Reproducibility and validity of healthy dietary indices derived from 24-hour dietary recalls

Background Healthy dietary indices are commonly used to investigate diet-disease associations, but the reproducibility and validity of these indices from 24-h dietary recalls (24HDRs) are undetermined. We aimed to investigate optimum number of 24HDRs for measuring healthy dietary indices including the Alternate Mediterranean Diet (AMED), Dietary Approaches to Stop Hypertension (DASH), and Healthy Eating Index-2015 (HEI-2015) in epidemiological studies. Methods From July 2021 to July 2022, 432 participants (median age: 48y, 57.2% women) completed 12-day 24HDRs, comprising four quarterly sets of three consecutive 24HDRs (two weekdays and one weekend day in each season). Energy-adjusted intraclass correlation coefficient (ICC) was used to evaluate reproducibility, while energy-adjusted deattenuated Spearman correlation coefficient (rc) were used to examine validity by comparing the estimates against those from the average of 12 24HDRs. Results The reproducibility and validity coefficient increased with the increasing number of 24HDRs. We observed a “moderate” reproducibility for three non-consecutive 24HDRs in assessing AMED, DASH and HEI-2015 (ICCs: 0.40-0.43). Healthy dietary indices from one 24HDR showed “moderate” validity (rc: 0.42-0.55). Higher body mass index, female sex, and autumn/winter of administration were associated with low reproducibility of AMED and/or HEI-2015. Conclusion Our results support that three non-consecutive 24HDRs provide acceptable estimates of AMED, DASH, and HEI-2015.