A comparison of change point models with application to longitudinal lung function measurements in children with...

Area of Science:

• Pulmonology
• Biostatistics
• Pediatric Medicine

Background:

• Cystic fibrosis (CF) is a genetic disorder causing progressive lung function loss.
• Adolescence is a critical period for accelerated lung function decline in CF.
• A hypothesis suggests a subgroup of CF patients experiences a slower, more stable decline.

Purpose of the Study:

• To model lung function decline in pediatric and adolescent CF patients.
• To compare four distinct change point models for analyzing lung function trajectories.
• To identify distinct patient subgroups based on lung function decline patterns.

Main Methods:

• Utilized data from the University of Colorado CF Children's Registry.
• Investigated four change point models: mixture-random, mixture-fixed, random, and fixed.
• Employed posterior predictive simulation for individual and population-level analysis.
• Conducted a simulation study to assess model misspecification effects.

Main Results:

• The mixed random change point model was identified as the most suitable.
• Approximately 30% of adolescents exhibited a steady annual decline of 0.5% in FEV1.
• Roughly 70% of adolescents experienced an accelerated decline of 4.4% in FEV1 annually.
• This accelerated decline began around the average age of 14.6 years.

Conclusions:

• The mixed random change point model effectively captures heterogeneous lung function decline in CF adolescents.
• Identified two distinct trajectories of lung function decline, highlighting personalized disease progression.
• Findings underscore the importance of age-specific monitoring and intervention strategies in CF care.