Improving Underestimation of HIV Prevalence in Surveys Using Time-Location Sampling

Area of Science:

• Epidemiology
• Biostatistics

Background:

• Time-location sampling (TLS) is crucial for recruiting most-at-risk populations (MARPs) for HIV surveillance.
• TLS methods can introduce biases, potentially leading to inaccurate HIV prevalence estimates.
• Improving estimation methods is vital for effective public health interventions.

Purpose of the Study:

• To evaluate the Calibration on Residuals (CARES) method for enhancing HIV estimates derived from TLS.
• To compare the accuracy of CARES against standard TLS for HIV prevalence estimation in men who have sex with men (MSM).

Main Methods:

• The study utilized a large dataset of 9591 MSM from the EMIS-2010 survey.
• Three simulated 'pseudo-populations' with varying HIV prevalences were created.
• 1000 TLS samples were drawn from each pseudo-population, and HIV prevalence was estimated using both TLS and CARES methods.

Main Results:

• The CARES method demonstrated improved accuracy in HIV prevalence estimation compared to the standard TLS method.
• CARES-derived estimates were consistently closer to the true HIV prevalence across simulated populations.
• This indicates a significant enhancement in the reliability of HIV estimates for MARPs.

Conclusions:

• The CARES method offers a statistically robust approach to refine HIV prevalence estimates obtained through TLS.
• This improved methodology can lead to more accurate assessments of HIV burden in MSM and other MARPs.
• Accurate HIV surveillance data is essential for targeted prevention and treatment strategies.