Self-reported data on diet is crucial to many trials that aim to change a person’s weight or eating behavior, but errors are an inevitable result of this type of measurement. To help solve this problem, Juned Siddique, DrPH, associate professor in Preventive Medicine-Biostatistics and Psychiatry and Behavioral Sciences, has received a National Institutes of Health (NIH) grant to develop a statistical framework for correcting measurement errors associated with self-reported diet assessments in longitudinal lifestyle intervention trials.
The National Heart, Lung, and Blood Institute will provide the three-year, $1.1 million grant.
In the study, Siddique and his team will examine the relationship between self-reported and objective measures of diet from four previous intervention trials to build the framework, which they hope will correct for errors in the dietary data.
Siddique explained how this work can help future investigators build more effective interventions for at-risk populations.
What are examples of lifestyle intervention trials that rely on self-reported diet assessment?
This work was directly motivated by my collaborations with Feinberg faculty. In my project, I will use data from two intervention studies conducted here at Feinberg. One, led by Bonnie Spring, PhD, was a randomized controlled trial designed to examine the feasibility and efficacy of an abbreviated smartphone-supported weight loss program. The other, led by Namratha Kandula, MD, MPH, was a pilot study designed to determine the feasibility and initial efficacy of a culturally-targeted, community-based lifestyle intervention to improve physical activity and diet behaviors among medically underserved South Asians. In both studies, participants were asked to report their dietary intake over the course of the interventions.
How are measurement errors made?
Measurement error in self-reported dietary data can be made in several different ways. Participants may not remember everything they ate, or they may overestimate or underestimate portion sizes. In intervention studies, participants may falsely report their diet in order to appear more compliant with the intervention. Or they may become more accurate over time due to repeated self-monitoring.
Why do measurement errors need to be corrected?
Not accounting for measurement error could result in a failure to identify an intervention that is effective. Or it may result in a determination that an intervention is effective, when in fact the effect is a result of measurement error.