Validation of a food frequency questionnaire to assess nutrient intakes in women participating in the PRedictors linking Obesity and gut MIcrobiomE (PROMIsE) study : a thesis presented in partial fulfilment of the requirements for the degree of Masters of Science in Nutrition and Dietetics, Massey University, Albany, New Zealand

Abstract

Background: Diet is a modifiable risk factor for a range of chronic diseases. Food frequency questionnaires (FFQ) are commonly used in epidemiological studies to investigate this relationship due to their ease of administration, low cost and ability to assess nutrient intake over an extended period of time. Like all dietary assessment tools FFQ’s are not free of error and need to be validated for use in their intended population. There is currently no FFQ that has been independently validated for use in the Pacific and New Zealand (NZ) European women in New Zealand. Aim To validate a semi-quantitative food frequency questionnaire in 18-45-year-old pre-menopausal adult NZ European and Pacific women participating in the PROMIsE study, living in the greater Auckland area. Methods Participants included 287 premenopausal women living in New Zealand of NZ European (n=161) and Pacific (n=126) ethnicity. Women completed an FFQ (NZWFFQ) designed to assess the dietary intake of 31 nutrients over the previous month and a five-day estimated food record (5d-FR). Relative validity was assessed by comparing the nutrient intakes of the NZWFFQ and 5d-FR using Wilcoxon signed rank test, Spearman’s correlation coefficients, cross-classification, weighted kappa statistic and Bland-Altman analysis. Validity was evaluated for crude and energy adjusted data for the total group and separately for NZ European and Pacific ethnicity. Results The nutrient intake of the NZWFFQ was higher than the 5d-FR overall for all nutrients (range: 6%-113% difference) except iodine (-16%). Correlation coefficients ranged from 0.07 for iodine in the unadjusted total group to 0.63 for alcohol. The highest energy correlation coefficients were for energy adjusted NZ European data (0.17-0.73) and were lowest for the unadjusted Pacific data (-0.02-0.47). Classification into same and adjacent quartiles of intake, and gross misclassification into opposite quartiles, were respectively 77.5% and 5.41% for the total group, 81% and 3.6% for the NZ European group, and 71.2% and 7.6% for the Pacific group for energy adjusted data. The weighted kappa showed slight to moderate agreement for the total group (0.12-0.47), slight to moderate agreement (0.16-0.54) for NZ European, and slight to fair agreement (-0.10-0.28) for the Pacific group. Bland-Altman analysis showed wide limits of agreement for nutrients in each group, with wider limits of agreement and larger mean differences for the Pacific group. Conclusion: The NZWFFQ gives good validity for ranking NZ European women by nutrient intake however did not compare well for ranking Pacific women by nutrient intake. As most nutrients were overestimated by the NZWFFQ it is not a suitable tool to use for estimating absolute nutrient intake.