Validation of a semi-quantitative food frequency questionnaire used among 1 1/2 - 3 year-old Norwegian children.

Abstract

Summary Background In 1999 the first national dietary survey among infants and toddlers in Norway was carried out. Two cohorts were implemented; Spedkost 1998/99 for 6 months and 12 months old children, and Småbarnskost 1999 for 24 months old children. In 2006/07, Spedkost and Småbarnskost was carried out again. However, the SFFQs used in Småbarnskost 2007 was revised and adjusted from the SFFQ used in 1999; therefore this validity study is necessary to investigate if the revised SFFQ used among 24 months old children in Småbarnskost 2007 is a good dietary assessment tool to be used in this age group. It is important to assess the dietary habits of children and follow the time-trend to monitor if any changes occur and uncover any trends that could constitute a potential health risk of their immediate and future health. Knowing the populations dietary habits is important in order to continuously update nutrition recommendations and guidelines. Validation of dietary assessment methods used in large surveys are important in order to clarify the strength of the results and investigate if the tools reflect the true dietary intake. Furthermore, validation studies are necessary so improvements of our dietary assessment techniques can be made in future surveys. In this validation study we aim to clarify if the SFFQ used in Småbarnskost 2007 is able to assess dietary intakes of the main food groups and macro- and micronutrients on the group level and ranking individuals according to high and low intakes. Subjects & Methods Forty-one children (n=41) aged 17 to 39 months participated. The total sample was initially of 42 subjects, but one child was excluded due to extreme and unrealistic reports of food intake from the SFFQ. A four day weighed food diary was chosen as a reference method to validate the SFFQ used in the survey Småbarnskost 2007 . Subjects were recruited through a selection of day-cares in Oslo, Norway. Parents filled out an accession form agreeing to participate and completed an SFFQ to assess the children s diet. Approximately 14 days later they completed a 4-day weighed food diary (WFD) (Wednesday-Saturday). Kitchen scales to assist the recording were handed out to each household and participating day-care. Parents recorded their child s diet at home, while day-care assistants recorded the toddler s food intake during the time spent in the day-care. The weighed food diaries were coded manually and the questionnaires were scanned and coded using syntax. Daily intakes of energy, nutrients and food items were calculated by a software data program using a food database developed at the Department of Nutrition, University of Oslo. Wilcoxon signed rank-test was performed where data was skewed (absolute nutrient intakes and intake of food items) and paired sample t-test was used when data was normally distributed (intake of macronutrients). Correlation coefficients and classification into tertiles were used to rank subjects according to intake, and Bland-Altman plots were used to check the level of agreement. Results Mean age of the total sample (n=41) is 26 months. Estimated intake of 13 out of 17 food groups were not significantly different between the SFFQ and the WFD; four food groups showed a significant difference; cereals, cakes, infant food and supplements. Median differences show underestimation of cereals and cakes from the SFFQ, while intake of infant foods and supplements were overestimated by the SFFQ. Fruit intake was marginally significantly different between the two methods and median difference showed overestimation with the SFFQ. Correlation coefficients for intake of main food groups ranged from -0.01 (fish) to 0.67 (supplements); 13 out of 17 correlations are significant. Subjects correctly classified in the same tertile of food intake ranged from 24 % for fish to 56 % for edible fats, with a median value of 46 %. The proportion of grossly misclassified subjects ranged from 2 % to 24 % with a median of 12 %. Energy requirements for the group were calculated using the Nordic Nutrition Recommendations (2004). Bland-Altman plot show large variation in energy intakes, between the two methods. Energy percentage intake of fibre and added sugar were significantly different between the SFFQ and the WFD. There was an underestimation of added sugar by the SFFQ and an overestimation of fibre compared to the WFD. Intake of total energy, percentages of energy from protein, total fat, saturated fat, monounsaturated fatty acids, polyunsaturated fatty acids, and carbohydrates were not significantly different between the SFFQ and the WFD. Comparing absolute intakes of macro- and micronutrients from the SFFQ and the WFD, show that for 16 out of 19 nutrients, there was no significant difference between the two dietary methods. Average (median) intake of fibre and added sugar were significantly different between the WFD and the SFFQ. The proportion of subjects correctly classified into tertiles ranged from 32 % (iron and saturated fat) to 56 % (vitamin A and C) with a median value of 45 %. The proportion of grossly misclassified subjects ranged from 7 % (calcium, added sugar and fibre) to 27 % (iron). Correlation coefficients ranged from -0.19 (iron) to 0.55 (added sugar) with a median of 0.19. Conclusion The results from the present study show that underestimation is as frequent as overestimation among individuals filling in the SFFQ. Moreover, the study indicates that the SFFQ may be a valuable tool for measuring average intake of several nutrients and food items. The capability of the questionnaire to rank children according to intake of nutrients and food items was rather low.