OBJECTIVE: The objective of this study was to compare estimated nutrient intakes from 3-day food records vs Food Frequency Questionnaire (FFQ) in a large cohort of individuals living with human immunodeficiency virus (HIV) infection. DESIGN: Dietary data from 315 HIV-positive participants enrolled in a longitudinal cohort study were collected. Nutrient intake data estimated from the Block FFQ were compared with multiple 3-day food records done over the same time period. SUBJECTS/ SETTING: Participants enrolled in Nutrition for Healthy Living, an ongoing cohort study based in Boston, MA, were studied. STATISTICAL ANALYSES PERFORMED: Using 3-day food records as the gold standard, we compared estimated nutrient intakes between the two assessment methods by (a) median of individual differences in intake, (b) correlation coefficients, (c) quintile ranking, and (d) intakes less than the Dietary Reference Intakes. Nutrient intake estimates between the two methods were compared using Spearman, Pearson, and deattenuated correlation coefficients. RESULTS: Median reported intakes of all macronutrients and most micronutrients were significantly less on the FFQ for both men and women. Deattenuated correlations were less than 0.5 for 86% of the men and for 68% of the women. Mean percent agreement for quintile ranking was 30%. CONCLUSIONS: In this large cohort of HIV-infected adults, we found significant differences between FFQs and food records in estimates of absolute dietary intakes, correlation coefficients, quintile rankings, and proportions of subjects consuming less than the Dietary Reference Intakes. These findings have implications for the interpretation and application of dietary intake data from FFQs in specialized populations.
OBJECTIVE: The objective of this study was to compare estimated nutrient intakes from 3-day food records vs Food Frequency Questionnaire (FFQ) in a large cohort of individuals living with human immunodeficiency virus (HIV) infection. DESIGN: Dietary data from 315 HIV-positive participants enrolled in a longitudinal cohort study were collected. Nutrient intake data estimated from the Block FFQ were compared with multiple 3-day food records done over the same time period. SUBJECTS/ SETTING:Participants enrolled in Nutrition for Healthy Living, an ongoing cohort study based in Boston, MA, were studied. STATISTICAL ANALYSES PERFORMED: Using 3-day food records as the gold standard, we compared estimated nutrient intakes between the two assessment methods by (a) median of individual differences in intake, (b) correlation coefficients, (c) quintile ranking, and (d) intakes less than the Dietary Reference Intakes. Nutrient intake estimates between the two methods were compared using Spearman, Pearson, and deattenuated correlation coefficients. RESULTS: Median reported intakes of all macronutrients and most micronutrients were significantly less on the FFQ for both men and women. Deattenuated correlations were less than 0.5 for 86% of the men and for 68% of the women. Mean percent agreement for quintile ranking was 30%. CONCLUSIONS: In this large cohort of HIV-infected adults, we found significant differences between FFQs and food records in estimates of absolute dietary intakes, correlation coefficients, quintile rankings, and proportions of subjects consuming less than the Dietary Reference Intakes. These findings have implications for the interpretation and application of dietary intake data from FFQs in specialized populations.