Assessing the Relationship between the Baseline Value of a Continuous Variable and Subsequent Change Over Time

Area of Science:

• Biostatistics
• Longitudinal Data Analysis
• Epidemiology

Background:

• Analyzing the relationship between baseline values and subsequent changes is common in cohort studies.
• Complexity arises with limited data waves, posing challenges for non-biostatisticians.
• Understanding these complexities is crucial for accurate longitudinal data interpretation.

Purpose of the Study:

• To review and clarify statistical methods for analyzing the association between baseline values and subsequent changes in continuous variables.
• To identify adequate statistical methods for cohort studies with limited data waves.
• To explain key issues like mathematical coupling, measurement error, and regression to the mean.

Main Methods:

• Simulated longitudinal data of body mass index (BMI) in children were used.
• Methods for analyzing the association between baseline value and subsequent change were reviewed, assuming linear growth.
• The study focused on scenarios with two data waves and compared different statistical approaches.

Main Results:

• Mathematical coupling, measurement error, individual change variability, and regression to the mean are key analytical challenges.
• Linear random effects models are suitable for analyses with more than two data waves.
• Blomqvist's method demonstrated accurate estimation of the regression coefficient when adjusting for measurement error variance with two data waves.

Conclusions:

• The choice of statistical method depends on the number of data waves, measurement error information, and individual change variability.
• Blomqvist's method is recommended for accurate analysis when only two data waves are available.
• Accurate analysis of baseline-change relationships in longitudinal studies requires careful consideration of data structure and potential biases.