Statistical Methods Used for the Evaluation of Reliability and Validity of Nutrition Assessment Tools Used in Medical Research

Vassiliki      Bountziouka; Demosthenes      B. Panagiotakos

doi:10.2174/138161210794455102

Abstract

The contribution of diet to the development of several chronic diseases, such as vascular disease, diabetes or lipid abnormalities has been established. Therefore, clinical trials dealing with these diseases need to adjust for individual dietary habits in order to account for potential confounding. Common practice in the majority of studies that collect dietary information is to record individual habits using specific questionnaires (e.g. Food Frequency questionnaires, FFQ). Nevertheless, a major challenge in nutrition assessment is the correct measurement of dietary exposure. This can be expressed as the reliability and validity of the retrieved information. These issues refer to how close the food records and the energy intake estimated by a tool represent actual food intake. To establish accuracy of a tool is of major importance in order to avoid inconsistent estimates of dietary intake that can distort any potential relation between diet, pharmacological treatment and disease. The aim of this review is to critically present commonly used statistical methods in reliability and validity studies.

Keywords: Reliability, validity, repeatability, accuracy, dietary assessment, clinical trials, Nutraceuticals and Chronic Illness, Nutrition Assessment, questionnaires (FFQ), memory lapses, Inter-rater reliability, Test-retest, Inter-method reliability, Internal consistency reliabil-ity, Correlation Coefficients, Spearman's rho correlation coefficient, Pearson's correlation coefficient, The intra-class correlation coefficient, Spearman-Brown reliability coefficient, Cronbach's Alpha, Kendall's Tau, Cohen's Kappa, The Bland and Altman Method, limits of agreement, fanning, Paired Comparisons