This project analysed data from two large-scale experiments involving six rounds of indoor pyrethroid spraying over a two-year period in the Peruvian Amazon city of Iquitos. We developed a spatial multilevel model to identify causes of Aedes aegypti population declines that were driven by (i) recent household use of ultra-low volume (ULV) insecticides and (ii) ULV use in neighbouring or nearby households. We compared the fit of the model to a range of possible spray effectiveness weighting schemes based on different temporal and spatial decay functions to capture the lagged effects of ULV insecticides.
Our results indicate that the reduction in A. aegypti abundance within a household was primarily due to spraying within the same household, while spraying in neighbouring households had no additional effect. The effectiveness of spraying activities should be assessed based on the time since the last spraying, as we did not find a cumulative effect from successive spraying. Based on our model, we estimated that spray effectiveness declined by 50% approximately 28 days after spraying.
Household Aedes aegypti mosquito population reductions were primarily dependent on the number of days since the last treatment in a given household, highlighting the importance of spray coverage in high-risk areas, with spray frequency dependent on local transmission dynamics.
Aedes aegypti is the primary vector of several arboviruses that can cause large epidemics, including dengue virus (DENV), chikungunya virus, and Zika virus. This mosquito species feeds primarily on humans and frequently feeds on humans. It is well adapted to urban environments [1,2,3,4] and has colonized many areas in the tropics and subtropics [5]. In many of these regions, dengue outbreaks recur periodically, resulting in an estimated 390 million cases annually [6, 7]. In the absence of a treatment or an effective and widely available vaccine, prevention and control of dengue transmission rely on reducing mosquito populations through various vector control measures, typically spraying insecticides that target adult mosquitoes [8].
In this study, we used data from two large-scale, replicated field trials of ultra-low volume indoor pyrethroid spraying in the city of Iquitos, in the Peruvian Amazon [14], to estimate the spatially and temporally lagged effects of ultra-low volume spraying on household Aedes aegypti abundance beyond the individual household. A previous study assessed the effect of ultra-low volume treatments depending on whether households were within or outside a larger intervention area. In this study, we sought to decompose treatment effects at a finer level, at the individual household level, to understand the relative contribution of within-household treatments compared to treatments in neighbouring households. Temporally, we estimated the cumulative effect of repeat spraying compared to the most recent spraying on reducing household Aedes aegypti abundance to understand the frequency of spraying needed and to assess the decline in spray effectiveness over time. This analysis can assist in the development of vector control strategies and provide information for the parameterization of models to predict their effectiveness [22, 23, 24].
Visual representation of the ring distance scheme used to calculate the proportion of households within a ring at a given distance from household i that were treated with insecticides in the week preceding t (all households i are within 1000 m of the buffer zone). In this example from L-2014, household i was in the treated area and the adult survey was conducted after the second round of spraying. The distance rings are based on the distances that Aedes aegypti mosquitoes are known to fly. Distance rings B are based on a uniform distribution every 100 m.
We tested a simple measure b by calculating the proportion of households within a ring at a given distance from household i that were treated with pesticides in the week preceding t (Additional file 1: Table 4).
where h is the number of households in ring r, and r is the distance between the ring and household i. The distances between rings are determined taking into account the following factors:
Relative model fit of the time-weighted within-household spray effect function. Thicker red lines represent the best-fitting models, where the thickest line represents the best-fitting models and the other thick lines represent models whose WAIC is not significantly different from the best-fitting model’s WAIC. B Decay function applied to days since last spray that were in the top five best-fitting models, ranked by average WAIC in both experiments
The estimated reduction in Aedes aegypti numbers per household is related to the number of days since the last spraying. The equation given expresses the reduction as a ratio, where the rate ratio (RR) is the ratio of the spraying scenario to the no-spray baseline.
The model estimated that spray effectiveness declined by 50% approximately 28 days after spraying, while Aedes aegypti populations had almost fully recovered approximately 50–60 days after spraying.
In this study, we describe the effects of indoor ultra-low volume pyrethroid spraying on household Aedes aegypti abundance as a function of the timing and spatial extent of spraying near the household. A better understanding of the duration and spatial extent of spraying effects on Aedes aegypti populations will help to identify optimal targets for spatial coverage and spraying frequency required during vector control interventions and inform modelling comparing different potential vector control strategies. Our results show that Aedes aegypti population reductions within a single household were driven by spraying within the same household, whereas spraying of households in neighbouring areas had no additional effect. The effects of spraying on household Aedes aegypti abundance were primarily dependent on the time since the last spraying and gradually decreased over 60 days. No further reduction in Aedes aegypti populations was observed as a result of the cumulative effect of multiple household sprayings. In short, the number of Aedes aegypti has decreased. The number of Aedes aegypti mosquitoes in a household depends mainly on the time that has passed since the last spraying in that household.
An important limitation of our study is that we did not control for the age of the adult Aedes aegypti mosquitoes collected. Previous analyses of these experiments [14] found a trend towards a younger age distribution of adult females (increased proportion of nulliparous females) in L-2014-treated areas compared to the buffer zone. Thus, although we did not find an additional explanatory effect of spraying in nearby households on A. aegypti abundance in a given household, we cannot be confident that there is no regional effect on A. aegypti population dynamics in areas where spraying occurs frequently.
Other limitations of our study include the inability to account for an emergency spraying conducted by the Ministry of Health approximately 2 months before the L-2014 experimental spraying due to a lack of detailed information on its location and timing. Previous analyses have shown that these sprays had similar effects across the study area, forming a common baseline for Aedes aegypti densities; indeed, Aedes aegypti populations began to recover when the experimental spraying was conducted [14]. Furthermore, the difference in results between the two experimental periods may be due to differences in study design and different susceptibility of Aedes aegypti to cypermethrin, with S-2013 being more sensitive than L-2014 [14]. We report the most consistent results from the two studies and include the model fitted to the L-2014 experiment as our final model. Given that the L-2014 experimental design is more appropriate for assessing the impact of recent spraying on Aedes aegypti mosquito populations, and that local Aedes aegypti populations had developed resistance to pyrethroids in late 2014 [41], we considered this model to be a more conservative choice and more appropriate to achieve the objectives of this study.
The relatively flat slope of the spray decay curve observed in this study may be due to a combination of the degradation rate of cypermethrin and mosquito population dynamics. The cypermethrin insecticide used in this study is a pyrethroid that degrades primarily through photolysis and hydrolysis (DT50 = 2.6–3.6 days) [ 44 ]. Although pyrethroids are generally considered to degrade rapidly after application and that residues are minimal, the degradation rate of pyrethroids is much slower indoors than outdoors, and several studies have shown that cypermethrin can persist in indoor air and dust for months after spraying [45,46,47]. Houses in Iquitos are often built in dark, narrow corridors with few windows, which may explain the reduced degradation rate due to photolysis [14]. In addition, cypermethrin is highly toxic to susceptible Aedes aegypti mosquitoes at low doses (LD50 ≤ 0.001 ppm) [48]. Due to the hydrophobic nature of residual cypermethrin, it is unlikely to affect aquatic mosquito larvae, explaining the recovery of adults from active larval habitats over time as described in the original study, with a higher proportion of non-oviparous females in treated areas than in buffer zones [14]. The life cycle of the Aedes aegypti mosquito from egg to adult can take 7 to 10 days depending on temperature and mosquito species.[49] The delay in recovery of adult mosquito populations may be further explained by the fact that residual cypermethrin kills or repels some newly emerged adults and some introduced adults from areas that have never been treated, as well as a reduction in egg laying due to the reduction in adult numbers [ 22 , 50 ].
Models that included the entire history of past household spraying had poorer accuracy and weaker effect estimates than models that included only the most recent spray date. This should not be taken as evidence that individual households do not need to be re-treated. The recovery of A. aegypti populations observed in our study, as well as in previous studies [14], shortly after spraying, suggests that households need to be re-treated at a frequency determined by local transmission dynamics to re-establish A. aegypti suppression. Spray frequency should be aimed primarily at reducing the probability of infection of female Aedes aegypti, which will be determined by the expected length of the extrinsic incubation period (EIP) – the time it takes for a vector that has gorged on infected blood to become infective to the next host. In turn, EIP will depend on the virus strain, temperature, and other factors . For example, in the case of dengue fever, even if insecticide spraying kills all infected adult vectors, the human population may remain infectious for 14 days and may infect newly emerging mosquitoes [54]. To control the spread of dengue fever, the intervals between sprayings should be shorter than the intervals between insecticide treatments to eliminate newly emerging mosquitoes that may bite infected hosts before they can infect other mosquitoes. Seven days can be used as a guideline and a convenient unit of measurement for vector control agencies. Thus, weekly insecticide spraying for at least 3 weeks (to cover the entire infectious period of the host) would be sufficient to prevent dengue fever transmission, and our results suggest that the effectiveness of the previous spraying would not be significantly reduced by that time [13]. Indeed, in Iquitos, health authorities successfully reduced dengue transmission during an outbreak by conducting three rounds of ultra-low-volume insecticide spraying in closed spaces over a period of several weeks to several months.
Finally, our results show that the impact of indoor spraying was limited to the households where it was carried out, and spraying of neighbouring households did not further reduce Aedes aegypti populations. Adult Aedes aegypti mosquitoes can remain near or inside the home where they hatch, aggregate up to 10 m away, and travel an average distance of 106 m.[36] Thus, spraying the area around a home may not have a significant effect on Aedes aegypti numbers in that home. This supports previous findings that spraying outside or around homes had no effect [18, 55]. However, as mentioned above, there may be regional effects on A. aegypti population dynamics that our model is unable to detect.
Post time: Feb-06-2025