Incidence of postnatal HIV transmission after an initial negative HEID PCR among HIV-exposed infants in Tanzania: a cohort study

Abstract Background Postnatal HIV transmission through breastfeeding remains poorly characterized within Tanzania's Prevention of Mother-to-Child Transmission (PMTCT) programme. Early infant diagnosis (EID) identifies most infections within the first 8 weeks of life; however, inadequate post-diagnosis follow-up has left population-level postnatal incidence largely unmeasured. Methods We conducted a retrospective cohort analysis linking Tanzania's national HIV Early Infant Diagnosis (HEID) records with maternal viral load (VL) data obtained from centralized laboratory information systems between December 2018 and February 2026. HIV-exposed infants with a negative (HEID) polymerase chain reaction (PCR) result at or before 8 weeks of age were eligible for inclusion (N = 98,456). The analytic cohort comprised those who had at least one subsequent (HEID) PCR test (n = 1,583; 1.6%). Baseline characteristics were compared against non-retained infants (n = 96,873) using chi-squared tests. Postnatal incidence and cumulative risk were estimated using exact Poisson 95% confidence intervals and Kaplan–Meier analysis, with loss to follow-up modelled as a competing risk to generate a lower-bound sensitivity estimate. Results The analytic cohort differed significantly from non-retained infants (p < 0.001), showing lower maternal viral suppression (43.6% vs. 60.0% with target not detected) and a higher proportion of missing VL data. Overall, 79.8% of maternal VL evaluations occurred after 6 weeks postpartum. Twenty infections occurred over 894.8 person-years, yielding an incidence of 2.24 per 100 person-years (95% CI: 1.37–3.45). Incidence increased from 1.25 per 100 person-years (95% CI: 0.40–2.91) among mothers with undetectable VL to 7.15 per 100 person-years (95% CI: 0.87–25.85) among those with VL ≥ 1,000 copies/mL. Cumulative incidence at 40 weeks reached 2.03% (95% CI: 1.13–3.62%), with a competing-risk lower bound of 1.2%. Conclusions This study provides the first longitudinal estimate of postnatal HIV transmission derived from a Tanzanian HEID cohort. The high proportion of eligible infants lacking follow-up (HEID) PCR test likely reflects both unintegrated point-of-care testing data and true clinical attrition. Given the highly selected analytic subpopulation, selection bias may limit population-level inference. While biomedical interventions remain central to PMTCT, achieving and sustaining the elimination of vertical transmission in Tanzania will require systemic monitoring enhancements—specifically, a unified, longitudinal mother–infant tracking system that integrates centralized and decentralized (point-of-care) HIV VL and HEID platforms with ePMTCT and CTC-2. Supported by optimized VL monitoring schedules and expanded testing coverage during pregnancy and breastfeeding, this system would close structural gaps in care retention and strengthen tracking of viral suppression and incident infections.