Combining growth curves when a longitudinal study switches measurement tools

Area of Science:

• Biostatistics
• Developmental Psychology
• Speech-Language Pathology

Background:

• Longitudinal studies tracking child development often require changing measurement tools.
• This necessitates complex data analysis to accurately assess growth trajectories.
• Existing methods struggle to account for measurement tool changes, hindering the identification of growth-influencing factors.

Purpose of the Study:

• To develop a flexible Bayesian hierarchical modeling framework for analyzing longitudinal growth data with changing measurement tools.
• To enable the accurate assessment of covariates impacting growth curves in pediatric populations.
• To specifically address challenges in analyzing speech perception outcomes in children with cochlear implants.

Main Methods:

• Developed a Bayesian hierarchical model to link individual growth curves across different measurement tools.
• Incorporated covariates to influence the shape of growth curves by leveraging information across all individuals (borrowing strength).
• Applied the framework to longitudinal speech perception data from pediatric cochlear implant users.

Main Results:

• The proposed model effectively integrates data from different measurement tools, providing a unified growth curve analysis.
• Demonstrated the ability to assess the impact of covariates, such as age at implantation, on growth trajectories.
• Facilitated a more robust comparison of growth rates between different age-at-implantation groups.

Conclusions:

• The Bayesian hierarchical modeling framework offers a powerful solution for analyzing longitudinal growth data with measurement tool transitions.
• This approach enhances the understanding of developmental trajectories and the factors influencing them in pediatric populations.
• Provides a valuable tool for researchers studying outcomes in children with cochlear implants, enabling more precise assessments of intervention impacts.