Skip to main content
Download PDF

Measuring mosquito control: adult-mosquito catches vs egg-trap data as endpoints of a cluster-randomized controlled trial of mosquito-disseminated pyriproxyfen

Parasites & Vectors volume 13, Article number: 352 (2020) Cite this article

Abstract

Background

Aedes aegypti and Culex quinquefasciatus are the main urban vectors of arthropod-borne viruses causing human disease, including dengue, Zika, or West Nile. Although key to disease prevention, urban-mosquito control has met only limited success. Alternative vector-control tactics are therefore being developed and tested, often using entomological endpoints to measure impact. Here, we test one promising alternative and assess how three such endpoints perform at measuring its effects.

Methods

We conducted a 16-month, two-arm, cluster-randomized controlled trial (CRCT) of mosquito-disseminated pyriproxyfen (MD-PPF) in central-western Brazil. We used three entomological endpoints: adult-mosquito density as directly measured by active aspiration of adult mosquitoes, and egg-trap-based indices of female Aedes presence (proportion of positive egg-traps) and possibly abundance (number of eggs per egg-trap). Using generalized linear mixed models, we estimated MD-PPF effects on these endpoints while accounting for the non-independence of repeated observations and for intervention-unrelated sources of spatial-temporal variation.

Results

On average, MD-PPF reduced adult-mosquito density by 66.3% (95% confidence interval, 95% CI: 47.3–78.4%); Cx. quinquefasciatus density fell by 55.5% (95% CI: 21.1–74.8%), and Ae. aegypti density by 60.0% (95% CI: 28.7–77.5%). In contrast, MD-PPF had no measurable effect on either Aedes egg counts or egg-trap positivity, both of which decreased somewhat in the intervention cluster but also in the control cluster. Egg-trap data, therefore, failed to reflect the 60.0% mean reduction of adult Aedes density associated with MD-PPF deployment.

Conclusions

Our results suggest that the widely used egg-trap-based monitoring may poorly measure the impact of Aedes control; even if more costly, direct monitoring of the adult mosquito population is likely to provide a much more realistic and informative picture of intervention effects. In our CRCT, MD-PPF reduced adult-mosquito density by 66.3% in a medium-sized, spatially non-isolated, tropical urban neighborhood. Broader-scale trials will be necessary to measure MD-PPF impact on arboviral-disease transmission.

Background

Urban mosquitoes are the main vectors of arthropod-borne viruses (arboviruses) causing disease in humans. In particular, Aedes aegypti and Ae. albopictus transmit, among other pathogens, the viruses causing dengue, Zika, chikungunya, or urban yellow fever, and members of the Culex pipiens complex including Cx. quinquefasciatus transmit (also among other pathogens) the viruses causing west Nile and Rift Valley fevers or Saint Louis encephalitis [1,2,3,4]. In the absence of specific antiviral treatments and (except for yellow fever and Japanese encephalitis) vaccines with true potential for broad public-health impact, mosquito control remains the cornerstone of arbovirus transmission control and arboviral-disease prevention [1, 5].

Although key to disease prevention, urban-mosquito control has met only limited success [1, 5, 6]. Traditional programmatic control tactics rely primarily on (i) treatment or removal/destruction of aquatic larval habitats by control agents and/or the public, which is hindered by the fact that detecting and gaining access to such habitats, especially the small, cryptic larval habitats of Aedes spp., can be very difficult; and (ii) adult-mosquito-targeted insecticide spraying, which often has only short-lived effects [5,6,7,8,9,10,11,12,13,14]. Among the several alternative control tactics under development/testing [6, 10], mosquito-disseminated pyriproxyfen (MD-PPF) specifically targets the challenge of larval-habitat detection and treatment [15,16,17,18,19,20,21,22,23,24,25,26,27,28,29,30,31,32,33,34]. It does so by attracting mosquitoes to surfaces coated with tiny PPF particles, which stick to the vectors’ bodies and are thus transferred by the mosquitoes themselves to otherwise untreated larval habitats [15,16,17,18].

PPF is an insect juvenile-hormone analogue that kills immature mosquitoes at minute doses and can be safely used in drinking water [35]. MD-PPF has yielded promising results in several semi-field and field trials based on the deployment of PPF ‘dissemination stations’, i.e. small artificial larval habitats that lure egg-laying mosquitoes and drive them to land/walk on PPF-coated surfaces. Most of the field trials reported to date, however, were too small-sized to provide useful operational guidance [18, 19, 21, 26, 27, 31], and the few trials conducted at the operationally-relevant scales of neighborhoods [25, 32] or towns [28] lacked truly independent controls. Two recent, moderately-sized (city block-scale), non-randomized trials were nominally controlled, but treatment and control blocks were spatially close to one another and, despite efforts to block migration, adult-mosquito exchange between blocks, including intervention ‘leakage’ into the control block, was likely [30, 33]. In sum, the lack of randomized, adequately controlled trials conducted at operationally relevant scales means that the evidence supporting dissemination-station-based MD-PPF as a useful means for urban-mosquito control remains relatively weak [6, 34, 36].

Here, we address this gap by presenting the results of a neighborhood-scale, parallel, two-arm cluster-randomized controlled trial (CRCT) of MD-PPF. In particular, we set to measure the impact of MD-PPF on local Aedes and Culex populations through both (i) active aspiration of adult mosquitoes, which directly measures adult-mosquito density [37,38,39,40], and (ii) egg-trap-based monitoring of female Aedes presence (proportion of positive egg-traps) and possibly abundance (number of eggs per egg-trap) [37, 41, 42]. Using a CRCT design, 16 months of field data, and a rigorous statistical-modeling strategy, we show that MD-PPF can significantly reduce adult-mosquito densities, yet Aedes egg-trap-based metrics may fail to detect this reduction.

Methods

Trial setting, design, and timeline

This study took place in São Sebastião (15°54′36″S, 47°46′1″W), a lower-middle class, urban administrative region of the Federal District, Brazil. We combined map and satellite-image visual appraisal, field visits, and interviews with local health officials to select two residential clusters of similar sizes and urbanization patterns (mostly single-family homes with basic sanitation) for our CRCT. These two clusters (Fig. 1) met three main requirements: (i) reciprocal geographical isolation: about 2.5 km apart and separated by stretches of non-built environment (a gallery-forest patch and a forested park) to prevent or minimize intervention ‘leaking’ (i.e. dispersal of PPF-carrying mosquitoes) into the control cluster; (ii) epidemiological similarity: broadly comparable recent histories of arboviral-disease incidence, as judged by local health-surveillance officials; and (iii) logistic feasibility: manageable sizes (~ 1500–2000 homes each) given the project’s logistic and financial constraints. Once the two candidate clusters were selected, we tossed a coin to randomly assign one of them to the intervention (‘Intervention cluster’ in Fig. 1; IC hereafter), and retained the other as our ‘Control cluster’ (Fig. 1; CC hereafter). Mosquito monitoring was run from January 2017 to April 2018 in both clusters; in the IC, we deployed MD-PPF in April 2017 (see below). The CRCT therefore included a 3-month baseline period (BP) and a 13-month intervention period (IP); Fig. 1 includes a schematic of the trial timeline. In each cluster, we selected 30 dwellings for adult-mosquito catches and 30 different dwellings for Aedes egg-trap monitoring (Fig. 1). We aimed at achieving fair spatial coverage, with a reasonable number of sampling dwellings given logistic/financial constraints, within each cluster; therefore, we did not do a priori power/sample-size calculations and did not use randomization to select sampling dwellings.

Fig. 1
figure 1

Trial setting, design, and timeline. Cluster selection and randomization plus trial timeline and overall weather conditions (rainfall and mean of daily minimum temperatures; weather conditions in December 2016 and April 2018 (lighter grey) were included only in some analyses) (a); and study clusters (b). One-hundred and fifty pyriproxyfen dissemination stations (red stars) were deployed over 13 months in the intervention cluster (b). Mosquito monitoring was run in both clusters from January 2017 to April 2018 (a) in 120 sampling dwellings: 60 for adult-mosquito catches with battery-powered aspirators (green dots) and 60 for Aedes egg-traps (blue dots) (b). Abbreviation: MD-PPF, mosquito-disseminated pyrpiproxyfen

Full size image

Adult-mosquito catch: aspiration

We used light-weight mosquito aspirators powered by 12-V, 5-Ah motorcycle batteries (Horst Armadilhas, São Paulo, Brazil) to collect adult mosquitoes; sampling took place once per month in each dwelling, with visits occurring between ~ 8–9 am and ~ 3–4 pm. After obtaining written informed consent, one field assistant dislodged mosquitoes from typical mosquito-resting surfaces (walls, curtains, plants, furniture, clothes, etc.) and a trained collector captured them with the aspirator. In each dwelling, aspiration covered the area around the house (including verandas/porches, patios, backyards, garages, gardens, etc.) and each room inside the house itself. At each sampling occasion, we recorded aspiration time to get a measure of mosquito-catch effort and noted the identity of the collector to check for possible among-collector variation in mosquito catches. Mosquitoes were stored in labeled (dwelling/date) pots, identified using standard keys [43], and counted.

Aedes egg-trap monitoring

Each month, also after informed consent, we used two Aedes egg-traps, one indoors and one outdoors, in each sampling dwelling. Aedes egg-traps were 1-L, black plastic cups (12 cm in diameter) each fitted with a 10 × 3-cm, rough-surface fiberboard paddle suitable for Aedes egg-laying and filled with ~ 800 ml of tap water and ~80 ml of hay infusion. We set Aedes egg-traps in mosquito-prone locations and out of the reach of children or pets; after five days of operation, we collected the traps and used a stereomicroscope to count Aedes eggs present on each (trap/date-labeled) paddle.

Intervention: mosquito-disseminated pyriproxyfen (MD-PPF)

We selected 150 dwellings of the IC to deploy, after getting written informed consent, one PPF ‘dissemination station’ in each from March 2017 to April 2018 (Fig. 1). These dwellings were roughly evenly distributed across the ~ 85.5-Ha IC; similar to [28], this yielded a density of about one dissemination station for every 10 dwellings (Fig. 1). PPF dissemination stations were 1.5-l, 15-cm diameter dark plastic cups filled with 400 ml of tap water; the inner wall was lined with black, Oxford-type cloth dusted with 5 g/m2 of PPF 0.5% (SumiLarv 0.5G® granules, Sumitomo Chemical, Tokyo, Japan) ground to talcum-like powder to allow dissemination by adult mosquitoes. We placed PPF dissemination stations in sun- and rain-protected spots out of the reach of children or pets, and serviced them (refilling water and re-dusting cloth with PPF) once per month; we also asked residents to check weekly that the cups contained water and to refill them if necessary. The CC remained without PPF dissemination stations (Fig. 1). We note that routine mosquito control by local health-surveillance agents was in place in both study clusters all through the trial. Control activities consisted primarily of active searches for larval habitats, which were physically destroyed or treated with SumiLarv 0.5G® PPF granules when detected; in addition, adult mosquito-killing ultra-low volume (ULV) malathion (aqueous emulsion 44%) was applied with truck-mounted sprayers (at a rate of 150 ml a.i./ha) when surveillance data suggested local transmission of Aedes-borne viruses [44, 45]. We recorded, for each dwelling and sampling occasion, whether ULV had been applied in the same city block (i) in the seven days before sampling (to test for immediate ULV effects), (ii) eight to 30 days before sampling (to test for lagged ULV effects), or (iii) > 30 days before sampling or never during the trial. This three-level factor variable (‘ulv’) allowed us to detect and, if present, adjust for ULV effects when assessing the MD-PPF effects of focal interest.

We finally note that, due to operational and logistic constraints, all field-team members had to be involved in both entomological monitoring and MD-PPF deployment; the trial, therefore, was cluster-randomized and controlled, but not blind.

Data analysis

In an initial descriptive/exploratory step, we summarized our data using tables and graphs; we present counts, means, standard deviations, quantiles, or, when appropriate, proportions with score 95% confidence intervals (CI). In a second, inferential step, we used generalized linear mixed models (GLMMs; [46]) to estimate intervention (MD-PPF) effects on each endpoint (adult-mosquito density, as directly measured by aspiration; or female Aedes presence and possibly abundance, as indirectly measured by, respectively, egg-trap positivity and eggs per egg-trap) while accounting for:

  1. (i)

    possible differences, unrelated to the intervention, between the IC and the CC;

  2. (ii)

    the non-independence of measures taken at the same sampling dwellings in different months and during the same month in different dwellings;

  3. (iii)

    possible temporal variation due to (a) monthly weather conditions (rainfall or temperature) or (b) ULV insecticide spraying; and

  4. (iv)

    other, unmeasured sources of spatial or temporal variation.

All GLMMs contained a ‘cluster × period’ interaction, where ‘cluster’ is either the control cluster (CC) or the intervention cluster (IC) and ‘period’ is either the baseline period (BP) or the intervention period (IP) (Fig. 1); the slope coefficient estimated for the ‘IC × IP’ term measures the (link-scale) change in the endpoint variable that can be attributed to MD-PPF. All our models included, in addition, two random-intercept terms: one on dwelling ID to adjust for repeated measures, and one on study month to adjust for temporal variation not explained by fixed effects (e.g. in models including weather or ULV spraying; see below). The structure of the focal model was

$$\begin{aligned} & {\text{Y }}\sim {\text{ cluster }} \times {\text{ period }} + {\text{ weather }} + {\text{ ULV }} + {\text{ random}}\left( {\text{dwelling}} \right) \\ & \quad + {\text{ random}}\left( {\text{month}} \right) \, + {\text{ residual error}}, \\ \end{aligned}$$

where the endpoint ‘Y’ is the result of either adult-mosquito aspiration (a count) or Aedes egg-trap monitoring—an egg count or a binary 1/0 indicator of trap positivity/negativity.

Adult-mosquito catch models used the negative binomial distribution (log link-function) and further adjusted for sampling effort via an offset variable specifying the time (in 10-min units and log-transformed) spent in each sampling occasion; we, therefore, effectively modeled mosquito-catch rates per 10 min aspiration. We used this strategy to model the counts of, first, all mosquito species considered together, and, then, of Ae. aegypti and Cx. quinquefasciatus separately. In a supplementary set of analyses, we modeled female- and male-mosquito catches separately; the data, however, became too sparse to separately model sex-by-species strata. Egg-trap monitoring, on the other hand, yields information on Aedes egg density (eggs per egg-trap) and trap positivity/negativity. We jointly analyzed these two variables using two-part, zero-inflated models [47] with a binomial (logit link) submodel for the probability that a trap is negative and a negative binomial (log link) submodel for the egg counts. Both submodels had random-intercept terms on dwelling ID and month as described above.

We adjusted for weather-related temporal variation using data provided by the Brazilian Instituto Nacional de Meteorologia (www.inmet.gov.br). In particular, we built covariates measuring temperature (mean of daily minimum, mean, and maximum) and total rainfall in the week before each sampling occasion (‘tmin_w’, ‘tmean_w’, ‘tmax_w’, ‘rain_w’), in the week before that (i.e. one-week-lagged; ‘tmin_2w’ etc.), and in the month before each sampling occasion (‘tmin_m’ etc.; see Fig. 1 and Additional file 1: Table S1). We fitted GLMMs with one of these weather covariates (standardized to mean 0 and SD 1) at a time, compared model performance using the Bayesian information criterion (BIC), and selected the smallest-BIC model as our top-performing ‘full’ model for each outcome [48]. We then investigated the importance of ULV-spraying effects by removing the ‘ulv’ covariate from each top-performing ‘full’ model (or submodel in zero-inflated GLMMs) and then checking whether this removal improved or worsened model performance, i.e. whether it reduced or increased the model’s BIC score [48]. Finally, we checked for among-collector variation in mosquito-aspiration results by refitting the top-performing Aedes + Culex model with an extra random term (‘collector’) and comparing both specifications using BIC [48].

We did all our analyses in R 3.6.3 [49], using packages stats 3.6.3 [49], Hmisc 4.3-1 [50], glmmTMB 1.0.1 [51], AICcmodavg 2.2-2 [52] and bbmle 1.0.23.1 [53]. We report, for each analysis, both the numerical results of the smallest-BIC model and the predictions of that model (at selected covariate values) computed with ggeffects 0.14.2 [54]; given our focus on population-level intervention effects, we present estimated marginal means with 95% confidence intervals—not prediction intervals, which take random-effect variances into account [54]. For completeness, below we will also comment on the results of selected, non-top-ranking GLMMs of special interest, in particular, models including ULV effects. All our analyses are on an intention-to-treat basis, i.e. disregarding the occasional malfunctioning of some (5.6% overall) dissemination stations.

Results

Adult-mosquito catch: aspiration

We gathered data in 957 sampling occasions (dwelling-month aspiration events) totaling 10,267 min of adult-mosquito aspiration; mean effort was 10.7 min (SD = 4.6; median = 10.0 min) of aspiration per sampling occasion. Overall, we caught 4356 adult mosquitoes (783 Ae. aegypti and 3573 Cx. quinquefasciatus; Table 1). Raw data are available in Additional file 2: Dataset S1.

Table 1 Adult-mosquito catches using battery-powered aspirators: summary statistics
Full size table

All mosquitoes

The average catch over sampling occasions was 3.41 ± 13.28 SD adult mosquitoes per 10 min aspiration (Table 1). At baseline, mosquito density appeared to be higher in the CC (5.81) than in the IC (3.63). In the IC, mean density fell by 68.6% (to 1.14 mosquitoes per 10 min aspiration) during the intervention, whereas density barely changed in the CC (Table 1). Table 1 presents a summary of observations, overall and stratified by cluster and period; for monthly results see Additional file 1: Table S2. Our BIC-based assessment selected average minimum temperatures in the month before sampling (‘tmin_m’) as the best-fitting weather covariate (Additional file 1: Table S3). This smallest-BIC ‘full’ model, however, performed substantially worse (BIC difference 10.1 units) than a simpler, alternative model excluding ULV effects (Table 2 and Additional file 1: Table S3). After adjusting for dwelling-level repeated measures and random month-to-month variation, as well as for the (positive) effect of warmer nights (as measured by ‘tmin_m’) and for intervention-unrelated differences between clusters, this model estimates a clearly (sensu [55]) negative effect of MD-PPF on adult-mosquito catches: βIC×IP = − 1.086, 95% CI: − 1.532 to − 0.641 (Table 2). This is equivalent to an incidence rate ratio e−1.086 = 0.337, which indicates that the intervention brought about a 100 − 33.7 = 66.3% reduction (95% CI: 47.3–78.4%) of mean adult-mosquito density (Table 2). Figure 2 shows the predictions of this top-performing model for selected ‘tmin_m’ values across trial clusters and periods. Adding the ‘collector’ random term to this model increased its BIC score by 5.8 units (Additional file 1: Table S3); among-collector variation was very small (SD = 0.13), and the βIC×IP estimate (− 1.030 ± 0.247 SE) was similar to that of the top-ranking model. We finally note that the ‘full’ model including ‘ulv’ estimated a nearly-zero lagged ULV effect (βULVlag = 0.020 ± 0.166 SE) and a negative, yet imprecise, immediate ULV effect (βULVweek = − 0.936; 95% CI: − 1.880–0.007); the effect of MD-PPF remained clearly [55] negative (βIC×IP = − 1.075 ± 0.243 SE) after ULV adjustment (see Additional file 1: Table S4). These results broadly mirrored those of modeling male and female mosquito catches separately; effect estimates from the top-ranking models were βIC×IP = − 0.715 ± 0.242 SE for females and βIC×IP = − 1.451 ± 0.297 SE for males (see details in Additional file 3: Tables S11 and S12).

Table 2 Adjusted effects of mosquito-disseminated pyriproxyfen on adult-mosquito catches (Aedes aegypti + Culex quinquefasciatus): top-ranking (smallest-BIC) generalized linear mixed model
Full size table
Fig. 2
figure 2

Effects of mosquito-disseminated pyriproxyfen on adult-mosquito catches (Aedes aegypti + Culex quinquefasciatus). Predictions of the top-ranking generalized linear mixed model at selected values (in green font) of covariate ‘tmin_m’ (mean of daily minimum temperatures in the month before sampling) across trial clusters (blue, control cluster; red, intervention cluster) and periods (lighter, baseline period; darker, intervention period). The right-hand panel shows in greater detail the predictions of the model at the observed mean ‘tmin_m’ value (17.4 °C). Abbreviations: BP, baseline period; IP, intervention period

Full size image

Aedes aegypti adults

At baseline, catches of adult Ae. aegypti were particularly productive in the IC; during the intervention period, the mean catch per 10 min aspiration fell by 81.1% in the IC and by 47.5% in the CC (Table 1; see also Additional file 1: Table S2). BIC scores identified rainfall in the month before sampling (‘rain_m’) as the best-fitting weather covariate; the performance of this rainfall ‘full’ model was again improved, albeit by a small amount (BIC difference 0.98 units), by removing ULV effects (Additional file 1: Table S5). The top-ranking Ae. aegypti model (Table 3) estimates a positive effect of rainfall and a negative effect of MD-PPF as measured by the ‘IC × IP’ slope coefficient, with a 60.0% decrease (95% CI: 28.7–77.5%) of mean Aedes catches attributable to the intervention (Table 3). Figure 3 presents a selected subset of this model’s predictions. The ‘full’ model including ULV effects suggests that, relative to no recent ULV spraying, mean Aedes catch was actually higher when ULV had been applied between 8 and 30 days before sampling (βULVlag = 0.607 ± 0.193 SE), but lower when ULV had been applied during the week before sampling (βULVweek = − 1.061 ± 0.511 SE); MD-PPF still had an overall negative, ULV-adjusted effect on adult Ae. aegypti density (a 44.7% decrease), but the 95% CI ran from a slight (1.92%) increase to a clear (70.0%) decrease (Additional file 1: Table S6).

Table 3 Adjusted effects of mosquito-disseminated pyriproxyfen on species-specific adult-mosquito catches (Aedes aegypti and Culex quinquefasciatus): top-ranking (smallest-BIC) generalized linear mixed models
Full size table
Fig. 3
figure 3

Effects of mosquito-disseminated pyriproxyfen on species-specific adult-mosquito catches (Aedes aegypti and Culex quinquefasciatus). Predictions of the top-ranking generalized linear mixed models at mean values (in green font) of covariates ‘rain_m’ (total rainfall in the month before sampling) and ‘tmin_m’ (mean of daily minimum temperatures in the month before sampling) across trial clusters (blue, control cluster; red, intervention cluster) and periods (lighter, baseline period; darker, intervention period). Abbreviations: BP, baseline period; IP, intervention period

Full size image

Culex quinquefasciatus adults

The density of adult southern house mosquitoes was particularly high in the CC both at baseline and during the intervention period; in the IC, mean adult Culex catches were 39.5% lower during MD-PPF deployment than at baseline (Table 1 and Additional file 1: Table S2). Our BIC-based assessment selected ‘tmin_m’ as the best-fitting weather covariate. However, the Culex GLMM that did not include ULV-spraying effects clearly outperformed (BIC difference 11.9 units) the ‘full’ model with both ‘tmin_m’ and ‘ulv’ (Additional file 1: Table S7). This top-performing model (Table 3) suggests, after adjusting for multiple additional sources of variation, that MD-PPF reduced adult Cx. quinquefasciatus density by about 55.4% (95% CI: 21.1–74.8%) (see Table 3 and Fig. 3). The ‘full’ Culex model estimates a near-zero lagged ULV effect (βULVlag = 0.009 ± 0.204 SE), a negative but imprecise immediate ULV effect (βULVweek = − 0.769 ± 0.561 SE), and a negative effect of MD-PPF deployment (βIC×IP = − 0.801 ± 0.314 SE) (Additional file 1: Table S8).

Aedes egg-trap monitoring

We gathered data from a total of 1879 Aedes egg-traps-week; the main descriptive results are summarized in Table 4. At baseline, both Aedes egg-trap positivity and mean Aedes egg counts per egg-trap were somewhat lower in the IC than in the CC (Table 4), despite substantially higher mean adult Aedes catches in the latter (Table 1). When comparing the baseline and intervention periods, the data suggest that Aedes egg-trap-based metrics decreased somewhat in the CC but remained largely stable (perhaps with a slight decline of egg-trap positivity) in the IC (Table 4; see Additional file 1: Table S9 for monthly results and Additional file 2: Dataset S1 for the raw data). Our BIC-based assessment selected ‘tmin_w’ as the best-fitting weather covariate; the top-performing ‘full’ model including also ULV effects (in both the egg-count and the trap-negativity submodels) was again outperformed by simpler models lacking ‘ulv’ in either submodel, and these models, in turn, were clearly outperformed (BIC differences > 9 units) by an even simpler model excluding ULV effects from both the negative binomial (egg-count) and the binomial (trap-negativity) submodels (Additional file 1: Table S10). This top-ranking model suggests (after multiple adjustments as noted above) that there were no measurable cluster, period, or MD-PPF effects on Aedes egg counts per egg-trap, which were however clearly higher when nights were warmer; the binomial submodel, on the other hand, predicts lower egg-trap positivity in the IC (11.9% at baseline, 10.1% during the intervention) than in the CC (23.1% and 16.2%, respectively) (Table 5 and Fig. 4).

Table 4 Egg-trap based monitoring of Aedes aegypti: summary statistics
Full size table
Table 5 Adjusted effects of mosquito-disseminated pyriproxyfen on Aedes egg-trap-derived endpoints: numerical results of the top-ranking (smallest-BIC) zero-inflated generalized linear mixed model
Full size table
Fig. 4
figure 4

Effects of mosquito-disseminated pyriproxyfen on Aedes egg-trap-derived endpoints. Predictions of the top-ranking zero-inflated generalized linear mixed model at mean values (in green font) of covariate ‘tmin_w’ (mean of daily minimum temperatures in the week before sampling) across trial clusters (blue, control cluster; red, intervention cluster) and periods (lighter, baseline period; darker, intervention period). Abbreviations: BP, baseline period; IP, intervention period

Full size image

Discussion

In this report, we present results of a parallel, two-arm, cluster-randomized controlled trial (CRCT) of mosquito-disseminated pyriproxyfen (MD-PPF); our study yielded two key findings and an additional, potentially useful insight. First, our MD-PPF-based intervention clearly reduced (by 66.3% on average) adult-mosquito density in a medium-sized, spatially non-isolated, lower-middle class urban neighborhood of central Brazil. Secondly, trial endpoints based on Aedes egg-trap monitoring failed to capture this reduction, which reached 60.0% on average for adult Ae. aegypti. Finally, our analyses indicate that pulses of ultra-low volume malathion spraying (ULV) had at most a very transient (about one-week) effect on local adult-mosquito density, with, again, no measurable impact on Aedes egg-trap-derived metrics.

Our CRCT provided strong evidence [36] that MD-PPF can considerably reduce adult-mosquito density at the neighborhood scale. Using rigorous statistical modeling to adjust for multiple potential sources of spatial-temporal variation (see Tables 2, 3 and 5, Additional file 1: Tables S1–S10, and “Methods”), we show that mean adult-mosquito density was nearly three times higher before than during MD-PPF deployment in the residential cluster randomized to receive the intervention; in contrast, adult-mosquito density remained stable across periods in the control cluster (Fig. 2). This key result was consistent across datasets (Figs. 2 and 3, Tables 2 and 3), suggesting broadly comparable intervention effects on adult Ae. aegypti and Cx. quinquefasciatus, the two most important urban vectors of disease-causing arboviruses [1,2,3,4].

Putting these results in the context of previous findings is at best problematic. Even in the case of field trials using PPF dissemination stations and measuring adult-mosquito density at scales similar to ours (city block or neighborhood), methodological issues complicate direct comparisons. For example, the clear reduction of adult-mosquito density we report is larger than found in similarly-sized MD-PPF field trials reporting no effects [30, 33] or in smaller field trials reporting moderate, variable effects [21, 32]. Beyond the facts that we (i) measured adult-mosquito density via active aspiration, not trapping, and (ii) used very simple dissemination stations with low-concentration (0.5% a.i.) PPF, these differences may reflect small trial size in [21, 32] or non-independence of intervention and control blocks in [30, 33]. In contrast, our current estimates of MD-PPF effects on adult-mosquito density are quantitatively below what would be expected if, as suggested by neighborhood- or town-scale trials using the same dissemination stations and PPF formulation, MD-PPF had reached most (> 90%) aquatic larval habitats and killed most (~ 70–90%) immature mosquitoes before adult emergence [25, 28]. It should be noted, first, that our IC was spatially isolated from the CC, but not from neighboring, untreated residential areas (Fig. 1); we therefore expected that mosquito immigration into the IC would weaken intervention effects. This was also the case in the study by Abad-Franch et al. [25] and probably in all other previous field trials of MD-PPF [19, 21, 23, 24, 26, 27, 30,31,32,33], with one exception, i.e. the only citywide study reported so far, in which mosquito immigration was particularly unlikely and MD-PPF had a particularly strong impact on local mosquito populations [28]. Secondly, we note that the intensity of the intervention was comparatively low in our trial, in the sense that we deployed ‘just’ 150 PPF dissemination stations (vs 1000 in [28]) at relatively low spatial densities (vs 100 spatially-clustered stations in [25]) and, importantly, serviced them every four weeks (vs fortnightly in [25, 28]).

We therefore stress that the effects of MD-PPF we report came about in spite of the spatial non-isolation of the IC and the relatively low intensity of the intervention including monthly servicing of PPF dissemination stations. This suggests that effective MD-PPF-based interventions may be even more readily scalable than indicated by previous trials using the same, very simple dissemination stations and PPF formulation we used here [25, 28]. Finally, and along these lines, it is clear that study-site peculiarities (in terms of, e.g. mosquito species composition or population density, availability of larval habitats, or climate), as well as other factors (e.g. dissemination-station design), may also contribute to variation among MD-PPF trial results. Our results come with the additional caveats that (i) our sampling dwellings were a roughly systematic, relatively small, nonrandom sample of the dwellings in the study clusters, and (ii) field and laboratory teams were not blinded to the intervention. The picture that is emerging, however, is one of overall support for the view of MD-PPF as a very promising addition to the urban-mosquito control toolkit [15,16,17,18,19,20,21,22,23,24,25,26,27,28,29,30,31,32,33,34, 56]. Because of the trial’s cluster-randomized controlled design and the thorough strategy used for data analysis, and in spite of its limitations, our study adds some of the strongest evidence yet in favor of this view.

One additional outcome of our CRCT was that Aedes egg-trap-based metrics failed to reflect the impact of MD-PPF on adult Ae. aegypti density as directly measured by mosquito aspiration (Tables 3 and 5, Figs. 3 and 4). Further, the observed baseline density of Aedes adult mosquitoes was clearly higher in the IC than in either the CC (both periods) or the IC during MD-PPF deployment, yet Aedes egg-trap positivity was not particularly high in the baseline IC data, and the number of Aedes eggs per egg-trap was actually the lowest we observed (see Tables 1 and 4). Three not mutually-exclusive hypotheses come to mind as plausible candidates to explain these observations. First, Aedes egg-trap positivity may remain largely unchanged if a small number of females can lay eggs in about as many sites as a substantially larger number of females [57] (Fig. 5). Secondly, low Aedes egg-trap positivity may be expected, irrespective of Aedes density, if so many suitable egg-laying sites are locally available that egg-traps are seldom chosen [37] (Fig. 5). Our field observations indicate that potential Aedes larval habitats, and in particular discarded trash items, may indeed have been more common in the IC than in the CC, but we did not measure this in a systematic way amenable to quantitative analysis. Finally, density-dependent egg-laying by Aedes females might yield a pattern of similar egg densities despite different adult abundances; this may occur, for example, if ‘lone’ Aedes females tend to lay many eggs in vacant egg-trap paddles, whereas females co-visiting a ‘crowded’ paddle tend to lay just a few eggs each [57,58,59,60,61] (Fig. 5). Whatever the actual underlying mechanism(s), our results suggest that the widely used egg-trap-based monitoring may measure poorly the impact of Aedes control; even if more costly, direct monitoring of the adult mosquito population is likely to provide a much more realistic and informative picture of intervention effects [5, 37,38,39,40, 62]. The fact that we found very little among-collector variation in mosquito-aspiration catches lends further support to this idea.

Fig. 5
figure 5

Three possible mechanisms underlying the observed mismatch between adult Aedes aegypti catches and egg-trap positivity (a) and egg density (b). a1: a small number of females may lay eggs in about as many sites as a large number of females; hence, egg-trap positivity remains largely unchanged; a2: if many alternative egg-laying sites are locally available, egg-traps may be relatively less attractive to Aedes females; hence, egg-trap positivity may be low regardless of adult-Aedes density; b: at high Aedes densities, females gather at ‘crowded’ egg-trap paddles and each female (black, blue, green, red) lays just a few eggs (left panel); at low adult densities, the earliest-arriving female (black) lays many eggs, and later-arriving females (red) lay just a few eggs (central panel); at very low Aedes densities, ‘lone’ females may lay many eggs each in otherwise vacant egg-trap paddles (right panel). Abbreviation: PPF, pyriproxyfen

Full size image

A final finding of our study was that ultra-low volume malathion spraying (ULV) had overall negligible effects on local mosquito populations, as shown by the fact that models including the ‘ulv’ covariate did not perform any better than models without it [48]. At most, we found some evidence that ULV may be associated with a very short-lived (about one-week) decrease of adult-mosquito densities [7,8,9, 14], possibly with a larger effect on Ae. aegypti than on Cx. quinquefasciatus. The seemingly ‘positive’ effect of ULV applied 8–30 days before sampling on Ae. aegypti (Additional file 1: Table S5) probably reflects the fact ULV is usually deployed when disease surveillance detects signals of local transmission of Aedes-borne viruses, an event that is presumably associated with high Aedes densities [62].

Conclusions

In summary, our cluster-randomized controlled trial provided strong evidence that mosquito-disseminated pyriproxyfen (MD-PPF) can significantly reduce adult-mosquito densities at the urban-neighborhood scale. MD-PPF had a clear negative impact on both Ae. aegypti and Cx. quinquefasciatus; importantly, it did so in spite of (i) very likely immigration of adult mosquitoes from adjacent, non-treated areas and (ii) a rather long (four-week) time-lag between PPF dissemination-station servicing visits. On the other hand, we found that egg-trap-based indices may perform poorly at measuring Aedes control; direct monitoring of adult-mosquito populations (e.g. with aspirators or traps) is probably much more informative about intervention effects. If egg-traps are to remain widely used in routine surveillance and in the study of Aedes ecology and control, the mechanisms that underlie the mismatch between egg-trap data and adult-mosquito density should be further elucidated. Finally, and in line with previous reports, we found that ‘pulses’ of ULV space spraying had a very limited, short-lived effect on adult-mosquito densities. This report, in sum, adds important new findings to the growing body of evidence suggesting that MD-PPF can develop into a major tool for urban-mosquito control. The next, decisive step should be to dependably measure the impact of MD-PPF, alone or in combination with other tactics [5, 6], on mosquito-borne disease transmission in the community. Crucially, our results and those from previous trials strongly suggest that, if it is to protect people from infection and disease, MD-PPF will have to be deployed over whole neighborhoods or entire towns. Extensive efforts to test this hypothesis are currently underway in Brazil.

Availability of data and materials

Data supporting the conclusions of this article are included in the article and its additional files.

Abbreviations

PPF:

pyriproxyfen

MD-PPF:

mosquito-disseminated pyriproxyfen

CRCT:

cluster-randomized controlled trial

CC:

control cluster

IC:

intervention cluster

BP:

baseline period

IP:

intervention period

ID:

identity of each sampling dwelling

ULV:

ultra-low volume insecticide spraying

a.i.:

active ingredient

ha:

hectare

GLMM:

generalized linear mixed model

BIC:

Bayesian information criterion

SE:

standard error

SD:

standard deviation

95% CI:

95% confidence interval

References

  1. Ferguson NM. Challenges and opportunities in controlling mosquito-borne infections. Nature. 2018;559:490–7.

    CAS  PubMed  Google Scholar 

  2. Brady OJ, Hay SI. The global expansion of dengue: how Aedes aegypti mosquitoes enabled the first pandemic arbovirus. Annu Rev Entomol. 2020;65:191–208.

    CAS  PubMed  Google Scholar 

  3. Paupy C, Delatte H, Bagny L, Corbel V, Fontenille D. Aedes albopictus, an arbovirus vector: from the darkness to the light. Microbes Infect. 2009;11:1177–85.

    CAS  PubMed  Google Scholar 

  4. Turell MJ. Members of the Culex pipiens complex as vectors of viruses. J Am Mosq Control Assoc. 2012;28(Suppl. 4):123–6.

    PubMed  Google Scholar 

  5. Morrison AC, Zielinski-Gutierrez E, Scott TW, Rosenberg R. Defining challenges and proposing solutions for control of the virus vector Aedes aegypti. PLoS Med. 2008;5:e68.

    PubMed  PubMed Central  Google Scholar 

  6. Achee NL, Grieco JP, Vatandoost H, Seixas G, Pinto J, Ching-Ng L, et al. Alternative strategies for mosquito-borne arbovirus control. PLoS Negl Trop Dis. 2019;13:e0006822.

    PubMed  PubMed Central  Google Scholar 

  7. Perich MJ, Tidwell MA, Williams DC, Sardelis MR, Pena CJ, Mandeville D, et al. Comparison of ground and aerial ultra-low volume applications of malathion against Aedes aegypti in Santo Domingo, Dominican Republic. J Am Mosq Control Assoc. 1990;6:1–6.

    CAS  PubMed  Google Scholar 

  8. Reiter P, Nathan MB. Guidelines for assessing the efficacy of insecticidal space sprays for control of the dengue vector Aedes aegypti. Geneva: World Health Organization; 2001. https://apps.who.int/iris/bitstream/handle/10665/67047/WHO_CDS_CPE_PVC_2001.1.pdf.

  9. Esu E, Lenhart A, Smith L, Horstick O. Effectiveness of peridomestic space spraying with insecticide on dengue transmission: systematic review. Trop Med Int Health. 2010;15:619–31.

    PubMed  Google Scholar 

  10. Achee NL, Gould F, Perkins TA, Reiner RC Jr, Morrison AC, Ritchie SA, et al. A critical assessment of vector control for dengue prevention. PLoS Negl Trop Dis. 2015;9:e0003655.

    PubMed  PubMed Central  Google Scholar 

  11. Bowman LR, Donegan S, McCall PJ. Is dengue vector control deficient in effectiveness or evidence? Systematic review and meta-analysis. PLoS Negl Trop Dis. 2016;10:e0004551.

    PubMed  PubMed Central  Google Scholar 

  12. Alvarado-Castro V, Paredes-Solís S, Nava-Aguilera E, Morales-Pérez A, Alarcón-Morales L, Balderas-Vargas NA, et al. Assessing the effects of interventions for Aedes aegypti control: systematic review and meta-analysis of cluster randomised controlled trials. BMC Public Health. 2017;17:384.

    PubMed  PubMed Central  Google Scholar 

  13. Bardach AE, García-Perdomo HA, Alcaraz A, López ET, Gándara RAR, Ruvinsky S, et al. Interventions for the control of Aedes aegypti in Latin America and the Caribbean: systematic review and meta-analysis. Trop Med Int Health. 2019;24:530–52.

    PubMed  Google Scholar 

  14. Gunning CE, Okamoto KW, Astete H, Vasquez GM, Erhardt E, Del Aguila C, et al. Efficacy of Aedes aegypti control by indoor ultra low volume (ULV) insecticide spraying in Iquitos, Peru. PLoS Negl Trop Dis. 2018;12:e0006378.

    PubMed  PubMed Central  Google Scholar 

  15. Itoh T. Control of DF/DHF vector, Aedes mosquito, with insecticides. Trop Med. 1993;35:259–67.

    Google Scholar 

  16. Dell Chism B, Apperson CS. Horizontal transfer of the insect growth regulator pyriproxyfen to larval microcosms by gravid Aedes albopictus and Ochlerotatus triseriatus mosquitoes in the laboratory. Med Vet Entomol. 2003;17:211–20.

    CAS  PubMed  Google Scholar 

  17. Sihuincha M, Zamora-Perea E, Orellana-Rios W, Stancil JD, López-Sifuentes V, Vidal-Oré C, et al. Potential use of pyriproxyfen for control of Aedes aegypti (Diptera: Culicidae) in Iquitos, Peru. J Med Entomol. 2005;42:620–30.

    CAS  PubMed  Google Scholar 

  18. Devine GJ, Perea EZ, Killeen GF, Stancil JD, Clark SJ, Morrison AC. Using adult mosquitoes to transfer insecticides to Aedes aegypti larval habitats. Proc Natl Acad Sci USA. 2009;106:11530–4.

    CAS  PubMed  Google Scholar 

  19. Caputo B, Ienco A, Cianci D, Pombi M, Petrarca V, Baseggio A, et al. The “auto-dissemination” approach: a novel concept to fight Aedes albopictus in urban areas. PLoS Negl Trop Dis. 2012;6:e1793.

    PubMed  PubMed Central  Google Scholar 

  20. Ohba S, Ohashi K, Pujiyati E, Higa Y, Kawada H, Mito N, et al. The effect of pyriproxyfen as a “population growth regulator” against Aedes albopictus under semi-field conditions. PLoS ONE. 2012;8:e67045.

    Google Scholar 

  21. Ponlawat A, Fansiri T, Kurusarttra S, Pongsiri S, McCardle PW, Evans BP, et al. Development and evaluation of a pyriproxyfen-treated device to control the dengue vector, Aedes aegypti (L.) (Diptera: Culicidae). Southeast Asian J Trop Med Public Health. 2013;44:167–78.

    PubMed  Google Scholar 

  22. Snetselaar J, Andriessen R, Suer RA, Osinga AJ, Knols BG, Farenhorst M. Development and evaluation of a novel contamination device that targets multiple life-stages of Aedes aegypti. Parasit Vectors. 2014;7:200.

    PubMed  PubMed Central  Google Scholar 

  23. Suman DS, Farajollahi A, Healy S, Williams GM, Wang Y, Schoeler G, et al. Point-source and area-wide field studies of pyriproxyfen autodissemination against urban container-inhabiting mosquitoes. Acta Trop. 2014;135:96–103.

    CAS  PubMed  Google Scholar 

  24. Mains JW, Brelsfoard CL, Dobson SL. Male mosquitoes as vehicles for insecticide. PLoS Negl Trop Dis. 2015;9:e0003406.

    PubMed  PubMed Central  Google Scholar 

  25. Abad-Franch F, Zamora-Perea E, Ferraz G, Padilla-Torres SD, Luz SLB. Mosquito-disseminated pyriproxyfen yields high breeding-site coverage and boosts juvenile mosquito mortality at the neighborhood scale. PLoS Negl Trop Dis. 2015;9:e0003702.

    PubMed  PubMed Central  Google Scholar 

  26. Chandel K, Suman DS, Wang Y, Unlu I, Williges E, Williams GM, et al. Targeting a hidden enemy: pyriproxyfen autodissemination strategy for the control of the container mosquito Aedes albopictus in cryptic habitats. PLoS Negl Trop Dis. 2016;10:e0005235.

    PubMed  PubMed Central  Google Scholar 

  27. Kartzinel MA, Alto BW, Deblasio MW 2nd, Burkett-Cadena ND. Testing of visual and chemical attractants in correlation with the development and field evaluation of an autodissemination station for the suppression of Aedes aegypti and Aedes albopictus in Florida. J Am Mosq Control Assoc. 2016;32:194–202.

    PubMed  Google Scholar 

  28. Abad-Franch F, Zamora-Perea E, Luz SLB. Mosquito-disseminated insecticide for citywide vector control and its potential to block arbovirus epidemics: entomological observations and modeling result from Amazonian Brazil. PLoS Med. 2017;14:e1002213.

    PubMed  PubMed Central  Google Scholar 

  29. Buckner EA, Williams KF, Marsicano AL, Latham MD, Lesser CR. Evaluating the vector control potential of the In2Care® mosquito trap against Aedes aegypti and Aedes albopictus under semifield conditions in Manatee County, Florida. J Am Mosq Control Assoc. 2017;33:193–9.

    PubMed  Google Scholar 

  30. Unlu I, Suman DS, Klinger K, Faraji A, Gaugler R. Effectiveness of autodissemination stations containing pyriproxyfen in reducing immature Aedes albopictus populations. Parasit Vectors. 2017;10:139.

    PubMed  PubMed Central  Google Scholar 

  31. Suman DS, Wang Y, Faraji A, Williams GM, Wiliges E, Gaugler R. Seasonal field efficacy of pyriproxyfen autodissemination stations against container-inhabiting mosquito Aedes albopictus under different habitat conditions. Pest Manag Sci. 2018;74:885–95.

    CAS  PubMed  Google Scholar 

  32. Seixas G, Paul REL, Pires B, Alves G, Jesus AD, et al. An evaluation of efficacy of the auto-dissemination technique as a tool for Aedes aegypti control in Madeira, Portugal. Parasit Vectors. 2019;12:202.

    PubMed  PubMed Central  Google Scholar 

  33. Unlu I, Rochlin I, Suman DS, Wang Y, Chandel K, Gaugler R. Large-scale operational pyriproxyfen autodissemination deployment to suppress the immature Asian tiger mosquito (Diptera: Culicidae) populations. J Med Entomol. 2020. https://doi.org/10.1093/jme/tjaa011.

    Article  PubMed  Google Scholar 

  34. Maoz D, Ward T, Samuel M, Müller P, Runge-Ranzinger S, Toledo J, et al. Community effectiveness of pyriproxyfen as a dengue vector control method: a systematic review. PLoS Negl Trop Dis. 2017;11:e0005651.

    PubMed  PubMed Central  Google Scholar 

  35. WHO. Pyriproxyfen in drinking-water: use for vector control in drinking-water sources and containers. Background document for development of WHO guidelines for drinking water quality. Geneva: World Health Organization; 2008.

  36. Wilson AL, Boelaert M, Kleinschmidt I, Pinder M, Scott TW, Tusting LS, et al. Evidence-based vector control? Improving the quality of vector control trials. Trends Parasitol. 2015;31:380–90.

    PubMed  Google Scholar 

  37. Focks DA. A review of entomological sampling methods and indicators for dengue vectors. Geneva: World Health Organization; 2003. https://apps.who.int/iris/bitstream/handle/10665/68575/TDR_IDE_DEN_03.1.pdf.

  38. Maciel-de-Freitas R, Peres RC, Alves F, Brandolini MB. Mosquito traps designed to capture Aedes aegypti (Diptera: Culicidae) females: preliminary comparison of Adultrap, MosquiTRAP and backpack aspirator efficiency in a dengue-endemic area of Brazil. Mem Inst Oswaldo Cruz. 2008;103:602–5.

    PubMed  Google Scholar 

  39. Vazquez P, Galvin WA, Kelly R, Kitron U. A new, cost-effective, battery-powered aspirator for adult mosquito collections. J Med Entomol. 2009;46:1256–9.

    Google Scholar 

  40. Koyoc-Cardeña E, Medina-Barreiro A, Cohuo-Rodríguez A, Pavía-Ruz N, Lenhart A, Ayora-Talavera G, et al. Estimating absolute indoor density of Aedes aegypti using removal sampling. Parasit Vectors. 2019;12:250.

    PubMed  PubMed Central  Google Scholar 

  41. Resende MC, Silva IM, Ellis BR, Eiras AE. A comparison of larval, ovitrap and MosquiTRAP surveillance for Aedes (Stegomyia) aegypti. Mem Inst Oswaldo Cruz. 2013;108:1024–30.

    PubMed  PubMed Central  Google Scholar 

  42. Nascimento KLC, da Silva JFM, Zequi JAC, Lopes J. Comparison between larval survey index and positive ovitrap index in the evaluation of populations of Aedes (Stegomyia) aegypti (Linnaeus, 1762) north of Paraná, Brazil. Environ Health Insights. 2020;14:178630219886570.

    Google Scholar 

  43. Consoli RAGB, Lourenço-de-Oliveira R. Principais mosquitos de importância sanitária no Brasil. 1st ed. Rio Janeiro: Fundação Oswaldo Cruz; 1994.

    Google Scholar 

  44. Ministério da Saúde, Brasil. Diretrizes Nacionais para a prevenção e controle de epidemias de dengue. Brasília: Ministério da Saúde; 2009. https:bvsms.saude.gov.br/bvs/publicacoes/diretrizes_nacionais_prevencao_controle_dengue.pdf.

  45. Ministério da Saúde, Brasil. Recomendações sobre o uso de malathion emulsão aquosa – EA 44% para o controle de Aedes aegypti em aplicações espaciais a ultra baixo volume (UBV). Brasília: Ministério da Saúde; 2014. https://www.saude.gov.br/images/pdf/2014/setembro/02/Recomenda—es-para-o-uso-de-malathion-EW.pdf.

  46. Bolker BM, Brooks ME, Clark CJ, Geange SW, Poulsen JR, Stevens MH, et al. Generalized linear mixed models: a practical guide for ecology and evolution. Trends Ecol Evol. 2009;24:127–35.

    PubMed  Google Scholar 

  47. Martin TG, Wintle BA, Rhodes JR, Kuhnert PM, Field SA, Low-Choy SJ, et al. Zero tolerance ecology: improving ecological inference by modelling the source of zero observations. Ecol Lett. 2005;8:1235–46.

    PubMed  Google Scholar 

  48. Aho K, Derryberry D, Peterson T. Model selection for ecologists: the worldviews of AIC and BIC. Ecology. 2014;95:631–6.

    PubMed  Google Scholar 

  49. R Development Core Team. R: A language and environment for statistical computing. Version 3.6.3. Vienna: R Foundation for Statistical Computing. 2020. https://www.R-project.org.

  50. Harrell FE. Hmisc: a package of miscellaneous R functions. 2014. http://biostat.mc.vanderbilt.edu/Hmisc.

  51. Brooks ME, Kristensen K, van Benthem KJ, Magnusson A, Berg CW, Nielsen A, et al. glmmTMB balances speed and flexibility among packages for zero-inflated generalized linear mixed modeling. The R Journal. 2017;9:378–400.

    Google Scholar 

  52. Mazerolle MJ. AICcmodavg: model selection and multimodel inference based on (Q)AIC(c). R package version 2.2-2. 2019. https://cran.r-project.org/package=AICcmodavg.

  53. Bolker BM. Package bbmle: tools for general maximum likelihood estimation v. 1.0.23.1. 2019. https://cran.r-project.org/web/packages/bbmle/index.html.

  54. Lüdecke D. ggeffects: tidy data frames of marginal effects from regression models. J Open Source Softw. 2018;3:772.

    Google Scholar 

  55. Dushoff J, Kain MP, Bolker BM. I can see clearly now: reinterpreting statistical significance. Methods Ecol Evol. 2019;10:756–9.

    Google Scholar 

  56. von Seidlein L, Kekulé AS, Strickman D. Novel vector control approaches: the future for prevention of Zika virus transmission? PLoS Med. 2017;14:e1002219.

    Google Scholar 

  57. Chadee DD. Oviposition strategies adopted by gravid Aedes aegypti (L.) (Diptera: Culicidae) as detected by ovitraps in Trinidad, West Indies (2002–2006). Acta Trop. 2009;111:279–83.

    CAS  PubMed  Google Scholar 

  58. Chadee DD, Corbet PS, Greenwood JJD. Egg-laying yellow fever mosquitoes avoid sites containing eggs laid by themselves or by conspecifics. Entomol Exp Appl. 1990;57:295–8.

    Google Scholar 

  59. Fonseca DM, Kaplan LR, Heiry RA, Strickman D. Density-dependent oviposition by female Aedes albopictus (Diptera: Culicidae) spreads eggs among containers during the summer but accumulates them in the fall. J Med Entomol. 2015;52:705–12.

    CAS  PubMed  Google Scholar 

  60. Chadee DD. Oviposition response of Aedes aegypti (L.) to presence of conspecific eggs in the field in Trinidad, West Indies. J Fla Mosq Control Assoc. 1993;64:63–6.

    Google Scholar 

  61. Apostol BL, Black WC 4th, Reiter P, Miller BR. Use of randomly amplified polymorphic DNA amplified by polymerase chain reaction markers to estimate the number of Aedes aegypti families at oviposition sites in San Juan, Puerto Rico. Am J Trop Med Hyg. 1994;51:89–97.

    CAS  PubMed  Google Scholar 

  62. Scott TW, Morrison AC. Aedes aegypti density and the risk of dengue-virus transmission. In: Takken W, Scott TW, editors. Ecological aspects for application of genetically modified mosquitoes. Dordrecht: Frontis; 2003. p. 187–206.

    Google Scholar 

Download references

Acknowledgements

We thank the study-site neighbors and local health-surveillance workers and officials (especially Milton L. Couthinho of the Diretoria de Vigilância Ambiental em Saúde, Secretaria de Saúde do Distrito Federal, Brazil) for their support. José B. Bezerra provided technical assistance throughout the trial.

Funding

The Fundação de Amparo à Pesquisa do Distrito Federal (FAP-DF), Brazil, provided primary funding (grant #6142). Additional support came from the Ministério da Saúde, Brazil, through the Programa Nacional de Controle da Dengue (Secretaria de Vigilância em Saúde, PNDC/SVS-MS; project 25792.000190/2016-92) and the Departamento de Ciência e Tecnologia (Decit-MS; project 25792.000131/2016-14). Scholarships from the Coordenação de Aperfeiçoamento de Pessoal de Nível Superior (CAPES), Brazil, finance code 001 (KKSG, HSV and BLC); FAP-DF (JPAC, VGCB and TTCM-S); and the Universidade de Brasília (TOA and GDG) are gratefully acknowledged. The funders had no role in the design of the study, in the collection, analysis or interpretation of the data, or in the preparation of the manuscript.

Author information

Author notes

  1. Klauss K. S. Garcia and Hanid S. Versiani contributed equally to this work

  2. Rodrigo Gurgel-Gonçalves and Fernando Abad-Franch are joint senior authors

Authors and Affiliations

  1. Núcleo de Medicina Tropical, Faculdade de Medicina, Universidade de Brasília, Brasilia, Brazil

    Klauss K. S. Garcia, João P. A. Conceição, Marcos T. Obara, Walter M. Ramalho, Bruno L. Carvalho, Rodrigo Gurgel-Gonçalves & Fernando Abad-Franch

  2. Secretaria de Vigilância em Saúde, Ministério da Saúde, Brasilia, Brazil

    Klauss K. S. Garcia, Thaís T. C. Minuzzi-Souza & Elisa N. Vianna

  3. Laboratório de Parasitologia Médica e Biologia de Vetores, Faculdade de Medicina, Universidade de Brasília, Brasilia, Brazil

    Hanid S. Versiani, Taís O. Araújo, João P. A. Conceição, Marcos T. Obara, Thaís T. C. Minuzzi-Souza, Gustavo D. Gomes, Elisa N. Vianna, Renata V. Timbó, Vinicios G. C. Barbosa, Maridalva S. P. Rezende, Luciana P. F. Martins, Glauco O. Macedo, Nadjar Nitz & Rodrigo Gurgel-Gonçalves

  4. Laboratório Interdisciplinar de Biociências, Faculdade de Medicina, Universidade de Brasília, Brasilia, Brazil

    Hanid S. Versiani & Nadjar Nitz

  5. Diretoria de Vigilância Ambiental em Saúde, Subsecretaria de Vigilância à Saúde, Secretaria de Estado de Saúde do Distrito Federal, Brasilia, Brazil

    Israel M. Moreira & Lorrainy A. Bartasson

  6. Instituto Leônidas e Maria Deane–Fiocruz Amazônia, Manaus, Brazil

    Sérgio L. B. Luz

Authors
  1. Klauss K. S. Garcia
    View author publications

    You can also search for this author in PubMed Google Scholar

  2. Hanid S. Versiani
    View author publications

    You can also search for this author in PubMed Google Scholar

  3. Taís O. Araújo
    View author publications

    You can also search for this author in PubMed Google Scholar

  4. João P. A. Conceição
    View author publications

    You can also search for this author in PubMed Google Scholar

  5. Marcos T. Obara
    View author publications

    You can also search for this author in PubMed Google Scholar

  6. Walter M. Ramalho
    View author publications

    You can also search for this author in PubMed Google Scholar

  7. Thaís T. C. Minuzzi-Souza
    View author publications

    You can also search for this author in PubMed Google Scholar

  8. Gustavo D. Gomes
    View author publications

    You can also search for this author in PubMed Google Scholar

  9. Elisa N. Vianna
    View author publications

    You can also search for this author in PubMed Google Scholar

  10. Renata V. Timbó
    View author publications

    You can also search for this author in PubMed Google Scholar

  11. Vinicios G. C. Barbosa
    View author publications

    You can also search for this author in PubMed Google Scholar

  12. Maridalva S. P. Rezende
    View author publications

    You can also search for this author in PubMed Google Scholar

  13. Luciana P. F. Martins
    View author publications

    You can also search for this author in PubMed Google Scholar

  14. Glauco O. Macedo
    View author publications

    You can also search for this author in PubMed Google Scholar

  15. Bruno L. Carvalho
    View author publications

    You can also search for this author in PubMed Google Scholar

  16. Israel M. Moreira
    View author publications

    You can also search for this author in PubMed Google Scholar

  17. Lorrainy A. Bartasson
    View author publications

    You can also search for this author in PubMed Google Scholar

  18. Nadjar Nitz
    View author publications

    You can also search for this author in PubMed Google Scholar

  19. Sérgio L. B. Luz
    View author publications

    You can also search for this author in PubMed Google Scholar

  20. Rodrigo Gurgel-Gonçalves
    View author publications

    You can also search for this author in PubMed Google Scholar

  21. Fernando Abad-Franch
    View author publications

    You can also search for this author in PubMed Google Scholar

Contributions

FA-F, RG-G, and SLBL conceived the study. RG-G and SLBL raised funds, and RG-G administered the project. FA-F, RG-G, MTO, WMR, LAB and SLBL contributed to the design of trial methods. KKS, HSV, TOA, JPAC, MTO, TTCM-S, GDG, ENV, RVT, VGCB, MSPR, LPFM, GOM, BLC, IMM, NN and RG-G performed research. RG-G, NN and WMR supervised students involved in field and laboratory research. FA-F, KKSG, HSV and RG-G curated the dataset and analyzed the data. FA-F and RG-G drafted the first version of the manuscript; all authors contributed to the interpretation of results and commented on manuscript drafts. All authors read and approved the final manuscript.

Corresponding authors

Correspondence to Rodrigo Gurgel-Gonçalves or Fernando Abad-Franch.

Ethics declarations

Ethics approval and consent to participate

RG-G holds a permanent license for mosquito collection (IBAMA/Sisbio 33156-3, Ministério do Meio Ambiente, Brazil); our study did not involve endangered or protected species. Prior to trial activities, we obtained written informed consent from all homeowners participating in either mosquito monitoring or PPF dissemination-station deployment.

Consent for publication

Not applicable.

Competing interests

The authors declare that they have no competing interests.

Additional information

Publisher's Note

Springer Nature remains neutral with regard to jurisdictional claims in published maps and institutional affiliations.

Supplementary information

Additional file 1: Table S1.

Weather covariate values. Table S2. Monthly adult-mosquito aspiration results. Table S3. Alternative models for the all-mosquito aspiration dataset. Table S4. ‘Full’ all-mosquito aspiration model with ULV. Table S5. Alternative models for the Aedes aegypti aspiration dataset. Table S6. ‘Full’ Aedes aspiration model with ULV. Table S7. Alternative models for the Culex quinquefasciatus aspiration dataset. Table S8. ‘Full’ Culex aspiration model with ULV. Table S9. Monthly Aedes egg-trap results. Table S10. Alternative models for the Aedes egg-trap dataset.

Additional file 2: Dataset S1.

Adult-mosquito aspiration data are provided in the sheet named ‘Aspiration’, and Aedes egg-trap data in the sheet named ‘Aedes egg traps’. The file includes a ‘Readme’ sheet with the meaning of each variable.

Additional file 3: Table S11.

Top-ranking (smallest-BIC) model for female adult-mosquito aspiration data analyzed separately. Table S12. Top-ranking (smallest-BIC) model for male adult-mosquito aspiration data analyzed separately.

Rights and permissions

Open Access This article is licensed under a Creative Commons Attribution 4.0 International License, which permits use, sharing, adaptation, distribution and reproduction in any medium or format, as long as you give appropriate credit to the original author(s) and the source, provide a link to the Creative Commons licence, and indicate if changes were made. The images or other third party material in this article are included in the article's Creative Commons licence, unless indicated otherwise in a credit line to the material. If material is not included in the article's Creative Commons licence and your intended use is not permitted by statutory regulation or exceeds the permitted use, you will need to obtain permission directly from the copyright holder. To view a copy of this licence, visit http://creativecommons.org/licenses/by/4.0/. The Creative Commons Public Domain Dedication waiver (http://creativecommons.org/publicdomain/zero/1.0/) applies to the data made available in this article, unless otherwise stated in a credit line to the data.

Reprints and permissions

About this article

Check for updates. Verify currency and authenticity via CrossMark

Cite this article

Garcia, K.K.S., Versiani, H.S., Araújo, T.O. et al. Measuring mosquito control: adult-mosquito catches vs egg-trap data as endpoints of a cluster-randomized controlled trial of mosquito-disseminated pyriproxyfen. Parasites Vectors 13, 352 (2020). https://doi.org/10.1186/s13071-020-04221-z

Download citation

  • Received:

  • Accepted:

  • Published:

  • DOI: https://doi.org/10.1186/s13071-020-04221-z

Keywords

Parasites & Vectors

ISSN: 1756-3305

Contact us