocial support and development of self-management skills, with primary healthcare providers offering promising opportunities for targeted interventions, including for instance by increasing contact of at-risk populations with the health care system (including through screening interventions [7,12,13]), improvements of awareness, equipment (e.g., blood pressure monitors) and access to medications among health care providers [44], updates/enhancements of the primary care guidelines to better reflect the needs of older persons [43] or other at-risk individuals (e.g., by integrating hypertension and HIV management [45]), improved linkage to care using cell phone and related technologies [46], and other improvements in the longitudinal control of hypertension [47,48].
Overall, our findings in this paper indicate that hypertension and related high cardiovascular risks are widespread, persistent and often not diagnosed or treated in this rural sub-Saharan population aged 45+. The prevalence, onset and persistence of hypertension cuts across all subgroups in this population-including, importantly, both women and men. While age is an important predictor of hypertension risk, as has been documented in other LMIC contexts [7], hypertension is already widespread in in middle ages 45-55 years. Hence, hypertension among adults 45+ in Malawi seems to be more similar to a "generalized epidemic" than in high-income countries where cardiovascular risk has strong socioeconomic gradients and untreated hypertension particularly prevalent in vulnerable subsets of older persons [49].
Yet, our findings do not provide a bleak picture about hypertension and cardiovascular disease risk in poor aging populations in SSA or elsewhere. In contrast, a recent body of research has started to provide evidence that relatively inexpensive screenings and referrals for hypertension and are effective approach to reduce the gaps in the cascade of care for hypertension [12,13]. Our findings highlight the urgency of building on this recent evidence and expand information about cardiovascular risk, screening for hypertension and available treatments for elevated blood pressure to the global poor. There are also primary healthcare intervention that can be scaled up at modest costs [42,43]. Such screening efforts or primary healthcare enhancements are likely to have large returns in terms of improving population health among older persons [37].
Some limitations of our study are noteworthy. While the MLSFH-MAC is based on a large, population-based sample of adults 45+ years old living in a rural SSA LIC context, it is not a nationally representative sample. This concern is somewhat ameliorated by the fact that the MLSFH-MAC cohort characteristics closely match that of cross-sectional nationally-representative samples [16] and the fact that roughly 85% of the Malawian population resides in rural areas [50]. Hence, our findings can likely be generalized to other rural areas in Malawi and similar low-income populations in southeastern SSA. The age range covered in our study (age 45+ years) is also comparable to other aging studies in SSA LICs [51], and given life expectancy trends in Malawi and SSA more generally, the study represents adequately the experience of the older population in the region. As in any longitudinal study, attrition is a concern for the MLSFH. However, non-mortality-related attrition has been very low in the MLSFH-MAC as a result of migration tracking and related fieldwork procedures; during 2012-18 the MLSFH-MAC retained 97% of surviving respondents [16]; importantly, non-mortality-related attrition 2013-17 is not associated with hypertension. Detailed analyses of attrition as part of the MLSFH cohort profiles [16,20] have also shown that, even though respondent characteristics often differ significantly between those who were lost to follow-up and those who were reinterviewed, the coefficient estimates for standard family background variables in regressions and probit equations for many health outcomes are not significantly affected by attrition.
Population-based blood pressure measurement in rural older and poor populations in challenging, and measurements of blood pressure on a single session-as is conducted in most population-based health surveys, including the MLSFH-does not provide a clinical diagnoses of hypertension. Yet the MLSFH-MAC adopted the protocol for Health and Retirement Study [25], and through extensive training and carefully-developed fieldwork logistics, the MLSFH-MAC not only achieved very high rates of consent to blood pressure measurements, and data quality of the blood pressure measures is generally high [16]. However, the MLSFH-MAC data lack details about how individuals are affected by hypertension, and among those who are treatment, what specific type of hypertension they receive and/or how frequently they are followed up by the health system.
Given the scarcity of longitudinal data on blood pressure and hypertension in SSA lowincome populations, the limitations of this study are relatively minor. The MLSFH-MAC is one of the rare cohort studies in a SSA LICs that provides detailed longitudinal information about blood pressure, hypertension knowledge and treatment, and our analyses are among the first that document the cohort patterns of hypertension in an aging SSA LIC population.