Validity of food frequency questionnaire-based estimates of long-term long-chain n-3 polyunsaturated fatty acid intake

To evaluate how long-term dietary intake of long-chain n-3 polyunsaturated fatty acids (LCn-3 PUFAs), estimated by repeated food frequency questionnaires (FFQs) over 15 years, is correlated with LCn-3 PUFAs in adipose tissue (AT). Subcutaneous adipose tissue was obtained in 2003-2004 (AT-03) from 239 randomly selected women, aged 55-75 years, after completion of a 96-item FFQ (FFQ-03). All participants had previously returned an identical FFQ in 1997 (FFQ-97) and a 67-item version in 1987-1990 (FFQ-87). Pearson product-moment correlations were used to evaluate associations between intake of total and individual LCn-3 PUFAs as estimated by the three FFQ assessments and AT-03 content (% of total fatty acids). FFQ-estimated mean relative intake of LCn-3 PUFAs (% of total fat intake) increased between all three assessments (FFQ-87, 0.55 +/- A 0.34; FFQ-97, 0.74 +/- A 0.64; FFQ-03, 0.88 +/- A 0.56). Validity, in terms of Pearson correlations between FFQ-03 estimates and AT-03 content, was 0.41 (95 % CI 0.30-0.51) for total LCn-3 PUFA and ranged from 0.29 to 0.48 for individual fatty acids; lower correlation was observed among participants with higher percentage body fat. With regard to long-term intake estimates, past dietary intake was also correlated with AT-03 content, with correlation coefficients in the range of 0.21-0.33 and 0.21-0.34 for FFQ-97 and FFQ-87, respectively. The correlations were improved by using average estimates from two or more FFQ assessments. Exclusion of fish oil supplement users (14 %) did not alter the correlations. These data indicate reasonable validity of FFQ-based estimates of long-term (up to 15 years) LCn-3 PUFA intake, justifying their use in studies of diet-disease associations.