Relating household entomological measures to individual malaria risk

Relating household entomological measures to individual malaria risk | Research Square window.SnipcartSettings = { analytics: { enabled: false } }; (function() { var accessVector = localStorage.getItem('access_vector') || ''; window.dataLayer = window.dataLayer || []; if (accessVector) { window.dataLayer.push({ user: { profile: { profileInfo: { snid: accessVector } } } }); } })(); (function(w,d,s,l,i){w[l]=w[l]||[];w[l].push({'gtm.start':new Date().getTime(),event:'gtm.js'});var f=d.getElementsByTagName(s)[0],j=d.createElement(s),dl=l!='dataLayer'?'&l='+l:'';j.async=true;j.src='https://www.googletagmanager.com/gtm.js?id='+i+dl;f.parentNode.insertBefore(j,f);})(window,document,'script','dataLayer','GTM-K279D39R'); Browse Preprints In Review Journals COVID-19 Preprints AJE Video Bytes Research Tools Research Promotion AJE Professional Editing AJE Rubriq About Preprint Platform In Review Editorial Policies Our Team Advisory Board Help Center Sign In Submit a Preprint Cite Share Download PDF Research Article Relating household entomological measures to individual malaria risk Max McClure, Emmanuel Arinaitwe, Moses R. Kamya, Philip J. Rosenthal, and 6 more This is a preprint; it has not been peer reviewed by a journal. https://doi.org/ 10.21203/rs.3.rs-5914493/v1 This work is licensed under a CC BY 4.0 License Status: Under Review Version 1 posted 12 You are reading this latest preprint version Abstract Background The gold standard measure of malaria exposure is the entomological inoculation rate (EIR), or the number of infectious bites an individual receives over a given period. Nevertheless, it is unclear whether household EIR reflects heterogeneity in individual infection risk. Methods To investigate this relationship, we used data collected from a cohort of 439 children aged 0.5-5 years in 239 households from 2011–2017 in three Ugandan districts: low-EIR Jinja, intermediate-EIR Kanungu and high-EIR Tororo. Participants underwent passive and quarterly active surveillance for clinical malaria, defined as fever with positive thick blood smear. Monthly vector densities and sporozoite rates in participating households were estimated using CDC light traps. We assessed the association between spatiotemporally smoothed household log 2 -transformed EIR and individual malaria incidence using Poisson generalized additive mixed effects models. Results Comparison across sites suggested an increasing relationship between average EIR and malaria incidence. Within-site relationships, however, varied by site, with a positive association in Kanungu (incidence rate ratio [IRR] 1.09, 95% credible interval 1.04–1.14) but none in Jinja (1.02, 0.774–1.26) or Tororo (1.02, 0.986–1.06). Conclusions These results show the relationship between measured EIR and malaria incidence may depend on local transmission dynamics and be strongest at intermediate EIR, while underscoring the challenges of using household-level measures of exposure. Figures Figure 1 Figure 2 Figure 3 Background Anopheline mosquitoes transmit malaria to humans, and exposure to Plasmodium -carrying mosquitoes corresponds to human malaria risk. WHO guidelines recommend surveillance of entomological proxies for transmission as a component of integrated vector management programs [1, 2]. However, current understanding of the quantitative relationship between mosquito exposure and human infection risk is surprisingly limited. The gold standard entomological measure of malaria exposure is the entomological inoculation rate (EIR), or the number of infectious bites received by an individual in a given time period [2]. EIR is typically approximated as the product of the malaria vector density (as determined by mosquito captures) and the proportion of tested mosquitoes positive for sporozoites [3]. EIR has a positive relationship to parasite prevalence [4] that has been validated with real-world data at community scales [5, 6], and an association with the incidence of blood stage infections down to the village or neighborhood level [7, 8], but it is unclear if it is able to capture heterogeneity in transmission driving incidence at smaller scales such as the household or the individual. Factors potentially obscuring such an association include measurement error and uncaptured spatiotemporal variability in entomological indices, behavioral heterogeneity between vector species, and behavioral and immunologic heterogeneity among human hosts. To better assess the association between EIR and individual malaria incidence, we analyzed data from completed longitudinal cohort studies that included passive and active clinical surveillance of participants and paired household-level entomological measurements at sites with varying transmission. We used flexible spatiotemporal models to smooth crude entomological measures and allow for nonlinear associations between exposure and human disease risk. These cohorts were well-suited to assess associations between mosquito exposure and clinical outcomes given their concurrent entomological and clinical data collection across a range of transmission intensities, a focus on children (with lower rates of acquired immunity compared to adults), and relatively stable transmission during the study period. Methods Study location The original Program for Resistance, Immunology, Surveillance and Modeling of Malaria (PRISM 1) cohort studies were conducted from 2011–2017 in three Ugandan subcounties representing a spectrum of malaria transmission settings: Walukuba, Jinja District, a peri-urban area near the northern shore of Lake Victoria with the lowest transmission; Kihihi, Kanungu District, a rural area in the southwest near the country’s border with the Democratic Republic of the Congo with relatively moderate transmission; and Nagongera, Tororo District, a rural area in the southeast near the country’s border with Kenya with the highest transmission. Plasmodium falciparum is the dominant malaria parasite throughout Uganda and was reported to account for over 98% of infections nationally at the time of the study [22]. During the data collection period, long-lasting insecticidal nets (LLIN) were distributed in Jinja in November 2013, Kanungu in June 2014, and Tororo in November 2013; LLINs were also provided to all study participants at the time of enrollment in the cohort studies. For this analysis, in Tororo only, we excluded any data collected after a participating household underwent indoor residual spraying (IRS), administered in the district since December 2014. IRS was not implemented in Jinja or Kanungu. Study design and data collection The study protocol is described in detail elsewhere [23]. Briefly, 100 households containing at least one resident 0.5–10 years of age and one adult resident were randomly selected from each subcounty. All children in each household between 0.5–10 years of age that met study eligibility criteria were enrolled. Household latitude and longitude were mapped using handheld GPS and projected to UTM zone 36 coordinates. Enrollment was dynamic over the course of the study – children from participating households joined the study as they became eligible and left as they aged out. Routine clinical visits were conducted every 3 months and included a standardized clinical assessment and collection of thick blood smears. Participants were encouraged to present to dedicated study clinics (one per site) for evaluation of any medical needs outside routine visits: if participants were febrile at the time of evaluation or reported fever within the past 24 hours, thick blood smears were obtained. Malaria was defined as a fever (tympanic temperature ≥ 38°C or reported fever within the past 24 hours) with a thick blood smear positive for malaria parasites [23]. Participants diagnosed with malaria were treated according to national treatment guidelines. Mosquito collections were conducted once a month using CDC light traps (CDC LT) set from 7pm to 7am 1 meter above the floor at the foot of the bed in one bedroom of each participating household. All mosquito species were identified morphologically. A random subset of mosquitoes from each capture (maximum 50 per collection) was stored on desiccant and tested for sporozoites using an enzyme-linked immunosorbent assay (ELISA) method [17, 24]. Results from light trap collections at these sites were previously shown to be strongly correlated with contemporaneous human-landing catches [17]. Statistical analyses The objective of this analysis was to characterize the association between entomological surveillance data and incidence of malaria. First, we developed multiple spatiotemporal models of entomological data to obtain smoothed estimates of household EIR. We then used the predicted EIRs from the best-fitting of these models to assess the relationship between EIR and malaria incidence. Modeling EIR over time and space To model total mosquito counts for the region’s two major vectors, Anopheles gambiae sensu lato and Anopheles funestus , we fit negative binomial spatiotemporal generalized additive models (GAMs) for each site, using thin plate splines for temporal smooths and either low rank gaussian process smooths with a power exponential correlation function or thin plate splines to describe the interaction of household projected coordinates. Additional model types were considered as detailed in the supplementary materials. We followed a similar process to model Anopheles -wide sporozoite rates for all mosquitoes that underwent ELISA: these were fit as spatiotemporal binomial GAMs. Independent spatial and temporal smooths were used as described above. In both cases, models without concerning over- or under-dispersion or heteroscedasticity were compared by Akaike information criterion (AIC). For each best-fitting model type, we then compared models specified with spatial smooths only, temporal smooths only, and spatial and temporal smooths by AIC. The best-fitting models after this step were used to generate daily predicted log 2 -transformed aEIR, calculated as the product of the predicted sporozoite rate and predicted vector count (subsequently referred to as modeled aEIR). We fit all GAMs with the mgcv package in R [25]. Residual diagnostics to assess concerning under- or over-dispersion, quantile deviations or influential outliers were performed using the DHARMa package [26]. Modeling the association between incidence of malaria and EIR Incident malaria was chosen as an admittedly imperfect proxy for incident infection: prior studies have demonstrated that the majority of asymptomatic infections in young children progress to symptomatic malaria, so we limited this analysis to children under 5 years of age [27]. The relationship between an individual’s household level aEIRs (log 2 -transformed as described above) and their daily malaria incidence was modeled for each site with a Poisson mixed effects GAM (GAMM) using a thin plate spline smooth in the mgcv package in R. The 14 days following an episode of malaria were excluded from analysis to account for the prophylactic effect of antimalarial treatment. aEIR was lagged by 14 days to account for the P. falciparum intrinsic incubation period; 28-day lags were also evaluated and yielded qualitatively similar results. All models controlled for participant age using a thin plate spline basis and included individual and household IDs as random effects, such that incidence models took the following form (where i represents date, j household, k an individual participant, µ j and γ k the household and individual random effects, and Y i,j,k a random variable: log(case count i,j,k ) = f(log 2 (14d-lagged aEIR i,j )) + f(age k ) + µ j + γ k , Y i,j,k ~Poi(case count i,j,k ) (1) To generate interpretable incidence rate ratios (IRRs), model fitting was repeated treating the association between log 2 -transformed modeled aEIR and incidence as linear on the log scale. To account for uncertainty in entomological parameter estimation, we generated prediction intervals for the entomological GAMs by drawing 1000 samples from the posterior of the fitted values of the models using the gratia package in R [28]. Binomial model samples were weighted according to the number of mosquitoes collected per household over the study period. We then refit the GAMMs with the covariates listed above to these draws and drew an additional 1000 samples from the posterior of the expected value of the model responses. Reported smooths and IRRs reflect the means and 2.5% and 97.5% quantiles of the pooled results and are adjusted for all listed covariates unless otherwise specified. While our main analysis was based on aEIRs derived from the best fitting models of mosquito counts and sporozoite rates, we also assessed the association between malaria incidence and crude aEIRs – defined as the products of vector count and sporozoite rate for each capture session – and modeled aEIRs that omitted either spatial or temporal smooths. We compared the fit of models incorporating different estimates of aEIR using the AIC. For these comparisons, we used the expected responses of entomological models rather than the pooled prediction intervals described above. Results Study population The PRISM 1 cohort study enrolled 454 participants in Jinja, 478 in Kanungu, and 470 in Tororo. To reduce the impact of immunity and host factors that may reduce the probability of an infection leading to clinical malaria, in this analysis we excluded members of the cohort who were older than 5 years of age or had documented sickle cell trait or disease. Our analysis included 439 participants from 239 households. Characteristics of the resulting study population are shown in Table 1 . Participants were followed for a median of 650 days (interquartile range [IQR] 324–1078), during which the median number of cases per person-year was 0 (IQR 0-0.481) in Jinja, 1.14 (0-2.56) in Kanungu, and 4.07 (2.14–7.41) in Tororo. Monthly trends in malaria incidence are shown in Fig. 1 d. Table 1 – Characteristics of study participants. Age refers to age in years at time of enrollment. Jinja Kanungu Tororo Overall No. participants 145 169 125 439 No. households 82 84 73 239 Age (IQR) 2.1 (1, 3.6) 2.1 (1.3, 3.4) 2.3 (1.1, 4) 2.2 (1.1, 3.6) % Male 48.3 49.7 54.4 50.6 Median days followed (IQR) 593 (346, 1079) 755 (490, 1151) 589 (246, 993) 651 (324.5, 1079) Median malaria cases per person (IQR) 0 (0, 1) 2 (0, 5) 5 (1, 9) 1 (0, 5) Median malaria cases per person-year (IQR) 0 (0, 0.481) 1.14 (0, 2.56) 4.07 (2.14, 7.41) 0.96 (0, 3.43) Median proportion of nights using LLIN (IQR) 1 (1, 1) 1 (1, 1) 1 (1, 1) 1 (1, 1) Variation in EIR over time and space Monthly entomological measures over the study period are summarized in Table 2 , with monthly subcounty-wide trends in Anopheles counts, sporozoite rate, crude annualized EIR (aEIR) and malaria incidence among cohort participants shown in Fig. 1 . As expected, the highest vector densities and sporozoite rates were recorded in Tororo, the second-highest in Kanungu, and the lowest in Jinja. Site-specific aEIRs calculated over the course of the study consequently followed the same pattern: 233 in Tororo, 20.3 in Kanungu, and 2.72 in Jinja. Table 2 – Summary statistics for monthly entomological measures. To generate aggregated measures, collection-level measurements were summed by household (“HH”) or site and divided by number of collections. Collection-level measures, which constituted the response variable of the smoothed entomological models used in our final analysis, are included to illustrate the zero inflation described in the main text. Jinja Kanungu Tororo Overall Collection median (IQR) HH median (IQR) Site Collection median (IQR) HH median (IQR) Site Collection median (IQR) HH median (IQR) Site Collection median (IQR) HH median (IQR) Site Vector count 0 (0,1) 0.429 (0.281,1) 0.89 0 (0,2) 1.4 (0.482,4.58) 3.41 10 (2, 39) 28.6 (19,41.8) 33.5 0 (0, 4) 2.51 (0.441, 19.8) 10.5 SR 0 (0,0) 0 (0,0.00147) 0.00837 0 (0,0) 0.00915 (0,0.0222) 0.0163 0 (0, 0) 0.0172 (0.0138,0.0236) 0.019 0 (0, 0) 0.0111 (0, 0.0208) 0.018 aEIR 0 (0,0) 0 (0,3.2) 2.72 0 (0,0) 8.15 (0,27) 20.3 0 (0,0) 178 (108,307) 233 0 (0, 0) 11.1 (0, 111) 69.3 No. collections NA 57 (33,57) 5210 NA 56 (56,57) 5410 NA 40 (39,40) 3860 NA 45 (39,57) 14500 The number of mosquitoes collected during any single collection was consistently low in Jinja and varied little over the study: the median number of mosquitoes collected per household (averaged over the entire study period) was 0.429 (IQR 0.281-1), but the median number at any given collection was 0 (IQR 0–1), and 0 mosquitoes were collected in 3697/5212 (70.9%) collections. Counts varied more widely at the two other sites: in Kanungu, the median per household was 1.4 (0.482–4.58) and median per collection 0 (0–2), with 0 mosquitoes collected in 2977/5414 (55.0%) collections; in Tororo, the median per household was 28.6 (19-41.8) and median per collection 10 (2–39), with 0 mosquitoes collected in 656/3858 (17.0%) collections. Median sporozoite rates by household were less variable: 0 (0–0) in Jinja, 0.00915 (0-0.0222) in Kanungu, and 0.0172 (0.0138–0.0236) in Tororo. Median sporozoite rate per collection was 0 (0–0) for all sites, as sporozoite rates were equal to 0 in the majority of captures in all three: 5175/5212 (99.3%) of captures in Jinja, 5209–5414 (96.2%) in Kanungu, and 2955/3858 (76.6%) in Tororo. The observed sparseness and variability prompted us to consider multiple spatiotemporally smoothed models of vector counts and sporozoite rates. For vector counts, the best-fitting models explained a moderate percentage of the deviance at all sites (Jinja: 45.0; Kanungu: 60.6; Tororo: 57.4) (Supplementary Table 1). For sporozoite rates, the best-fitting models explained a small percentage of the deviance, particularly in Jinja (Jinja: 1.83; Kanungu: 11.1; Tororo: 11.8) (Supplementary Table 1). Predictions generated from the best vector density and sporozoite rate models are overlaid on Figs. 1 a-c. At all sites, we observed biannual peaks in vector counts corresponding roughly to the March-May and August-October rainy seasons. Increases in sporozoite rates corresponded to decreases in vector count, as expected in an aging mosquito population. Vector counts were higher at the eastern border of the study site in Jinja, nearer to Lake Victoria, and at the northern border in Kanungu, where altitude was lower, but were patchy in Tororo. Sporozoite rates were higher in the southwest in Jinja, patchy in Kanungu, and lacked notable spatial structure in Tororo (Supplementary Fig. 1). Temporal and spatial trends in EIR were similar to those for vector counts. Association between entomological metrics and malaria incidence We next evaluated the association between aEIR and malaria incidence. Figure 2 shows the association between average household aEIR and the average incidence of malaria experienced by individuals over the course of the study, for both crude (2a) and modeled (2b) aEIRs. Although analyzing data from the three sites together suggests a positive relationship between aEIR and malaria incidence, much of this association might be attributed to between-site differences, since both aEIR and incidence were lowest in Jinja, moderate in Kanungu and highest in Tororo. To investigate whether the relationship observed when pooling data from all sites applied within sites, we fit models of associations between modeled aEIR and incidence separately for each site (Fig. 3 ). We found a positive association between modeled aEIR and incidence in Kanungu, but not in Jinja and Tororo. Treating the association between log 2 -transformed spatiotemporally modeled aEIR and incidence as linear on the log scale, we estimated that, on average, malaria incidence increased by 9% (IRR 1.09, 95% interval 1.04–1.14) in Kanungu with each doubling of EIR. At both Jinja and Tororo, the 95% credible interval of predicted IRRs for modeled aEIR crossed 1 (Jinja: mean 1.02, 95% interval 0.774–1.26; Tororo: 1.02, 0.986–1.06). Overall, these results were qualitatively similar to trends suggested by average aEIR-incidence plots. In both Kanungu and Tororo, incidence additionally increased with age among younger children before saturating at older ages. No significant relationship was seen in Jinja (Supplementary Fig. 2). We also investigated whether alternative estimates of aEIR could better capture the variance in incidence. We fit incidence models to crude aEIRs, aEIRs generated from temporal vector count and sporozoite rate smooths (ignoring space), and aEIRs generated from spatial vector count and sporozoite rate smooths (ignoring time), comparing the results by AIC (Supplementary Table 2). The spatiotemporal smooths were the best- or second best-performing models for all sites: in Jinja, the best-performing model by AIC was fit to spatially smoothed aEIRs that ignored time; in Kanungu, models fit to spatiotemporally smoothed aEIRs and temporally smoothed aEIRs that ignored space performed equally well; in Tororo, the best-performing model was fit to temporally smoothed aEIRs. Models fit to crude aEIRs did not converge in Jinja and were the worst-performing by AIC at the other two sites. Regardless of the aEIR estimate used, all models were consistent with a positive association in Kanungu (with the exception of the poorly fitting spatial-only model) but showed no or little association in Tororo and Jinja (Supplementary Figs. 3–4). All models explained a relatively small percentage of the deviance: in Jinja 16.8–21.0%; in Kanungu 9.56–11.6%; and in Tororo 5.48–6.60% (Supplementary Table 2). Discussion Modeling based on data from clinical surveillance and entomological measures derived from light trap captures, we found that relationships between household aEIR and individual malaria incidence varied between three study sites in Uganda where transmission varies 100-fold. Overall results suggested a positive association between aEIR and malaria risk, but analyses stratified by site showed that only in Kanungu, an intermediate-EIR and -malaria incidence site, higher household aEIRs were associated with increasing individual malaria incidence. In Jinja (low EIR and low incidence) and Tororo (high EIR and high incidence), associations were weak or absent. There are several possible explanations for this potentially counterintuitive result. First, previous studies have suggested an underlying sigmoid relationship between EIR and malaria risk [5, 7], which is intuitively appealing: regardless of the exact shape of the EIR-incidence association, there is a maximum number of malaria episodes an individual can experience in a year. In keeping with this pattern, the positive association between EIR and incidence observed at our intermediate-EIR site might correspond to the steep part of an EIR-incidence curve, while the absent associations at the low- and high-EIR sites might correspond to the behavior of the curve near its minimum and carrying capacity, respectively. This explanation revives the possibility of a general relationship between entomological measures and disease incidence, though it would not account for the lack of overlap between the expected incidences at the three sites as seen in Fig. 3 . Second, it is possible that features unique to Kanungu allowed us to detect an association between entomological measures and disease incidence. Kanungu is the only of the three sites with an altitude gradient. This geographical feature likely contributes both to a broader within-site aEIR range, across which an association between entomological metrics and incidence may be more easily captured, and to stronger seasonal and spatial trends of mosquito exposure that more closely correlate with corresponding trends in clinical malaria incidence. In this setting, information borrowed from nearby houses and dates would be more informative than at other sites, possibly decreasing variance in our estimates. Third, it is clear from the small proportion of variance explained by our incidence models that there were important drivers of malaria incidence not reflected by entomological surveillance data. Potential candidates include variations in human behavior, immunity, and mosquito feeding patterns. Differences in behavior may mean that household aEIR measured in Jinja and Tororo did not reflect household members’ exposure to infected mosquitoes. Imported cases are one potentially important contributor to local malaria incidence: recent overnight travel has been associated with increased malaria risk at all three sites, but was found to be more common in Jinja and Tororo than in Kanungu [9]. LLIN adherence may similarly decouple household EIR from incidence, although bednet distribution was previously shown to have only a modest effect on malaria risk at the study sites despite high reported rates of adherence [10]. Anti-parasite and/or anti-disease immunity is also likely to drive incidence patterns, particularly in the high-transmission setting of Tororo. While we attempted to minimize the impact of such immunity by restricting the age of the study population to those under five and controlling for the effect of age in the final analysis, these approaches are imperfect. The nonlinear relationship we recovered between age and malaria incidence in Kanungu and Tororo is consistent with prior analyses of children under five years in this cohort [11]. Taking advantage of molecular methods to identify incident infections, rather than incident disease, may address some of these concerns by accounting for asymptomatic infections, including superinfections [12]. Finally, local mosquito feeding behavior may also have differed in ways not captured, or captured differentially by site, in our surveillance data, including variation in biting time and location relative to human behaviors. These differences could conceivably stem from environmental heterogeneity, or from differences in species composition within the An. gambiae s.l. species complex. Both An. gambiae sensu stricto and An. arabiensis are endemic to the study sites, with An. arabiensis exhibiting less anthropophilic and endophagic tendencies [13, 14] and a recent temporal association with lower malaria risk in Tororo [15]. The lack of association between entomological and clinical metrics at two of our sites may also reflect the imprecision of entomological data derived from a single CDC LT per household-month, with sporozoite rates estimated from an even smaller subset. Whether CDC LT data themselves are problematic is unclear: EIRs derived from CDC LT data have been comparable with gold standard human-landing catches both in prior analyses of PRISM 1 cohort data and in subsequent entomological studies conducted in Tororo [16, 17], but other studies comparing CDC LT and human-landing catches of anophelines have noted significant differences in overall vector densities, species composition, sporozoite rates, and parous rates [18–21]. To summarize, our study’s use of concurrent longitudinal spatiotemporal entomological and clinical data across a wide range of transmission intensities afforded us a uniquely detailed view of the relationship between these two markers of exposure. Nevertheless, the relative lack of overlap in EIRs between the three sites limited our ability to distinguish between potential explanations for the EIR-incidence pattern we observed: either a general, potentially sigmoid, EIR-incidence relationship, or site-specific differences in exposure patterns, host immunity, and/or vector characteristics. Limitations of our study included a limited ability to characterize direct mosquito-human exposure as would be afforded by simultaneous human behavioral observations and human-landing catches, a focus on malaria incidence rather than incident P. falciparum infection, and a lack of sub-species complex mosquito species identification that might have obscured significant differences in the relative roles played by vector species. Entomological data collections are inherently noisy and sparse relative to the exposure patterns they are meant to reflect, and a fine-scale EIR-incidence association may not exist except in the highly favorable setting offered by a site like Kanungu. Conclusions In conclusion, despite strong theoretical support for a general relationship between the aEIR and malaria incidence, household-level EIRs estimated from smoothed mosquito surveillance data were significantly associated with individual malaria incidences in an intermediate-transmission site, but not at low- or high-transmission sites. Further assessment of this relationship using data collected at a finer temporal scale with molecular identification of new infections may be helpful to tease apart this heterogeneity. Abbreviations aEIR annualized entomological inoculation rate AIC Akaike information criterion CDC LT CDC light trap EIR entomological inoculation rate ELISA enzyme-linked immunosorbent assay GAM generalized additive model GAMM generalized additive mixed model IQR interquartile range IRR incidence rate ratio IRS indoor residual spraying LLIN long-lasting insecticidal net PRISM Program for Resistance, Immunology, Surveillance and Modeling of Malaria Declarations Ethics approval and consent to participate: Ethical approval was obtained from the Makerere University School of Medicine Research and Ethics Committee, the Uganda National Council for Science and Technology, the London School of Hygiene and Tropical Medicine Ethics Committee, the School of Biological and Biomedical Sciences Ethics Committee, Durham University and the University of California, San Francisco Committee on Human Research. Consent for publication: Not applicable Availability of data and materials: The datasets used and/or analyzed during the current study are publicly available at ClinEpiDB: https://clinepidb.org/ce/app/workspace/analyses/DS_0ad509829e Competing interests: The authors have no competing interests to declare. Funding: Funding was provided by the National Institutes of Health as part of the International Centers of Excellence in Malaria Program (2U19AI089674) and the UCSF Biology of Infectious Diseases Training Program (2T32AI007641-21). Authors’ contributions: Conceptualization – MM, IRB, BG Data curation – KM, AM Formal analysis – MM Funding acquisition – PJR, MRK, GD, BG Investigation – EA, IRB, MRK, PJR, GD, JN, KM Methodology – MM, IRB Project administration – EA, JN, MRK, GD Writing (original draft) - MM Writing (review and editing) – MM, EA, MRK, PJR, JN, KM, AM, GD, BG, IRB Acknowledgements: We would like to thank the study participants and their families, as well as the study team, Makerere University-UCSF Research Collaboration, and Infectious Diseases Research Collaboration. Portions of this work were performed on the Wynton HPC Co-Op cluster, which is supported by UCSF research faculty and UCSF institutional funds. References World Health Organization: Malaria surveillance, monitoring and evaluation: a reference manual. 2018. Shaukat AM, Breman JG, McKenzie FE: Using the entomological inoculation rate to assess the impact of vector control on malaria parasite transmission and elimination. Malaria journal 2010, 9: 1-12. Garrett-Jones C, Ferreira Neto JA, Organization WH: The prognosis for interruption of malaria transmission through assessment of the mosquito's vectorial capacity. World Health Organization; 1964. Ross RS: The prevention of malaria. England1911. Smith D, Dushoff J, Snow R, Hay S: The entomological inoculation rate and Plasmodium falciparum infection in African children. Nature 2005, 438: 492-495. Beier JC, Killeen GF, Githure JI: entomologic inoculation rates and Plasmodium falciparum malaria prevalence in Africa. The American journal of tropical medicine and hygiene 1999, 61: 109-113. Charlwood J, Smith T, Lyimo E, Kitua A, Masanja H, Booth M, Alonso P, Tanner M: Incidence of Plasmodium falciparum infection in infants in relation to exposure to sporozoite-infected anophelines. The American journal of tropical medicine and hygiene 1998, 59: 243-251. Beier JC, Oster CN, Onyango FK, Bales JD, Sherwood JA, Perkins PV, Chumo DK, Koech DV, Whitmire RE, Roberts CR: Plasmodium falciparum incidence relative to entomologic inoculation rates at a site proposed for testing malaria vaccines in western Kenya. The American journal of tropical medicine and hygiene 1994, 50: 529-536. Arinaitwe E, Dorsey G, Nankabirwa JI, Kigozi SP, Katureebe A, Kakande E, Rek J, Rosenthal PJ, Drakeley C, Kamya MR: Association between recent overnight travel and risk of malaria: a prospective cohort study at 3 sites in Uganda. Clinical Infectious Diseases 2019, 68: 313-320. Katureebe A, Zinszer K, Arinaitwe E, Rek J, Kakande E, Charland K, Kigozi R, Kilama M, Nankabirwa J, Yeka A: Measures of malaria burden after long-lasting insecticidal net distribution and indoor residual spraying at three sites in Uganda: a prospective observational study. PLoS medicine 2016, 13: e1002167. Rodriguez-Barraquer I, Arinaitwe E, Jagannathan P, Kamya MR, Rosenthal PJ, Rek J, Dorsey G, Nankabirwa J, Staedke SG, Kilama M: Quantification of anti-parasite and anti-disease immunity to malaria as a function of age and exposure. Elife 2018, 7: e35832. Briggs J, Teyssier N, Nankabirwa JI, Rek J, Jagannathan P, Arinaitwe E, Bousema T, Drakeley C, Murray M, Crawford E: Sex-based differences in clearance of chronic Plasmodium falciparum infection. Elife 2020, 9: e59872. Mahande A, Mosha F, Mahande J, Kweka E: Feeding and resting behaviour of malaria vector, Anopheles arabiensis with reference to zooprophylaxis. Malaria journal 2007, 6: 1-6. Tirados I, Costantini C, Gibson G, Torr SJ: Blood-feeding behaviour of the malarial mosquito Anopheles arabiensis: implications for vector control. Med Vet Entomol 2006, 20: 425-437. Kamya MR, Nankabirwa JI, Arinaitwe E, Rek J, Zedi M, Maiteki-Sebuguzi C, Opigo J, Staedke SG, Oruni A, Donnelly MJ, et al: Dramatic resurgence of malaria after 7 years of intensive vector control interventions in Eastern Uganda. PLOS Glob Public Health 2024, 4: e0003254. Mawejje HD, Asiimwe JR, Kyagamba P, Kamya MR, Rosenthal PJ, Lines J, Dorsey G, Staedke SG: Impact of different mosquito collection methods on indicators of Anopheles malaria vectors in Uganda. Malaria Journal 2022, 21: 388. Kilama M, Smith DL, Hutchinson R, Kigozi R, Yeka A, Lavoy G, Kamya MR, Staedke SG, Donnelly MJ, Drakeley C: Estimating the annual entomological inoculation rate for Plasmodium falciparum transmitted by Anopheles gambiae sl using three sampling methods in three sites in Uganda. Malaria journal 2014, 13: 1-13. Rubio-Palis Y, Curtis C: Evaluation of different methods of catching anopheline mosquitoes in western Venezuela. Journal of the American Mosquito Control Association 1992, 8: 261-267. Mbogo CN, Glass GE, Forster D, Kabiru EW, Githure JI, Ouma JH, Beier JC: Evaluation of light traps for sampling anopheline mosquitoes in Kilifi, Kenya. J Am Mosq Control Assoc 1993, 9: 260-263. Hii J, Smith T, Mai A, Ibam E, Alpers M: Comparison between anopheline mosquitoes (Diptera: Culicidae) caught using different methods in a malaria endemic area of Papua New Guinea. Bulletin of entomological research 2000, 90: 211-219. Lines J, Curtis C, Wilkes T, Njunwa K: Monitoring human-biting mosquitoes (Diptera: Culicidae) in Tanzania with light-traps hung beside mosquito nets. Bulletin of entomological research 1991, 81: 77-84. Uganda Bureau of Statistics - UBOS, ICF Macro: Uganda Malaria Indicator Survey (MIS) 2009. Calverton, Maryland, USA: UBOS and ICF Macro; 2010. Kamya MR, Arinaitwe E, Wanzira H, Katureebe A, Barusya C, Kigozi SP, Kilama M, Tatem AJ, Rosenthal PJ, Drakeley C: Malaria transmission, infection, and disease at three sites with varied transmission intensity in Uganda: implications for malaria control. The American journal of tropical medicine and hygiene 2015, 92: 903. Wirtz R, Duncan J, Njelesani E, Schneider I, Brown A, Oster C, Were JB, Webster H: ELISA method for detecting Plasmodium falciparum circumsporozoite antibody. Bulletin of the World Health Organization 1989, 67: 535. Wood SN: Fast stable restricted maximum likelihood and marginal likelihood estimation of semiparametric generalized linear models. Journal of the Royal Statistical Society Series B: Statistical Methodology 2011, 73: 3-36. Hartig F: DHARMa: residual diagnostics for hierarchical (multi-level/mixed) regression models. R Packag version 020 2018. Missinou MA, Lell B, Kremsner PG: Uncommon asymptomatic Plasmodium falciparum infections in Gabonese children. Clinical Infectious Diseases 2003, 36: 1198-1202. Simpson G, Singmann H: R Package: gratia. ggplot ‐based graphics and other useful functions for GAMs fitted using mgcv, 0.1 ‐0 (ggplot ‐based graphics and utility functions for working with GAMs fitted using the mgcv package). DOI Unavailable 2018. Additional Declarations No competing interests reported. Supplementary Files Supp115.docx Cite Share Download PDF Status: Under Review Version 1 posted Editorial decision: Revision requested 28 Apr, 2025 Reviews received at journal 22 Apr, 2025 Reviewers agreed at journal 06 Apr, 2025 Reviewers agreed at journal 02 Apr, 2025 Reviews received at journal 21 Mar, 2025 Reviewers agreed at journal 10 Mar, 2025 Reviewers agreed at journal 03 Feb, 2025 Reviewers agreed at journal 01 Feb, 2025 Reviewers invited by journal 29 Jan, 2025 Editor assigned by journal 28 Jan, 2025 Submission checks completed at journal 28 Jan, 2025 First submitted to journal 27 Jan, 2025 You are reading this latest preprint version Research Square lets you share your work early, gain feedback from the community, and start making changes to your manuscript prior to peer review in a journal. As a division of Research Square Company, we’re committed to making research communication faster, fairer, and more useful. We do this by developing innovative software and high quality services for the global research community. Our growing team is made up of researchers and industry professionals working together to solve the most critical problems facing scientific publishing. Also discoverable on Platform About Our Team In Review Editorial Policies Advisory Board Help Center Resources Author Services Accessibility API Access RSS feed Manage Cookie Preferences © Research Square 2026 | ISSN 2693-5015 (online) Privacy Policy Terms of Service Do Not Sell My Personal Information {"props":{"pageProps":{"initialData":{"identity":"rs-5914493","acceptedTermsAndConditions":true,"allowDirectSubmit":false,"archivedVersions":[],"articleType":"Research Article","associatedPublications":[],"authors":[{"id":408772883,"identity":"eca3414a-4828-409e-8136-9470415e777f","order_by":0,"name":"Max McClure","email":"data:image/png;base64,iVBORw0KGgoAAAANSUhEUgAAAZAAAAAyAQMAAABI0h/eAAAABlBMVEX///8AAABVwtN+AAAACXBIWXMAAA7EAAAOxAGVKw4bAAAA00lEQVRIiWNgGAWjYBADGSBmfAAk+IHYgCgtPEDMDFIq2UCKFjYJorTotp8xYLpRY8djcPvwsWqemjoJBvbmbRL4tJidSUtgzjmWzGNwLi3tNs+xwxIMPMfK8Gs5kHyAObfhAI/BGR6z27wNB+oYJHLM8Gs5/7ABqoX/WzFvA9Bh8m8IaLmBsIWNmbeBWYJBgoeQlmcJh0F+kTzDZiw5B+gXNp60Ygv8DssxfJxTYyfHd4b54Yc3wBDjZz+88QY+LSBwAIXHRkj5KBgFo2AUjALCAAAuKEHMimQSKwAAAABJRU5ErkJggg==","orcid":"","institution":"University of California, San Francisco","correspondingAuthor":true,"prefix":"","firstName":"Max","middleName":"","lastName":"McClure","suffix":""},{"id":408772884,"identity":"2a627f15-1ded-41dd-b349-d5a383c708f8","order_by":1,"name":"Emmanuel Arinaitwe","email":"","orcid":"","institution":"Infectious Diseases Research Collaboration","correspondingAuthor":false,"prefix":"","firstName":"Emmanuel","middleName":"","lastName":"Arinaitwe","suffix":""},{"id":408772885,"identity":"115d4c37-4bca-4f9b-8aa4-d0d48618162d","order_by":2,"name":"Moses R. Kamya","email":"","orcid":"","institution":"Makerere University College of Health Sciences","correspondingAuthor":false,"prefix":"","firstName":"Moses","middleName":"R.","lastName":"Kamya","suffix":""},{"id":408772886,"identity":"6bc8eb2c-bcd0-4356-8551-d5923a595d62","order_by":3,"name":"Philip J. Rosenthal","email":"","orcid":"","institution":"University of California, San Francisco","correspondingAuthor":false,"prefix":"","firstName":"Philip","middleName":"J.","lastName":"Rosenthal","suffix":""},{"id":408772887,"identity":"89f64660-ea80-4cd8-b33b-04680b2b721e","order_by":4,"name":"Joaniter Nankabirwa","email":"","orcid":"","institution":"Makerere University College of Health Sciences","correspondingAuthor":false,"prefix":"","firstName":"Joaniter","middleName":"","lastName":"Nankabirwa","suffix":""},{"id":408772889,"identity":"7da730e7-ebaa-40ff-932d-aeca6a53da05","order_by":5,"name":"Kilama Maxwell","email":"","orcid":"","institution":"Infectious Diseases Research Collaboration","correspondingAuthor":false,"prefix":"","firstName":"Kilama","middleName":"","lastName":"Maxwell","suffix":""},{"id":408772890,"identity":"63936f0a-4a6d-49f6-b167-eeea170075f0","order_by":6,"name":"Alex Musiime","email":"","orcid":"","institution":"Ministry of Health","correspondingAuthor":false,"prefix":"","firstName":"Alex","middleName":"","lastName":"Musiime","suffix":""},{"id":408772892,"identity":"2258bdde-5cd5-4f53-aec2-c189438e101d","order_by":7,"name":"Grant Dorsey","email":"","orcid":"","institution":"University of California, San Francisco","correspondingAuthor":false,"prefix":"","firstName":"Grant","middleName":"","lastName":"Dorsey","suffix":""},{"id":408772894,"identity":"81e5663a-fe36-429a-9a90-5642d82e9641","order_by":8,"name":"Bryan Greenhouse","email":"","orcid":"","institution":"University of California, San Francisco","correspondingAuthor":false,"prefix":"","firstName":"Bryan","middleName":"","lastName":"Greenhouse","suffix":""},{"id":408772897,"identity":"d7f01eeb-6b7d-49a3-aeef-b3f851d7a252","order_by":9,"name":"Isabel Rodriguez-Barraquer","email":"","orcid":"","institution":"University of California, San Francisco","correspondingAuthor":false,"prefix":"","firstName":"Isabel","middleName":"","lastName":"Rodriguez-Barraquer","suffix":""}],"badges":[],"createdAt":"2025-01-27 18:53:03","currentVersionCode":1,"declarations":"","doi":"10.21203/rs.3.rs-5914493/v1","doiUrl":"https://doi.org/10.21203/rs.3.rs-5914493/v1","draftVersion":[],"editorialEvents":[],"editorialNote":"","failedWorkflow":false,"files":[{"id":75405817,"identity":"954d1186-cc64-4020-aa28-4bdfe1b64f25","added_by":"auto","created_at":"2025-02-04 08:49:03","extension":"png","order_by":1,"title":"Figure 1","display":"","copyAsset":false,"role":"figure","size":39541,"visible":true,"origin":"","legend":"\u003cp\u003eMonthly vector count, sporozoite rate, annualized entomological inoculation rate (aEIR) and malaria incidence (cases per person-year) by site. Points with whiskers represent mean monthly crude data and associated uncertainty, while lines with confidence bands represent mean spatiotemporal model outputs and associated uncertainty. For vector counts, whiskers show 95% confidence intervals for summed \u003cem\u003eAnopheles gambiae \u003c/em\u003es.l. and \u003cem\u003eAnopheles funestus \u003c/em\u003es.l.\u003cem\u003e \u003c/em\u003ecounts modeled as a Poisson process. For sporozoite rate and malaria incidence, whiskers show 95% confidence intervals from the exact binomial test. For aEIR, whiskers represent 2.5% and 97.5% quantiles of collection-level aEIRs. Confidence bands for all GAM outputs represent the 2.5% and 97.5% quantiles of pooled draws as described in the text.\u003c/p\u003e","description":"","filename":"floatimage1.png","url":"https://assets-eu.researchsquare.com/files/rs-5914493/v1/125c2998c07f9a81de26bfb6.png"},{"id":75407998,"identity":"fc68a63c-6239-4043-8c3b-7421e16418f6","added_by":"auto","created_at":"2025-02-04 08:57:03","extension":"png","order_by":2,"title":"Figure 2","display":"","copyAsset":false,"role":"figure","size":57241,"visible":true,"origin":"","legend":"\u003cp\u003eAssociation between household mean crude (a) or modeled (b) annualized entomological inoculation rate (aEIR) and individuals’ annual malaria incidence experienced over the course of the study, grouped by site. Plots are restricted to individuals followed for at least 365 days. Point sizes represent the duration of follow-up time in days. In panel b, a line of best fit with 95% confidence intervals is overlaid across all sites.\u003c/p\u003e","description":"","filename":"floatimage2.png","url":"https://assets-eu.researchsquare.com/files/rs-5914493/v1/2076c88096d67d07429e2baf.png"},{"id":75405815,"identity":"d3c48d59-f8d5-4ee6-861b-85b7cb2dfc4c","added_by":"auto","created_at":"2025-02-04 08:49:03","extension":"png","order_by":3,"title":"Figure 3","display":"","copyAsset":false,"role":"figure","size":12068,"visible":true,"origin":"","legend":"\u003cp\u003eIndividual malaria incidence as a smooth function of modeled aEIR with a 14-day lag, grouped by site. Lines show means and ribbons show the 2.5% and 97.5% quantiles of the expected incidence. The density plot at the upper margin indicates the distribution of expected modeled aEIRs by site. The range of predictions for each site is restricted to the 2.5% and 97.5% quantiles of the expected EIRs for that site.\u003c/p\u003e","description":"","filename":"floatimage3.png","url":"https://assets-eu.researchsquare.com/files/rs-5914493/v1/ed4b971396db9784691c39db.png"},{"id":75409671,"identity":"265228de-21e1-4509-a222-1218c33a01c3","added_by":"auto","created_at":"2025-02-04 09:05:04","extension":"pdf","order_by":0,"title":"","display":"","copyAsset":false,"role":"manuscript-pdf","size":2045228,"visible":true,"origin":"","legend":"","description":"","filename":"manuscript.pdf","url":"https://assets-eu.researchsquare.com/files/rs-5914493/v1/d10ea51d-b186-4a5a-95d4-6770fa0247e5.pdf"},{"id":75405820,"identity":"6374ed94-afb5-4d23-8aec-6a6d76804e84","added_by":"auto","created_at":"2025-02-04 08:49:03","extension":"docx","order_by":0,"title":"","display":"","copyAsset":false,"role":"supplement","size":405239,"visible":true,"origin":"","legend":"","description":"","filename":"Supp115.docx","url":"https://assets-eu.researchsquare.com/files/rs-5914493/v1/d0547f2890d4784df01ae6bc.docx"}],"financialInterests":"No competing interests reported.","formattedTitle":"Relating household entomological measures to individual malaria risk","fulltext":[{"header":"Background","content":"\u003cp\u003eAnopheline mosquitoes transmit malaria to humans, and exposure to \u003cem\u003ePlasmodium\u003c/em\u003e-carrying mosquitoes corresponds to human malaria risk. WHO guidelines recommend surveillance of entomological proxies for transmission as a component of integrated vector management programs [1, 2]. However, current understanding of the quantitative relationship between mosquito exposure and human infection risk is surprisingly limited.\u003c/p\u003e \u003cp\u003eThe gold standard entomological measure of malaria exposure is the entomological inoculation rate (EIR), or the number of infectious bites received by an individual in a given time period [2]. EIR is typically approximated as the product of the malaria vector density (as determined by mosquito captures) and the proportion of tested mosquitoes positive for sporozoites [3]. EIR has a positive relationship to parasite prevalence [4] that has been validated with real-world data at community scales [5, 6], and an association with the incidence of blood stage infections down to the village or neighborhood level [7, 8], but it is unclear if it is able to capture heterogeneity in transmission driving incidence at smaller scales such as the household or the individual. Factors potentially obscuring such an association include measurement error and uncaptured spatiotemporal variability in entomological indices, behavioral heterogeneity between vector species, and behavioral and immunologic heterogeneity among human hosts.\u003c/p\u003e \u003cp\u003eTo better assess the association between EIR and individual malaria incidence, we analyzed data from completed longitudinal cohort studies that included passive and active clinical surveillance of participants and paired household-level entomological measurements at sites with varying transmission. We used flexible spatiotemporal models to smooth crude entomological measures and allow for nonlinear associations between exposure and human disease risk. These cohorts were well-suited to assess associations between mosquito exposure and clinical outcomes given their concurrent entomological and clinical data collection across a range of transmission intensities, a focus on children (with lower rates of acquired immunity compared to adults), and relatively stable transmission during the study period.\u003c/p\u003e \n\n \n\n\n\n \n\n "},{"header":"Methods","content":"\u003ch3\u003eStudy location\u003c/h3\u003e\u003cp\u003eThe original Program for Resistance, Immunology, Surveillance and Modeling of Malaria (PRISM 1) cohort studies were conducted from 2011–2017 in three Ugandan subcounties representing a spectrum of malaria transmission settings: Walukuba, Jinja District, a peri-urban area near the northern shore of Lake Victoria with the lowest transmission; Kihihi, Kanungu District, a rural area in the southwest near the country’s border with the Democratic Republic of the Congo with relatively moderate transmission; and Nagongera, Tororo District, a rural area in the southeast near the country’s border with Kenya with the highest transmission. \u003cem\u003ePlasmodium falciparum\u003c/em\u003e is the dominant malaria parasite throughout Uganda and was reported to account for over 98% of infections nationally at the time of the study [22].\u003c/p\u003e\u003cp\u003eDuring the data collection period, long-lasting insecticidal nets (LLIN) were distributed in Jinja in November 2013, Kanungu in June 2014, and Tororo in November 2013; LLINs were also provided to all study participants at the time of enrollment in the cohort studies. For this analysis, in Tororo only, we excluded any data collected after a participating household underwent indoor residual spraying (IRS), administered in the district since December 2014. IRS was not implemented in Jinja or Kanungu.\u003c/p\u003e\u003ch2\u003eStudy design and data collection\u003c/h2\u003e\u003cp\u003eThe study protocol is described in detail elsewhere [23]. Briefly, 100 households containing at least one resident 0.5–10 years of age and one adult resident were randomly selected from each subcounty. All children in each household between 0.5–10 years of age that met study eligibility criteria were enrolled. Household latitude and longitude were mapped using handheld GPS and projected to UTM zone 36 coordinates. Enrollment was dynamic over the course of the study – children from participating households joined the study as they became eligible and left as they aged out.\u003c/p\u003e\u003cp\u003eRoutine clinical visits were conducted every 3 months and included a standardized clinical assessment and collection of thick blood smears. Participants were encouraged to present to dedicated study clinics (one per site) for evaluation of any medical needs outside routine visits: if participants were febrile at the time of evaluation or reported fever within the past 24 hours, thick blood smears were obtained. Malaria was defined as a fever (tympanic temperature \u003cspan type=\"Underline\" class=\"Underline\" name=\"Emphasis\"\u003e≥\u003c/span\u003e 38°C or reported fever within the past 24 hours) with a thick blood smear positive for malaria parasites [23]. Participants diagnosed with malaria were treated according to national treatment guidelines.\u003c/p\u003e\u003cp\u003eMosquito collections were conducted once a month using CDC light traps (CDC LT) set from 7pm to 7am 1 meter above the floor at the foot of the bed in one bedroom of each participating household. All mosquito species were identified morphologically. A random subset of mosquitoes from each capture (maximum 50 per collection) was stored on desiccant and tested for sporozoites using an enzyme-linked immunosorbent assay (ELISA) method [17, 24]. Results from light trap collections at these sites were previously shown to be strongly correlated with contemporaneous human-landing catches [17].\u003c/p\u003e\u003ch3\u003eStatistical analyses\u003c/h3\u003e\u003cp\u003eThe objective of this analysis was to characterize the association between entomological surveillance data and incidence of malaria. First, we developed multiple spatiotemporal models of entomological data to obtain smoothed estimates of household EIR. We then used the predicted EIRs from the best-fitting of these models to assess the relationship between EIR and malaria incidence.\u003c/p\u003e\u003ch3\u003eModeling EIR over time and space\u003c/h3\u003e\u003cp\u003eTo model total mosquito counts for the region’s two major vectors, \u003cem\u003eAnopheles gambiae\u003c/em\u003e sensu lato and \u003cem\u003eAnopheles funestus\u003c/em\u003e, we fit negative binomial spatiotemporal generalized additive models (GAMs) for each site, using thin plate splines for temporal smooths and either low rank gaussian process smooths with a power exponential correlation function or thin plate splines to describe the interaction of household projected coordinates. Additional model types were considered as detailed in the supplementary materials.\u003c/p\u003e\u003cp\u003eWe followed a similar process to model \u003cem\u003eAnopheles\u003c/em\u003e-wide sporozoite rates for all mosquitoes that underwent ELISA: these were fit as spatiotemporal binomial GAMs. Independent spatial and temporal smooths were used as described above.\u003c/p\u003e\u003cp\u003eIn both cases, models without concerning over- or under-dispersion or heteroscedasticity were compared by Akaike information criterion (AIC). For each best-fitting model type, we then compared models specified with spatial smooths only, temporal smooths only, and spatial and temporal smooths by AIC. The best-fitting models after this step were used to generate daily predicted log\u003csub\u003e2\u003c/sub\u003e-transformed aEIR, calculated as the product of the predicted sporozoite rate and predicted vector count (subsequently referred to as modeled aEIR).\u003c/p\u003e\u003cp\u003eWe fit all GAMs with the mgcv package in R [25]. Residual diagnostics to assess concerning under- or over-dispersion, quantile deviations or influential outliers were performed using the DHARMa package [26].\u003c/p\u003e\u003ch3\u003eModeling the association between incidence of malaria and EIR\u003c/h3\u003e\u003cp\u003eIncident malaria was chosen as an admittedly imperfect proxy for incident infection: prior studies have demonstrated that the majority of asymptomatic infections in young children progress to symptomatic malaria, so we limited this analysis to children under 5 years of age [27].\u003c/p\u003e\u003cp\u003eThe relationship between an individual’s household level aEIRs (log\u003csub\u003e2\u003c/sub\u003e-transformed as described above) and their daily malaria incidence was modeled for each site with a Poisson mixed effects GAM (GAMM) using a thin plate spline smooth in the mgcv package in R. The 14 days following an episode of malaria were excluded from analysis to account for the prophylactic effect of antimalarial treatment. aEIR was lagged by 14 days to account for the \u003cem\u003eP. falciparum\u003c/em\u003e intrinsic incubation period; 28-day lags were also evaluated and yielded qualitatively similar results. All models controlled for participant age using a thin plate spline basis and included individual and household IDs as random effects, such that incidence models took the following form (where \u003cem\u003ei\u003c/em\u003e represents date, \u003cem\u003ej\u003c/em\u003e household, \u003cem\u003ek\u003c/em\u003e an individual participant, \u003cem\u003eµ\u003c/em\u003e\u003csub\u003e\u003cem\u003ej\u003c/em\u003e\u003c/sub\u003e and \u003cem\u003eγ\u003c/em\u003e\u003csub\u003e\u003cem\u003ek\u003c/em\u003e\u003c/sub\u003e the household and individual random effects, and \u003cem\u003eY\u003c/em\u003e\u003csub\u003e\u003cem\u003ei,j,k\u003c/em\u003e\u003c/sub\u003e a random variable:\u003c/p\u003e\u003cp\u003e \u003cem\u003elog(case count\u003c/em\u003e \u003csub\u003e \u003cem\u003ei,j,k\u003c/em\u003e \u003c/sub\u003e \u003cem\u003e) = f(log\u003c/em\u003e \u003csub\u003e \u003cem\u003e2\u003c/em\u003e \u003c/sub\u003e \u003cem\u003e(14d-lagged aEIR\u003c/em\u003e \u003csub\u003e \u003cem\u003ei,j\u003c/em\u003e \u003c/sub\u003e \u003cem\u003e)) + f(age\u003c/em\u003e \u003csub\u003e \u003cem\u003ek\u003c/em\u003e \u003c/sub\u003e \u003cem\u003e) + µ\u003c/em\u003e \u003csub\u003e \u003cem\u003ej\u003c/em\u003e \u003c/sub\u003e \u003cem\u003e+ γ\u003c/em\u003e\u003csub\u003e\u003cem\u003ek\u003c/em\u003e\u003c/sub\u003e, \u003cem\u003eY\u003c/em\u003e\u003csub\u003e\u003cem\u003ei,j,k\u003c/em\u003e\u003c/sub\u003e\u003cem\u003e~Poi(case count\u003c/em\u003e\u003csub\u003e\u003cem\u003ei,j,k\u003c/em\u003e\u003c/sub\u003e\u003cem\u003e)\u003c/em\u003e (1)\u003c/p\u003e\u003cp\u003eTo generate interpretable incidence rate ratios (IRRs), model fitting was repeated treating the association between log\u003csub\u003e2\u003c/sub\u003e-transformed modeled aEIR and incidence as linear on the log scale.\u003c/p\u003e\u003cp\u003eTo account for uncertainty in entomological parameter estimation, we generated prediction intervals for the entomological GAMs by drawing 1000 samples from the posterior of the fitted values of the models using the gratia package in R [28]. Binomial model samples were weighted according to the number of mosquitoes collected per household over the study period. We then refit the GAMMs with the covariates listed above to these draws and drew an additional 1000 samples from the posterior of the expected value of the model responses. Reported smooths and IRRs reflect the means and 2.5% and 97.5% quantiles of the pooled results and are adjusted for all listed covariates unless otherwise specified.\u003c/p\u003e\u003cp\u003eWhile our main analysis was based on aEIRs derived from the best fitting models of mosquito counts and sporozoite rates, we also assessed the association between malaria incidence and crude aEIRs – defined as the products of vector count and sporozoite rate for each capture session – and modeled aEIRs that omitted either spatial or temporal smooths. We compared the fit of models incorporating different estimates of aEIR using the AIC. For these comparisons, we used the expected responses of entomological models rather than the pooled prediction intervals described above.\u003c/p\u003e"},{"header":"Results","content":"\u003cdiv id=\"Sec8\" class=\"Section2\"\u003e \u003ch2\u003eStudy population\u003c/h2\u003e \u003cp\u003eThe PRISM 1 cohort study enrolled 454 participants in Jinja, 478 in Kanungu, and 470 in Tororo. To reduce the impact of immunity and host factors that may reduce the probability of an infection leading to clinical malaria, in this analysis we excluded members of the cohort who were older than 5 years of age or had documented sickle cell trait or disease.\u003c/p\u003e \u003cp\u003e Our analysis included 439 participants from 239 households. Characteristics of the resulting study population are shown in Table\u0026nbsp;\u003cspan refid=\"Tab1\" class=\"InternalRef\"\u003e1\u003c/span\u003e. Participants were followed for a median of 650 days (interquartile range [IQR] 324\u0026ndash;1078), during which the median number of cases per person-year was 0 (IQR 0-0.481) in Jinja, 1.14 (0-2.56) in Kanungu, and 4.07 (2.14\u0026ndash;7.41) in Tororo. Monthly trends in malaria incidence are shown in Fig.\u0026nbsp;\u003cspan refid=\"Fig1\" class=\"InternalRef\"\u003e1\u003c/span\u003ed.\u003c/p\u003e \u003cp\u003e \u003cdiv class=\"gridtable\"\u003e\u003ctable float=\"Yes\" id=\"Tab1\" border=\"1\"\u003e \u003ccaption language=\"En\"\u003e \u003cdiv class=\"CaptionNumber\"\u003eTable 1\u003c/div\u003e \u003cdiv class=\"CaptionContent\"\u003e \u003cp\u003e\u0026ndash; Characteristics of study participants. Age refers to age in years at time of enrollment.\u003c/p\u003e \u003c/div\u003e \u003c/caption\u003e \u003ccolgroup cols=\"5\"\u003e \u003cdiv align=\"left\" class=\"colspec\" colname=\"c1\" colnum=\"1\"\u003e\u003c/div\u003e \u003cdiv align=\"left\" class=\"colspec\" colname=\"c2\" colnum=\"2\"\u003e\u003c/div\u003e \u003cdiv align=\"left\" class=\"colspec\" colname=\"c3\" colnum=\"3\"\u003e\u003c/div\u003e \u003cdiv align=\"left\" class=\"colspec\" colname=\"c4\" colnum=\"4\"\u003e\u003c/div\u003e \u003cdiv align=\"left\" class=\"colspec\" colname=\"c5\" colnum=\"5\"\u003e\u003c/div\u003e \u003cthead\u003e \u003ctr\u003e \u003cth align=\"left\" colname=\"c1\"\u003e\u0026nbsp;\u003c/th\u003e \u003cth align=\"left\" colname=\"c2\"\u003e \u003cp\u003eJinja\u003c/p\u003e \u003c/th\u003e \u003cth align=\"left\" colname=\"c3\"\u003e \u003cp\u003eKanungu\u003c/p\u003e \u003c/th\u003e \u003cth align=\"left\" colname=\"c4\"\u003e \u003cp\u003eTororo\u003c/p\u003e \u003c/th\u003e \u003cth align=\"left\" colname=\"c5\"\u003e \u003cp\u003eOverall\u003c/p\u003e \u003c/th\u003e \u003c/tr\u003e \u003c/thead\u003e \u003ctbody\u003e \u003ctr\u003e \u003ctd align=\"left\" colname=\"c1\"\u003e \u003cp\u003eNo. participants\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c2\"\u003e \u003cp\u003e145\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c3\"\u003e \u003cp\u003e169\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c4\"\u003e \u003cp\u003e125\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c5\"\u003e \u003cp\u003e439\u003c/p\u003e \u003c/td\u003e \u003c/tr\u003e \u003ctr\u003e \u003ctd align=\"left\" colname=\"c1\"\u003e \u003cp\u003eNo. households\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c2\"\u003e \u003cp\u003e82\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c3\"\u003e \u003cp\u003e84\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c4\"\u003e \u003cp\u003e73\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c5\"\u003e \u003cp\u003e239\u003c/p\u003e \u003c/td\u003e \u003c/tr\u003e \u003ctr\u003e \u003ctd align=\"left\" colname=\"c1\"\u003e \u003cp\u003eAge (IQR)\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c2\"\u003e \u003cp\u003e2.1 (1, 3.6)\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c3\"\u003e \u003cp\u003e2.1 (1.3, 3.4)\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c4\"\u003e \u003cp\u003e2.3 (1.1, 4)\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c5\"\u003e \u003cp\u003e2.2 (1.1, 3.6)\u003c/p\u003e \u003c/td\u003e \u003c/tr\u003e \u003ctr\u003e \u003ctd align=\"left\" colname=\"c1\"\u003e \u003cp\u003e% Male\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c2\"\u003e \u003cp\u003e48.3\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c3\"\u003e \u003cp\u003e49.7\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c4\"\u003e \u003cp\u003e54.4\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c5\"\u003e \u003cp\u003e50.6\u003c/p\u003e \u003c/td\u003e \u003c/tr\u003e \u003ctr\u003e \u003ctd align=\"left\" colname=\"c1\"\u003e \u003cp\u003eMedian days followed (IQR)\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c2\"\u003e \u003cp\u003e593 (346, 1079)\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c3\"\u003e \u003cp\u003e755 (490, 1151)\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c4\"\u003e \u003cp\u003e589 (246, 993)\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c5\"\u003e \u003cp\u003e651 (324.5, 1079)\u003c/p\u003e \u003c/td\u003e \u003c/tr\u003e \u003ctr\u003e \u003ctd align=\"left\" colname=\"c1\"\u003e \u003cp\u003eMedian malaria cases per person (IQR)\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c2\"\u003e \u003cp\u003e0 (0, 1)\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c3\"\u003e \u003cp\u003e2 (0, 5)\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c4\"\u003e \u003cp\u003e5 (1, 9)\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c5\"\u003e \u003cp\u003e1 (0, 5)\u003c/p\u003e \u003c/td\u003e \u003c/tr\u003e \u003ctr\u003e \u003ctd align=\"left\" colname=\"c1\"\u003e \u003cp\u003eMedian malaria cases per person-year (IQR)\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c2\"\u003e \u003cp\u003e0 (0, 0.481)\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c3\"\u003e \u003cp\u003e1.14 (0, 2.56)\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c4\"\u003e \u003cp\u003e4.07 (2.14, 7.41)\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c5\"\u003e \u003cp\u003e0.96 (0, 3.43)\u003c/p\u003e \u003c/td\u003e \u003c/tr\u003e \u003ctr\u003e \u003ctd align=\"left\" colname=\"c1\"\u003e \u003cp\u003eMedian proportion of nights using LLIN (IQR)\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c2\"\u003e \u003cp\u003e1 (1, 1)\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c3\"\u003e \u003cp\u003e1 (1, 1)\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c4\"\u003e \u003cp\u003e1 (1, 1)\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c5\"\u003e \u003cp\u003e1 (1, 1)\u003c/p\u003e \u003c/td\u003e \u003c/tr\u003e \u003c/tbody\u003e \u003c/colgroup\u003e \u003c/table\u003e\u003c/div\u003e \u003c/p\u003e \u003cp\u003e \u003c/p\u003e \u003c/div\u003e\n\u003ch3\u003eVariation in EIR over time and space\u003c/h3\u003e\n\u003cp\u003eMonthly entomological measures over the study period are summarized in Table\u0026nbsp;\u003cspan refid=\"Tab2\" class=\"InternalRef\"\u003e2\u003c/span\u003e, with monthly subcounty-wide trends in \u003cem\u003eAnopheles\u003c/em\u003e counts, sporozoite rate, crude annualized EIR (aEIR) and malaria incidence among cohort participants shown in Fig.\u0026nbsp;\u003cspan refid=\"Fig1\" class=\"InternalRef\"\u003e1\u003c/span\u003e. As expected, the highest vector densities and sporozoite rates were recorded in Tororo, the second-highest in Kanungu, and the lowest in Jinja. Site-specific aEIRs calculated over the course of the study consequently followed the same pattern: 233 in Tororo, 20.3 in Kanungu, and 2.72 in Jinja.\u003c/p\u003e \u003cp\u003e \u003cdiv class=\"gridtable\"\u003e\u003ctable float=\"Yes\" id=\"Tab2\" border=\"1\"\u003e \u003ccaption language=\"En\"\u003e \u003cdiv class=\"CaptionNumber\"\u003eTable 2\u003c/div\u003e \u003cdiv class=\"CaptionContent\"\u003e \u003cp\u003e\u0026ndash; Summary statistics for monthly entomological measures. To generate aggregated measures, collection-level measurements were summed by household (\u0026ldquo;HH\u0026rdquo;) or site and divided by number of collections. Collection-level measures, which constituted the response variable of the smoothed entomological models used in our final analysis, are included to illustrate the zero inflation described in the main text.\u003c/p\u003e \u003c/div\u003e \u003c/caption\u003e \u003ccolgroup cols=\"13\"\u003e \u003cdiv align=\"left\" class=\"colspec\" colname=\"c1\" colnum=\"1\"\u003e\u003c/div\u003e \u003cdiv align=\"left\" class=\"colspec\" colname=\"c2\" colnum=\"2\"\u003e\u003c/div\u003e \u003cdiv align=\"left\" class=\"colspec\" colname=\"c3\" colnum=\"3\"\u003e\u003c/div\u003e \u003cdiv align=\"left\" class=\"colspec\" colname=\"c4\" colnum=\"4\"\u003e\u003c/div\u003e \u003cdiv align=\"left\" class=\"colspec\" colname=\"c5\" colnum=\"5\"\u003e\u003c/div\u003e \u003cdiv align=\"left\" class=\"colspec\" colname=\"c6\" colnum=\"6\"\u003e\u003c/div\u003e \u003cdiv align=\"left\" class=\"colspec\" colname=\"c7\" colnum=\"7\"\u003e\u003c/div\u003e \u003cdiv align=\"left\" class=\"colspec\" colname=\"c8\" colnum=\"8\"\u003e\u003c/div\u003e \u003cdiv align=\"left\" class=\"colspec\" colname=\"c9\" colnum=\"9\"\u003e\u003c/div\u003e \u003cdiv align=\"left\" class=\"colspec\" colname=\"c10\" colnum=\"10\"\u003e\u003c/div\u003e \u003cdiv align=\"left\" class=\"colspec\" colname=\"c11\" colnum=\"11\"\u003e\u003c/div\u003e \u003cdiv align=\"left\" class=\"colspec\" colname=\"c12\" colnum=\"12\"\u003e\u003c/div\u003e \u003cdiv align=\"left\" class=\"colspec\" colname=\"c13\" colnum=\"13\"\u003e\u003c/div\u003e \u003cthead\u003e \u003ctr\u003e \u003cth align=\"left\" colname=\"c1\" morerows=\"1\" rowspan=\"2\"\u003e\u0026nbsp;\u003c/th\u003e \u003cth align=\"left\" colspan=\"3\" nameend=\"c4\" namest=\"c2\"\u003e \u003cp\u003eJinja\u003c/p\u003e \u003c/th\u003e \u003cth align=\"left\" colspan=\"3\" nameend=\"c7\" namest=\"c5\"\u003e \u003cp\u003eKanungu\u003c/p\u003e \u003c/th\u003e \u003cth align=\"left\" colspan=\"3\" nameend=\"c10\" namest=\"c8\"\u003e \u003cp\u003eTororo\u003c/p\u003e \u003c/th\u003e \u003cth align=\"left\" colspan=\"3\" nameend=\"c13\" namest=\"c11\"\u003e \u003cp\u003eOverall\u003c/p\u003e \u003c/th\u003e \u003c/tr\u003e \u003ctr\u003e \u003cth align=\"left\" colname=\"c2\"\u003e \u003cp\u003eCollection median (IQR)\u003c/p\u003e \u003c/th\u003e \u003cth align=\"left\" colname=\"c3\"\u003e \u003cp\u003eHH median (IQR)\u003c/p\u003e \u003c/th\u003e \u003cth align=\"left\" colname=\"c4\"\u003e \u003cp\u003eSite\u003c/p\u003e \u003c/th\u003e \u003cth align=\"left\" colname=\"c5\"\u003e \u003cp\u003eCollection median (IQR)\u003c/p\u003e \u003c/th\u003e \u003cth align=\"left\" colname=\"c6\"\u003e \u003cp\u003eHH median (IQR)\u003c/p\u003e \u003c/th\u003e \u003cth align=\"left\" colname=\"c7\"\u003e \u003cp\u003eSite\u003c/p\u003e \u003c/th\u003e \u003cth align=\"left\" colname=\"c8\"\u003e \u003cp\u003eCollection median (IQR)\u003c/p\u003e \u003c/th\u003e \u003cth align=\"left\" colname=\"c9\"\u003e \u003cp\u003eHH median (IQR)\u003c/p\u003e \u003c/th\u003e \u003cth align=\"left\" colname=\"c10\"\u003e \u003cp\u003eSite\u003c/p\u003e \u003c/th\u003e \u003cth align=\"left\" colname=\"c11\"\u003e \u003cp\u003eCollection median (IQR)\u003c/p\u003e \u003c/th\u003e \u003cth align=\"left\" colname=\"c12\"\u003e \u003cp\u003eHH median (IQR)\u003c/p\u003e \u003c/th\u003e \u003cth align=\"left\" colname=\"c13\"\u003e \u003cp\u003eSite\u003c/p\u003e \u003c/th\u003e \u003c/tr\u003e \u003c/thead\u003e \u003ctbody\u003e \u003ctr\u003e \u003ctd align=\"left\" colname=\"c1\"\u003e \u003cp\u003eVector count\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c2\"\u003e \u003cp\u003e0 (0,1)\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c3\"\u003e \u003cp\u003e0.429 (0.281,1)\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c4\"\u003e \u003cp\u003e0.89\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c5\"\u003e \u003cp\u003e0 (0,2)\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c6\"\u003e \u003cp\u003e1.4 (0.482,4.58)\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c7\"\u003e \u003cp\u003e3.41\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c8\"\u003e \u003cp\u003e10 (2, 39)\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c9\"\u003e \u003cp\u003e28.6 (19,41.8)\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c10\"\u003e \u003cp\u003e33.5\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c11\"\u003e \u003cp\u003e0 (0, 4)\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c12\"\u003e \u003cp\u003e2.51 (0.441, 19.8)\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c13\"\u003e \u003cp\u003e10.5\u003c/p\u003e \u003c/td\u003e \u003c/tr\u003e \u003ctr\u003e \u003ctd align=\"left\" colname=\"c1\"\u003e \u003cp\u003eSR\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c2\"\u003e \u003cp\u003e0 (0,0)\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c3\"\u003e \u003cp\u003e0 (0,0.00147)\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c4\"\u003e \u003cp\u003e0.00837\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c5\"\u003e \u003cp\u003e0 (0,0)\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c6\"\u003e \u003cp\u003e0.00915 (0,0.0222)\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c7\"\u003e \u003cp\u003e0.0163\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c8\"\u003e \u003cp\u003e0 (0, 0)\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c9\"\u003e \u003cp\u003e0.0172 (0.0138,0.0236)\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c10\"\u003e \u003cp\u003e0.019\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c11\"\u003e \u003cp\u003e0 (0, 0)\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c12\"\u003e \u003cp\u003e0.0111 (0, 0.0208)\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c13\"\u003e \u003cp\u003e0.018\u003c/p\u003e \u003c/td\u003e \u003c/tr\u003e \u003ctr\u003e \u003ctd align=\"left\" colname=\"c1\"\u003e \u003cp\u003eaEIR\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c2\"\u003e \u003cp\u003e0 (0,0)\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c3\"\u003e \u003cp\u003e0 (0,3.2)\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c4\"\u003e \u003cp\u003e2.72\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c5\"\u003e \u003cp\u003e0 (0,0)\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c6\"\u003e \u003cp\u003e8.15 (0,27)\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c7\"\u003e \u003cp\u003e20.3\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c8\"\u003e \u003cp\u003e0 (0,0)\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c9\"\u003e \u003cp\u003e178 (108,307)\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c10\"\u003e \u003cp\u003e233\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c11\"\u003e \u003cp\u003e0 (0, 0)\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c12\"\u003e \u003cp\u003e11.1 (0, 111)\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c13\"\u003e \u003cp\u003e69.3\u003c/p\u003e \u003c/td\u003e \u003c/tr\u003e \u003ctr\u003e \u003ctd align=\"left\" colname=\"c1\"\u003e \u003cp\u003eNo. collections\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c2\"\u003e \u003cp\u003eNA\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c3\"\u003e \u003cp\u003e57 (33,57)\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c4\"\u003e \u003cp\u003e5210\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c5\"\u003e \u003cp\u003eNA\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c6\"\u003e \u003cp\u003e56 (56,57)\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c7\"\u003e \u003cp\u003e5410\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c8\"\u003e \u003cp\u003eNA\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c9\"\u003e \u003cp\u003e40 (39,40)\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c10\"\u003e \u003cp\u003e3860\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c11\"\u003e \u003cp\u003eNA\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c12\"\u003e \u003cp\u003e45 (39,57)\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c13\"\u003e \u003cp\u003e14500\u003c/p\u003e \u003c/td\u003e \u003c/tr\u003e \u003c/tbody\u003e \u003c/colgroup\u003e \u003c/table\u003e\u003c/div\u003e \u003c/p\u003e \u003cp\u003eThe number of mosquitoes collected during any single collection was consistently low in Jinja and varied little over the study: the median number of mosquitoes collected per household (averaged over the entire study period) was 0.429 (IQR 0.281-1), but the median number at any given collection was 0 (IQR 0\u0026ndash;1), and 0 mosquitoes were collected in 3697/5212 (70.9%) collections. Counts varied more widely at the two other sites: in Kanungu, the median per household was 1.4 (0.482\u0026ndash;4.58) and median per collection 0 (0\u0026ndash;2), with 0 mosquitoes collected in 2977/5414 (55.0%) collections; in Tororo, the median per household was 28.6 (19-41.8) and median per collection 10 (2\u0026ndash;39), with 0 mosquitoes collected in 656/3858 (17.0%) collections.\u003c/p\u003e \u003cp\u003eMedian sporozoite rates by household were less variable: 0 (0\u0026ndash;0) in Jinja, 0.00915 (0-0.0222) in Kanungu, and 0.0172 (0.0138\u0026ndash;0.0236) in Tororo. Median sporozoite rate per collection was 0 (0\u0026ndash;0) for all sites, as sporozoite rates were equal to 0 in the majority of captures in all three: 5175/5212 (99.3%) of captures in Jinja, 5209\u0026ndash;5414 (96.2%) in Kanungu, and 2955/3858 (76.6%) in Tororo.\u003c/p\u003e \u003cp\u003eThe observed sparseness and variability prompted us to consider multiple spatiotemporally smoothed models of vector counts and sporozoite rates. For vector counts, the best-fitting models explained a moderate percentage of the deviance at all sites (Jinja: 45.0; Kanungu: 60.6; Tororo: 57.4) (Supplementary Table\u0026nbsp;1). For sporozoite rates, the best-fitting models explained a small percentage of the deviance, particularly in Jinja (Jinja: 1.83; Kanungu: 11.1; Tororo: 11.8) (Supplementary Table\u0026nbsp;1).\u003c/p\u003e \u003cp\u003ePredictions generated from the best vector density and sporozoite rate models are overlaid on Figs.\u0026nbsp;\u003cspan refid=\"Fig1\" class=\"InternalRef\"\u003e1\u003c/span\u003ea-c. At all sites, we observed biannual peaks in vector counts corresponding roughly to the March-May and August-October rainy seasons. Increases in sporozoite rates corresponded to decreases in vector count, as expected in an aging mosquito population. Vector counts were higher at the eastern border of the study site in Jinja, nearer to Lake Victoria, and at the northern border in Kanungu, where altitude was lower, but were patchy in Tororo. Sporozoite rates were higher in the southwest in Jinja, patchy in Kanungu, and lacked notable spatial structure in Tororo (Supplementary Fig.\u0026nbsp;1). Temporal and spatial trends in EIR were similar to those for vector counts.\u003c/p\u003e\n\u003ch3\u003eAssociation between entomological metrics and malaria incidence\u003c/h3\u003e\n\u003cp\u003eWe next evaluated the association between aEIR and malaria incidence. Figure\u0026nbsp;\u003cspan refid=\"Fig2\" class=\"InternalRef\"\u003e2\u003c/span\u003e shows the association between average household aEIR and the average incidence of malaria experienced by individuals over the course of the study, for both crude (2a) and modeled (2b) aEIRs. Although analyzing data from the three sites together suggests a positive relationship between aEIR and malaria incidence, much of this association might be attributed to between-site differences, since both aEIR and incidence were lowest in Jinja, moderate in Kanungu and highest in Tororo.\u003c/p\u003e \u003cp\u003e \u003c/p\u003e \u003cp\u003eTo investigate whether the relationship observed when pooling data from all sites applied within sites, we fit models of associations between modeled aEIR and incidence separately for each site (Fig.\u0026nbsp;\u003cspan refid=\"Fig3\" class=\"InternalRef\"\u003e3\u003c/span\u003e). We found a positive association between modeled aEIR and incidence in Kanungu, but not in Jinja and Tororo. Treating the association between log\u003csub\u003e2\u003c/sub\u003e-transformed spatiotemporally modeled aEIR and incidence as linear on the log scale, we estimated that, on average, malaria incidence increased by 9% (IRR 1.09, 95% interval 1.04\u0026ndash;1.14) in Kanungu with each doubling of EIR. At both Jinja and Tororo, the 95% credible interval of predicted IRRs for modeled aEIR crossed 1 (Jinja: mean 1.02, 95% interval 0.774\u0026ndash;1.26; Tororo: 1.02, 0.986\u0026ndash;1.06). Overall, these results were qualitatively similar to trends suggested by average aEIR-incidence plots. In both Kanungu and Tororo, incidence additionally increased with age among younger children before saturating at older ages. No significant relationship was seen in Jinja (Supplementary Fig.\u0026nbsp;2).\u003c/p\u003e \u003cp\u003e \u003c/p\u003e \u003cp\u003eWe also investigated whether alternative estimates of aEIR could better capture the variance in incidence. We fit incidence models to crude aEIRs, aEIRs generated from temporal vector count and sporozoite rate smooths (ignoring space), and aEIRs generated from spatial vector count and sporozoite rate smooths (ignoring time), comparing the results by AIC (Supplementary Table\u0026nbsp;2). The spatiotemporal smooths were the best- or second best-performing models for all sites: in Jinja, the best-performing model by AIC was fit to spatially smoothed aEIRs that ignored time; in Kanungu, models fit to spatiotemporally smoothed aEIRs and temporally smoothed aEIRs that ignored space performed equally well; in Tororo, the best-performing model was fit to temporally smoothed aEIRs. Models fit to crude aEIRs did not converge in Jinja and were the worst-performing by AIC at the other two sites. Regardless of the aEIR estimate used, all models were consistent with a positive association in Kanungu (with the exception of the poorly fitting spatial-only model) but showed no or little association in Tororo and Jinja (Supplementary Figs.\u0026nbsp;3\u0026ndash;4). All models explained a relatively small percentage of the deviance: in Jinja 16.8\u0026ndash;21.0%; in Kanungu 9.56\u0026ndash;11.6%; and in Tororo 5.48\u0026ndash;6.60% (Supplementary Table\u0026nbsp;2).\u003c/p\u003e"},{"header":"Discussion","content":"\u003cp\u003eModeling based on data from clinical surveillance and entomological measures derived from light trap captures, we found that relationships between household aEIR and individual malaria incidence varied between three study sites in Uganda where transmission varies 100-fold. Overall results suggested a positive association between aEIR and malaria risk, but analyses stratified by site showed that only in Kanungu, an intermediate-EIR and -malaria incidence site, higher household aEIRs were associated with increasing individual malaria incidence. In Jinja (low EIR and low incidence) and Tororo (high EIR and high incidence), associations were weak or absent. There are several possible explanations for this potentially counterintuitive result.\u003c/p\u003e \u003cp\u003eFirst, previous studies have suggested an underlying sigmoid relationship between EIR and malaria risk [5, 7], which is intuitively appealing: regardless of the exact shape of the EIR-incidence association, there is a maximum number of malaria episodes an individual can experience in a year. In keeping with this pattern, the positive association between EIR and incidence observed at our intermediate-EIR site might correspond to the steep part of an EIR-incidence curve, while the absent associations at the low- and high-EIR sites might correspond to the behavior of the curve near its minimum and carrying capacity, respectively. This explanation revives the possibility of a general relationship between entomological measures and disease incidence, though it would not account for the lack of overlap between the expected incidences at the three sites as seen in Fig.\u0026nbsp;\u003cspan refid=\"Fig3\" class=\"InternalRef\"\u003e3\u003c/span\u003e.\u003c/p\u003e \u003cp\u003eSecond, it is possible that features unique to Kanungu allowed us to detect an association between entomological measures and disease incidence. Kanungu is the only of the three sites with an altitude gradient. This geographical feature likely contributes both to a broader within-site aEIR range, across which an association between entomological metrics and incidence may be more easily captured, and to stronger seasonal and spatial trends of mosquito exposure that more closely correlate with corresponding trends in clinical malaria incidence. In this setting, information borrowed from nearby houses and dates would be more informative than at other sites, possibly decreasing variance in our estimates.\u003c/p\u003e \u003cp\u003eThird, it is clear from the small proportion of variance explained by our incidence models that there were important drivers of malaria incidence not reflected by entomological surveillance data. Potential candidates include variations in human behavior, immunity, and mosquito feeding patterns.\u003c/p\u003e \u003cp\u003eDifferences in behavior may mean that household aEIR measured in Jinja and Tororo did not reflect household members\u0026rsquo; exposure to infected mosquitoes. Imported cases are one potentially important contributor to local malaria incidence: recent overnight travel has been associated with increased malaria risk at all three sites, but was found to be more common in Jinja and Tororo than in Kanungu [9]. LLIN adherence may similarly decouple household EIR from incidence, although bednet distribution was previously shown to have only a modest effect on malaria risk at the study sites despite high reported rates of adherence [10].\u003c/p\u003e \u003cp\u003eAnti-parasite and/or anti-disease immunity is also likely to drive incidence patterns, particularly in the high-transmission setting of Tororo. While we attempted to minimize the impact of such immunity by restricting the age of the study population to those under five and controlling for the effect of age in the final analysis, these approaches are imperfect. The nonlinear relationship we recovered between age and malaria incidence in Kanungu and Tororo is consistent with prior analyses of children under five years in this cohort [11]. Taking advantage of molecular methods to identify incident infections, rather than incident disease, may address some of these concerns by accounting for asymptomatic infections, including superinfections [12].\u003c/p\u003e \u003cp\u003eFinally, local mosquito feeding behavior may also have differed in ways not captured, or captured differentially by site, in our surveillance data, including variation in biting time and location relative to human behaviors. These differences could conceivably stem from environmental heterogeneity, or from differences in species composition within the \u003cem\u003eAn. gambiae\u003c/em\u003e s.l. species complex. Both \u003cem\u003eAn. gambiae\u003c/em\u003e sensu stricto and \u003cem\u003eAn. arabiensis\u003c/em\u003e are endemic to the study sites, with \u003cem\u003eAn. arabiensis\u003c/em\u003e exhibiting less anthropophilic and endophagic tendencies [13, 14] and a recent temporal association with lower malaria risk in Tororo [15].\u003c/p\u003e \u003cp\u003eThe lack of association between entomological and clinical metrics at two of our sites may also reflect the imprecision of entomological data derived from a single CDC LT per household-month, with sporozoite rates estimated from an even smaller subset. Whether CDC LT data themselves are problematic is unclear: EIRs derived from CDC LT data have been comparable with gold standard human-landing catches both in prior analyses of PRISM 1 cohort data and in subsequent entomological studies conducted in Tororo [16, 17], but other studies comparing CDC LT and human-landing catches of anophelines have noted significant differences in overall vector densities, species composition, sporozoite rates, and parous rates [18\u0026ndash;21].\u003c/p\u003e \u003cp\u003eTo summarize, our study\u0026rsquo;s use of concurrent longitudinal spatiotemporal entomological and clinical data across a wide range of transmission intensities afforded us a uniquely detailed view of the relationship between these two markers of exposure. Nevertheless, the relative lack of overlap in EIRs between the three sites limited our ability to distinguish between potential explanations for the EIR-incidence pattern we observed: either a general, potentially sigmoid, EIR-incidence relationship, or site-specific differences in exposure patterns, host immunity, and/or vector characteristics. Limitations of our study included a limited ability to characterize direct mosquito-human exposure as would be afforded by simultaneous human behavioral observations and human-landing catches, a focus on malaria incidence rather than incident \u003cem\u003eP. falciparum\u003c/em\u003e infection, and a lack of sub-species complex mosquito species identification that might have obscured significant differences in the relative roles played by vector species. Entomological data collections are inherently noisy and sparse relative to the exposure patterns they are meant to reflect, and a fine-scale EIR-incidence association may not exist except in the highly favorable setting offered by a site like Kanungu.\u003c/p\u003e"},{"header":"Conclusions","content":"\u003cp\u003eIn conclusion, despite strong theoretical support for a general relationship between the aEIR and malaria incidence, household-level EIRs estimated from smoothed mosquito surveillance data were significantly associated with individual malaria incidences in an intermediate-transmission site, but not at low- or high-transmission sites. Further assessment of this relationship using data collected at a finer temporal scale with molecular identification of new infections may be helpful to tease apart this heterogeneity.\u003c/p\u003e"},{"header":"Abbreviations","content":"\u003cdiv class=\"DefinitionList\"\u003e \u003cdiv class=\"DefinitionListEntry\"\u003e \u003cdiv class=\"Term\"\u003eaEIR\u003c/div\u003e \u003cdiv class=\"Description\"\u003e \u003cp\u003eannualized entomological inoculation rate\u003c/p\u003e \u003c/div\u003e \u003c/div\u003e \u003cdiv class=\"DefinitionListEntry\"\u003e \u003cdiv class=\"Term\"\u003eAIC\u003c/div\u003e \u003cdiv class=\"Description\"\u003e \u003cp\u003eAkaike information criterion\u003c/p\u003e \u003c/div\u003e \u003c/div\u003e \u003cdiv class=\"DefinitionListEntry\"\u003e \u003cdiv class=\"Term\"\u003eCDC LT\u003c/div\u003e \u003cdiv class=\"Description\"\u003e \u003cp\u003eCDC light trap\u003c/p\u003e \u003c/div\u003e \u003c/div\u003e \u003cdiv class=\"DefinitionListEntry\"\u003e \u003cdiv class=\"Term\"\u003eEIR\u003c/div\u003e \u003cdiv class=\"Description\"\u003e \u003cp\u003eentomological inoculation rate\u003c/p\u003e \u003c/div\u003e \u003c/div\u003e \u003cdiv class=\"DefinitionListEntry\"\u003e \u003cdiv class=\"Term\"\u003eELISA\u003c/div\u003e \u003cdiv class=\"Description\"\u003e \u003cp\u003eenzyme-linked immunosorbent assay\u003c/p\u003e \u003c/div\u003e \u003c/div\u003e \u003cdiv class=\"DefinitionListEntry\"\u003e \u003cdiv class=\"Term\"\u003eGAM\u003c/div\u003e \u003cdiv class=\"Description\"\u003e \u003cp\u003egeneralized additive model\u003c/p\u003e \u003c/div\u003e \u003c/div\u003e \u003cdiv class=\"DefinitionListEntry\"\u003e \u003cdiv class=\"Term\"\u003eGAMM\u003c/div\u003e \u003cdiv class=\"Description\"\u003e \u003cp\u003egeneralized additive mixed model\u003c/p\u003e \u003c/div\u003e \u003c/div\u003e \u003cdiv class=\"DefinitionListEntry\"\u003e \u003cdiv class=\"Term\"\u003eIQR\u003c/div\u003e \u003cdiv class=\"Description\"\u003e \u003cp\u003einterquartile range\u003c/p\u003e \u003c/div\u003e \u003c/div\u003e \u003cdiv class=\"DefinitionListEntry\"\u003e \u003cdiv class=\"Term\"\u003eIRR\u003c/div\u003e \u003cdiv class=\"Description\"\u003e \u003cp\u003eincidence rate ratio\u003c/p\u003e \u003c/div\u003e \u003c/div\u003e \u003cdiv class=\"DefinitionListEntry\"\u003e \u003cdiv class=\"Term\"\u003eIRS\u003c/div\u003e \u003cdiv class=\"Description\"\u003e \u003cp\u003eindoor residual spraying\u003c/p\u003e \u003c/div\u003e \u003c/div\u003e \u003cdiv class=\"DefinitionListEntry\"\u003e \u003cdiv class=\"Term\"\u003eLLIN\u003c/div\u003e \u003cdiv class=\"Description\"\u003e \u003cp\u003elong-lasting insecticidal net\u003c/p\u003e \u003c/div\u003e \u003c/div\u003e \u003cdiv class=\"DefinitionListEntry\"\u003e \u003cdiv class=\"Term\"\u003ePRISM\u003c/div\u003e \u003cdiv class=\"Description\"\u003e \u003cp\u003eProgram for Resistance, Immunology, Surveillance and Modeling of Malaria\u003c/p\u003e \u003c/div\u003e \u003c/div\u003e \u003c/div\u003e"},{"header":"Declarations","content":"\u003cp\u003eEthics approval and consent to participate: Ethical approval was obtained from the Makerere University School of Medicine Research and Ethics Committee, the Uganda National Council for Science and Technology, the London School of Hygiene and Tropical Medicine Ethics Committee, the School of Biological and Biomedical Sciences Ethics Committee, Durham University and the University of California, San Francisco Committee on Human Research.\u003c/p\u003e\n\u003cp\u003eConsent for publication: Not applicable\u003c/p\u003e\n\u003cp\u003eAvailability of data and materials: The datasets used and/or analyzed during the current study are publicly available at ClinEpiDB: https://clinepidb.org/ce/app/workspace/analyses/DS_0ad509829e\u003c/p\u003e\n\u003cp\u003eCompeting interests: The authors have no competing interests to declare.\u0026nbsp;\u003c/p\u003e\n\u003cp\u003eFunding: Funding was provided by the National Institutes of Health as part of the International Centers of Excellence in Malaria Program (2U19AI089674) and the UCSF Biology of Infectious Diseases Training Program (2T32AI007641-21).\u003c/p\u003e\n\u003cp\u003eAuthors’ contributions:\u003c/p\u003e\n\u003cp\u003eConceptualization – MM, IRB, BG\u003c/p\u003e\n\u003cp\u003eData curation – KM, AM\u003c/p\u003e\n\u003cp\u003eFormal analysis – MM\u003c/p\u003e\n\u003cp\u003eFunding acquisition – PJR, MRK, GD, BG\u003c/p\u003e\n\u003cp\u003eInvestigation – EA, IRB, MRK, PJR, GD, JN, KM\u003c/p\u003e\n\u003cp\u003eMethodology – MM, IRB\u003c/p\u003e\n\u003cp\u003eProject administration – EA, JN, MRK, GD\u003c/p\u003e\n\u003cp\u003eWriting (original draft) - MM\u003c/p\u003e\n\u003cp\u003eWriting (review and editing) – MM, EA, MRK, PJR, JN, KM, AM, GD, BG, IRB\u003c/p\u003e\n\u003cp\u003eAcknowledgements: We would like to thank the study participants and their families, as well as the study team, Makerere University-UCSF Research Collaboration, and Infectious Diseases Research Collaboration. Portions of this work were performed on the Wynton HPC Co-Op cluster, which is supported by UCSF research faculty and UCSF institutional funds.\u003c/p\u003e"},{"header":"References","content":"\u003col\u003e\n\u003cli\u003eWorld Health Organization: \u003cstrong\u003eMalaria surveillance, monitoring and evaluation: a reference manual.\u003c/strong\u003e 2018.\u003c/li\u003e\n\u003cli\u003eShaukat AM, Breman JG, McKenzie FE: \u003cstrong\u003eUsing the entomological inoculation rate to assess the impact of vector control on malaria parasite transmission and elimination.\u003c/strong\u003e \u003cem\u003eMalaria journal \u003c/em\u003e2010, \u003cstrong\u003e9:\u003c/strong\u003e1-12.\u003c/li\u003e\n\u003cli\u003eGarrett-Jones C, Ferreira Neto JA, Organization WH: \u003cstrong\u003eThe prognosis for interruption of malaria transmission through assessment of the mosquito\u0026apos;s vectorial capacity.\u003c/strong\u003e World Health Organization; 1964.\u003c/li\u003e\n\u003cli\u003eRoss RS: \u003cem\u003eThe prevention of malaria.\u003c/em\u003e England1911.\u003c/li\u003e\n\u003cli\u003eSmith D, Dushoff J, Snow R, Hay S: \u003cstrong\u003eThe entomological inoculation rate and Plasmodium falciparum infection in African children.\u003c/strong\u003e \u003cem\u003eNature \u003c/em\u003e2005, \u003cstrong\u003e438:\u003c/strong\u003e492-495.\u003c/li\u003e\n\u003cli\u003eBeier JC, Killeen GF, Githure JI: \u003cstrong\u003eentomologic inoculation rates and Plasmodium falciparum malaria prevalence in Africa.\u003c/strong\u003e \u003cem\u003eThe American journal of tropical medicine and hygiene \u003c/em\u003e1999, \u003cstrong\u003e61:\u003c/strong\u003e109-113.\u003c/li\u003e\n\u003cli\u003eCharlwood J, Smith T, Lyimo E, Kitua A, Masanja H, Booth M, Alonso P, Tanner M: \u003cstrong\u003eIncidence of Plasmodium falciparum infection in infants in relation to exposure to sporozoite-infected anophelines.\u003c/strong\u003e \u003cem\u003eThe American journal of tropical medicine and hygiene \u003c/em\u003e1998, \u003cstrong\u003e59:\u003c/strong\u003e243-251.\u003c/li\u003e\n\u003cli\u003eBeier JC, Oster CN, Onyango FK, Bales JD, Sherwood JA, Perkins PV, Chumo DK, Koech DV, Whitmire RE, Roberts CR: \u003cstrong\u003ePlasmodium falciparum incidence relative to entomologic inoculation rates at a site proposed for testing malaria vaccines in western Kenya.\u003c/strong\u003e \u003cem\u003eThe American journal of tropical medicine and hygiene \u003c/em\u003e1994, \u003cstrong\u003e50:\u003c/strong\u003e529-536.\u003c/li\u003e\n\u003cli\u003eArinaitwe E, Dorsey G, Nankabirwa JI, Kigozi SP, Katureebe A, Kakande E, Rek J, Rosenthal PJ, Drakeley C, Kamya MR: \u003cstrong\u003eAssociation between recent overnight travel and risk of malaria: a prospective cohort study at 3 sites in Uganda.\u003c/strong\u003e \u003cem\u003eClinical Infectious Diseases \u003c/em\u003e2019, \u003cstrong\u003e68:\u003c/strong\u003e313-320.\u003c/li\u003e\n\u003cli\u003eKatureebe A, Zinszer K, Arinaitwe E, Rek J, Kakande E, Charland K, Kigozi R, Kilama M, Nankabirwa J, Yeka A: \u003cstrong\u003eMeasures of malaria burden after long-lasting insecticidal net distribution and indoor residual spraying at three sites in Uganda: a prospective observational study.\u003c/strong\u003e \u003cem\u003ePLoS medicine \u003c/em\u003e2016, \u003cstrong\u003e13:\u003c/strong\u003ee1002167.\u003c/li\u003e\n\u003cli\u003eRodriguez-Barraquer I, Arinaitwe E, Jagannathan P, Kamya MR, Rosenthal PJ, Rek J, Dorsey G, Nankabirwa J, Staedke SG, Kilama M: \u003cstrong\u003eQuantification of anti-parasite and anti-disease immunity to malaria as a function of age and exposure.\u003c/strong\u003e \u003cem\u003eElife \u003c/em\u003e2018, \u003cstrong\u003e7:\u003c/strong\u003ee35832.\u003c/li\u003e\n\u003cli\u003eBriggs J, Teyssier N, Nankabirwa JI, Rek J, Jagannathan P, Arinaitwe E, Bousema T, Drakeley C, Murray M, Crawford E: \u003cstrong\u003eSex-based differences in clearance of chronic Plasmodium falciparum infection.\u003c/strong\u003e \u003cem\u003eElife \u003c/em\u003e2020, \u003cstrong\u003e9:\u003c/strong\u003ee59872.\u003c/li\u003e\n\u003cli\u003eMahande A, Mosha F, Mahande J, Kweka E: \u003cstrong\u003eFeeding and resting behaviour of malaria vector, Anopheles arabiensis with reference to zooprophylaxis.\u003c/strong\u003e \u003cem\u003eMalaria journal \u003c/em\u003e2007, \u003cstrong\u003e6:\u003c/strong\u003e1-6.\u003c/li\u003e\n\u003cli\u003eTirados I, Costantini C, Gibson G, Torr SJ: \u003cstrong\u003eBlood-feeding behaviour of the malarial mosquito Anopheles arabiensis: implications for vector control.\u003c/strong\u003e \u003cem\u003eMed Vet Entomol \u003c/em\u003e2006, \u003cstrong\u003e20:\u003c/strong\u003e425-437.\u003c/li\u003e\n\u003cli\u003eKamya MR, Nankabirwa JI, Arinaitwe E, Rek J, Zedi M, Maiteki-Sebuguzi C, Opigo J, Staedke SG, Oruni A, Donnelly MJ, et al: \u003cstrong\u003eDramatic resurgence of malaria after 7 years of intensive vector control interventions in Eastern Uganda.\u003c/strong\u003e \u003cem\u003ePLOS Glob Public Health \u003c/em\u003e2024, \u003cstrong\u003e4:\u003c/strong\u003ee0003254.\u003c/li\u003e\n\u003cli\u003eMawejje HD, Asiimwe JR, Kyagamba P, Kamya MR, Rosenthal PJ, Lines J, Dorsey G, Staedke SG: \u003cstrong\u003eImpact of different mosquito collection methods on indicators of Anopheles malaria vectors in Uganda.\u003c/strong\u003e \u003cem\u003eMalaria Journal \u003c/em\u003e2022, \u003cstrong\u003e21:\u003c/strong\u003e388.\u003c/li\u003e\n\u003cli\u003eKilama M, Smith DL, Hutchinson R, Kigozi R, Yeka A, Lavoy G, Kamya MR, Staedke SG, Donnelly MJ, Drakeley C: \u003cstrong\u003eEstimating the annual entomological inoculation rate for Plasmodium falciparum transmitted by Anopheles gambiae sl using three sampling methods in three sites in Uganda.\u003c/strong\u003e \u003cem\u003eMalaria journal \u003c/em\u003e2014, \u003cstrong\u003e13:\u003c/strong\u003e1-13.\u003c/li\u003e\n\u003cli\u003eRubio-Palis Y, Curtis C: \u003cstrong\u003eEvaluation of different methods of catching anopheline mosquitoes in western Venezuela.\u003c/strong\u003e \u003cem\u003eJournal of the American Mosquito Control Association \u003c/em\u003e1992, \u003cstrong\u003e8:\u003c/strong\u003e261-267.\u003c/li\u003e\n\u003cli\u003eMbogo CN, Glass GE, Forster D, Kabiru EW, Githure JI, Ouma JH, Beier JC: \u003cstrong\u003eEvaluation of light traps for sampling anopheline mosquitoes in Kilifi, Kenya.\u003c/strong\u003e \u003cem\u003eJ Am Mosq Control Assoc \u003c/em\u003e1993, \u003cstrong\u003e9:\u003c/strong\u003e260-263.\u003c/li\u003e\n\u003cli\u003eHii J, Smith T, Mai A, Ibam E, Alpers M: \u003cstrong\u003eComparison between anopheline mosquitoes (Diptera: Culicidae) caught using different methods in a malaria endemic area of Papua New Guinea.\u003c/strong\u003e \u003cem\u003eBulletin of entomological research \u003c/em\u003e2000, \u003cstrong\u003e90:\u003c/strong\u003e211-219.\u003c/li\u003e\n\u003cli\u003eLines J, Curtis C, Wilkes T, Njunwa K: \u003cstrong\u003eMonitoring human-biting mosquitoes (Diptera: Culicidae) in Tanzania with light-traps hung beside mosquito nets.\u003c/strong\u003e \u003cem\u003eBulletin of entomological research \u003c/em\u003e1991, \u003cstrong\u003e81:\u003c/strong\u003e77-84.\u003c/li\u003e\n\u003cli\u003eUganda Bureau of Statistics - UBOS, ICF Macro: \u003cstrong\u003eUganda Malaria Indicator Survey (MIS) 2009.\u003c/strong\u003e Calverton, Maryland, USA: UBOS and ICF Macro; 2010.\u003c/li\u003e\n\u003cli\u003eKamya MR, Arinaitwe E, Wanzira H, Katureebe A, Barusya C, Kigozi SP, Kilama M, Tatem AJ, Rosenthal PJ, Drakeley C: \u003cstrong\u003eMalaria transmission, infection, and disease at three sites with varied transmission intensity in Uganda: implications for malaria control.\u003c/strong\u003e \u003cem\u003eThe American journal of tropical medicine and hygiene \u003c/em\u003e2015, \u003cstrong\u003e92:\u003c/strong\u003e903.\u003c/li\u003e\n\u003cli\u003eWirtz R, Duncan J, Njelesani E, Schneider I, Brown A, Oster C, Were JB, Webster H: \u003cstrong\u003eELISA method for detecting Plasmodium falciparum circumsporozoite antibody.\u003c/strong\u003e \u003cem\u003eBulletin of the World Health Organization \u003c/em\u003e1989, \u003cstrong\u003e67:\u003c/strong\u003e535.\u003c/li\u003e\n\u003cli\u003eWood SN: \u003cstrong\u003eFast stable restricted maximum likelihood and marginal likelihood estimation of semiparametric generalized linear models.\u003c/strong\u003e \u003cem\u003eJournal of the Royal Statistical Society Series B: Statistical Methodology \u003c/em\u003e2011, \u003cstrong\u003e73:\u003c/strong\u003e3-36.\u003c/li\u003e\n\u003cli\u003eHartig F: \u003cstrong\u003eDHARMa: residual diagnostics for hierarchical (multi-level/mixed) regression models.\u003c/strong\u003e \u003cem\u003eR Packag version 020 \u003c/em\u003e2018.\u003c/li\u003e\n\u003cli\u003eMissinou MA, Lell B, Kremsner PG: \u003cstrong\u003eUncommon asymptomatic Plasmodium falciparum infections in Gabonese children.\u003c/strong\u003e \u003cem\u003eClinical Infectious Diseases \u003c/em\u003e2003, \u003cstrong\u003e36:\u003c/strong\u003e1198-1202.\u003c/li\u003e\n\u003cli\u003eSimpson G, Singmann H: \u003cstrong\u003eR Package: gratia. ggplot\u003c/strong\u003e\u003cstrong\u003e‐based graphics and other useful functions for GAMs fitted using mgcv, 0.1\u003c/strong\u003e\u003cstrong\u003e‐0 (ggplot\u003c/strong\u003e\u003cstrong\u003e‐based graphics and utility functions for working with GAMs fitted using the mgcv package).\u003c/strong\u003e \u003cem\u003eDOI Unavailable \u003c/em\u003e2018.\u003c/li\u003e\n\u003c/ol\u003e"}],"fulltextSource":"","fullText":"","funders":[],"hasAdminPriorityOnWorkflow":false,"hasManuscriptDocX":true,"hasOptedInToPreprint":true,"hasPassedJournalQc":"","hasAnyPriority":false,"hideJournal":false,"highlight":"","institution":"","isAcceptedByJournal":true,"isAuthorSuppliedPdf":false,"isDeskRejected":"","isHiddenFromSearch":false,"isInQc":false,"isInWorkflow":false,"isPdf":false,"isPdfUpToDate":true,"isWithdrawnOrRetracted":false,"journal":{"display":true,"email":"[email protected]","identity":"malaria-journal","isNatureJournal":false,"hasQc":true,"allowDirectSubmit":false,"externalIdentity":"malj","sideBox":"Learn more about [Malaria Journal](http://malariajournal.biomedcentral.com/)","snPcode":"12936","submissionUrl":"https://submission.nature.com/new-submission/12936/3","title":"Malaria Journal","twitterHandle":"@malariajournal","acdcEnabled":true,"dfaEnabled":true,"editorialSystem":"em","reportingPortfolio":"BMC/SO AJ","inReviewEnabled":true,"inReviewRevisionsEnabled":true},"keywords":"","lastPublishedDoi":"10.21203/rs.3.rs-5914493/v1","lastPublishedDoiUrl":"https://doi.org/10.21203/rs.3.rs-5914493/v1","license":{"name":"CC BY 4.0","url":"https://creativecommons.org/licenses/by/4.0/"},"manuscriptAbstract":"\u003ch2\u003eBackground\u003c/h2\u003e \u003cp\u003eThe gold standard measure of malaria exposure is the entomological inoculation rate (EIR), or the number of infectious bites an individual receives over a given period. Nevertheless, it is unclear whether household EIR reflects heterogeneity in individual infection risk.\u003c/p\u003e\u003ch2\u003eMethods\u003c/h2\u003e \u003cp\u003eTo investigate this relationship, we used data collected from a cohort of 439 children aged 0.5-5 years in 239 households from 2011\u0026ndash;2017 in three Ugandan districts: low-EIR Jinja, intermediate-EIR Kanungu and high-EIR Tororo. Participants underwent passive and quarterly active surveillance for clinical malaria, defined as fever with positive thick blood smear. Monthly vector densities and sporozoite rates in participating households were estimated using CDC light traps. We assessed the association between spatiotemporally smoothed household log\u003csub\u003e2\u003c/sub\u003e-transformed EIR and individual malaria incidence using Poisson generalized additive mixed effects models.\u003c/p\u003e\u003ch2\u003eResults\u003c/h2\u003e \u003cp\u003eComparison across sites suggested an increasing relationship between average EIR and malaria incidence. Within-site relationships, however, varied by site, with a positive association in Kanungu (incidence rate ratio [IRR] 1.09, 95% credible interval 1.04\u0026ndash;1.14) but none in Jinja (1.02, 0.774\u0026ndash;1.26) or Tororo (1.02, 0.986\u0026ndash;1.06).\u003c/p\u003e\u003ch2\u003eConclusions\u003c/h2\u003e \u003cp\u003eThese results show the relationship between measured EIR and malaria incidence may depend on local transmission dynamics and be strongest at intermediate EIR, while underscoring the challenges of using household-level measures of exposure.\u003c/p\u003e","manuscriptTitle":"Relating household entomological measures to individual malaria risk","msid":"","msnumber":"","nonDraftVersions":[{"code":1,"date":"2025-02-04 08:48:59","doi":"10.21203/rs.3.rs-5914493/v1","editorialEvents":[{"type":"communityComments","content":0},{"type":"decision","content":"Revision requested","date":"2025-04-28T18:47:08+00:00","index":"","fulltext":""},{"type":"editorInvitedReview","content":"","date":"2025-04-22T19:41:03+00:00","index":"hide","fulltext":""},{"type":"reviewerAgreed","content":"87752883044431979409043093976168930007","date":"2025-04-06T17:01:15+00:00","index":"hide","fulltext":""},{"type":"reviewerAgreed","content":"86446199827879079227346847522253389525","date":"2025-04-02T19:16:01+00:00","index":"hide","fulltext":""},{"type":"editorInvitedReview","content":"","date":"2025-03-21T09:42:11+00:00","index":"hide","fulltext":""},{"type":"reviewerAgreed","content":"70863490493016178275505492386515290180","date":"2025-03-10T05:32:27+00:00","index":"hide","fulltext":""},{"type":"reviewerAgreed","content":"108448283863214960157454830155800807253","date":"2025-02-03T19:18:49+00:00","index":"hide","fulltext":""},{"type":"reviewerAgreed","content":"157963943968925824449479509909575630195","date":"2025-02-01T12:48:29+00:00","index":"hide","fulltext":""},{"type":"reviewersInvited","content":"","date":"2025-01-29T18:06:29+00:00","index":"","fulltext":""},{"type":"editorAssigned","content":"","date":"2025-01-28T10:55:09+00:00","index":"","fulltext":""},{"type":"checksComplete","content":"","date":"2025-01-28T10:54:06+00:00","index":"","fulltext":""},{"type":"submitted","content":"Malaria Journal","date":"2025-01-27T18:37:32+00:00","index":"","fulltext":""}],"status":"published","journal":{"display":true,"email":"[email protected]","identity":"malaria-journal","isNatureJournal":false,"hasQc":true,"allowDirectSubmit":false,"externalIdentity":"malj","sideBox":"Learn more about [Malaria Journal](http://malariajournal.biomedcentral.com/)","snPcode":"12936","submissionUrl":"https://submission.nature.com/new-submission/12936/3","title":"Malaria Journal","twitterHandle":"@malariajournal","acdcEnabled":true,"dfaEnabled":true,"editorialSystem":"em","reportingPortfolio":"BMC/SO AJ","inReviewEnabled":true,"inReviewRevisionsEnabled":true}}],"origin":"","ownerIdentity":"77626321-5a5b-41ee-a0ea-742eec8e1663","owner":[],"postedDate":"February 4th, 2025","published":true,"recentEditorialEvents":[],"rejectedJournal":[],"revision":"","amendment":"","status":"under-review","subjectAreas":[],"tags":[],"updatedAt":"2025-08-06T19:38:10+00:00","versionOfRecord":[],"versionCreatedAt":"2025-02-04 08:48:59","video":"","vorDoi":"","vorDoiUrl":"","workflowStages":[]},"version":"v1","identity":"rs-5914493","journalConfig":"researchsquare"},"__N_SSP":true},"page":"/article/[identity]/[[...version]]","query":{"redirect":"/article/rs-5914493","identity":"rs-5914493","version":["v1"]},"buildId":"8U1c8b4HqxoKbykW_rLl7","isFallback":false,"isExperimentalCompile":false,"dynamicIds":[84888],"gssp":true,"scriptLoader":[]}