Introduction

Neglected tropical diseases (NTDs) are a heterogeneous group of helminthic, bacterial, viral, fungal and protozoan infections that cause chronic and debilitating disability [1]. However, research and development to combat NTDs have notoriously been underfunded [2]. NTDs persist in areas where access to clean water, hygienic conditions, and health care are limited. As such, they are most prevalent in low-income countries [1]. Indeed, more than 70% of countries with endemic NTDs are classified as low-income or lower middle-income economies [3]. Infection with an NTD may affect cognitive and physical development and can result in permanent physical disability. Therefore, NTDs do not just coexist in poverty, they further propagate the cycle of poverty by hindering economic potential [4, 5].

Lymphatic filariasis (LF), an NTD, can result in irreversible disability, most often manifested as elephantiasis, lymphedema, and hydrocele [6]. With more than a billion people at-risk and 120 million people thought to be infected across 73 countries [7], LF is estimated to account for 2.74 million disability-adjusted life years (DALYs) (1.73m-4.00m) [8]. When incorporating the mild and moderate depression associated with LF-related disability, the health burden due to LF may be upwards of 5 million DALYs [9].

However, while LF can result in profound morbidity, it is inefficiently transmitted, with an estimated 15,500 infective mosquito bites thought necessary to generate one transmittable infection [10]. LF is also preventable through once yearly treatment with antifilarials distributed through mass drug administration (MDA) [6]. This, coupled with the fact that LF does not have a significant animal reservoir, led the International Task Force on Disease Elimination to classify LF as a potentially eradicable disease [11, 12]. In response, the World Health Organization (WHO) began the Global Program to Eliminate LF (GPELF), which aims for the global elimination of LF by 2020 [6]. In the fifteen years since the inception of the GPELF more than five billion antifilarial treatments have been distributed in 58 LF-endemic countries [13].

Successfully eradicating a disease has innumerable long-term health benefits, and is also a classic example of a global public good [14–16]. Eradicating an NTD, like LF, has additional societal benefits, including improvements towards equity, fairness, and social justice [17]. However, disease elimination and eradication initiatives require substantial social and political commitments, as well as significant financial and economic investments. Given the increasingly intense competition for global health resources, the decision on where to invest funds needs to be based upon solid evidence [18]. In order to provide evidence to decision makers about the investment required to eliminate and eradicate LF, we used a micro-costing approach to analyze the financial and economic costs of interrupting LF transmission in all endemic countries under varying levels of MDA intensity, as well as the subsequent costs of conducting post-MDA surveillance.

Methods

Determination of financial costs

In 2013, the median GDP per capita among LF endemic countries was estimated at 1,245.51 USD, ranging from 226.46 USD in Malawi to 25,140.29 USD in Brunei [29]. Given the highly variable economic conditions across LF endemic countries and lack of expenditure data available for this analysis, we proceeded to estimate unit costs for all local, non-tradable goods and services for each LF endemic country by adjusting detailed expenditure budgets from the LF elimination program in Uganda, a country with a GDP per capita estimated at 571.96 USD [29], by country-specific comparative price levels (i.e., purchasing power parity (PPP) adjusted exchange rates) [30]. Tradable goods were assumed to already be at market value and were thus left unadjusted. Table 1 provides a list of the primary activities considered in calculating the financial costs, as well as the average cost per district in the base case.

Download:

• PPT
  PowerPoint slide
• PNG
  larger image
• TIFF
  original image

Table 1. Average costs per district, base case.

https://doi.org/10.1371/journal.pntd.0005934.t001

Determination of economic costs

Economic costs, which were assumed to encompass financial costs as well as the value of volunteer time and donated pharmaceuticals, were also estimated in order to have a more comprehensive understanding of the projected investment needed to eliminate and eradicate LF [31]. A schematic of the algorithm used for calculating the financial and economic costs is depicted in Fig 1.

Download:

• PPT
  PowerPoint slide
• PNG
  larger image
• TIFF
  original image

Fig 1. Financial and economic costing algorithm.

https://doi.org/10.1371/journal.pntd.0005934.g001

Value of donated pharmaceuticals.

The opportunity costs of the donated drugs used in the GPELF were accounted for by valuing each 400 mg tablet of albendazole at 0.19 USD, 50 mg tablet of DEC at 0.0025 USD, and 3 mg tablet of ivermectin at 0.50 USD, which were the suggested manufacturer prices prior to being donated [32–34]. An additional economic cost of 0.0018 USD was assumed to the value of each tablet for insurance and shipping costs, which also are currently absorbed by the drug manufacturers [34]. While the WHO specifies each treatment to include either 6 mg DEC/kg of body weight or 150 μg ivermectin/kg of body weight plus 400 mg ALB, for the purposes of this global level exercise, we assume all annual MDA treatments to be comprised of one tablet of ALB with either three tablets of ivermectin or seven tablets of DEC.

Value of volunteer time.

The value of donated time was evaluated by correlating the time CDDs were presumed to volunteer under each scenario with country-specific or, when necessary, region-specific daily per worker agriculture wage estimates taken from the World Bank’s World Development Indicators Online, inflated to 2012 [35]. Two CDDs were assumed to be sufficient to dispense MDA in each village [36]. Drawing from the results of previous time studies, CDDs were assumed to volunteer 5.5 days on mobilization and sensitization, 4.6 days conducting pre-MDA census activities, and 17.8 days on drug distribution [37].

Results

The total financial investment to implement the elimination scenario is projected at 929 million USD (891m-965m). To expand the campaign to all endemic countries at an average rate of scale-up (eradication I scenario) would require 1.29 billion USD (1.24bn-1.34bn), an increase of about 360 million USD (346m-374m) over the elimination scenario (Fig 2). The decrease in scenario duration inherent in the eradication II scenario (intensified scale-up) comes with decreased costs, estimated at 1.27 billion USD (1.22bn-1.32bn), while instantaneously scaling up MDA to all LF endemic countries is projected to require an investment of 1.24 billion USD (1.18bn-1.28bn). The AFRO region accounts for 62–68% of the financial costs, with Southeast Asia requiring between 22–30% of the projected investment (Table 2).

Download:

• PPT
  PowerPoint slide
• PNG
  larger image
• TIFF
  original image

Fig 2. Incremental financial costs (elimination scenario as comparator).

https://doi.org/10.1371/journal.pntd.0005934.g002

Download:

• PPT
  PowerPoint slide
• PNG
  larger image
• TIFF
  original image

Table 2. Total financial costs by region (in millions USD).

https://doi.org/10.1371/journal.pntd.0005934.t002

Providing MDA to the entire at-risk population immediately, as assumed under the eradication III scenario, requires a significant initial investment, but within 10 years’ time, the annual cost of implementing the scenario becomes less than the alternatives (Fig 3). The sharp increase in financial costs four years from the start of the eradication III scenario corresponds to the start of MDA to all at-risk populations in countries that were previously delayed due to mapping.

Download:

• PPT
  PowerPoint slide
• PNG
  larger image
• TIFF
  original image

Fig 3. Financial costs by year, discounted at 3%.

https://doi.org/10.1371/journal.pntd.0005934.g003

The average unit financial cost for undertaking each of the scenarios ranges from 0.27 USD in the elimination scenario to 0.31 USD in both the eradication II and III scenarios. However, as the scenarios progress, the unit costs increase substantially. This is due to the fact that the number of people to be treated (the denominator of the estimate) decreases, though the cost associated with some of the core activities—including coordination and strengthening partnerships, administration, and data management—are assumed to remain relatively constant. As an example of this, Fig 4 depicts the unit financial costs seen under eradication I, which, by 2050, extend to more than 1,700 USD per person treated.

Download:

• PPT
  PowerPoint slide
• PNG
  larger image
• TIFF
  original image

Fig 4. Financial cost per person treated, eradication I.

https://doi.org/10.1371/journal.pntd.0005934.g004

Capacity strengthening proves to be the most costly activity, representing between 53–55% of the overall financial costs, while advocacy and communication (22–24%); ongoing surveillance (6%); monitoring, evaluation, and supervision (8%); and drug delivery (9%) account for most of the remaining costs (Table 3).

Download:

• PPT
  PowerPoint slide
• PNG
  larger image
• TIFF
  original image

Table 3. Percentage of financial costs by activity.

https://doi.org/10.1371/journal.pntd.0005934.t003

Our assessment of the financial costs of treating a population of 1 million at-risk for LF with and without L. loa coendemicity in the Democratic Republic of Congo indicates that areas of meso L. loa are anticipated to only result in an increase in monitoring and evaluation, thereby having little effect on the overall costs. In comparison to a population of comparable size without L. loa, hyper L. loa endemicity is associated with an increase of approximately 15% in the overall costs of the program due to the increase in data management, drug delivery, and monitoring and evaluation activities.

When the economic costs are considered, the costs of all scenarios are substantially higher (5.2 billion USD for the elimination scenario). Extending the coverage to all endemic countries is estimated to require around 7.9 billion USD (7.5bn–8.0bn), or 45% more than under the elimination scenario (Table 2). Depending on the scenario, between 48–53% of the economic costs are due to the value of the donated drugs (Fig 5).

Download:

• PPT
  PowerPoint slide
• PNG
  larger image
• TIFF
  original image

Fig 5. Economic costs by component, discounted at 3%.

https://doi.org/10.1371/journal.pntd.0005934.g005

Discussion

This study used a micro-costing approach to provide estimates on the financial and economic investments required to eliminate and eradicate lymphatic filariasis. Across the eradication scenarios, faster rates of MDA scale-up are associated with decreased costs (1.2 billion USD in the eradication III scenario versus 1.3 billion USD in the eradication I scenario). The projected economic costs of eradication range from 7.5 billion to 8.0 billion USD, about half of which is due to the value of the donated drugs. These results serve to further highlight both the importance of higher rates of MDA scale-up as well as the crucial partnership between the GPELF and the drug donation programs. These projections may also be important in order to convince pharmaceutical companies to continue donating the drugs necessary to eliminate LF. While the cost to eliminate LF is less than that to eradicate, it must be recognized that deciding to pursue elimination rather than eradication signifies the continuation of LF-related costs indefinitely, and comes at a health burden to populations that remain untreated.

With a dearth of evidence on the costs of implementing morbidity management programs [39], and given that the aim of our study was to assess the costs of interrupting transmission of the causative agent of LF, we did not include the costs of morbidity management in our estimates. Had morbidity management been included in the scenarios, the overall costs would have certainly increased, though presumably with reduced times to eradication associated with reduced morbidity management costs.

Our cost estimates for reaching LF elimination and eradication inherently assume that the GPELF strategy, if carried out sufficiently, will lead to the interruption of LF transmission in all areas. In our analysis we have also assumed that accelerating the rate of MDA scale-up will result in decreased marginal costs (the cost of one additional treatment) by assuming there to be spare capacity among fixed costs (cars, equipment, etc.). However, in some areas, additional capital may be required and, therefore, marginal costs would increase. In such a case, accelerating eradication might not be the most cost-saving option in the short term.

Additionally, our analysis does not take into account the cost for certifying elimination on a country level nor the activities involved in globally assessing whether eradication has been achieved. As currently experienced by the Global Polio Eradication Initiative, the costs for finding and ascertaining the last cases to reach eradication are substantial [40, 41]. While our analysis does not consider the costs associated with finding the last cases of LF infection, the unit costs for the final populations treated in the eradication I scenario are orders of magnitude higher than the average unit costs—beyond 1,700 USD per person treated. This is due to fewer people to be treated (the denominator of the calculation) rather than higher overall costs of running the program. It has previously been recognized that when the cost of disease eradication comes within reach, the unit costs associated with prevention become decreasingly attractive. However, at that point, it is crucial not to lose momentum, nor investment, otherwise there is great risk of failure [42, 43]. Developing realistic cost projections from the start of the program could help mitigate the risk of donor fatigue towards the endgame of disease eradication. Further, when disease eradication is within reach, shifting the focus from unit costs per person treated to the costs per case averted may also help to sustain global commitments [44].

Our analysis found a 15% increase in financial costs to treat an area of hyper L. loa endemicity in the DRC. This estimate does not take into account increases in advocacy and communication in these areas, which may be necessary to achieve the targeted levels of coverage. Further, the costs for medical transport and additional medications that might be needed to treat patients suffering from severe adverse events have not been incorporated in this analysis. Costing studies for such post-MDA response activities have not previously been carried out, though such costs are likely to vary by the incidence of SAEs, the geographic location, and the intensity of response required. If substantial and significant response is required in many areas, implementing the eradication scenarios would certainly result in higher costs than projected in this baseline study.

A number of methodological uncertainties in our study must be mentioned. Country-specific cost data was mostly unavailable and, consequently, was largely imputed from Ugandan data. By extrapolating cost data across countries and regions, we inherently made the assumption that each LF endemic country implements the GPELF strategy as in Uganda (for example, using volunteer CDDs, similar amount of trainings, etc.). Moreover, we assumed that the number of resources required to carry out the PELF for a certain population in Uganda remains relatively constant (varying by +/-10%) both across time and across countries. As described above, the variation assumed in our model was included probabilistically in both unit costs and quantities.

We chose to have modest uncertainty assumptions, assuming a variation of 10% probabilistically for both costs and quantity estimates. Though 10% variation has a large impact on the individual unit costs, the impact on total costs was less than expected. As such, we chose to increase the variability to 30% during additional uncertainty analyses. However, this resulted in only a modest increase in the uncertainty intervals (see S3 Table and S1 Fig). It is, therefore, likely that the narrow uncertainty intervals found in this analysis are the result of assumptions made during earlier analyses to estimate the number of annual rounds of treatment required to interrupt LF transmission. Namely, we assumed the required number of annual rounds of treatments to be fixed at the upper bound of our model-based projections in order to provide as conservative an estimate as we could. Varying the estimated number of treatments across their credible interval may have instead provided a more complete picture.

Additionally, our MDA round estimates are considerably longer than the 4–6 years currently specified by the GPELF, instead ranging from 6 rounds of MDA in areas where Anopheles spp are the primary vector, treatment is through DEC+ALB, and the baseline MF prevalence is 5–10%, to 15 rounds in areas where Culex spp are the primary vector, IVM+ALB is the treatment provided and the baseline mf prevalence is 15–20% [19, 45]. These extended MDA durations have a number of implications, but most relevant in terms of costing is that both the financial and economic estimates are likely to be much more conservative than estimates developed using an assumed 4–6 years of MDA, since the overall costs are heavily dependent on the duration of the program. Our approach, therefore, resulted in conservative estimates at the cost of capturing more realistic uncertainty intervals. In our baseline analysis we chose to make rather crude assumptions of independence between inputs as well as assumptions of the uncertainty and distributions of the model inputs. For the purposes of assessing the relative costs of scaling up MDA at different rates we felt that this was sufficient in order to avoid introducing increased complexity, and consequently, further uncertainty in our model.

Ideally, LF-specific expenditure data would have been collected in all 72 LF-endemic countries. Having such data would have also allowed for an important assessment on the variance of costs from the beginning stages of implementation towards the ‘end game’ of reaching surveillance. However, 14 of these endemic countries have never carried out MDA for LF [13]. Moreover, undertaking a study that accurately collected such data would potentially begin to rival the time and cost of running the PELF programs in many of the countries to begin with. Further, a recent study by Brady et al. found no significant difference between grant budgets and actual expenditures when analyzing costs of Transmission Assessment Surveys [46]. Thus, despite the large number of assumptions inherent in our approach, our costing model allowed for the development of comparable cost estimates on a country, regional, and global level.

We conducted a probabilistic sensitivity analysis in order to overcome some of the limitations inherent in our costing parameters. In order to provide more robust estimates on the costs of achieving elimination and eradication, costing studies in areas with the highest burden of LF could be undertaken in order to reduce the level of uncertainty in costs. While improved cost data could be used to inform policy and improve planning, the cost of acquiring additional data should be weighed against the value of such data [47]. An additional use for the results found in our economic analyses, therefore, could be to assess the value of additional investments for LF, including the collection of expenditure data, as well as investments in diagnostics, drugs, and surveillance tools to help advance LF eradication.

Our findings on the costs to eliminate and eradicate LF represent very achievable investments. Our cost estimates would have likely been even lower, though, if we assumed some level of integration or cost-sharing between other disease initiatives. Many countries have, in fact, integrated similar activities across vertical programs, including onchocerciasis, trachoma, and schistosomiasis, and others have paired drug delivery for MDA with other community distribution campaigns, including insecticide treated nets (ITNs) for malaria and ongoing vaccine programs. In so doing, the overall costs per program, indeed, generally decreased and efficiency reportedly improved [2, 48–51]. However, it is important to note that, historically, other disease elimination programs that have proven to be successful have generally proceed as vertical programs [52]. Additionally, given the cost involved with each round of MDA, it could be cost saving to undertake TAS sooner in order to assess whether the interruption of transmission had been achieved. However, this approach could prove to be a difficult balance, since prematurely stopping MDA could result in resurgence of infection [53], ultimately leading to an increase in the cost of reaching eradication.

We further did not include the cost of vector control in this costing study, assuming instead the cost to be covered by other disease control programs (for example: malaria or dengue control). While vector control is not a core strategy of the GPELF in non l. loa endemic areas, previous studies have indicated that including vector control could decrease the number of MDA rounds needed to interrupt transmission, thereby reducing the overall duration of the program [54], and potentially also the cost.

A comparison of our costs against other costs is important for validation, though challenging due to differing methodologies. Though not inclusive of overhead costs, a study from two states in Nigeria found the cost associated with conducting MDA for the prevention of LF to be between 0.02 USD and 0.12 USD per treatment delivered [55]. A multi-country costing study conducted by Goldman et. al found unit financial costs to range from 0.06 USD in Burkina Faso to 2.23 USD in Haiti [34]. A separate study in Haiti reported the cost per person treated to be 1.44 USD [56]. Thus, in comparison to other MDA costing studies, our average unit financial cost estimates are well within the range of previously reported studies.

Knowing the global costs of the program will help decision makers assess the feasibility and rationale of investing in LF eradication, while simultaneously helping to facilitate planning and the development of strategies and policies. However, successfully eradicating LF depends on more than the monetary investment. Political will, continued community ownership, and the feasibility of the campaign all need to be taken into account [57]. However, if successful, disease eradication not only results in innumerable long-term health benefits, but also savings to the health system, gains in productivity, and improvements in social justice [17, 58]. The decision of whether disease eradication should be pursued, therefore, needs to be approached with a comprehensive understanding of the many complex issues at play.

Supporting information

Acknowledgments

We are grateful to the members of this project’s technical advisory group for sharing their insights and suggestions that improved this work throughout the process: (alphabetically listed): Jan Agosti, Uche Amazigo, Maria-Gloria Basáñez, Moses Bockarie, Stephen Cochi, Ciro de Quadros, David B. Evans, Amadou Garba, Lance Gordon, Lee Hall, Ian Hastings, Mamoun Homeida, Adrian Hopkins, Julie Jacobson, Dean Jamison, Tom Kanyok, Claudio Lanata, Mike J. Lehane, Julia Lupp, Mwele Ntuli Malecela, David Molyneux, Hassane Mahamat, Eric Ottesen, Regina Rabinovich, J. H. F. Remme, Jose Ignacio Santos, Lorenzo Savioli, Pere Simarro, Wilma Stolk, Mark Taylor, Stephen Torr, Hugo Turner and Laurent Yameogo. We also greatly appreciate the statistical support provided by Dr Felix Day, Dr. Fiona Vanobberghen, and Dr. Peter Pemberton-Ross.