Reproducibility and relative validity of a brief quantitative food frequency questionnaire for assessing fruit and vegetable intakes inNorth‐African women

Abstract

In the context of a rapidly increasing prevalence of noncommunicable diseases, fruit and vegetables could play a key preventive role. To date, there is no rapid assessment tool available for measuring the fruit and vegetable intakes of North-African women. The present study aimed to investigate the reproducibility and relative validity of an eight-item quantitative food frequency questionnaire that measures the fruit and vegetable intakes (FV-FFQ) of Moroccan women. During a 1-week period, 100 women, living in the city of Rabat, Morocco (aged 20-49years) completed the short FV-FFQ twice: once at baseline (FV-FFQ1) and once at the end of the study (FV-FFQ2). In the mean time, participants completed three 24-h dietary recalls. All questionnaires were administered by interviewers. Reproducibility was assessed by computing Spearman's correlation coefficients, intraclass correlation (ICC) coefficients and kappa statistics. Relative validity was assessed by computing Wilcoxon signed-rank tests and Spearman's correlation coefficients, as well as by performing Bland-Altman plots. In terms of reproducibility, Spearman's correlation coefficient was 0.56; ICC coefficient was 0.68; and weighted kappa was 0.35. In terms of relative validity, compared with the three 24-h recalls, the FV-FFQ slightly underestimated mean fruit and vegetable intakes (-10.9%; P=0.006); Spearman's correlation coefficient was 0.69; at the individual level, intakes measured by the FV-FFQ were between 0.39 and 2.19 times those measured by the 24-h recalls. The brief eight-item FV-FFQ is a reliable and relatively valid tool for measuring mean fruit and vegetable intakes at the population level, although this is not the case at the individual level.