Reproducibility and validity of dietary polyphenols derived from 24-h dietary recalls in the Anhui Lifestyle Validation Study

While 24-h dietary recalls (24HDR) are increasingly used to assess polyphenol-disease associations, their performance for measuring total dietary polyphenols and their subclasses remains unknown. In the Anhui Lifestyle Validation Study, 432 community-dwelling adults completed 12-d dietary recalls, structured as four seasonal sets of three consecutive 24HDR (2 weekdays and 1 weekend day each) between July 2021 and July 2022. Reproducibility of dietary total polyphenols, flavonoids, lignans and phenolic acids assessed by different number of non-consecutive 24HDR was evaluated by intraclass correlation coefficient (ICC). Relative validity was evaluated by comparing estimates from different numbers of 24HDR with the internal reference using deattenuated Spearman's correlation coefficients (rc). The reproducibility from two 24HDR was weak for these four polyphenol categories (ICC: 0·10-0·39), while three 24HDR yielded a reasonable reproducibility for total polyphenols (ICC: 0·40, 95 % CI 0·31-0·50), flavonoids (ICC: 0·42, 95 % CI 0·33-0·51) and phenolic acids (ICC: 0·45, 95 % CI 0·16-0·66), but not for lignans (ICC: 0·11, 95 % CI 0·01-0·26). Increasing days up to 4-6 seemed not to appreciably improve the reproducibility (ICC range: 0·13-0·17 for lignans, 0·41-0·56 for others). The relative validity against the internal reference of these four polyphenol categories was moderate for two 24HDR (rc: 0·50-0·74) and improved for ≥ 3 24HDR (r c : 0·76-1·00). Older age, female sex and weekend administration were associated with lower reproducibility of three 24HDR-based assessments of phenolic acids or flavonoids. Overall, our results suggest that three non-consecutive 24HDR may have acceptable performance in measuring most dietary polyphenol intake except for lignans.