Assessing treatment efficacy for interval-censored endpoints using multistate semi-Markov models fit to multiple data...

Area of Science:

• Biostatistics
• Epidemiology
• Infectious Disease Modeling

Background:

• The REGEN-2069 trial evaluated REGEN-COV's protective efficacy (PE) against SARS-CoV-2 in high-risk households.
• Complex biomedical endpoints often involve multiple, interval-censored data streams.
• Accurate modeling is crucial for understanding treatment effects in infectious diseases.

Purpose of the Study:

• To introduce a computationally efficient and general approach for analyzing complex biomedical data using multistate semi-Markov models.
• To estimate the protective efficacy (PE) of REGEN-COV against asymptomatic SARS-CoV-2 infection.
• To assess REGEN-COV's impact on seroconversion and viral shedding duration.

Main Methods:

• Utilized multistate semi-Markov models for interval-censored data.
• Employed a Monte Carlo expectation-maximization algorithm with importance sampling.
• Developed an efficient algorithm to handle intermittently observed and complexly coarsened data.

Main Results:

• REGEN-COV significantly reduced the risk of asymptomatic SARS-CoV-2 infection.
• The treatment decreased the duration of viral shedding.
• Lower seroconversion rates were observed in asymptomatically infected participants receiving REGEN-COV.

Conclusions:

• The developed semi-Markov modeling approach is computationally efficient and generalizable.
• REGEN-COV demonstrated efficacy in reducing SARS-CoV-2 transmission risk and disease duration.
• The method allows for fitting semi-parametric models to complex, intermittently observed biomedical data.