BackgroundVulnerable children from poor communities with high HIV and Tuberculosis(TB) burdens were negatively impacted by COVID-19 lockdowns. Concern was raised about the extent of this impact and anticipated post-pandemic surges in mortality.MethodsInterrupted time series segmented regression analyses were done using routinely collected facility-level data of children admitted for medical conditions at four South African referral hospitals. Monthly admission and mortality data over a 60-month period from 01 April 2018 to 31 January 2023 was analysed using models which included dummy lockdown level variables, a dummy post-COVID period variable, Fourier terms to account for seasonality, and excess mortality as a proxy for healthcare burden.FindingsOf the 45 015 admissions analysed, 1237(2·75%) demised with significant decreases in admissions during all the lockdown levels, with the most significant mean monthly decrease of 450(95%, CI=657·3, -244·3) p<0·001 in level 5 (the most severe) lockdown. There was evidence of seasonality on a six-month scale during the pre-and post-COVID periods in total admissions (p=0·002), under-one-year-olds (p=0·034) and under-five-year-olds (p=0·004), which was disrupted by the COVID lockdowns. No significant mortality changes during any of the lockdown levels were found. Post-pandemic surges in admissions or mortality were not identified in children with acute gastroenteritis, acute pneumonia and severe acute malnutrition.InterpretationPaediatric admissions from communities with high levels of HIV and TB decreased during COVID-19 lockdowns, but there was no decrease in in-hospital deaths. Two postulates for these findings are the altered transmission dynamics of childhood infections and an exacerbation of existing delays in accessing healthcare. Anticipated post-pandemic surges in communicable diseases and mortality have not transpired, indicating a need for studies to identify the impact of persisting challenges in healthcare provision and that of pandemic control strategies on vulnerable populations.