Leveraging Routinely Collected Program Data to Inform Extrapolated Size Estimates for Key Populations in Namibia:...

Area of Science:

• Epidemiology
• Biostatistics
• Public Health

Background:

• Estimating key populations like female sex workers (FSW) and men who have sex with men (MSM) is crucial for effective HIV program planning and resource allocation.
• Small area estimation (SAE) can bridge data gaps in areas lacking direct population size estimates (PSEs) by utilizing auxiliary data.
• Routinely collected program data have not been extensively used as auxiliary data for subnational PSEs of key populations.

Purpose of the Study:

• To systematically generate regional size estimates for FSW and MSM in Namibia.
• To integrate routinely collected HIV program data into estimation methods.
• To inform HIV program strategies through consensus-based approaches with local stakeholders.

Main Methods:

• Utilized a consensus-informed estimation approach with local stakeholders.
• Integrated quarterly HIV program data from service providers.
• Employed Bayesian triangulation to weight existing PSEs with program data.
• Generated SAEs using simple imputation, stratified imputation, and Poisson regression.
• Selected final estimates through an iterative qualitative ranking process.

Main Results:

• National estimates for FSW ranged from 4,777 to 13,148 (1.5%-3.6% of females aged 15-49).
• National estimates for MSM ranged from 4,611 to 10,171 (0.7%-1.5% of males aged 15-49).
• Inclusion of program data as priors influenced proportions, e.g., FSW proportion increased from 1.9% to 2.8% via simple imputation.
• Stratified imputation using HIV prevalence or population density showed varied impacts on FSW and MSM proportions by region and density.

Conclusions:

• SAE approaches successfully combined epidemiologic and program data for subnational key population size estimates in Namibia.
• Estimates demonstrated significant sensitivity to the inclusion of program data.
• Program data offer a valuable supplemental source for aligning PSEs with real-world HIV programs, especially where data collection is challenging.
• Further research is needed to optimize the inclusion and validation of program data for key population size estimation and enhance HIV response efforts.