Volatile pyrethroid spatial repellents for preventing mosquito bites: a systematic review and meta-analysis

Abstract

Background: Volatile pyrethroid spatial repellents (VPSRs) can prevent mosquito-borne diseases including malaria and dengue fever, but the use of varied evaluation methods has resulted in a lack of clarity regarding their protective efficacy (PE) against contact with mosquitoes. This systematic review and meta-analysis consolidates the entomological evidence base on the PE of VPSRs against Anopheles, Aedes, and Culex mosquitoes and different test methods used.

Methods: We identified studies completed between January 2000 and September 2023 by searching through databases, conference abstracts, and personal correspondences. Included studies were semi-field or field studies that measured the PE of VPSRs using human landing catch (HLC) of mosquito landings on human legs and/or mosquito trap density, the number of mosquitoes captured using traps per set time period, compared to control groups. The systematic review summarised study-level data using a generalised linear mixed model with random effects. The meta-analysis pooled individual mosquito-level data and weather data on temperature, humidity, and wind from satellites, analysing PE subgrouped by product format, active ingredient, mosquito capture method used, mosquito species, and indoor vs outdoor setting. Risk of bias was assessed using a SYRCLE tool adapted for mosquito studies. Additional studies published from October 2023 to July 2025 were summarised. PROSPERO registration: CRD42021268852.

Findings: 58 eligible publications showed that VPSRs provided an average of 56% (95% CI 50, 62%) PE from mosquito bites. Meta-analysis of individual mosquito-level data from 50 (86%) of eligible studies involving 1,703,120 mosquitoes showed that PE was highest when measured using HLC, with similar results seen in semi-field (58%, 95% CI 54, 62%) and field studies (50%, 95% CI 40, 59%). Differences between indoor (54%, 95% CI 18, 68%) and outdoor settings (56%, 95% CI 51, 60%) were unclear. Species-level differences were observed with low PE seen in Anopheles funestus (31%, 95% CI 19, 43%); the potential for cross-resistance to solid-state pyrethroids is unclear. Efficacy was not sensitive to combined weather effects.

Interpretation: VPSRs offer protection from contact with mosquitoes, with semi-field studies reflecting field data and species-level differences observed. HLC provided the best quality data. Additional field studies that evaluate outdoor protection in malaria-endemic settings are needed, especially in West African, South American, and Southeast Asian settings.

Funding: National Institutes of Health (National Institute of Allergy and Infectious Diseases (K01AI156182)) and "Accelerate to Eliminate Malaria" program.

Keywords: Meta-analysis; Mosquito; Spatial emanator; Spatial repellent; Vector control; Volatile pyrethroid.

Fig. 1

Flow diagram of studies identified and assessed for inclusion eligibility. PRISMA chart shows literature search results on the left and grey literature search findings on right, including reasons for exclusion throughout the screening process. Study databases and registers identified 1198 abstracts and 171 studies were identified via other methods. After screening, 64 reports were included for data extraction in tables, 58 were included in the study-level forest plot, and 50 received original mosquito-level data that were pooled in the meta-analysis.

Fig. 2

Study-level forest plotof VPSR protective efficacy subgrouped by intervention format (Part 1). Each line represents one study. Average protective efficacy signified by red line based on 1,388,993 mosquitoes (84 studies, combining 39 studies in figure 2 and 45 in Fig. 3). Point estimates are blue boxes sized by weight including confidence intervals (horizontal blue line); exact values for these are provided in the column titled ‘Protective efficacy with 95% CI’. Point estimates to the left of the red line show lower efficacy than average, those to the right have higher efficacy. Some publications had more than one experimental study; the column ‘Experiment’ indicates variables specific to that study. Red diamonds show the average efficacy for each product format. The ‘Weight (%)’ column corresponds to the percentage that each study contributes to the average protective efficacy (total 100% when adding values from figures 2 and 3). The test of θi = θj is a test of homogeneity that uses Cochran's Q test (chi-square), values of p < 0.05 indicates that the studies are statistically different from one another. The test of θ = 0 is a z-statistic based on a Wald test that assesses whether overall pooled effects in a group of studies is equal to zero; values of p < 0.05 indicate that overall effects are statistically significantly not equal to zero. These two θ tests are computed for each subgroup. This figure is divided into two parts, with the second half presented in Fig. 3.

Fig. 3

Study-level forest plot of VPSR protective efficacy subgrouped by intervention format (Part 2). This figure is divided into two parts, with the first half presented in Fig. 2. Each line represents one study. Average protective efficacy signified by red line based on 1,388,993 mosquitoes (84 studies, combining 39 studies in Fig. 2 and 45 in figure 3). Point estimates are blue boxes sized by weight including confidence intervals (horizontal blue line); exact values for these are provided in the column titled ‘Protective efficacy with 95% CI’. Point estimates to the left of the red line show lower efficacy than average, those to the right have higher efficacy. Some publications had more than one experimental study; the column ‘Experiment’ indicates variables specific to that study. Red diamonds show the average efficacy for each product format. The ‘Weight (%)’ column corresponds to the percentage that each study contributes to the average protective efficacy (total 100% when adding values from figures 2 and 3). The test of θi = θj is a test of homogeneity that uses Cochran's Q test (chi-square), values of p < 0.05 indicates that the studies are statistically different from one another. The test of θ = 0 is a is a z-statistic based on a Wald test that assesses whether overall pooled effects in a group of studies is equal to zero; values of p < 0.05 indicate that overall effects are statistically significantly not equal to zero. These two θ tests are computed for each subgroup. Overall summary statistics of all 84 studies in figures 2 and 3 are provided at the bottom of this figure. For overall pooled estimates, the Q test is a test of group differences where if p < 0.05, the subgroup effect sizes are statistically different from one another. Since p = 0.25, the overall effect sizes from the subgroups are not considered to be different from one another.

Fig. 4

Pooled meta-analysis of VPSR efficacy subgrouped by select parameters. Combined standardised individual mosquito-level data of 1,703,120 mosquitoes from 50 studies analysed by subgroups; red line represents average protective efficacy, with efficacy to the right signifying higher efficacy, and lower efficacy to the left. Pooled estimates in blue boxes sized by weight, with horizontal blue lines indicating confidence intervals.

Fig. 5

Funnel plot to assess for publication bias. The blue dots each correspond to a study in the forest plots in Fig. 2, Fig. 3 (84 studies total), the red line is the overall effect (56%), which uses a Restricted Maximum Likelihood (REML) estimate indicating its precision through overall effect size using confidence intervals after controlling for within and between study variability. The included studies shown in blue are mostly between the confidence intervals indicated by grey lines in the shape of an upside-down funnel, suggesting there is no publication bias. Imputed studies shown in yellow estimate the results of missing studies that would address publication bias if it was present; since these have higher protective efficacy of volatile pyrethroid spatial repellents than published studies, publication bias is unlikely.