Relative validity of an online short FFQ assessing the Dutch adapted version of the Mediterranean-DASH Intervention for Neurodegenerative Delay (MIND) diet in older adults at risk of cognitive decline

Abstract A short FFQ was developed for online assessment of adherence to the Dutch Mediterranean-DASH Diet Intervention for Neurodegenerative Delay (MIND-NL) diet, a culturally adapted version of the original American MIND diet. This study aimed to evaluate the relative validity of this short FFQ for assessing adherence to the MIND-NL diet, as scored by the MIND-NL score, compared with 3-d food records among community-dwelling older adults at risk of cognitive decline ( n 1078; 67·4 ( sd 4·6) years; 64 % female). A combination of statistical methods was used to assess the relative validity: presence of bias by Bland–Altman analysis; strength of association with Kendall’s Tau-b and Spearman correlation coefficients and levels of agreement with Wilcoxon signed rank test, cross-classification and weighted Kappa (κ) statistics. The Kendall’s Tau-b correlation for the MIND-NL score was 0·33 (95 % CI: 0·29, 0·37; de-attenuated Tau-b: 0·45). Individual MIND-NL diet component score correlations ranged from 0·05 to 0·56, with 12 out of 15 of the MIND-NL diet components adequately correlated (> 0·20). The average MIND-NL scores for the short FFQ (8·4 ( sd 1·8) points) and food records (6·7 ( sd 1·7) points) showed to be significantly different ( P < 0·001). The Kappa (κ) coefficient for tertile classification of the MIND-NL score was 0·29 (95 % CI: 0·25, 0·33), indicating an acceptable level of agreement in ranking participants beyond chance. Acceptable agreements (κ > 0·20) were observed for 10 out of 15 MIND-NL diet components. Taking all analyses together, the short FFQ showed acceptable validity for ranking older adults at risk of cognitive decline according to their adherence to the MIND-NL diet.