Agreement between a brief food frequency questionnaire and diet records using two statistical methods

Area of Science:

• Nutrition Science
• Dietary Assessment Methods

Background:

• Accurate dietary assessment is crucial for understanding nutrient intake and health outcomes.
• Food Frequency Questionnaires (FFQs) are commonly used but require validation for specific nutrients like beta carotene (BC) and retinol.

Purpose of the Study:

• To evaluate the intra- and inter-method reliability of a semi-quantitative FFQ for BC and retinol intake.
• To compare the efficacy of limits of agreement (LOA) and correlation coefficients in assessing FFQ reliability.

Main Methods:

• A cross-sectional dietary intake study was conducted.
• Data from 83 adults (57 men, 26 women) aged 28-72 years in Western Australia were analyzed.
• The FFQ was administered twice (FFQ1, FFQ2) one year apart, with four 1-week diet records (DRs) collected during the interim.

Main Results:

• Intra-method reliability (FFQ1 vs. FFQ2) showed mean agreement of 120% for BC and 98% for retinol, with wide LOA.
• Inter-method reliability (FFQ2 vs. DR) showed mean agreement of 149% for BC and 63% for retinol, also with wide LOA.
• Correlation coefficients (0.36 for BC, 0.51 for retinol) between FFQ2 and DR were significant but did not align with LOA findings and varied by demographics and energy intake.

Conclusions:

• The limits of agreement method indicated less than ideal intra- and inter-method reliability for the FFQ.
• The FFQ demonstrated significant overestimation of BC and underestimation of retinol compared to diet records.
• Relying solely on correlation coefficients can be misleading; LOA provides a more comprehensive assessment of FFQ reliability and validity.