3.1. Literature review

A rapid literature review was initially conducted to summarize and update available cost-effectiveness evidence data for malaria control and elimination. Several literature reviews have previously been conducted on the economics of malaria prevention and treatment (Shretta et al., Reference Shretta, Avanceña and Hatefi2016; Conteh et al., Reference Conteh, Shuford, Agboraw, Kont, Kolaczinski and Patouillard2021). This review therefore focused on new articles published after 2019. Details on the literature review can be found in Appendix A.

3.2. Transmission model

The fundamental epidemiological and basic economic model used here is the Single Patch Plasmodium falciparum (SPPf) tool. This spatially explicit, compartmental, nonlinear, ordinary differential equation transmission model is an extension of previously published models and has been implemented in R and C++ (White et al., Reference White, Maude, Pongtavornpinyo, Saralamba, Aguas and Van Effelterre2009; Silal et al., Reference Silal, Little, Barnes and White2014; Silal Reference Silal2019; Silal et al., Reference Silal, Shretta, Celhay, Mercado, Saralamba and Maude2019). The economic evaluation presented here uses the outputs of this transmission model as described below.

Key features of the model include four infection classes representing infections that are severe, clinical, asymptomatic and detectable by microscopy, and asymptomatic and undetectable by microscopy, with each infection class having an associated infectiousness based on infectivity data. The probability of individuals entering each class of infection is dependent on their immunity status. It is assumed that untreated individuals will transition from higher to lower severity infection classes as they recover and that they can be boosted to higher severity classes through superinfection. It is assumed that treated individuals test positive for histidine-rich protein 2 (HRP2) after clearance of asexual parasitaemia for different durations depending on the detection limit of the test used. Other additional features were subnational climatic variation (seasonality) and importation of infection. More details on the model and the parameters driving the model can be found on GitHub (2020).

3.3. Data

The data used to calibrate the model were obtained from several sources. The main estimates for cases and deaths stem from the latest updated World Malaria Report 2022, covering the period 2000–2021. To mitigate skewing the model outputs with the malaria program disruptions caused by COVID-19, data points beyond 2019 were not used for the model. When unavailable in the newest update, we have also extracted specific information from the World Malaria Reports for the period 2001 to 2021. The data collected covers Non-community cases; Community cases; Number of LLINs sold or delivered; Number of people protected by IRS; Reported fatalities due to malaria; Population at risk (high, low transmission, and active foci); Coverage of first-line treatment; and Coverage of RDT (years available).

Owing to differing reporting standards and interpretations of community cases, both community and non-community cases were grouped together. Where parameters driving the model could not be estimated from available data, they were sourced from existing literature.

The scenarios modeled including the assumptions are shown in Table 1.

Table 1.

Scenarios modeled.

Note: *LLIN effectiveness, usage × proportion of bites averted.

In all countries, interventions to increase use beyond the estimated proportions implemented in 2019 were added to simulate increased net use. The interventions modeled were a combination of activities of a “hang-up campaign” as well as SBC and IEC, where LLIN coverage and use increased by 10 percentage points by 2030.

4.1.1. Direct cost savings to the health service

The total direct cost savings resulting from fewer malaria cases was estimated using data from published literature at the national level (see Appendix B). Where no data were available, proxies were used from other countries or the literature. The findings reflect the vertical costs to the malaria program and the publicly funded system costs of implementing the malaria intervention. Cost estimates expressed in international (PPP) US$ value were converted to 2021 constant US$ values.

4.1.2. Direct cost savings to households

Malaria exerts a significant financial burden on households. Malaria patients often pay for transportation to access health facilities, diagnostic services, and medicines. In many countries in Africa, although testing and treatment for malaria and antimalarials are free, prepaid, or covered by capitation of the National Health Insurance Schemes, malaria patients still incur out-of-pocket expenditures (OOP) (Nabyonga et al, Reference Nabyonga Orem, Mugisha, Okui, Musango and Kirigia2013; RBM, 2015). To estimate direct household costs for malaria, the number of reported outpatient (OP) and inpatient (IP) malaria cases was multiplied by the mean OOP spending, which included the cost of transportation (separately for OP and IP cases).

4.1.3. Indirect benefits to society

The economic impact of malaria extends beyond the health system. Patients forego income while recovering from malaria, caregivers looking after ill children and the elderly also lose out on potential earnings, and children missing out on school affect human capital accumulation. Premature deaths also cost society through losses in lifetime productivity and in the value that people place on living longer, healthier lives.

To evaluate the economic impact of malaria-related morbidity, the income lost for malaria patients and caregivers was estimated. The estimated income per worker was derived from GDP per capita adjusted for labor force participation and labor share of GDP. The resulting figure was used as a proxy for lost worker income, the time value of non-working adults (15 years and older) was reduced by 50%, and a zero value of time was assigned to children under 15 years old. The incidence of malaria for each country reported in Global Burden of Disease for 2019 (IHME, 2021) was used to estimate the share of children and adults, respectively. For each age group, the value of the lost productivity was multiplied by the duration of OP and IP illness from published literature and the number of reported OP and IP cases. In addition, the effect of reduced productivity from “presenteeism” was calculated by assuming that adults returning to work after malaria illness would be 50% less productive for an additional three days.

Averted mortality is valued using a standardized approach across all Halftime SDG papers, which follows the recommendations of Robinson et al. (Reference Robinson, Hammitt, Cecchini, Chalkidou, Claxton, Cropper, Hoang-Vu Eozenou, de Ferranti, Deolalikar, Guanais, Jamison, Kwon, Lauer, O’Keeffe, Walker, Whittington, Wilkinson, Wilson and Wong2019).

To estimate the value of averted mortality, we use the U.S. Value of Statistical Life (VSL) US$ 9.4 million (2015 US$) as reference, which represents approximately 160 times income as measured by income per capita PPP. The relationship is transferred to the entire low- and lower–middle-income population via the ratio of GDP per capita, using an income elasticity of 1.5.

To estimate these values, we take the population-weighted GDP per capita figure in 2020 Int$ for the group of LLMCs and the United States of America, and estimate the VSL at time t = 0, 2020.

$$ {VSL}_t={\left(\frac{GDP\;{pc}_{LLMC,t}}{GDP\;{pc}_{USA,t}}\right)}^{e-1}\ast 160 \ast GDP\;{pc}_{LLMC,t} $$

Following Cropper et al. (Reference Cropper2019), we estimate each subsequent VSL in the time series according to the following formula:

$$ {VSL}_{t+1}={VSL}_t \ast {\left(1+{g}_t\right)}^e $$

where gt is the real GDP per capita growth rate between period t and t + 1 (SSP Database, IIASA GDP Model, Scenario SSP2_v9_130219) and e = 1.5. The value per statistical life year (VSLY) is calculated by dividing the VSL by half the life expectancy at birth.

The GDP growth in this group of countries outpaces population growth so that VSLY grows rapidly over time. In constant 2021 US$ values, the benefit of averting a life year lost (VSLY) is US$ 3,732 (2023), US$ 5,049 (2025), and US$ 6,062 (2030).

Using the distribution of malaria deaths between age groups by country reported in the Global Burden of Disease (GBD 2019), and assuming 2.5 years as the average death among children under 5 years, 12 years among children aged 5–19, and half the remaining life expectancy for adults over 20 years. The average life expectancy of males and females was used to estimate the number of years of life lost and then multiplied by the value of an additional life year (VSLY) for low-income and low–middle-income countries (all deaths valued equal). Data on life expectancy was retrieved from World Bank data.

4.2. Cost projections

Unit costs (see Table 3) were used in the SPPf model to calculate the cost of the scenarios and the additional costs of the LLIN and SBCC scale-up scenario compared to the baseline scenario.

Table 3.

Unit costs used for estimating intervention scale-up costs.

Table 4.

Summarized results of incremental costs and benefits of the LLIN and SBCC scale-up scenario compared to baseline (2023–2030).

4.3. Benefits estimation

The benefits of each scenario were estimated as the sum of the direct cost savings to the health system from reduced use of outpatient and inpatient health services and reduction in the cost of delivering malaria control activities, the direct cost savings to households, and the indirect cost savings of reduced morbidity and mortality from malaria calculated above. These were computed using the outputs of the transmission model: the malaria cases and deaths averted in the scale-up scenario compared to the baseline scenario were calculated and valued using the same methods described previously for estimating the economic burden of malaria (see Table 2).

Each of these was estimated for each of the 29 countries and added together to obtain the total cases and deaths averted, the total costs, and the total benefits.

The Net Present Value (NPV) was calculated to obtain the present value of the future revenue generated from reducing the burden of malaria using standard economic techniques. The purpose was to give a true picture of the financial value of an investment today. The timeframe used for calculating the NPV was 7 years (2023–2030) and an 8% discount rate was applied.

4.4. Benefit–cost ratio

The BCR is interpreted as the economic return from every additional dollar spent on malaria above the baseline scenario. To calculate the BCR, the NPV of the incremental benefits of the scale-up scenario compared to baseline was divided by the NPV of the incremental cost of the scale-up scenario (compared to the baseline).

4.5. Sensitivity analysis

A stochastic sensitivity analysis on the epidemiological and cost outputs of the malaria transmission model was performed. The minimum, median, and maximum malaria cases and deaths predicted by the model for each scenario were used to calculate the minimum, median, and maximum costs. Three hundred random samples were drawn, which generated a range of costs. From the range of costs generated, the minimum, maximum, and median percentiles are presented.

4.6. Limitations

This report has several limitations. Due to time and resource constraints, the transmission model generated national transmission-based estimates based on the World Malaria Report. Higher levels of spatial heterogeneity would need to be modeled to enable more accurate subnational estimates of benefits and costs. The costs of interventions have been estimated based on available published data and proxies when data were unavailable. For example, the costs of outpatients and inpatients were derived from WHO/CHOICE. As countries move closer to elimination, the impact of active surveillance on both the epidemiology and the cost will also need to be included. This was not included due to a lack of historical data to enable fitting the model for impact or cost.

While employee absenteeism was included in the estimates of benefits, the analysis did not include the economic benefits conferred by reductions in school absenteeism and subsequent improvements in cognitive development due to the limited empirical evidence to enable converting these estimates to wages earned (Kuecken et al., Reference Kuecken, Thuilliez and Valfort2020). Other benefits not included include potential benefits on tourism and the impact of economic development and housing improvements on malaria transmission, as well as regional or cross-border externalities.

Households spend substantial amounts of money on malaria preventive tools such as insecticide sprays and repellants. These costs were not included in this study, thereby possibly underestimating the direct household costs of malaria. In addition, infection with malaria is likely to result in a higher likelihood of death from other causes such as HIV and newborn mortality. These additional impacts are not included.

Last, the effectiveness of LLINs at reducing bites is assumed to be 40%. However, this may be an overestimate given recent concerns with pyrethroid resistance and net durability (Kilian et al., Reference Kilian, Obi, Mansiangi, Abílio, Haji and Blaufuss2021). New, more costly nets are likely to be needed in the future, and resistance management strategies will need to be deployed. To accommodate additional costs of maintaining effectiveness, we calculated the average price of an LLIN assuming 30% of the standard nets are replaced with chlorfenapir nets, and in addition, adopted the higher-end range of the ITN and SBCC scale-up cost estimate.

5.1. Rapid literature review

In total, 53 articles were screened for eligibility. After screening, 48 articles were included in the analysis, with the majority of articles published in 2020 and 2021 (19 and 16, respectively). Reasons for exclusion were opinion paper (1), discrete choice experiment (1), protocol (1), severe malaria incidence (1), and Plasmodium vivax (1). The total number of countries included in all studies was 24, with the majority of countries being in sub-Saharan Africa. The majority of the studies were cost-effectiveness analyses (80.9%), with the least being cost-saving analyses and investment cases (4.3% each). Some 83% of studies were focused on malaria control, while 17% were focused on malaria elimination. The number of studies with more than one economic outcome reported was just 18. The studies employed heterogeneous inputs and methodologies preventing cross-comparisons and an overall synthesis of all the outputs. Summaries of the review are presented in Appendix A.

These and previously published studies affirm that interventions to prevent malaria, particularly the use of LLINs, are highly cost-effective across different settings using different distribution channels. The use of LLINs in combination with improved SBCC is therefore considered in this article to be among the most cost-effective policies for scaling up in the control setting at the present time.

5.2. Transmission model predictions and projections

5.2.1. Baseline response

Maintaining the interventions (LLIN distribution, IRS, SMC) and health-system access and performance at 2019 levels does not change the transmission intensity. Figure 1 shows that malaria is predicted to continue unabated, with no further decrease expected until 2030 (the endpoint of the model). The slight upward trend in cases and deaths reflects a growing population, rather than an increased incidence of malaria.

Figure 1.

Baseline clinical cases and deaths per year.

5.2.2. Scale-up LLIN and SBCC coverage by 10 percentage points

Figure 2 illustrates the projected clinical cases and deaths with scaled-up LLIN and SBCC with the baseline (where other interventions were held constant). In the LLIN and SBCC scenario, clinical cases fell from 4.17 billion to 3.10 billion, and deaths from 4,823,000 to 3,486,000. Scale-up and better use of LLINs resulted in a projected 1.07 billion clinical cases and 1,337,000 deaths averted cumulatively over eight years.

Figure 2.

LLIN and SBCC scenario versus baseline scenario.

5.3. Cost projections

To account for potential underestimation of the cost of combating pyrethroid resistance and to maintain the effectiveness of LLINs throughout the period, the upper bound range of the cost estimate for the LLIN and SBCC program produced by the SPPf model is used for reporting the main scenario. The medium cost was used for all other cost estimates.

Adding up all the costs of malaria interventions for maintaining the 2019 levels and the resulting costs of treatment to the health system and out-of-pocket expenses for households, the total estimated present value for 2023 to 2030 discounted at 8% is US$ 53.1 billion (min-max range US$ 51.7–54.4 billion). The total cost of the LLIN and SBCC scenario was estimated to be US$ 49.3 billion (min-max range US$ 47.1–50.6 billion) between 2023 and 2030.

Comparing the two scenarios, the incremental costs of scaling up the LLIN and SBCC program is US$ 5.7 billion in total over 7 years discounted at 8%. The undiscounted costs gradually increase by year as more nets are purchased and distributed with social and behavior change communication (see Figure 3).

Figure 3.

Total incremental costs of increasing the coverage of LLINs and SBCC.

The incremental costs for treating malaria cases for the health system and out-of-pocket for households decrease as LLIN and SBCC scale-up reduces the number of malaria cases. Therefore, the total net cost of the LLIN and SBCC scenario is lower than the cost of the baseline scenario.

In the cost–benefit analysis, the cost savings obtained from reduced outpatient and inpatient health-system expenditures due to diminishing cases and reduced out-of-pocket household expenses are added to the benefits. These financial benefits of scaling up LLINs and SBCC will outweigh the expenses for additional LLINs and SBCC in the year 2026. Figure 4 illustrates the total costs of increasing the coverage of LLINs (same as Figure 3) and the total financial cost savings. Costs rise throughout the period of scale-up due to increased investments for LLIN purchase, distribution, and use, while healthcare cost savings increase even more over the entire period as fewer and fewer people get sick.

Figure 4.

Year-by-year comparison of incremental costs for LLIN and SBCC, and expenses saved due to reduction in malaria cases.

5.4. Benefits estimation

In 2023–2030, the LLIN and SBCC scenario will generate economic benefits of US$ 275.4 billion (NPV 8%). The majority of the benefit is derived from life years saved, US$ 225.9 billion, the avoided productivity loss for patients and caregivers adds US$ 41.7 billion in economic benefits, and the avoided healthcare system spending and out-of-pocket expenses for malaria treatment adds financial benefits of US$ 7.7 billion (NPV 8%) (Figure 5).

Figure 5.

Mortality benefits, averted productivity losses, and expenses saved due to reduction in malaria cases from scale-up in LLIN and SBCC compared to baseline.

5.5. Benefit–cost ratio

Implementing the LLIN and SBCC scenario (in addition to the baseline scenario of maintaining coverage) over the period 2023–2030 is estimated to produce a return on investment (BCR) of 48:1 (the high-end model cost range for ITNs was used for a moderate estimate due to the pyrethroid resistance challenges, therefore the BCR range is 48–57). The BCR estimates for the 29 individual countries range from 9 to 128 (see Appendix C).