Relative Validity and Reproducibility of Dietary Measurements Assessed by a Semiquantitative Food Frequency Questionnaire among Chinese Healthy Adults

This study aimed to evaluate the relative validity and reproducibility of a semiquantitative food frequency questionnaire (SFFQ) in adult populations in China. Among the 49 recruited healthy participants (age range: 20–60 years), the relative validity of a 79-item SFFQ was assessed in two ways: (1) by comparing its dietary intake estimates with those from the average measurements of three inconsecutive 24 h dietary records (24-HDRs); and (2) by comparing its estimates of dietary fatty acids with the measured plasma levels of fatty acids. The reproducibility of the SFFQ was evaluated by a comparison of two SFFQ measurements from the same participants collected one year apart. In the relative validity study, the average Spearman correlation coefficient (r) was 0.27 among 18 prespecified food group intakes estimated from the SFFQ and the 24-HDRs; nevertheless, that of five food group intakes (e.g., red meat and seafood) was higher (all rs > 0.40, p < 0.05). In addition, a moderate correlation between the SFFQ estimate of polyunsaturated fatty acid intakes (energy-adjusted percentage of total fatty acids) and its plasma level was observed (r = 0.42, p < 0.05). Regarding the one-year reproducibility of the SFFQ-assessed intakes, the average rank intraclass correlation coefficient (ICC) was 0.35 for the 18 food group estimates. In particular, moderately reproducible estimates of seven food group intakes (e.g., refined grains and red meat, all ICCs ≥ 0.40, p < 0.05) were observed. In conclusion, the SFFQ provides valid and reproducible estimates of dietary intakes for various food groups in general and performs well as a potential tool for estimating habitual dietary intakes of some unsaturated fatty acids.