Measuring concurrency using a joint multistate and point process model for retrospective sexual history data

Area of Science:

• Epidemiology
• Mathematical Biology
• Statistics

Background:

• Measuring concurrency (overlapping sexual partnerships) is challenging but crucial for understanding HIV transmission.
• Previous methods for estimating concurrency have limitations in accounting for partnership dynamics and data dependencies.

Purpose of the Study:

• To introduce a novel joint multistate and point process model for accurately estimating concurrency.
• To analyze the relationship between concurrency, partnership dynamics, and HIV status.

Main Methods:

• Developed a joint multistate and point process model where states represent the number of concurrent partnerships.
• Modeled discrete events for partnerships starting and ending on the same date ('one-offs').
• Applied the model to epidemiological data from men who have sex with men (MSM) in Los Angeles.

Main Results:

• The new model provides estimators for concurrent partnership distribution and mean sojourn times.
• Higher point prevalence of concurrency was observed among men later diagnosed as HIV positive.
• One-off partnerships were linked to increased rates of subsequent partnership dissolution.

Conclusions:

• The proposed modeling approach offers a more robust method for quantifying concurrency.
• Findings suggest a link between higher concurrency and HIV positivity in the studied MSM population.
• Understanding partnership dynamics, including one-offs, is important for HIV prevention strategies.