Browse the corpus

Randomized Trial of Targeted Indoor Spraying to Prevent Aedes-Borne Diseases. BACKGROUND: Targeted indoor residual spraying focuses insecticide applications on common resting surfaces of Aedes aegypti mosquitoes (an arboviral disease vector) in houses, such as exposed lower sections of walls and under furniture. METHODS: We conducted a two-group, parallel, unblinded, cluster-randomized trial in Merida, Mexico, to quantify the efficacy of targeted indoor residual spraying for preventing aedes-borne diseases (chikungunya, dengue, or Zika). Children 2 to 15 years of age were enrolled from households in 50 clusters of five-by-five city blocks. Households in 25 clusters received an annual application of targeted indoor residual spraying (intervention) before each season of aedes-borne disease (July through December). All clusters received routine Ministry of Health vector control. The primary end point was laboratory-confirmed, symptomatic aedes-borne disease. Community effects were assessed with the use of geolocated national surveillance data. RESULTS: A total of 4461 children were monitored for up to three seasons (2021, 2022, and 2023). The indoor density of A. aegypti mosquitoes was 59% (95% confidence interval [CI], 51 to 65) lower with the intervention than with control. A total of 422 cases of aedes-borne disease were confirmed, primarily dengue in 2023. In the per-protocol analysis of cluster centers, 91 cases occurred among 1038 participants in the intervention group and 89 cases among 1037 participants in the control group (efficacy, -12.8%; 95% CI, -60.7 to 23.0). In an intention-to-treat analysis of entire clusters, 198 cases occurred among 2239 participants in the intervention group and 199 cases among 2222 participants in the control group (efficacy, 3.9%; 95% CI, -28.1 to 26.7). Adjustment of analyses for mobility or demographic characteristics did not change results. On the basis of 150 cases in the intervention clusters and 202 in the control clusters that were geolocated, the estimated community effect of the intervention was 24.0% (95% CI, 6.0 to 38.6). Two cases of multisymptom adverse events (e.g., nausea, watery eyes, diarrhea, and vomiting) were associated with the intervention. CONCLUSIONS: Despite lower entomologic indexes with targeted indoor residual spraying than with routine vector control, the cumulative incidence of aedes-borne diseases was not significantly lower with targeted indoor residual spraying. (Funded by the National Institutes of Health and the Innovative Vector Control Consortium; ClinicalTrials.gov number, NCT04343521.).

The TIRS trial was a two-arm, parallel, unblinded, cluster randomized controlled trial18 conducted in an ABD hot-spot of Mérida, Yucatan State, Mexico.19 With 50 clusters of 5×5 city blocks each, 25 TIRS clusters received annual household TIRS application before the start of each ABD season in July plus routine vector control carried out by the Yucatan Ministry of Health (MOH), while 25 control clusters received routine vector control only (Figure 1).19 MOH vector control actions were primarily reactive to an ABD outbreak, including truck-mounted ultra-low volume (ULV) spraying, larval source management, indoor space spraying, and outdoor insecticide spraying. The TIRS trial did not interfere with the deployment of these actions. The trial’s primary aim was to evaluate the efficacy of TIRS by quantifying the hazard rate of ABD within a cohort of children over consecutive ABD seasons. To minimize mosquito contamination from untreated areas nearby, the primary analysis followed a “fried egg design”20 where the entire cluster was treated, but outcome measurement focused on children residing in the center 3×3 blocks.18 The trial protocol was approved by the institutional review board at all collaborating institutions.18 Informed consent was obtained from all participants or their guardians, as detailed below. All authors vouch for the accuracy and completeness of the data and for the fidelity of the trial to the protocol.

The TIRS trial was a two-arm, parallel, unblinded, cluster randomized controlled trial18 conducted in an ABD hot-spot of Mérida, Yucatan State, Mexico.19 With 50 clusters of 5×5 city blocks each, 25 TIRS clusters received annual household TIRS application before the start of each ABD season in July plus routine vector control carried out by the Yucatan Ministry of Health (MOH), while 25 control clusters received routine vector control only (Figure 1).19 MOH vector control actions were primarily reactive to an ABD outbreak, including truck-mounted ultra-low volume (ULV) spraying, larval source management, indoor space spraying, and outdoor insecticide spraying. The TIRS trial did not interfere with the deployment of these actions. The trial’s primary aim was to evaluate the efficacy of TIRS by quantifying the hazard rate of ABD within a cohort of children over consecutive ABD seasons. To minimize mosquito contamination from untreated areas nearby, the primary analysis followed a “fried egg design”20 where the entire cluster was treated, but outcome measurement focused on children residing in the center 3×3 blocks.18 The trial protocol was approved by the institutional review board at all collaborating institutions.18 Informed consent was obtained from all participants or their guardians, as detailed below. All authors vouch for the accuracy and completeness of the data and for the fidelity of the trial to the protocol. Covariate-constrained randomization was used to identify two sets of 25 clusters with acceptable balance across selected census-tract variables21. As described,22 study teams visited cluster households to obtain informed consent for household participation in entomological surveys and receipt of TIRS. For children aged 2–15 years, consent and assent were obtained to participate in the prospective monitoring of ABD through yearly blood-draws for quantifying seroconversions and yearly mobility surveys (see Supplement). A fair coin flip was performed to determine cluster allocation to the TIRS or control arm. Details of the study design are available in the trial protocol, posted as a supplementary file with this article at nejm.org.

rough yearly blood-draws for quantifying seroconversions and yearly mobility surveys (see Supplement). A fair coin flip was performed to determine cluster allocation to the TIRS or control arm. Details of the study design are available in the trial protocol, posted as a supplementary file with this article at nejm.org. Spraying was conducted between May and early July each year (2021–2023) to achieve >60% of all residential premises treated per TIRS cluster (Figure S1).16 The organophosphate insecticide pirimiphos-methyl (Actellic 300CS, Syngenta) was selected after initial studies showed high Ae. aegypti susceptibility23 and an estimated residual effect of up to 7 months.16 During the trial, Ae. aegypti susceptibility to the insecticide was monitored using standard CDC bioassays.23 TIRS application followed WHO and Pan American Health Organization (PAHO) best practices.24 Any adverse health effects associated with TIRS were solicited and recorded. The primary endpoint was laboratory-confirmed, symptomatic, ABD in cohort children. Weekly contacts (rotating home visits, phone calls, and mass SMS) and a toll-free hotline were used to detect symptoms.18 The clinical team activated an Aedes-borne virus (ABV) testing protocol for symptoms consistent with any ABD (Figure S2). To characterize potential locations of ABD exposure, a retrospective mobility survey25 was administered.

me visits, phone calls, and mass SMS) and a toll-free hotline were used to detect symptoms.18 The clinical team activated an Aedes-borne virus (ABV) testing protocol for symptoms consistent with any ABD (Figure S2). To characterize potential locations of ABD exposure, a retrospective mobility survey25 was administered. To measure community impact, a secondary endpoint utilized a national surveillance database of laboratory-confirmed ABD cases geolocated within study clusters, regardless of age (see Supplement),4,26 To measure entomological impact (secondary endpoint), teams conducted monthly Prokopack27 mosquito collections post-TIRS in a subsample of 750 houses per arm and measured Ae. aegypti entomological indices, including density (mean number of females per house). Entomological endpoints and additional secondary endpoints, which will be covered in future reports, are summarized in Table S1.

d monthly Prokopack27 mosquito collections post-TIRS in a subsample of 750 houses per arm and measured Ae. aegypti entomological indices, including density (mean number of females per house). Entomological endpoints and additional secondary endpoints, which will be covered in future reports, are summarized in Table S1. Assuming 4% ABD incidence, an ICC of 0.035, and 20% loss to follow-up, the trial required 92 age-eligible children enrolled per cluster for an overall sample size of 50 clusters and 4600 children to have 80% power to detect a 70% reduction in ABD with a two-sided 0.05 test.18 The primary efficacy analysis measured time-to-first ABD in children in the 3×3 cluster center in households receiving the intervention per protocol, restricting to the active surveillance period of July 1-December 31 of each year. A secondary analysis measured efficacy in the entire cluster including all randomized participants per an intention-to-treat principle. For each, efficacy was estimated as one minus the unadjusted Cox model hazard ratio with a permutation-based confidence interval over the constrained randomization space. Kaplan-Meier cumulative incidence estimates with robust variance after each season are reported. A pre-specified secondary analysis included Cox model-adjustment for baseline covariates of age and the presence of protective household window screens.28

-based confidence interval over the constrained randomization space. Kaplan-Meier cumulative incidence estimates with robust variance after each season are reported. A pre-specified secondary analysis included Cox model-adjustment for baseline covariates of age and the presence of protective household window screens.28 Community impact was estimated as one minus the incidence rate ratio for TIRS versus control from a Poisson regression, using the cumulative number of surveillance cases geolocated to each cluster as cluster-level data. In the absence of a precise denominator per cluster, this approach treats the 5×5 city block clusters as having similar population sizes. Entomological efficacy was quantified as one minus the incidence rate ratio for TIRS versus control from a negative binomial model, with the outcome of Ae. aegypti density and random effect for trial cluster. The study data were maintained in REDCap database and administered by a database manager.18 All statistical analyses were conducted with R software, version 4.4.1.29

Covariate-constrained randomization was used to identify two sets of 25 clusters with acceptable balance across selected census-tract variables21. As described,22 study teams visited cluster households to obtain informed consent for household participation in entomological surveys and receipt of TIRS. For children aged 2–15 years, consent and assent were obtained to participate in the prospective monitoring of ABD through yearly blood-draws for quantifying seroconversions and yearly mobility surveys (see Supplement). A fair coin flip was performed to determine cluster allocation to the TIRS or control arm. Details of the study design are available in the trial protocol, posted as a supplementary file with this article at nejm.org.

Spraying was conducted between May and early July each year (2021–2023) to achieve >60% of all residential premises treated per TIRS cluster (Figure S1).16 The organophosphate insecticide pirimiphos-methyl (Actellic 300CS, Syngenta) was selected after initial studies showed high Ae. aegypti susceptibility23 and an estimated residual effect of up to 7 months.16 During the trial, Ae. aegypti susceptibility to the insecticide was monitored using standard CDC bioassays.23 TIRS application followed WHO and Pan American Health Organization (PAHO) best practices.24 Any adverse health effects associated with TIRS were solicited and recorded.

The primary endpoint was laboratory-confirmed, symptomatic, ABD in cohort children. Weekly contacts (rotating home visits, phone calls, and mass SMS) and a toll-free hotline were used to detect symptoms.18 The clinical team activated an Aedes-borne virus (ABV) testing protocol for symptoms consistent with any ABD (Figure S2). To characterize potential locations of ABD exposure, a retrospective mobility survey25 was administered. To measure community impact, a secondary endpoint utilized a national surveillance database of laboratory-confirmed ABD cases geolocated within study clusters, regardless of age (see Supplement),4,26 To measure entomological impact (secondary endpoint), teams conducted monthly Prokopack27 mosquito collections post-TIRS in a subsample of 750 houses per arm and measured Ae. aegypti entomological indices, including density (mean number of females per house). Entomological endpoints and additional secondary endpoints, which will be covered in future reports, are summarized in Table S1.

Assuming 4% ABD incidence, an ICC of 0.035, and 20% loss to follow-up, the trial required 92 age-eligible children enrolled per cluster for an overall sample size of 50 clusters and 4600 children to have 80% power to detect a 70% reduction in ABD with a two-sided 0.05 test.18 The primary efficacy analysis measured time-to-first ABD in children in the 3×3 cluster center in households receiving the intervention per protocol, restricting to the active surveillance period of July 1-December 31 of each year. A secondary analysis measured efficacy in the entire cluster including all randomized participants per an intention-to-treat principle. For each, efficacy was estimated as one minus the unadjusted Cox model hazard ratio with a permutation-based confidence interval over the constrained randomization space. Kaplan-Meier cumulative incidence estimates with robust variance after each season are reported. A pre-specified secondary analysis included Cox model-adjustment for baseline covariates of age and the presence of protective household window screens.28

Enrollment began on October 1, 2020, with 4792 children enrolled.22 Movement out of the study area was common during this period of heightened COVID-19 restrictions,22 leading to early losses that were mitigated by further enrollment prior to the start of active surveillance (Figure 2). At the start of Year 1 surveillance on July 1, 2021, there were 2124 children (1087 in TIRS, 1037 in control) in the cluster center cohort, and there were 4461 children (2239 in TIRS, 2222 in control) in the entire cluster cohort. The trial was conducted over three ABD seasons, through December 31, 2023, during which time 908 children (20%) were censored from the entire cohort, primarily due to moving out of the study area (Figure S3). The baseline characteristics of participating households and children were balanced across the arms.22 The representativeness of the trial population is discussed in Table S2. Sex and age at enrollment for the children contributing to key efficacy analyses are summarized in Table S3. Median age was 8.7 years in both trial arms, with 51.5% male. Baseline serum samples from a subset of 1,399 children tested by commercially available ELISAs found that 45.1% were seropositive for dengue (DENV) or Zika (ZIKV) viruses and 24.0% were seropositive for chikungunya virus (CHIKV), increasing sharply with age and balanced across the study arms.22

l arms, with 51.5% male. Baseline serum samples from a subset of 1,399 children tested by commercially available ELISAs found that 45.1% were seropositive for dengue (DENV) or Zika (ZIKV) viruses and 24.0% were seropositive for chikungunya virus (CHIKV), increasing sharply with age and balanced across the study arms.22 A total of 27,833 TIRS applications were conducted during the trial, reaching 90–96% coverage of TIRS houses with enrolled children and >60% of all premises during 2021–2023 (Table S4, Table S5). Reasons for refusal are summarized in the Supplement. Spraying took an average of 4.9 minutes and consumed an average of 0.47 L of insecticide per house. Pirimiphos-methyl susceptibility monitoring showed no evidence of insecticide resistance throughout the trial. Two residents reported adverse events possibly related to the TIRS intervention, including colic/abdominal pain, nausea, watery eyes, and runny nose (Table S6). The participants were referred to the nearest hospital and their symptoms resolved within a day with no sequelae.

insecticide resistance throughout the trial. Two residents reported adverse events possibly related to the TIRS intervention, including colic/abdominal pain, nausea, watery eyes, and runny nose (Table S6). The participants were referred to the nearest hospital and their symptoms resolved within a day with no sequelae. During the ABD seasons, mean Ae. aegypti density was maintained at low levels in TIRS households, comparable to low ABD periods (Figure 3), in contrast to a rapid spike in the control arm one month after baseline. From monthly samples of the 750 houses selected from each arm, total Ae. aegypti collected were 8,275 in TIRS and 18,761 in control homes (Table S7). There was a 59% (95% CI: 51%, 65%) overall reduction in Ae. aegypti density in the TIRS arm with the largest reduction during the peak Ae. aegypti months of July to October (Figure 3, Table S8). Mean mortality of Ae. aegypti females using WHO cone bioassays shows high insecticidal activity during the 6 months post-TIRS (Figure S4). During the trial, MOH truck-mounted ULV spraying with the insecticide malathion occurred at a rate of 3.5 applications per cluster (3.1 TIRS and 3.8 control) in 2021, increasing to 12.6 (12.0 TIRS and 13.2 Control) in 2022 and 18.9 (18.8 TIRS and 18.9 Control) in 2023 in response to large dengue outbreaks.

During the ABD seasons, mean Ae. aegypti density was maintained at low levels in TIRS households, comparable to low ABD periods (Figure 3), in contrast to a rapid spike in the control arm one month after baseline. From monthly samples of the 750 houses selected from each arm, total Ae. aegypti collected were 8,275 in TIRS and 18,761 in control homes (Table S7). There was a 59% (95% CI: 51%, 65%) overall reduction in Ae. aegypti density in the TIRS arm with the largest reduction during the peak Ae. aegypti months of July to October (Figure 3, Table S8). Mean mortality of Ae. aegypti females using WHO cone bioassays shows high insecticidal activity during the 6 months post-TIRS (Figure S4). During the trial, MOH truck-mounted ULV spraying with the insecticide malathion occurred at a rate of 3.5 applications per cluster (3.1 TIRS and 3.8 control) in 2021, increasing to 12.6 (12.0 TIRS and 13.2 Control) in 2022 and 18.9 (18.8 TIRS and 18.9 Control) in 2023 in response to large dengue outbreaks. Between July 1, 2021, and December 31, 2023, the clinical team received 1902 symptom reports, and 825 (43%) met the criteria for ABV testing (Figure S5, Table S9). A total of 422 ABD cases were confirmed in the cohort (213 in TIRS, 209 in control), with 403 testing ABV negative (187 in TIRS, 216 in control) (Table S10). Of the 422 confirmed ABV positive, there were 297 DENV positive (serotypes DENV2 or DENV3), 94 ZIKV positive, 1 CHIKV positive, 20 DENV/ZIKV positive, and 10 DENV/CHIKV positive (Table S11), and 59% were confirmed by acute and convalescent testing for IgM and 41% by quantitative polymerase chain reaction (qPCR) (Table S12). In 2021, amid heightened COVID-19 restrictions, transmission was minimal, with 23 confirmed ABD cases (Table S12). Transmission was more typical in 2022, with 89 confirmed ABD cases. In 2023, a large DENV3 outbreak occurred, with 310 confirmed ABD cases. The median age of ABD cases was 10.0 years (Figure S6) and 49% were in males. There were no cases of severe ABD based on WHO definitions.

, with 23 confirmed ABD cases (Table S12). Transmission was more typical in 2022, with 89 confirmed ABD cases. In 2023, a large DENV3 outbreak occurred, with 310 confirmed ABD cases. The median age of ABD cases was 10.0 years (Figure S6) and 49% were in males. There were no cases of severe ABD based on WHO definitions. For the primary efficacy analysis of time-to-first confirmed ABD in the cluster center analyzed per-protocol, there were 91 and 89 confirmed cases in the TIRS and control arms, respectively (Table 1). Cumulative incidence was 11.5% (95% CI: 8.4%, 14.5%) in the TIRS arm and 10.1% (95% CI: 7.8%, 12.3%) in the control arm (Figure 4). The estimated TIRS efficacy was −12.8% (95% CI: −60.7%, 23.0%). Analyzing the cluster center by intention-to-treat returned a similar efficacy estimate of - 8.3% (95% CI: −57.9%, 26.8%). For the secondary efficacy analysis of time-to-first confirmed ABD in the entire cluster analyzed per intention-to-treat, there were 198 and 199 confirmed cases in the TIRS and control arms, respectively (Table 1, Table S13). Estimated cumulative incidence was 10.1% (95% CI: 8.4%, 11.8%) in the TIRS arm and 10.5% (95% CI: 8.6%, 12.3%) in the control arm (Figure 4). The estimated TIRS efficacy was 3.9% (95% CI: −28.1%, 26.7%). Pre-specified adjustment for age and the presence of screens had no impact (estimated efficacy 4.0% (95% CI: −24.5%, 26.0%). Patterns varied by year, with estimated TIRS efficacy at the end of the first two ABD seasons of 24.7% (95% CI: −10.2%, 48.5%), but with a reversed pattern in the third (outbreak) season (Table S14).

specified adjustment for age and the presence of screens had no impact (estimated efficacy 4.0% (95% CI: −24.5%, 26.0%). Patterns varied by year, with estimated TIRS efficacy at the end of the first two ABD seasons of 24.7% (95% CI: −10.2%, 48.5%), but with a reversed pattern in the third (outbreak) season (Table S14). Human mobility varied substantially over the three ABD seasons (Figure S7, Figure S8). Percent time spent at home was similar across TIRS treatment and control clusters and between symptomatic participants with and without a confirmed ABD illness (Figure S9). The mobility-adjusted estimated efficacy was similar to unadjusted efficacy (see Supplement). Between July 1, 2021, and December 31, 2023, the national surveillance system captured 4882 confirmed dengue cases (1 in 2021, 268 in 2022, 4613 in 2023) and no chikungunya or Zika cases in Mérida (Figure S10). The dengue cases had a median age of 20 years (range: <1 to 92 years). Restricting to dengue cases geolocated within the study clusters during active surveillance, the following cases were detected: 2021: 0 in TIRS, 0 in control; 2022: 5 in TIRS, 19 in control; 2023: 144 in TIRS, 183 in control. Across years, the estimated community impact of TIRS was 24.0% (95% CI: 6.0%, 38.6%) (Table 1).

A total of 27,833 TIRS applications were conducted during the trial, reaching 90–96% coverage of TIRS houses with enrolled children and >60% of all premises during 2021–2023 (Table S4, Table S5). Reasons for refusal are summarized in the Supplement. Spraying took an average of 4.9 minutes and consumed an average of 0.47 L of insecticide per house. Pirimiphos-methyl susceptibility monitoring showed no evidence of insecticide resistance throughout the trial. Two residents reported adverse events possibly related to the TIRS intervention, including colic/abdominal pain, nausea, watery eyes, and runny nose (Table S6). The participants were referred to the nearest hospital and their symptoms resolved within a day with no sequelae. During the ABD seasons, mean Ae. aegypti density was maintained at low levels in TIRS households, comparable to low ABD periods (Figure 3), in contrast to a rapid spike in the control arm one month after baseline. From monthly samples of the 750 houses selected from each arm, total Ae. aegypti collected were 8,275 in TIRS and 18,761 in control homes (Table S7). There was a 59% (95% CI: 51%, 65%) overall reduction in Ae. aegypti density in the TIRS arm with the largest reduction during the peak Ae. aegypti months of July to October (Figure 3, Table S8). Mean mortality of Ae. aegypti females using WHO cone bioassays shows high insecticidal activity during the 6 months post-TIRS (Figure S4).

9% (95% CI: 51%, 65%) overall reduction in Ae. aegypti density in the TIRS arm with the largest reduction during the peak Ae. aegypti months of July to October (Figure 3, Table S8). Mean mortality of Ae. aegypti females using WHO cone bioassays shows high insecticidal activity during the 6 months post-TIRS (Figure S4). During the trial, MOH truck-mounted ULV spraying with the insecticide malathion occurred at a rate of 3.5 applications per cluster (3.1 TIRS and 3.8 control) in 2021, increasing to 12.6 (12.0 TIRS and 13.2 Control) in 2022 and 18.9 (18.8 TIRS and 18.9 Control) in 2023 in response to large dengue outbreaks.

Between July 1, 2021, and December 31, 2023, the clinical team received 1902 symptom reports, and 825 (43%) met the criteria for ABV testing (Figure S5, Table S9). A total of 422 ABD cases were confirmed in the cohort (213 in TIRS, 209 in control), with 403 testing ABV negative (187 in TIRS, 216 in control) (Table S10). Of the 422 confirmed ABV positive, there were 297 DENV positive (serotypes DENV2 or DENV3), 94 ZIKV positive, 1 CHIKV positive, 20 DENV/ZIKV positive, and 10 DENV/CHIKV positive (Table S11), and 59% were confirmed by acute and convalescent testing for IgM and 41% by quantitative polymerase chain reaction (qPCR) (Table S12). In 2021, amid heightened COVID-19 restrictions, transmission was minimal, with 23 confirmed ABD cases (Table S12). Transmission was more typical in 2022, with 89 confirmed ABD cases. In 2023, a large DENV3 outbreak occurred, with 310 confirmed ABD cases. The median age of ABD cases was 10.0 years (Figure S6) and 49% were in males. There were no cases of severe ABD based on WHO definitions.

For the primary efficacy analysis of time-to-first confirmed ABD in the cluster center analyzed per-protocol, there were 91 and 89 confirmed cases in the TIRS and control arms, respectively (Table 1). Cumulative incidence was 11.5% (95% CI: 8.4%, 14.5%) in the TIRS arm and 10.1% (95% CI: 7.8%, 12.3%) in the control arm (Figure 4). The estimated TIRS efficacy was −12.8% (95% CI: −60.7%, 23.0%). Analyzing the cluster center by intention-to-treat returned a similar efficacy estimate of - 8.3% (95% CI: −57.9%, 26.8%). For the secondary efficacy analysis of time-to-first confirmed ABD in the entire cluster analyzed per intention-to-treat, there were 198 and 199 confirmed cases in the TIRS and control arms, respectively (Table 1, Table S13). Estimated cumulative incidence was 10.1% (95% CI: 8.4%, 11.8%) in the TIRS arm and 10.5% (95% CI: 8.6%, 12.3%) in the control arm (Figure 4). The estimated TIRS efficacy was 3.9% (95% CI: −28.1%, 26.7%). Pre-specified adjustment for age and the presence of screens had no impact (estimated efficacy 4.0% (95% CI: −24.5%, 26.0%). Patterns varied by year, with estimated TIRS efficacy at the end of the first two ABD seasons of 24.7% (95% CI: −10.2%, 48.5%), but with a reversed pattern in the third (outbreak) season (Table S14).

specified adjustment for age and the presence of screens had no impact (estimated efficacy 4.0% (95% CI: −24.5%, 26.0%). Patterns varied by year, with estimated TIRS efficacy at the end of the first two ABD seasons of 24.7% (95% CI: −10.2%, 48.5%), but with a reversed pattern in the third (outbreak) season (Table S14). Human mobility varied substantially over the three ABD seasons (Figure S7, Figure S8). Percent time spent at home was similar across TIRS treatment and control clusters and between symptomatic participants with and without a confirmed ABD illness (Figure S9). The mobility-adjusted estimated efficacy was similar to unadjusted efficacy (see Supplement).

Between July 1, 2021, and December 31, 2023, the national surveillance system captured 4882 confirmed dengue cases (1 in 2021, 268 in 2022, 4613 in 2023) and no chikungunya or Zika cases in Mérida (Figure S10). The dengue cases had a median age of 20 years (range: <1 to 92 years). Restricting to dengue cases geolocated within the study clusters during active surveillance, the following cases were detected: 2021: 0 in TIRS, 0 in control; 2022: 5 in TIRS, 19 in control; 2023: 144 in TIRS, 183 in control. Across years, the estimated community impact of TIRS was 24.0% (95% CI: 6.0%, 38.6%) (Table 1).

Annual treatment of households with TIRS prior to the ABV transmission season reduced household Ae. aegypti density but did not significantly reduce ABD hazard rate in children. While the analyses based on cohort surveillance data were not statistically significant, a reduction in dengue incidence was observed in a pre-specified randomized analysis of national surveillance data covering a wider age range than the study cohort. The trial results highlight the need to better understand how entomological efficacy translates into epidemiological impact,30,31 particularly what outcomes and thresholds reliably indicate decreased ABV transmission risk. We observed measurable reductions in household mosquito density and mosquito age-structure.15,16 Yet the trial did not replicate the high estimated effectiveness from an observational analysis in Australia.12 In that context, TIRS was deployed reactively to a dengue outbreak via contact tracing to households. In our trial, most cases came from a large 2023 DENV3 outbreak. It could be that even greater reductions in household mosquitoes at the household, neighborhood, or regional level may be required. Mobility analyses further reveal that more time was spent out of the home in 2023 than 2021 and 2022. Out-of-cluster ABD exposure may be a key contributing factor to the reduced efficacy observed in the cohort, although adjustment for mobility did not alter the efficacy estimate.

ood, or regional level may be required. Mobility analyses further reveal that more time was spent out of the home in 2023 than 2021 and 2022. Out-of-cluster ABD exposure may be a key contributing factor to the reduced efficacy observed in the cohort, although adjustment for mobility did not alter the efficacy estimate. A protective community impact was observed in national surveillance data. This is an encouraging signal, although the reason for differences between this and the primary study endpoint remain unclear. The national surveillance data captured older populations, who are more likely to experience their second or third DENV infection, which is more likely to be symptomatic. There may be different patterns of mobility by age, but the data to test this hypothesis were not generated in the trial that only tracked mobility of children. The high coverage of TIRS (62–75%) could have led to cluster-level reductions in ABD transmission beyond what we measured within the few houses with enrolled children.

y be different patterns of mobility by age, but the data to test this hypothesis were not generated in the trial that only tracked mobility of children. The high coverage of TIRS (62–75%) could have led to cluster-level reductions in ABD transmission beyond what we measured within the few houses with enrolled children. The trial quantified efficacy over three epidemiologically distinct seasons. ABV transmission was low during 2021, coinciding with COVID-19 lockdowns and low human mobility.32 Transmission was typical of non-outbreak years during 2022, and very high during 2023 due to a large DENV3 outbreak, which baseline data showed our cohort had low immunity against.22 Notably, TIRS had the highest impact during the first two ABD seasons. The effect was lowest during the 2023 outbreak, during which Mérida’s MOH deployed one of the largest vector control operations in history across the city, including treatment and control clusters.

data showed our cohort had low immunity against.22 Notably, TIRS had the highest impact during the first two ABD seasons. The effect was lowest during the 2023 outbreak, during which Mérida’s MOH deployed one of the largest vector control operations in history across the city, including treatment and control clusters. The WHO prioritizes evidence from randomized controlled trials with entomological and epidemiological outcomes to inform policy recommendations.32 Yet conducting such trials is challenging, even without an unprecedented pandemic complicating trial operations. The trial incorporated innovative design elements tailored for ABV transmission.25,33 As efficacy estimates were similar when comparing the center versus entire cluster, the fried egg structure seems to have exerted little impact in our setting. Future secondary analyses will explore household-level entomological indices and ABV infection in mosquitoes, providing insight on the possible relationship between entomological and epidemiological estimates. Future studies using a mathematical model simulating the trial should reveal why the results were as observed, and what effectiveness may be expected from more widespread use of TIRS.

ces and ABV infection in mosquitoes, providing insight on the possible relationship between entomological and epidemiological estimates. Future studies using a mathematical model simulating the trial should reveal why the results were as observed, and what effectiveness may be expected from more widespread use of TIRS. The combined trial results suggest that deploying one TIRS application preventively can have a measurable public health value when considering its entomological and community impacts. As additional ABV prevention tools become available (e.g., Wolbachia, spatial repellents, vaccines), studying their scalability and potential synergistic interactions when combined34 will provide unique opportunities to achieve the ambitious WHO target of reducing dengue case fatality to zero by 2030.35 Disclosure forms provided by the authors are available with the full text of this article at NEJM.org.