Summary of findings
We have previously shown that there is sub-optimal control of hypertension in countries at all economic levels [3]. This paper confirms that, while there are gaps along the entire management pathway, the greatest problems in most high- and middle-income countries lie in awareness and control. Once a patient is diagnosed, they are very likely to be treated but not necessarily controlled. In contrast, in most of the low-income countries there tends to be steady attrition from awareness through treatment to control. What this study adds is that the gaps from detection through treatment to control vary among people with different levels of material resources within each country and within groups of countries at similar levels of development. This emphasizes the importance of country-specific information that can inform the national policies needed to implement universal health coverage, now a target in the 2015 Sustainable Development Goals [25].
Some countries, such as Iran or OPT, exhibit relatively equitable outcomes. Others, such as the Philippines, Colombia, India and Pakistan, do not. However, both levels and distribution of outcomes are important; Tanzania and Zimbabwe achieve very poor levels of control, but for everyone. Finally, Canada, Sweden and Poland exhibit distributions that are actually pro-poor.
Limitations
Other international comparative studies of hypertension management outcomes exist [4, 26], but PURE is unique in its geographical reach, both in including countries at all levels of development, and in the detail collected on individuals and the communities they live in. However, it is subject to a number of limitations. In addition to the recognized biases associated with using self-reported data, PURE’s design as a cohort study necessitated inclusion of subjects in communities who could feasibly and affordably be followed up over many years. Hence, the sampling framework in each country was not designed to be nationally representative, and sample sizes were relatively modest in some countries, such as Zimbabwe. Whilst analysis of the PURE household sample has showed good concordance with the national age, sex, urban/rural, education, and mortality profiles of the study countries, suggesting that there was no systemic bias in data collection [10], caution is still warranted when interpreting and extrapolating our findings to national level.
Also, the individual element reported here was limited to subjects aged 35 to 70, as they are at highest risk of cardiovascular disease. This means that our estimates of hypertension prevalence, its detection, treatment and control will be higher than those derived from individual surveys of people at all ages. Some of the most marginalized groups, such as migrant workers, will also have been missed. As with all research on inequalities, it is important to recognize the scope for artefacts where sampling methods exclude those at the extremes of the distribution, although considerable care was taken when recruiting communities to achieve representation and avoid this problem.
The asset-based wealth index, while facilitating comparisons across such a diverse range of countries, is only one approach to measuring inequalities and is subject to certain limitations. We have also adopted a parsimonious approach when estimating concentration indices, avoiding over-adjustment for other variables. Although such variables may be important determinants of inequalities, examinations of their effects would be best studied when limited to a single country or to a few similar ones as any determinant of inequality is, to some extent, context specific, such as membership of a particular ethnic group.
Notwithstanding these limitations, the data presented here provide important new information on the scale and nature of inequalities at each stage of the hypertension management pathway, presenting the data by individual country rather than in groups within the same category of economic development. This provides a much more finely grained picture, highlighting important differences in otherwise similar countries. This is critical, given that health policy development takes place at the national level.
Comparisons with other research
Our findings are broadly consistent with the existing, albeit somewhat fragmentary research, looking primarily at differences in prevalence of hypertension or adherence to treatment, with few international comparative studies and none, to our knowledge, that look at inequalities using comparable measures, such as the concentration index. Thus, we found 92% of Canadians aware of hypertension were being treated, compared with 82% in the Canadian Community Health Survey [27]. While the rates of awareness, treatment and control in Sweden were substantially lower than the other high-income PURE countries, they were similar to those in an earlier study that used data from the WHO MONICA (Multinational MONItoring of trends and determinants in CArdiovascular disease) Project [28]. In Iran, data from the National Surveillance of Risk Factors for NCDs also showed a similarly large attrition in the pathway of hypertensive individuals from detection through treatment to control (33% treated and 12% controlled, respectively, compared with 51 to 18% in our study) [29]. A recent paper combined data from SAGE (WHO Study on global AGEing and adult health) and COURAGE (Collaborative Research on Ageing in Europe), with data on hypertension prevalence by household wealth [5]. Among the countries in common with PURE, similar levels of hypertension awareness and control were reported: in China, 43% were aware and 8% controlled vs. 42 and 8% in our study; in India, 38% aware and 14% controlled vs. 42 and 14% in our study; and in South Africa, 38% aware and 8% controlled vs. 31 and 6% in our study.
Studies examining wealth-related inequalities are much fewer in number, however the pro-rich inequalities observed for detection and control observed in our study were consistent with previous findings in China, India and South Africa [5]; as were other study findings regarding hypertension treatment rates in Brazil [30], China [31], Iran [29] and Tanzania [32]. Finally, a recent Canadian study of adults using national data found no evidence for income-related inequality with respect to hypertension detection or control [33], while we observed a small, albeit significant, pro-poor distribution for detection but not control.
Interpretation
The most important message from this study is that generalizations based on research in a few countries must be cautious. We describe considerable diversity in the patterns observed even in countries at similar levels of development, both at what point on the clinical pathway the problems are greatest and in how detection, treatment and control vary by household wealth. Consequently, further research within individual countries is needed to understand better how characteristics of national health systems influence treatment seeking behavior for hypertension. Studies such as this one can only be a first step in developing appropriate policies.
Nonetheless, the observed gaps and inequalities still highlight the importance of ensuring universal access to affordable, efficient and contextually appropriate methods to detect hypertension in middle age, and when detected, to provide continuous access to skilled health workers and treatment, including medications. As such, good population-level hypertension control is unlikely to be achieved unless approaches to universal health care include mechanisms to ensure access to both primary care and essential drugs for all.
As alluded to above, further research is being taken forward within PURE in the form of detailed health system analyses. Those completed so far reveal complex pictures relating to both demand for treatment, influenced by cost of drugs and by how hypertension – a chronic, asymptomatic, but hazardous condition – is conceptualized, and by supply of trained health professionals and medicines [34, 35]. Other data from the Environmental Profile of a Community’s Health study (EPOCH), nested within PURE, also provide insights. Among the upper-middle income countries, treatment rates were highest in Brazil and Poland, where all pharmacies surveyed in PURE communities had four common CVD medicines in stock and the median cost, as a percentage of ability, to pay was 3 and 1% respectively. On the other hand, the lowest treatment rates were in Malaysia and South Africa, where availability was, respectively, 37 and 33% and the cost as a percentage of ability to pay was 9 and 32% [36]. This has clear consequences for inequalities in treatment and control, particularly as effective blood pressure control typically requires use of two or more medicines [37]. Thus, wealth-related inequality in hypertension control was pro-rich in Malaysia and South Africa, and in many other lower middle- and low-income countries. In contrast, it was relatively equitable in Iran, where access to affordable medicines is good [38].
Some other observations are possible based on knowledge of the health systems involved. In general, the percentage of participants aware of hypertension and on treatment tends to be higher in countries with universal health care, but only if it renders long-term medication affordable. Universal health care should also reduce inequalities where levels of coverage are high. Brazil has implemented wide-ranging healthcare reforms, including provision of common medications for free, with one survey finding that over 90% of users of the public system, the Family Health Strategy, who were aware of hypertension were taking medicine [39]. Yet, while China has three insurance schemes that also provide some degree of coverage for most of the population, medicines are paid for out of pocket and can be expensive [40]. In contrast, though India has yet to implement universal health coverage, it has a thriving domestic generics industry with high levels of availability, but affordability is poor [36].