NUTRITIONAL NEUROSCIENCE

https://doi.org/10.1080/1028415X.2021.1885241

Maternal metal intake during pregnancy and childhood behavioral problems in

Japan: the Kyushu Okinawa Maternal and Child Health Study
a,b,c a,b,c
Yoshihiro Miyake , Keiko Tanaka , Hitomi Okuboa,d, Satoshi Sasakie, Akiko Tokinobua,c and
Masashi Arakawaf,g
a
Department of Epidemiology and Preventive Medicine, Ehime University Graduate School of Medicine, Ehime, Japan; bResearch Promotion
Unit, Translational Research Center, Ehime University Hospital, Ehime, Japan; cCenter for Data Science, Ehime University, Ehime, Japan;
d
Department of Health Promotion, National Institute of Public Health, Saitama, Japan; eDepartment of Social and Preventive Epidemiology,
School of Public Health, The University of Tokyo, Tokyo, Japan; fWellness Research Fields, Faculty of Global and Regional Studies, University of
the Ryukyus, Okinawa, Japan; gThe Department of Cross Cultural Studies, Osaka University of Tourism, Okinawa, Japan

ABSTRACT KEYWORDS
Objectives: Lower maternal metal intake during pregnancy might affect childhood development. Behavioral problems;
The current prebirth cohort study investigated the relationship between maternal intake of zinc, Japanese children;
magnesium, iron, copper, and manganese during pregnancy and behavioral problems in magnesium; maternal intake;
prebirth cohort
Japanese children aged five years.
Methods: Subjects were 1199 mother–child pairs. Dietary intake during the preceding month was
assessed using a diet history questionnaire. Emotional, conduct, hyperactivity, and peer problems
and low prosocial behavior were assessed using the Strengths and Difficulties Questionnaire.
Maternal age, gestation at baseline, region of residence, number of children, maternal and
paternal education, household income, maternal depressive symptoms during pregnancy,
maternal alcohol intake during pregnancy, maternal smoking during pregnancy, child’s birth
weight, child’s sex, breastfeeding duration, smoking in the household during the first year of
life, and some dietary confounders that were associated with outcomes under study in this
population were adjusted for.
Results: Compared with maternal magnesium intake during pregnancy in the first quartile,
magnesium intake in the second, third, and fourth quartiles was independently inversely related
to childhood hyperactivity problems, but not to emotional, conduct, or peer problems or low
prosocial behavior: the adjusted odds ratio between extreme quartiles was 0.48 (95%
confidence interval: 0.23−0.99, P for trend = 0.04). No evident associations were observed
between maternal intake of zinc, iron, copper, or manganese during pregnancy and childhood
emotional, conduct, hyperactivity, or peer problems or low prosocial behavior.
Conclusions: The present study suggests that higher maternal magnesium intake during
pregnancy is inversely associated with hyperactivity problems in Japanese children.

Introduction
Lower maternal metal intake during pregnancy might
affect childhood cognitive, motor, and socio-emotional
development [1]. A prebirth cohort study in Singapore
showed that higher maternal zinc concentrations during
pregnancy were positively associated with four-year-old
children’s picture description and spatial identification
abilities and that higher maternal magnesium concen-
trations were positively associated with the children’s
ability to identify letters and writing [2]. A 2019 meta-
analysis that included seven studies demonstrated that
subjects with attention deficit hyperactivity disorder
(ADHD) had lower serum magnesium levels compared
with healthy controls [3]. Magnesium is suspected to
play an important role in the development and function
of nerve tissue [4]. Findings on possible associations
between zinc, iron, and copper concentrations and
ADHD symptoms have been contradictory, but zinc,
iron, and copper levels might be related to dopamine
levels [4]. To our knowledge, no epidemiological studies
have examined the relationship between maternal
intake of zinc, magnesium, iron, copper, or manganese
during pregnancy and the risk of childhood behavioral
problems. In an Australian birth cohort study of 684
children, which assessed mental health symptoms
using the Youth Self-Report at the 14- and 17-year fol-
low-ups, higher intake of magnesium, but not zinc, at
age 14 and age 17 was significantly associated with
reduced externalizing behaviors, while there were no
significant associations between intake of magnesium

CONTACT Yoshihiro Miyake yoshihiro.miyake.ls@ehime-u.ac.jp Department of Epidemiology and Preventive Medicine, Ehime University Graduate
School of Medicine, Ehime, 791-0295, Japan
© 2021 Informa UK Limited, trading as Taylor & Francis Group
2 Y. MIYAKE ET AL.
or zinc and internalizing behavior problems [5]. A
Spanish case–control study of 100 patients diagnosed
with ADHD and 100 healthy children with a mean
age of 11 years showed that the daily intake of zinc,
magnesium, and iron was significantly higher in the
control group than in the ADHD group [6]. Here, we
investigated the relationship between maternal intake
of zinc, magnesium, iron, copper, and manganese
during pregnancy and the risk of behavioral problems
in Japanese children at five years of age using data
from the Kyushu Okinawa Maternal and Child Health
Study (KOMCHS).
Methods
Study population
Study subjects were mother–child pairs from the
KOMCHS, a prospective prebirth cohort study [7–12].
Details of the baseline survey of the KOMCHS have
been described previously [7]. Briefly, between April
2007 and March 2008, 423 obstetric hospitals in seven
prefectures on Kyushu Island in southern Japan, with a
total population of approximately 13.26 million, and in
Okinawa Prefecture, an island chain in the southwest of
Japan with a total population of nearly 1.37 million, pro-
vided a set of leaflets explaining the KOMCHS, an appli-
cation form to take part in the KOMCHS, and a self-
addressed and stamped return envelope to as many preg-
nant women as possible. Pregnant women who wanted to
participate in the KOMCHS filled out the application
form and mailed it to the data management center.
Using the contact information in this form, research tech-
nicians gave each pregnant woman a detailed explanation
of the KOMCHS by telephone and sent her a self-admi-
nistered questionnaire after obtaining her agreement.
Ultimately, 1757 pregnant women between the fifth and
39th week of pregnancy gave their written informed con-
sent to take part in the KOMCHS and completed this self-
administered questionnaire at baseline. Of the 1757 preg-
nant women, 1590, 1527, 1430, 1362, 1305, 1264, and
1201 mother–child pairs took part in all surveys from
the baseline survey to the second (after delivery), third
(around four months postpartum), fourth (around 12
months postpartum), fifth (around 24 months postpar-
tum), sixth (around 36 months postpartum), seventh
(around 48 months postpartum), and eighth (around
60 months postpartum) surveys, respectively. After two
pairs with missing data on household income were
excluded, the final sample for the present analysis com-
prised 1199 pairs. The ethics committees of Fukuoka Uni-
versity School of Medicine and Ehime University
Graduate School of Medicine approved the KOMCHS.
Measurements
At the time of each survey, the study’s data management
center sent a self-administered questionnaire to the
study subjects who then filled out the questionnaire
and returned it to the data management center.
Research technicians completed missing or illogical
data by telephone interview.
In the baseline survey, the first part of the question-
naire elicited data on maternal age, gestation, region
of residence, number of children, maternal and paternal
education, household income, and depressive symp-
toms. Depressive symptoms were measured using a
Japanese version [13] of the 20-item Center for Epide-
miologic Studies Depression Scale (CES-D) instrument
[14]. The range of possible scores on this scale is 0–
60, with higher scores indicating more depressive symp-
toms. A cut-off of 16 was used to determine the presence
of depressive symptoms [13,14].
The second part of the questionnaire at baseline con-
sisted of a semi-quantitative, comprehensive diet history
questionnaire (DHQ) designed to assess the dietary
intake of Japanese adults over the previous month [15–
21]. Pregnant women were asked to report their average
consumption frequency of eight categories of foods and
beverages, ranging from ‘never’ to ‘≥ 2 times/d’ for
foods and from ‘< 1 time/wk’ to ‘≥ 6 times/d’ for bev-
erages, and their relative portion size in comparison
with a standard portion size according to five categories:
‘50% smaller or less,’ ‘20%–30% smaller,’ ‘the same,’
‘20%–30% larger,’ and ‘50% larger or more’ [15]. Esti-
mates of daily intake for 150 food and beverage items,
as well as for energy, nutrients, and alcohol, were calcu-
lated using an ad hoc computer algorithm for the DHQ,
which was based on the Standard Tables of Food Compo-
sition in Japan [22]. In a validation study of 92 Japanese
women aged 31–69 years, the Pearson’s correlation
coefficients between the DHQ and 16-day semi-weighed
dietary records were 0.60 for zinc, 0.64 for magnesium,
0.67 for iron, 0.66 for copper, 0.48 for manganese, 0.52
for vitamin C, 0.55 for folate, 0.68 for vitamin B6, 0.57
for vitamin B2, and 0.56 for calcium [17]. All dietary fac-
tors were adjusted for total energy intake according to the
residual method [23]. The DHQ included questions on
consumption of six types of alcoholic beverages: beer,
Japanese sake (rice wine), shochu (a distilled alcoholic
beverage made in Japan), chuhai (a beverage containing
shochu and carbonated water), whiskey, and wine. Data
on dietary supplements were not used in the calculation
of dietary intake because of the lack of a reliable compo-
sition table in Japan.
The questionnaire in the second survey inquired
about the baby’s sex, birth weight, date of birth, and
 NUTRITIONAL NEUROSCIENCE 3

Table 1. Characteristics of 1199 parent-child pairs in relation to quartile (Q) of maternal magnesium intake during pregnancya.
Total Magnesium
(n = 1199)b Q1 Q2 Q3 Q4 P for trendc
Baseline characteristics
Maternal age, years 32.0 (29.0−34.0) 31.0 32.0 32.0 32.0 0.001
Gestation, weeks 17.0 (14.0−21.0) 17.0 17.0 17.0 18.0 0.08
Region of residence, % 0.96

Fukuoka Prefecture 57.8 55.9 56.0 64.3 55.0

Other than Fukuoka Prefecture in Kyushu 32.8 32.1 35.0 28.0 36.0

Okinawa Prefecture 9.4 12.0 9.0 7.7 9.0

Number of living children already born to same mother, % 0.44

0 40.4 37.8 39.0 44.0 40.7

1 40.0 39.8 44.0 39.0 37.3

≥2 19.6 22.4 17.0 17.0 22.0

Maternal education, years, % 0.003
< 13 20.9 28.4 18.0 19.3 17.7
13−14 33.3 31.1 34.7 36.3 31.0
≥ 15 45.9 40.5 47.3 44.3 51.3
Paternal education, years, % 0.06

< 13 28.6 32.8 26.7 30.3 24.7

13−14 14.4 15.4 15.7 10.0 16.7

≥ 15 57.0 51.8 57.7 59.7 58.7
Household income, yen/year, % 0.06

< 4,000,000 32.2 37.5 34.3 28.7 28.3

4,000,000−5,999,999 37.5 36.1 30.3 42.0 41.3

≥ 6,000,000 30.4 26.4 35.3 29.3 30.3

Maternal depressive symptoms during pregnancy, % 18.2 24.4 18.3 18.7 11.3 < 0.0001

Maternal alcohol intake during pregnancy, % 13.2 15.7 13.0 11.0 13.0 0.25

Maternal daily intake

Total energy, kJ 7127 (6117−8465) 7239 6810 7052 7439 0.58

Zinc, mg 7.2 (6.6−7.8)

Magnesium, mg 211.8 (187.8−241.2)

Iron, mg 6.6 (5.7−7.5)

Copper, mg, 1.0 (0.9−1.2)

Manganese, mg, 3.6 (2.8−4.5)

d
Vitamin C, mg 96.8 (76.5−123.9) 76.2 93.4 106.1 118.1 < 0.0001

Folate, μgd 277.4 (233.6−329.1) 210.7 261.4 296.4 348.9 < 0.0001
d
Vitamin B6, mg 0.9 (0.8−1.1) 0.8 0.9 1.0 1.2 < 0.0001

Vitamin B2, mgd 1.3 (1.1−1.5) 1.1 1.3 1.3 1.4 < 0.0001
d
Calcium, mg 481.6 (398.2−587.0) 367.3 462.2 532.8 570.7 < 0.0001

Characteristics at the postnatal assessment

Maternal smoking during pregnancy, % 7.3 11.4 6.7 6.7 4.3 0.002
Birth weight, g 3012 (2772−3246) 3034 3020 2980 3011 0.03
Male gender, % 47.4 44.8 46.3 45.3 53.0 0.07
Breastfeeding duration, months, % 0.002
<6 10.8 14.4 12.3 10.0 6.7

(Continued)
4 Y. MIYAKE ET AL.
Table 1. Continued.
Total
(n = 1199)b
≥6 89.2
Smoking in household during the first year of life, % 27.4
a
Values are medians for continuous variables and percentages of subjects for categorical variables.
b
Values are medians (interquartile ranges) for continuous variables and percentages of subjects for categorical variables.
c
For continuous variables, a linear trend test was used; for categorical variables, a Mantel-Haenszel χ2-test was used.
d
Food and nutrient intake levels were adjusted for total energy intake using the residual method.
maternal smoking during pregnancy. The question-
naires in the third and fourth surveys included ques-
tions on household smoking and breastfeeding
duration. In the eighth survey, childhood behavioral
problems at the age of five years were assessed using
the Japanese parent–report version of the Strengths
and Difficulties Questionnaire (SDQ) for three- to 16-
year-old children [24]. The SDQ consists of 25 items
rated on a three-point Likert Scale: not true = 0, some-
what true = 1, and certainly true = 2. It has five scales:
an emotional problems scale, a conduct problems
scale, a hyperactivity scale, a peer problems scale, and
a prosocial scale. Each scale has five items; thus, each
scale was scored between 0 and 10. A high score on
the prosocial scale is considered a behavioral strength,
whereas high scores on the other four scales indicate
behavioral problems. These scale scores were categor-
ized into three levels representing ‘normal,’ ‘borderline,’
and ‘abnormal’ degrees of difficulty according to cut-off
points that had previously been established in a sample
of Japanese children [25]. Because of the small number
of children with abnormal scores in the present study,
the five scale scores were dichotomized so that children
with borderline and abnormal scores were considered as
a single group to be compared with children with nor-
mal scores; we defined emotional problems, conduct
problems, hyperactivity problems, peer problems, or
low prosocial behavior as present when a child had a
borderline or abnormal score in the respective scale.
Statistical analysis
Maternal metal intake during pregnancy was categor-
ized at quartile points according to its distribution
among the 1199 mothers. Maternal age, gestation at
baseline, region of residence at baseline, number of chil-
dren at baseline, maternal and paternal education,
household income, maternal depressive symptoms
during pregnancy, maternal alcohol intake during preg-
nancy, maternal smoking during pregnancy, child’s
birth weight, child’s sex, breastfeeding duration, and
smoking in the household during the first year of life
were selected a priori as potential non-dietary
Magnesium
Q1 Q2 Q3 Q4 P for trendc
85.6 87.7 90.0 93.3
30.4 27.3 28.7 23.3 0.08
confounding factors. Maternal age, gestation, and
birth weight were used as continuous variables. The fol-
lowing dietary confounding factors that were signifi-
cantly associated with childhood conduct or
hyperactivity problems or low prosocial behavior in
this population were also adjusted for as continuous
variables: maternal intake of vitamin B2 and calcium
during pregnancy was adjusted for in analyses of the
association with childhood emotional problems;
maternal intake of vitamin C, vitamin B6, and calcium
during pregnancy was adjusted for in analyses of the
association with childhood hyperactivity problems;
and maternal intake of vitamin C, folate, and vitamin
B6 during pregnancy was adjusted for in analyses of
the association with childhood low prosocial behavior
[10–12].
Multiple logistic regression analysis was employed to
estimate adjusted odds ratios (ORs) and 95% confidence
intervals (CIs) for each behavioral problem for each
quartile of dietary intake of each metal under study,
and the lowest quartile was used as the reference. The
linear trend was tested using a logistic regression
model assigning consecutive integers (1−4) to the quar-
tiles of maternal metal intake. All statistical analyses
were carried out using the SAS software package version
9.4 (SAS Institute Inc., Cary, NC, USA).
Results
Among the 1199 children aged 59–71 months, the
prevalence values of emotional problems, conduct pro-
blems, hyperactivity problems, peer problems, and low
prosocial behavior were 12.9%, 19.4%, 13.1%, 8.6%,
and 29.2%, respectively. Median age of the 1199
mothers at baseline was 32.0 years (interquartile range
[IQR]: 29.0−34.0) (Table 1). During pregnancy, the
mothers’ median daily total energy intake and median
daily energy-adjusted intake of zinc, magnesium, iron,
copper, and manganese were 7127 kJ (IQR: 6117
−8465), 7.2 mg (IQR: 6.6−7.8), 211.8 mg (IQR: 187.8
−241.2), 6.6 mg (IQR: 5.7−7.5), 1.0 mg (IQR: 0.9
−1.2), and 3.6 mg (IQR: 2.8−4.5), respectively. Maternal
magnesium intake during pregnancy was positively
Table 2. Odds ratios (ORs) and 95% confidence intervals (CIs) for behavioral problems assessed by the Strength and Difficulties Questionnaire in 1199 children aged 5 years according to
quartile of maternal metal intake during pregnancy.
Emotional problems Conduct problems Hyperactivity problems Peer problems Low prosocial behavior
Variablesa Risk (%) Adjusted OR (95% CI)b Risk (%) Adjusted OR (95% CI)c Risk (%) Adjusted OR (95% CI)d Risk (%) Adjusted OR (95% CI)c Risk (%) Adjusted OR (95% CI)e
Zinc
Q1 (6.2) 18.4 1.00 22.4 1.00 15.4 1.00 9.0 1.00 30.4 1.00
Q2 (6.9) 12.0 0.67 (0.41−1.08) 21.0 1.06 (0.71−1.59) 12.0 0.90 (0.54−1.49) 9.0 1.14 (0.64−2.03) 32.3 1.30 (0.89−1.89)
Q3 (7.4) 10.7 0.60 (0.36−0.99) 15.3 0.75 (0.48−1.15) 12.3 1.01 (0.58−1.74) 7.3 0.91 (0.50−1.67) 26.3 1.10 (0.73−1.66)
Q4 (8.2) 10.7 0.67 (0.39−1.16) 19.0 1.08 (0.71−1.64) 12.7 1.29 (0.69−2.40) 9.0 1.24 (0.69−2.23) 27.7 1.52 (0.95−2.41)
P for trend 0.11 0.85 0.42 0.65 0.17
Magnesium
Q1 (173.0) 15.1 1.00 24.4 1.00 19.4 1.00 11.0 1.00 36.1 1.00
Q2 (200.3) 12.0 0.94 (0.56−1.57) 16.7 0.66 (0.43−0.99) 11.7 0.55 (0.33−0.91) 7.3 0.71 (0.39−1.26) 27.3 0.80 (0.54−1.17)
Q3 (225.1) 12.7 0.98 (0.57−1.68) 17.7 0.78 (0.51−1.17) 11.0 0.49 (0.27−0.88) 7.3 0.67 (0.37−1.20) 26.7 0.87 (0.56−1.33)
Q4 (266.8) 12.0 1.23 (0.69−2.20) 19.0 0.91 (0.60−1.37) 10.3 0.48 (0.23−0.99) 8.7 0.89 (0.50−1.57) 26.7 1.02 (0.60−1.73)
P for trend 0.49 0.77 0.04 0.61 0.92
Iron
Q1 (5.2) 15.1 1.00 20.7 1.00 17.7 1.00 10.4 1.00 34.1 1.00
Q2 (6.2) 11.0 0.89 (0.52−1.49) 18.3 0.92 (0.60−1.39) 14.7 0.94 (0.58−1.52) 7.0 0.72 (0.39−1.30) 29.0 1.06 (0.72−1.57)
Q3 (6.9) 14.3 1.23 (0.73−2.08) 20.7 1.06 (0.70−1.59) 9.3 0.54 (0.31−0.93) 10.0 1.03 (0.60−1.77) 28.3 1.22 (0.79−1.89)
Q4 (8.2) 11.3 1.06 (0.59−1.89) 18.0 1.02 (0.67−1.55) 10.7 0.72 (0.39−1.32) 7.0 0.72 (0.39−1.29) 25.3 1.47 (0.83−2.59)
P for trend 0.58 0.77 0.09 0.50 0.18
Copper
Q1 (0.9) 15.7 1.00 22.7 1.00 16.4 1.00 10.4 1.00 34.1 1.00
Q2 (1.0) 11.7 0.85 (0.51−1.39) 17.0 0.76 (0.50−1.16) 14.3 0.94 (0.58−1.52) 8.7 0.89 (0.50−1.57) 28.0 0.81 (0.56−1.18)
Q3 (1.1) 12.7 0.90 (0.55−1.47) 19.3 0.94 (0.62−1.42) 13.3 0.93 (0.56−1.55) 8.3 0.86 (0.48−1.52) 30.7 1.11 (0.75−1.65)
Q4 (1.3) 11.7 0.90 (0.54−1.51) 18.7 0.95 (0.63−1.44) 8.3 0.57 (0.31−1.05) 7.0 0.71 (0.39−1.29) 24.0 0.87 (0.55−1.38)
P for trend 0.74 0.96 0.11 0.28 0.95
Manganese
Q1 (2.4) 13.7 1.00 19.4 1.00 12.7 1.00 10.4 1.00 31.8 1.00
Q2 (3.2) 14.7 1.16 (0.71−1.89) 19.0 0.97 (0.64−1.48) 16.0 1.45 (0.89−2.36) 7.7 0.70 (0.39−1.25) 32.0 1.08 (0.75−1.56)
Q3 (4.0) 13.3 1.03 (0.62−1.71) 22.0 1.28 (0.85−1.93) 11.0 1.06 (0.62−1.79) 9.0 0.89 (0.51−1.55) 27.3 0.99 (0.67−1.45)
Q4 (5.2) 10.0 0.76 (0.42−1.38) 17.3 0.86 (0.56−1.32) 12.7 1.35 (0.79−2.32) 7.3 0.72 (0.39−1.29) 25.7 0.98 (0.64−1.49)
P for trend 0.39 0.83 0.52 0.41 0.83
a
Quartile medians in mg/day adjusted for energy intake using the residual method are given in parentheses.
b
Adjustment for maternal age, gestation at baseline, region of residence at baseline, number of children at baseline, maternal and paternal education, household income, maternal depressive symptoms during pregnancy,
maternal alcohol intake during pregnancy, maternal smoking during pregnancy, child’s birth weight, child’s sex, breastfeeding duration, smoking in the household during the first year of life, and maternal intake of
vitamin B2 and calcium during pregnancy.
c
Adjustment for maternal age, gestation at baseline, region of residence at baseline, number of children at baseline, maternal and paternal education, household income, maternal depressive symptoms during pregnancy,
maternal alcohol intake during pregnancy, maternal smoking during pregnancy, child’s birth weight, child’s sex, breastfeeding duration, and smoking in the household during the first year of life.
d
Adjustment for maternal age, gestation at baseline, region of residence at baseline, number of children at baseline, maternal and paternal education, household income, maternal depressive symptoms during pregnancy,
maternal alcohol intake during pregnancy, maternal smoking during pregnancy, child’s birth weight, child’s sex, breastfeeding duration, smoking in the household during the first year of life, and maternal intake of
vitamin C, vitamin B6, and calcium during pregnancy.
e
Adjustment for maternal age, gestation at baseline, region of residence at baseline, number of children at baseline, maternal and paternal education, household income, maternal depressive symptoms during pregnancy,
maternal alcohol intake during pregnancy, maternal smoking during pregnancy, child’s birth weight, child’s sex, breastfeeding duration, smoking in the household during the first year of life, and maternal intake of
vitamin C, folate, and vitamin B6 during pregnancy.
NUTRITIONAL NEUROSCIENCE
5
6 Y. MIYAKE ET AL.
associated with maternal age, maternal educational
level, maternal intake levels of vitamin C, folate, vitamin
B6, vitamin B2, and calcium during pregnancy, and
breastfeeding duration and inversely related to maternal
depressive symptoms during pregnancy, maternal
smoking during pregnancy, and child’s birth weight.
After adjustment for potential non-dietary and diet-
ary confounding factors, compared with maternal mag-
nesium intake during pregnancy in the first quartile,
maternal magnesium intake levels in the second, third,
and fourth quartiles were independently inversely
related to the risk of hyperactivity problems in the chil-
dren, showing a clear inverse exposure–response
relationship: the adjusted OR between extreme quartiles
was 0.48 (95% CI: 0.23−0.99, P for trend = 0.04) (Table
2). Maternal magnesium intake during pregnancy was
not independently associated with the risk of childhood
emotional, conduct, or peer problems or low prosocial
behavior. No evident associations were observed
between maternal intake of zinc, iron, copper, or
manganese during pregnancy and the risk of childhood
emotional, conduct, hyperactivity, or peer problems or
low prosocial behavior.
Discussion
To our knowledge, the current prebirth cohort study is
the first to find an independent inverse association
between maternal magnesium intake during pregnancy
and the risk of childhood hyperactivity problems, but
not emotional, conduct, or peer problems or low proso-
cial behavior. Maternal intake levels of zinc, iron, cop-
per, or manganese during pregnancy were not related
to any of the childhood behavioral problems under
study. The current results are in partial agreement
with those of an Australian study that reported a signifi-
cant inverse relationship between adolescents’ intake of
magnesium, but not zinc, at ages 14 and 17 years and
externalizing, but not internalizing, behavior problems
[5] and with those of a Spanish study showing signifi-
cant inverse associations between subjects’ intake of
zinc, magnesium, and iron and ADHD [6].
Magnesium is a cofactor in more than 300 enzyme
systems that regulate diverse biochemical reactions in
the body, including protein synthesis, muscle and
nerve transmission, neuromuscular conduction, signal
transduction, blood glucose control, and blood pressure
regulation [26]. Magnesium also plays a key role in the
active transport of calcium and potassium ions across
cell membranes, a process that is important for nerve
impulse conduction, muscle contraction, vasomotor
tone, and normal heart rhythm [26]. A French study
showed that intra-erythrocyte Mg2+ levels were lower
in 52 ADHD children than in healthy adult subjects,
then confirmed that, after combined Mg2+/vitamin B6
supplementation (100 mg/day) for 3–24 weeks, intra-
erythrocyte Mg2+ levels among the ADHD children
rose nearly to the levels seen in healthy adult subjects
and the children’s hyperexcitability symptoms (physical
aggression, instability, inattention in school, hypertony,
spasm, and myoclony) diminished [27]. Given that ionic
magnesium depletion could contribute to the onset of
hyperactivity problems, higher maternal magnesium
intake during pregnancy might offer persistent protec-
tive effects against the development of childhood hyper-
activity problems.
One methodological advantage of the current inves-
tigation is that the data were prospectively collected
from a relatively large sample over a long interval
extending from the prenatal period to five years of
age. Another is that adjustment was made for potential
non-dietary and dietary confounding factors after
extensively detailed information was collected on these
factors.
The current study should be interpreted with certain
limitations in mind, however. We cannot exclude the
possibility of residual confounding, namely, distortion
that might remain after controlling for the confounders
under study. Although the validity of the DHQ regard-
ing intake levels of the metals under study seems reason-
able, as mentioned above, the DHQ can only
approximate consumption and was designed to assess
dietary intake for one month prior to the question-
naire’s completion. As the DHQ was answered any-
where between the 5th and 39th week of pregnancy, it
cannot reflect maternal dietary intake over the entire
course of the pregnancy. Maternal participants likely
have the same probability of being misclassified in
relation to their metal intake status regardless of
whether their children will later develop emotional,
conduct, hyperactivity, or peer problems or low proso-
cial behavior, and this type of non-differential mis-
classification would lead to an underestimation of the
strength of the association between exposure and out-
come, that is, it would dilute the effect of the exposure.
With respect to the SDQ, we are fully aware that
parents’ reports on their children’s development may
be biased. Moreover, it is unknown whether the cut-
off points chosen for the dichotomization of the five
scale scores derived from the SDQ are reasonable,
although they were chosen based on a previous study
performed in Japan [25]. Misclassification of the out-
come was likely to occur at similar rates among those
exposed and those unexposed; this type of non-differen-
tial outcome misclassification would attenuate the esti-
mated effect toward the null.

Of the 1757 participants in the baseline survey of the
KOMCHS, 556 mother–child pairs were not included in
the current study due to being lost to follow-up by the
eighth survey. No differences were found between the
556 non-participants and the 1201 participants in the
eighth survey with respect to the distribution of number
of children, depressive symptoms during pregnancy,
and alcohol intake during pregnancy. Compared with
non-participants in the eighth survey, participants
were more likely to be older, to have participated in
the baseline survey earlier in their gestation, to live in
Fukuoka Prefecture, and to report high maternal and
paternal educational levels and high household income.
At baseline, we could not estimate the participation rate
because data on the number of pregnant women who
were provided with a set of leaflets explaining the
KOMCHS, an application form, and a self-addressed
and stamped return envelope by the 423 collaborating
obstetric hospitals were not available. Of the 1757
mothers at baseline, 978 mothers lived in Fukuoka Pre-
fecture. According to data collected by the government
of Fukuoka Prefecture, the number of childbirths was
46,393 in 2007 and 46,695 in 2008, indicating that the
participation rate must have been low. The present
study’s subjects were also probably not representative
of Japanese women in the general population. Accord-
ing to a population census conducted in 2000 in
Fukuoka Prefecture, the percentages of women aged
30–34 years with < 13, 13−14, ≥ 15, and an unknown
number of years of education were 52.0%, 31.5%,
11.8%, and 4.8%, respectively [28]. The corresponding
figures for our study were 20.9%, 33.3%, 45.9%, and
0.0%, respectively.
In conclusion, the current prebirth cohort study pro-
vides evidence that higher maternal magnesium intake
during pregnancy is independently associated only
with a reduced risk of hyperactivity problems in Japa-
nese children. No relationships were evident between
maternal intake of zinc, iron, copper, or manganese
during pregnancy and any of the childhood behavioral
problems under study. Further epidemiological studies
should be conducted to replicate our analyses, and
further studies are also needed to clarify the underlying
biological mechanisms of this inverse association.
Availability of data and materials
The datasets are not publicly available due to ethical
considerations. The authors will support access to the
data where requests are reasonable and consistent with
the ethical approval of the Ethics Committee of the
Ehime University Graduate School of Medicine.
NUTRITIONAL NEUROSCIENCE 7
Contributors
The authors would like to thank the Kyushu Branch of
the Japan Allergy Foundation, the Fukuoka Association
of Obstetricians & Gynecologists, the Okinawa Associ-
ation of Obstetricians & Gynecologists, the Miyazaki
Association of Obstetricians & Gynecologists, the Oita
Association of Obstetricians & Gynecologists, the
Kumamoto Association of Obstetricians & Gynecolo-
gists, the Nagasaki Association of Obstetricians & Gyne-
cologists, the Kagoshima Association of Obstetricians &
Gynecologists, the Saga Association of Obstetricians &
Gynecologists, the Fukuoka Society of Obstetrics and
Gynecology, the Okinawa Society of Obstetrics and
Gynecology, the Fukuoka City Government, and the
Fukuoka City Medical Association for their valuable
support.
Disclosure statement
No potential conflict of interest was reported by the author(s).
Ethics approval
The ethics committees of Fukuoka University School of
Medicine and Ehime University Graduate School of
Medicine approved the Kyushu Okinawa Maternal
and Child Health Study.
Funding
This study was supported by JSPS KAKENHI grant numbers
19590606JP, 20791654JP, 21590673JP, 22592355JP,
22119507JP, 24390158JP, 25463275JP, 25670305JP,
17K12011JP, and 17H04135JP; by Health and Labour Sciences
Research Grants for Research on Allergic Disease and Immu-
nology and Health Research on Children, Youth and Families
from the Ministry of Health, Labour and Welfare, Japan; by
Meiji Co. Ltd.; and by the Food Science Institute Foundation;
Japan Society for the Promotion of Science London.
Notes on contributors
Yoshihiro Miyake is a Professor in the Department of Epide-
miology and Preventive Medicine, Ehime University Gradu-
ate School of Medicine, Ehime, Japan. He has a MD, PhD
and an epidemiologist. He is a principal investigator of several
projects such as the Kyushu Okinawa Maternal and Child
Health Study, Aidai Cohort Study, and Japan Ulcerative Coli-
tis Study.
Keiko Tanaka is a Associate Professor in the Department of
Epidemiology and Preventive Medicine, Ehime University
Graduate School of Medicine, Ehime, Japan. She has a DDS,
PhD and an epidemiologist. She is a co-principal investigator
of several projects such as the Kyushu Okinawa Maternal and
8 Y. MIYAKE ET AL.
Child Health Study, Aidai Cohort Study, and Japan Ulcerative
Colitis Study.
Hitomi Okubo has a PhD and a nutritional epidemiologist.
She has an interest in maternal and child health.
Satoshi Sasaki is a Professor in the Department of Social and
Preventive Epidemiology, School of Public Health, The Uni-
versity of Tokyo, Tokyo, Japan. He has a MD, PhD and a
nutritional epidemiologist. He developed the Diet History
Questionnaire.
Akiko Tokinobu has a Bachelor of Ayurvedic Medicine and
Surgery and PhD and is a Research Associate in the Depart-
ment of Epidemiology and Preventive Medicine, Ehime Uni-
versity Graduate School of Medicine, Ehime, Japan.
Masashi Arakawa is a Professor in the Wellness Research
Fields, Faculty of Global and Regional Studies, University of
the Ryukyus, Okinawa, Japan and has a PhD. He has an inter-
est in health tourism.
ORCID
Yoshihiro Miyake http://orcid.org/0000-0003-1244-4488
Keiko Tanaka http://orcid.org/0000-0002-3748-5799
References
[1] Gernand AD, Schulze KJ, Stewart CP, West Jr KP,
Christian P. Micronutrient deficiencies in pregnancy
worldwide: health effects and prevention. Nat Rev
Endocrinol. 2016;12(5):274–89.
[2] Lai JS, Cai S, Feng L, Shek LP, Yap F, Tan KH, et al.
Associations of maternal zinc and magnesium with
offspring learning abilities and cognitive development at
4 years in GUSTO. Nutr Neurosci. 2019 Jul 22: 1–10.
doi:10.1080/1028415X.2019.1643624. Epub ahead of print.
[3] Effatpanah M, Rezaei M, Effatpanah H, Effatpanah Z,
Varkaneh HK, Mousavi SM, et al. Magnesium status
and attention deficit hyperactivity disorder (ADHD): a
meta-analysis. Psychiatry Res. 2019;274:228–34.
[4] Robberecht H, Verlaet AAJ, Breynaert A, De Bruyne T,
Hermans N. Magnesium, iron, zinc, copper and sel-
enium status in attention-deficit/hyperactivity disorder
(ADHD). Molecules. 2020;25(19):4440.
[5] Black LJ, Allen KL, Jacoby P, Trapp GS, Gallagher CM,
Byrne SM, et al. Low dietary intake of magnesium is
associated with increased externalising behaviours in
adolescents. Public Health Nutr. 2015;18(10):1824–30.
[6] Durá-Travé T, Gallinas-Victoriano F. Caloric and nutri-
ent intake in children with attention deficit hyperactivity
disorder treated with extended-release methylphenidate:
analysis of a cross-sectional nutrition survey. JRSM
Open. 2014;5(2):204253331351769.
[7] Miyake Y, Tanaka K, Okubo H, Sasaki S, Arakawa M.
Fish and fat intake and prevalence of depressive symp-
toms during pregnancy in Japan: baseline data from
the Kyushu Okinawa maternal and child health study.
J Psychiatr Res. 2013;47:572–8.
[8] Miyake Y, Tanaka K, Okubo H, Sasaki S, Arakawa M.
Maternal fat intake during pregnancy and behavioral
problems in 5-y-old Japanese children. Nutrition.
2018;50:91–6.
[9] Miyake Y, Tanaka K, Okubo H, Sasaki S, Arakawa M.
Maternal caffeine intake in pregnancy is inversely
related to childhood peer problems in Japan: the
Kyushu Okinawa maternal and child health study.
Nutr Neurosci. 2019;22:817–24.
[10] Miyake Y, Tanaka K, Okubo H, Sasaki S, Arakawa M.
Maternal B vitamin intake during pregnancy and child-
hood behavioral problems in Japan: the Kyushu
Okinawa maternal and child Health study. Nutr
Neurosci. 2018 Nov 19: 1–8. doi:10.1080/1028415X.
2018.1548139. [Epub ahead of print].
[11] Takahashi K, Tanaka K, Nakamura Y, Okubo H, Sasaki
S, Arakawa M, Miyake Y. Calcium intake during preg-
nancy is associated with decreased risk of emotional
and hyperactivity problems in five-year-old Japanese
children. Nutr Neurosci. 2019: 1–8. doi:10.1080/
1028415X.2019.1676971. [Epub ahead of print].
[12] Miyake Y, Tanaka K, Okubo H, Sasaki S, Arakawa M.
Maternal consumption of vegetables, fruit, and antioxi-
dants during pregnancy and risk of childhood behav-
ioral problems. Nutrition. 2020;69:110572.
[13] Shima S, Shikano T, Kitamura T, Asai M. New self-rated
scale for depression (in Japanese). Jpn J Clin Psychiatry.
1985;27:717–23.
[14] Radloff LS. The CES-D scale: a self-report depression
scale for research in the general population. Appl
Psychol Meas. 1977;1:385–401.
[15] Sasaki S, Yanagibori R, Amano K. Self-administered diet
history questionnaire developed for health education: a
relative validation of the test-version by comparison
with 3-day diet record in women. J Epidemiol.
1998;8:203–15.
[16] Sasaki S, Ushio F, Amano K, Morihara M, Todoriki T,
Uehara Y, et al. Serum biomarker-based validation of
a self-administered diet history questionnaire for
Japanese subjects. J Nutr Sci Vitaminol (Tokyo).
2000;46:285–96.
[17] Kobayashi S, Honda S, Murakami K, Sasaki S, Okubo H,
Hirota N, et al. Both comprehensive and brief self-admi-
nistered diet history questionnaires satisfactorily rank
nutrient intakes in Japanese adults. J Epidemiol.
2012;22:151–9.
[18] Shiraishi M, Haruna M, Matsuzaki M, Murayama R,
Sasaki S, Murashima S. Validity and reproducibility of
folate and vitamin B12 intakes estimated from a self-
administered diet history questionnaire in Japanese
pregnant women. Nutr J. 2012;11:15.
[19] Shiraishi M, Haruna M, Matsuzaki M, Murayama R,
Yatsuki Y, Sasaki S. Estimation of eicosapentaenoic
acid and docosahexaenoic acid intakes in pregnant
Japanese women without nausea by using a self-admi-
nistered diet history questionnaire. Nutr Res.
2013;33:473–8.
[20] Shiraishi M, Haruna M, Matsuzaki M, Murayama R,
Sasaki S. Validity of a diet history questionnaire estimat-
ing β-carotene, vitamin C and α-tocopherol intakes in
Japanese pregnant women. Int J Food Sci Nutr.
2013;64:694–9.
[21] Shiraishi M, Haruna M, Matsuzaki M, Murayama R,
Kitanaka S, Sasaki S. Validity of a self-administered

diet history questionnaire for estimating vitamin D
intakes of Japanese pregnant women. Matern Child
Nutr. 2015;11:525–36.
[22] Science and Technology Agency. Standard tables of food
composition in Japan, fifth revised and enlarged edition.
(in Japanese). Tokyo: Printing Bureau of the Ministry of
Finance; 2005.
[23] Willett W, Stampfer MJ. Total energy intake: impli-
cations for epidemiologic analyses. Am J Epidemiol.
1986;124:17–27.
[24] Goodman R. The strengths and difficulties question-
naire: a research note. J Child Psychol Psychiatry.
1997;38:581–6.
[25] Matsuishi T, Nagano M, Araki Y, Tanaka Y, Iwasaki M,
Yamashita Y, et al. Scale properties of the Japanese
NUTRITIONAL NEUROSCIENCE 9
version of the strengths and difficulties questionnaire
(SDQ): a study of infant and school children in commu-
nity samples. Brain Dev. 2008;30:410–5.
[26] Gröber U, Schmidt J, Kisters K. Magnesium in preven-
tion and therapy. Nutrients. 2015;7(9):8199–226.
[27] Mousain-Bosc M, Roche M, Rapin J, Bali JP. Magnesium
VitB6 intake reduces central nervous system hyperexcit-
ability in children. J Am Coll Nutr. 2004;23(5):545S–8S.
[28] Statistics Bureau, Ministry of Public Management, Home
Affairs, Posts and Telecommunications. 2000 population
census of Japan, Vol. 3-2-40, labour force status of popu-
lation, industry (major groups) of employed persons, and
education, Fukuoka-ken. Tokyo: Statistics Bureau,
Ministry of Public Management, Home Affairs, Posts
and Telecommunications; 2002.
Copyright of Nutritional Neuroscience is the property of Taylor & Francis Ltd and its content
may not be copied or emailed to multiple sites or posted to a listserv without the copyright
holder's express written permission. However, users may print, download, or email articles for
individual use.