The study population comprised 20 intervention clusters per site (n = 60 clusters), ranging from 48,241 individuals in Benin to 68,457 in India at baseline (Table 1). Age distribution differed substantially between sites, with 77.4% of study area residents in India being adults, compared to 51.3% in Malawi. The population at the Benin site was the most diverse in terms of language and religion, with 10.6% speaking a minority language and 41.4% practicing a minority religion. Benin had the lowest proportion of individuals identified as migratory at baseline (1.5%, cluster range 0.4–4%) compared to 2.9% in India, and 3.8% in Malawi.
Table 1 Demographic characteristics of DeWorm3 intervention clusters from the baseline censusMean population density within 0.5 km of households was highest at the Benin site, which includes the town of Comé, at a cluster median of 4149 individuals/km2. However, median population density differed substantially within sites, with a greater than ten-fold difference across clusters in Benin (488–8058) and India (323–4206) and a six-fold difference in Malawi (587–3843).
Treatment coveragePer-protocol coverage was high across all sites and rounds (Fig. 1). In Benin, mean per-protocol coverage at the cluster level ranged from 82% in MDA5 to 92% in MDA3. In Malawi, mean coverage ranged from 78% in MDA2 to 92% in MDA5. In India, mean coverage exceeded 90% (ranging from 91% at MDA2 to 95% at MDA3) at all rounds except MDA5, which was conducted in March 2020 and cut short due to a government COVID-19 lockdown order. Coverage increased over time in many settings (Supplementary Materials 4).
Fig. 1Per-protocol coverage (Panel A), directly observed treatment (DOT) coverage (Panel B), and treatment uptake (Panel C) across the 20 intervention arm clusters at each DeWorm3 site, by cluster and round of mass drug administration (MDA)
Treatment uptake among eligible individuals reached during MDA exceeded 95% in all clusters across all sites and rounds, with the exception of a single cluster in Malawi during MDA2 where it was 93%. DOT coverage was more varied, ranging from 54% to 97% in Benin, 48% to 96% in Malawi and 47% to 95% in India. DOT coverage was lower at rounds 2, 4, and 6 in Malawi and Benin, as those rounds coincided with treatment of SAC in schools and a smaller number eligible for cMDA, and DOT was consistently higher among children than adults (Fig. 1).
Cluster-level drivers of coverageCoverage varied by MDA round at all sites (Table 2). In fully adjusted cluster-level models, coverage was significantly higher in Benin at MDA3 than MDA1; in India, coverage was significantly lower at MDA5 than MDA1; and in Malawi, MDA3-MDA6 all had higher coverage than MDA1. In Benin, proportion speaking a minority language was associated with lower coverage [−2.3% (95% CI: −0.5, −4.1%) per category], and number of radio announcements with higher coverage [3.3% (95% CI: 0.7, 5.8%) per quartile]. In contrast, in India, the proportion speaking a minority language was associated with higher coverage [1.2% (95% CI: 0.1, 2.3%) per category].
Table 2 Predictors of per-protocol MDA coverage in DeWorm3 at the cluster level, by siteUnadjusted models are available in Supplementary Materials 5.
Individual-level correlates of non-treatment amongst adultsIn fully mutually adjusted models, odds of non-treatment were higher among younger adults than those 50 and older, with the exception of 40–49 year olds in India [odds ratio (OR) = 0.77, 95% CI: 0.71, 0.83, Table 3]. Non-treatment was highest among those 20–29 years old at all three sites (OR = 2.44, 95% CI: 2.26, 2.64 in Benin; OR = 1.74, 95% CI: 1.62, 1.88 in India; and OR = 4.25, 95% CI: 3.93, 4.59 in Malawi). While there was no significant difference by gender in India, non-treatment was lower among women in Benin (OR = 0.83, 95% CI: 0.79, 0.88) and in Malawi (OR = 0.22, 95% CI: 0.21, 0.24). Migration was associated with non-treatment at all three sites, ranging from 3.99-fold odds in Benin (95% CI: 3.73, 4.27) to 6.34-fold (95% CI: 5.93, 6.79) in Malawi. Compared to unmarried adults, monogamously married adults were less likely to be untreated in India and Benin (OR = 0.62, 95% CI: 0.59, 0.66; and OR = 0.79, 95% CI: 0.75, 0.84, respectively), while polygamous marriage was associated with lower odds of non-treatment in Benin (OR = 0.71, 95% CI: 0.60, 0.83), but higher in Malawi (OR = 1.21, 95% CI: 1.10, 1.33). Compared to adults with less than primary school education, odds of non-treatment were lower among those with primary school education in India (OR = 0.92, 95% CI: 0.86, 0.98). Otherwise, education was positively associated with non-treatment, particularly in Malawi where tertiary education was associated with three-fold odds of non-treatment (OR = 3.06, 95% CI: 1.90, 4.91). Individuals with missing or unknown information on marital status and education were more likely to be untreated than those with complete information.
Table 3 Individual-level correlates of non-treatment amongst adults in DeWorm3 intervention clustersOdds of non-treatment were lower among adults from wealthier households in Benin and India, with 11–20% lower odds for the three highest wealth quintiles compared to the lowest in Benin, and 20–27% lower odds for all four quintiles compared to the lowest in India. Living in a minority language speaking household was associated with 48–49% greater odds of non-treatment at all three sites; in contrast, no association was found with minority religion. Population density within 0.5 km of the household was associated with reduced odds of non-treatment in Benin, but increased odds in India (OR = 0.98, 95% CI: 0.96, 0.99; and OR = 1.02, 95% CI: 1.00, 1.04 per 1000 individuals/0.5 km radius, respectively).
Individual-level correlates of non-treatment amongst childrenIn the fully adjusted model, compared to school-attending SAC, school-attending PSAC had higher odds of non-treatment in India (Table 4, OR = 1.26, 95% CI: 1.09, 1.46) but lower odds in Malawi (OR = 0.77, 95% CI: 0.70, 0.85), and there was no significant difference in Benin. School-attending young adults were more likely to be untreated in Malawi (OR = 1.80, 95% CI: 1.68, 1.93) but not in India. However, odds of non-treatment were consistently higher among children in all age categories who were not attending school than school-attending SAC at all three sites, ranging from 18% increased odds for non-school-attending SAC in Benin (OR = 1.18, 95% CI: 1.10, 1.27) to 3.80-fold odds among non-school-attending young adults in India (OR = 3.80, 95% CI: 3.36, 4.29).
Table 4 Individual-level correlates of non-treatment amongst children in DeWorm3 intervention clustersGirls had consistently increased odds of non-treatment compared to boys at all three sites (Benin: OR = 1.14, 95% CI: 1.07, 1.22; India: OR = 1.13, 95% CI: 1.04, 1.24; Malawi: OR = 1.09, 95% CI: 1.03, 1.14).
Of factors potentially associated with treatment access, migration had the strongest association with non-treatment, with 4.87-fold odds in Benin (95% CI: 4.38, 5.41), 5.20-fold in India (95% CI: 4.64, 5.83), and 6.68-fold in Malawi (95% CI: 6.11, 7.31). In general, increased household wealth was associated with decreased odds of non-treatment, though there was not a dose-dependent trend. Compared to the lowest wealth quintile, children from the highest three quintiles in Benin had 12–20% decreased odds of non-treatment, in India all four quintiles had 14–27% decreased odds, and in Malawi the second, third and fourth had 19–24% decreased odds. Children from households speaking minority languages had substantially greater odds of non-treatment at all three sites, 80% in Benin (OR = 1.80, 95% CI: 1.62, 2.00), 56% in India (OR = 1.56, 95% CI: 1.18, 2.07), and 2.32-fold in Malawi (OR = 2.32, 95% CI: 1.99, 2.72). In contrast, children belonging to minority religion households had decreased odds of being untreated in Benin (OR = 0.81, 95% CI: 0.75, 0.87) but increased odds in India (OR = 1.31, 95% CI: 1.03, 1.67), and no difference compared to majority religion households in Malawi.
In contrast to adults’ own treatment, children’s treatment was positively associated with adult education in the household. Compared to children in households where adults had no primary school education, children living with adults with a primary or middle school education in India and Malawi were less likely to be untreated (OR = 0.68, 95% CI: 0.58, 0.80; and OR = 0.88, 95% CI: 0.83, 0.94), as were those living with adults with a secondary education in Benin and India (OR = 0.88, 95% CI: 0.81, 0.95; and OR = 0.64, 95% CI: 0.54, 0.75, respectively). Unknown adult education level was associated with increased odds of non-treatment in Benin and Malawi (OR = 6.08, 95% CI: 3.33, 11.08; and OR = 5.27, 95% CI: 3.79, 7.32).
Population density was associated with increased odds of being untreated at all three sites; 4% per 1000 residents within 0.5 km in Benin (OR = 1.04, 95% CI: 1.03, 1.05), and 3% in India (OR = 1.03, 95% CI: 1.00, 1.07) and Malawi (OR = 1.03, 95% CI: 1.00, 1.06).
留言 (0)