Hypertension education and adherence in South Africa: a cost-effectiveness analysis of community health workers

The burden of non-communicable diseases (NCDs) in low and middle income countries (LMIC) is very high and compounds the effects of the already high burden of infectious diseases [1]. Of the NCDs, hypertension is a major burden in general and in particular in South Africa with trends showing a growth of 20% for both men and women over the past decade [2]. Of further significance is that adherence to hypertension medication varies significantly from under 10% in the lowest socio-economic status (SES) quintile compared to 80% in the highest quintile [3]. Furthermore, HIV/AIDS can now be regarded as a treatable chronic illness, with the expectation that persons with HIV/AIDS will live longer and lead more active lives. This will increase their exposure to CVD risk including hypertension [4].

However, effective management of patients who are at high risk for NCDs in low-resource settings is challenging due to limited human and financial resources [5]. In response, the South African Department of Health has recently acknowledged the need for improved community based care for NCDs and is currently undergoing a major ‘re-engineering’ of its Primary Health Care (PHC) system [6]. The goal of the restructuring is to ensure the service capacity necessary to manage the ongoing infectious disease challenges in addition to the rising demands from increasing hypertension and other NCDs with limited numbers of doctors and nurses. The restructuring of the PHC will include the training of over 50,000 Community Health Workers (CHWs.) In addition to the traditional training for maternal and child health, HIV/AIDS, and tuberculosis, authorities have considered that training in Cardiovascular Disease management is necessary.

It is thus worthwhile to consider which lessons from prior initiatives with CHWs may be applicable to Cardiovascular Disease interventions. Currently, the utilization of CHWs in many low and middle income countries tends to focus on infectious disease management. Where CHWs have been used to manage NCDs, this has largely been for improving adherence and lifestyle choices, or screening for cancer. Finally, the WHO has also articulated the explicit recommendation for the existence of referral systems as part of managing care and for the appropriate training of health workers to use them [7]. However, to date, there is no evaluation of the cost-effectiveness of CHWs in aiding adherence to medications for cardiovascular disease (CVD).

We therefore propose to evaluate the cost-effectiveness of training CHWs to help with the adherence of stabilized hypertensive patients within the PHC system. If effective, this would serve two purposes. First, patients with the diagnosis of hypertension may benefit from improved control of their hypertension. Second, the staff at primary care facilities will be able to manage newly diagnosed conditions and not be overwhelmed, as patients are cared for in the community rather than depending on frequent PHC visits.

The cost of the intervention, including training, of CHWs visiting the household of a hypertensive patient two times per year is about $8 per patient per year. However, the costs are somewhat offset by reductions in 2% of non-fatal cardiovascular disease events over a life-time. As a result of these offsets, the net annual cost is less than $0.50 per person with only an additional life-time cost of $6.56 per patient. Table 5 lists the life time costs, effects, and incremental cost-effectiveness ratio for the intervention and the current standard. The CHW intervention leads to an incremental cost-effectiveness ratio of $320/DALY averted. This is well below a willingness to pay threshold of $2154 (Afro E region of WHO) or $10,000 (South Africa GDP per capita) for a “very good buy” set by the WHO’s CHOICE program.

The cost-effectiveness of the intervention is somewhat sensitive to the cost of the intervention. One of the largest drivers of variation in costs was the number of households that could be visited by CHWs in regions of differing population density (Table 6). In urban areas the cost-effectiveness of the intervention was $17/DALY averted, compared with the ratio of $1529/DALY averted in deep rural areas where the number of visits per day were estimated to be half of that in an urban area. The cost-effectiveness of the intervention in widely spread ‘average’ rural areas lies between the two at about $772/DALY averted, which is about twice the national average.

When we tested the sensitivity of the CHW intervention to the cost per patient, the intervention remained cost effective (Figure 1). Above our base case annual cost of $8 per patient the intervention remained cost-effective even with a near doubling of the cost at $15 per patient ($1900/DALY). Once the annual cost per patient was below $6.50, the CHW intervention became “cost-saving” (i.e. it both saved costs and increased life-expectancy). The results were also favorable across the full range of estimates of the benefit of the intervention on blood pressure reduction (Figure 2). At the levels below the base case assumption of 4 mmHg the intervention remained cost-effective down to a level of only 2 mmHg reduction where it was just under $2000/DALY averted. Once the blood pressure reduction was above 4.98 mmHg, the CHW intervention became cost-saving. Results were not sensitive to changes in the cost of hospitalizations or CHW salary; that is, the ICERs remained below $2000/DALY averted.

The results were not sensitive to a change in the mortality rate from myocardial infarction of 50% more or less than the baseline rate assumed in the model (ICERs for the CHW intervention compared to the standard of care ranged from ($305-$335/DALY averted. The results were also not sensitive to a cost of an MI ranging 50% above and below the base case value (range of ICERs $316-$326/DALY averted). When we assessed the predictive accuracy of the Framingham risk score, we found the results were somewhat sensitive to whether it over or underestimates risk in South Africans but all results remained well under the willingness to pay of $10,000 per DALY averted. If the risk score overestimates by up to 50% relatively an individual’s risk the ICER for the CHW intervention increases to $2991/DALY averted. If it underestimates the risk by 50%, the ICER for the CHW intervention becomes cost-saving, that is saves lives and costs less than the standard of care.

The results of the probabilistic sensitivity analysis (Figure 3) showed the results to be quite robust. The mean ICER was $223/DALY with 95% of the results remaining in an interval ranging from $0.40-$402/DALY averted. The maximum ICER comparing the CHW intervention to the standard of care was $441/DALY averted. The minimum values showed that the CHW intervention was cost-saving.

Our analyses have shown that an intervention of training CHWs to educate patients about the risk of hypertension and the benefits of life-style changes and adherence to medications would lead to cost-effective prevention of cardiovascular diseases at about $320/DALY averted. Even with a conservative estimate of a 2 mmHg reduction in systolic blood pressure, or almost a doubling of the cost of the intervention, the incremental cost-effectiveness ratio (ICER) remains below $2000/DALY averted. The intervention would lead to approximately 2% reduction in strokes and a 1% reduction in ischemic heart disease. Overall these estimates likely underestimate the benefit of the intervention as we did not include reductions in congestive heart failure and end-stage renal disease that are also affected by increased blood pressure. Another underestimation of the benefits includes the fact that when CHWs who encourage the adherence to one medication or lifestyle change they likely have an impact on other chronic conditions such as dyslipidemia and diabetes and thus could further reduce CVD and other chronic conditions. An emphasis on overall CVD risk has been shown to be more cost-effective [43] than just focusing on blood pressure and we ultimately suggest that CHW training on multiple CVD risk factors together may have a greater impact.

The WHO considers interventions to be cost-effective if they fall below three times gross-national income per capita. The ICERs we report in the main analysis and across the range of sensitivity analyses all fell below this threshold for South Africa as well as for most of the sub-Saharan African countries. Out main finding would be a “very good buy” according to WHO CHOICE criteria of less than one times gross national income (GNI) per capita in South Africa and neighboring countries. These values make this strategy a likely good option for many African and potentially other low and middle income countries.

Overall, there is a paucity of reported research on CHWs for CVD in developing countries; however, there is ample information about their use for infectious disease interventions in these settings. Several studies in developing countries have identified elements that are essential to the success of CHW interventions, such as community involvement in selecting CHWs, integration with existing health systems, adequate and ongoing training, and proper supervision [44, 45]. Monetary payment for CHW services are useful and increasingly favored [46], though non-monetary incentives also help to reduce turnover [47]. These findings, which are similar to those from a review of CHWs in the United States [48], are likely to apply to CHW interventions for NCDs as well. Even a doubling of the CHW salary that we used would make this an attractive intervention.

Our study adds to the limited data available on the cost-effectiveness of task-shifting. Task-shifting from physicians to nurses in managing NCDs such as hypertension has been shown to be effective in several countries [49]. A review of the evidence regarding nurse-led interventions reveals that nurses are effective at the management of diabetes in primary care, outpatient, and community settings [50]; and in reducing hospitalizations, days spent in hospital, multiple readmissions, patient care, and cost-savings, even after factoring in the cost of the intervention [51]. One study in Pakistan [52] suggested that in addition to the education by community members, the general practitioners (GPs) taking care of the patients also had to have an adequate understanding of managing hypertension. In this study, GPs plus community education led to up to 10 mmHg reductions in blood pressure. While we believe the training in South Africa for hypertension management among doctors is adequate, nurse training in chronic diseases in many regions remains deficient. Still, the lack of human resources in LMICs overall, negatively impacts the ability of nurses to manage NCDs and the deployment of CHWs to offset this burden on nurses will be important. On average a hypertensive patient visits the primary health center 12 times a year to pick up medications and sees his or her nurse or physician 3–4 times per year. If the CHW visits can cut out 1–2 of these provider visits, then nurses would be freed up to either take on newly diagnosed patients with NCDs or better manage the already challenging case-load.

Another question is whether the CHWs can manage the workload. There are approximately 7.5 million South Africans with hypertension according to the 2003 WHO definition [53]. With only 42% aware of their diagnosis and only 15% controlled, there remain approximately 2.5 million who could benefit from improved hypertension management in the public sector that are already identified. Upwards of 50,000 CHWs are anticipated to be trained over the long term with the roll-out of primary health care re-engineering in South Africa. In this scenario, at capacity, each CHW takes responsibility for approximately 50 patients with hypertension. This is about one per 6 households that are expected to be managed by the CHW. The CHW could devote two of her expected 15–30 household visits per week (depending whether deep rural or urban) and comfortably accomplish this goal. Further, other care can be provided in this visit to increase economies of scale such as reminders about medications for other chronic conditions. In addition it appears that the intervention would even be cost-effective in remote areas where visits per CHW may be more challenging to achieve due to geography and population density.

The study has several limitations. First, the benefits of the intervention are based on studies outside of South Africa where, we used estimates from both a US study and one based in Taiwan. Other studies though have shown that task-shifting lead to improved treatment adherence for HIV medications and for mental health in South Africa and may be cost-effective [54‐56]. A 20% increase in adherence may lead to a 4 mmHg blood pressure reduction [57, 58]. Additional gains with lifestyle advice and visit reminders could lead to further reductions. A second limitation is the use of WHO CHOICE results for event cost estimates. Until such local South African data can be procured through additional studies, we are reliant on WHO provided estimates. However, our cost estimates are quite conservative and any increase in the cost of CVD events would lead to more favorable cost –effectiveness ratios. Furthermore, the ICERs were quite robust even when a doubling of the cost of the intervention was assessed. A third limitation is our assumption that the benefit would be the same for all patients. However, there may be differences in responsiveness to the intervention based on geographic location, level of prior experience with CHWs in certain areas of South Africa (primarily urban), or income of the patients. Further work is needed to assess if these differences exist. One of the key conclusions is that this CHW intervention addressing both adherence and identification has the potential to be especially important for the lowest quintiles of the population, many of whom live in the rural and deep rural areas of South Africa.

This study found that CHWs could potentially have a significant impact on chronic conditions in South Africa and other middle-income countries leading to improved blood pressure control and reduced strokes and myocardial infarctions. The demands on CHWs in South Africa and elsewhere are going to continue to grow as the health transition unfolds and CVD burden grows while health care professional numbers continue to be less than adequate for the need. Questions regarding how best CHWs can contribute to the overall care and education of the public will remain. However, it appears that training CHWs to improve community knowledge and individual adherence to medications for hypertension (and potentially other chronic non-communicable conditions) may be one valuable use of scarce human resources for one of the leading causes of death and disability in South African adults.

Stephen M Tollman: http://​www.​indepth-network.​org