Group level validation of protein intakes estimated by 24-hour diet recall and dietary questionnaires against 24-hour urinary nitrogen in the European Prospective Investigation into Cancer and Nutrition (EPIC) calibration study

Abstract

A calibration approach was developed to correct for systematic between-cohort dietary measurement errors in the European Prospective Investigation into Cancer and Nutrition (EPIC), a large multicenter cohort study. To validate the 24-h diet recalls (24-HDRs) as reference measurements for between-cohort calibration, we estimated the agreement between center mean nitrogen (N) and total energy intakes and mean 24-h urinary N. Similar analyses using N and energy intake data from different dietary questionnaires (DQs) used at study baseline were conducted to estimate the effect of the calibration approach. This study was conducted between 1995 and 1999, and involved 1103 volunteers of both genders from 12 centers participating in European Prospective Investigation into Cancer and Nutrition. Pearson's correlation coefficients were weighted for study center sample size. When both genders were considered together (n = 22), the correlation coefficients between the center mean log-transformed urinary estimates and the center mean log-transformed dietary N estimates from the 24-HDRs were 0.86 and 0.94 after exclusion of outliers. The corresponding correlation with the DQs was 0.53. When center mean total energy intakes were regressed on center mean urinary N, the correlation remained slightly higher with 24-HDRs (0.91; 0.95 after exclusion of outliers) than DQs (0.86). When stratified by gender, these correlations were systematically higher in men than women with both dietary methods. The beta regression coefficients were not significantly different from 1 when mean N (or total energy intakes) from 24-HDR or DQ were regressed on urinary estimates, except with N from 24-HDRs in men and, in most cases, after adjustment for age, body mass index, and sex with both genders together. This suggests that overall the systematic bias across centers is of uniform magnitude. Although relatively high correlations were observed between urinary N and both dietary methods in men, the errors in DQs tend to vary in both directions (under- and over-reporting) in contrast with 24-HDRs in women. This observation may have implications on the dietary measurement error characteristics and support the potential benefit of between-cohort calibration.