Validation of a Food Frequency Questionnaire assessing the sugar intakes of Pacific Islanders in Auckland, New Zealand

Background: Evidence linking fructose intakes to gout, type 2 diabetes and obesity is growing. This is of particular interest in Pacific people who have higher rates of these diseases compared to other ethnic groups in New Zealand. Research to examine the links between sugar and non-communicable diseases in the Pacific population is limited, however, by lack of a validated, culturally appropriate instrument to assess sugar intakes in this group. Objective: To test the relative validity and reliability of a Food Frequency Questionnaire (FFQ) designed to measure usual sugar intakes in Pacific Islanders living in Auckland, New Zealand. Methods: 90 Pacific Island participants were recruited over three phases during 2013 and 2014 from the Auckland region. Participants took part in four dietary assessment interviews over four weeks: the sugars FFQ was administered in weeks one and four. Three 24-hour recalls were administered in weeks one, two and three. Validity of the sugars FFQ for assessing sugar intakes from various food categories was assessed by cross classification agreement, Kappa scores, Spearman’s and intraclass correlation coefficients and Bland-Altman tests of agreement with mean 24-hour recall intakes. Reliability was assessed by intraclass correlation coefficients comparing sugar intakes estimated from the two sugars FFQ administrations. Results: Relative validity of the sugars FFQ was moderate-good. Correct classification into the same or adjacent quartiles for sugars intakes from all sources and from sugar sweetened beverages (SSB) were 86% and 91%, respectively, and Kappa scores for both categories were 0.54, indicating moderate agreement. Intraclass correlation coefficients were 0.81 and 0.86 and Spearman’s correlation coefficients were between 0.65 and 0.74, respectively. Sugar intakes from fresh and canned fruit showed less reliability with 76-80% correctly classified into the same or adjacent quartiles and Kappa scores of 0.25-0.33 indicating fair agreement. Bland-Altman analysis indicated good agreement at the group level for total sugars; fructose and glucose intakes from all sources however the wide limits of agreement from all sugar intakes indicating poor agreement at the individual level. For servings of fruit 88% of participants were correctly classified into the same or adjacent quartiles and Spearman’s and intra-class correlation coefficients were 0.5 and 0.68, respectively. Reproducibility of the sugars FFQ was good with intraclass correlation coefficients over 0.7 for sugar intakes from all categories. Conclusion: The relative validity and repeatability of the sugars FFQ in Pacific populations was good and comparable with (or better than) other validated FFQs assessing sugars intakes. The sugars FFQ could be used to identify high, moderate and low sugar consumers in large population studies conducted amongst Pacific populations