点击显示 收起
1 From the Chronic Disease Nutrition Branch (HMB, CG, and MR) and the Maternal and Child Nutrition Branch (MEC), Division of Nutrition and Physical Activity, Centers for Disease Control and Prevention, Atlanta, GA
2 The findings and conclusions in this report are those of the authors and do not necessarily represent the views of the funding agency. 3 Reprints not available. Address correspondence to H Michels Blanck, 4770 Buford Highway NE, MS K-26, Atlanta, GA 30341-3717. E-mail: hblanck{at}cdc.gov
ABSTRACT
Background: Patients with hemochromatosis are instructed to avoid taking supplemental iron. Whether supplemental iron intakes lead to higher iron status among healthy persons remains less clear.
Objective: The objective was to ascertain whether supplemental iron intakes are associated with increases in iron transport (transferrin saturation) and stores (serum ferritin) among US adults aged 19 y.
Design: We analyzed data for 5948 adults from whom a fasting serum sample was collected during the third National Health and Nutrition Examination Survey. We used multivariable linear regression and analysis of variance to assess the association of supplemental iron intake with iron transport and stores among men (aged 19–30 y or >30 y) and women (nonpregnant premenopausal or postmenopausal); multiple comparison tests were also performed.
Results: Healthy adults who took supplements containing average daily amounts of iron at 3 times the recommended dietary allowance (RDA) did not have significantly higher iron transport or stores than did those who did not take supplements. In younger men, the intake of >32 mg Fe/d (>4x RDA) was associated with mean transport iron concentrations that were significantly higher than those in persons who took 0 to 24 mg Fe/d. In older men, the intake of >32 mg Fe/d (>4x RDA) was associated with mean iron stores that were significantly higher than those in persons who took 0 to 24 mg Fe/d; a similar result was observed in postmenopausal women, but it was of borderline statistical significance.
Conclusion: Supplement users should be made aware of the amount of iron necessary to satisfy dietary requirements and informed of the possible influence that excess iron intake can have on body iron stores and health.
Key Words: Iron supplements • ferritin • transferrin saturation • adults • National Health and Nutrition Examination Survey • NHANES • recommended dietary allowances • RDAs
INTRODUCTION
In persons with a genetic predisposition to absorb more iron than is normal, iron may accumulate in body tissues (iron overload) over many decades, which may lead to tissue and organ damage (1, 2). These patients have hemochromatosis and are instructed to avoid taking supplements containing iron (3). The issue of whether supplemental iron intake leads to higher iron status in a population-based sample of persons remains less certain.
Fleming et al (4) found that subjects aged 68–93 y in the Framingham Heart Study cohort who took 30 mg supplemental iron/d had significantly higher iron stores than did subjects taking no supplemental iron. Garry et al (5) found a similar result in a longitudinal study of persons aged 60–93 y who took 18 mg Fe/d. Liu et al (6) found a significant trend in plasma ferritin concentrations with increasing supplemental iron use among healthy postmenopausal women enrolled in the Nurses' Health Study, including a suggestion of elevated ferritin in those who consumed 21.5 mg Fe/d. These studies were conducted in the United States among populations who, on average, consumed more than the recommended dietary allowance (RDA) of iron from their diet. Current iron RDAs are 8, 18, and 8 mg/d for men aged >19 y, women aged 19–50 y, and women aged >50 y, respectively (7). However, previous research may not be generalizable to younger populations, to persons at greater risk of iron deficiency, or to persons who have high iron requirements, such as premenopausal women.
Previous data from the third National Health and Nutrition Examination Survey (NHANES III; 1988–1994) indicated that serum ferritin (SF) concentrations did not differ between persons who were in the lower and upper quartiles of total daily dietary intakes of iron (7). These data, however, excluded persons who consumed supplements. The objective of our analyses was to ascertain whether supplemental iron intakes are associated with changes in measures of iron transport or iron stores among relatively healthy US adults aged 19 y.
SUBJECTS AND METHODS
Study population
Data from NHANES III represent the total noninstitutionalized civilian US population aged 2 mo. The National Center for Health Statistics of the Centers for Disease Control and Prevention (CDC) collected the data through household interviews and physical examinations at a mobile examination center.
A detailed description of the NHANES III plan, operation, and sample design is provided elsewhere (8). Briefly, after being interviewed at home, participants were invited to attend 1 of 3 physical examination sessions. Those attending the morning session were asked to fast for 12 h before the session. Those attending the afternoon or evening sessions were asked to fast for 6 h. Examination data were available for 16955 participants aged 19 y.
Written informed consent was obtained from all participants. The study was approved by the institutional review board of the CDC.
Biochemical variables
The methods used to collect data on iron measures and other biochemical markers, including serum vitamin C, were reported previously (9). All iron assays were conducted at the NHANES laboratory in the CDC's National Center for Environmental Health. Serum iron concentration and total-iron-binding capacity were colorimetrically measured with an Alpken RFA analyzer (Alpkem, Clackamas, OR); a 1% thiourea solution was added to complex Cu++ to prevent copper interference. The percentage of transferrin saturation (TS) was calculated by dividing the serum iron concentration by the total-iron-binding capacity and multiplying by 100. The SF concentration was measured by using the Quantimmune IRMA kit (BioRad Laboratories, Hercules, CA).
Because ferritin is an acute phase protein, it can be an unreliable indicator of body iron stores if inflammation due to infection, liver disease, or malignancy is present (10). We therefore defined an inflammatory condition as the presence of any of the following: an abnormal white blood cell count, a high C-reactive protein concentration, or a high liver enzyme concentration. An elevated white blood cell count was defined as a count of >10.6 x 109/L in men and >11.0 x 109/L in women (11). A high serum C-reactive protein concentration was defined as a concentration of >1.0 mg/dL (12). A high liver enzyme concentration was defined as a concentration twice the upper limit of the reference range (1st–99th percentile in NHANES III): ie, a serum alanine aminotransferase concentration 80 U/L in men and 62 U/L in women, a serum aspartate aminotransferase concentration 74 U/L in men and 62 U/L in women, or an alkaline phosphatase concentration 234 U/L in both men and women.
Iron supplements
Participants' daily intake of supplemental iron was estimated from their responses to a series of questions, asked during the household interview, about the dietary supplements (vitamins or minerals or both) or antacids, including those prescribed and those not prescribed by a doctor, that they had taken within the previous month. Participants who answered that they were taking supplemental iron were asked to show the supplement containers and queried about dosages and duration of use. A detailed description of the questions and coding used appears in the NHANES III supplementary data file (13).
For the purpose of our analyses, we defined "iron supplement use" as the consumption of any supplement that contained iron. We estimated the participants' average daily intake of iron according to the duration of supplemental iron use. For example, if the total supplemental iron intake for 1 mo was estimated to be 4800 mg, the average daily intake was calculated as 160 mg Fe/d.
Dietary data
We included a number of dietary factors in our analysis because of their potential enhancement or inhibition of iron absorption. A food-frequency questionnaire was used to determine respondents' diet. Respondents were told to recall their diet over the past month and to report the frequency with which they ate certain foods. Their average weekly servings of red meat, a major contributor of heme iron, were ascertained from questions about hamburger, steaks, roast beef, and meatloaf, and their average weekly servings of caffeinated coffee and tea were ascertained from questions about those beverages. Separately from the food-frequency questionnaire, participants were asked about the number of alcoholic drinks they had consumed in the past 12 mo. Total energy intake (in kcal) was calculated from their responses to one 24-h dietary recall.
Nondietary data
Participants' age in years, race-ethnicity, current hormone replacement therapy status (yes or no), and blood donation in the past 12 mo (yes or no) were based on self-reports. Body mass index was calculated by dividing participants' measured weight (in kg) by measured height (in m2) (14). Menopausal status was determined from the following questions: "Have you had a period in the past 12 mo?" "Have you had a hysterectomy?" and "Were both ovaries removed, or only one?" Self-reported treatment of anemia was determined by an affirmative response to the question, "Are you now or in the past 3 mo have you been on treatment for anemia, sometimes called ‘tired blood’ or ‘low blood’ (include diet, iron pills, iron shots, or transfusions as treatment)?"
Sample exclusions
The first set of exclusion criteria were specific to the primary exposure (supplemental iron) and primary outcomes (TS percentage and SF concentration). In preliminary analyses, we found significant differences in iron measures between fasting and nonfasting participants (data not shown). Therefore, respondents whose data supplement use (n = 151) or serum iron marker (n = 872) were missing and those who did not provide a fasting serum sample (n = 6686) were excluded. Participants who reported anemia treatment (n = 1460) or who did not indicate they had undergone anemia treatment (n = 890) were excluded from the analysis, because anemia of various causes has been associated with altered TS percentages and SF concentrations (15, 16). Similarly, participants with hepatitis or without hepatitis status information (n = 1599) were excluded, because positive hepatitis status can be associated with liver damage that is attributable to concurrent viral disease and alcohol or drug exposure (17). Participants were considered to have an active hepatitis infection if they had hepatitis B core antibody, hepatitis B surface antibody, or hepatitis C antibody. In addition, women were excluded if they were pregnant (n = 299) or breastfeeding (n = 96) or if their pregnancy or breastfeeding status was unknown (n = 34). Participants lacking information on inflammatory conditions, serum vitamin C (mg/dL), alcohol consumption, total energy, body mass index, blood donations, menopause, or hormone replacement therapy were also excluded (n = 593). After these exclusions, the final analytic sample consisted of 5948 persons. Compared with the group that was excluded, the group in the final analytic sample had a significantly larger proportion of males (45.1% and 51.6%, respectively; P < 0.0001), a larger proportion of non-Hispanic whites (73.2% and 80.4%, respectively; P < 0.0001), a smaller proportion of non-Hispanic blacks (12.9% and 8.1%, respectively; P < 0.0001), and a smaller proportion of persons from other racial-ethnic groups (8.6% and 6.5%, respectively; P = 0.006).
Statistical analyses
For analytic purposes, we weighted data in all statistical analyses to account for the complex NHANES sample design. Analyses were done by using SAS software (version 9.1; SAS Institute, Cary, NC; 18) and SAS-callable SUDAAN software (version 9.0; Research Triangle Institute, Research Triangle Park, NC; 19). Because the distribution of SF values was positively skewed, we applied a natural logarithmic transformation to the measurements to normalize the distributions before examining differences in means. We set significance at P < 0.05 for all comparisons. We assessed whether age or menopausal status modified the association between iron supplement use and iron measures. For men, we chose an age of 30 y because of reports that median SF concentrations stabilize in men after age 32 y (20). We found a significant interaction between iron supplement use and age for TS in men (P = 0.0001) and a significant interaction between age and menopausal status for SF in both sexes (men: P = 0.0097; women: P = 0.0173). Therefore, we created 4 subgroups: premenopausal and postmenopausal women and men aged 19–30 y (younger men) and men aged >30 y (older men).
We categorized the intake of iron supplement by using the sex-specific RDA for iron. The iron RDA is 8, 18, and 8 mg/d for men aged 19 y, women aged 19–50 y, and women aged > 50 y, respectively (7). We used RDAs of 18 and 8 mg/d for premenopausal and postmenopausal women, respectively, because of the effect that monthly blood loss has on female iron stores (21). We therefore categorized estimated daily supplemental iron intakes into 5 categories: none, up to the RDA, 3x RDA, > 3x–4x RDA, and > 4x RDA. Therefore, for all men and postmenopausal women, our high-intake category corresponded to > 32 mg Fe/d, which does not differ significantly from the high-intake category of 30 mg Fe/d used by Fleming et al (4). For premenopausal women, our high-intake category corresponded to >72 mg Fe/d.
We used linear regression models to calculate least-squares multivariate conditional means (adjusted means) and accompanying 95% CIs for TS percentages and conditional geometric means and 95% CI for SF concentrations in each of the 5 categories of supplemental iron. We analyzed overall differences within subgroups by using analysis of variance (ANOVA) [in SUDAAN based on the Satterthwaite adjusted F test (22)]. If the ANOVA was significant, multiple comparison tests to ascertain the significant differences between category means were conducted by using Tukey's test (in SUDAAN by calculating the t statistic and testing against the critical values of the studentized range statistic). Covariates were included in multivariate modeling to control for possible confounding effects. In a secondary analysis that assessed whether inflammation due to acute or chronic disease altered the association between supplemental iron intake and iron status, we ran final models excluding the 840 persons with an inflammatory condition.
RESULTS
The sample consisted of 5948 persons divided into subgroups of younger men (n = 773), older men (n = 2235), premenopausal women (n = 1553), and postmenopausal women (n = 1387) (Table 1). In the previous month, 18.5% of younger men, 16.1% of older men, 22.6% of premenopausal women, and 21.8% of postmenopausal women were consuming supplements containing iron.
View this table:
TABLE 1. . Characteristics of eligible participants in the third National Health and Nutrition Examination Survey by group1
The median intake of iron in all supplement users was 17.8 mg/d (Table 2, Table 3). The proportion of users whose average iron intake was at or below the RDA was 18.1% of younger men, 21.7% of older men, 59.3% of premenopausal women, and 22.5% of postmenopausal women. The proportion of users whose average dose was > 4x RDA was 12.0% of younger men, 12.7% of older men, 4.2% of premenopausal women, and 8.2% of postmenopausal women.
View this table:
TABLE 2. . Frequency, median dose, and median duration of consumption by men of supplements that contain iron, by category of average daily supplement use1
View this table:
TABLE 3. . Frequency, median dose, and median duration of consumption by women of supplements that contain iron, by category of average daily supplement use 1
We assessed the iron supplement source (multivitamin containing iron or a separate iron supplement) and found that most of the iron was from multivitamins (96.6% of young men, 92.6% of older men, 87.9% of premenopausal women, and 94.9% of postmenopausal women). However, the use of a separate iron supplement was predominant in those from all subgroups who took a mean of > 3x RDA (61.6% of young men, 53.9% of older men, 69.6% of premenopausal women, and 50.1% of postmenopausal women). The median duration of iron supplement use was 12 mo in younger men, 24 mo in older men, 12 mo in premenopausal women, and 24 mo in postmenopausal women.
ANOVAs were significant for supplement use and TS percentage in younger men (Table 4). Multiple comparison tests found significant differences between the adjusted mean TS percentage of younger men who took > 4x RDA and that in those who took between no iron and 3x RDA. We found no significant association between supplemental iron intake and TS percentage in older men and premenopausal or postmenopausal women.
View this table:
TABLE 4. . Mean (and 95% CI) transferrin saturation by average daily use of supplements containing iron in men and women1
ANOVAs were not significant for supplement use and SF in younger men, but there was an association in older men (Table 5). Multiple comparison tests found significant differences between the adjusted mean SF concentrations in older men who took >3x–4x and 3x RDA and between those in older men who took >4x and 3x RDA (including no iron use). In addition, the adjusted mean SF concentration in those who took up to the RDA was significantly lower than that in the other subgroups. In postmenopausal women, a borderline significant ANOVA result was found (P = 0.0636). Because of this suggestive finding, we also assessed differences in means by using multiple comparison tests; we found significant differences between the adjusted mean SF concentrations in those who took >4x RDA and in those who took 3x RDA (including no iron use). Sensitivity analyses from which persons with an inflammatory condition were excluded produced similar results (data not shown).
View this table:
TABLE 5. . Geometric mean (and 95% CI) serum ferritin concentrations by average daily use of supplements containing iron in men and women1
DISCUSSION
In our study of free-living American adults, we did not find a difference in serum iron measures between those who took supplements containing iron that was on average 3x RDA and those who had lower daily iron use (including nonusers). We found that, in the older men (those 30 y old), average iron supplementation of >32 mg Fe/d (>4x RDA) was associated with higher SF values than was supplementation of 24 mg Fe/d (including nonuse). A similar suggestive effect was found in postmenopausal women. The dose of >32 mg Fe/d is somewhat higher than that found in 2 other studies that reported significant increases in ferritin concentration in postmenopausal women who consumed average iron supplement doses of >18 and >21.5 mg (5, 6).
In addition, in younger men (those aged 19–30 y), we found that, whereas consumption of iron >4x RDA (>32 mg) from supplements was associated with higher transport iron than was the consumption of 24 mg Fe/d, the greater consumption did not appear to affect iron stores. Although iron stores may increase with higher consumption, the average ferritin concentration was remarkably consistent across the consumption levels in younger men. Whereas the overall ANOVA was not significant in younger, premenopausal women, a mean supplemental iron consumption of >3–4x RDA (ie, >54–72 mg/d) was associated with higher iron stores than were seen in persons who consumed 0 to 54 mg Fe/d. We did not find an association between SF concentrations and the consumption of >4x RDA for iron (ie, >72 mg/d), but only 10 women reported consuming supplements at this highest level.
Consumption of supplements containing iron was not uncommon: >15% of men and >20% of women reported consuming supplements with iron in the previous month. Among iron supplement users, the median reported duration of supplement consumption was 12 mo in younger men and premenopaual women and 24 mo in older men and postmenopausal women. These reported durations represent a longer use of prophylactic iron than many health care professionals may prescribe for anemia, which is typically treated by using 60 mg elemental iron for 2 mo to improve the hemoglobin concentration, but which may continue for another 2 mo to build up iron stores in the body (23).
Strengths of our study include its use of data from a large representative national survey that included premenopausal women and younger men and the use of a software package that could manage the complex sampling design. Unlike many mail or phone studies, NHANES participants bring their supplement bottles to the exam and specify the amounts that they took during the previous month. We controlled for several potential confounders including red-meat consumption, serum vitamin C, and blood donation in the past year. Prior studies that focused on the elderly did not adjust for phlebotomy. We also conducted a sensitivity analysis that excluded participants who may have inflammatory conditions that would elevate SF concentrations and found similar results to those in our overall analysis.
Our assessment of supplement intake, however, does have limitations. For example, daily iron supplement intake was calculated by using an indirect method that assumed that the use during a 1-mo period could be applied to a single 24-h period. This calculation may not represent daily amounts for persons who do not take supplements consistently. It is also possible that some persons in our referent group had used iron supplements in the previous 6–12 mo but had stopped taking them 1 mo before the survey. Such persons may have had elevated iron measures because of this prior usage, which would reduce our ability to observe differences between groups. In addition, we do not know the composition of the multiple vitamins (eg, some complete multivitamin-mineral supplements contain calcium or other minerals that might affect iron absorption) or the timing of supplements (eg, taken with a meal or in a fasted state).
Our analysis is also subject to other limitations. We did not control for occult blood loss, which may reduce body iron stores, because of its low incidence in our study sample (<2%), and small sample sizes in some groups limited the statistical power to find differences in iron measures. We also did not have individual data on hemochromatosis diagnosis or HFE gene status. We know, however, from previous studies of the NHANES III data that < 1% of persons in the United States are homozygous for the C282Y HFE mutation, 2% are homozygous for the H63D HFE mutation, and 2% are compound heterozygous (24). Another study found that these groups are at increased risk of elevated transport iron, but the risks associated with simple heterozygosity for these mutations were less clear (25). Because most of the subjects in the current study are not homozygous or compound heterozygous for these mutations, it is unlikely that observed effects were related to genetic susceptibility for hemochromatosis.
Because we conducted > 32 comparisons of iron status, one would expect that at least 1 or 2 of the comparisons would be statistically significant by chance. However, 4 of the differences we found were statistically significant. Given that older men and postmenopausal women are at increased risk of iron overload and given the consistency of the elevations in iron status among those whose average daily consumption of iron from supplements was > 4x RDA, it is more doubtful that these differences are due to chance.
In 2001, the tolerable upper intake level (UL) for iron was reduced from 60 mg/d to 45 mg/d on the basis of the relation of supplemental iron preparations and reversible gastrointestinal distress (7). Our results suggest that average daily iron supplementation at levels below and including the UL for iron are related to higher concentrations of transport iron in younger men and of storage iron in older men and in postmenopausal women. Previous researchers have shown that an increase of 1 ng/mL in the ferritin concentration is equivalent to an increase of 8 mg in body storage iron (26). In our study, the ferritin concentrations in older men and postmenopausal women who consumed > 32 mg iron/d were, on average, 65–91 ng/mL higher than those in similar subjects who consumed no supplements. The difference in iron stores between nonusers and subjects who consumed > 32 mg Fe/d is equivalent to 520–729 mg storage iron, or the amount contained in 2–3 units of blood (a 450–500-mL unit of blood typically contains 200–250 mg iron). Although this quantity represents a substantial amount of storage iron, it is not clear that these levels of intake would result in systemic iron overload in healthy persons who have no other risk factors for iron overload. None of the conditional mean SF values for the subgroups met sex-specific clinical cutoffs for iron elevation (SF > 200 µg/L in premenopausal females and >300 µg/L in postmenopausal females and males) (27).
Among healthy persons, the associations between iron intake, iron overload, and chronic disease are suggestive but inconclusive (7), and they have not been used in determining the iron UL. However, iron overload was recently reported in a 61-y-old woman with a 27-y history of oral supplementation with 325 mg FeSO4 twice a day, along with 15000 mg vitamin C (28). The woman was found to have iron overload despite being wild-type for the HFE gene. Although extreme cases like this are rare (29, 30), it suggests that iron overload can result from excess iron supplementation.
In conclusion, our findings suggest that, in a population such as that of the United States, which typically consumes a mixture of heme- and non-heme dietary iron sources, iron supplementation < 3x RDA was not related to higher iron status in relatively healthy persons. In older men and postmenopausal women, supplemental iron intake > 3x RDA was positively related to higher storage iron.
Excessive iron intake may cause gastrointestinal distress, including constipation, and may reduce the absorption of zinc and certain prescription medications (31-33). In addition, some observational studies suggest that higher iron stores are deleterious to health (34-37). Therefore, middle-aged and older men and postmenopausal women who use supplements containing iron should be made aware of the amount of iron necessary to satisfy dietary requirements and of the possible effect of excess iron supplementation on body iron stores and health.
ACKNOWLEDGMENTS
We thank Joel Kimmons for his thoughtful comments on an earlier draft of the manuscript.
All authors were involved in the planning of these analyses and the interpretation of results. Data analysis was the primary responsibility of CG. HMB was responsible for the first draft of the manuscript. All authors were closely involved in revisions and final approval. None of the authors had a personal or financial conflict of interest.
REFERENCES