Lifestyle interventions influence relative errors in self-reported diet intake of sodium and potassium

PURPOSE: To characterize the distribution of errors in self-reported sodium and potassium dietary intakes relative to mure objective urine measures am ong participants receiving lifestyle interventions. METHODS: We analyzed longitudinal data from 900 individuals with hypertensi on who had been enrolled in a randomized controlled clinical trial to estab lish with ther usual care or three lifestyle interventions (weight loss, so dium reduction, and combined weight loss and sodium reduction) could effect ively substitute for phamacotherapy. Repeated standardized 24-hour diet rec alls and 24-hour urine collections were collected over up to three years of follow-up to estimate sodium and potassium intakes. By contrasting self-re ported and urine-based sodium and potassium data collected before and durin g interventions, we examined the relative impact of intervention assignment on estimated intakes, repeatability, and multivariate measurement error. RESULTS: Relative to urine based measures, mean self-reported sodium intake s were biased about 10% lower among participants assigned to combined weigh t loss and sodium reduction, hut were unaffected by the other interventions . The repeatability of self-report measures increased slightly with time, p articularly among participants assigned to sodium interventions. Errors in self-reported sodium and potassium intakes were correlated before the start of the intervention, but became uncorrelated among individuals assigned to sodium restriction interventions. CONCLUSIONS: Lifestyle interventions may influence not only diet intake, hu t also the measurement of diet intake. Ann Epidemiol 2001;11:85-93. (C) 200 1 Elsevier Science Inc. All rights reserved.