Children living in low-resource settings are frequently gut-colonized with multidrug-resistant bacteria. We explored whether breastfeeding may protect against childrens incident gut colonization with extended-spectrum beta-lactamase-producing Escherichia coli (ESBL-Ec) and Klebsiella, Enterobacter, or Citrobacter spp. (ESBL-KEC). We screened 937 monthly stool samples collected from 112 children aged 1-16 months during a 2016-19 prospective cohort study of enteric infections in peri-urban Lima. We used 52,816 daily surveys to examine how exposures to breastfeeding in the 30 days prior to a stool sample were associated with childrens risks of incident gut-colonization, controlling for antibiotic use and other covariates. We sequenced 78 ESBL-Ec from 47 children to explore their diversity. Gut-colonization with ESBL-Ec was increasingly prevalent as children aged, approaching 75% by 16 months, while ESBL-KEC prevalence fluctuated between 18% and 36%. Through 6 months of age, exclusively providing human milk in the 30 days prior to a stool sample did not reduce childrens risk of incident gut-colonization with ESBL-Ec or ESBL-KEC. From 6 to 16 months of age, every 3 additional days of breastfeeding in the prior 30 days was associated with 6% lower risk of incident ESBL-Ec gut-colonization (95% CI: 0.90, 0.98, p = .003). No effects were observed on incident ESBL-KEC colonization. We detected highly diverse ESBL-Ec among children and few differences between children who were predominantly breastfed (mean age: 4.1 months) versus older children (10.8 months). Continued breastfeeding after 6 months conferred protection against childrens incident gut colonization with ESBL-Ec in this setting. Policies supporting continued breastfeeding should be considered in efforts to combat antibiotic resistance.

BACKGROUND: Neurological opportunistic infections cause significant morbidity and mortality in people with human immunodeficiency virus (HIV) but are difficult to diagnose. METHODS: One hundred forty people with HIV with acute neurological symptoms from Iquitos, Peru, were evaluated for cerebral toxoplasmosis with quantitative polymerase chain reaction (qPCR) of cerebrospinal fluid (CSF) and for cryptococcal meningitis with cryptococcal antigen test (CrAg) in serum or CSF. Differences between groups were assessed with standard statistical methods. A subset of samples was evaluated by metagenomic next-generation sequencing (mNGS) of CSF to compare standard diagnostics and identify additional diagnoses. RESULTS: Twenty-seven participants were diagnosed with cerebral toxoplasmosis by qPCR and 13 with cryptococcal meningitis by CrAg. Compared to participants without cerebral toxoplasmosis, abnormal Glasgow Coma Scale score (P = .05), unilateral focal motor signs (P = .01), positive Babinski reflex (P = .01), and multiple lesions on head computed tomography (CT) (P = .002) were associated with cerebral toxoplasmosis. Photophobia (P = .03) and absence of lesions on head CT (P = .02) were associated with cryptococcal meningitis. mNGS of 42 samples identified 8 cases of cerebral toxoplasmosis, 7 cases of cryptococcal meningitis, 5 possible cases of tuberculous meningitis, and incidental detections of hepatitis B virus (n = 1) and pegivirus (n = 1). mNGS had a positive percentage agreement of 71% and a negative percentage agreement of 91% with qPCR for T gondii. mNGS had a sensitivity of 78% and specificity of 100% for Cryptococcus diagnosis. CONCLUSIONS: An infection was diagnosed by any method in only 34% of participants, demonstrating the challenges of diagnosing neurological opportunistic infections in this population and highlighting the need for broader, more sensitive diagnostic tests for central nervous system infections.