United States: The National Health and Nutrition Examination Survey

Information about the U.S. National Health and Nutrition Examination Survey (NHANES) is available from the survey's main website,¹ and is therefore only summarized here. In the 1960s, the U.S. government initiated the National Health Examination Survey to assess the health status of individuals ages 6 months through 74 years. Nutritional intake was added as a survey component in the 1970s, beginning with the first NHANES, known as NHANES I (1971 to 1974). NHANES II covered the time period 1976 to 1980, and NHANES IIII encompassed 1988 to 1994. NHANES has reflected a continuous and standardized data collection based on a representative sample of the U.S. population and provides critical diet and health measures for federal program planning and policy making. The survey relies on the gold standard for dietary intake measures, two or more 24-hour dietary recalls per person (IOM, 2000). The U.S. Department of Agriculture's (USDA's) food composition database has provided the sources of information that allow the estimates of food intake collected in the NHANES to be translated into quantitative nutrient intake (Bodner-Montville et al., 2006; Briefel, 2006).

In 1999, the survey became a continuous program that has a changing focus on a variety of health and nutrition measurements to meet emerging needs;² the survey data are reported on the basis of 2-year periods. The survey now examines a nationally representative sample of about 5,000 persons each year. These persons are located in counties across the country, 15 of which are visited each year.³ The NHANES and related food intake surveys conducted by the USDA were integrated in 2002; at that time, the dietary reports from the integrated survey became known as What We Eat in America (WWEIA). For this report data for the 2003 to 2004 and 2005 to 2006 period were used because the data for the 2007 and 2008 period did not become available until after the committee had completed its deliberations. Calcium intake has been estimated since NHANES I. Intakes for vitamin D were first published in 2009 and currently are available for the 2003–2006 survey period. The NHANES is said to “follow the sun” in that the survey is generally conducted in the southern states during the winter months and in the northern states in the summer months.

The NHANES is unique in that it collects and tracks both total intake and health measures in a national sample of Americans, and provides an important aspect of the nation's health monitoring system. As would be expected, total intake estimates are limited by survey respondents' abilities to accurately report foods and amounts consumed and by the accuracy, specificity, and timeliness of the food composition databases linked to foods reported in the survey. Respondents are also prone to under-reporting intake (IOM, 2000). Issues related to estimation of usual intake from WWEIA-NHANES have been reviewed by others (Dwyer et al., 2003), including the challenges of updating food composition tables and addressing the under-reporting of intake amounts by participants.

In the case of nutrient intake data for the United States, calcium and vitamin D intake estimates from the WWEIA report series⁴ have formed the basis of an expanded analysis conducted and made available by the National Cancer Institute (NCI) of the National Institutes of Health. The NCI analysis was used in this report, as described below.

Canada: Canadian Health Measures Survey and Canadian Community Health Survey

Data relevant to the Canadian intake of calcium and vitamin D from foods, as well as measures of serum 25OHD concentrations¹⁰ for a representative sample of Canadians, are available from national surveys conducted by the Government of Canada. These are described below.

Approach Used

An earlier IOM committee addressed applications of the DRIs in dietary assessment and described statistical approaches to estimating the prevalence of inadequate intakes, specifically the probability approach and a shortcut to the probability approach called the Estimated Average Requirement (EAR) cut-point method (IOM, 2000). These approaches are based on a distribution of usual intakes, and by definition the prevalence of inadequate intakes for a group is the proportion of the group with intakes below the median requirement (or EAR). The 2000 IOM report also points out that it is inappropriate to compare usual nutrient intakes with the Recommended Dietary Allowance (RDA), because this approach will lead to estimates of inadequacy that are too large.

Based on the 2000 IOM report cited above (IOM, 2000), whenever possible, the assessment of apparent dietary adequacy should consider relevant biological parameters. In the case of vitamin D, an important biological parameter reflective of dietary exposure—serum 25OHD concentrations— was available and could be compared to values that the committee estimated to be approximately equivalent to an EAR or an RDA. However, the existing statistical models provided in the 2000 IOM report (IOM, 2000) address only dietary intake data and do not provide a basis for considering a biological parameter such as serum measures in order to specify the prevalence of inadequate intakes in population groups. Further, the apparent discrepancy between the intake data for vitamin D, as described below, and the biological parameter was also concerning and decreased the confidence in the appropriateness of estimating prevalence of inadequacy based on a distribution of vitamin D intakes. Therefore, a descriptive rather than an analytical approach is used for the vitamin D intake assessment.

U.S. Calcium Intake

Estimated calcium intakes from food sources only, by intake percentile groups, are shown as bar graphs in Figure 7-1. The prevalence of dietary inadequacy for a group can be estimated by the proportion of the group with intakes less than the EAR (IOM, 2000). The 5th percentile of intake for children 1 to 3 years of age is approximately equal to their EAR of 500 mg/day, implying a low prevalence of inadequacy (less than 5 percent). However, for all other age and gender groups of children, the prevalence is at least 25 percent, because intake at the 25th percentile is below the EAR. For adults, the prevalence of inadequacy from food sources alone is high.

FIGURE 7-1

Estimated calcium intakes in the United States from food sources only, by intake percentile groups, age, and gender. NOTE: F = female; M = male; y = years. SOURCE: NHANES 2003–2006 as analyzed by Bailey et al. (2010). Data used to create figure (more...)

As shown in Figure 7-2, the addition of information about calcium intake from supplements to the data set, thereby allowing an estimate of total intake, appears to impact primarily women over 50 years of age. All life stage groups show a slight increase when supplements are taken into account, but women 51 to 70 years of age demonstrate an estimated median total calcium intake (i.e., from foods plus supplements) of 1,044 mg/day compared with 755 mg/day from foods alone, while median intake for women more than 70 years of age was 983 mg/day compared to 706 mg/day without supplements. Thus, when intake from supplements is considered, the prevalence of dietary inadequacy for older women is approximately 50 percent (i.e., intake at the 50th percentile is approximately equal to the EAR of 1,000 mg/day calcium in Figure 7-2, Panel B).

FIGURE 7-2

Estimated total calcium intakes in the United States from food and supplements, by intake percentile groups, age, and gender. NOTE: F = female; M = male; y = years. SOURCE: NHANES 2003–2006 as analyzed by Bailey et al. (2010). Data used to create (more...)

Total calcium intakes at the 95th percentile are below the UL of 2,000 mg of calcium per day for most of the adult life stage groups, implying that less than 5 percent are at risk of excessive intake. The exception is older women, who have estimated total calcium intakes at the 95th percentile of 2,364 mg/day for those 51 to 70 years of age, and 2, 298 mg/day for those more than 70 years of age, so more than 5 percent are at risk of excessive intakes. By contrast, the 95th percentile of calcium consumption from food sources alone are 1,353 and 1,337 mg/day for these two life stage groups, respectively.

Canadian Calcium Intake

Estimates of calcium intake from foods for Canadians appear to be similar to those reported for the United States, although the median intake drops at a younger age for men, at the 31- to 50-year life stage as compared to the 51- to 70-year life stage in the United States (Figure 7-3). Overall, estimated intakes of calcium from foods in Canada appear to be slightly lower than those reported for the United States. Although differences in survey methodologies could be responsible for some of the difference, the surveys use very similar methodologies and work to ensure uniformity as much as possible. A more likely possibility is that the differences are attributable to food fortification practices. In Canada, calcium may only be added to a limited number of foods. Flour, cornmeal, plant-based beverages, and orange juice may be fortified with calcium, but not breakfast cereals and bread. However, discretionary fortification with calcium is widespread in the United States and can encompass breakfast cereals, breads, and an array of beverages.

FIGURE 7-3

Estimated calcium intakes in Canada from food sources only, by intake percentile groups, age, and gender. NOTE: F = female; M = male; y = years. SOURCE: Statistics Canada, Canadian Community Health Survey (CCHS), Cycle 2.2, Nutrition 2004. Data used to (more...)

At the time of this study, only intake data for foods were available for Canadians; estimates of total calcium intake (i.e., foods plus supplements) had not yet been compiled. After the completion of the study, information on total intake was published (Garriguet, 2010).

U.S. Vitamin D Intakes and Serum 25OHD Concentrations

Figure 7-4 shows U.S. vitamin D intake from foods alone. Median vitamin D intake levels for males ranged from 272 to 396 International Units (IU)/day depending upon life stage group. For females, median vitamin D intakes spanned between 160 and 260 IU/day. When intake from supplements is considered to provide total intakes (Figure 7-5), all life stage groups for both male and female Americans show a slight increase in values. The most marked increase is among older women, as was the case for calcium. For women 51 to 70 years of age, median intake of vitamin D from both food and supplements increases to 308 IU/day, compared with vitamin D intake from foods alone, at 140 IU/day. For women more than 70 years of age, the increase in median intake associated with supplement use is an additional 196 IU/day (356 IU with supplements vs. 160 IUs from foods alone).

FIGURE 7-4

Estimated vitamin D intakes in the United States from food sources only, by intake percentile groups, age, and gender. NOTE: F = female; IU = International Units; M = male; y = years. SOURCE: NHANES 2005–2006 as analyzed by Bailey et al. (2010). (more...)

FIGURE 7-5

Estimated total vitamin D intakes in the United States from food and supplements, by intake percentile groups, age, and gender. NOTE: F = female; IU = International Units; M = male; y = years. SOURCE: NHANES 2005–2006 as analyzed by Bailey et (more...)

As shown in Figure 7-5, the 95th percentiles for total vitamin D (foods plus supplements) for males and females range between 568 and 940 IU/day, with both this high and low value found among the female life stage groups. Persons in the 95th percentile for total intake did not appear to exceed the UL for their group.

The comparison between vitamin D intake estimates and serum 25OHD concentrations is worthy of note, but it is important to recognize that this comparison, although interesting, is somewhat problematic because the only possible comparison is based on group means, rather than on data linked to individuals. Moreover, as pointed out previously (IOM, 2000; Dwyer et al., 2003), estimates of intake tend to reflect an underestimation. With these caveats, the comparison is presented in Table 7-3. Shown are the average intakes for the various life stage groups, along with the average serum 25OHD levels for those life stage groups. For this table, serum 25OHD concentration data from the 2005 to 2006 surveys were used rather than those from the 2003 to 2006 data set because intake estimates for total vitamin D (i.e., the NCI method) are currently available only for the 2005 to 2006 data.

TABLE 7-3

Mean Vitamin D Intake and Mean Serum 25OHD Concentrations for the United States, 2005–2006, by Life Stage Groups.

Assuming that a serum 25OHD level of 40 nmol/L is consistent with a desirable median intake,¹⁶ the comparison would suggest that, on average, persons may be experiencing intakes below the reference values, but are exhibiting serum 25OHD levels above 40 nmol/L. In fact, all are above the 50 nmol/L concentration, the level associated with the RDA. There is an additional factor to consider in this comparison, in that the NHANES data are generally collected during the summer months in the northern regions of the United States and in the winter months in the southern regions; this introduces the variable of sun exposure into the comparison in that it decreases the likelihood that individuals surveyed will be experiencing low levels of sun exposure. As an informal conceptual check, it is possible to adjust these data so as to roughly simulate a reduction in serum 25OHD levels consistent with the difference between the summer zenith and the winter nadir. Specifically, if the estimate that there is a one-third difference in serum 25OHD levels between the winter nadir and summer zenith as described in Chapter 3 is applied to this comparison, reducing these serum 25OHD levels by one-third results in a range of serum 25OHD levels from a low of 37 nmol/L (women > 70 years) to a high of 47 nmol/L (found in four life stage groups), which are still very close to, and in many cases above, a 40 nmol/L concentration consistent with an estimated average required intake. Given the observation made in Chapter 5 that the seasonal decline during the winter may differ between those with high and low initial baseline values, the correction applied using a 30 percent reduction may overestimate the decline in those at lower baseline 25OHD levels below 50 nmol/L. Moreover, this adjustment is excessive because for those persons living in the southern United States, their serum 25OHD measures were taken generally during the winter, not summer, months; the one-third reduction is therefore an over-correction in this case. However, because it is not possible using the data available to the committee to distinguish between values taken in the summer in northern areas and in the winter in southern areas, the adjustment cannot be further refined.

Although serum 25OHD levels from the 2005 to 2006 period in the United States are the data used for Table 7-3 because total intake data are available only for 2005 to 2006, serum 25OHD levels are available for the 2003 to 2006 data set, the two most current surveys, which, when combined, provide a larger data set. For comparison, these are shown in Table 7-4 and appear to reflect values very similar to those reported for 2005 to 2006 alone. No effort has been made to consider vitamin D intake for this period (2003 to 2006) because only data from the WWEIA are available for 2003 to 2006, which would not provide information on total intake (foods plus supplements).

TABLE 7-4

Mean Serum 25OHD Concentrations for the United States, 2003–2006, by Life Stage Group.

Canadian Vitamin D Intakes and Serum 25OHD Concentrations

As previously mentioned, the available Canadian survey data provided information on vitamin D intake from foods alone; quantified information on vitamin D supplement intake among Canadians, and thus on total intake from foods and supplements, was not available at the time of the study. Figure 7-6 outlines the estimated intakes of vitamin D from foods alone, which overall tend to be slightly higher than those reported for the United States. Median vitamin D intakes ranged from a low of 176 IU/day (women 51 to 70 years) to a high of 264 IU/day (boys 9 to 13 years). Similar to the U.S. population, persons in the 95th percentile of intake of vitamin D from foods would be expected to be considerably below the UL for their life stage.

FIGURE 7-6

Estimated vitamin D intakes in Canada from food sources only, by intake percentile groups, age, and gender. NOTE: F = female; IU = International Units; M = male; y = years. SOURCE: Statistics Canada, Canadian Community Health Survey (CCHS), Cycle 2.2, (more...)

Comparison between mean intakes of vitamin D and mean serum 25OHD concentrations for Canadians is problematic. For Canada, intake estimates are provided for the survey year 2004 based on the CCHS, whereas the serum 25OHD concentrations available reflect data from the 2007 to 2009 CHMS. The mean serum 25OHD levels for Canadians are shown in Table 7-5, and no effort has been made to compare these with intake estimates. As a general matter, average serum 25OHD concentrations of Canadians are above both the 40 and 50 nmol/L concentration levels. Although average intakes of vitamin D among Canadians from foods alone (i.e., not taking into account supplements) are less than the EAR, measures of serum 25OHD levels are well above the 40 nmol/L level consistent with the EAR. Again, as described earlier, the ability to interpret the prevalence of inadequacy based on serum 25OHD concentrations using the methodology as established in the 2000 IOM report (IOM, 2000) is unclear.

TABLE 7-5

Serum 25OHD Levels for Canadians by Percentile Group, Age, and Gender.