Reproducibility and validity of low-carbohydrate diet and low-fat diet scores derived from 24-hour dietary recalls.

Abstract

The reproducibility and validity of low-carbohydrate diets (LCD) and low-fat diets (LFD) from 24-hour dietary recall (24HDR) remain unclear. We evaluated the performance of 24HDR-based assessments of 6 dietary patterns (overall, healthful, and unhealthful LCD/LFD) to determine the optimum number of 24HDR in epidemiologic studies. Participants (n = 432, 185 men and 247 women) aged 18-84 years from the Anhui Lifestyle Validation Study (ALVS) completed 4 quarterly 3 consecutive 24HDRs (2 weekdays and 1 weekend day) during July 2021 to July 2022. Reproducibility was evaluated by intraclass correlation coefficient (ICC). Validity was evaluated by comparing the estimates against those from the average of 12 24HDRs using deattenuated Spearman correlation coefficient (rc). Reproducibility and validity increased with an escalating number of 24HDRs. Dietary pattern scores from 2 24HDRs showed moderate reproducibility (ICC: 0.40-0.47), except unhealthful LFD with the ICC of 0.37, while 4 24HDRs yielded a moderate reproducibility for all dietary patterns (ICC: 0.43-0.59). The validity of the 6 dietary patterns was moderate for 1 24HDR (rc: 0.44-0.58) and moderate or strong for ≥2 24HDRs (rc: 0.56-0.98). Two 24HDRs show reasonable performance in measuring most LCD/LFD scores, and 4 24HDRs are sufficient for all dietary patterns.