Magnesium requirements: new estimations for men and women by
cross-sectional statistical analyses of metabolic magnesium balance
data1⫺4
Curtiss D Hunt and LuAnn K Johnson
ABSTRACT
Background: Current recommendations for magnesium require-
ments are based on sparse balance data.
Objective: To provide new estimates of the average magnesium
requirement for men and women, we pooled magnesium data from
27 different tightly controlled balance studies conducted at the US
Department of Agriculture, Agricultural Research Service, Grand
Forks Human Nutrition Research Center, Grand Forks, ND.
Design: Magnesium balance data (magnesium intake Ҁ [fecal mag-
nesium ѿ urinary magnesium]) (664 data points) were collected
from 243 subjects (women: n ҃ 150; weight: 71.6 앐 16.5 kg; age:
51.3 앐 17.4 y; men: n ҃ 93, weight: 76.3 앐 12.5 kg; age: 28.1 앐
8.1 y). Data from the last 6 –14 d of each dietary period (욷28 d) of
each study were analyzed and were excluded if individual intakes of
calcium, copper, iron, phosphorus, or zinc fell below respective
estimated average requirements (EARs) or exceeded 99th percen-
tiles of usual intakes of those elements (iron: above the upper limit)
from the 1994 Continuing Survey of Food Intakes by Individuals.
Daily intakes of magnesium ranged between 84 and 598 mg. The
relation between magnesium intake and magnesium output was in-
vestigated by fitting random coefficient models.
Results: The models predicted neutral magnesium balance [defined
as magnesium output (Y) equal to magnesium intake (M)] at mag-
nesium intakes of 165 mg/d [95% prediction interval (PI): 113, 237
mg/d; Y ҃ 19.8 ѿ 0.880 M], 2.36 mg · kgҀ1 · dҀ1 (95% PI: 1.58, 3.38
mg · kgҀ1 · dҀ1; Y ҃ 0.306 ѿ 0.870 M), or 0.075 mg · kcalҀ1 · dҀ1
(95% PI: 0.05, 0.11 mg · kcalҀ1 · dҀ1; Y ҃ 0.011 ѿ 0.857 M). Neither
age nor sex affected the relation between magnesium intake and
output.
Conclusion: The findings suggest a lower magnesium requirement
for healthy men and women than estimated previously. Am J
Clin Nutr 2006;84:843–52.
KEY WORDS Magnesium intake, magnesium excretion,
magnesium balance, magnesium requirement
INTRODUCTION
Most Americans consume less than the estimated average
requirement (EAR) for magnesium (1) as determined by analysis
of the National Health and Nutrition Examination Survey
(NHANES) 2001–2002 data set (2). For example, 64% of women
aged 51–70 y do not attain the magnesium EAR for that sex-age
group (265 mg/d) (1). Instead, their estimated mean magnesium
intake during the NHANES 2001–2002 survey was 246 mg/d.
Am J Clin Nutr 2006;84:843–52. Printed in USA. © 2006 American Society for Nutrition
Differential analysis of the NHANES 1999 –2000 data set (3)
indicated that the mean magnesium intake for American women
(white) aged 51–70 y was 238 mg/d (1). For Mexican and African
American women of the same age group, magnesium intakes
were lower: 185 and 169 mg/d, respectively (4). For white, Mex-
ican, and African women, intakes were lower still for those re-
Downloaded from www.ajcn.org by guest on December 19, 2010
porting no use of dietary supplements: 222, 176, and 150 mg/d,
respectively. Also, some evidence exists that enteric magnesium
absorption decreases with age and urinary magnesium excretion
increases (5, 6).
The data available to estimate the magnesium average require-
ment are extremely limited. For example, the magnesium EARs
for adult women of all ages are based nearly exclusively on the
findings from one study (7) of adults who consumed self-selected
diets in a free-living environment. Because the magnesium in-
takes of American adults are below the EAR, further research is
warranted that tests the appropriateness of the current EAR.
Estimations of magnesium adequacy typically use balance stud-
ies in which positive magnesium balance is associated with ad-
equacy (8 –11). However, individual magnesium balance studies
seldom test 쏜2 magnesium intakes, such that titration of the
magnesium requirement is not possible. Magnesium intakes
around predicted zero balance are needed to model the precise
relation between mineral intake and loss and retention near zero
balance (12).
To provide a better estimate of the adult magnesium require-
ment, the present study used magnesium balance data generated
previously in a series of tightly controlled metabolic in-house
1
From the US Department of Agriculture, Agricultural Research Service,
Grand Forks Human Nutrition Research Center, Grand Forks, ND.
2
Mention of a trademark or proprietary product does not constitute a
guarantee or warranty of the product by the US Department of Agriculture
and does not imply its approval to the exclusion of other products that may
also be suitable. The US Department of Agriculture, Agricultural Research
Service, Northern Plains Area, is an equal opportunity, affirmative action
employer, and all agency services are available without discrimination.
3
Supported by the USDA Agricultural Research Service (ARS) program
“Mineral Intakes for Optimal Bone Development and Health,” Current Re-
search Information System (CRIS) no. 5450-51000-039-00D, as part of the
author’s official duties.
4
Reprints not available. Address correspondence to CD Hunt, USDA ARS
Grand Forks Human Nutrition Research Center, PO Box 9034, Grand Forks,
ND 58202-9034. E-mail: chunt@gfhnrc.ars.usda.gov.
Received March 30, 2006.
Accepted for publication June 1, 2006.
843
844 HUNT AND JOHNSON
feeding studies conducted at the US Department of Agriculture,
Agricultural Research Service, Grand Forks Human Nutrition
Research Center, Grand Forks, ND, between 1976 and 2001. The
studies were designed originally to determine specific outcomes
of nutritional challenges on physiologic function. Twenty-seven
of 42 studies (13–35) measured magnesium balance and incor-
porated design components, such as a control (nutritionally re-
plete) dietary period, relevant for estimating the magnesium re-
quirement by cross-sectional statistical analysis. A broad range
of magnesium intakes were used to estimate the amount of di-
etary magnesium needed to maintain zero magnesium balance in
healthy persons.
SUBJECTS AND METHODS
Common design characteristics of the 27 individual original
metabolic studies used in the present study are summarized be-
low. Details about each study are available in the references cited
in Table 1.
Subjects
Each of the original metabolic studies (Table 1) was reviewed
and approved separately by the University of North Dakota Pro-
tection of Human Subjects Committee (studies 1-6) or the Uni-
versity of North Dakota Institutional Review Board (studies 7-27
and the current study). Subjects were informed verbally and in
writing about the purpose and design for each original metabolic
study and provided written informed consent to participate in
protocols that followed the guidelines of the Declaration of Hel-
sinki about the use of human subjects.
Recruitment
In all studies, healthy women and men were recruited by public
advertisement and selected to enter each study after they were
informed in detail of the nature of the research, including the risks
and benefits.
Inclusion criteria
Healthy subjects were screened onsite and selected on the
basis of medical data (no evidence of alcoholism; normal bone,
kidney, thyroid, and liver functions; normal blood pressure and
fasting glucose; no chronic medication use; negative lung scan),
psychological history [free of psychopathology as determined by
the Minnesota Multiphasic Personality Inventory (NCS Assess-
ment, Minneapolis, MN) and an extensive in-house psycholog-
ical history questionnaire and clinical interview], and diet history
(no pertinent food allergies or refusal to eat required foods). Each
postmenopausal woman agreed either to discontinue hormone
replacement therapy (HRT) before joining a specific study if she
and her physician agreed that it was not harmful to stop the HRT
or to maintain HRT throughout the study.
Living environment
Accepted male [n ҃ 93; weight (x៮ 앐 SD): 76.3 앐 12.5 kg; age:
28.1 앐 8.1 y (range: 19 – 65 y)] and female [n ҃ 150; weight:
71.6 앐 16.5 kg; age: 51.3 앐 17.4 y (range: 19 –77 y)] subjects
resided for the entire length (typically 6 m) of the individual
studies in the metabolic ward at the Grand Forks Human Nutri-
tion Research Center. Subject ethnicity was predominantly white
(n ҃ 227) with additional participation by blacks (n ҃ 6), Amer-
ican Indians or Alaskans (n ҃ 5), Asians (n ҃ 2), Hispanics (n ҃
2), and undeclared ethnicity (n ҃ 1). The ward provided an
environment for strict control of food consumption, physical
activity, and data collection. Each subject was provided with a
private bedroom with cable television, radio with wake-up alarm,
intercom to a 24-h central nurse station, telephone, and a semi-
private bathroom. Activity areas and the nurse station were ad-
jacent to the private bedrooms. Subjects were allowed to leave
the immediate living or dining areas or facility only when ac-
companied by a chaperone to ensure compliance with study
protocols. Meal consumption was observed by specially trained
dietary staff members, and irregularities were recorded. Subjects
agreed to use only personal care products and in the amounts
approved by the principal investigators and to limit and standard-
ize extraneous chemical exposure. Subjects were not allowed to
use tobacco or medicinal marijuana or consume alcohol (except
for specified ethanol tolerance tests). For most studies, individ-
ually prescribed physical activity was performed multiple times
per week to maintain initial body composition and physical work
Downloaded from www.ajcn.org by guest on December 19, 2010
capacity.
Diet
Composition
Basal diets were composed of ordinary Western foods, some-
times supplemented with experimental foods (eg, fructose corn-
bread, egg white drinks, casein biscuits), and fed as a 6-d (study
1) or 3-d (studies 2-27) menu rotation to provide variety, but in
a manner that assured variations in nutrient intake were not con-
sequential. Standard temperatures and cooking times were ad-
hered to for each recipe on the menu cycles. Salt and pepper were
served in constant amounts, selected by each volunteer, through-
out individual studies. The limited menus were supplemented as
needed with some nutrients in constant amounts to maintain
nutritional adequacy. Because the study intervals included win-
ter months in North Dakota, when sunlight exposure is limited,
all subjects received a daily supplement of cholecalciferol (10 or
20 ␮g). Dietary iron (as ferrous gluconate) was provided in
excess of the recommended dietary allowance to mitigate the
decline in iron status as a result of phlebotomy during the studies.
Preparation and use
Subjects consumed only and all foods, beverages (including
water), and vitamin, mineral, or other supplements provided by
the center. The minimum length of any dietary period for any
study was 18 d. Whenever possible, food was purchased in single
lots sufficient to last for several months to ensure minimal vari-
ation in food types. All food was weighed proportionally with a
1% rounding error during preparation in the metabolic kitchen
and was consumed completely by the subjects with the aid of
spatulas and rinse bottles. Deionized water was consumed ad
libitum. Initial energy requirement for each subject was deter-
mined by using the Harris and Benedict equation (37) and adding
a uniform amount (between 50% and 70%) of basal energy ex-
penditure for normal physical activity. Except for the weight loss
studies (studies 20 and 24), energy intake was adjusted in stan-
dardized increments [typically 0.84 MJ (200 kcal)] during the
course of each experiment to maintain body weight (measured
daily) within 앐2% of admission weight. This study used balance
 MAGNESIUM BALANCE AND REQUIREMENT 845
TABLE 1
Characteristics of metabolic studies available for cross-sectional statistical analysis1

Study no., study design, and reference2 Study period Sex All subjects Qualified subjects3

1. Plant fiber: bioavailability of minerals in food (13) 5/76–2/82 M

Age (y) 25.6 앐 8.8 (18.0–52.0)4 27.1 앐 10.3 (19.0–52.0)
Weight (kg) 78.5 앐 15.2 (58.9–135.8) 77.8 앐 14.6 (58.9–107.9)
Height (cm) 177.1 앐 7.6 (152.8–193.4) 174.4 앐 4.6 (166.3–182.0)
n 29 10
2. Dietary fiber: stool output (14) 5/77–12/80 M
Age (y) 29.6 앐 11.1 (19.0–64.0) 33.9 앐 14.6 (19.0–64.0)
Weight (kg) 77.4 앐 12.3 (60.3–107.9) 79.2 앐 15.2 (62.7–107.9)
Height (cm) 176.4 앐 9.3 (152.9–194.7) 173.2 앐 10.8 (152.9–194.0)
n 28 9
3. Copper intake: copper balance, absorption, and 1/80–12/82 M
indicators of status (15)
Age (y) 24.6 앐 4.8 (19.0–32.0) 24.6 앐 4.8 (19.0–32.0)
Weight (kg) 79.1 앐 10.2 (69.0–105.0) 79.1 앐 10.2 (69.0–105.0)
Height (cm) 175.5 앐 5.2 (164.6–182.1) 175.5 앐 5.2 (164.6–182.1)
n 11 11
4. Zinc intake: whole-body surface loss of zinc (16) 7/79–9/83 M
Age (y) 30.9 앐 9.9 (19.0–57.0) 36.7 앐 18.0 (23.0–57.0)

Downloaded from www.ajcn.org by guest on December 19, 2010

Weight (kg) 69.4 앐 10.7 (58.3–97.2) 62.9 앐 5.4 (58.3–68.9)
Height (cm) 172.2 앐 5.3 (164.1–182.9) 170.2 앐 5.6 (164.1–175.0)
n 12 3
5. Physical performance and carbohydrate and lipid 9/80–12/80 M
metabolism (17)
Age (y) 23.7 앐 6.4 (20.0–31.0) 25.5 앐 7.8 (20.0–31.0)
Weight (kg) 70.6 앐 5.1 (67.3–76.4) 67.7 앐 0.5 (67.3–68.0)
Height (cm) —5 —5
n 3 2
6. Browned and unbrowned corn products: bioavailability 1/81–4/81 M
of zinc (18)
Age (y) 35.3 앐 20.1 (22.0–65.0) 35.3 앐 20.1 (22.0–65.0)
Weight (kg) 73.9 앐 6.9 (68.0–83.7) 73.9 앐 6.9 (68.0–83.7)
Height (cm) 176.4 앐 9.4 (165.0–186.4) 176.4 앐 9.4 (165.0–186.4)
n 4 4
7. Intrinsically and extrinsically labeled meals: 65Cu 10/82–12/84 M
absorption (19)
Age (y) 27.2 앐 9.1 (19.0–49.0) 27.2 앐 9.6 (19.0–49.0)
Weight (kg) 67.9 앐 6.4 (57.5–80.1) 67.2 앐 6.3 (57.5–80.1)
Height (cm) 174.7 앐 4.6 (167.7–184.0) 174.7 앐 4.8 (167.7–184.0)
n 11 10
8. Fat, vitamin E, and zinc intakes: copper and iron 7/82–6/83 M
absorption and retention6
Age (y) 28.3 앐 5.7 (22.0–37.0) 28.0 앐 6.1 (22.0–37.0)
Weight (kg) 76.2 앐 14.5 (48.3–91.0) 77.3 앐 15.6 (48.3–91.0)
Height (cm) 177.4 앐 7.3 (165.4–189.2) 177.3 앐 8.0 (165.4–189.2)
n 7 6
9. Dietary Maillard products: iron and zinc absorption and 8/83–5/84 M
retention6
Age (y) 26.1 앐 6.1 (20.0–39.0) 26.1 앐 6.1 (20.0–39.0)
Weight (kg) 73.4 앐 13.5 (51.4–91.4) 73.4 앐 13.5 (51.4–91.4)
Height (cm) 175.4 앐 5.6 (165.2–186.6) 175.4 앐 5.6 (165.2–186.6)
n 9 9
10. Folic acid supplements: zinc and iron absorption (20) 1/84–7/84 M
Age (y) 29.1 앐 5.3 (19.0–36.0) 29.1 앐 5.3 (19.0–36.0)
Weight (kg) 85.9 앐 22.1 (64.8–134.7) 85.9 앐 22.1 (64.8–134.7)
Height (cm) 180.6 앐 10.3 (161.7–193.6) 180.6 앐 10.3 (161.7–193.6)
n 8 8
11. Copper and sucrose interactions6 5/84–12/84 M
Age (y) 26.6 앐 4.5 (21.0–32.0) 26.6 앐 4.5 (21.0–32.0)
Weight (kg) 70.8 앐 3.4 (67.3–75.8) 70.8 앐 3.4 (67.3–75.8)
Height (cm) 180.8 앐 7.1 (170.1–189.9) 180.8 앐 7.1 (170.1–189.9)
n 7 7
(Continued)
846 HUNT AND JOHNSON

TABLE 1 (Continued)

Study no., study design, and reference2 Study period Sex All subjects Qualified subjects3

12. Intrinsically and extrinsically labeled meat, liver, and 1/85–7/85 F

peanut and sunflower butters: 65Cu absorption (21)
Age (y) 57.6 앐 5.3 (49.0–66.0) 57.6 앐 5.3 (49.0–66.0)
Weight (kg) 75.1 앐 12.4 (57.4–90.8) 75.1 앐 12.4 (57.4–90.8)
Height (cm) 165.3 앐 6.4 (155.4–172.5) 165.3 앐 6.4 (155.4–172.5)
n 7 7
13. Marginal zinc intakes: ethanol metabolism (22) 1/85–7/85 F
Age (y) 58.2 앐 3.4 (55.0–63.0) 60.3 앐 2.5 (58.0–63.0)
Weight (kg) 66.9 앐 13.5 (53.6–86.7) 66.4 앐 17.8 (53.6–86.7)
Height (cm) 158.8 앐 5.4 (150.5–165.0) 159.5 앐 1.7 (158.0–161.3)
n 5 3
14. Ascorbic acid and copper intakes: indicators of copper 7/85–6/86 F
nutriture (23)
Age (y) 26.8 앐 4.8 (20.0–37.0) 25.1 앐 3.5 (20.0–29.0)
Weight (kg) 61.3 앐 10.0 (46.0–76.0) 65.5 앐 9.9 (48.2–76.0)
Height (cm) 163.0 앐 4.0 (154.6–168.5) 163.0 앐 5.2 (154.6–168.5)
n 13 7
15. Short-term and long-term variability of nutritional status 7/85–12/85 F
indexes (24)

Downloaded from www.ajcn.org by guest on December 19, 2010

Age (y) 30.6 앐 8.0 (23.0–44.0) 30.6 앐 8.0 (23.0–44.0)
Weight (kg) 86.9 앐 7.5 (76.8–92.7) 86.9 앐 7.5 (76.8–92.7)
Height (cm) 170.2 앐 2.2 (168.1–172.9) 170.2 앐 2.2 (168.1–172.9)
n 5 5
16. Aluminum, boron, and magnesium intakes: boron, 6/86–12/86 F
calcium, and magnesium absorption and retention (25)
Age (y) 58.3 앐 10.2 (39.0–81.0) 58.3 앐 10.2 (39.0–81.0)
Weight (kg) 67.8 앐 10.4 (54.6–88.3) 67.8 앐 10.4 (54.6–88.3)
Height (cm) 163.3 앐 7.9 (146.4–180.0) 163.3 앐 7.9 (146.4–180.0)
n 15 15
17. Calcium and manganese intakes: menstrual cycle 7/87–12/87 F
symptoms (26)
Age (y) 26.2 앐 6.7 (18.0–41.0) 26.9 앐 6.5 (19.0–41.0)
Weight (kg) 61.8 앐 10.8 (46.4–87.0) 62.0 앐 11.2 (46.4–87.0)
Height (cm) 164.0 앐 7.4 (145.6–174.1) 163.7 앐 7.6 (145.6–174.1)
n 14 13
18. Boron and magnesium intakes: central nervous system 1/88–7/88 F
activity (27)
Age (y) 61.9 앐 8.2 (49.0–77.0) 61.9 앐 8.2 (49.0–77.0)
Weight (kg) 65.4 앐 11.2 (52.4–91.0) 65.4 앐 11.2 (52.4–91.0)
Height (cm) 163.3 앐 6.7 (155.1–179.7) 163.3 앐 6.7 (155.1–179.7)
n 14 14
19. Magnesium intakes: magnesium status indicators (28) 7/89–12/89 F
Age (y) 61.8 앐 8.2 (47.0–75.0) 61.8 앐 8.2 (47.0–75.0)
Weight (kg) 64.4 앐 12.4 (49.9–96.8) 64.4 앐 12.4 (49.9–96.8)
Height (cm) 158.9 앐 5.0 (149.7–167.7) 158.9 앐 5.0 (149.7–167.7)
n 13 13
20. Weight loss: menstrual symptomatology (29)7 7/90–12/90 F
Age (y) 26.8 앐 4.4 (21.0–38.0) 26.8 앐 4.4 (21.0–38.0)
Weight (kg) 93.1 앐 13.8 (75.2–115.9) 93.1 앐 13.8 (75.2–115.9)
Height (cm) 162.9 앐 6.3 (155.0–176.5) 162.9 앐 6.3 (155.0–176.5)
n 14 14
21. Oxidant stress by dietary factors: copper status 1/91–8/91 M
indicators (30)
Age (y) 31.0 앐 4.8 (24.0–37.0) 31.0 앐 4.8 (24.0–37.0)
Weight (kg) 82.8 앐 7.5 (73.0–97.7) 82.8 앐 7.5 (73.0–97.7)
Height (cm) 181.4 앐 7.2 (177.0–197.1) 181.4 앐 7.2 (177.0–197.1)
n 7 7
22. Meat consumption: zinc absorption and iron status (31) 7/92–12/92 F
Age (y) 62.5 앐 6.2 (51.0–70.0) 62.5 앐 6.2 (51.0–70.0)
Weight (kg) 68.3 앐 10.8 (52.7–88.5) 68.3 앐 10.8 (52.7–88.5)
Height (cm) 159.7 앐 6.2 (145.1–172.7) 159.7 앐 6.2 (145.1–172.7)
n 14 14
(Continued)
 MAGNESIUM BALANCE AND REQUIREMENT 847
TABLE 1 (Continued)

Study no., study design, and reference2 Study period Sex All subjects Qualified subjects3

23. Copper intakes: copper status indicators (32) 1/93–7/93 F

Age (y) 62.5 앐 7.4 (49.0–75.0) 63.3 앐 7.2 (49.0–75.0)
Weight (kg) 67.2 앐 10.4 (45.6–83.1) 68.1 앐 10.3 (45.6–83.1)
Height (cm) 158.2 앐 6.1 (148.5–166.9) 158.9 앐 5.8 (148.5–166.9)
n 13 12
24. Weight loss: regional body composition7 (33) 1/94–6/94 F
Age (y) 29.6 앐 4.2 (25.0–38.0) 28.3 앐 3.4 (25.0–33.0)
Weight (kg) 100.3 앐 16.9 (78.0–132.5) 102.8 앐 21.9 (78.4–132.5)
Height (cm) 167.8 앐 6.4 (159.3–182.5) 170.5 앐 8.0 (160.2–182.5)
n 12 6
25. Fructose and magnesium intakes: macromineral 7/94–12/94 M
homeostasis (34)
Age (y) 30.4 앐 5.5 (22.0–40.0) 30.4 앐 5.5 (22.0–40.0)
Weight (kg) 78.9 앐 13.1 (53.4–94.8) 78.9 앐 13.1 (53.4–94.8)
Height (cm) 178.2 앐 9.0 (164.0–199.0) 178.2 앐 9.0 (164.0–199.0)
n 14 14
26. Magnesium and copper intakes: magnesium status 1/95–12/95 F
indicators (35)
Age (y) 63.8 앐 8.6 (47.0–78.0) 66.9 앐 6.9 (50.0–74.0)

Downloaded from www.ajcn.org by guest on December 19, 2010

Weight (kg) 67.4 앐 11.5 (50.1–89.9) 63.7 앐 11.3 (50.1–82.1)
Height (cm) 160.1 앐 6.0 (149.2–176.0) 159.9 앐 6.8 (149.4–176.0)
n 25 13
27. Magnesium intakes: neuronal function6 1/00–6/00, F
Age (y) 1/01–6/01 59.6 앐 7.2 (49.0–70.0) 59.6 앐 7.2 (49.0–70.0)
Weight (kg) 74.8 앐 14.5 (53.5–112.6) 74.8 앐 14.5 (53.5–112.6)
Height (cm) 164.3 앐 8.3 (151.0–180.0) 164.3 앐 8.3 (151.0–180.0)
n 14 14
1
Metabolic studies done at the US Department of Agriculture, Agricultural Research Service, Grand Forks Human Nutrition Research Center, Grand
Forks, ND, between 1976 and 2001 that measured individual dietary magnesium intake and fecal and urine magnesium output.
2
For the study design, the experimental variable or variables are given, and the primary outcome measure or measures follow the colon. The reference
number is in parentheses.
3
Data from a specific dietary period for a person were excluded when intakes of calcium, copper, iron, phosphorus, or zinc fell below the respective
estimated average requirements or exceeded the respective 99th percentiles of usual intakes from the 1994 Continuing Survey of Food Intakes by Individuals
(for iron, above the upper limit) (1, 36) to avoid confounding the results with concurrent nutritional stress. To maximize consistency in data across persons,
balance periods 쏝6 d or 쏜 12 d in length were eliminated. To meet the design criteria suggested by the Food and Nutrition Board (1), the minimum acceptable
dietary adaptation period was 12 d.
4
x៮ 앐 SD; range in parentheses (all such values).
5
Height data were not available.
6
Publication was not available.
7
Study used data from maintenance diets only.

data from the weight loss experiments collected only during the
initial maintenance dietary periods.
Magnesium supplementation
In most of the studies (studies 1-16, 20, and 24), the basal diet
was considered magnesium replete and not supplemented with
magnesium. In the remainder of the studies, the basal diet was
supplemented with magnesium as magnesium gluconate (except
for one) as needed to meet the magnesium recommended dietary
allowance (study 17: 57 mg Mg/d; study 21: 171 mg Mg/d; study
23: 114 mg Mg/d) or a specific experimental dietary magnesium
value (study 16: 200 mg Mg/d; study 18: 200 mg Mg/d; study 19:
200 mg Mg/d; study 25: 207 mg Mg/d; study 26: 206 or 284 mg
Mg/d; study 27: 100 or 200 mg Mg/d). For study 22, a magnesium
supplement of 50 mg/d was provided as magnesium citrate dibasic.
Magnesium balance method
Dietary analysis
Balance data from the last 6 –14 d of each dietary period of
each study were analyzed. A duplicate diet for each subject in
studies 1-14 and one representative duplicate diet supplying 8.4
MJ (2000 kcal) for all subjects in studies 14-27 were prepared
daily for analysis. For all studies, the daily duplicate diets were
blended, and aliquots (6% of total weight) of the daily meals were
mixed well and made into 6- or 7-d composites before freezing.
Aliquots of all prescribed dietary supplements, discretionary
foods, and transient medications were measured for mineral con-
tent, and the mineral contributions were included as part of the
dietary magnesium intake.
Urinary and fecal analyses
Total urine was collected by polypropylene funnels into 4-L
polypropylene containers (6 mL of 6N HCl as “Baker Technical”
grade added to prevent bacterial growth; JT Baker Inc, Phillips-
burg, NJ). Fecal output, excluding remains left on toilet paper,
was collected directly in plastic bags throughout each study with
precautions to avoid trace mineral contamination. All excreta
were cooled immediately after collection in a double-doored
refrigerator that provided discreet sample transfer to technical
staff. The last 6- or 7-d urine composites of each dietary period
848 HUNT AND JOHNSON
for each volunteer were prepared by combining proportional
aliquots of daily urines and freezing them until analysis.
Weighed fecal specimens were frozen, lyophilized, and then
combined in toto in a plastic bag to prepare 6- or 7-d composites
and were subsequently pulverized by a rolling pin and mixed by
hand shaking for each volunteer.
Exercise
For all studies, all subjects were required to exercise a mini-
mum of 15 min at 50% maximum work capacity on an ergocycle 3
times each week. Additional exercise was prescribed as needed to
maintain body weight within 2% of initial weight. Voluntary walk-
ing regimens did not affect mandatory exercise prescriptions.
Magnesium balance determination
Dietary and fecal sample digestion and analyses
Aliquots of the diet and fecal composites were digested with
perchloric and nitric acids in glass beakers. Magnesium content
of dietary, urinary, and fecal digestates was determined by flame
atomic absorption spectrophotometry (studies 1-6) or by induc-
tively coupled plasma emission spectroscopy (studies 7-27) with
aqueous calibration standards. Urine samples were diluted in
0.5% lanthanum chloride before magnesium measurement by
flame atomic absorption spectrophotometry to mask interfer-
ences. Methodologic precision and accuracy of the digestion and
analytic procedures were evaluated by concurrent analysis of the
National Institute of Standards and Technology bovine liver
standards (Standard Reference Materials no. 1577), pool sam-
ples, and replicate samples containing added metal.
Magnesium balance calculation
Whole-body surface losses of magnesium for young men were
determined (unpublished data, 1989) in one of the metabolic
studies (38). Each subject showered, then put on a cotton suit of
long underwear and a protective covering over briefs and socks,
all provided by the metabolic unit. After 48 h, the suits were
removed, and the subjects stood in a plastic tub and shower with
warmed deionized water. Shaving or application of any skin care
product other than the wash soap provided by staff members was
not allowed for 4 d before or during the sweat test. Whole-body
surface losses of magnesium were negligible (4.1 mg/d).
The last balance periods (울2) were selected from each avail-
able dietary period to provide 1–9 observations per subject. Mag-
nesium balance was calculated by differences between dietary
intake and fecal and urinary losses. Magnesium balance calcu-
lations did not include whole body surface (negligible; 4.1 mg/d
for young men), phlebotomy [negligible; 0.019 mg Mg/mL se-
rum (39)], menstrual [negligible; 2.3 mg/d (range: 6.5-0.3 mg/
d)], or seminal losses.
Data management and statistical analysis
Confidentiality considerations
For each original study, a password secure, confidential com-
puter file maintained the linkage between subject name and iden-
tification number, with access limited to select center staff mem-
bers. This linkage was separate from research data files and was
kept for several reasons: to provide subjects with individual study
results and to provide governmental and institutional auditors
with access as required by law. For the present study, the linkage
between preexisting data files and subject was broken by the
following method. The database for the original studies consisted
of multiple data files, each keyed by using a unique individual
identification number. A separate file was created that consisted
only of records with current identification numbers along with
new, randomly generated identification numbers. Subsequently,
the relevant data files in the preexisting database were copied into
a new database; as the file was copied, the original identification
number was replaced by the randomly generated identification
number. Subject names and birth dates were not copied to the
new database. After the new database was created, the file con-
taining the existing identification numbers and the randomly
generated identification numbers were deleted permanently. No
hard copy of this file was ever generated. In summary, all linkage
was broken between the preexisting data and the original iden-
tification numbers with no possibility to associate any original
data with a specific subject.
Outliers
Data from a specific dietary period for a subject were excluded
Downloaded from www.ajcn.org by guest on December 19, 2010
when intakes of calcium, copper, iron, phosphorus, or zinc fell
below the respective EARs (calcium: 5th percentile of intake) or
exceeded the respective 99th percentiles of usual intakes from the
1994 Continuing Survey of Food Intakes by Individuals (iron:
above the upper limit) (1, 36) to avoid confounding the results
with concurrent nutritional stress. To maximize consistency in
data across subjects, balance periods 쏝 6 d or 쏜 12 d were
eliminated. To meet the design criteria suggested by the Food and
Nutrition Board (FNB) (1), the minimum dietary adaptation pe-
riod was 12 d (median: 31 d; maximum: 109 d).
Model
The relation between magnesium intake [in mg/d (model A),
mg · kgҀ1 · dҀ1 (model B), or mg · kcalҀ1 · dҀ1 (model C)] and
magnesium output (fecal ѿ urinary excretion) was investigated
by using the following linear mixed-effect model (40):
Yij ⫽ ␣ ⫹ ␤Xij ⫹ ai* ⫹ bi*Xij ⫹ eij (1)
where Yij is the jth magnesium output measurement on the ith
subject, Xij is the jth magnesium intake measurement on the ith
subject, i ҃ 1, ѧ, 243 subjects, j ҃ 1,ѧ, ni values, and
冉 冊
ai*
bi*
⬃ iid N 冋冉 冊 冘册
0
0 , ,
冘 冉 ⫽ 冊␴2a ␴ab
␴ab ␴2b and eij ⬃ iid N共0,␴2e 兲 (2)
The fixed effects part of the model is ␣ ѿ ␤Xij, whereas the
random-effects part of the model is a*i ѿ b*i Xij ѿ eij. PROC
MIXED in SAS (version 9.1; SAS Institute, Cary, NC) was used
to fit all models. An unstructured variance-covariance matrix, ⌺,
was specified for the slopes and intercepts. As specified in Equa-
tion 1, both intercepts and slopes are considered random effects.
To test whether random intercepts and slopes were necessary,
additional models were fit, allowing only random intercepts and
only random slopes. Akaike’s Information Criteria was used to
compare models. In all cases, the model that resulted in the
smallest Akaike’s Information Criteria value allowed both the
slopes and intercepts to be random effects.
 MAGNESIUM BALANCE AND REQUIREMENT 849
Fixed coefficients were added to the models to allow for sep-
arate intercepts and slopes for men and women to determine
whether sex differences existed in the requirement estimates.
The model used was as follows:
Yij ⫽ ␣ ⫹ ␤1Xij ⫹ ␤2Gi ⫹ ␤3XijGi ⫹ ai* ⫹ bi*Xij ⫹ eij
(3)
where Gi is 1 if subject i was male or 0 if subject i was female. To
test whether there was an age effect on magnesium requirements,
age was added to the model as follows:
Yij ⫽ ␣ ⫹ ␤1Xij ⫹ ␤2Ai ⫹ ␤3Xij Ai ⫹ ␤4Xij AiGi ⫹ ai* ⫹ bi*Xij
⫹ eij (4)
where Ai is the age of subject i. The men in this study had a
relatively narrow age range compared with the women. There-
FIGURE 1. Relation between magnesium output (fecal ѿ urinary excre-
fore, models that included age as a predictor were also tested by tion) and magnesium intake, both expressed as mg/d, for women [F; n ҃ 150;
using data only from women to assess whether the results ob- weight: 71.6 앐 16.5 kg; age: 51.3 앐 17.4 y (range: 19 –77 y)] and men [E;
tained were simply artifacts of the different age distributions. n ҃ 93; weight: 76.3 앐 12.5 kg; age: 28.1 앐 8.1 y (range: 19 – 65 y)] who
The 95% CIs (41) and 95% prediction intervals (PIs) (42, 43) participated in tightly controlled feeding studies conducted in a metabolic

Downloaded from www.ajcn.org by guest on December 19, 2010

were calculated and graphed for the final models. The magne-
sium requirement was defined as the point where magnesium
intake equaled magnesium output. The 95% PI around this point
was obtained by treating this as a calibration problem (ie, inverse
prediction of magnesium intake given magnesium output) and
calculating the PIs by using methods for random coefficient
models (43).
RESULTS
A total of 664 observations from 243 subjects (women, n ҃
150; men, n ҃ 93) were available for statistical analysis. Daily
intakes of magnesium ranged between 84 and 598 mg. To facil-
itate application of the findings to the general population, the data
were examined by the 3 statistical models.
unit (664 observations). Magnesium output (Y) increased linearly with in-
creases in magnesium intake (P ҃ 0.0001; n ҃ 664). Across all subjects,
neutral magnesium balance was maintained at magnesium intakes (M) of 165
mg/d [95% prediction interval (PI): 113, 237 mg/d (thick dotted lines); Y ҃
19.8 ѿ 0.880 M]. Solid thin lines indicate 95% CIs around the regression line.
Magnesium intake compared with output: statistical
model C
Magnesium output increased linearly with increases in mag-
nesium intake (P ҃ 0.0001) when both were expressed as mg ·
kcalҀ1 · dҀ1 (Figure 3). Across all subjects, magnesium intake
equaled magnesium output at 0.075 mg · kcalҀ1 · dҀ1 (Y ҃
0.011 ѿ 0.857 M; 95% PI: 0.05, 0.11 mg · kcalҀ1 · dҀ1). The

Magnesium intake compared with output: statistical

model A
Magnesium output (Y) increased linearly with increases in
magnesium intake (M) when intake was expressed as mg/d (P ҃
0.0001) (Figure 1). Magnesium intake equaled magnesium out-
put (crossed line of unity; neutral balance) at 165 mg/d (Y ҃
19.8 ѿ 0.880 M; 95% PI: 113, 237 mg/d). The relation between
magnesium output and intake, expressed in mg/d, was not sex
dependent (P ҃ 0.52). Adjusting for magnesium intake, age did
not affect magnesium output (P ҃ 0.14).

Magnesium intake compared with output: statistical

model B
Magnesium output increased linearly with increases in mag-
nesium intake expressed as mg · kg body wtҀ1 · dҀ1 (P ҃ 0.0001)
(Figure 2). Accordingly, magnesium intake equaled magnesium
output at 2.36 mg · kg body wtҀ1 · dҀ1 (Y ҃ 0.306 ѿ 0.870 M; 95%
PI: 1.58, 3.38 mg · kg body wtҀ1 · dҀ1). Sex did not affect the
relation between magnesium intake and output (P ҃ 0.23) when
the relation was expressed as mg · kg body wtҀ1 · dҀ1. Adjusting
for magnesium intake, the effect of age was not statistically
significant (P ҃ 0.17) when the relation was expressed as mg · kg
body wtҀ1 · dҀ1.
FIGURE 2. Relation between magnesium output (fecal ѿ urinary excre-
tion) and magnesium intake, both expressed as mg · kg body wtҀ1 · dҀ1, for
women [F; n ҃ 150; weight: 71.6 앐 16.5 kg; age: 51.3 앐 17.4 y (range:
19 –77 y)] and men [E; n ҃ 93; weight: 76.3 앐 12.5 kg; age: 28.1 앐 8.1 y
(range: 19 – 65 y)] who participated in tightly controlled feeding studies
conducted in a metabolic unit (664 observations). Magnesium output (Y)
increased linearly with increases in magnesium intake (P ҃ 0.0001; n ҃ 664).
Across all subjects, neutral magnesium balance was maintained at magne-
sium intakes (M) of 2.36 mg · kg body wtҀ1 · dҀ1 [95% prediction interval
(PI): 1.58, 3.38 mg · kg body wtҀ1 · dҀ1 (thick dotted lines); Y ҃ 0.306 ѿ
0.870 M]. Solid thin lines indicate 95% CIs around the regression line.
850 HUNT AND JOHNSON
FIGURE 3. Relation between magnesium output (fecal ѿ urinary excre-
tion) and magnesium intake, both expressed as mg · kcalҀ1 · dҀ1, for women
[F; n ҃ 150; weight: 71.6 앐 16.5 kg; age: 51.3 앐 17.4 y (range: 19 –77 y)]
and men [E; n ҃ 93; weight: 76.3 앐 12.5 kg; age: 28.1 앐 8.1 y (range: 19 – 65
y)] who participated in tightly controlled feeding studies conducted in a
metabolic unit (664 observations). Magnesium output (Y) increased linearly
with increases in magnesium intake (P ҃ 0.0001; n ҃ 664). Across all
subjects, neutral magnesium balance was maintained at magnesium intakes
(M) of 0.075 mg · kcalҀ1 · dҀ1 [95% prediction interval (PI): 0.05, 0.11 mg
· kcalҀ1 · dҀ1 (thick dotted lines); Y ҃ 0.011 ѿ 0.857 M]. Solid thin lines
indicate 95% CIs around the regression line.
relation between magnesium output and intake did not depend on
sex (P ҃ 0.18) when expressed as mg · kcalҀ1 · dҀ1. Adjusting for
magnesium intake, the effect of age did not affect the relation
between magnesium intake and output (P ҃ 0.17) when the
relation was expressed as mg · kcalҀ1 · dҀ1.
Magnesium balance (mg/d) compared with age
Most male and female subjects were aged 19 – 40 y and 50 –75
y, respectively (Figure 4). To explore further the possible effects
of age, magnesium balance was modeled by using age as the
FIGURE 4. Relation between magnesium balance (mg/d) and age for
women [F; n ҃ 150; weight: 71.6 앐 16.5 kg; age: 51.3 앐 17.4 y (range:
19 –77 y)] and men [E; n ҃ 93; weight: 76.3 앐 12.5 kg; age: 28.1 앐 8.1 y
(range: 19 – 65 y)] who participated in tightly controlled feeding studies
conducted in a metabolic unit. There was no significant relation between
balance and age (P ҃ 0.9; n ҃ 664).
predictor. No significant relation between balance and age (in
mg/d) (P ҃ 0.9) was indicated.
DISCUSSION
This study expands the magnesium balance data needed to
generate better estimates of the adult magnesium requirement.
As outlined by the FNB (1), the number of magnesium balance
studies that meet minimum design criteria is insufficient. This
study exploited the metabolic data collected from studies in
which only healthy volunteers participated, adequate dietary ad-
aptation (44) was assured by examining only dietary periods
쏜 27 d, magnesium intakes below and near the presumed re-
quired amounts were included, the mineral content of the drink-
ing water was measured, foods and beverages were prepared by
professional staff members for delivery within 1% weighing
error, duplicate diets were prepared by professional staff mem-
bers, food consumption was quantitative and carried out under
visual supervision, fecal and urine collections were continuous to
ensure familiarity with the procedure, and samples were ana-
lyzed by state-of-the-art technology. Excluding data when in-
Downloaded from www.ajcn.org by guest on December 19, 2010
takes of measured minerals fell below the EARs eliminated the
possibility of confounding the data set with potential catabolic
states that result in release of magnesium from tissues (44).
Magnesium homeostasis
Characteristics of the statistical models indicate strong ho-
meostatic control of magnesium metabolism. The regression
lines crossed respective lines of unity at highly acute angles, a
characteristic that indicates magnesium balance is highly resis-
tant to change across a broad range of typical dietary magnesium
intakes. The homeostatic mechanisms seemed particularly active
below neutral magnesium balance. Within the range of dietary
magnesium intakes examined, the data did not reach an asymp-
tote, indicating no change in magnesium fractional intestinal
absorption over the typical range of dietary magnesium intakes.
Estimations of the average magnesium requirement
The statistical models used in the present study predict neutral
magnesium balance at magnesium intakes of 165 mg/d for
healthy persons regardless of age or sex. Therefore, compared
with the existing EAR, the new estimate of the magnesium re-
quirement is 35– 48% lower for women and 50 –53% lower for
men. The diets and environments used by the metabolic studies
were a reasonable approximation of the free-living environment.
For example, the amount and type (swimming, walking, station-
ary biking) exercise was prescribed per individual and therefore
varied considerably among subjects.
An EAR includes an adjustment for an assumed bioavailabil-
ity of the relevant nutrient (1). Therefore, the findings from the
present study may have particular relevance to the general pop-
ulation because most of the magnesium intake values examined
were generated from studies in which the diets were not supple-
mented with magnesium. Equally important, all diets were com-
posed of ordinary Western foods and constructed in such a man-
ner as to ensure a high degree of compositional heterogeneity
across meals and studies.
Some magnesium intake values used in the present study were
based on diets composed of ordinary Western foods and supple-
mented with magnesium (all data points in studies 17 and 21-23,
and 앒50% of all data points from studies 16, 18, 19, and 25-27).
 MAGNESIUM BALANCE AND REQUIREMENT
For all those studies except study 22, the supplemental magne-
sium was supplied as magnesium gluconate. If the bioavailability
of magnesium in magnesium gluconate is greater than that of
dietary magnesium, inclusion of such data should artificially
reduce the estimated magnesium requirement. However, find-
ings from the relevant metabolic studies indicate that providing
supplemental magnesium as magnesium gluconate did not im-
prove magnesium absorption. Instead, this form of supplemen-
tation reduced the percentage of magnesium intake excreted in
the urine (25, 27).
Comparative magnesium balance studies
Magnesium balance was considered the only reliable indicator
of magnesium status when setting the current adult magnesium
EARs (1). The FNB examined 9 magnesium balance studies (7,
9 –11, 45– 49) and excluded 2 because of inadequate assessment
of dietary magnesium near the predicted requirement (11) or
insufficient adaptation periods (45).
For adult women of all ages, the magnesium EARs are based
nearly exclusively on the findings from one study (7) of adults
(aged 20-53 y; 16 men; 18 women) who consumed self-selected
diets in a free-living environment. Subjects were responsible for
collection and transport of food and excreta, and magnesium
intakes decreased substantially for the female subjects during the
balance period. A separate balance study conducted elsewhere
with free-living female students aged 22–28 y fed controlled
diets for 22 d reported that mean magnesium intakes of 243–252
mg/d were sufficient to maintain balance close to equilibrium
when subjects were fed high-protein and low- or high-fiber diets
but were not as close when fed a high-fiber, low-protein diet (46).
For men, estimations of the magnesium requirement rely
heavily on results from the primary study used to establish the
EAR for women (7). Magnesium intakes decreased substantially
during the balance period, and mean magnesium balance was
negative. A different metabolic study (49) with 5 men diagnosed
with either psychoneurosis or osteoporosis showed that a mag-
nesium intake of 240 앐 24 mg/d resulted in negative magnesium
balance. Finally, a low-fiber diet providing 356 앐 10 mg Mg/d
resulted in a positive magnesium balance, whereas a high-fiber
diet that provided less magnesium (322 앐 12 mg/d) resulted in
negative magnesium balance (47). Comparable results were re-
ported from a similar study (48). For American men aged 욷 19 y,
the 5th and 95th percentile of dietary fiber intakes is 8.5 and 31.0
g/d (2). In the present study, diets supplied between 앒3.5 and
앒20 g fiber/d.
Magnesium balance and age or sex
Age did not affect the relation between magnesium intake and
output, similar to the findings of an earlier study (7). A separate
study (5) found no association with age on urinary magnesium
excretion (fecal magnesium excretion not measured), when ex-
pressed per millimole of creatinine. The effect of age on body
stores of magnesium in healthy older adults was examined pre-
viously (50, 51). An observed increase in percentage retention of
an intravenous magnesium load test in older adults (50) may have
been confounded by the severe unequal distribution of men and
women in the experimental group such that the findings may
relate more to sex or body size differences than to age. A separate
study reported a decrease in erythrocyte magnesium in the older
851
adults compared with younger adults (51), but findings on eryth-
rocyte magnesium content were not considered sufficiently pre-
dictive to be of use in establishing an EAR (1).
Findings that the relation between magnesium intake and out-
put is not affected by sex concur with earlier findings of balance
from a study with subjects on self-selected diets (7). In a separate
study, sex did not affect urinary magnesium excretion after cor-
rections were made for body size (5).
In summary, findings from the present study augment the
sparse information available for calculating the EAR for mag-
nesium. They suggest a lower magnesium requirement for
healthy men and women than estimated previously.
We thank the principal investigators of the original metabolic studies
conducted at the USDA ARS Grand Forks Human Nutrition Research Cen-
ter: Wesley K Canfield, Janet R Hunt, Phyllis E Johnson, Henry C Lukaski,
Glenn I Lykken, David B Milne, Forrest H Nielsen, James G Penland, and
Harold H Sandstead. We also thank Emily Nielsen, Bonita Hoverson, Debbie
Krause, Angela Scheett, and Rodger Sims for compilation of the data.
CDH was responsible for the conception of the experimental design, data
selection, data analysis, and the writing of the manuscript. LKJ was respon-
sible for data selection, data analysis, and statistical modeling procedures and
Downloaded from www.ajcn.org by guest on December 19, 2010
was involved in the preparation of the article. None of the authors had any
financial or personal conflict of interest.
REFERENCES
1. Food and Nutrition Board, Institute of Medicine. Dietary reference in-
takes for calcium, phosphorus, magnesium, vitamin D, and fluoride.
Washington, DC: National Academy Press, 1997.
2. Moshfegh A, Goldman J, Cleveland L. What we eat in America,
NHANES 2001-2002: usual nutrient intakes from food compared to
dietary reference intakes. Internet: http://www.ars.usda.gov/SP2User
Files/Place/12355000/pdf/usualintaketables2001-02.pdf (accessed 30
September 2005).
3. Ford ES, Mokdad AH. Dietary magnesium intake in a national sample of
US adults. J Nutr 2003;133:2879 – 82.
4. Ford ES, Mokdad AH, Giles WH, Brown DW. The metabolic syndrome
and antioxidant concentrations: findings from the Third National Health
and Nutrition Examination Survey. Diabetes 2003;52:2346 –52.
5. Lowik MR, van Dokkum W, Kistemaker C, Schaafsma G, Ockhuizen T.
Body composition, health status and urinary magnesium excretion
among elderly people (Dutch Nutrition Surveillance System). Magnes
Res 1993;6:223–32.
6. Martin BJ. The magnesium load test: experience in elderly subjects.
Aging (Milano) 1990;2:291– 6.
7. Lakshmanan FL, Rao RB, Kim WW, Kelsay JL. Magnesium intakes,
balances, and blood levels of adults consuming self-selected diets. Am J
Clin Nutr 1984;40:1380 –9.
8. Hunt SM, Schofield FA. Magnesium balance and protein intake level in
adult human female. Am J Clin Nutr 1969;22:367–73.
9. Mahalko JR, Sandstead HH, Johnson LK, Milne DB. Effect of a mod-
erate increase in dietary protein on the retention and excretion of Ca, Cu,
Fe, Mg, P, and Zn by adult males. Am J Clin Nutr 1983;37:8 –14.
10. Schwartz R, Apgar BJ, Wien EM. Magnesium absorption in human
subjects from leafy vegetables, intrinsically labeled with stable 26Mg.
Am J Clin Nutr 1984;39:571– 6.
11. Schwartz RP, Apgar BJ, Wien EM. Apparent absorption and retention of
Ca, Cu, Mg, Mn, and Zn from a diet containing bran. Am J Clin Nutr
1986;43:444 –55.
12. Beaton GH. Statistical approaches to establish mineral element recom-
mendations. J Nutr 1996;126:2320 – 8S.
13. Munoz JM, Sandstead HH, Jacob RA, et al. Effects of some cereal brans
and textured vegetable protein on plasma lipids. Am J Clin Nutr 1979;
32:580 –92.
14. Tucker DM, Sandstead HH, Logan GM, Jr, et al. Dietary fiber and
personality factors as determinants of stool output. Gastroenterology
1981;81:879 – 83.
15. Milne DB, Johnson PE, Klevay LM, Sandstead HH. Effect of copper
852 HUNT AND JOHNSON
intake on balance, absorption, and status indices of copper in men. Nutr
Res 1990;10:975– 86.
16. Canfield WK, Bolonchuck W, Henriksen L, et al. Cholesterol in men fed
a low zinc diet for four months. Clin Res 1982;30:762A (abstr).
17. Lukaski HC, Bolonchuk WW, Klevay LM, Mahalko JR, Milne DB,
Sandstead HH. Influence of type and amount of dietary lipid on plasma
lipid concentrations in endurance athletes. Am J Clin Nutr 1984;39:35–
44.
18. Lykken GI, Mahalko J, Johnson PE, et al. Effect of browned and un-
browned corn products intrinsically labeled with 65Zn on absorption of
65Zn in humans. J Nutr 1986;116:795– 801.
19. Johnson PE, Lykken GI. Copper-65 absorption by men fed intrinsically
and extrinsically labeled whole wheat bread. J Agric Food Chem 1988;
36:537– 40.
20. Milne DB, Lukaski HC, Johnson PE. Effect of folic acid supplements on
zinc balance and metabolism in men fed diets adequate in zinc. J Trace
Elem Exp Med 1990;3:319 –26.
21. Johnson PE, Stuart MA, Hunt JR, Mullen L, Starks TL. 65Copper ab-
sorption by women fed intrinsically and extrinsically labeled goose
meat, goose liver, peanut butter and sunflower butter. J Nutr 1988;118:
1522– 8.
22. Milne DB, Canfield WK, Gallagher SK, Hunt JR, Klevay LM. Ethanol
metabolism in postmenopausal women fed a diet marginal in zinc. Am J
Clin Nutr 1987;46:688 –93.
23. Lukaski HC, Klevay LM, Milne DB. Effects of dietary copper on human
autonomic cardiovascular function. Eur J Appl Physiol Occup Physiol
1988;58:74 – 80.
24. Gallagher SK, Johnson LK, Milne DB. Short-term and long-term vari-
ability of indices related to nutritional status. I: Ca, Cu, Fe, Mg, and Zn.
Clin Chem 1989;35:369 –73.
25. Hunt CD, Herbel JL, Nielsen FH. Metabolic response of postmenopausal
women to supplemental dietary boron and aluminum during usual and
low magnesium intake: boron, calcium, and magnesium absorption and
retention and blood mineral concentrations. Am J Clin Nutr 1997;65:
803–13.
26. Penland JG, Johnson PE. Dietary calcium and manganese effects on
menstrual cycle symptoms. Am J Obstet Gynecol 1993;168:1417–23.
27. Nielsen FH. The alteration of magnesium, calcium and phosphorus me-
tabolism by dietary magnesium deprivation in postmenopausal women
is not affected by dietary boron deprivation. Magnesium Res. 2004;17:
197–210.
28. Nielsen FH. Effects in healthy postmenopausal women of magnesium
deprivation. FASEB J 1990;4:A1044 (abstr).
29. Penland JO, Klevay LM. Caloric restriction and increased exercise in
mildly obese women. 1. Effects on menstrual symptomatology and
plasma monoamine oxidase activity. Proc N D Acad Sci 1992;46:52
(abstr).
30. Nielsen FH. Oxidant stress effects on the clinical and nutritional signif-
icance of trace elements. In: Abdulla M, Vohora SB, Athar M, eds. Trace
and toxic elements in nutrition and health. New Delhi, India: Wiley
Eastern, 1993:227– 41.
31. Hunt JR, Gallagher SK, Johnson LK, Lykken GI. High- versus low-meat
diets: effects on zinc absorption, iron status, and calcium, copper, iron,
magnesium, manganese, nitrogen, phosphorus, and zinc balance in post-
menopausal women. Am J Clin Nutr 1995;62:621–32.
32. Milne DB, Nielsen FH. Effects of a diet low in copper on copper-status
indicators in postmenopausal women. Am J Clin Nutr 1996;63:358 – 64.
33. Siders WA, Lukaski HC. Waist:hip ratio and regional body composition
changes caused by caloric and exercise intervention. Proc N D Acad Sci
1995;49:28 (abstr).
34. Milne DB, Nielsen FH. The interaction between dietary fructose and
magnesium adversely affects macromineral homeostasis in men. J Am
Coll Nutr 2000;19:31–7.
35. Nielsen FH, Milne DB. Some magnesium status indicators and oxidative
metabolism responses to low-dietary magnesium are affected by dietary
copper in postmenopausal women. Nutrition 2003;19:617–26.
36. Food and Nutrition Board, Institute of Medicine. Dietary reference in-
takes for vitamin A, vitamin K, arsenic, boron, chromium, copper, io-
dine, iron, manganese, molybdenum, nickel, silicon, vanadium, and
zinc. Washington, DC: National Academy Press, 2001.
37. Harris JA, Benedict FG. A biometric study of basal metabolism in man.
Philadelphia, PA: Lippincott, 1919.
38. Hunt CD, Johnson PE, Herbel J, Mullen LK. Effects of dietary zinc
depletion on seminal volume and zinc loss, serum testosterone concen-
trations, and sperm morphology in young men. Am J Clin Nutr 1992;
56:148 –57.
39. Lukaski HC, Nielsen FH. Dietary magnesium depletion affects meta-
bolic responses during submaximal exercise in postmenopausal women.
J Nutr 2002;132:930 –5.
40. St-Pierre NR. Invited review: integrating quantitative findings from
Downloaded from www.ajcn.org by guest on December 19, 2010
multiple studies using mixed model methodology. J Dairy Sci 2001;84:
741–55.
41. Kleinbaum DG, Kupper LL. Applied regression analysis and other mul-
tivariable methods. North Scituate, MA: Duxbury Press, 1978.
42. Oman SD. Calibration with random slopes. Biometrika 1998;85:439 –
49.
43. Vecchia DF, Iyer HK, Chapman PL. Calibration with randomly chang-
ing standard curves. Technometrics 1989;31:83–90.
44. Shils ME, Rude RK. Deliberations and evaluations of the approaches,
endpoints and paradigms for magnesium dietary recommendations. J
Nutr 1996;126:2398 – 403S.
45. Greger JL, Baier MJ. Effect of dietary aluminum on mineral metabolism
of adult males. Am J Clin Nutr 1983;38:411–9.
46. Wisker E, Nagel R, Tanudjaja TK, Feldheim W. Calcium, magnesium,
zinc, and iron balances in young women: effects of a low-phytate barley-
fiber concentrate. Am J Clin Nutr 1991;54:553–9.
47. Kelsay JL, Behall KM, Prather ES. Effect of fiber from fruits and veg-
etables on metabolic responses of human subjects, II. Calcium, magne-
sium, iron, and silicon balances Am J Clin Nutr 1979;32:1876 – 80.
48. Kelsay JL, Prather ES. Mineral balances of human subjects consuming
spinach in a low-fiber diet and in a diet containing fruits and vegetables.
Am J Clin Nutr 1983;38:12–9.
49. Spencer H, Fuller H, Norris C, Williams D. Effect of magnesium on the
intestinal absorption of calcium in man. J Am Coll Nutr 1994;13:485–92.
50. Gullestad L, Nes M, Ronneberg R, Midtvedt K, Falch D, Kjekshus J.
Magnesium status in healthy free-living elderly Norwegians. J Am Coll
Nutr 1994;13:45–50.
51. Paolisso G, Sgambato S, Gambardella A, et al. Daily magnesium sup-
plements improve glucose handling in elderly subjects. Am J Clin Nutr
1992;55:1161–7.