Characterizing Measurement Error in Dietary Sodium in Longitudinal Intervention Studies

Area of Science:

• Nutritional epidemiology
• Biomarker validation
• Longitudinal data analysis

Background:

• Previous research on measurement error in nutrition has focused on single time points or basic demographics.
• Few studies have explored how measurement error structures evolve in longitudinal randomized trials.
• Understanding this dynamic is vital for correcting self-reported dietary data and interpreting intervention effects.

Purpose of the Study:

• To investigate the relationship between urinary sodium biomarkers and self-reported sodium intake.
• To determine if this measurement error relationship varies across time and treatment conditions in longitudinal trials.

Main Methods:

• Utilized internal longitudinal validation data from two randomized controlled trials.
• Employed mixed-effects regression models with flexible error variance-covariance structures.
• Tested interactions between time, treatment condition, and self-reported intake.

Main Results:

• No evidence found that measurement error varies with self-reported sodium levels.
• Significant evidence indicated that urinary sodium can differ by time or treatment condition, independent of self-reported values.
• A consistent final model was achieved for both validation datasets.

Conclusions:

• Measurement error in longitudinal nutritional intervention trials may differ across time and treatment groups.
• Researchers must consider the potential for differential measurement error, rather than assuming it is non-differential.
• Future studies should incorporate data collection strategies that capture the dynamic nature of measurement error, including validation data across time and groups.