Development, validation and test-retest reliability of a short food frequency questionnaire that measures sugar intake in Pukapuka Pacific Islanders living in Auckland, New Zealand

Background: Sugar intakes have increased worldwide in parallel with the obesity, diabetes and cardiovascular disease epidemics indicating that sugar may be partly responsible. In New Zealand (NZ) the prevalence of these conditions are higher in the Pacific Island population than in the general population. Thus there is a need to develop a valid and reliable instrument to measure sugar intakes to establish whether high sugars intake, particularly fructose, are associated with chronic disease risk. The food frequency questionnaire (FFQ) is a relatively quick, simple and cost effective method to measure food and beverage consumption in large epidemiological studies. Objective: The present study aimed to develop, pretest and conduct a preliminary investigation into the validity and test-retest reliability of a 33-item, open-ended, self-administered, semi-quantitative food frequency questionnaire (FFQ) designed to measure usual sugars intakes over the previous month in Pukapuka Pacific Islanders living in South Auckland, NZ. Design: Two versions of the FFQ (open- and closed-ended) were pretested among a sample of ten Pacific Island adults living in South Auckland using cognitive interviews. Minor amendments were made to the preferred open-ended format prior to validation. Thirty-three Pukapuka Pacific Islanders, aged 18-59 years, then completed the final Pukapukan Kai FFQ administered on two occasions three weeks apart to assess test-retest reliability. The relative validity of the FFQ was assessed by comparing the data from the second administration of the FFQ (FFQ2) to the mean of three 24-hour (24h) recalls collected over three weeks prior to FFQ2. Validation was assessed using paired t-tests, Bland-Altman analysis, cross-classification, the weighted kappa and spearman correlations (SC). Differences in the mean between FFQ administrations and intraclass correlation coefficients (ICCs) were used to assess the reliability of the FFQ. Results: The FFQ adequately estimated the group’s geometric mean intake of fructose, glucose and total sugars (but not sucrose, P