Efficacy of Aedes aegypti control by indoor Ultra Low Volume (ULV) insecticide spraying in Iquitos, Peru

Abstract

Background: Aedes aegypti is a primary vector of dengue, chikungunya, Zika, and urban yellow fever viruses. Indoor, ultra low volume (ULV) space spraying with pyrethroid insecticides is the main approach used for Ae. aegypti emergency control in many countries. Given the widespread use of this method, the lack of large-scale experiments or detailed evaluations of municipal spray programs is problematic.

Methodology/principal findings: Two experimental evaluations of non-residual, indoor ULV pyrethroid spraying were conducted in Iquitos, Peru. In each, a central sprayed sector was surrounded by an unsprayed buffer sector. In 2013, spray and buffer sectors included 398 and 765 houses, respectively. Spraying reduced the mean number of adults captured per house by ~83 percent relative to the pre-spray baseline survey. In the 2014 experiment, sprayed and buffer sectors included 1,117 and 1,049 houses, respectively. Here, the sprayed sector's number of adults per house was reduced ~64 percent relative to baseline. Parity surveys in the sprayed sector during the 2014 spray period indicated an increase in the proportion of very young females. We also evaluated impacts of a 2014 citywide spray program by the local Ministry of Health, which reduced adult populations by ~60 percent. In all cases, adult densities returned to near-baseline levels within one month.

Conclusions/significance: Our results demonstrate that densities of adult Ae. aegypti can be reduced by experimental and municipal spraying programs. The finding that adult densities return to approximately pre-spray densities in less than a month is similar to results from previous, smaller scale experiments. Our results demonstrate that ULV spraying is best viewed as having a short-term entomological effect. The epidemiological impact of ULV spraying will need evaluation in future trials that measure capacity of insecticide spraying to reduce human infection or disease.

Fig 1. Map of experiment areas.

Detail for (A) S-2013 and (B) L-2014 experimental areas, showing individual houses. Color shows sector. (C) City of Iquitos. Black box highlights experimental areas. (D) Regional map. Black circle highlights Iquitos. See also S6 Fig.

Fig 2. Experiment timeline.

Each box shows one circuit. With one exception (L-2014 C2), each house was visited (and possibly surveyed) at least once per circuit. Except where noted, each circuit consisted of one or more spatially systematic subcircuits. Each subcircuit lasted approximately one calendar week. See S7 Fig. for survey maps.

Fig 3. Sampling and spraying.

(A) Number of houses per week sprayed and/or surveyed. Circuits are labeled (e.g., C1), with date ranges shown by horizontal bars. Containers were not surveyed during spray periods. The first two emergency (citywide) spray events (red +) occurred within the same calendar week, but are plotted separately here. (B) Spray coverage by spray cycle. Percent houses sprayed is shown in text. Top row: emergency (citywide) spraying. Bottom row: experimental spraying.

Fig 4. Model estimates of Ae. aegypti adults per house (AA/HSE, top row) and proportion infested houses (PrIH = AHI).

A separate generalized linear model (GLM) was constructed for each experiment (column) and for each measured response (row). (A) AA/HSE, negative binomial GLM (NB-GLM). (B) PrIH, logistic GLM (L-GLM). Models describe response of measure (row) to time period (X-axis) and treatment sector (color & shape). Shading indicates spray events: experimental spraying (light) and citywide spraying (dark). Vertical bars show 95% CI; non-overlapping CI indicate highly significant difference. Letters (s, t) indicate significant differences between pairwise contrasts: s, between sector (within time, S2 Table); t, between time (within spray sector, relative to baseline C1, S3 Table). See also S6A and S6B Table, S7A and S7B Table, and S5 Fig.

Fig 5. Contrast ratios of AA/HSE, based on NB-GLM.

Results from NB-GLM models (Fig 4A). Top row (between-sector): Spray/Buffer. Bottom row (between-time, within spray sector): contrast relative to baseline (C1). Vertical bars show 95% CI. Horizontal dashed line indicates null hypothesis of equality (H₀: ratio = 1). Asterisks (*) indicate significant difference between pairwise contrasts (reject H₀).

Fig 6. Detailed time series of AA/HSE response to ULV spraying, aggregated by week.

X-axis shows week start date. Color and symbol shape shows sector (orange triangle: spray sector). Point size shows number of surveyed houses. Vertical dashed lines show approximate dates of experimental spraying (spray sector only). Vertical colored bars show bootstrap 95% CI (1e+04 draws per circuit).