Statistical approaches for assessing the relative validity of a food-frequency questionnaire: use of correlation coefficients and the kappa statistic

Abstract

AbstractObjective:To compare different statistical methods for assessing the relative validity of a self-administered, 150-item, semi-quantitative food-frequency questionnaire (FFQ) with 4-day weighed diet records (WR).Design:Subjects completed the Scottish Collaborative Group FFQ and carried out a 4-day WR. Relative agreement between the FFQ and WR for energy-adjusted nutrient intakes was assessed by Pearson and Spearman rank correlation coefficients, the percentages of subjects classified into the same and opposite thirds of intake, and Cohen's weighted kappa.Subjects:Forty-one men, mean age 36 (range 21-56) years, and 40 women, mean age 33 (range 19-58) years, recruited from different locations in Aberdeen, Scotland.Results:Spearman correlation coefficients tended to be lower than Pearson correlation coefficients, and were above 0.5 for 10 of the 27 nutrients in men and 17 of the 27 nutrients in women. For nutrients with Spearman correlation coefficients above 0.5, the percentage of subjects correctly classified into thirds ranged from 39 to 78%, and weighted kappa values ranged from 0.23 to 0.66.Conclusions:Both Spearman correlation coefficients and weighted kappa values are useful in assessing the relative validity of estimates of nutrient intake by FFQs. Spearman correlation coefficients above 0.5, more than 50% of subjects correctly classified and less than 10% of subjects grossly misclassified into thirds, and weighted kappa values above 0.4 are recommended for nutrients of interest in epidemiological studies.