As described previously [13, 14], EAG surveys received approval from the Haitian Ministry of Public Health and Population Bioethics Committee (Comité National de Bioéthique), the Institutional Review Boards of Tulane University and the London School of Hygiene and Tropical Medicine. This activity did not constitute engagement in human subjects research as determined by the US Centers for Disease Control and Prevention’s Center for Global Health human subjects office (2016–135a). Consent for school EAGs was collected from school directors and following community meetings to inform parents and allow opt-out. At health facility and church EAGs, adult participants provided consent directly; consent for children (<18 years) was provided by a parent or guardian, and children above 6 years gave assent to participant. Individuals aged 16 or 17 who were married, head of household, or a parent were considered mature minors and consented directly. Thumbprint consent or assent (countersigned by a witness) was used for illiterate participants. Individuals under 6 months of age or requiring immediate medical attention were excluded. A single finger-prick was performed on consenting participants to collect capillary blood for the high-sensitivity rapid diagnostic test (hsRDT, also known as ultrasensitive RDT, Alere Malaria Ag P.f.; 05FK141; Standard Diagnostics), and dried blood spots (DBS) on filter paper (Whatman 903; GE Healthcare).

The Verrettes and La Chapelle communes of the Artibonite department in Haiti are characterized by mountainous terrain with the Artibonite river forming the northern border of both communes. Enrollment sites encompassed rural, semirural, and urban populations within the Artibonite valley as well as the higher-elevation sides of the valley. Participants were enrolled at 3 different types of venues: health facilities, schools, and churches, with further detail in Supplementary Material. A sample of 21 schools and 9 churches was selected using stratified random sampling in Artibonite (Figure 1). As previously reported, among individuals for whom household data were available, the median distance traveled to schools was 1.1 km (interquartile range [IQR], 0.5–1.8 km) and to health facilities was 2.4 km (IQR, 0.7–4.9 km) [13]. Data collection was performed in April and May 2017, which is the beginning of the rainy season in Haiti. A questionnaire was administered to all participants to gather information about their history offever, treatment-seeking practices, travel, and sociodemographic characteristics. DBS were shipped immediately to the national laboratory in Port-au-Prince, Haiti for serological data collection from May to June 2017.

Three- to 7-day-old unfed An. albimanus mosquitoes (laboratory strain STECLA) had whole salivary glands dissected and frozen for later use. Whole salivary glands were homogenized with glass tissue grinder in phosphate buffered saline pH 7.2, and freeze-thawed twice for further protein dissociation. Total protein concentration of this SGE homogenate was determined by bicinchoninic acid (BCA) assay (Pierce BCA Protein Assay Kit; ThermoFisher).

Malaria antigens and SGE homogenate proteins were covalently bound to magnetic microbeads (Luminex) through the sulfo-NHS/EDC (N-hydroxysulfosuccinimide/1-ethyl-3-(3-dimethylaminopropyl) carbodiimide) intermediate reaction as described previously [15, 16], and fully described in Supplementary Material.

DBS processing and IgG detection assay was performed in flat-bottom BioPlex Pro 96-well plates (Bio-Rad) as described previously [16], and fully outlined in Supplementary Material. Background mean fluorescence intensity (MFI) from wells incubated with buffer B was subtracted from each sample to give a final value of MFI background. Positive and negative IgG controls were added to each assay plate to ensure performance and identify if assay plates needed rerunning as described previously [17]. To optimize the protein coupling concentration to provide the most robust anti-SGE IgG assay signal, a random panel of samples from 78 Haitian participants was tested against beads coupled to different concentrations of anti-SGE homogenate (Supplementary Figure 1).

Data analysis was performed using SAS version 9.4. Log-transformed MFI background values represent anti-SGE IgG. Student t test, ANOVA, and Tukey pairwise comparisons assessed differences in anti-SGE IgG response by participant-level covariates. Nonparametric Mann-Whitney, Kruskal-Wallis, and Dunn tests of comparisons assessed differences by enrollment site-level environmental covariates using an empirical Bayes estimate of mean anti-SGE IgG by site. Remotely sensed data for environmental covariates were obtained from outside sources and values for each site were sampled using QGIS (3.20.3-Odense) (Supplementary Table 1). A multilevel logistic regression model estimated odds of an above-the-median anti-SGE IgG response by adjusting for participant- and enrollment site-level covariates. An α level of .05 was used for all statistical tests.

Five different protein concentrations of SGE homogenate (7.5, 15, 30, 60, and 120 μg/mL) were conjugated to different bead regions to directly compare IgG assay signal among these concentrations and choose the best for large-scale bead conjugation. As shown in Supplementary Figure 1, among the panel of 78 random Haitian blood samples, the 30 μg/mL SGE homogenate concentration consistently gave the highest MFI background assay signal, and thus was chosen for the bead conjugation for the remainder of the study.

Within the Artibonite EAG, the median age of study participants was 14 years and 60.5% of participants were female (Table 1). Participants enrolled with the highest frequency at health facilities (40.9%), compared to schools (34.6%) and churches (24.5%). As indicated by a positive hsRDT result, 0.9% of participants had active P. falciparum malaria infection at the time of enrollment. In total, 2.9% of all participants had a fever at the time of data collection, while 14.0% had a fever within the 2 weeks prior to collection; among those with a positive hsRDT, 7.7% had current fever and 36.1% had a history of fever. Most participants (58.0%) indicated they had used a bed net the night before enrollment. For the 1021 participants for whom these data were available, anti-SGE IgG was higher among those who used a bed net (mean = 6.91, SD = 2.36) versus those who had not (mean = 6.68, SD = 2.24); however, this difference was not statistically significant (t = −1.60, P = .11).

Median log SGE IgG MFI background level of the study population was 6.0 with a strong peak for IgG response occurring between the ages of 6 and 7 years (Figure 2). The 6 to 7-year age group showed a median value of 9.3, while the lowest median value, 4.9, occurred in the 6-month to 1-year age group (Figure 2A). Beyond age 30 years, SGE responses remained consistently low for all ages with a median of 5.1 for >30 year olds (Figure 2B). The mean anti-SGE IgG value of males was significantly higher than that of females in both Verrettes (t = 4.09, P < .0001) and La Chapelle (t = 3.25, P = .001) (Supplementary Figure 2A). Differences in anti-SGE IgG response by household size (categories of 1–4, 5–7, or more than 7 people) were significant in La Chapelle (F = 4.60, P = .01), but not in Verrettes (F = 1.86, P = .16; Supplementary Figure 2B). Mean SGE IgG level was not significantly different based on malaria infection (t = 0.46, P = .65; Supplementary Figure 2C). There were no significant differences in mean anti-SGE IgG by sex based on age group (t_{age 0–5} = −1.52, P = .13; t_{age 5–10}= 0.76, P = .45; t_{age 10–15} = −0.62, P = .53; t_age >₁₅ = −0.01, P = .99; Supplementary Figure 2D). Distribution of anti-SGE IgG levels by commune of enrollment and sex are shown in Supplementary Figure 3.

SGE IgG by site of enrollment ranged from 4.8 to 8.0 (Figure 3A). Anti-SGE IgG monotonically declined with increasing elevation (r_s = −0.22, P = .21; Figure 3B). When elevation was categorized into 3 levels (<100 m, 100–200 m, >200 m), anti-SGE IgG at elevations between 100 and 200 m was significantly higher compared to the levels seen at enrollment sites above 200 m (P = .01; Figure 3C). When comparing all sites below 200 m to any sites above 200 m, this difference was statistically significant (P = .03). Differences in anti-SGE IgG by other environmental covariates were also examined based on each covariate’s median value (listed in Supplementary Table 2). Anti-SGE IgG was significantly higher at sites that were less than 2.3 km away from the nearest water body compared to those farther than 2.3 km (P = .04). When categorizing distance to the nearest water body into quartiles (0.9–2.0 km, >2.0–2.3 km, >2.3–2.6 km, and >2.6 km), the only significant difference in pairwise comparisons was for sites with water bodies 2.0–2.3 km away versus those with water bodies greater than 2.6 km away (P = .002). There was no difference in anti-SGE IgG based on population density above versus below the median of 424 people/km² (P = .71), by normalized difference vegetation index above versus below 0.55 (P = .34), by air temperature above versus below 25.9°C (P = .54), or by average rainfall above versus below 297 mm (P = .10).

When controlling for other covariates, only age group and seropositivity to 3 P. falciparum antigen targets (circumsporozoite protein [CSP], reticulocyte binding-like protein homologue 2 [Rh2_2030], and schizont egress antigen [SEA-1]) were significantly associated with a high anti-SGE IgG response (Table 2 and Supplementary Table 3). The adjusted odds of a high IgG response to SGE for participants aged 6 months to 5 years were nearly 4-fold greater (adjusted odds ratio [aOR], 3.72; 95% confidence interval [CI], 2.29–6.04) when compared to participants over the age of 15 years. For 6–15 year olds, odds of an elevated IgG response were over 5 times higher (aOR, 5.40; 95% CI, 3.68–7.93) compared to >15 year olds. Among the 3 P. falciparum antigens, seropositivity to CSP and SEA-1 were both positively associated with increased odds of high anti-SGE IgG compared to those who were not seropositive to each antigen. Seropositivity to Rh2_2030 was associated with approximately 60% decreased odds of the high SGE IgG when compared to those not seropositive to Rh2_2030 (aOR, 0.39; 95% CI, .22–.69). All 3 antigens showed significant differences in mean anti-SGE IgG levels when each antigen’s log-transformed MFI background values were categorized into quartiles (Supplementary Figure 4).