STRENGTHS AND LIMITATIONS OF THIS STUDY
-
A cost–utility analysis comparing medical treatments for hypertension was undertaken for the first time in Zambia.
-
We use direct cost data specific to the Zambia context.
-
Utility values, transition probability between health statuses and case fatality rate were not available for the Zambia context; thus, we used secondary data and tested the sensitivity of results to the chosen parameters.
-
The model focuses only on first-line pharmaceutical interventions at the primary level and does not consider preventive or other kinds of interventions.
Introduction
Although hypertension (HTN) can be easily detected through blood pressure (BP) measurements and is often treated effectively using medications, it is estimated that 700 million cases remain untreated globally.1 2 With 62% of cerebrovascular diseases and 49% of ischaemic heart diseases estimated to be attributable to suboptimal BP control,3 4 HTN treatment is a key topic of interest among researchers and practitioners.5–7 The NCD Global Monitoring Framework coordinated by the WHO uses the proportion of individuals taking antihypertensive treatment or counselling among those at a higher risk of cardiovascular disease (CVD) to monitor progress in reducing the disease burden associated with Noncommunicable disease (NCD) in each country.8
In Zambia, one of the sub-Saharan African countries, the demand for quality HTN services is rising.9 10 According to the latest WHO STEP wise approach to surveillance,11 a large number of people has been diagnosed with HTN: 19.1% of the adult population (20.5% of men and 17.6% of women) had elevated BP, defined as systolic BP (SBP) ≥140 mm Hg and/or diastolic BP≥90 mm Hg. Other studies reported higher morbidity levels; the prevalence of HTN in adults ranged between 25.8% and 32.8% in rural areas12 13 and 34.8% in urban areas.14 Considering this, the improvement of health services to prevent and treat HTN has been identified by the Zambian Ministry of Health (MoH) as one of the key priority in the National Health Strategic Plan 2022–2026.15 While there is a promising policy agenda, the cost-effectiveness of HTN treatment has never been evaluated in Zambia. The evidence on the cost-effectiveness of treatment for HTN in sub-Saharan Africa is mixed.16–27 Most of the studies suggested that pharmaceutical treatment should be cost-effective compared with no treatment,17 21–23 25 especially among people with moderate and severe HTN. However, there is no consensus on which pharmaceutical intervention is the most cost-effective, with high levels of uncertainty related to specific assumptions regarding the cost data. For example, the unit cost of outpatient department (OPD) visits per patient is US$7.1 and US$29–US$55 for medication in Tanzania,17 28 compared with US$4.3 and US$1–US$7 in Zambia, respectively, found in our previous study.29 Similarly, the admission cost for CVD differs significantly between South Africa (US$563616) and Zambia (US$120–US$14129).
In this study, we investigate the cost-effectiveness of HTN treatment by using the cost data collected independently in the local health facilities, focusing on the first-line pharmaceutical interventions at the primary level in rural Zambia. The results from this study will provide the government and MoH with evidence-based recommendations to assess the feasibility and affordability of HTN health policies.
Methods
A cost–utility analysis was undertaken from the perspective of healthcare providers to compare the cost-effectiveness of interventions for HTN in the adult population in Zambia. The economic evaluation methods and results were reported following the International Society for Pharmacoeconomics and Outcomes Research (ISPOR) Consolidated Health Economic Evaluation Reporting Standards.30
The Chongwe District in Zambia was selected as the setting and location to conduct the economic analysis due to the availability of unit cost, health service data for cardiovascular diseases (CVDs) from previous studies29 31 . This data was obtained as part of the activities in the Project for Strengthening Basic Health Care Services Management for Universal Health Coverage conducted by the Japanese International Cooperation Agency. The Chongwe District is a typical rural district in Zambia, located next to the capital city. According to the latest Census,32 the population’s average age was 21.1 (SD: 0.25) in 2010. The age distribution was similar between the Chongwe District and the national total. The average prevalence of HTN was 19.1% (95% CI 17.5% to 20.6%).11 The CVD risk distribution at the age of 40 was different across genders (26% low risk, 21% moderate risk and 12% high risk for males and 21%, 15%, and 5% for females).31 All detailed data on age distribution, HTN and CVD risks for both genders are provided in online supplemental table S1–16.
Supplemental material
Population
The target population for this study was patients from the Chongwe District aged ≥40 years with HTN, as the WHO recommends routine BP screening for individuals in this age group due to their higher risk of CVD.33–35 Zambia’s 2016 Standard Treatment Guideline (STG) aligns with this recommendation.36 HTN is defined as SBP≥140 mm Hg and/or DBP≥90 mm Hg, according to the HEART technical package.34 It was assumed that people are eligible for preventive cardiology intended as the practice of early, primary and secondary prevention of all CVDs37 at the age of 40 years and that thereafter they receive treatment for their remaining life spans. This assumption is in line with the WHO package34 that proposed the routine screening of HTN among persons aged >40 years, as a high-risk population for CVDs, as well as the clinical protocol in Zambia36 that promotes a screening for HTN for this population subgroup.
A hypothetical cohort of 1000 male and female patients according to gender distribution in the district (online supplemental table S4) was entered into the model for the treatment of HTN. The starting age in the model was 40 years (in accordance with the aforementioned WHO and Zambian STG recommendations33–36). To provide a relevant policy message, the Zambian guidelines do not suggest different treatments for different levels of risk or sex; hence, the main findings of this study relate to the general population in the Chongwe District. However, an exploratory analysis was performed on clinically relevant subpopulations based on the three risk levels (low, moderate and high) and sex, resulting in a total of six subgroup analyses.
Intervention and comparators
A Markov model assessed the cost-effectiveness of intervention (ie, drug combination to treat HTN recommended by the WHO HEARTS technical package34 and Zambia STG,36 as listed in table 1) versus comparator (no pharmaceutical intervention). The type, name and recommended dose of each drug available for HTN in Zambia38 are presented in online supplemental table S1.
List of interventions
Cycle length and time horizon
A lifetime horizon was chosen to capture and evaluate the full costs and effects over a lifetime, in line with the National Institute for Health and Care Excellence (NICE) recommendations39 and ISPOR good practices.40 Specifically, the time horizon was set at 30 years because the life expectancy was 29.5 years at the age of 40 years for both genders, according to the 2010 census life table.32 The lehgthe of each sycle was a 1-year.17
Discount and currency rates
An annual discount rate of 3.5% rate was applied to costs and utilities as recommended by NICE.39 The currency rate adopted was US$1=ZK10.59 in 30 June 2019.
Model structure
The CVD prevention model is a probabilistic Markov-type model representing long-term costs and outcomes associated with treatments and comparator, considering a hypothetical Zambian population. The model (figure 1) was adapted from a previous study17, which investigated the cost-effectiveness of first-line pharmaceutical interventions for HTN in the Tanzanian population. Six mutually exclusive health states were considered: ‘no history of CVD,’ ‘history of myocardial infarction or congestive heart disease (CHD)’ and ‘history of stroke’. Finally, ‘death’ was modelled as an absorbing state, and stroke was modelled in three different health states based on severity: ‘mild,’ ‘moderate’ or ‘severe,’ to be consistent with disability weights reported in the Global Burden of Disease (GBD) 2010.41 In the model, a cohort starts from healthy individuals aged 40 (ie, ‘no history of CVD’). The cohort transitioned between different health states according to age-specific risks for each type of clinical event and considering also risk reduction from the interventions. Each health state is associated with a specific cost and disutility (as detailed in table 2). After each cycle, patients can stay in their state or change state in line with the model transition probabilities. We assume that if an individual experienced several CVD events, we used the cost and disutility related to the most severe event. Also, our model assumes that patients can either stay in their health state or move to a more severe status. A 70-year time horizon was considered to reflect life expectancy at the age of 40 years in Zambia, according to the 2010 census life table.32 In each model, ‘no treatment’ was the baseline strategy, which was compared with different interventions for the primary prevention of CVD. To explore the effects of the interventions in different populations, three distinct submodels were created to represent three CVD risk levels: low risk, moderate risk and high risk, according to the WHO CVD risk chart at the age of 40 years.42
Model structure. CHD, congestive heart disease.
Model parameters
Model parameters
Transition probabilities
The annual risk of acute myocardial infarction (AMI), CHD and stroke was estimated using the non-laboratory-based risk chart developed by the WHO Risk Chart Working Group42 (risk equation shown in online supplemental figure 1).
Supplemental material
The reason for choosing a non-laboratory-based risk chart was that laboratory testing was not necessarily available in all health facilities in the Chongwe District. The annual risks of AMI or CHD and stroke are described by the WHO CVD risk in persons aged 40 years in online supplemental table S7. Using the weights and the transition probabilities among different risk categories in online supplemental table S8–15, the transition probability was calculated for the general population in Chongwe District aged 40 years (table 2).
The case fatality rates of AMI and Stroke were extracted from the previous literature17 43–45 and are outlined in table 3. Background mortality in Zambia was obtained from the Life Table of the 2010 Census,32and the annual mortality of people after the first event of CHD and stroke was assumed to increase according to previous studies44 45 as described in table 3. Postevent background mortality was assumed to be twice as high as the original mortality trend after a CHD and stroke event. Besides, the transition probability from stroke to the three severity levels (mild, moderate and severe) was collected from previous studies17 46 47 (table 2).
Result of cost and effectiveness by the Markov model among the general population
Health outcome
Disability-adjusted life-years (DALYs) were calculated using data from the GBD 201041 for years lost due to disability and Zambia age-specific life expectancy data for years of life lost.32 Table 3 summarises the data.
Effect of interventions
The relative risks (RRs) of the drug classes were retrieved from a systematic review,48 meta-analysis49and other published sources.50 The details are listed in table 3. The effects of combination interventions were calculated multiplicatively using RRs (RR1×RR2) as described in a previous study.17 The effects of each intervention are summarised in online supplemental table S15.
Treatment cost
The standard price of each HTN drug was obtained from the catalogue of Medical Store Limited,38 which was owned by the Zambian government. In the cost-effectiveness analysis, a cost estimated for the Zambia context was used. This cost is significantly lower compared with estimated cost from similar studies and context21; therefore, a sensitivity analysis considering a higher cost was performed. Table 3 lists the medications used in this study and their associated costs.
Regarding health service costs, unit costs of OPD visits and laboratory testing for HTN by different health levels were obtained from previous publications,29–31 considering the Zambia context. While OPD unit costs included capital and labour costs, laboratory unit costs included capital, labour and material costs. The material cost for the OPD consultation was added to the relevant OPD unit cost, according to the selected intervention. The proportion of OPD visits per health facility level was also collected from the same previous study,31 (14.5% for first-level hospitals, 46.7% for health centres and 38.8% for health posts). Using OPD visit weights by health level, the average and range of unit costs for HTN in the entire Chongwe District were calculated. Eastern OPD visits included screening, laboratory testing and consultations. The unit cost per BP screening was OPD unit cost×0.125, and table 3 summarises the unit costs and their ranges. The frequency of follow-up recommended by the Zambian government is once every 1–3 months. For each expected frequency, the annual cost of HTN health services per patient was calculated as shown in table 3 and online supplemental table S17.
Regarding the cost of treatment for AMI, CHD and stroke, admission costs, including capital, labour and material costs, were extracted from a previous study29 (table 2). Special treatments, such as coronary artery bypass grafting, percutaneous transluminal coronary angiography and other intensive care, were not considered because these medical technologies were not available in Zambia during the study period(2019). For secondary prevention, it was assumed that a dose of 75 mg of acetylsalicylic acid was initiated for all patients in addition to HTN drugs in the intervention before the event. Table 2 summarises treatment costs after CVD events. All costs were discounted at the 2019 level according to World Bank inflation rates.51
Assumptions
Once individuals develop AMI or stroke, they continue to receive interventions for primary prevention until death. Life expectancy is estimated under poststroke or post-AMI conditions without secondary prevention.
Analytical methods
Main analysis
The incremental effect, cost and incremental cost-effectiveness ratios (ICERs) were first calculated for the six mutually exclusive interventions (table 1) in two different analyses regarding the general population and then according to the three levels of risk of HTN and between sexes. Strategies with ICERs below the GDP per capita in Zambia (approximately US$1500 in 2019)51 were considered cost-effective.
Sensitivity analysis
Deterministic and probabilistic sensitivity analyses were performed. A deterministic sensitivity analysis was performed on the following parameters: the effects of drugs on stroke and CHD, cost of drugs, the utility of CHD and stroke, initial treatment cost of CVD, cost of aspirin treatment after the event and cost of OPD visits. Parameter for both sensitivity analyses are fully reported in online supplemental table S18 and 19. For the deterministic sensitivity analysis, a Tornado diagram was produced, and for the probabilistic sensitivity analysis, an incremental cost-effectiveness scatter plot and cost-effectiveness acceptability curves were produced using TreeAge Pro.
Budget impact
A budget impact analysis was also performed using the results of the Markov model and the data of the HTN population of the Chongwe District (online supplemental table S3).
Results
Cost-effectiveness analysis of treatment for HTN medications among the general population
The results for the general population are presented in table 3. The dominant treatment is a combination of diuretics and calcium blockers. The estimated ICER was US$1114.65 per averted DALY, compared with no treatment. As the incremental cost-effectiveness per DALY threshold was set at US$1500 based on Zambian GDP per capita,51 this combination therapy was considered a cost-effective intervention.
Deterministic sensitivity analysis: the general population
The Tornado diagram between the no treatment and dominant intervention, which was a combination therapy of diuretics and calcium channel blockers (CCB), is shown in online supplemental figure 2. The midpoint of the ICER was US$1129.06 per averted DALY. As the cost of drugs changed from 1/10th to 10 times the standardised price, the ICER ranged from approximately US$1100–US$1350 per averted DALY. The ICER score changed from US$800 to US$1400 per averted DALY, and the parameter with the greatest impact on the ICER was the cost of OPD visits, which depended on the frequency of visits. If the frequency was less than every month, the ICERs were lower than US$1500 per averted DALY.
Supplemental material
Probabilistic sensitivity analysis for the general population
Online supplemental figure 3 depicts the incremental cost-effectiveness scatter plot of the combination of the dominant treatment strategy (combination of diuretics and calcium blockers) for HTN compared with no treatment using 1000 Monte Carlo iterations. The dotted line represents a willingness-to-pay (WTP) threshold of US$1500. Most dots were below the WTP threshold and the ICER from the PSA analysis comparing Diu+CCB and no treatment was US$880/averted DALY. We also reported the 95% uncertainty interval estimates for ICERs from the PSA: (248–2118). Figure 2 displays the cost-effectiveness acceptability curves representing the probability of cost-effectiveness for each level of WTP. No treatment is the most cost-effective strategy if the WTP is below approximately US$1200, and combination therapy with diuretics and calcium blockers becomes the most cost-effective intervention if the WTP is beyond this value.
Supplemental material
Cost-effectiveness acceptability curve among men at high risk. ACEI, ACE inhibitor; ARB, angiotensin receptor blocker; CCB, calcium channel blockers.
Subgroup analysis
The target population was categorised into six groups according to sex and three CVD risks: mild, moderate or high. Online supplemental table S20 shows the cost-effectiveness results for males and females at each level of the risk subgroup. In all subgroups, the dominant treatment was a combination therapy with diuretics and calcium blockers. However, compared with the absence of treatment, this intervention was not cost-effective across all subgroups. The ICERs between combination therapy and no treatment are summarised in online supplemental table S22. The cost-effectiveness acceptability curves for the six subgroups are illustrated in online supplemental figure 4. When the cost-effectiveness threshold of ICER is set at US$1500 per averted DALY based on GDP per capita,51 combination therapy is cost-effective among both sexes with high and moderate CVD risk. However, in a population with low CVD risk, no treatment should be considered a cost-effective strategy.
Supplemental material
Budget impact
Online supplemental table S22 shows the estimated annual budget impact if all HTN patients aged 40 years and above in Chongwe District received the most cost-effective treatment: diuretics and calcium blockers. The annual total budget impact was estimated at US$1 015 605, including labour, capital and material costs. Taking into account only material costs, the annual total budget was US$29 435.5, which needs to be covered by the Chongwe District as the national government would be responsible for labour and capital costs.
Discussion
This study examined the cost-effectiveness of several first-line HTN therapy options recommended by the WHO using novel data coming from primary care in rural Zambia. According to the findings, the most cost-effective first-line medication for HTN in a rural Zambian setting was a combination therapy with diuretics and calcium blockers for the general population. Considering subpopulations by gender and three levels of risk, this combination of therapies was still cost-effective for all high-risk individuals (male and female) and for male moderate-risk individuals. The threshold of ICER was set as US$1500, which is approximately equivalent to the GDP per capita in Zambia in 2019.51 For other subgroups (male and female low-risk and female moderate risk) ‘no treatment’ was the most cost-effective strategy. In a previous study investigating the cost-effectiveness of first-line antihypertensive drugs in Tanzania,17 the combination of ACE inhibitor+diuretic and triple therapy of ACE inhibitor+diuretic+calcium blocker was the most cost-effective interventions for populations with low to moderate risk and high to very-high risk, respectively. However, triple therapy was not included in this study because this package was not recommended as a first-line drug by WHO HEARTS guidelines.35 Thus, the results were not comparable among high-risk populations. However, in the low-risk population, our results, finding that the combination of diuretics and calcium blockers was the most cost-effective, differ from the Tansania study.17 As the drug effectiveness parameters were similar between the two studies, this discrepancy is likely due to the cost of each HTN drug. The cost of ACE inhibitor (ACEI) and CCB in another study21 in Tanzania. However, as our cost source derived from the standard prices published by the governmental medical supply organisation in Zambia36 (the annual costs of ACEI and CCB are estimated as US$29.3 and US$54.8 to US$1.16 and US$3.47, respectively), we believe that our estimates are more likely to provide robust results for the Zambian context.
According to the sensitivity analysis exhibited in figure 2, most of the changes in the parameters did not increase the ICERs beyond US$1500, which is Zambia’s GDP per capita. While only the costs of OPD visits could affect the selection of cost-effective interventions, combination therapy remained a cost-effective intervention if the frequency of OPD visits was less than once per month. Therefore, considering both one way and PSA (with CEAC showing the highest probability of cost-effectiveness of our dominant strategy at a WTP higher than 1300), among the general population in the Chongwe District, the results of cost-effectiveness analysis for pharmaceutical intervention for HTN seem robust.
Regarding budget impact, US$29 435.5 would be required as the annual total material costs for the district if all target populations could receive the most cost-effective pharmaceutical treatment, which is a combination of diuretics and CCB. Most material costs are derived from the cost of the drugs. According to the National Budget Book of the Ministry of Finance in Zambia,52 the annual budget for service delivery was US$241 481, excluding human resource and health system management costs. The budget impact of hypertensive drugs would account for more than 10% of the total district health office budget. Considering other health services, this appears to be a potential financial burden if the district needs to cover these investments. Support at the national or provincial level is critical for maintaining the supply of hypertensive drugs, based on the estimation of the necessary budget for HTN control so that every patient can access healthcare services.
The strength of this study is that it provides novel evidence for the cost-effectiveness analysis of HTN pharmaceutical treatments in the WHO guidelines using locally obtained data in Zambia. Consequently, building evidence on primary data from Zambia, this study is likely more robust and reliable for policy purposes in the country, as other existing evidence created on secondary data or with data from neighbouring countries. Additionally, we conducted a budget impact analysis highlighting the resources required for HTN services in rural sub-Saharan African countries. This information could contribute to create a tangible plan for the expenditure framework for comprehensive essential health services, including HTN, at the district level.
However, this study has several limitations. First, secondary prevention and special treatment for CDH, AMI and stroke were not considered because these treatments were not available in Zambia. However, as medical technologies are being developed in Zambia, these advanced interventions for CVD have to potential to be incorporated into future national guidelines. With this perspective in mind, a cost-effectiveness analysis that includes advanced treatments for CVD should be considered in the future. Second, there was limited information about the necessary parameters for economic analysis, such as utility value, transition probability between health statuses, case fatality rate related to CVD, patient costs and updated Census data since 2010 in Zambia or sub-Saharan Africa. Standardised values of these parameters specific to the sub-Saharan African setting should be investigated to identify priority areas for investment based on more reliable data. According to the deterministic analysis in figure 2 among the general population, the costs of interventions (and specifically the cost of drugs) could significantly influence the results of the cost-effectiveness analysis to calculate ICERs. Future studies should review market prices in specific sub-Saharan countries. Finally, this study did not consider other policy combined with the pharmaceutical treatment, such as reducing sodium intake, the promotion of physical activity and sensitisation to promote awareness and maintain high adherence to HTN treatment. The cost-effectiveness of comprehensive interventions, including policy-level actions, health promotion activities, pharmaceutical treatment and advanced health technology, should be investigated in future studies to assess the feasibility of introducing CVD programmes in sub-Saharan Africa, including Zambia.
Conclusions
The most cost-effective first-line medication for HTN among the six types of drug combinations in rural Zambia is the combination therapy of diuretics and calcium blockers for the general population. When analysing the subpopulation results, the most cost-effective intervention was still the combination of diuretics and calcium blockers for the subpopulations, except for men with low risk and women with moderate and low risk. Among these populations, no treatment is cost-effective. Besides, as the costs of HTN drugs are affordable, it is worth considering the promotion of delivering HTN medications even in rural areas. However, as this study analysed only first-line medications for HTN, further studies are imperative to assess the feasibility of introducing a comprehensive CVD programme in Zambia, including community sensitisation, lifestyle modification, secondary prevention and advanced treatment.
Data availability statement
No additional data available.
Ethics statements
Patient consent for publication
Acknowledgments
The authors wish to thank Dr Mzaza Nthele (Director, Department of Clinical Care, Diagnostic Service), Dr Consity Mwale (Lusaka Provincial Health Director), Dr Joseph Kabungo (Chongwe District Health Director), Dr Arthur Mumba (Medical Superintendent, Chongwe District Hospital), Mr. Andrew Chibangula (Planner in Chongwe District Health Office) and Mrs Tasila Sompa (Information Officer in Chongwe District Health Office) of the Ministry of Health, Zambia, for setting up the research environment. Furthermore, the authors wish to acknowledge the support from Dr John Musuku, Dr Evans Mulendele, Dr Dominique Chimanuka, Dr Nchimunya Machila and Ngosa Mumba (University Teaching Hospital in Zambia) for the development of the design, the data collection and analysis.
This post was originally published on https://bmjopen.bmj.com