A joint model for a longitudinal outcome and a progressive multistate model under a mixed observation scheme

Area of Science:

• Biostatistics
• Longitudinal Data Analysis
• Survival Analysis

Background:

• Joint models are crucial for understanding correlated longitudinal and event-time outcomes, like diabetic microvascular complications.
• Existing models lack the capability to handle interval-censored entry times common in multistate processes and cohort studies.

Purpose of the Study:

• To develop a novel joint model accommodating interval-censored multistate event-time data.
• To accurately assess the association between longitudinal trajectories and multistate disease progression.

Main Methods:

• Formulated a shared random effects joint model.
• Incorporated linear mixed-effects and proportional intensities progressive three-state Markov submodels.
• Utilized maximum likelihood estimation and assessed parameter bias and confidence interval coverage via simulation.

Main Results:

• The proposed model effectively handles interval-censored entry times in multistate processes.
• Compared to existing methods, the new model provides less biased estimates of association.
• Demonstrated the association between retinopathy trajectory and neuropathy progression in diabetic patients.

Conclusions:

• The developed joint model offers a robust framework for analyzing complex longitudinal and multistate event-time data.
• The findings suggest potential efficiencies in screening and monitoring diabetic complications by linking retinopathy and neuropathy.
• This approach improves upon existing methods by accurately accounting for interval-censored data in multistate models.