Vector control for malaria prevention during humanitarian emergencies: a systematic review and meta-analysis

Summary Background Humanitarian emergencies can lead to population displacement, food insecurity, severe health system disruptions, and malaria epidemics among individuals who are immunologically naive. We aimed to assess the impact of different vector control interventions on malaria disease burden during humanitarian emergencies. Methods In this systematic review and meta-analysis, we searched ten electronic databases and two clinical trial registries from database inception to Oct 19, 2020, with no restrictions on language or study design. We also searched grey literature from 59 stakeholders. Studies were eligible if the population was affected by a humanitarian emergency in a malaria endemic region. We included studies assessing any vector control intervention and in which the primary outcome of interest was malaria infection risk. Reviewers (LAM, JF-A, KC, BP, and LP) independently extracted information from eligible studies, without masking of author or publication, into a database. We did random-effects meta-analyses to calculate pooled risk ratios (RRs) for randomised controlled trials, odds ratios (ORs) for dichotomous outcomes, and incidence rate ratios (IRR) for clinical malaria in non-randomised studies. Certainty of evidence was evaluated using the Grading of Recommendations Assessment, Development and Evaluation (GRADE) approach. This study is registered with PROSPERO, CRD42020214961. Findings Of 12 475 studies screened, 22 studies were eligible for inclusion in our meta-analysis. All studies were conducted between Sept 1, 1989, and Dec 31, 2018, in chronic emergencies, with 616 611 participants from nine countries, evaluating seven different vector control interventions. Insecticide-treated nets significantly decreased Plasmodium falciparum incidence (RR 0·55 [95% CI 0·37–0·79]; high certainty) and Plasmodium vivax incidence (RR 0·69 [0·51–0·94]; high certainty). Evidence for an effect of indoor residual spraying on P falciparum (IRR 0·57 [95% CI 0·53–0·61]) and P vivax (IRR 0·51 [0·49–0·52]) incidence was of very low certainty. Topical repellents were associated with reductions in malaria infection (RR 0·58 [0·35–0·97]; moderate certainty). Moderate-to-high certainty evidence for an effect of insecticide-treated chaddars (equivalent to shawls or blankets) and insecticide-treated cattle on malaria outcomes was evident in some emergency settings. There was very low certainty evidence for the effect of insecticide-treated clothing. Interpretation Study findings strengthen and support WHO policy recommendations to deploy insecticide-treated nets during chronic humanitarian emergencies. There is an urgent need to evaluate and adopt novel interventions for malaria control in the acute phase of humanitarian emergencies.


Introduction
Humanitarian emergencies, of either natural or anthropogenic origins, can lead to large-scale population movement, food insecurity, and severe health system disruptions. As of 2022, the UN High Commissioner for Refugees estimates that there are 89·3 million people who have been forcibly displaced worldwide, including 53·2 million people who have been internally displaced, 21·3 million refugees, and 4·6 million asylum-seekers. 1 Humanitarian emergencies in Venezuela have led to a displacement of 4·4 million Venezuelan people and a 1200% increase in malaria cases between 2000 and 2020 in Venezuela, which is a stark reminder of how easily reversible malaria control gains are. 2 Venezuela now accounts for 73% of total malaria deaths on the continent. 3 Globally, almost twothirds of people affected by humanitarian emergencies inhabit malaria endemic areas, 4 particularly the WHO African region, which currently accounts for 94% of all malaria cases and deaths. 3 Mass displacement can increase the risk of severe malaria epidemics, especially when populations with little or no previous disease exposure move into areas of more intense transmission or when individuals with subclinical infections transit into urban settings. Inadequate water, sanitation, and hygiene (WASH) facilities, drainage, and waste management systems all contribute to high levels of vector breeding and increased malaria transmission. Limited, disrupted, and overburdened national malaria control programmes and health services result in insufficient access to treatment and control measures, with poor health outcomes worsened by concomitant infectious diseases, malnutrition, security concerns, 5 trauma, and anaemia. 6 During the initial phase of a humanitarian emergency, the priorities for malaria control are prompt and effective diagnosis and treatment. 6 Selection of complementary vector control interventions in these situations depends on malaria infection risk, human and local vector population behaviours, available logistical support, and type of shelter. 7 In some emergencies, effective case management can be supplemented with distribution of insecticide-treated nets (ITNs), first targeting the most susceptible populations, such as pregnant women and children younger than 5 years, with the end goal of achieving and maintaining universal coverage. 8-10 Indoor residual spraying (IRS) is generally not feasible when dwellings are scattered widely, of a temporary nature, or constructed with surfaces that are unsuitable for spraying, 11 but it is more appropriate for protecting larger populations where housing is permanent and structurally sound, as are more often established when emergencies progress into the post-acute and recovery phases. 6, 12 Other innovative community-level and personal vector control interventions have been designed specifically for humanitarian emergencies, including shelter materials, [13][14][15] bed sheets, [16][17][18] blankets, 19 and clothing that are treated with insecticides, 20 as well as topical repellents for individual use 21,22 or to treat community livestock. 23 However, evidence for the effectiveness of vector control tools deployed during humanitarian emergencies has been considered insufficient by WHO to develop policy recommendations. Recommendations for ITNs and IRS during humanitarian emergencies are based on their proven efficacy in non-emergency situations, without consideration of the unique operational challenges associated with emergency settings. 7 We did a systematic review and meta-analysis of the effect of different vector control interventions on malaria disease burden during humanitarian emergencies.

Search strategy and selection criteria
In this systematic review and meta-analysis, we adhered to the updated PRISMA 2020 guidelines. 24, 25 We searched Cochrane Infectious Disease Group Specialized Register,

Research in context
Evidence before this study Humanitarian emergencies (natural, climate-related, or manmade) can increase risk of malaria epidemics and the incidence of severe disease, particularly when individuals with immunological naivety are displaced into high malaria transmission areas and when vector breeding increases due to flooding, loss of infrastructure, and inappropriate waste drainage and management systems. When reviewed previously by WHO, the evidence for vector control tools deployed during humanitarian emergencies was considered insufficient to develop formal policy. Few recommendations were given based on proven intervention effectiveness during non-emergency situations. We searched PubMed, without restriction on language or study design, using the terms "humanitarian emergency", "malaria", "vectors" or "mosquito" or "Anopheles", "insecticide-treated net", "indoor residual spraying", "insecticide-treated plastic sheeting", "insecticidetreated clothing", "insecticide-treated blankets", "topical repellents" for articles published between database inception and Oct 15, 2020. We found no published synthesis of the evidence base for malaria control interventions during emergency settings.

Added value of this study
We investigated the effect of different vector control interventions on malaria disease burden during humanitarian emergencies to provide a comprehensive global evidence summary and to highlight the complexities associated with generating robust evidence for vector control tools in such settings. Insecticide-treated nets (ITNs) were confirmed as effective for preventing both Plasmodium falciparum and Plasmodium vivax malaria infection in chronic emergencies with adequate shelter. The evidence for an effect of indoor residual spraying (IRS) on preventing malaria infection during humanitarian emergencies was weaker. Other experimental vector control tools, including insecticide-treated chaddars or blankets and insecticide treatment of cattle, also showed promise in individual studies and might be transferable to acute emergencies or outdoor conditions. Key differences in vector and human behaviours might influence the effectiveness of these interventions between emergency settings and additional randomised studies are warranted.

Implications of all the available evidence
No formal guidelines existed for malaria vector control in humanitarian emergencies. The results of this systematic review and meta-analysis support the use of ITNs in these settings, where shelter types and sleeping arrangements are appropriate for their use. We identified multiple barriers affecting the evaluation of vector control tools in humanitarian emergencies, including the rapid onset of these events, ethical design of appropriate control groups, inequitable access of study participants to concomitant improvements in malaria diagnosis and treatment, heterogeneities in refugee camp infrastructure, and recent requirements of two randomised controlled trials with epidemiological endpoints to provide evidence for approval of new classes of vector control tool. Most importantly, there is a clear and urgent need to evaluate novel, innovative, emergency-specific vector control interventions to complement ITN distributions, IRS campaigns, and parallel strategies to improve policy evaluation processes, expand market access, and stimulate the development of these niche vector control products. Study findings have now contributed directly to the latest WHO guidelines for malaria vector control.  26 Studies retrieved were eligible for inclusion in the systematic review and meta-analysis if they satisfied all criteria: the study population consisted of individuals of all age groups or children of specified age groups (ie ≤5 years or 6-15 years) who were affected by humanitarian emergencies (any phase) in malaria endemic regions; any malaria-specific vector control intervention was evaluated; and the primary outcome of interest was malaria infection risk, measured as case incidence, infection incidence, or parasite prevalence. An expanded definition of a humanitarian emergency can be found in the appendix (p 21). Studies retrieved were eligible for inclusion in secondary analyses if none of the primary malaria risk indicators were measured but at least one of the following secondary outcomes were reported: all-cause mortality, incidence of severe malaria, anaemia prevalence (<10 g/dL), entomological inoculation rate, adult mosquito density, sporozoite rate, intervention durability, occurrence of adverse events, and user acceptability and usage of interventions. The following study designs were eligible for inclusion in the primary meta-analysis: clusterrandomised, controlled before-and-after, cross-sectional, non-randomised crossover, case-control, and cohort studies, case series, interrupted time series, and programmatic evaluations. A detailed study design is available from the review protocol. 27

Data analysis
Two review authors (LAM and JF-A) independently extracted information from eligible studies, without masking to author or publication, into a database. Disagreements in data extraction were resolved by discussion and consensus between the two review authors, with arbitration by a third (KC, BP, or LP) when necessary. Original study authors and research groups were contacted in cases of missing data or ambiguous reporting. Information on data extraction can be found in the appendix (pp 22).
Analyses were structured first by type of vector control intervention, second by outcome (separating Plasmodium falciparum and Plasmodium vivax), and third by study design. If a combination of vector control interventions were used (eg, IRS with ITNs), these were considered as separate interventions. Studies that reported sufficient data to calculate crude effects and studies that reported crude or adjusted effect measures with 95% CIs were included in the quantitative analysis.
Epidemiological data were combined in meta-analyses. Meta-analyses of both crude and adjusted results were reported. Random-effects models were used to calculate pooled effect measures (risk ratios [RRs] for randomised controlled trials, odds ratios [ORs] for dichotomous outcomes in non-randomised studies, and incidence rate ratios [IRRs] for clinical malaria incidence in nonrandomised studies). Study effects were combined in the meta-analysis using the generic inverse method for nonrandomised studies with adjusted results. 28 For clusterrandomised controlled trials or non-randomised cluster trials, adjusted measures of effect were extracted. If study authors did not perform any adjustment for clustering, raw data were adjusted using an intraclass correlation coefficient (ICC). If no ICC was reported, this was estimated with reference to similar studies. If the ICC was estimated, sensitivity analyses were done to investigate the robustness of our analyses.
Risk of bias for randomised controlled trials was assessed using the Cochrane risk of bias tool. 29 Risk of bias for non-randomised studies was assessed using the Newcastle-Ottawa scale. 30 There were insufficient studies identified to assess publication bias; however, we had planned to assess this by visual inspection of funnel plots and Egger test for funnel plot asymmetry. 31 Certainty of the evidence was evaluated for all comparisons using the Grading of Recommendations, Assessment, Development and Evaluation (GRADE) approach. 32 We did analyses in Review Manager (version 54.1). This study is registered with PROSPERO, CRD42020214961.

Role of the funding source
The funder of the study had no role in study design, data collection, data analysis, data interpretation, or writing of the report.
Due to the absence of a published protocol for all studies, there were some concerns regarding the risk of bias in the selection of the reported result for cRCTs. However, in all other domains, the risk of bias was low for all studies apart from the study by Rowland and colleagues (1999), 16 in which there were concerns related to bias due to deviations from the intended intervention (appendix p 51). There was a high risk of bias in two case-control studies, 39,46 which were scored low on the Newcastle-Ottawa scale in the representativeness of cases, selection of controls, and ascertainment of exposure (appendix p 52). There was a moderate risk of bias for the 11 cross-sectional, controlled before-and-after,  and before-and-after studies. 8, 12,20,[37][38][39][40][41][42]44,46 Bias was attributed to the absence of information on non-responding participants, exposures not being formally validated through a measurement tool, and because some studies did not control for any potential confounders (appendix p 53). There was a low risk of bias in the one cohort study 14 (appendix p 54). There were insufficient studies to test for asymmetry in the meta-analysis for any outcome measurement per intervention. The GRADE assessments indicated moderate-to-high certainty evidence for the effect of ITNs, insecticide-treated chaddars and blankets, topical repellents, and insecticide-treated cattle on malaria outcomes during humanitarian emergencies.
The GRADE approach indicated very low or low certainty evidence for the effect of IRS, IRS and ITNs, insecticidetreated clothing, and insecticide-treated plastic sheeting on malaria outcomes during humanitarian emergencies.

Discussion
Study findings showed high certainty evidence that ITN deployment in chronic humanitarian emergencies can reduce P falciparum incidence by 45% and P vivax by 31%. Similar effect sizes have been reported from recent meta-analyses of ITNs used in non-emergency malaria-endemic settings. 47 The contemporary effectiveness of ITNs is predicated on several factors, including

Total (95% CI)
Heterogeneity: τ 2 =0·04; χ 2 =3·61, df=2 (p=0·16); I 2 =45% Test for overall effect: Z=2·34 (p=0·019)  net access, use, and durability, insecticide retention and bioefficacy, and vector susceptibility. 48 49 There were some indications that IRS can prevent P falciparum and P vivax infection during humanitarian emergencies. However, the certainty of evidence was very low, aligning with previous results from nonemergency settings, which highlighted the need for additional trials to quantify the effect size of IRS in different transmission settings. 51 Based on ITN and IRS efficacy during non-emergency situations, WHO has suggested these two tools for use in emergencies, and our study findings strengthen and support this policy for ITNs where sleeping arrangements are appropriate (ie, where there is somewhere to hang a bed net). Compared with ITNs, IRS requires fewer behavioural changes, which can impede intervention use during humanitarian emergencies, and the arsenal of available insecticides is greater, an asset with regards to the pervasive problem of insecticide resistance among major malaria vector populations. 52 However, IRS does require planning, timely application, specialist experience, greater logistical resources for implementation, and sustained donor initiatives. 53 One strategy to circumvent some of these limitations has been to exploit utilitarian emergency materials, such as plastic sheeting, tarpaulins, and tents, as longerterm mechanisms of insecticide delivery. 13,15,[54][55][56][57][58][59] In this systematic review, we only identified one study that reported very low certainty evidence of the effect of insecticide-treated plastic sheeting on P falciparum case incidence in west Africa during a chronic emergency. 14 This concept has also been extended to treating personal material items during emergencies, including clothing, blankets, and bedsheets, with pyrethroid insecticides, 16,18,20,60 with moderate evidence for a reduction in P falciparum case incidence from a household-randomised trial of permethrin-treated chaddars in Pakistan 16 and lower certainty evidence for insecticide-treated clothing to prevent malaria infection in Kenya. 20 As none of these products are currently commercially available, there is an urgent need to design and further evaluate novel, innovative, and emergency-specific vector control tools to complement ITN distributions and IRS campaigns. The simultaneous deployment of several malaria control tools has been identified as a priority in humanitarian emergencies among experts, with insecticide-treated covers and blankets considered favourable because of their transportability and flexibility for use in mobile populations. 53 The private sector has largely failed to  develop these niche vector control tools, particularly those that might not have a market share for routine use in stable malaria endemic settings. 61 Topical repellents and insecticide-treated cattle were also associated with significant reductions in malaria incidence in the Eastern Mediterranean and South-East Asia with moderate-to-high certainty evidence. 21,23,36 However, by comparison with sub-Saharan Africa, there are key differences in malaria vector populations between these areas, particularly exophilic or exophagic and anthropophilic or zoophilic tendencies, which might influence the efficacy of these vector control interventions. Insecticide-treated cattle was an effective intervention in Pakistan, 23,62 where vector populations are highly zoophilic and the study population lived in established housing settlements. It could be assumed that this strategy would also show promise in areas where Anopheles arabiensis is a major vector species in sub-Saharan Africa. [63][64][65] Topical repellents have not always performed well in malariaendemic parts of Africa, largely due to issues of IEC/BCC, standardised repellent formulation, user compliance, and duration of protection, which can all contribute to participants being unprotected at peak vector biting times. 66,67 Differences in vector behaviour were also apparent in some of the eligible ITN studies from South-East Asia, where the evidence for an effect of ITNs on malaria transmission was weak in rural, non-conflict areas due to early biting and exophilic or exophagic behaviours of most primary vectors. 68,69 A similar absence of protection from malaria was observed in populations of internally displaced people in the same region. 10 In several cases, published data were limited to observational studies or retrospective programmatic evaluations, 8,12,[37][38][39][42][43][44]46 or from randomised studies, which were done in much more stable internally displaced people and refugee settlements, established for a number of years. 10,11,16,22,23,[34][35][36]45 We were only able to identify one study done in acute settings, which was done during a tropical cyclone in Vanuatu and was excluded because it had no malaria events. 33 Most eligible randomised studies were done in chronic humanitarian emergencies in the Eastern Mediterranean and in South-East Asia, with comparatively fewer done in sub-Saharan Africa, where most of the malaria disease burden is now concentrated 3 and substantial numbers of humanitarian emergencies are ongoing. 1 In addition, most eligible studies were done between 1990 and mid-2000s. Due to the age of these studies, some key details, such as reporting methods of cluster randomisation, were not always described adequately because of the available reporting guidelines at the time, thereby downgrading their overall GRADE assessment. 70 Furthermore, all eligible studies evaluated interventions using insecticides (primarily pyrethroids), which might not be used for malaria vector control due to widespread insecticide resistance. [71][72][73] New dual-active ingredient ITNs, piperonyl butoxide ITNs, novel IRS compounds, attractive toxic sugar baits, 74 and spatial repellents 75 are undergoing phase 3 trials in nonemergency settings to determine their efficacy against pyrethroid-resistant malaria vector populations. [76][77][78][79][80] Additional randomised trials evaluating the efficacy of these new tools, and other existing ones, are warranted in emergency settings, spanning a range of geographical areas, malaria vector species, and insecticide resistance intensities, to strengthen the evidence basis for use of vector control tools to prevent malaria infection among internally displaced people and refugees.

Contributors
LAM and MR conceived of the study concept. LAM, JF-A, BP, and MR developed the study design. LAM, JF-A, KC, BP, and LP screened the literature and extracted data from the articles. JF-A and KC led the data analysis, with support from LAM, BP, LP, and MR. LAM wrote the first draft of the manuscript, which was revised by all authors. All authors reviewed and agreed with the final version of the report. All authors had access to all the data in the study and had final responsibility for the decision to submit for publication. LAM, JF-A, and KC accessed and verified the data.

Declaration of interests
We declare no competing interests.

Data sharing
As this study is a systematic review and meta-analysis, there is no primary data to be shared. Raw extracted data are available from the corresponding author on reasonable request.