a1 Cardiovascular Risk and Nutrition Research Group (CARIN-ULEC), IMIM-Hospital del Mar Barcelona, Biomedical Research Park (Parc de Recerca Biomèdica de Barcelona – PRBB), c/Doctor Aiguader 88, 08003 Barcelona, Spain
a2 CIBER Physiopathology of Obesity and Nutrition (CIBEROBN), Instituto de Salud Carlos III, Madrid, Spain
a3 Program of Research in Inflammatory and Cardiovascular Disorders, IMIM-Hospital del Mar, Barcelona, Spain
a4 Center for Research in Environmental Epidemiology (CREAL), Biomedical Research Park (Parc de Recerca Biomèdica de Barcelona – PRBB), Barcelona, Spain
a5 CIBER Epidemiology and Public Health (CIBERESP), Instituto de Salud Carlos III, Madrid, Spain
a6 Centro de Salud La Marina, IDIAP Jordi Gol, Barcelona, Spain
a7 Cardiovascular Epidemiology and Genetics Research Group (EGEC-ULEC), IMIM-Hospital del Mar, Barcelona, Spain
Objective The aim of the present study was to assess the concurrent and construct validity of two diet-quality indices, a modified Mediterranean diet score (mMDS) and a Mediterranean-like diet score (MLDS) additionally incorporating unhealthy food choices, as determined by an FFQ.
Design A validation study assessing FFQ intake estimates compared with ten or more unannounced 24 h recalls. Pearson's correlation coefficients, intraclass correlation coefficients (ICC), Bland–Altman plots and the limits of agreement method were used to assess the between-method agreement of scores. Construct validity was shown using associations between nutrient intakes derived from multiple 24 h recalls and the mMDS and MLDS derived from the FFQ.
Setting Gerona, Spain.
Subjects A total of 107 consecutively selected participants from a population-based cross-sectional survey.
Results Pearson's correlations for the energy-adjusted mMDS and MLDS compared with multiple recalls were 0·48 and 0·62, respectively. The average FFQ energy-adjusted mMDS and MLDS were 102 % and 98 % of the recall-based mMDS and MLDS estimates, respectively. The FFQ under- and overestimated dietary recall estimates of the energy-adjusted MLDS by 28 % and 25 %, respectively, with slightly wider boundaries for the mMDS (31 % and 34 %). The ICC, which assesses absolute agreement, was similar to Pearson's correlations (mMDS = 0·48 and MLDS = 0·61). The mean differences between methods were similar across the range of average ratings for both scores, indicating the absence of bias. The FFQ-derived mMDS and MLDS correlated in the anticipated directions with intakes of eleven (73·3 %) and thirteen of fifteen nutrients (86·7 %), respectively.
Conclusions The FFQ provides valid estimates of diet quality as assessed by the mMDS and MLDS.
(Received November 10 2010)
(Accepted April 26 2011)
(Online publication July 11 2011)
† A full roster of REGICOR investigators and collaborators can be found at www.REGICOR.org/regicor.inv