A likelihood approach to incorporating self-report data in HIV recency classification.

Area of Science:

• Epidemiology
• Biostatistics
• Public Health

Background:

• Estimating new HIV infections is crucial for public health surveillance but challenging due to difficulties in distinguishing recent from long-term infections.
• Biomarkers and self-reported testing history are increasingly available in bio-behavioral surveys, offering potential for improved HIV recency classification.

Purpose of the Study:

• To develop and validate a probabilistic model for classifying HIV recency status using available data.
• To improve the accuracy and efficiency of estimating recent HIV infections in population-based surveys.

Main Methods:

• A likelihood-based probabilistic model was developed using nationally representative data from the Population-based HIV Impact Assessment (PHIA) Project.
• The model incorporates individuals with known recency status and those with undetermined status, integrating biomarker data and self-reported testing history.
• The proposed model was compared against logistic regression and binary classification trees using Malawi PHIA data and simulated data.

Main Results:

• The novel probabilistic model demonstrated more efficient and less biased parameter estimates compared to current methods.
• The model showed robustness to potential reporting errors and model misspecification.
• This approach offers a more reliable method for classifying HIV recency status.

Conclusions:

• The developed probabilistic model provides a more accurate and robust approach to estimating recent HIV infections.
• This method can enhance the precision of HIV surveillance and impact assessments.
• Utilizing biomarker and testing history data within a probabilistic framework is key to advancing HIV recency classification.