Useful and extraneous variability in longitudinal assessment of lung function.

Area of Science:

• Pulmonary Medicine
• Biostatistics
• Epidemiology

Background:

• Longitudinal measurements are vital for assessing lung function changes due to treatments or environmental exposures.
• Variability in lung function estimates arises from true individual differences (signal) and measurement error (noise).

Purpose of the Study:

• To define and quantify "signal decay" in longitudinal lung function studies.
• To identify factors contributing to signal decay and propose mitigation strategies.

Main Methods:

• Conceptual framework defining signal and noise in longitudinal data.
• Analysis of how measurement imprecision, study length, subject attrition, and testing frequency affect signal decay.

Main Results:

• Imprecise individual rate estimations are a primary driver of signal decay.
• Increased subject attrition also significantly contributes to signal decay and introduces bias.
• Increasing study length can compensate for signal decay, while frequent testing offers limited benefit for signal but aids in bias reduction.

Conclusions:

• Signal decay, caused by measurement noise and attrition, diminishes the interpretability of longitudinal lung function data.
• Optimizing study design through longer durations and careful attrition management is crucial for robust findings.
• Interval testing can mitigate secular bias and data loss.