Background: Longitudinal population-based cohorts are critical in HIV surveillance programs in Africa but continued rapid population growth poses serious challenges to maintaining cohort representativeness with limited resources. In one such cohort, we evaluated if systematic exclusion of some residents due to growing population size biases key HIV metrics like prevalence and viremia.
Methods: Data were obtained from the Rakai Community Cohort study (RCCS) in south central Uganda, an open population-based cohort which began excluding some residents of newly constructed household structures within its surveillance boundaries in 2008. We evaluated the extent to which changing inclusion criteria may bias recent population HIV seroprevalence and viremia estimates from the RCCS using ensemble machine learning models fit to 2019-2020 RCCS census and survey data.
Results: Of the 24,729 census-eligible residents, 2,920 (12%) were living within new household structures and excluded. Predicted seroprevalence for excluded residents was 11.4% (95% Confidence Interval: 10.2, 12.3) compared to 11.8% in the observed sample. However, predicted seroprevalence for younger excluded residents 15-24 years was 5.1% (3.6, 6.1), which was significantly higher than that in the observed sample for the same age group (2.6%). Over all ages, predicted prevalence of viremia in excluded residents (2.8% [2.2, 3.3]) was higher than that in the observed sample (1.7%), resulting in a somewhat higher overall population viremia estimate of 1.9% [1.8, 2.0]).
Conclusions: Exclusion of residents in new households may modestly bias HIV viremia estimates and some age-specific seroprevalence estimates in the RCCS. Overall HIV seroprevalence estimates were not significantly affected.