Leishmania: Parasite and Vector

Leishmania is of the protozoan genus trypanosomatida. The parasite resides intracellularly causing the disease leishmaniasis, and is transmitted between hosts by the bite of the female sandfly (genus Phlebotomus in the Old World localities of Europe, Africa and Asia, and Lutzomyia in the New World – Americas and Oceania). The primary hosts are vertebrates and commonly infected animals include humans, domestic dogs and cats, opossums, the crab-eating fox and the common black rat [1].

The sandfly comprises five genera and over 700 species. Approximately 30 species are thought to be implicated in transmission of Leishmania parasites [2]. Sandflies are found around human settlements and breed in organic matter such as leaf litter, manure and in rodent burrows [3], and are approximately a third of the size of mosquitos, measuring between 2–3 mm in length [4]. They are often categorised by virtue of where they bite (being classified as either endophagic (biting indoors) or exophagic (biting outdoors)), as well as where they rest (either endophilic (resting indoors) or exophilic (resting outdoors)) [5].

In humans, the clinical forms of the leishmaniases are broadly categorised into visceral leishmaniasis (VL) and cutaneous leishmaniasis (CL) (cutaneous forms presenting in a spectrum ranging from cutaneous, mucosal and diffuse cutaneous) [6].

Leishmania occurs in five continents and is endemic in 98 countries [7]. The World Health Organisation (WHO) estimate that 350 million people are at risk of contracting leishmaniasis [6]. Approximately 58,000 cases of visceral leishmaniasis and 220,000 cutaneous cases are officially reported each year. However it is thought that only around two thirds of countries actually report incidence data, with the sparsest data from Africa [7]. Based on assessments of under-reporting, 0.2–0.4 million new cases of VL and 0.7–1.2 million new cases of CL are estimated to occur every year [7].

There are several reasons for under-reporting of official Leishmania cases [2];

• Most official data are compiled through passive case detection
• Passive case detection relies on individuals presenting themselves for medical attention; however since leishmaniasis is associated with poverty, many patients do not have access to health care and do not, therefore, come to medical attention
• Many cases are misdiagnosed, undiagnosed or unreported, especially when availability of diagnostic tests is poor
• Leishmaniasis is often not a legally notifiable disease
• The number of asymptomatic infections (or individuals with sub-clinical infection - which may act as a reservoir) are not reported and could be an important driver for future infections

After malaria and African trypanosomiasis (sleeping sickness), the leishmaniases are the third most important vector-borne disease and are ranked ninth in terms of global burden of disease of all infectious and parasitic diseases [8]; accounting for more than 57,000 deaths per year and an estimated 2–2.4 million Disability Adjusted Life Years (DALYs) lost [8], [9].

When discussing the eco-epidemiology of leishmaniasis, there are considered to be four main forms of disease which are named with respect to the mode of transmission; Zoonotic Visceral Leishmaniasis (ZVL), Anthroponotic Visceral Leishmaniasis (AVL), Zoonotic Cutaneous Leishmaniasis (ZCL) and Anthroponotic Cutaneous Leishmaniasis (ACL). In anthroponotic forms, which are mostly located in Old World foci, humans are considered to be the main reservoir for infection whereas in zoonotic forms (mainly in New World areas such as Brazil), animals are thought to be the major source of parasites [10].

Following the 2007 World Health Assembly, where a resolution by member states to improve research on prevention, control and management of leishmaniasis was approved, the WHO convened the Expert Committee on Leishmaniasis in March 2010. The resulting technical report “Control of the Leishmaniases” [6] was the first updated report in more than 20 years. It details recommendations for diagnosis, treatment and prevention.

Although this report makes a start on describing the problem, as well as bringing together and directing global efforts to reduce the burden of the Leishmaniases, it highlights the fact that the disease is very complicated. There are various different diagnostic modalities, treatments and preventative methods, being more or less useful depending on the Leishmania species, the vector characteristics, the host immunity levels, the main reservoir, as well as the socio-economic and political makeup of the locality.

The fact that the true burden of disease is not accurately known (due in part, to reliance on passive case detection, as well as an array of issues with different diagnostic modalities), further complicates any efforts to deploy efficient methods of prevention and disease management and to secure funding for research.

Human VL, is mainly caused by two species of Leishmania parasites, each having a characteristic regional distribution, as described by Gill & Beeching [11]

• L. infantum is the causative agent in the Mediterranean, Middle East, Central Asia, China and Central and South America
• L. donovani in India and East Africa

VL may also be caused by L. tropica in the Old World and L. amazonesis in the New World, and is fatal in 85–90% of untreated cases and up to 50% of treated cases [11].

Approximately 90% of CL occurs in Afghanistan, Pakistan, Syria, Saudi Arabia, Algeria, Islamic Republic of Iran, Brazil, and Peru [12].

CL occurs in a spectrum of clinical presentations ranging from ulceration of the skin only (cutaneous), to various degrees of mucosal involvement (diffuse cutaneous and mucocutaneous).

Distinct sub-genii of Leishmania are thought to cause different cutaneous clinical presentations, usually divided into Old and New World regions. However Leishmania species which are implicated in VL can cause cutaneous disease (and vice versa, especially when the individual has co-infections) [11].

Mortality associated with CL is not significant; however the morbidity, in the form of disfigurement, with subsequent social stigmatisation which arises from cutaneous lesions and the resulting scars is very important. In endemic areas many people have the belief that CL can be transferred through physical contact [12] resulting in restriction of social participation. Arguably, equally as important in terms of burden of disease as the health and economic effects, are the detrimental impacts on quality of life and mental health resulting from social stigma [13].

Diagnosis

Clinical diagnosis of VL is often confused with other diseases such as malaria, schistosomiasis, African trypanosomiasis, miliary tuberculosis and malnutrition, and for CL, with tropical ulcers, leprosy and skin cancer [14].

Importantly, infection does not always result in clinical presentation of symptoms. The ratio of asymptomatic infections to clinical infections is thought to vary between 1∶2.6 to 50∶1 [15]. This presents a major problem for organisations relying on passive case detection when determining the true burden of disease and the size of the reservoir for future infections in areas of anthroponotic transmission. It is also interesting for future disease control developments to understand the genetically determined immunological factors that regulate clinical manifestation in humans [15].

The gold standard for confirmation of Leishmania infection is visualisation of parasites by microscopy; in a tissue smear such as a splenic aspirate, bone marrow or liver biopsy for VL [14], and scrapings or fluid from cutaneous sores in the case of CL [3]. Potential complications with obtaining tissue smears and biopsies, as well as the need for specialist medical staff and equipment mean that less invasive but equally as sensitive and specific diagnostic tools are needed for diagnosis of VL. Polymerase Chain Reaction (PCR) detection of parasite DNA in blood or organs is highly sensitive and specific but the high cost and need for specialised equipment and staff limit its use to hospitals and research centres [15].

Serological diagnostic methods are increasingly being used for diagnosis of VL but are not suitable for diagnosis of CL. One major problem with these methods is that serum antibodies remain in the body after successful treatment for several years, and therefore relapses cannot be detected using the same method [15]. Another problem is that a proportion of the population of an endemic area will test positive for serum antibodies even though they have no history of clinical leishmaniasis due to asymptomatic infection [16]. This makes it difficult to differentiate asymptomatic infected individuals from successfully treated, and from apparently cured individuals who may relapse in the future. There are a range of serologic assays including ELISA, IFAT, DAT, rK39.

More recently new techniques including loop-mediated isothermal amplification (LAMP), Nucleic Acid Sequence-Based Assay (NASBA) and Latex Agglutination Test (KAtex) have been developed but as yet have not been used in the trials reviewed here.

Skin hypersensitivity tests such as the Montnegro or Leishmanin skin test (MST/LST) are used to detect cell-mediated immunity using intradermal injection of Leishmania antigen. A negative response is normally seen during active infection with VL, with a switch to a positive skin test after cure. A positive result is also seen after asymptomatic infection [6]. A lack of sensitivity (14%) has been seen in LST for diagnosis of VL in India, limiting its utility [17].

The number of diagnostic tests available, as well as the variation in antibodies and antigens used in tests, coupled with the appearance of counterfeit immunochromatographic tests on the Indian subcontinent [6], shows the need for not only more reliable field-appropriate diagnostics, but standardisation and regulation of those already in use.

In areas of anthroponotic leishmaniasis, effective treatment of VL will help to decrease the human reservoir, however in areas of zoonotic transmission, treatment of humans will not solve the problem of potential human reinfection from an infected animal reservoir.

With the lack of effective drugs, prohibitive treatment costs and the possibility of relapse and resistance, clearly there is a need for a more effective way to stop the cycle of infection in both zoonotic and anthroponotic transmission. Potential areas of intervention include targeting the vector or animal reservoir population, or attempting to either prevent humans being bitten, being infected, or from developing clinical symptoms. Figure 1 illustrates the potential areas of intervention.

Download:

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

Figure 1. Range of preventative interventions.

Potential areas of intervention include; targeting the vector population, the animal reservoir population, or attempting to either prevent humans being bitten, being infected, or from developing clinical symptoms.

https://doi.org/10.1371/journal.pntd.0002278.g001

This literature review aims to provide a comprehensive account of the methods used to prevent human infection with Leishmania by performing a systematic search of all interventions aimed at reducing human disease incidence.

Search Strategy

Medline, EMBASE, CENTRAL, Web of Science, LILACS and WHOLIS were searched using terms relating to the keywords “leishmania”, “leishmaniasis”, “kala azar” (see Supplemental Materials Text S1 for full search terms for each database). Hand-searching of references of relevant studies and review articles was also performed and relevant articles retrieved. Numbers of studies retrieved, included and excluded were documented and recorded using a flow chart of stages of inclusion following the recommendations of the Preferred Reporting Items for Systematic Reviews and Meta-Analyses (PRISMA) group [18] (Figure 2).

Download:

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

Figure 2. PRISMA flow chart of included studies.

PRISMA flow chart documenting numbers of studies retrieved, included and excluded at each stage of the literature search.

https://doi.org/10.1371/journal.pntd.0002278.g002

Inclusion and Exclusion Criteria

Study Selection

After all titles and abstracts resulting from the electronic and hand searching were assessed against the predetermined inclusion and exclusion criteria by use of a standardised study eligibility form, the full article was retrieved.

The full paper was assessed against the inclusion/exclusion criteria. Any paper excluded at this stage was documented with a reason for the exclusion.

In order to extract relevant data as systematically as possible, an electronic data extraction template form was used. Due to the anticipated heterogeneity of the study types and outcomes investigated, the template section headings were designed to be broad in order to capture as much information as possible.

Data Synthesis

Based on scoping searches, it was not anticipated that meta-analysis would be possible due to the array of interventions and outcome measurements. Studies were analysed by narrative synthesis, making use of subgroup divisions based on which species (human, animal reservoir or vector) was the focus of the intervention. Within these groups there were further subdivisions based on the outcome measurements investigated. The primary measurement of interest was human infection rates. In many cases however, intermediate surrogate measurements were investigated which have differing applicability to potential changes in human incidence/prevalence of disease. Studies were grouped and analysed based on these outcome measures.

Animal Reservoir Control

This section includes animal elimination, canine vaccines, use of insecticides on dogs; including insecticide-impregnated dog collars, spot-on insecticides (drops applied to skin underneath hair on neck of dog), as well as whole-body insecticide use.

A total of 34 studies were retrieved, of which;

• Four investigated animal elimination
• Eight investigated insecticide-impregnated dog collars
• Six used spot-on insecticide treatments
• One used whole body insecticide use
• Seventeen investigated canine vaccines

Only four of these studies investigated human-specific outcomes as a result of interventions directed at animal reservoir control. Table 1 provides a summary of human outcomes for animal reservoir control studies.

One rodent culling intervention in Iran measured human CL by parasitological confirmation of skin lesion smear [28]. One insecticide-impregnated dog collar study investigated child VL seroconversion using DAT in Iran [29]. One dog culling intervention in Brazil studied human VL seropositivity using ELISA [30], and another dog culling intervention identified paediatric VL cases by passive case detection of clinical disease using health records in Brazil [31].

Three of the four studies reported statistically significant reductions in human Leishmania infection between the intervention and control areas. Two of these were animal elimination programmes, and one was an insecticide-impregnated dog collar study.

The only study to measure human infection using the gold standard of parasite visualisation was that of Ershadi et al. [28] who used poisoned grain to eliminate the rodent population around one intervention village and compared human infection to one control village where rodents were not eliminated. This study suffers from lack of generalizability not only to areas of anthroponotic transmission but also to areas of zoonotic transmission where transmission is believed to be driven by dogs. As with all the animal culling studies reviewed here, as well as all but one insecticide study, the report makes no reference to randomisation of areas. Ershadi et al. do however state that pre-intervention rates of infection were similar between the two areas investigated, but no data to support this assertion were presented. Pre-intervention numbers of active rodent burrows vary between intervention and control areas (between two and 18 times more burrows were reported in the control area than in the intervention areas).

Dietze et al. [30] reported that dog elimination in two valleys in Brazil did not result in a significant difference in human seropositivity as measured by ELISA when compared to one control valley. Although the major mode of transmission in Brazil is thought to be zoonotic (with dogs as the major reservoir), the authors hypothesised a greater than previously thought role for humans as the significant reservoir for VL, adding to the uncertainty in the literature that dogs are the most important driver of transmission in areas historically considered zoonotic. The group used active case detection by serology census of humans, using ELISA at six and 12 months following culling of all seropositive dogs in two intervention areas and following no intervention in one control area. The authors did not specify if the eliminated dogs were domestic or feral and no reference to randomisation of intervention and control areas is made. Absence of data on initial numbers of dogs in each valley and how many were eliminated makes it difficult to know if the areas are comparable. If only domestic animals were studied, this would leave a large section of the feral dog population unaccounted for and if the feral population was included, this would make follow-up even more difficult.

Ashford et al. [31] studied paediatric VL cases using passive case detection of clinical disease using health records following dog elimination. The authors used the measurement of cases per 1000 inhabitants, however the actual numbers of inhabitants (and therefore the actual numbers of cases) in each of the two neighbourhoods studied are not stated. In the four years of follow up, the authors reported nine cases per 1000 in the intervention area compared to 35 per 1000 in the control area. As with all studies using passive case detection, potential bias is introduced if one area is systematically better or worse at detecting and reporting cases. This may be linked to other factors such as educational level and socio-economic status of the population studied as well as quality and accessibility of healthcare. There is also a possibility, as with any intervention study, that increased activity focussing on preventative methods may influence the population to take other precautions against acquisition of the disease. None of the studies included in this section reported these kinds of population data. Interestingly, the decrease in incidence of dog infection, as measured by seroconversion, did not differ significantly between the two groups thus adding to the lack of clarity regarding the relationship between canine and human Leishmania infection. Like Dietze et al. [30], Ashford et al. [31] do not specify if the elimination programme included domestic and/or feral dog populations.

One of the major problems faced by groups carrying out any kind of animal elimination programme is the lack of control over the numbers of animals actually present in an area. In the case of the rodent elimination study, Ershadi et al. [28] used poisoned grain around rodent burrows. Active burrows were counted and treated every six months for three years if the number of re-opened burrows was 30% or more of initial numbers. Only once in three years did the researchers not need to re-bait burrows with poison meaning that burrows were re-opened rapidly and that the population of rodents may have attained original numbers quite quickly post-intervention. The same is seen of the dog elimination studies by Ashford et al. [31] and Diezte et al. [30] where follow-up of dogs was compromised by virtue of the dog population being both poorly understood through lack of regular censuses, and being dynamic with variable births/deaths/inward and outward movement. It is not clear from either dog culling study whether the elimination included all dogs or just domestic animals.

The only other study in this section which investigated human infection was an insecticide-impregnated dog collar study by Gavgani et al [29] which used DAT to detect seroconversion in children. The group used a matched-cluster randomised trial of 18 villages paired on pre-intervention child VL prevalence. Collars were only fitted to domestic dogs, however the authors note that due to elimination of stray dogs being a generic disease control practise in Iran, the feral dog population is very small and was not considered an issue. Although the intervention is associated with a statistically significant decrease in child VL prevalence during the one year follow up, the actual numbers of seroconversions are low – 17 in the nine intervention villages, and 26 in the nine control villages, and because of this the authors recommend caution be used when interpreting the results of the study.

The lack of studies measuring human infection following intervention resulted in only four out of 34 studies being included in this part of the discussion. Three of those reported a positive effect of the intervention they studied and one reported no statistically significant difference in the intervention and control group. Because of the small number of studies and the potential for bias, limited conclusions can be drawn as to the efficacy of interventions aimed at reducing animal reservoir infection with Leishmania on reducing disease burden in humans. In order to address this gap in knowledge, more and larger studies investigating human infection are needed. Ideally these would be cluster randomised in order to attempt to account for known and unknown confounders. Although the fundamental aim would be to reduce the burden of human disease, a secondary outcome would be to clarify whether the canine population is actually driving Leishmania transmission in areas historically described as zoonotic, or whether the human reservoir is more important. Without this evidence, it is not possible to determine if there is any use in allocating resources towards controlling animal reservoir populations.

Vector Population Control

The search for vector population control interventions identified studies relating to indoor and area-wide insecticide spraying, and a range of different interventions broadly termed here ‘Environmental Management’.

A total of 22 studies were retrieved, of which;

• Sixteen studied insecticide spraying of houses, outbuildings and area-wide spraying
• Four studied plastering of walls of houses with mud and/or lime to prevent entry of sandflies
• One investigated use of insecticide on termite mounds and animal burrows
• One measured the efficacy of educating school children on environmental risk factors

Of the 22 studies included in this section, only four investigated human-specific outcome measures following interventions directed at controlling the vector population. All of these involved insecticide spraying of houses and other buildings. Table 2 provides a summary of human outcomes for vector population control studies.

Two studies in Brazil measured seroconversion by ELISA; one measured VL in children under 12 years of age [32] whilst the other included all age groups [33]. One study in the Peruvian Amazon measured human CL by clinical signs [34], and one measured human CL by self-reporting in Afghanistan [35]. Two of the four studies reported statistically significant reductions in human Leishmania infection between the intervention and control areas, and two did not.

As a general criticism of studies in this category, only four measured human outcomes, and none studied human infection by parasite visualisation. Only three studies used a cluster randomised study design, with the majority not randomising study areas at all.

Both studies using the most robust diagnostic method (serology) reported no significant difference between the control and intervention groups.

Souza et al. [32] studied VL in children under 12 years in three control and three intervention areas subject to intradomiciliary pyrethroid spraying every six months. The control group is described by the authors as ‘normal activity’ - however this is not defined, nor is it specified if the study area used any preventative interventions as part of their ‘normal’ activity. The researchers needed parental consent to enrol in the study; it is not reported if there was any systematic difference between families who allowed their children to be enrolled, compared to those who did not, or whether there were differences in enrolment between intervention and control areas. Children were tested serologically for leishmaniasis using ELISA every 12 months for 2 years. Two different pyrethroid insecticides are used but there is no mention of which were used when and where. No reference is made to randomization of areas although the authors do state that background infection rates were approximately equal in all three areas.

Nery Costa et al. [33] used random allocation of 34×200 m² areas into four interventions (see also multiple interventions section) in order to study VL in the population living in each area. The intervention area used an unspecified insecticide applied to the inside walls of houses and animal pens whereas the control area used an unspecified insecticide applied to the inside walls of houses only. Background VL prevalence was 42% in the intervention areas and 31% in the control area. It is not clear whether this initial difference was accounted for in the final analysis. Of the 213 seronegative individuals included in the study (authors do not specify how many individuals were in each group), 120 (56%) were followed up during the 6 month study. No indication of drop-out rates by intervention is given and so potential bias could be introduced if the majority of drop-outs occurred in one particular group.

Davies et al. [34] used semi-randomisation (some were randomised, some were matched based on pre-intervention measures) of houses in 3 villages allocated to either intervention (sprayed) or control (unsprayed) arms. Spraying did not occur at the same time for all areas (one village was first sprayed one year later than the other two). Houses allocated to the intervention group were sprayed at six-monthly intervals on four occasions for two villages and only three occasions for the third village. No explanation for differences in randomisation or spraying regimens was given. Cases were detected by active case finding with clinical diagnosis of active lesions. The group reported a significant difference in CL cases over two years between the control houses and the intervention areas (24 versus nine respectively); the differences were more pronounced when cases detected within three months and then six months of the start of the study were excluded . The authors' rationale for removing cases diagnosed within three and six months is that cases within the first six months may reflect infection events which occurred prior to the commencement of the intervention programme. No evidence of existence of a period of time between infection occurring and a positive serology result (pre-patent period) is referenced by the authors. Such post hoc analytical decisions may introduce bias. There is evidence in the literature of varied infection rates over time in the same area [36] and epidemics occurring in certain years [37]. Therefore removing data without evidence of a significant pre-patent period of infection, as well as starting the interventions at different times for data that would later be pooled, may not have been appropriate.

One general criticism relating to many of the studies included in this section is that reports exist of interventions (especially residual insecticide spraying) having the effect of directing sandflies into non-sprayed areas [38]. This could be a particular issue when the unit of study is a room or a house, or where areas allocated to intervention and control are directly adjacent. Sandflies are thought to be able to travel anywhere up to 960 m in 36 hours [39]. Therefore, adjacent areas may be more at risk of increased sandfly abundance, and therefore risk of infection, if an intervention has a repellent, but not harmful, effect on sandfly activity.

Finally, Reyburn et al. [35] conducted a large (3666 participants) cluster randomised trial of management of CL in Afghanistan, with four arms (see also multiple interventions section). The sample size was chosen based on power calculations (which none of the other studies reported) and follow-up house visits were conducted at eight, 10 and 15 months post-intervention. The units studied were blocks of 10 houses. Like Nery Costa et al. ¹⁷, Reyburn et al., also included a safeguard against interventions overlapping into adjacent blocks by including ‘reservoir’ (untreated and not included in the study) blocks. All self-reported lesions were visually inspected before being reported as a case. The authors do state that the intention to confirm diagnosis by smear was not possible due to deterioration of the security situation in Kabul.

All survey workers were blinded to the intervention, and a census was conducted prior to the programme to ensure areas were matched for pre-intervention prevalence. Intervention houses were sprayed only once during the 15 month study period and control and reservoir houses were also offered insecticide spraying of the living areas of their houses although the authors do not report how many houses accepted the intervention. This fact calls into question the validity of the control areas to which the intervention is being compared, however the authors report that since the concentration of the insecticide applied to control houses was so low, it likely did not have an effect on results. The group added that if sandflies had been diverted from intervention to control houses then the protective effect of spraying could have been over-estimated. A loss to follow-up of approximately 45% was reported due to the security situation, however possible bias introduced by this was not considered.

Although Davies et al., [34] do not specify that case detection was self-reported, it is difficult with CL diagnosis that relies on clinical signs, to be certain that all cases have been reported. Reyburn et al., [35] rely on household questionnaire and follow up of clinical signs. Ultimately if someone wanted to hide the fact that they had a skin lesion and the lesion was not on an exposed part of the body, the information could be concealed from researchers, even when active case detection was employed. None of the papers included in this section, or indeed in this review, address the fact that CL diagnosis, especially because of the social stigma lesions can bring to the sufferer [13], could be particularly susceptible to bias in case finding, in comparison to population-wide serological diagnostics.

In summary, only four studies measure human outcome following insecticide spraying of houses and buildings. Two of these use robust diagnostic methodologies and show no difference between intervention and control, and two use less robust methods of diagnosis and show a statistically significant effect. None of the studies measuring environmental management study human-specific outcomes.

Based on this evidence, there is not enough information to determine if insecticide spraying is advantageous in reducing the burden of disease in humans.

In terms of generalizability, indoor insecticide spraying programmes are only useful where the sandfly is likely to come into contact with the walls that are actually sprayed, which necessitates endophagic or endophilic (either biting or resting indoors) sandfly species. It is not clear from any of the papers reviewed here that the authors took this into consideration.

Further research would be needed in order to characterise the feeding, resting and breeding habits of different sandfly species in each endemic area, prior to implementation of a preventative intervention. In addition, more studies are needed which measure human infection after vector control interventions.

Finally, there is evidence in the literature that sandfly resistance to insecticides is emerging, especially to DDT [4] which is still used in large-scale spraying programmes in India [40]. There is also evidence that spraying campaigns are more effective directly before transmission seasons which vary depending on location [40]. Care needs to be taken to ensure that spraying programmes are thoroughly investigated with respect to transmission seasons and sandfly activity in order to have the highest likelihood of being effective. Insecticide spraying is still used during leishmaniasis epidemics [37] and misuse may result in wide-spread resistance which would render them useless when they are needed most.

Human Reservoir Control

This section includes studies describing personal protection measures for humans such as treated and untreated bed nets, barrier nets, insecticide-impregnated fabrics such as curtains, clothing and bed sheets, and use of soap containing insecticide. It also includes studies on human vaccines. Table 3 provides a summary of human outcomes for human reservoir control studies.

Based on the search criteria, a total of 34 studies were retrieved and included in this section. Five studies which investigated insecticide-impregnated bed nets or other fabrics have multiple arms in order to examine multiple interventions, often including non-impregnated nets and fabrics as a control group. Because of this, there are more interventions than studies. One study [41] investigated the concomitant use of insecticide-impregnated bed nets and curtains and so will be included in the multiple interventions section.

• Sixteen studies investigated insecticide treated bed nets
• Seven examined untreated bed nets
• One study used large area-wide insecticide-impregnated barrier nets
• Two reported on insecticide-impregnated clothing
• Two investigated insecticide-impregnated curtains
• One examined insecticide-impregnated bed sheets
• One study used an insecticide-containing soap
• Eleven studied human vaccines

Limitations of Review

Due to the heterogeneity of study designs and outcome measures, a full and formal quality assessment of studies was not carried out. Although the variable methodological quality seen in the studies reviewed here is addressed in this discussion section, this was not carried out systematically. It may be advantageous for future work to systematically rate methodological quality and take this into consideration when reporting results. The authors are aware of one potentially relevant paper published in a Chinese language journal but were unable to obtain the reference.

Summary and Conclusions

At the time of writing, published protocols highlight the intention to investigate aspects of prevention of Leishmania infection in humans [59]–[61]. To date however, there are no published reviews which address preventative methods in their entirety. This review provides a comprehensive overview of all interventions against human leishmaniasis, highlighting fundamental gaps in knowledge, and suggesting directions for future research.

Four broad categories of preventative interventions were identified in this review, investigating a heterogeneous mix of outcome measures and using a variety of different methods.

This review emphasizes the absence of high quality evidence demonstrating the impact of interventions on the prevalence or incidence of human Leishmania infection assessed using reliable diagnostic modalities.

Research identified within this review includes intervention strategies ranging from protection of humans against infection, to interventions aimed one stage upstream of human infection (targeting the sandfly vector), and even further, to interventions targeting animal reservoir species.

Conflicting data on the impact of dogs in transmission of leishmaniasis to humans, along with lack of generalizability of interventions directed at vector control, point towards gaps in fundamental knowledge of the biology of transmission. Despite this weak evidence base, many countries continue to invest heavily in preventative methods focussed on control of leishmaniasis in dogs.

Based on our current lack of understanding of the transmission of Leishmania, it seems salient to focus scant resources on prevention of human infection, as opposed to interventions which attempt to address upstream risk.

The absence of a promising prophylactic human vaccine candidate, along with the scarcity of human vaccine studies, indicates that an effective (and cost-effective) human vaccine is unlikely to be forthcoming in the near future. Nonetheless, with no reliable intervention having been identified and the current treatment options for leishmaniasis being expensive, with serious side effects and emerging resistance, there is clearly a need for a more integrated focus within the international community to direct resources towards development of a human vaccine.

Final Conclusions