Niger is implementing nationwide mass drug administration (MDA) with azithromycin among children aged 1-59 months to reduce child mortality. Prior trials used population-based samples to monitor antimicrobial resistance (AMR), but clinic-based samples may be more resource-efficient. We leveraged baseline data collection from a cluster-randomized trial to conduct a cross-sectional study to assess whether clinic and population-based samples differ in antibiotic resistance genes (ARGs).

We collected rectal samples from 3 groups of children aged 1-59 months across 39 Centres de Santé Intégrés (CSI, primary health centers). The sick group included 30 children/CSI presenting with diarrheal diseases at the clinics. The healthy group included 30 children/CSI attending routine clinic visits. The population-based group included 10 children randomly selected from 3 villages per CSI.

To account for younger children in the sick group, we age-matched the comparison groups to the sick group by predefined age strata, sampling without replacement. Samples were then pooled by CSI and sequenced. Differential ARG abundance was assessed using DESeq2 with permuted p-values. We also adjusted for antibiotic use in the pairwise comparison between the sick and healthy groups.

After matching, 3,528 children (median age 13 months [IQR: 7–24]; 52.5% male) were included. The abundance of macrolide ARGs was higher in the sick group than in the population-based (P=0.001) and healthy groups  (P<0.001); there was no difference between the population-based and healthy groups (P=0.38). The same trend persists for beta-lactams (Figure 1).

Children presenting to clinics with diarrheal disease exhibited higher macrolide and beta-lactam resistance compared to non-sick children, whereas healthy and population-based groups were comparable. Clinic-based AMR has been proposed as a way to monitor MDA, interpretation should include that sick children have higher AMR loads than population-based children.