Public Health Nutrition

Research Article

Misclassification associated with measurement error in the assessment of dietary intake

Carl de Moora1 c1, Tom Baranowskia2, Karen W Cullena2 and Theresa Nicklasa2

a1 Departments of Behavioral Science and Biostatistics, The University of Texas M.D. Anderson Cancer Center, 1515 Holcombe Boulevard – Box 243, Houston, TX 77030, USA

a2 Children's Nutrition Research Center, Department of Pediatrics, Baylor College of Medicine, Houston, TX, USA

Abstract

Objective: Dietary assessment has been used for certification to receive food supplements or other nutrition services and to provide feedback for educational purposes. The proportion of individuals correctly certified as eligible is a function of the amount of error that exists in the dietary measures and the level of dietary intake used to establish eligibility. Whether individuals are correctly counselled to increase or decrease the consumption of selected foods or nutrients is a function of the same factors. It is not clear, however, what percentage of individuals would be correctly classified under what circumstances. The objective of this study is to demonstrate the extent to which measurement error and eligibility criteria affect the accuracy of classification.

Design: Hypothetical distributions of dietary intake were generated with varying degrees of measurement error. Different eligibility criteria were applied and the expected classification rates were determined using numerical methods.

Setting and subjects: Simulation study.

Results: Cut points of dietary intake at decreasing levels below the 50th percentile of true intake were associated with lower sensitivity and predictive value positive rates, but higher specificity and predictive value negative rates. The correct classification rates were lower when two cut points of dietary intake were used. Using a single cut point that was higher than the targeted true consumption resulted in higher sensitivity but lower predictive value positive, and lower specificity but higher predictive value negative.

Conclusions: Current methods of dietary assessment may not be reliable enough to attain acceptable levels of correct classification. Policy-makers and educators must consider how much misclassification error they are willing to accept and determine whether more intensive methods are necessary.

(Received August 01 2002)

(Accepted November 11 2002)

Correspondence

c1 *Corresponding author: Email cdemoor@mdanderson.org

0Comments