Preventing cardiovascular disease in rural South Africa
Date
2018
Authors
Maredza, Mandy
Journal Title
Journal ISSN
Volume Title
Publisher
Abstract
Background: The epidemiological, nutritional and demographic transitions occurring in developing countries, have resulted in an unprecedented rise in noncommunicable diseases, in particular, cardiovascular disease (CVD). Consequently, CVD are contributing significantly to disease burden in populations where infectious diseases and maternal and perinatal conditions still persist. Rural areas which are least equipped to deal with this burden are no exception; recent evidence points towards a similar trend of increased CVD, particularly stroke. Understanding the disease burden and cost-effective strategies for prevention is therefore of critical importance. Such information should assist policymakers to justify policy decisions on interventions to prevent cardiovascular disease.
Aims: To better understand the epidemiological profile and pattern of stroke and its risk factors in rural South Africa so as to select interventions that are effectively targeted, are cost-effective, and have the potential to reduce the cardiovascular disease burden in key populations.
Methods: The study utilised two methodological approaches to address the research aims. First, existing validated methods and epidemiological tools were used. The Global Burden of Disease Study approach was the framework for analysing years lost due to premature mortality (YLL), years lived with disability (YLD) and disability adjusted life years (DALYs). The mathematical tool - Dismod II - was used to estimate the disease incidence. The World Health Organization’s comparative risk assessment framework was used to estimate burden attributable to metabolic risk factors; namely high blood pressure, raised blood glucose and excess body mass index. All data for the analyses were derived from the Agincourt health and sociodemographic surveillance system, which covers a population of over 100,000 individuals in rural North-east, South Africa, adjacent to Southern Mozambique. Extrapolation to ‘mostly rural’ South Africa was based on the conservative assumption that all municipalities that are considered ‘mostly rural’ share an epidemiological profile similar to that of Agincourt sub-district; thus only population figures (by age and sex) adjusted the incidence and DALYs. Second, an economic model to assess the cost-effectiveness of CVD prevention interventions was developed, customised for the Agincourt sub-district population. This is a Markov model which simulates disease progression in a cohort of people, starting from a healthy (disease-free) state until death or 100 years of age by accounting for changing risk factor profiles in each age-sex cohort. The application of the model is illustrated by estimating the cost-effectiveness of alternative pharmaceutical interventions for CVD prevention. Interventions are targeted to groups with varying degrees of absolute CVD risk over the next 10 years of ‘<10%’ (low risk), ‘≥10% & <20%’ (medium risk) and ‘≥20%’ (high risk). A fourth target group was individuals with untreated stage 2 hypertension.
Results: Baseline burden of CVD was substantial: crude stroke incidence rate was 244 per 100,000 person years in Agincourt sub-district. An estimated 33, 500 strokes occurred in 2011 in “predominantly” rural municipalities of South Africa, a population of some 13,000,000 people. Crude stroke mortality was 114 per 100,000 personyears in 2007–11 in Agincourt sub-district whilst 1,070 DALYs (CI 750 - 1680) were lost due to stroke. Preventable risk factors were responsible for a significant proportion of the stroke burden: Among males, 40% of the stroke burden was attributable to high blood pressure. Similarly, 38% of the stroke burden in females was attributable to high blood pressure. There was marked variation in stroke burden attributable to excess body mass index (BMI) by gender. Approximately 11.4% of the stroke burden in males was attributable to excess BMI compared to 22.5% in females. Despite some uncertainty, it appears that diabetes plays a relatively small contributory role to the overall CVD burden. Furthermore, it has been assumed that cholesterol does not play an important role in this setting based on previously published data. By combining actual health care utilization estimates from the Agincourt sub-district with incidence and prevalence data, we estimated that direct costs of stroke treatment comprised 1-3% of the sub-district expenditure in 2013. Average costeffectiveness ratios (ACERs) for all pharmacotherapies, across all target groups, fell in the range of US$160-500 per DALY averted. Incremental cost-effectiveness ratios (ICERs) for the optimal interventions (i.e. after removal of costlier yet less effective ‘dominated’ interventions) were US$156 (combination of β-blocker and diuretic) and US$373 (polypill – a single tablet containing a statin and three antihypertensive agents) per DALY averted for the high-risk group. ICERs indicate the additional cost needed to avert an additional DALY as successively less cost-effective strategies are adopted. The same interventions were cost-effective for the medium-risk group and patients with stage 2 hypertension. The optimal treatment pathway for low risk individuals (10-year risk of <10%) included low-dose diuretic with an ICER of US$228; combination of β-blocker and diuretic (ICER: US$454 per DALY averted); polypill (ICER: US$683 per DALY averted) and ACE-I/diuretic combination which yielded an ICER of US$2,752 per DALY averted. ICERs were sensitive to discount rates, doubling the costs of drugs and halving the drug treatment effects. The budget impact analysis indicated that giving the optimal interventions of β-blocker /Diuretic and polypill to at least 90% of individuals with high CVD risk in the entire rural South Africa would cost the government, per year, approximately US$4.06 million, US$6.28 million respectively.
Conclusions: The burden of stroke in Agincourt HDSS as estimated in this study is considerable and is currently propelled by, amongst other risk factors, the high prevalence of hypertension and obesity. If we assume similar prevalence and distribution of risk factors for the other 69 rural municipal areas, then the effect on rural South Africa is equally substantial. By applying the custom-built economic model, we show that several pharmaceutical options are potentially cost-effective and affordable in reducing CVD. Furthermore, the cost-effectiveness modelling showed that a total risk factor approach is more cost-effective than a single risk factor approach. The model, which is an output of this thesis can be applied across other health and socio-demographic surveillance sites to build an augmented dataset that will assist in better profiling of CVD prevention across other sub-Saharan African sites.
Description
A Thesis Submitted to the Faculty of Health Sciences, University of the Witwatersrand, Johannesburg, in fulfillment of the requirements for the degree of Doctor of Philosophy, Jphannesburg October 2018