Background

Lymphatic filariasis (LF), is a human disease caused by parasitic helminths (Wuchereria bancrofti, Brugia malayi and Brugia timori). These filarial worms are transmitted via infected mosquitoes.

There are 73 endemic countries at-risk of LF, and before widespread control approximately 120 million people worldwide were infected - of whom 40 million were suffering from overt clinical disease [1, 2]. Clinical disease can manifest as painful severe swelling due to lymphedema (an accumulation of lymphatic fluid generally in the limbs), hydrocele (fluid accumulation in the scrotal sac) and episodes of acute adenolymphangitis [1, 2].

In 1997, the World Health Assembly passed Resolution 50.29, calling for the elimination of LF as a public health problem [3]. Following on from this, in 2000 the World Health Organization (WHO) established the Global Programme to Eliminate Lymphatic Filariasis (GPELF) with the goal of eliminating the disease as a public health problem by 2020 [4, 5]. The programme has two parallel goals [4, 5]:

  1. (i)

    To use community-wide annual mass drug administration (MDA) to interrupt transmission, using a combination of albendazole and ivermectin in areas co-endemic with onchocerciasis, and albendazole and diethylcarbamazine (DEC) elsewhere.

  2. (ii)

    To alleviate suffering by managing morbidity and preventing disability in clinical LF patients.

These goals are supported by the WHO’s 2020 Neglected Tropical Disease (NTD) Road Map [6] and the London Declaration on NTDs [7].

Some countries are acknowledged as having eliminated LF as a public health problem [8]. However, it is recognised that we are not currently on track to meet these goals in many settings, and achieving elimination may require alternative approaches [9,10,11].

One particular challenge facing LF elimination efforts in Africa is areas co-endemic with onchocerciasis and the tropical eye worm Loa loa (which causes loiasis). Traditionally, onchocerciasis is managed with annual or biannual (twice yearly) ivermectin treatment. However, due to the potential for severe and often fatal encephalopathic reactions to ivermectin in patients with high L. loa microfilaria loads, this therapeutic approach is not permissible in many loiasis co-endemic areas [12]. To facilitate LF elimination in these problematic co-endemic zones of central Africa, the WHO has proposed an alternative strategy that involves biannual albendazole monotherapy together with the expanded use of bed nets [13]. It is also important to restate that DEC can cause severe adverse reactions in individuals with heavy Onchocerca volvulus infections and that it is not used in onchocerciasis-endemic areas [14, 15].

As we move forward towards elimination, we need to better understand the cost-effectiveness of both the current and the potential alternative control strategies. The aim of this paper is to provide a systematic review of economic evaluations which have already been conducted for LF interventions and to summarise the key knowledge and research gaps in this area.

Systematic review

Search strategy and methodology

A systematic review of the literature was conducted in December 2016 using the PubMed (MEDLINE) and ISI Web of Science electronic databases. Variants of the following search terms were used to find relevant papers: lymphatic filariasis, cost(s), cost-benefit, cost-effectiveness, economic(s), economic evaluation. No date or language stipulations were applied to the searches. A more detailed summary of the search terms and the PRISMA checklist are supplied in Additional file 1.

The titles and abstracts of all the identified papers were examined initially for relevance and then the bibliographies of papers suitable for inclusion were scanned for studies not originally retrieved from the databases. The full selection process is outlined in Fig. 1. This process was performed in duplicate.

Fig. 1
figure 1

Decision tree outlining the inclusion and exclusion of the identified studies. Some studies reported both cost-benefit and cost-effectiveness estimates. Several ‘grey literature’ texts (including policy reports) which were not found within the databases, were also identified (using Google Scholar and the bibliographies of other papers). A PRISMA checklist is provided in Additional file 1

Full size image

Summary of the identified studies

We identified 12 different primary sources reporting the results of economic evaluations of LF interventions. A summary of the studies is presented in Tables 1, 2. The majority of the estimates were evaluating MDA, though it was not always clear which drug combination was being investigated. Only two studies were identified that investigated the cost-effectiveness or cost-benefit of morbidity management strategies (Tables 1, 2).

Table 1 Summary of the identified cost-effectiveness analyses
Full size table
Table 2 Summary of the identified cost-benefit analyses and estimates of the economic benefits of interventions
Full size table

Due to the different aims of the identified studies, a variety of different effectiveness measures were used by the different analyses - including the cost to elimination, cost per disability-adjusted life year (DALY) averted, the benefit-cost ratio, the cost per case cured. Several studies [2, 16, 17] used DALYs averted as the effectiveness measure to quantify the health impact of MDA - therefore their outcomes are directly comparable to each other. The cost-effectiveness ratios varied depending on which costs were included and the time horizon of the analysis (Table 1). However, they all would class MDA for LF as either cost-effective or highly cost-effective based on the thresholds for low-income countries established by the World Bank (≤ US$ 251 per DALY averted = cost-effective [18], and ≤ US$ 42 per DALY = highly cost-effective [18] (adjusting for inflation - 2016 prices) [19]). Stone et al. [20] also used DALYs averted as an effectiveness metric, and estimated the incremental cost-effectiveness of three different scenarios for accelerating the rate of MDA coverage scale-up (Table 1). Within this study, they also estimated the savings to the health system and the gains in worker productivity (Table 2).

Chu et al. [21] and Turner et al. [22] projected that the MDA provided under the GPELF would result in substantial economic benefits. The clear majority (> 80%) of this estimated economic benefit resulted from the prevention of the potential productivity/income losses associated with LF morbidity (indirect costs, Table 3). These studies were based on the same framework, and an explanation for the differences in the results is outlined in Turner et al. [22]. Stillwaggon et al. [23] also found notable economic benefits and productivity gains resulting from a community-based lymphedema management programme in India (Table 2).

Table 3 Glossary
Full size table

Other studies have also highlighted the importance of the productivity losses associated with LF morbidity [24, 25]. For example, it has been estimated that in India, between 3.8–8.0% of the potential male labour input was being lost due to LF morbidity [26, 27] - subsequently valued at US$ 704 million per year (1995 prices) [28]. A similar value has been reported for Ghana, were over 7% of potential male labour was estimated to be lost due to chronic LF [29]. It is noteworthy that non-filarial elephantiasis (podoconiosis) has also been found to be associated with significant productivity losses [30].

It should be highlighted that these types of economic burden/benefit estimates are highly dependent on assumptions regarding the effect of clinical disease on productivity [21, 31], the number of years of productive life lived with clinical disease, and employment rates. In addition, when comparing these estimates, it is particularly important to consider which method and wage source has been used to value the productivity losses, as these can be highly variable even when referring to the same type of profession (highlighted in Additional file 1: Table S1). Furthermore, it is important to note whether lost wages were adjusted for future inflation or for future real wage growth (such as in [23]) as this could result in higher economic benefits/burden estimates. All of the studies that we found investigating the economic benefits resulting from LF interventions used the human capital approach to value the prevented productivity losses. This takes the patient’s perspective for valuing lost productivity and therefore counts any hour not worked by the patient as an hour lost - not accounting for the possibility that absent workers may be replaced (Table 3) [32]. It is worth noting that an alternative method known as the friction cost approach takes the employer’s perspective and therefore only counts as lost, the hours not worked before another employee takes over the patient’s work [32]. If this approach had been used, the estimated economic benefits could have been significantly lower [33]. There is continued debate regarding which approach is most appropriate [32]. Interestingly, the second US public health service panel on “cost-effectiveness in health and medicine” recently recommended using the human capital approach [34].

Only five cost-effectiveness estimates were identified which evaluated alternative interventions to the currently recommended strategies (outlined in Table 1). Furthermore, no studies were found that evaluated interventions specific for loiasis co-endemic areas.

The majority of the estimates had either no sensitivity analysis conducted or only univariate sensitivity analysis (where the impact of changing one parameter at a time is evaluated). The two main exceptions to this were Stone et al. [20] and Stolk et al. [35].

The assumed costs of mass drug administration

Delivery costs

When comparing the different studies, it is important to consider that there is variation in the assumed delivery costs of MDA, even for estimates pertaining to the same country. The majority of the studies were based on the same relatively small number of costing studies (Tables 1, 2), and several of the cost-effectiveness/cost-benefit estimates were not based on published costing studies/data. This meant it was not always clear which costs were being included in the analyses, at times making it difficult to judge the generalizability of these studies.

It is also important to recognise whether or not the studies are using financial or economic cost data (Table 3). The following were the studies that clearly stated that they are using economic costs for the investigated intervention in at least a subset of the analysis [17, 20, 21, 35,36,37]. However, even in these cases it was not always clear which economic costs were being included. For example, the economic value of the volunteer community drug distributors’ time was not always included within the economic costs.

Drug costs

Depending on the perspective of the analysis, the value of the donated drugs may also be included as an economic cost. Several of the identified studies considered the economic value of the donated drugs within their economic evaluation - which increases the intervention’s cost (Table 4) and therefore decreases the estimated cost-effectiveness/cost-benefit (Table 1). However, it is important to note that there was variation in the assumed economic value of the drugs, and in some cases the official figures have changed over time. For example, in 2009 GlaxoSmithKline changed their valuation of donated albendazole to US$ 0.045 per tablet from $0.19 per tablet (GSK, unpublished) [38]. A summary of the economic value of the drugs assumed by Turner et al. [17] is outlined in Table 5.

Table 4 Summary of the average treatment costs of the GPELF (2000–2014)
Full size table
Table 5 Drug costs and their economic value
Full size table

Turner et al. [17] found that when only considering countries using the ivermectin and albendazole regimen, that the GPELF would no longer be classed as cost-effective when using the World Bank thresholds (although only marginally and it remained highly cost-effective based on the WHO-CHOICE thresholds [39]). This is due to the higher economic value of ivermectin (Table 5). Despite this result, the GPELF was found to be clearly cost-effective as a whole [17]. Stolk et al. [35] also found that including the value of the donated drugs, decreased the potential economic benefits of increasing the treatment frequency to twice a year. It should be noted that it is difficult to estimate the true economic value of these donated drugs [17]. Furthermore, it is important to consider that the foundation of the GPELF is based on the long-term and sustained commitment of drug donations of ivermectin and albendazole for as long as needed until the elimination of LF is achieved [40], and the majority of the required DEC is being donated up to 2020 (Table 5). It should also be noted that drug donations are the primary basis for many NTD MDA programmes.

Limitations

A potential source of bias within this review is that the employed search strategy could not always retrieve economic evaluations outside of published papers (i.e. grey literature such as policy documents and reports). This bias was minimised by searching the bibliographies of selected studies and the use of Google Scholar. This resulted in four publications being added to the initial compilation.

It should be noted that there could be a degree of publication bias, with economic evaluations with negative or unfavorable results being less likely to be published.

The cost-effectiveness of control versus elimination

When comparing the different studies, it is important to consider the time horizon used for the analysis and whether the study is evaluating morbidity control or the elimination of transmission. Michael et al. [36] found that a MDA programme’s cost per case cured can be higher when its aim is to eliminate transmission compared to when its aim is only morbidity control. The analysis highlighted that a MDA programme’s peak cost-effectiveness can occur at a point before full disease control is achieved. This is because, as the prevalence of infection decreases, the incremental cost per additional infection cured can increase steeply for each subsequent MDA round (illustrated in Fig. 2). However, depending on the time horizon and assumptions of the analysis, it is possible that an elimination campaign will become more cost-effective in the long-term and potentially even cost-saving (Fig. 2). For example, Remme et al. [16] found that with a 30-year time horizon, an elimination strategy would be more cost-effective than a morbidity control strategy (where transmission is brought to low levels but not interrupted). This was because, though an elimination strategy is more expensive to run, after elimination has been achieved, MDA and its associated costs stop. In contrast, for the control scenario, transmission is not broken so the costs associated with MDA are incurred for the full-time horizon (Table 1). Due to this, the control scenario ultimately has a higher total cost over the 30 year time horizon (even though it was initially cheaper). It is important to highlight that in these studies, the potential cost savings resulting from achieving elimination/eradication are not infinite [20, 41], as the costs being considered are restricted within the study’s time horizon and are often discounted into the future.

Fig. 2
figure 2

A theoretical diagram of the potential cost, effectiveness and cost-effectiveness of a mass drug administration programme before and after elimination. Note that this figure is illustrative and not based on primary data. The time horizon for the cost-effectiveness analysis is the duration over which the outcomes and costs are calculated. Both the cost and effects are being discounted into the future at a rate of 3%

Full size image

These principles are highlighted in Fig. 2. In this hypothetical example, the cumulative cost of the programme steadily increases over time but then increases at a faster rate during the final phase of the programme - due to the costs associated with scaling-up into harder-to-reach areas, and the cost of the surveys needed to confirm the programme can be stopped, i.e. post-MDA surveillance. After elimination is certified, the cumulative costs stop increasing. In contrast, the cumulative effectiveness of the programme also increases over time, but shows a degree of diminishing returns (because as the intervention progresses fewer cases are prevented with each subsequent MDA round). As a result of these relationships, the cost-effectiveness of the programme is not constant and is highly dependent on the time horizon of the analysis. In this example, as the time horizon is increased, the cost-effectiveness will initially increase during the first phase of the programme but then start to decrease due to the diminishing returns in effectiveness (as the level of infection/transmission is reduced) and then decrease further when the costs rise during the final phase of the programme. After elimination is certified, the cost-effectiveness will steadily increase with the time horizon, as the costs have stopped but the benefits continue to accumulate (though they are discounted into the future). In this context, it is important to highlight that instantaneous cost-effectiveness ratios (i.e. comparing the costs and benefits at one selected time point) are not particularly informative, and it is the total cost and total effect for the assumed time horizon that should be evaluated.

It is noteworthy that alternative interventions aimed at accelerating and sustaining elimination may only have small “incremental health gains” but a large influence on the programme’s overall total cost (as seen for onchocerciasis [42]). In such cases, an incremental cost-effectiveness ratio in terms of the cost per additional DALY averted may not reflect the true value of these novel interventions. Kastner et al. [41] also highlighted that the number of DALYs averted may not be the best measure to assess the possible benefits of disease eradication - as the long-term consequences and broader benefits are not necessarily fully captured. A cost-benefit analysis may be more useful in capturing these benefits more fully.

Areas that need further research

The results of the review indicate that the standard LF control strategies are consistently found to be cost-effective or cost-saving. However, there are some important inconsistencies and research gaps that need to be addressed as we move forward towards the 2020 goals, particularly regarding the evaluation of alternative elimination strategies.

In the following section we outline several key research needs.

Settings co-endemic with loiasis

Due to the potential for life-threatening adverse events in intensely infected L. loa patients, alternative strategies to address the elimination of LF where loiasis is prevalent have been proposed [12]. In 2013, the Strategic and Technical Advisory Group for NTDs (STAG) recommended albendazole monotherapy combined with coordinated vector control in areas co-endemic with loiasis [13]. The impact of this albendazole monotherapy strategy is currently being evaluated in parts of central Africa [13, 43] as is a “Test-to-Exclude” from treatment approach [44]. However, none of the identified economic evaluations focused on strategies for these co-endemic areas, and policy for these settings is a notable research gap for LF elimination. This gap is not necessarily surprising, as currently the main objective and focus for these areas is still to find strategies that work and are safe.

It should be highlighted that the novel strategies (such as the “Test-to-Exclude” from treatment approach) in these settings could be more expensive than conventional MDA strategies. It will be important to consider the value of these interventions not only in reducing the burden in co-endemic areas, but also in their capacity to help enable the global elimination goals to be reached and the reduced risk that sustained transmission in these co-endemic settings results in the re-establishment of transmission in neighbouring areas.

It is important to consider that loiasis is a vector-borne disease (transmitted by Chrysops spp.) and another potential solution for these areas is to use vector control to reduce its transmission - reducing the overall burden of L. loa in these population and hence to risk of the severe adverse events associated with high microfilaria loads [45].

Morbidity management strategies

A key element of the WHO’s strategy to combat LF involves increased morbidity management and disability prevention activities [4, 46]. However, we identified only two studies in this area - one on lymphedema management and one on hydrocele surgery (Tables 1, 2).

To allow for more economic evaluations of LF morbidity management strategies (across a range of settings), more data are urgently needed assessing their costs, resource requirements, clinical effectiveness, and the incidence of complications/relapse for the different potential techniques.

Methodological issues and data needs

Treatment delivery costs

The costs of MDA delivery vary in different regions (highlighted by a multi-country costing study by Goldman et al. [47] and the systematic review by Keating et al. [24]). Understanding this variation and quantifying its impact is an important research gap for future studies - as it potentially affects the generalisability of cost-effectiveness/cost-benefit analysis [48]. In particular, one of the key drivers in the variation in delivery costs is the economies of scale associated with MDA [49,50,51] - the reduction in the cost per treatment as a result of increasing the scale of the programme (Fig. 3). However, the majority of studies identified in this systematic review assumed a constant cost per treatment and did not take into account the potential changes over time or scale (Tables 1, 2). The economies of scale associated with MDA are vital to consider when projecting the future costs of LF control, as well as when estimating the incremental costs of adopting alternative strategies. Furthermore, additional clarity regarding which costs are being included in the analysis will be important in future studies.

Fig. 3
figure 3

Observed economies of scale and scope associated with preventive chemotherapy. Data adapted from Evans et al. [51]. Costs are in 2008 and 2009 US$ prices

Full size image

There are few costing studies investigating alternative strategies (such as increasing the treatment frequency [52]) [53]. In these cases, it is vital to consider the generalizability of the estimated difference in cost between the alternative and standard strategies across different programmatic settings. This is particularly significant if the costs of the alternative strategy have been estimated within a randomised control trial.

It should be noted that the unit delivery costs for the programmes will likely increase considerably as they approach the “last mile” towards elimination. This is because of the increase in the costs resulting from expanding the programmes to target harder-to-reach areas/groups (diseconomies of scale) and costs relating to conducting transmission assessment surveys (TAS). This has been seen in other interventions - particularly elimination campaigns [54,55,56,57]. Furthermore, it is important to note that as programmes start closing down implementation units, their costs will not decrease linearly (Fig. 3).

Programme integration

A notable research gap is the lack of understanding of the costs of integrated NTD control [24, 58] and how integration may influence the cost and cost-effectiveness of implementing different control strategies (economies of scope) (Fig. 3). Evans et al. [51] found that integrating MDA for LF with that for schistosomiasis, STH and onchocerciasis in Nigeria reduced the cost per treatment by 41% (not including the drug and overhead costs). The role and impact of this economies of scope should be considered further in future analyses.

Ancillary benefits of LF control programmes

The GPELF uses broad-spectrum antiparasitic drugs, and consequently, it has substantial auxiliary benefits on other parasitic diseases such as onchocerciasis, scabies, and the soil-transmitted helminths (STH) (described in more detail in [2, 22]). These auxiliary benefits are not typically included in economic evaluations of LF control programmes, which therefore underestimates their cost-effectiveness and cost-benefit. Furthermore, the end of LF-related MDA programmes is likely to have a considerable effect on STH transmission and prevalence, and this potentially increased risk of STH recrudescence needs to be evaluated [59].

Metrics and cost-effectiveness thresholds

The wide range of effectiveness metrics used by the different studies hinders direct comparison of their results. This has been noted for other NTDs as well [50].

The ideal choice of metric for evaluating control strategies will often be the number of DALYs averted, as it allows the cost-effectiveness estimates to be directly compared to that of other healthcare interventions. This makes it possible to have standardised thresholds for policymakers, which class whether or not an intervention is cost-effective - which is rarely possible when reporting a disease specific cost per infection case averted. However, it is important to restate that, as discussed in the “The cost-effectiveness of control versus elimination” section, DALYs averted and incremental cost-effectiveness ratios may not reflect the true value of alternative interventions aimed at accelerating and sustaining elimination or disease eradication. In addition, DALYs are not without limitations, and their design contains inherent flaws that fail to acknowledge the implications of local context on disease burden [60], which is particularly important for NTDs which are most prevalent in poor populations. Furthermore, clinical LF has an impact on the quality of life for patients as well as their families, which is not fully captured by a DALY weight. It is also important to consider that due to a lack of data, features of the disease burden are ignored. For example, all of the current DALY estimates for LF assume it is not associated with any excess mortality (which could underestimate its burden). It is also worth noting that Ton et al. [61] found that accounting for the mental illness that can be experienced by LF patients and their caregivers significantly increased the DALY burden estimates related to LF. This has not currently been included in any the economic evaluations of LF control, which therefore underestimates its cost-effectiveness/cost-benefit. Non-filarial elephantiasis (podoconiosis) has also been found to be associated with depression [62].

There is debate and uncertainty surrounding the most appropriate cost per DALY averted thresholds for defining which interventions are classed as cost-effective [63, 64]. It should be noted that the thresholds established by the World Bank [18] are more conservative than the thresholds set by WHO-CHOICE [39] (a cost per DALY averted > 3 times the national gross domestic product (GDP) per capita = not cost-effective; between 1 and 3 times the national GDP per capita = “cost-effective”; and < 1 times the national GDP per capita = “very cost-effective”). However, these WHO thresholds are now widely considered to be too high [63,64,65,66] and are rarely used for NTD interventions. A recent analysis indicated that a cost per DALY averted threshold closer to ½ the national per capita GDP would be more appropriate for low-income countries [67]. Interestingly, a subsequent study used a threshold of US$ 200 per DALY averted to identify priority interventions for consideration in low-income countries [68].

Reporting standards for economic evaluations

Elements of the studies were not always clear, and at times important pieces of information were not reported. Moving forward it would be beneficial if studies were to adhere more to standardised guidelines (such as CHEERS [69]) regarding what should be reported within the manuscript.

Evaluation of alternative interventions

Though we found five cost-effectiveness estimates relating to alternative strategies to the standard dual drug MDA strategy (Table 1), there are still notable research gaps in this area. In particular, the following are some key interventions that will require further economic evaluation in the future.

Anti-Wolbachia therapy and other novel drug treatments

A novel approach for treating LF involves using tetracycline antibiotics (such as doxycycline), to target the parasites Wolbachia endosymbionts which are essential for worm fertility and survival [70, 71]. A six-week course of doxycycline has been reported as a safe and well-tolerated treatment for LF, with significant activity against the adult worms [71]. Treatment also improves mild to moderate lymphoedema independent of ongoing infection [72]. An important benefit of this intervention is that it can also be used to treat onchocerciasis and is safe in loiasis co-endemic areas (as L. loa do not have any Wolbachia). One of the primary goals of the Anti-Wolbachia Consortium (A-WOL) is to identify drugs or regimens that reduce the period of treatment from weeks to days [71].

Other potential macrofilaricides should also be evaluated if they become available [73,74,75,76,77,78,79].

Triple drug administration

Triple drug administration with ivermectin, albendazole and DEC (IDA) has been shown to keep participants free of microfilariae for up to two years after treatment [80]. In contrast, within the same study over 90% of the control group (who received the standard dual drug therapy) tested positive for microfilaria after only one year [80]. This shows that IDA is a more effective treatment strategy and a potential method for accelerating transmission elimination (this is supported by mathematical modelling studies [81]). However, this strategy is not currently applicable to most of sub-Saharan Africa, as DEC is non-permissible for use in onchocerciasis endemic areas, and ivermectin is not recommended where intense loiasis transmission occurs [15]. Alternative approaches to manage these programmatic exceptions have been proposed [15, 44]. For example:

  1. (i)

    A Test-to-Exclude from treatment strategy is currently being evaluated in loiasis-endemic areas [44]. However, were this strategy to be widely adopted, an increase in operational costs of the LF elimination strategy would be expected.

  2. (ii)

    Pre-treatment with ivermectin in onchocerciasis endemic areas followed by the IDA regimen is also being considered (a “pretreat and treat” approach) [15]. Such an approach would have substantial benefits for LF elimination and, possibly, onchocerciasis elimination, but would likely also incur an increase in programmatic costs.

Although IDA has the potential to be a game changer for LF elimination, more research is required to determine if there is a safe and effective way to use it in co-endemic settings before it is approved for these areas [15]. In particular, the restrictions regarding the use of DEC in onchocerciasis-endemic areas would need to be addressed through robust and extensive studies showing that IDA can be used safely in these settings [15].

Vector control

The potential impact of vector control on LF transmission has been illustrated by several studies [82]. For example, a study in the Gambia, which found that even without MDA, LF transmission may have been interrupted through the extensive and long-term (decades) use of insecticide-treated nets for malaria control [83]. A malaria eradication campaign in the Solomon Islands was also found to result in the interruption of LF transmission in the absence of MDA [84]. In addition, Nsakashalo-Senkwe et al. [85] found a significant decline in LF transmission associated with the nationwide scale-up of insecticide-treated nets in Zambia. These studies highlight how the expansion of insecticide-treated nets for malaria control since 2000 [86], could have had a notable impact on LF transmission in some settings [87]. A more detailed review of the role of vector control in the GPELF is provided by Bockarie et al. [82].

Due to the long-life expectancy of the adult worms and the delay between infection and morbidity, the use of vector control as a standalone strategy would result in a lag before any significant effect on the prevalence of infection and morbidity is seen [88]. This finding is mainly because vector control programmes only reduce exposure to new infections and do not have a direct effect on the established infections within the host population. Although the established adult worms will die naturally within their hosts, this occurs slowly due to their long-life expectancy [88]. However, in combination with MDA, vector control could potentially be beneficial in accelerating progress to elimination, preventing transmission hotspots and reducing the risk of the re-establishment of the transmission cycle from imported cases [82, 87,88,89]. This indicates that in the context of economic evaluations, the true potential benefits of combining vector control with MDA are long-term - in contrast to additional short-term reductions in morbidity or infection. This means that economic evaluations of vector control would require a long-time horizon for the analysis and a model accounting for the possibility of elimination to capture its full long-term benefit.

It is noteworthy that the only study we identified evaluating the cost-effectiveness of integrating vector control with MDA (which found that it did not appear to be cost-effective in the investigated setting [37]) had only a five-year time horizon (Table 1). Due to this, the potential longer-term benefits of vector control were not necessarily fully captured.

In the context of further economic evaluations of vector control for LF, it is essential to note that its benefit will be highly dependent on the local species of vector. For example, bednets will not be effective in areas where the predominant vector species bites during the day. This highlights the importance of not overgeneralizing the results of studies and policy in this area. It is also important to consider issues relating to insecticide resistance and the additional benefits of vector control on other vector-borne diseases (such as dengue and malaria) [90].

Diagnostics and surveillance strategies

As well as new interventions, we need to evaluate novel diagnostics and surveillance strategies. The importance of this research area is highlighted by a recent study which demonstrated resurgence of transmission six years after stopping MDA [91]. When considering new surveillance strategies, it is important to note the potential need to integrate surveillance for other NTDs (such as STH) [92, 93]. Only one of the studies [20] we identified explicitly considered the cost of post-MDA surveillance.

Conclusions

LF occurs across a wide and diverse range of epidemiological settings, making it difficult to draw conclusions regarding the value of LF interventions as a whole from studies based in a single country or setting. Also, due to the different aims of the identified studies and the different approaches used, it can be difficult to directly compare the results of the different studies. However, overall this systematic review highlights that the WHO recommended strategies for LF elimination are consistently found to be cost-effective or cost-saving across a wide range of settings and assumptions. This finding has important implications for advocacy groups and potential funders. However, there are several important research gaps that need to be addressed as we move forward towards the 2020 milestones and beyond. These include the evaluation of alternative interventions (such as IDA, anti-Wolbachia therapy and vector control). Furthermore, elements of the studies were not always clear, and at times important pieces of methodological information were not reported. Moving forward it would be beneficial if studies adhered more to standardised guidelines for reporting cost-effectiveness analysis - allowing easier comparison of the different studies results.