Cell phone use and behavioural problems in young children

Abstract

Potential health effects of cell phone use in children have not been adequately examined. As children are using cell phones at earlier ages, research among this group has been identified as the highest priority by both national and international organisations. The authors previously reported results from the Danish National Birth Cohort (DNBC), which looked at prenatal and postnatal exposure to cell phone use and behavioural problems at age 7 years. Exposure to cell phones prenatally, and to a lesser degree postnatally, was associated with more behavioural difficulties. The original analysis included nearly 13 000 children who reached age 7 years by November 2006.
To see if a larger, separate group of DNBC children would produce similar results after considering additional confounders, children of mothers who might better represent current users of cell phones were analysed. This 'new' dataset consisted of 28 745 children with completed Age-7 Questionnaires to December 2008.
The highest OR for behavioural problems were for children who had both prenatal and postnatal exposure to cell phones compared with children not exposed during either time period. The adjusted effect estimate was 1.5 (95% CI 1.4 to 1.7).
The findings of the previous publication were replicated in this separate group of participants demonstrating that cell phone use was associated with behavioural problems at age 7 years in children, and this association was not limited to early users of the technology. Although weaker in the new dataset, even with further control for an extended set of potential confounders, the associations remained.

Full text

Cell phone use and behavioural problems in
young children
Hozefa A Divan,
1
Leeka Kheifets,
2
Carsten Obel,
3
Jørn Olsen
2,3
ABSTRACT
Background Potential health effects of cell phone use in
children have not been adequately examined. As children
are using cell phones at earlier ages, research among
this group has been identified as the highest priority by
both national and international organisations. The authors
previously reported results from the Danish National Birth
Cohort (DNBC), which looked at prenatal and postnatal
exposure to cell phone use and behavioural problems at
age 7 years. Exposure to cell phones prenatally, and to
a lesser degree postnatally, was associated with more
behavioural difficulties. The original analysis included
nearly 13 000 children who reached age 7 years by
November 2006.
Methods To see if a larger, separate group of DNBC
children would produce similar results after considering
additional confounders, children of mothers who might
better represent current users of cell phones were
analysed. This ‘new’ dataset consisted of 28 745 children
with completed Age-7 Questionnaires to December 2008.
Results The highest OR for behavioural problems were
for children who had both prenatal and postnatal
exposure to cell phones compared with children not
exposed during either time period. The adjusted effect
estimate was 1.5 (95% CI 1.4 to 1.7).
Conclusions The findings of the previous publication
were replicated in this separate group of participants
demonstrating that cell phone use was associated with
behavioural problems at age 7 years in children, and this
association was not limited to early users of the
technology. Although weaker in the new dataset, even
with further control for an extended set of potential
confounders, the associations remained.
Technological developments now bring social and
economic benefits to large sections of society;
however, the health consequences of these devel-
opments can be difficult to predict. Sources of radio
frequency electromagnetic fields (EMF) have been
present as a result of radio and TV broadcasts since
the early 20th century. EMF from radar and related
technologies has been present since the mid-20th
century. EMF from cellular communications has
come along in the last quarter of the previous
century, and in just the past few years, sources of
EMF from Wi-Fi, RFID and other novel technolo-
gies have come into existence. All these sources
have increased considerably since first emerging,
and increases in cellular communications and other
radio frequency technologies in the past decade
have been particularly rapid. The worldwide
proportion subscribing to cell phones has increased
from 5% approximately 15 years ago to well over
70% who are current users.
1
If cell phones lead to
adverse health effects associated with their use,
then the potential rise in health burden could be
significant.
Exposure to cell phones is increasingly becoming
prevalent among children at younger ages. Also of
concern is use by expectant mothers. Previously, we
reported an association from the Danish National
Birth Cohort (DNBC) for prenatal and postnatal
exposure to cell phone use and behavioural prob-
lems at age 7 years among nearly 13 000 Danish
children born between 1997 and 1999.
2
In order to determine whether our original
results were a chance finding, or because the initial
sample were ‘early adopters of technology’who are
more likely to have behavioural problems and
whose children are likely to have similar behav-
ioural patterns, we analysed a new and separate
group of mothers and children from the DNBC.
Compared with our previous efforts we further
adjusted for an extended set of potential
confounders, including variables that reflect
mother’s attention towards the child early in life.
METHODS
From March 1996 to November 2002, the DNBC
recruited nearly 100 000 pregnant mothers with the
intent to follow these women and their offspring
longitudinally in a life-course perspective.
3 4
Mothers reported detailed information on lifestyle
factors, dietary habits and environmental exposures
collected by means of four telephone interviewsd
two during pregnancy and two within 18 months
postpartum.
5
With the resources of the various
administrative, health and socioeconomic registers,
information was linkable for cohort participants by
means of a ‘central person register’number that is
assigned to all Danish persons.
6
In this analysis,
information from the Danish Medical Birth
Registry was linked with DNBC data.
7
When offspring reached 7 years of age, a new
questionnaire was administered to mothers on
information pertaining to the health of her child.
Questions on cell phone use among children, as
well as among mothers during pregnancy, were
asked. More detailed prenatal cell phone use infor-
mation included: historical use of cell phone by
mother (year of first regular use, amount of use
during pregnancy); use of hands-free equipment by
mother (proportion of time); use of hands-free
equipment during pregnancy and location of the
phone (handbag or pant/shirt/jacket pocket) and
current use of cell phones.
The Age-7 Questionnaire also included data on
social conditions, family lifestyle and diseases in
childhood, including behavioural problems as
defined by the strengths and difficulties question-
naire (SDQ).
8 9
Mothers completed a list of 25
<Additional tables are
published online only. To view
these files please visit the
journal online (http://jech.bmj.
com).
1
Division of Biostatistics,
Department of Preventive
Medicine, Keck School of
Medicine of the University of
Southern California, Los
Angeles, California, USA
2
Department of Epidemiology,
School of Public Health,
University of California, Los
Angeles, California, USA
3
Institute of Public Health,
University of Aarhus, Aarhus,
Denmark
Correspondence to
Dr Leeka Kheifets, UCLA Pub
Hlth-Epid, Box 951772, 73-320
CHS, Los Angeles, CA
90095-1772, USA;
kheifets@ucla.edu
Accepted 22 September 2010
Divan HA, Kheifets L, Obel C, et al.J Epidemiol Community Health (2010). doi:10.1136/jech.2010.115402 1 of 6
Research report
JECH Online First, published on December 7, 2010 as 10.1136/jech.2010.115402
Copyright Article author (or their employer) 2010. Produced by BMJ Publishing Group Ltd under licence.

questions with scaled responses (very true, partly true, or not
true) regarding their child’s behaviour. Scores were summed over
a particular group of questions assessing for overall and specific
behavioural problems or disorders with a priori defined cut-off
points. Based on the score, children were classified as normal
(0e13), borderline (14e16), or abnormal (17e40) for having
‘overall behavioural problems’.
Our analysis included comparisons between covariates
(potential confounders) and prenatal and postnatal cell phone
exposure. Covariates of interest included: child’s gender;
mother’s age at birth; father’s age at birth; mother’s history of
psychiatric problems (self-reported from Age-7 Questionnaire);
mother’s history of psychiatric, behavioural, or cognitive prob-
lems as child (self-reported from prenatal interviews); father’s
history of psychiatric, behavioural, or cognitive problems as
child (spousal report from prenatal interviews); social occupa-
tional status; prenatal smoking (entire pregnancy, early preg-
nancy, or not a smoker); prenatal alcohol (entire, early, or late
pregnancy only, or not at all) and prenatal marijuana use (yes or
no); prenatal stress (14-point summary score categorised as low
(0e4), medium (5), high (6e14)); prenatal physical activity
(entire, early, or late pregnancy, or no activity); other sources of
prenatal ionising and non-ionising radiation (ie, x-rays, ultra-
sound); parity; gestational age; birth weight; postpartum stress
(15-point summary score categorised as low (0e3), medium
(4), high (5e15)); child breastfed for at least the first 6 months
(yes or no); hours spent with child daily by age 6 and
18 months; and child in daycare by 18 months.
An ordinal logistic regression model was used to estimate the
odds of the overall behavioural problems (0, normal; 1, border-
line; 2, abnormal) according to prenatal and postnatal exposure
to cell phones. Regression models were adjusted for covariates
from the original analysis such as child’s gender, mother’s age at
birth, mother’s social occupational status, prenatal smoking,
and mother’s history of psychiatric problems. Regression
modelling also considered covariates not included in the previous
publication such as both parents’history of psychiatric, cogni-
tive, or behavioural problems as a child, a combined social
occupational status, prenatal alcohol and drug use, prenatal
physical activity, other prenatal radiation sources, father’s age at
birth, gestational age, parity, birth weight, postpartum stress,
breast feeding, hours spent daily by ages 6 and 18 months, and
child in daycare by 18 months. Certain covariates (risk factors)
were not associated with the outcome of interest in this analysis
and were eliminated. Also certain covariates were not associated
with exposure and were found not to be statistically significant
(p >0.05) in a multivariate model that included the exposure
and outcome. The log likelihood ratio test was utilised during
the model building process to develop a parsimonious model by
manually eliminating variables at a p value greater than 0.05.
Proxies of prenatal exposure intensity (times per day spoken,
location of the phone when not used, proportion of time the
phone was turned on, and use of an earpiece with cell phone)
were used to evaluate possible doseeresponse patterns.
Depending on the characteristic, the reference category was
defined as the lowest possible category (ie, no use, 0e1 times per
day spoken). For location of phone when not in use, the refer-
ence category was ‘carried in bag’versus ‘carried in dress/pant
pocket’.
Previously, we reported data for 13 159 Danish children born
between 1997 and 1999.
2
In this analysis, a ‘new’and separate
dataset of Danish children (born 1998e2002) was utilised. For
comparison, results from 12 796 of the ‘original’children were
included after excluding 363 ‘original’children who were born as
part of a set of twins or triplets. This analysis for comparability
purposes includes singleton, live births in both datasets: ‘original’
and ‘new’, and concludes with an analysis of the ‘combined’
datasets.
Human ethics review approvals were obtained from the
Danish Data Protection Agency (Datatilsynet) and the Univer-
sity of California, Los Angeles (UCLA) Office for the Protection
of Research Subjects.
RESULTS
Results are presented for the ‘original’(n¼12 796) and ‘new’
(n¼28 745) datasets for all DNBC singleton, live births followed
up to age 7 years. In both datasets 30.5% (original) and 35.2%
(new) of children, were using a cell phone at the age of 7 years,
but less than 1% used a cell phone for more than 1 h per week in
both datasets. In the original dataset, 10.1% of children had both
prenatal and postnatal (joint) exposure, whereas among the new
dataset of children, 17.9% were jointly exposed. Respectively,
53.3% and 39.5% of children in the original and new datasets
Table 1 Association of prenatal and postnatal exposure to cell phone
use with overall behavioural problems by type of dataset and by birth
year
Original dataset New dataset
OR aORy95% CI OR aORy95% CI
All birth years*
n 12 796z28 745x
Prenatal and postnatal exposure 2.2 1.9 1.5 to 2.3 1.9 1.5 1.3 to 1.7
Prenatal exposure only 1.7 1.5 1.3 to 1.9 1.5 1.3 1.1 to 1.5
Postnatal exposure only 1.2 1.2 1.0 to 1.4 1.2 1.2 1.0 to 1.4
No exposure 1.0 1.0 e1.0 1.0 e
1998
n 5685z1090x
Prenatal and postnatal exposure 2.4 2.0 1.5 to 2.8 3.4 3.4 1.5 to 7.9
Prenatal exposure only 1.4 1.2 0.8 to 1.7 1.7 1.5 0.6 to 3.8
Postnatal exposure only 1.2 1.1 0.8 to 1.4 2.0 2.1 1.0 to 4.4
No exposure 1.0 1.0 e1.0 1.0 e
1999
n 7076z4214x
Prenatal and postnatal exposure 2.1 1.8 1.3 to 2.4 2.2 1.9 1.4 to 2.7
Prenatal exposure only 2.0 1.8 1.3 to 2.4 1.5 1.4 1.0 to 1.9
Postnatal exposure only 1.3 1.2 1.0 to 1.6 1.3 1.2 0.9 to 1.7
No exposure 1.0 1.0 e1.0 1.0 e
2000
ne13 115x
Prenatal and postnatal exposure ee e 1.9 1.4 1.1 to 1.7
Prenatal exposure only ee e 1.5 1.3 1.0 to 1.6
Postnatal exposure only ee e 1.2 1.2 0.9 to 1.5
No exposure ee e 1.0 1.0 e
2001
ne9682x
Prenatal and postnatal exposure ee e 1.8 1.4 1.1 to 1.8
Prenatal exposure only ee e 1.5 1.3 1.1 to 1.7
Postnatal exposure only ee e 1.0 1.0 0.7 to 1.4
No exposure ee e 1.0 1.0 e
*Includes the years 1997 (n¼24) and/or 2002 (n¼635).
yAdjusted for sex of child, mother’s age at birth, mother’s socio-occupational status,
smoking during pregnancy, and mother’s psychiatric history.
zSingleton, live births; our previous analysis included 13 159 children of singleton and
multiple, live births; 363 children of multiple births were not included in this analysis.
xSingleton, live births.
aOR, adjusted odds ratio.
2 of 6 Divan HA, Kheifets L, Obel C, et al.J Epidemiol Community Health (2010). doi:10.1136/jech.2010.115402
Research report

had neither prenatal nor postnatal cell phone use exposure. Tests
for trend indicated that patterns of cell phone use did change
with birth year for both datasets.
Regarding overall behavioural problems, 93.5% (original) and
93.0% (new) of children had no recorded behavioural problems.
In both 3.3% were considered borderline and 2.9% (original) and
3.1% (new) of children scored as abnormal.
In table 1, the joint exposures were positively associated with
overall behavioural problems. These estimates for both datasets
were adjusted for the original set of covariates as previously
published.
2
The highest OR for behavioural problems were
observed for children who had a joint exposure compared
with no exposure. Adjusting for potential confounders moved
the results towards the null. In the new dataset, the joint
exposure association with overall behavioural problems by
birth year decreased from 1998 to 2001 yet remained incom-
patible with the null. Unadjusted and adjusted models with an
interaction term (cell phone use and birth year) were tested
and found not to be statistically significant at the p¼0.05 level
(not shown).
Figure 1 shows the association between certain adjusted cova-
riates and the joint exposure to cell phone use for both datasets.
Children with prenatal and postnatal exposures were more often
in the lower social occupational status, to have mothers who
smoked during pregnancy, to have younger mothers and to
have mothers with higher prenatal stress scores. The percentage
of parents for whom childhood history of psychiatric, cognitive,
or behavioural problems was unknown was higher among the
‘original’dataset (early adopters of cell phone technology).
Supplementary tables A1 and A2 (available online only)
present all of the covariates considered in this analysis by levels
of exposure (no cell phone exposure, prenatal exposure only,
postnatal exposure only, or joint exposure) for both datasets.
Worth noting are the greater number children whose mothers
Figure 1 Prenatal and postnatal (joint)
exposure to cell phone use by type of
dataset and percentage distributions for
selected covariates: parents’ combined
social ccupational status (A); mother’s
history of psychiatric, cognitive or
behavioural problems as a child (B);
father’s history of psychiatric, cognitive
or behavioural problems as a child (C);
mother’s smoking status during
pregnancy (D); mother’s age at child’s
birth (E); and mother’s stress score
during pregnancy (F).
Divan HA, Kheifets L, Obel C, et al.J Epidemiol Community Health (2010). doi:10.1136/jech.2010.115402 3 of 6
Research report

reported ‘not smoking’during pregnancy with no cell phone
exposure or with prenatal exposure only compared with children
with postnatal only or joint exposure in both datasets. Another
difference is the greater percentage of children at birth with
mothers 15e24 years of age with joint exposure compared with
prenatal only, postnatal only, or no exposure in both datasets.
Table 2 presents estimates for the ‘combined’dataset (‘orig-
inal’and ‘new’,n¼41 541). The adjusted OR in the ‘combined’
dataset for overall behavioural problems score was 1.6 for the
joint exposure. Upon further adjustment, the OR for prenatal
and postnatal exposure was 1.5 (95% CI 1.4 to 1.7). This final
model was adjusted for sex of child, mother’s age at birth,
mother’s and father’s history of psychiatric, cognitive or
behavioural problems as a child, combined socio-occupational
status, gestational age, mother’s prenatal stress, and child
breastfed up to 6 months of age. For prenatal or postnatal
exposure only, the adjusted OR were 1.4 (95% CI 1.2 to 1.5) and
1.2 (95% CI 1.0 to 1.3), respectively.
When analyses were stratified by the modelled covariates, the
associations between cell phone use and overall behavioural
problems remained across the strata (table 3). These results
demonstrate that the selected covariates confound the associa-
tion between cell phone use and behavioural problems if not
controlled, yet complete confounder control is unlikely due to
residual confounding caused by measurement error for these
covariates. For nearly all strata of covariates, the highest OR
were for those with the joint exposure.
To estimate mother ’s inattention we looked at variables such
as breastfeeding up to 6 months of age, reported number of
hours spent with child at ages 6 and 18 months, and whether
child was in regular daycare by 18 months of age. In table 4, the
only covariates that were associated with overall behavioural
problems were breastfeeding up to 6 months of age and
spending less time daily with the child at 6 months.
In the combined dataset, considering prenatal cell phone use
characteristics (independent of postnatal use by child), almost
85% of mothers carried their cell phone in a bag during preg-
nancy rather than on their person or elsewhere, and nearly 80%
reported not using an earpiece (not shown). In table 5, more
than 10% of children with prenatal exposure had mothers who
reported speaking four times per day or more during their
pregnancy and 48.5% reported having the phone turned on at all
times. For prenatal exposures, regardless of control for postnatal
Table 2 Association of prenatal and postnatal exposure to cell phone
use with overall behavioural problems by birth year among all children
(combined dataset, n¼41 541)
OR aORy95% CI ORz95% CI
All birth years* (n¼41 541)
Prenatal and postnatal exposure 2.0 1.6 1.4 to 1.8 1.5 1.4 to 1.7
Prenatal exposure only 1.5 1.4 1.2 to 1.5 1.4 1.2 to 1.5
Postnatal exposure only 1.2 1.2 1.0 to 1.3 1.2 1.0 to 1.3
No exposure 1.0 1.0 e1.0 e
1998 (n¼6775)
Prenatal and postnatal exposure 2.5 2.2 1.6 to 3.0 2.2 1.7 to 3.0
Prenatal exposure only 1.4 1.2 0.9 to 1.7 1.3 1.0 to 1.8
Postnatal exposure only 1.3 1.2 0.9 to 1.6 1.3 1.0 to 1.7
No exposure 1.0 1.0 e1.0 e
1999 (n¼11 290)
Prenatal and postnatal exposure 2.2 1.9 1.5 to 2.3 1.8 1.5 to 2.3
Prenatal exposure only 1.8 1.6 1.3 to 2.0 1.5 1.3 to 1.9
Postnatal exposure only 1.3 1.2 1.0 to 1.5 1.2 1.0 to 1.5
No exposure 1.0 1.0 e1.0 e
2000 (n¼13 115)
Prenatal and postnatal exposure 1.9 1.4 1.1 to 1.7 1.3 1.1 to 1.6
Prenatal exposure only 1.5 1.3 1.0 to 1.6 1.2 1.0 to 1.5
Postnatal exposure only 1.2 1.2 0.9 to 1.5 1.2 1.0 to 1.5
No exposure 1.0 1.0 e1.0 e
2001 (n¼9682)
Prenatal and postnatal exposure 1.8 1.4 1.1 to 1.8 1.4 1.1 to 1.7
Prenatal exposure only 1.5 1.4 1.1 to 1.7 1.4 1.1 to 1.7
Postnatal exposure only 1.0 1.0 0.7 to 1.4 1.0 0.8 to 1.4
No exposure 1.0 1.0 e1.0 e
*Includes the years 1997 (n¼24) and/or 2002 (n¼635).
yAdjusted for sex of child, mother’s age at birth, mother’s socio-occupational status,
smoking during pregnancy, and mother’s psychiatric history.
zAdjusted for sex of child, mother’s age at birth, mother’s and father’s history of
psychiatric, cognitive or behavioural problems as a child, combined socio-occupational
status, gestational age, mother’s prenatal stress, and child breastfed up to 6 months of age.
aOR, adjusted odds ratio.
Table 3 Association of overall behavioural problems with prenatal and
postnatal exposure to cell phone use stratified by covariates
Prenatal and
postnatal exposure
Prenatal
exposure only
Postnatal
exposure only
OR (95% CI) OR (95% CI) OR (95% CI)
Social occupational status (combined)
High level
(n¼27170)
1.6 (1.4 to 1.9) 1.2 (1.0 to 1.4) 1.1 (0.9 to 1.3)
Medium level
(n¼11 185)
2.1 (1.7 to 2.5) 1.9 (1.6 to 2.3) 1.4 (1.2 to 1.7)
Low level (n¼1374) 1.8 (1.2 to 2.6) 2.2 (1.5 to 3.4) 1.2 (0.7 to 1.9)
Sex of child
Boy (n¼21 284) 2.1 (1.9 to 2.4) 1.6 (1.4 to 1.8) 1.3 (1.1 to 1.5)
Girl (n¼20 237) 1.9 (1.6 to 2.3) 1.4 (1.1 to 1.6) 1.3 (1.1 to 1.6)
Mother’s history of psychiatric, cognitive, or behavioural problems as a child
Yes (n¼4579) 2.4 (1.9 to 3.1) 1.7 (1.3 to 2.2) 1.2 (0.8 to 1.5)
No (n¼28 411) 1.7 (1.4 to 1.9) 1.5 (1.3 to 1.7) 1.1 (1.0 to 1.3)
Father’s history of psychiatric, cognitive, or behavioural problems as a child
Yes (n¼3378) 2.2 (1.6 to 2.9) 2.1 (1.5 to 2.9) 1.0 (0.7 to 1.5)
No (n¼29 034) 1.8 (1.6 to 2.0) 1.4 (1.2 to 1.6) 1.2 (1.0 to 1.3)
Mother’s age at child’s birth (years)
15e24 (n¼3453) 1.8 (1.4 to 2.3) 1.6 (1.2 to 2.1) 0.8 (0.6 to 1.2)
25e29 (n¼15 868) 1.7 (1.4 to 2.0) 1.5 (1.2 to 1.7) 1.2 (1.0 to 1.4)
30e34 (n¼15 904) 1.7 (1.4 to 2.1) 1.3 (1.1 to 1.7) 1.2 (1.0 to 1.5)
35e39 (n¼5625) 1.7 (1.2 to 2.4) 1.2 (0.8 to 1.7) 1.6 (1.2 to 2.3)
40 or older (n¼691) 1.2 (0.3 to 4.5) 2.6 (1.1 to 6.4) 1.4 (0.5 to 4.2)
Gestational age at birth (weeks)
<37 (n¼1979) 2.1 (1.4 to 3.1) 2.1 (1.4 to 3.2) 1.2 (0.8 to 1.9)
37e41 (n¼35 686) 1.9 (1.7 to 2.1) 1.5 (1.3 to 1.7) 1.0 (0.5 to 1.7)
42 or greater
(n¼3769)
2.2 (1.6 to 3.1) 1.4 (1.0 to 2.0) 1.2 (1.1 to 1.4)
Mother’s stress score during pregnancy
Low (0e4)
(n¼36 085)
1.8 (1.6 to 2.0) 1.4 (1.3 to 1.6) 1.2 (1.1 to 1.4)
Medium (5)
(n¼1430)
2.2 (1.4 to 3.4) 1.6 (1.0 to 2.6) 1.0 (0.5 to 1.7)
High (6e14)
(n¼1693)
2.9 (2.0 to 4.2) 2.5 (1.6 to 3.8) 1.2 (0.8 to 1.9)
Child breastfed for at least the first 6 months
Yes (n¼25 066) 1.7 (1.5 to 2.0) 1.5 (1.3 to 1.7) 1.3 (1.1 to 1.5)
No (n¼7629) 1.8 (1.4 to 2.1) 1.4 (1.1 to 1.7) 0.9 (0.7 to 1.1)
4 of 6 Divan HA, Kheifets L, Obel C, et al.J Epidemiol Community Health (2010). doi:10.1136/jech.2010.115402
Research report

exposure, adjusted OR for the overall behavioural problems score
tended to be greater with higher potential for fetal exposure.
Proxies for intensity of mother’s phone use during pregnancy
did exhibit doseeresponse associations, and tests for trend were
statistically significant.
DISCUSSION
Using a new group of participants from the DNBC, we repli-
cated our previously reported study on prenatal and postnatal
(joint) exposure to cell phones. Our results make it unlikely that
the first finding was by chance, but our estimate was higher in
the ‘original’dataset (adjusted OR 1.9) compared with the ‘new’
dataset (adjusted OR 1.5).
Many including ourselves have raised concerns regarding the
role of uncontrolled confounding as well as unmeasured
confounding in the original analysis.
10
Here, we examined
numerous other covariates that were not considered previously.
With the addition of these variables, the association still
remained. Although we took a larger set of potential
confounders into consideration there was no appreciable effect
on the results.
We also hypothesised that greater cell phone use during
pregnancy may be indicative of mother’s inattention in rearing
her child, thus providing an alternative explanation for the
positive association with behavioural problems in children. As
this study was not designed to observe direct motherechild
interactions or how much attention a mother gave her child, we
used measures of breastfeeding and hours spent per day as proxy
measures for this covariate. Breastfeeding was inversely associ-
ated (OR 0.5 in the combined dataset) with overall behavioural
problems but did not diminish the association between cell
phone exposure and the outcome when included. If breast-
feeding and time spent with child are good measures of mothers’
attention then we believe that our results do not support
inattention as a likely explanation for the observed association.
It has been suggested that our initial results were due to
characteristics of early technology adopters of cell phones and
that these parents’behaviour may strongly influence and predict
overall behavioural problems in their children. These findings are
not limited to a unique group of parents in the early part of our
cohort, but are replicated in a more general population of Danish
mothers who used cell phones during pregnancy.
There were concerns that the SDQ as an instrument might be
too non-specific and biased if mothers have children with other
serious mental and health conditions before SDQ administra-
tion. However, our work and the work of others indicates both
the internal validity of SDQ and its ability to predict clinical
diagnosis for overall behavioural problems.
8911
We also do not believe that differential recall bias explains the
observed associations. We have tested this exposure assessment
method with other outcomes and did not find an association
(data not shown). It is highly unlikely that reporting prenatal or
postnatal cell phone use would be influenced by the mother’s
knowledge or suspicion of her child’s behavioural status and not
by more debilitating neurological outcomes such child’s history
of febrile seizures or epilepsy, which we looked at.
Modelling specific absorption rates (SAR) of radiofrequency
fields to the womb of pregnant mothers suggest that exposures
are likely to be low and not high enough to elevate the body
temperature,
12e14
but modelling is based on numerous
assumptions and extrapolations. In addition, possible non-
thermal effects of radiofrequency fields remain of interest. In
a recent letter to the editor, Hocking
15
cites a review article by
Brzezinski
16
that suggests talking on a cell phonedplaced on the
side of the head by the ear and jawdmay lead to increased
melatonin secretion due to the excitation of nearby post-
ganglionic nerves that lead to the pineal gland, which is
responsible for producing melatonin. One of the many things
that this hormone does is to inhibit the secretion of gonado-
tropin-releasing hormone thus directly affecting steroid metab-
olism within the ovaries and progesterone synthesis. It is
believed that diverse changes in maternal metabolism or the sex
hormone environment can affect the development of the fetal
brain thus leading to behavioural problems.
15
Vrijheid et al
17
recently published results reporting no asso-
ciation between prenatal exposure to cell phone use and
neurodevelopment at 14 months among a smaller pregnancy
Table 4 Association of proxy covariates for mother’s inattention with
overall behavioural problems in children
Original dataset
(n[12 796)
New dataset
(n[28 745)
Combined
dataset
(n[41 541)
OR 95% CI OR 95% CI OR 95% CI
Child breastfed for at least the first 6 months
Yes 0.6 0.5 to 0.7 0.5 0.5 to 0.6 0.5 0.5 to 0.6
No 1.0 e1.0 e1.0 e
Reported amount of hours spent per day with child at age 6 months interview
<1 1.5 1.1 to 2.0 1.5 1.2 to 1.9 1.5 1.2 to 1.8
1e7 1.1 0.9 to 1.4 1.3 1.1 to 1.5 1.2 1.1 to 1.4
8 or more 1.0 e1.0 e1.0 e
Reported amount of hours spent per day with child at age 18 months interview
<1 1.1 0.9 to 1.4 0.9 0.8 to 1.1 1.0 0.9 to 1.1
1e4 1.1 0.8 to 1.5 1.2 1.0 to 1.4 1.2 1.0 to 1.3
5 1.0 0.8 to 1.3 0.9 0.8 to 1.1 1.0 0.8 to 1.1
6e7 1.0 0.8 to 1.3 0.8 0.6 to 0.9 0.8 0.7 to 1.0
8 or more 1.0 e1.0 e1.0 e
Child in regular daycare outside the home at age 18 months interview
Yes 1.1 0.8 to 1.4 0.9 0.8 to 1.1 1.0 0.8 to 1.1
No 1.0 e1.0 e1.0 e
Table 5 Association of characteristics of mother’s cell phone use
during pregnancy with overall behavioural problems score in children
with prenatal exposure (n¼13 938)
No. (%) OR aOR* (95% CI) aORy(95% CI)
Times spoken daily
0e1 7268 (52.2) 1.0 1.0 1.0
2e3 3703 (26.6) 1.4 1.2 (1.0 to 1.4) 1.2 (1.0 to 1.4)
4+ 1409 (10.8) 1.7 1.4 (1.2 to 1.7) 1.4 (1.2 to 1.7)
Missing 1458 (10.4) ee e
p for trend e0.09 0.07 0.07
Percentage of time turned on (%)
0 1098 (7.9) 1.0 1.0 1.0
<50 1788 (12.8) 1.6 1.4 (1.0 to 2.1) 1.4 (1.0 to 2.1)
50e99 4201 (30.1) 2.2 1.7 (1.2 to 2.3) 1.7 (1.2 to 2.3)
100 6750 (48.5) 2.8 2.0 (1.4 to 2.7) 2.0 (1.4 to 2.7)
Missing 101 (0.7) ee e
p for trend e<0.0001 0.003 0.004
*Adjusted for sex of child, mother’s age at birth, mother’s and father’s history of
psychiatric, cognitive or behavioural problems as a child, combined socio-occupational
status, gestational age, mother’s prenatal stress and child breastfed up to 6 months of age.
yAdjusted for sex of child, mother’s age at birth, mother’s and father’s history of
psychiatric, cognitive or behavioral problems as a child, combined socio-occupational
status, gestational age, mother’s prenatal stress, child breastfed up to 6 months of age and
postnatal exposure to cell phones.
aOR, adjusted odds ratio.
Divan HA, Kheifets L, Obel C, et al.J Epidemiol Community Health (2010). doi:10.1136/jech.2010.115402 5 of 6
Research report

cohort. Their findings point to the possibility that exposure may
have specificity for a particular outcome such as behavioural
problems, which probably has a different causal pathway than
infant neurodevelopment delays.
Whereas it is unlikely that mothers would erroneously recall
using or not using a cell phone, more detailed information such
as trimester of use was difficult to recall. We assume that
reported use correlates with levels of radiofrequency field expo-
sure, which are truly unknown, and prenatal exposure is
dichotomised, whereas the true exposure is a continuous value.
Data from the Age-7 Questionnaire represents nearly 60e65%
of mothers and children eligible to participate. This is down
from 80% participation for the 6 and 18 month interviews. In
this research a proportional odds model for an ordinal logistic
regression was utilised to understand behavioural problems. If
the proportional odds assumption was truly incorrect, then
model misspecification bias would have been introduced. This
can be explored further through multinomial logit analyses in
which such an assumption is not necessary.
Although it is premature to interpret these results as causal,
we are concerned that early exposure to cell phones could carry
a risk, which, if real, would be of public health concern given the
widespread use of this technology. Even with limited scientific
investigations into this research hypothesis, given that expo-
sures to children and fetuses are easily reduced at virtually no
cost, precautionary measures might be warranted. It is our hope
that other scientists will attempt to replicate or refute the
findings of our research based upon similar study designs. Also,
prospective and detailed ascertainment would greatly improve
exposure measurement quality. A random subsample, who are
offered clinical evaluation for behavioural problems, would be
another enhancement. Adequate populations of both exposed
and unexposed are needed, but as cell phone technology is
widely used, researchers will find it difficult to enrol these
shrinking, unexposed populations.
Acknowledgements The authors would like to thank the coordinator of the data
collection, Inge Kristine Meder, data analysts, Inge Eisensee and Lone Fredslund
Møller, and the participating mothers.
Funding The Age-7 Questionnaire was financially supported by the Lundbeck
Foundation (195/04) and the Danish Medical Research Council (SSVF 0646). Support
for this work was provided by the UCLA School of Public Health, research innovation
seed grant (4565963LK19914) and by an NIH/NIEHS grant (R21ES016831).
Competing interests None.
Ethics approval This study was conducted with the approval of the Danish Data
Protection Agency (Datatilsynet) and the UCLA Office for the Protection of Research
Subjects.
Provenance and peer review Not commissioned; externally peer reviewed.
REFERENCES
1. International Telecommunication Union. Measuring the information society: the
ICT development index. 2009. http://www.itu.int/ITU-D/ict/publications/idi/2009/
index.html (accessed 30 Sep 2009).
2. Divan HA, Kheifets L, Obel C, et al. Prenatal and postnatal exposure to cell phone
use and behavioral problems in children. Epidemiology 2008;19:523e9.
3. Olsen J, Melbye M, Olsen SF, et al. The Danish National Birth Cohortdits
background, structure and aim. Scand J Public Health 2001;29:300e7.
4. Bech BH, Nohr EA, Vaeth M, et al. Coffee and fetal death: a cohort study with
prospective data. Am J Epidemiol 2005;162:983e90.
5. Danish National Birth Cohort. Data collection instruments. http://www.ssi.dk/
sw9314.asp (accessed 12 Mar 2008).
6. Pedersen CB, Gøtzsche H, Møller JO, et al. The Danish Civil Registration System.
A cohort of eight million persons. Dan Med Bull 2006;53:441e9.
7. Knudsen LB, Olsen J. The Danish Medical Birth Registry. Dan Med Bull
1998;45:320e3.
8. Goodman R. The strengths and difficulties questionnaire: a research note. J Child
Psychol Psychiatry 1997;38:581e6.
9. Goodman R, Ford T, Simmons H, et al. Using the strengths and difficulties
questionnaire (SDQ) to screen for child psychiatric disorders in a community sample.
Br J Psychiatry 2000;177:534e9.
10. Australian Centre for RF Bioeffects Research. FAQs and facts: position
statementdprenatal and postnatal exposure to cell phone use and behavioral
problems in children. 2008. http://www.acrbr.org.au/FAQ.aspx (accessed 12 Oct
2008).
11. Obel C, Linnet KM, Henriksen TB, et al. Smoking during pregnancy and
hyperactivityeinattention in the offspringdcomparing results from three Nordic
cohorts. Int J Epidemiol 2009;38:698e705.
12. Dimbylow P. SAR in the mother and foetus for RF plane wave irradiation. Phys Med
Biol 2007;52:3791e802.
13. Dimbylow PJ, Nagaoka T, Xu XG. A comparison of foetal SAR in three sets of
pregnant female models. Phys Med Biol 2009;54:2755e67.
14. Wiart J, Hadjem A, Wong MF, et al. Analysis of RF exposure in the head tissues of
children and adults. Phys Med Biol 2008;53:3681e95.
15. Hocking B. Maternal cell phone use and behavioral problems in children.
Epidemiology 2009;20:312.
16. Brzezinski A. Melatonin in humans. N Engl J Med 1997;16:186e95.
17. Vrijheid M, Martinez D, Forns J, et al. Prenatal exposure to cell phone use and
neurodevelopment at 14 months. Epidemiology 2010;21:1e4.
What is already known on this subject
Previous studies of cell phone use have emphasised health
effects in adults. Yet the most susceptible population to envi-
ronmental exposures are children. This past decade has seen
a great increase worldwide in cell phone use and access. During
this same period, an equally important public health outcome that
has increased in prevalence is childhood behavioural problems.
What this study adds
There is an association between prenatal as well as postnatal use
and behavioural problems by age 7 years among a general
population of mothers who are cell phone users. These results
replicate the findings of an association observed among only
early technology adopters. These new results also reduce the
likelihood that these are chance findings or findings that did not
adequately consider the influence of other important factors for
behavioural problems. These results should not be interpreted as
demonstrating a causal link between cell phone use and adverse
health effects for children, but if realdand given the nearly
universal use of cell phonesdthe impact on the publics’ health
could be of concern.
6 of 6 Divan HA, Kheifets L, Obel C, et al.J Epidemiol Community Health (2010). doi:10.1136/jech.2010.115402
Research report