Biochemical validation of food frequency questionnaire-estimated carotenoid, alpha-tocopherol, and folate intakes among African Americans and non-Hispanic Whites in the Southern Community Cohort Study

Abstract

Few food frequency questionnaires (FFQs) have been developed specifically for use among African Americans, and reports of FFQ performance among African Americans or low-income groups assessed using biochemical indicators are scarce. The authors conducted a validation study within the Southern Community Cohort Study to evaluate FFQ-estimated intakes of alpha-carotene, beta-carotene, beta-cryptoxanthin, lutein/zeaxanthin, lycopene, folate, and alpha-tocopherol in relation to blood levels of these nutrients. Included were 255 nonsmoking participants (125 African Americans, 130 non-Hispanic whites) who provided a blood sample at the time of study enrollment and FFQ administration in 2002-2004. Levels of biochemical indicators of each micronutrient (alpha-tocopherol among women only) significantly increased with increasing FFQ-estimated intake (adjusted correlation coefficients: alpha-carotene, 0.35; beta-carotene, 0.28; beta-cryptoxanthin, 0.35; lutein/zeaxanthin, 0.28; lycopene, 0.15; folate, 0.26; alpha-tocopherol, 0.26 among women; all P's < 0.05). Subjects in the top decile of FFQ intake had blood levels that were 27% (lycopene) to 178% (beta-cryptoxanthin) higher than those of subjects in the lowest decile. Satisfactory FFQ performance was noted even for participants with less than a high school education. Some variation was noted in the FFQ's ability to predict blood levels for subgroups defined by race, sex, and other characteristics, but overall the Southern Community Cohort Study FFQ appears to generate useful dietary exposure rankings in the cohort.

Figure 1.

Plots of the average levels of selected biochemical indicators (y-axis) across quintiles of food frequency questionnaire (FFQ)-estimated dietary intake (x-axis), Southern Community Cohort Study, 2002–2004. Points represent the average FFQ-estimated intake and the average biochemical indicator level for each FFQ-determined quintile of intake. Superimposed is the predicted line from a univariate linear regression model with FFQ-estimated intake as the independent variable and the biochemical indicator as the dependent variable.