Adjusting vector surveillance for human behaviors reveals Anopheles funestus drove a resurgence in malaria despite IRS with clothianidin in Uganda

Categories: Disease & Virus

May 22, 2025

Characteristics of household cohorts

Household demographic data was only gathered from the 12 households that took part in the HBOs in Phase 2. At that time, households had a mean number of 4.0 inhabitants (range: 1 to 8), 56% were female and median age was 13 years (interquartile range: 7 to 35). These households had a mean of 1.2 bednets (range: 0 to 2) and 1.6 sleeping rooms (range: 1 to 4). These characteristics were comparable to those of the 80 households in the clinical cohort which are detailed elsewhere⁴.

Entomological and clinical indicators

The HLCs captured 1023 An. gambiae s.l. and 357 An. funestus sensu lato mosquitoes in Phase 1 and 2461 An. gambiae s.l. and 507 An. funestus s.l. in Phase 2. Of the 403 An. gambiae s.l. mosquitoes tested by PCR from the first half of Phase 2, 73.5% were An. arabiensis and 26.5% were An. gambiae sensu stricto (henceforth referred to as An. gambiae). There was no significant difference between the proportion of An. arabiensis collected indoors and outdoors (data not shown). All 126 An. funestus s.l. tested by PCR during Phase 2 were determined to be An. funestus sensu stricto (henceforth referred to as An. funestus). Figure 1 summarizes female Anopheles mosquitoes captured by species overlaid with clinical malaria incidence (as measured in the clinical cohort) and vector control interventions. Malaria incidence decreased from November 2020 through February 2021 coinciding with declining abundance of all three Anopheles species. During this time period, the HBR for An. funestus was 1.7 mosquitoes per person per night (20.5% of the total). From March to November 2021, following the second round of IRS with Fludora Fusion, abundances of all species increased slightly, though the An. funestus HBR was similar at 1.5 per person per night (29.2% of total). When collections resumed in November 2022, abundances remained high, but now the HBR from An. funestus had increased to 2.3 (43.0% of the total). Following IRS with Actellic in February to March 2023, coinciding with a dramatic reduction in incidence, the relative abundances of An. arabiensis and An. gambiae were much greater than An. funestus whose HBR had fallen to 0.2 per night (3.4% of the total) (Fig. 1).

Human and vector behaviors

During Phase 2, HBOs were performed on the same nights and in the same households as the HLCs from June 2023 until September 2023. The proportion of individuals observed to be outside decreased from 82% between 1800 and 1900 h to 50% between 2000 and 2100 h to nearly 0% by 2300 h (Fig. 2). By 2300 h, observed ITN use had reached its maximum at around 50%, with the other 50% of the population inside, presumably sleeping, without an ITN. Hourly all-species HBRs were higher outdoors than indoors, with a mean of 0.62 mosquitoes per person per hour (ppph) outdoors (95% CI: 0.57 to 0.68) compared to 0.27 ppph (95% CI: 0.24 to 0.30) indoors (Fig. 2).

Human exposure to mosquito bites Human-behavior adjusted HBRs, aHBRs, were calculated as an estimate of hourly exposure by accounting for observed temporal and spatial human presence (including ITN usage behaviors). The proportion of exposure to each vector species occurring outdoors increased over time, with a pronounced increase after IRS with Actellic (Fig. 3: Panel A). Exposure to An. gambiae, even during the high malaria incidence period, was markedly lower than the other two species. After Actellic was reintroduced exposure to An. funestus dropped dramatically and exposure to An. arabiensis and An. gambiae remained consistent, but shifted to earlier hours of the evening and outdoors. Compared to Phase 1, the proportion of exposure due to An. funestus was higher during the resurgence and then dropped precipitously after the re-introduction of Actellic (Fig. 3: Panel B).

Estimates of overall nightly exposure derived from the aHBRs were lower than raw HBRs, but An. funestus made up essentially the same proportions in both. In terms of exposure to all Anopheles species, there were no significant differences in aHBRs comparing the first to any other time period after accounting for human behaviors and adjusting for season, proximity to the border and clustering at the household level. (Table 1). When evaluating aHBRs by species, there was a marked decrease in the An. funestus aHBR after the switch to Actellic compared to the first time period with a RR 0.18 (95% CI: 0.05 to 0.69), but no significant increase during the period with high transmission (Table 2). The estimated aHBR from An. arabiensis was more than 40% lower during the malaria resurgence in the third time period compared to the first (RR: 0.59, 95% CI: 0.37 to 0.94). There were no significant differences in aHBRs from An. gambiae across any of the time periods (joint significance p-value = 0.486).

Table 1 Human biting and infective biting rates from all Anopheles species adjusted for host availability from observed human behaviors.

Table 2 Human biting and infective biting rates adjusted for host availability from observed human behaviors by species.

Vector infectivity rates

Sporozoite rates in mosquitoes from indoor CDC light traps in the clinical cohort were consistently higher in An. funestus than An. gambiae s.l. (Table 3). Sporozoite rates across all species were highest between November 2022 and February 2023, when An. funestus positivity rate was 3.2% (95% CI: 2.3–4.0%), An. arabiensis was 1.8% (1.4 to 2.2%) and An. gambiae was 2.2% (1.8–2.7%). When Actellic was reintroduced in most of the district in March 2023, sporozoite rates dropped in all species.

Table 3 Sporozoite rates by species from biweekly indoor CDC light traps in clinical cohort households.

Exposure to infective bites

We estimated nightly adjusted infective biting rates that considered both species-specific sporozoite rates and species-specific HBRs adjusted for human behaviors. This measure, not the same as entomological inoculation rate (EIR), incorporates both vector infectivity and human-behavior adjusted exposure to estimate mean individual exposure to infective bites. We found a nearly two times higher overall Anopheles infective biting rate during the resurgence (November 2022-February 2023) compared to the first time period (November 2020-February 2021) with a risk ratio (RR) of 1.92 (95% CI: 1.25 to 2.95). The proportion of infective biting occurring indoors was similar across the first three time periods (Fig. 4), including during the malaria resurgence when 63.6% (95% CI: 55.6–71.1%) of the estimated 0.08 (0.04 to 0.11) infective bites per person per night occurred indoors. The magnitude of infective biting dropped after the introduction of Actellic in March 2023 to 0.02 (95% CI: 0.01 to 0.03) bites per person per night, with 35.9% (95% CI: 28.9 to 43.0%) of these occurring indoors. Compared to the first time period, infective biting rates were more than halved after the switch back to Actellic with a RR of 0.48 (95% CI: 0.25 to 0.92) (Table 1). These observations track malaria incidence trends over those time periods (Fig. 1).

Compared to the first time period, exposure to infective bites from An. funestus was 4.3 (95% CI: 1.8 to 10.3) times higher during the malaria resurgence. These then decreased by 80% after Actellic was introduced and malaria incidence fell (RR: 0.2, 95% CI: 0.1 to 0.7) (Table 2). Adjusted infective biting due to An. arabiensis also declined markedly after Actellic was reintroduced compared to the first time period (RR: 0.3; 95% CI: 0.2 to 0. 6), despite aHBR not being statistically different (p = 0.330). Exposure to infective bites from An. gambiae was not significantly different across any of the time periods (p = 0.850). The proportion of infective biting due to An. funestus, however, showed a marked increase during the time of the malaria resurgence to 43.3% (95% CI: 34.9–51.8%). This was significantly higher than either the first (p = 0.011) or second time (p = 0.003) periods. Furthermore, after the introduction of Actellic, the proportion of overall infective biting due to An. funestus dropped to 10.2% (95% CI: 5.5–14.8%), which was the lowest of all time periods (p = 0.017). Of note, the proportions of exposure attributable to An. funestus across all three metrics, the raw HBR, aHBR and the infective biting rate, were essentially the same in each time period.

Causal analysis of IRS product on An. funestus adjusted landing rates

We performed a difference-in-difference analysis restricted to eight households that were all near the border with Busia District where IRS has never been implemented. This analysis took advantage of the fact that three of these households received a second round of the clothianidin-based Sumishield after February-March 2023, while the other five households received the organophosphate Actellic like most households in the rest of the Tororo District. This analysis showed that, after controlling for household-level characteristics, secular trends and unobserved time-invariant factors, there was no significant difference between households that received Sumishield or Actellic in nightly aHBRs when combining all Anopheles species (RR: 1.3; 95% CI: 0.5 to 3.4) (Table 4). However, households that received Sumishield compared to Actellic had 8.6 times higher An. funestus-specific aHBRs (95% CI: 2.0 to 37.0).

Table 4 Difference-in-difference analysis of households receiving actellic compared to households continuing to receive Sumishield in Feb-Mar 2023 IRS round.

Source link