Analysis of longitudinal substance use outcomes using ordinal random-effects regression models

Area of Science:

• Biostatistics
• Longitudinal Data Analysis
• Substance Use Research

Background:

• Longitudinal substance use data present analytical challenges due to missingness and individual variability.
• Traditional regression models may not adequately capture complex temporal patterns and varying covariate effects.

Purpose of the Study:

• To describe the application of random-effects regression models (RRM) for analyzing longitudinal substance use outcomes.
• To present categorical versions of RRM, focusing on ordinal outcomes with covariate effects varying across outcome categories.

Main Methods:

• Utilized random-effects regression models (RRM) to analyze longitudinal substance use data.
• Focused on categorical RRM, specifically an ordinal RRM allowing covariate effects to vary across outcome cutpoints.
• Illustrated the model's application using data from a smoking cessation study.

Main Results:

• Demonstrated the capability of RRM to handle incomplete longitudinal data and estimate individual change trajectories.
• Showcased how ordinal RRM can effectively model substance use outcomes with distinct categories.
• Highlighted the utility of varying covariate effects for nuanced interpretation of treatment impacts, such as differential effects on abstinence levels.

Conclusions:

• Random-effects regression models provide a flexible and powerful framework for analyzing complex longitudinal substance use data.
• The proposed ordinal RRM with varying covariate effects offers enhanced insights into treatment efficacy across different outcome strata.
• This methodology is valuable for researchers studying addiction and cessation interventions.