Using different approaches to assess the reproducibility of a culturally sensitive quantified food frequency questionnaire

Abstract

Objective: To report on the use of different approaches to assess the reproducibility of a culturally sensitive quantified food frequency questionnaire (QFFQ) used for assessment of the habitual dietary intake of Setswana-speaking adults in the North West Province of South Africa. Method: A previously developed and validated QFFQ was completed by trained fieldworkers. Portion sizes were estimated using different methods. Food intake was coded and analysed for nutrient intake per day for each subject. The first interview (n = 1 888) took place during the baseline data collection period. For the second interview (n = 175), a convenient sample from the subjects who had completed the first interview was collected and the interview was conducted within four to six weeks of the first interview. Results: There were good correlations between the first and second QFFQ for all the nutrients (p < 0.0001). The Wilcoxon signed-rank test showed that there were no significant differences in the median intake between the two administrations, except for energy and total fat. The Bland-Altman plots showed good agreement. Between 41% and 58% of the subjects were correctly classified into the same quartile, with less than 3% grossly misclassified. The weighted κ statistics showed moderate agreement between the two applications. Conclusion: Our results show that more than one statistical approach is needed to assess the reproducibility of a QFFQ. The reproducibility of this culturally sensitive QFFQ was good.