to analyze data for this study are described in Appendix E. Information from the analyses is summarized below, and more detailed data tables can be found in Appendix F.

Although data based on the “disappearance” of sodium in the food supply, as described in Chapter 2, can provide some information, two general methods of assessing the population’s intake of sodium are considered to provide reasonably accurate estimates: (1) dietary self-reports (interviews, food records, diaries, food frequency questionnaires of individuals) and (2) urinary sodium measures of individuals.

The more accurate and reliable method of estimating sodium intake is the analysis of urine collected during a 24-hour period, which reflects about 90 percent or more of the ingested sodium (Clark and Mossholder, 1986; Luft et al., 1982; McCullough et al., 1991; Schachter et al., 1980). However, such measures are not currently included in national surveys carried out in the United States. Therefore, available information on the U.S. population’s sodium intake is based currently on national survey data derived from self-reported dietary intake of respondents. These large-scale national surveys provide representative estimates for the total population and large race/ethnic subgroups. However, NHANES data sets from 2003–2004 and 2005–2006 were combined for this study to provide larger sample sizes for subgroup analysis (see Appendix F). Clinical trials and smaller-scale studies can also provide dietary information for subgroups or special populations that cannot be gleaned from national surveys, but these cannot be relied upon to be representative.

For population-level or group intake estimates, multiple 24-hour dietary recalls are the preferred method (IOM, 2000). Other methods are feasible, but require greater respondent effort and may alter behavior (e.g., food records and diaries) or overestimate food and energy intake (e.g., food frequency questionnaires) (Thompson and Subar, 2008). The strengths of the 24-hour dietary recall include the use of a standardized protocol to quantify the types and amounts of foods consumed over the course of a day, reduced respondent burden, and the provision of valid dietary intake estimates for groups and usual nutrient intake if two or more 24-hour recalls are collected for at least a subsample of the group. Also, individual intake data permit calculation of intake distributions for groups so that the prevalence of high and low intake can be estimated. Additionally, they reflect the sodium content of foods as consumed.

The major limitation of any dietary intake method is that there is some degree of misreporting and measurement error (Thompson and Subar, 2008). For example, overweight persons may underreport intake, omitting certain foods or reducing the reported amounts; furthermore, parents may