Namibia Small Area Estimation: Final Report

The objective is to inform national and sub-national HIV programming and strengthen capacity of country-level stakeholders to support evidence-based approaches in addressing data gaps for key populations. Small area estimation (SAE) techniques are applied to estimate parameters for "small areas" of interest, where empiric estimates may have unacceptably large standard errors or where there are no empiric data. The report postulates that SAE of key-population-related outcomes will yield substantial reductions in the standard errors of population size estimates (PSE) and better inform regional measures of HIV prevalence and ART coverage in Namibia. Data collation and synthesis involved collecting available direct PSE for key populations, including female sex workers (FSW), men who have sex with men (MSM), and transgender women (TGW). Direct estimates were obtained from three different studies in 2014, 2015, and 2020. Auxiliary data sources accessed in partnership with key stakeholders include the Namibia Population-Based HIV Impact Assessment (NAMPHIA), Risk Assessment Screening Tool (RAST) Database, PEPFAR KP DATIM Dataset, KP-STAR Dataset, Spectrum Estimates, and 2011 Census Report. A quality assessment tool was applied to all systematically identified HIV-related data for key populations in Namibia. The tool evaluated the general study design, study implementation, and outcome-specific quality of each study. A total of 14 data sources were found eligible for quality assessment, with most data sources for key populations being of "fair" quality. Methodologies for extrapolation were applied at the regional level per existing PSE and available auxiliary data. Approaches include age stratification of existing direct estimates, simple imputation, stratified imputation, and regression modeling. Preliminary extrapolated estimates were reviewed by the Namibia USAID Mission and implementing partners, and feedback was integrated into the final estimates. The consensus estimator tool was used to harmonize existing direct estimates for each key population. The tool leverages prior information about the known distribution of each population to inform a singular estimate for each region where data are available. The tool also weights each direct estimate differently based on a user-defined measure of confidence. Simple imputation was conducted for all key populations, while stratified imputation was conducted for FSW and MSM. For simple imputation, the mean size estimate proportion was calculated for the regions per consensus estimates across regions where direct size estimates were available. Stratified imputation was based on the median HIV prevalence in regions with IBBS direct estimates. The report provides a comprehensive overview of the small area estimation methods used to estimate key population sizes in Namibia. The findings of this report will inform national and sub-national HIV programming and strengthen capacity of country-level stakeholders to support evidence-based approaches in addressing data gaps for key populations.