Systematic review of indoor residual spray efficacy and effectiveness against Plasmodium falciparum in Africa

Abstract

Indoor residual spraying (IRS) is an important part of malaria control. There is a growing list of insecticide classes; pyrethroids remain the principal insecticide used in bednets but recently, novel non-pyrethroid IRS products, with contrasting impacts, have been introduced. There is an urgent need to better assess product efficacy to help decision makers choose effective and relevant tools for mosquito control. Here we use experimental hut trial data to characterise the entomological efficacy of widely-used, novel IRS insecticides. We quantify their impact against pyrethroid-resistant mosquitoes and use a Plasmodium falciparum transmission model to predict the public health impact of different IRS insecticides. We report that long-lasting IRS formulations substantially reduce malaria, though their benefit over cheaper, shorter-lived formulations depends on local factors including bednet use, seasonality, endemicity and pyrethroid resistance status of local mosquito populations. We provide a framework to help decision makers evaluate IRS product effectiveness.

Fig. 1

Summary estimates for the level of mosquito mortality (a), exophily (b), blood-feeding (c) and insecticide-induced deterrence (d) as assessed in experimental hut trials with different indoor residual spray chemical classes measured within 2 months of spraying. Black box-plot show the binomial logistic model predictions (median (dark point), 25th and 75th uncertainty intervals indicated by box, 5th and 95th uncertainty intervals indicated by whiskers) which are weighted for the number of mosquitoes caught in experimental huts. The symbols show the raw data (also provided in Supplementary Data 1, analysis 1) and are classified according to the type of experimental hut (shape of symbol) and the hut substrate (symbol fill) as noted in the key in c. Point colour indicates the mosquito species complex, be it blue (A. funestus s.l.) or red (A. gambiae s.l.). Mosquitoes in the A. gambiae s.l. complex which were identified as A. arabiensis are shown in green. Supplementary Fig. 2–5 show the Bayesian posterior predictive fits against disaggregated data

Fig. 2

Summary of the temporal entomological impact of different IRS compounds. Probability of mosquitoes dying (top row), successfully blood-feeding (surviving and feeding) (row 2) or being deterred (row 3) in experimental hut trials over time. Row 4 summarises the best fit probability outcomes per feeding attempt for a mosquito to successfully blood-feed (red), exit without feeding (orange), be deterred before entering (green) or be killed (blue) for the different IRS products; pirimiphos-methyl: Actellic®300CS (column 1), pyrethroids: lambda-cyhalothrin, deltamethrin and alpha-cyhalothrin (column 2), bendiocarb: bendiocarb (1 spray round per year) (column 3) and neonicotinoids: clothianidin, SumiShield®50WG (column 4). Symbol shapes indicate the different studies (legend key references study numbers in Table 2 corresponding to 1, 2, 14 and 15, 3, 4 and 13, 5, 6 and 16, 7, 17 and 18, 8, 9, 10, 11 and 12 are previously unpublished data). Solid lines indicate the best fit statistical model to the mean data, weighted by sample size in different studies, and the dark-shaded area shows the 90% credible intervals around these best fit lines. The maximum and minimum data for each unique time point, for each IRS product, are fitted to capture the uncertainty in predicted performance of IRS products over time, these ranges are shown as pale polygons in rows 1–3 for each product. There is much uncertainty in the measurement for deterrence (row 3) because huts testing products that are sprayed onto walls cannot be easily rotated, we therefore simply fit to the initial deterrence measured and consider the depreciation of the deterrence effect to match that of mortality (further detailed in Supplementary Methods). Supplementary Fig. 6 shows individual study fits for these data

Fig. 3

The impact of pyrethroid resistant mosquitoes on the efficacy of pyrethroid-IRS. a The association between pyrethroid resistance in a mosquito population (measured as percentage survival over 24-h after a 60-min-exposure to a standard dose pyrethroid in the bioassay test) and 24-h mosquito survival immediately (time t = 1 day) after IRS spraying in an experimental hut trial (Supplementary Data 1, analysis 3a). Data represent A. gambiae s.l. complex (no circle), A. funestus s.l. data (symbol circled in blue) or A. arabiensis (circled in orange). The pyrethroid active ingredient tested in the bioassay either matched (purple) or mis-matched (red) the pyrethroid active ingredient used in the hut trial, or bioassay data was taken from a second study reported at the time of the IRS hut trial (green) (Supplementary Data 1, analysis 3a). The relationship between 24-h mosquito survival in a standard pyrethroid hut trial and the probability that a mosquito, on entering a hut, will successfully blood-feed (b) or preferentially enter a hut (c) without IRS (deterrence). In b and c, the 18 data from 13 studies with standard pyrethroid discriminating dose bioassay data (purple) are shown together with the 21 data from 11 studies for pyrethroid-IRS with time series but no bioassay measurement (blue) (Supplementary Data 1, analysis 3). Any symbol not noted in the key is included in addition to studies listed in Table 2 within Supplementary Data 1, analysis 3b. d The relationship between mosquito survival and longevity of the IRS. The time taken in days until less than 50% of mosquitoes die within 24-h (y-axis); the longer this duration, the longer the activity of the IRS (see Methods). e Summary of how pyrethroid resistance is predicted to influence the probability that a host-seeking mosquito will be killed (blue), deterred from entering (green), exit without feeding (orange) or successfully feed and survive (red) during a single feeding attempt in a hut freshly sprayed with pyrethroid-IRS (t = 1 day). Panel f shows the 3D relationship for pyrethroid resistance (50% survival at bioassay), mosquito outcome (colours as per panel e) and time since spraying

Fig. 4

Comparison of the predicted impact of IRS on malaria prevalence compared to that measured in randomised control trials. a Comparison of best fit standard LLINs (solid red line) vs standard LLINs + a single round of Actellic® 300CS (solid orange line). Shaded area indicates uncertainty in model predictions. The paler shaded area around the IRS lines shows additional uncertainty driven by variability in IRS efficacy (as illustrated in Fig. 2, Supplementary Table 3). The thin grey lines, noted in the key, denote IRS predictions parameterised separately for the individual experimental hut studies (Supplementary Table 4). b Comparison of standard LLINs (solid red line) vs standard LLINs + two rounds of bendiocarb (solid blue line). All models were parameterised using the data listed in Table 3 and estimates of the seasonality of transmission within the Kagera district and fitted to baseline prevalence (open symbols). Predictions of any change in prevalence for the respective age cohort measured were then made. The observed estimates for prevalence obtained during cross-sectional surveys of each RCT are plotted as closed symbols (in b vertical lines indicating 95% confidence intervals reported in the RCT)

Fig. 5

The additional impact of adding IRS to bed nets. The predicted number of malaria cases averted by annual application of IRS to a population with an existing level of bednet use (0–100% cover, y-axes) and a defined level of pyrethroid resistance (measured as percentage survival in a standard pyrethroid discriminating dose bioassay, x-axes). Clinical cases averted are measured per 1000 people per year, following standard LLIN distribution in a moderate endemicity area (30% prevalence in 2–10-year olds in the absence of interventions) with perennial transmission (a–c), highly seasonal transmission (d–f). In all panels IRS is applied, untargeted, to 80% of the population using either a long-lasting IRS product (for example Actellic®) (a, d), a short-acting IRS product (for example bendiocarb, applied annually) (b, e) or a pyrethroid-IRS product (for example deltamethrin (c, f)). Long-lasting products avert more cases though short-lasting products perform substantially better in highly seasonal settings