Author Affiliations: Centers for Disease Control and Prevention, Atlanta, Georgia (Drs Yang, Kuklina, Hong, and Khoury, and Mr Liu); Department of Epidemiology, Emory University, Atlanta (Dr Flanders); and Departments of Nutrition and Epidemiology, Harvard School of Public Health, Boston, Massachusetts (Dr Hu).
We appreciate the letters by Jürgens and Cohen et al regarding our article.1 Compared with earlier analyses,2 we included an additional follow-up (2000-2006), leading to substantially improved statistical power. We also estimated usual intakes of sodium and potassium using validated National Cancer Institute methodology that allows to adjust for day-to-day intake variation, minimizing bias in estimation of nutrient-health associations.3 As expected, a single 24-hour recall led to substantial misclassifications; approximately 40% and 32% of subjects were reclassified by comparing quartiles of day 1 vs the usual sodium and potassium intakes, respectively (eTable 1). In addition, our findings were consistent with another study using the 24-hour urinary excretion of sodium and potassium, which is a more reliable measurement of the intake.4 Although some of the associations in our study appear to be stronger in certain subgroups such as overweight individuals, none of the test results for interactions were significant.
Yang Q, Liu T, Kuklina EV, Flanders WD, Hong Y, Khoury MJ, Hu FB. Method of Estimating Sodium Intake and Its Possible Influence on NHANES III Outcome—Reply. Arch Intern Med. 2011;171(22):2063-2064. doi:10.1001/archinternmed.2011.591