A Quantitative Investigation of Differences in Mediterranean Diet Definitions in Nutrition Research (P13-026-19). (13th June 2019)

• Record Type: Journal Article
• Authors: Lingle, Daniel; Dickinson, Stephanie; Vorland, Colby; Brown, Andrew
• Is Part Of: Current developments in nutrition. Volume 3(2019)Supplement 1
• Publication Date: 2019-06-13
• Subjects: Nutrition -- Periodicals; Nutritional Physiological Phenomena; Nutrition; Periodicals; Periodicals; Fulltext; Internet Resources; Periodicals; 612.3
• DOI: 10.1093/cdn/nzz036.P13-026-19
• Languages: English
• ISSNs: 2475-2991

Abstract

Abstract: Objectives: There are myriad operational definitions of the Mediterranean Diet (MedDiet) utilized in nutrition research. Variability in MedDiet definitions creates uncertainty when conducting and evaluating research, and may lead to consumer confusion and practitioner inconsistency. Therefore, we investigated and quantified heterogeneity in MedDiets. Methods: We compared and contrasted how MedDiets were operationalized from studies within the meta-analysis conducted by Sofi et al (2014) using median intakes from males from the datasets used in each study (results for females forthcoming). To quantify similarity among different diet definitions, we used two multivariate distances: Jaccard Distance (DJ ), which calculates a range from 0 (identical) to 1 (unique) based on the binary presence or absence of dietary components in the definition, and standardized Euclidean distance (DE ), which quantifies the distance between diets in multi-dimensional space from 0 (identical) with no upper bound. DJ were calculated before and after an iterative harmonization process, which collapsed similar dietary categories together. Results: Nine unique MedDiet definitions were extracted and investigated from 25 studies. Seven foods available for DE included legumes, cereals, fruit, vegetables, fish, meat, and dairy. Additional categories were available for DJ . In our preliminary findings, threshold values varied from study-to-study within and between food categories, including over a degree of magnitude difference for some foods among studies. Prior to harmonization, there were 20 food categories (e.g., one study used meat, another used meat and poultry, and another used red and processed meats). DJ ranged from 0.22 to 0.77 prior to harmonization. Harmonization reduced the number of categories to 10, and variability among constructs decreased, with DJ shrinking to 0.0 to 0.33 after. DE ranged from 0.07 (almost identical) to 6.49 – about one standard deviation for each component. Conclusions: There is substantial variability across MedDiet definitions within nutrition research as evaluated by which foods were included, thresholds for compliance within individual food categories, and binary and continuous multi-variate distances. Funding Sources: NIH/NHLBI R25HL124208; NIH/NIDDK R25DK099080.