Mixed effects logistic regression models for longitudinal ordinal functional response data with multiple-cause...

Area of Science:

• Gerontology
• Biostatistics
• Public Health

Background:

• Analyzing functional limitations in aging populations is crucial for understanding health trajectories.
• Informative drop-out (death, loss-to-follow-up) complicates longitudinal data analysis in aging studies.

Purpose of the Study:

• To investigate the association between current and previous functional limitations.
• To examine the modifying role of physical activity and informative drop-out on this association.
• To develop a statistical model accommodating longitudinal ordinal outcomes and multiple drop-out causes.

Main Methods:

• Utilized a mixed-effects logistic model with shared random effects for ordinal functional outcomes and discrete-time survival profiles.
• Modeled multiple causes of informative drop-out (death, unknown loss-to-follow-up) within a shared parameter selection framework.
• Assessed model robustness by varying assumptions regarding random effects distribution and drop-out components.

Main Results:

• Initial functional limitation strongly predicts future functional limitation in older adults.
• This association is significantly stronger for individuals with lower physical activity levels.
• The link between current and previous within-individual functional status trajectories is weaker and more sensitive to model assumptions.

Conclusions:

• Baseline functional limitation is a key predictor of future limitations, with physical activity playing a crucial moderating role.
• The proposed statistical model effectively handles longitudinal functional data with informative drop-out.
• Promoting physical activity may be vital for mitigating the progression of functional limitations in aging populations.