While several studies have examined the validity of using an FFQ to estimate dietary PUFA intake by comparing the FFQ to biomarkers such as adipose tissue and erythrocyte membranes in adults, no studies have considered a population of children where data are collected via FFQ using parental report. It is useful to examine the utility of an FFQ in children regarding this question because they may have low or infrequent intakes of fish, in particular, which may not be adequately picked up by 24-hour recall or diet record that are collected only periodically for a small number of days.
We compared intake of fatty acids as measured by multiple FFQ with fatty acid content of erythrocyte membranes over the same time periods for a cohort of healthy children aged 1 to 11 years, at increased risk for type 1 diabetes. Intakes of omega-3 fatty acids and, in particular, marine PUFAs were strongly associated with levels of these PUFAs in the erythrocyte membranes.
Our results are consistent with other studies that compared PUFA intake as assessed by an FFQ to biomarkers among adults. One such study compared fatty acid composition of adipose tissue to dietary intake (as a percentage of total fat) as estimated from an FFQ for 86 adults and reported correlations of 0.42 for DHA and 0.55 for EPA (Tjonneland et al., 1993
). Similar to our study, they found that non-marine omega-3 fatty acids and omega-6 fatty acids were less correlated. Fish intake, as assessed by FFQ, was significantly associated with EPA and DHA in platelet phospholipids in adult males (Li et al., 2001
). High correlations were found in a study comparing intake of PUFAs (as a percent of total fat) assessed by an FFQ to serum phospholipids levels in adults in a high fish-consuming population, for EPA (ρ = 0.59) and DHA (ρ = 0.49) (Kobayashi et al., 2003
These results all suggest that the Willett FFQ is good at picking up omega-3 fatty acids, especially marine PUFAs. Our Pearson correlation estimates indicated that the FFQ is not as good at detecting omega-6 fatty acids, and in particular arachidonic acid. One possible explanation is that there may be unknown upper limits to omega-3 and omega-6 fatty acid levels in erythrocyte membranes. Such an upper limit could result in a lower membrane fatty acid level than expected given fatty acid exposure; thereby underestimating the correlation of the membrane content with levels measured using FFQ.
Another possible explanation is that the FFQ is adequately measuring omega-6 fatty acid intake, but that this intake would not be reflected in the erythrocyte membrane content because of the preferential incorporation of omega-3 fatty acids into the membranes (Willett, 1998
). Membrane fatty acid composition is determined by the interplay between available fatty acids from a dietary source and further metabolism of these fatty acids, such as the competition between n-6 and n-3 fatty acids. To look at this issue, we estimated Pearson correlation using the mixed model methods described above for omega-6 total fatty acid by quartiles of marine fatty acid (lower two quartiles versus upper two quartiles) and found that the correlation between intake and membrane levels was higher among the lower two marine PUFA quartiles (ρ = 0.26) than among the highest two quartiles (ρ = 0.07), suggesting that this may be the case. Romon et al. (1995)
found a negative correlation between fish intake and RBC arachidonic acid. In experimental studies, increased n-3 fatty acid intake decreased arachidonic acid content of membrane phospholipids (Wander et al., 1991
We observed slightly stronger correlations between erythrocyte membrane fatty acid content and intake in 6–11 year old children compared with 1–5 year olds. This may reflect a better ability of the FFQ to record more adult-like diets compared with early childhood diets. It is not known whether there could also be a biological reason for the differences in correlation by age group
This study fills a gap in the literature by validating the FFQ among a cohort of children. Also, the methods used strengthen the conclusions because multiple measures were used for each child while accounting for within subject variation over time. One limitation of this study was that the cohort included children who were all at higher risk for type I diabetes compared to the general population. These children and their families all knew they were at higher risk for diabetes and therefore, their diets may not have been representative of children the same age in the general population. However, very few of these children were autoimmune (n=72), of whom 23 went on to develop diabetes over the course of the study. Moreover, when these children were removed from the analyses, we saw similar results as those presented herein (data not shown).
In conclusion, our validation study finds that the FFQ performs well in measuring intake of marine PUFAs among children.