Electromagnetic Hypersensitivity: A Systematic Review of Provocation Studies

Abstract

The objectives of this study were to assess whether people who report hypersensitivity to weak electromagnetic fields (EMFs) are better at detecting EMF under blind or double-blind conditions than nonhypersensitive individuals, and to test whether they respond to the presence of EMF with increased symptom reporting.
An extensive systematic search was used to identify relevant blind or double-blind provocation studies. This involved searching numerous literature databases and conference proceedings, and examining the citations of reviews and included studies. The results of relevant studies were tabulated and metaanalyses were used to compare the proportions of "hypersensitive" and control participants able to discriminate active from sham EMF exposures.
Thirty-one experiments testing 725 "electromagnetically hypersensitive" participants were identified. Twenty-four of these found no evidence to support the existence of a biophysical hypersensitivity, whereas 7 reported some supporting evidence. For 2 of these 7, the same research groups subsequently tried and failed to replicate their findings. In 3 more, the positive results appear to be statistical artefacts. The final 2 studies gave mutually incompatible results. Our metaanalyses found no evidence of an improved ability to detect EMF in "hypersensitive" participants.
The symptoms described by "electromagnetic hypersensitivity" sufferers can be severe and are sometimes disabling. However, it has proved difficult to show under blind conditions that exposure to EMF can trigger these symptoms. This suggests that "electromagnetic hypersensitivity" is unrelated to the presence of EMF, although more research into this phenomenon is required.

Full text

Electromagnetic Hypersensitivity: A Systematic Review of Provocation Studies
G. JAMES RUBIN,PHD, JAYATI DAS MUNSHI, MBBS, AND SIMON WESSELY,MD
Objectives: The objectives of this study were to assess whether people who report hypersensitivity to weak electromagnetic fields
(EMFs) are better at detecting EMF under blind or double-blind conditions than nonhypersensitive individuals, and to test whether
they respond to the presence of EMF with increased symptom reporting. Methods: An extensive systematic search was used to
identify relevant blind or double-blind provocation studies. This involved searching numerous literature databases and conference
proceedings, and examining the citations of reviews and included studies. The results of relevant studies were tabulated and
metaanalyses were used to compare the proportions of “hypersensitive” and control participants able to discriminate active from
sham EMF exposures. Results: Thirty-one experiments testing 725 “electromagnetically hypersensitive” participants were
identified. Twenty-four of these found no evidence to support the existence of a biophysical hypersensitivity, whereas 7 reported
some supporting evidence. For 2 of these 7, the same research groups subsequently tried and failed to replicate their findings. In
3 more, the positive results appear to be statistical artefacts. The final 2 studies gave mutually incompatible results. Our
metaanalyses found no evidence of an improved ability to detect EMF in “hypersensitive” participants. Conclusions: The
symptoms described by “electromagnetic hypersensitivity” sufferers can be severe and are sometimes disabling. However, it has
proved difficult to show under blind conditions that exposure to EMF can trigger these symptoms. This suggests that “electro-
magnetic hypersensitivity” is unrelated to the presence of EMF, although more research into this phenomenon is required. Key
words: electromagnetic hypersensitivity, provocation studies, systematic review.
EHS ⫽electromagnetic hypersensitivity; EMF ⫽electromagnetic
field; SMD ⫽standardized mean difference; UMTS ⫽universal
mobile telecommunications system (a “third-generation” mobile
phone signal); VDU ⫽visual display unit.
INTRODUCTION
“Electromagnetic hypersensitivity” (EHS) is a relatively
new phenomenon in which sufferers report a range of
symptoms that are apparently triggered by the presence of
weak electromagnetic fields (EMFs) (1). These symptoms
show no cohesive pattern (2,3) but are typified by nonspecific
sensations such as sleep disturbance, headaches, fatigue, and
subjective cognitive problems (2,3). In its more severe form,
EHS can be disabling, preventing sufferers from pursuing
normal work or social lives.
The electromagnetic triggers for the symptoms reported by
EHS sufferers are diverse. Although a list of the more com-
mon would include visual display units (VDUs), mobile
phones, mobile phone base stations, fluorescent lighting, over-
head power lines, and household items such as televisions and
microwave ovens (3), not all sufferers report being sensitive to
all potential triggers and many report only 1 or 2 specific
electrical items as being problematic. The EMFs emitted by
these devices vary considerably and encompass frequencies in
the radio, microwave, kilohertz, and extremely low-frequency
ranges (3). In almost all cases, however, the intensity of the
EMFs that seem to trigger EHS symptoms is far below that
known to cause physiological changes in animal models (4,5).
The underlying causes of EHS are the subject of considerable
debate. On the one hand, it has been proposed that biophysical
factors may make a minority of people particularly sensitive to
EMF, with mechanisms involving the release of histamine by
mast cells (6) having been suggested. On the other hand, it has
also been proposed that the condition may be more psychologic
than physicochemical (7), with symptom amplification and clas-
sic conditioning being important. Clearly, the appropriate treat-
ment and management of EHS will depend on which of these
models is correct. The best way to determine this is to examine
the results of blind and double-blind experimental provocation
studies. These experiments typically expose volunteers with self-
reported EHS to 2 conditions, an active condition in which weak
EMFs are presented and an inactive condition in which they are
not. Two outcomes can then be examined: the participant’s
ability to correctly discriminate active from inactive (their “elec-
tromagnetic sensibility”) and the participant’s tendency to expe-
rience more symptoms in the active condition (their “electromag-
netic hypersensitivity”) (8). Although electromagnetic sensibility
may be a necessary precondition for EHS, it is certainly not
sufficient as there is evidence that healthy individuals can display
heightened sensibility without necessarily experiencing symp-
toms as a result (8).
Two previous reviews have looked at provocation studies
for EHS in some detail. In 1997, a report for the European
Commission (1) described the results of 13 such experiments
and concluded that, although “‘electromagnetic hypersensi-
tive’ people do react in these provocation studies, [. . . ]these
reactions have not been shown to be related to the fields.”
More recently, a systematic review considered the results of 8
experimental studies published in peer-reviewed journals be-
fore coming to a similar conclusion (9).
The systematic review reported here attempted to identify
all blind or double-blind provocation studies for EHS that
could help to answer the following questions: are people who
are apparently hypersensitive to weak EMFs better at detect-
ing these fields under blind or double-blind conditions than
nonhypersensitive individuals, and do they respond to the
presence of weak EMFs with increased symptom reporting?
METHODS
Search Strategy for the Identification of Studies
The following electronic databases were searched for potentially relevant
studies: AMED, ASSIA, Cinahl, the Cochrane Collaboration Library, Em-
From the Mobile Phones Research Unit, Division of Psychological Medi-
cine, Institute of Psychiatry and Guy’s, King’s and St. Thomas’ School of
Medicine, King’s College London, UK.
Address correspondence and reprint requests to Gideon James Rubin, BSc,
MSc, PhD, Mobile Phones Research Unit, New Medical School Building,
Bessemer Road, London SE5 9PJ, UK. E-mail: g.rubin@iop.kcl.ac.uk
Received for publication May 10, 2004; revision received September 23,
2004.
DOI: 10.1097/01.psy.0000155664.13300.64
224 Psychosomatic Medicine 67:224–232 (2005)
0033-3174/05/6702-0224
Copyright © 2005 by the American Psychosomatic Society

base, Index to Theses, the Institute of Electrical and Electronics Engineers
EMF research database, ISI Proceedings, ISI Web of Science, Medline,
Psychinfo, and the World Health Organisation EMF research database. These
were searched from inception to January 2004 for a wide range of MeSH or
free-text key words relating to EHS, including, for example, “electrical
sensitivity,” “electromagnetic hypersensitivity,” “electro-smog,” “techno-
stress,” “screen dermatitis,” and “environmental illness.” The databases were
also searched for papers using combinations of relevant stimulus MeSH or
free-text key words (eg, “mobile phone,” “computer,” “power line”) and
hypersensitivity MeSH or free-text key words (eg, “allergy,” “hypersensitiv-
ity,” “intolerance”).
In addition, the Bioelectromagnetics Society conference proceedings for
1996 to 2003 were handsearched, as were the documents available on the
COST 281 web site. The reference sections of any pertinent studies and
reviews were also examined for additional references.
Inclusion Criteria
Only blind or double-blind experimental provocation studies were eligible
for inclusion in the review, with a provocation study defined as any experi-
ment in which the participants were systematically exposed to higher EMFs
in 1 experimental condition than in another. Because we were interested
primarily in the effects of ambient EMF, studies in which an electrical current
was directly applied to participants were excluded (e.g. 10).
Eligible studies were required to test a discrete sample of participants who
reported symptoms associated with low-intensity manmade electromagnetic
stimuli. This attribution had to have been explicitly made by either the
participants themselves or the experimenters. So, for example, studies inves-
tigating symptomatic VDU users in which no attribution of the symptoms to
the VDU had been explicitly made (e.g. 11) were excluded, as were studies in
which participants reported symptoms that were suspected to be the result of
visual or ergonomic features of the VDU. Studies that examined the putative
effects of EMFs on healthy volunteers alone were also excluded.
Finally, only studies reporting data relating to certain outcomes were
included in the review. In particular, we were only interested in experiments
that examined outcomes that are central to the self-diagnosis of EHS, ie,
subjective symptoms, physical signs such as observer-rated skin rash and
subjective perception of whether EMF is present. Proxy measures of ill health
such as blood tests or skin biopsies are less relevant to this process of
self-diagnosis and are not covered in this article.
Data Extraction and Analysis
Details regarding the following aspects of each included study were
extracted and tabulated: sample (n, details regarding any control sample),
stimuli (type of provocation), design (number and length of provocations per
participant), sensibility results (total number of provocations correctly iden-
tified as active or inactive by participants, number of participants classified as
able to discriminate active from inactive), and hypersensitivity results (type of
self-reported symptoms or other relevant health outcomes measured, signifi-
cance level of any difference).
Metaanalyses were conducted to determine whether the proportion of
EHS participants apparently able to discriminate active from inactive condi-
tions was greater than the corresponding proportion of control participants.
Review Process
The literature searches, assessments of inclusion, and data extraction were
conducted by Gideon James Rubin with uncertainties resolved through con-
sultation with the 2 coauthors. Where uncertainty existed as to whether 2
papers reported data from the same experiment, clarification was sought from
the authors of the original papers.
RESULTS
Search Results
In total, approximately 8600 titles or abstracts were exam-
ined from which 497 papers were selected as potentially
relevant to the review and examined in full. Of these, 372
were excluded because they were review papers, editorials,
duplicate publications, or did not include a sample of people
whose symptoms were explicitly attributed to EMF. A further
83 did not describe provocation studies, and 13 reported
unblinded provocations.
Details relating to 31 individual provocation experiments
involving 725 EHS participants were included. Thirteen ex-
periments (213 EHS participants) used VDU-related provoca-
tions as their active exposure condition, 7 (161 EHS partici-
pants) examined mobile phone-related provocations, 10 (315
EHS participants) examined other EMF provocations for par-
ticipants with generalized EHS, and 1 (36 EHS participants)
tested an apparently healthy group of volunteers categorized
as electromagnetically hypersensitive based on a median split
for the results of a single questionnaire item.
Visual Display Unit-Related Experiments
Details regarding the 13 VDU-related experiments are
given in Table 1. Of these, 1 reported significantly worse
symptoms in the active condition than in the sham condition
for 1 of the 10 tests conducted (standardized mean difference
[SMD] ⫽1.0, p⬍.01) (12), although the authors suggested
that this result probably reflected a type 1 error caused by
multiple significance tests. A second study found that more
EHS participants reported a reduction in skin “tingling, prick-
ling or itching” after 2 weeks of work with an activated
electric-conductive filter fitted to their computer screen com-
pared with 2 weeks of work with a deactivated “placebo” filter
fitted (13). However, this effect was small (mean difference ⫽
0.1 on a 10-point scale), no other symptoms were affected, and
the authors were subsequently unable to replicate this finding
in a larger study with a longer exposure period (14). Of the
remaining 10 studies, none found any evidence that EHS
participants had greater sensibility than healthy controls or
experienced more symptoms in active compared with inactive
conditions (7,15–22).
Two of the 99 EHS participants (2.0%) for whom infor-
mation was available in these studies appeared to be reliably
able to discriminate EMF from sham conditions, compared
with 1 of 32 healthy controls (3.1%). This difference was not
statistically significant (chi-square ⫽0.13, df ⫽1, p⫽.72).
Mobile Phone-Related Experiments
Details of the 7 studies relating to mobile phone hypersen-
sitivity are given in Table 2. One of these reported that a single
participant of the 7 tested could detect whether a mobile phone
hidden inside a bag was “on” or “off” 9 times out of 9 (23).
Unfortunately, this study did not include a nonsensitive con-
trol group. Moreover, in an as-yet unpublished study, the same
group previously used similar methods to test “about 70” EHS
participants 3 to 12 times each without finding any who could
reliably make this discrimination (Johansson, personal com-
munication), suggesting that the significant finding may be a
statistical artefact caused by repeated testing.
In another study, Zwamborn et al. (24) reported that expo-
sure to a universal mobile telecommunications system
ELECTROMAGNETIC HYPERSENSITIVITY
225Psychosomatic Medicine 67:224–232 (2005)

TABLE 1. Provocation Studies for Visual Display Unit-Related Sensitivities
Reference Sample Active Stimulus Number and Length of
Exposures
Results (all p⬎.05 unless otherwise indicated)
Total Number of Correct
Discriminations Between
Active and Sham (number
of participants apparently
demonstrating
electromagnetic
sensibility)
Comparison Between
Electromagnetic Field Conditions for
Self-Reported Symptoms
Comparison Between
Electromagnetic Field
Conditions for Other
Health Outcomes
Nilsen, 1982 (17) 5 EHS Exposure to a VDU Two 6-hr provocations, 1 active
and 1 inactive
Not measured None Observer-rated skin rash
Swanbeck, 1989 (18) 30 EHS Exposure to moderate
EMF VDU and low
EMF VDU
One 3-hr provocation to each
VDU
Not measured Heat or reddening, itching, stinging,
oedema, “others”
None
Hamnerius, 1993 (19) 30 EHS Exposure to VDU-like
magnetic fields
Up to 8 1-hr provocations, in
pairs of active and inactive
EHS: 38/80 (0/30) Unspecified symptoms Erythema, observer-
rated skin redness
Hellbom, 1993 (22) 6 EHS Exposure to a VDU Four 30-min provocations; 2
active, 2 inactive
EHS: 8/24 (0/6) Symptoms—unspecified in
translation
None
Sandstom, 1993 (20) 22 EHS Various VDU
provocations
Varying durations of exposure Not measured Skin tightness, heat, itching,
pricking, aching, other
None
Oftedal, 1995 (13) 19 EHS Normal office work with
an active or inactive
VDU filter
Participants worked for 2 weeks
with each filter
Not measured Heat, itching (p⫽.03), skin
tightness, tenderness, redness,
blisters/acne, desquamation
Observer-rated
dermatologic status
Sjoberg, 1995 (12) 7 EHS, 5 healthy controls for a
subset of provocations
Exposure to 3 different
VDU strengths
Four active and 4 inactive 1-hr
provocations for each
exposure
EHS: 99/176 (0/7);
control: not reported
(0/5)
“Comparison with symptoms from
(liquid crystal display) work” (p
⬍.01); 9 other unspecified
symptoms
None
Andersson, 1996 (21) 16 EHS Exposure to a computer
and VDU
At least 2 pairs of 30-min
provocations (1 active and 1
inactive) before and after a
period of cognitive
behavioral therapy
EHS: 41/80 (1/16) Unspecified symptoms previously
reported by the participants as
elicited by EMF
None
Keisu, 1996 (15) 1 EHS, 1 healthy control Exposure to a personal
computer
Ten provocations, each
randomized to active or
inactive
EHS: 6/10 (0/1); control:
5/10 (0/1)
Not measured None
Oftedal, 1999 (14) 38 EHS Normal office work with
an active or inactive
VDU filter
Participants worked for 3
months with each filter
Not measured Heat, itching, skin tightness, skin
redness, eye stinging, eye pain,
eye redness, eye tiredness,
sensitivity to light, headaches,
dizziness, tingling, fatigue
None
Flodin, 2000 (16) 15 EHS, 26 healthy controls Different provocations,
most using a VDU
Two active and 2 sham
exposures of up to 1 hr each
EHS: 29/60 (1/15);
control: 30/60 (1/26)
Ten unspecified symptoms relating
to skin, mouth, airways,
abdominal sensations, mental
sensations
None
Lonne-Rahm, 2000a (7) 12 EHS, 12 healthy controls Exposure to a computer
and VDU, with or
without the presence
of a stressor
One 30-min provocation to
each of 4 conditions (VDU
on or off, stressors present or
absent)
EHS: 22/48 (0/12);
control: not reported
Facial skin sensations, stress level,
tiredness; no interactions were
found between stressor and EMF
condition
None
Lonne-Rahm, 2000b (7) 12 EHS, 12 healthy controls Exposure to a computer
and VDU, with or
without the presence
of a stressor
One 30-min provocation to
each of 4 conditions (VDU
on or off, stressors present or
absent)
EHS: 29/52 (0/12);
control: not reported
Facial skin sensations, stress level,
tiredness; no interactions were
found between stressor and EMF
condition
None
EHS ⫽electromagnetic hypersensitivity; EMF ⫽electromagnetic field; VDU ⫽visual display unit.
G. J. RUBIN et al.
226 Psychosomatic Medicine 67:224–232 (2005)

TABLE 2. Provocation Studies for Mobile Phone-Related Sensitivities
Reference Sample Active Stimulus Number and Length of
Exposures
Results (all p⬎.05 unless otherwise indicated)
Total Number of Correct
Discriminations Between
Active and Sham (number
of participants apparently
demonstrating
electromagnetic
sensibility)
Type of Self-Reported
Symptoms Measured and
Comparison Between Active
and Inactive Conditions
Other Health Outcomes
Measured and Comparison
Between Active and Inactive
Conditions
Johansson, 1995
(23)
7 EHS Mobile phone hidden
inside a bag
Participants exposed up to 9
times each; each exposure
randomized as active or
inactive; exposures lasted
for twice the time necessary
to provoke symptoms
during a nonblind
provocation
EHS: 25/37 (1/7 [p⫽
.002])
None None
Radon, 1998
(25)
11 EHS GSM 900 signal A series of 12 trials, each
consisting of 3 2-min
exposures, 1 active and 2
inactive
EHS: 54/132 (0/11) None None
Raczek, 2000
(27)
16 EHS GSM 900 signal A series of 21 trials, each
consisting of 3 3-min
exposures, 1 active and 2
inactive
EHS: 94/336 (0/16) None None
Barth, 2000 (26) 1 EHS Mobile phone Patient exposed to 15 active
provocations and 16
inactive provocations
EHS: 13/31 (0/1) None None
Hietanen, 2002
(28)
20 EHS Analog, GSM 900 and
GSM 1800 signals
One 30-min exposure to each
condition
EHS: not reported (0/20) Number of symptoms reported
by participants during
experiment was greater in
inactive condition (no
statistical analysis)
None
Johansson, 2003
(unpublished
data)
70 EHS Mobile phones Between 3 and 12
provocations per
participant; exposures
lasted for twice the time
necessary to provoke
symptoms during a
nonblind provocation
EHS: not reported (0/70) Unspecified symptoms None
Zwamborn, 2003
(24)
36 EHS, 36
healthy
controls
GSM 900, GSM 1800,
and UMTS mobile
phone base station
signals
Each volunteer exposed to 3
45-min provocations, 1
inactive and 2 active
Not measured Anxiety (p⬍.05) 关B兴, somatic
symptoms (p⬍.05) 关B兴,
inadequacy (p⬍.05) 关B, E兴,
depression, hostility (p⬍.05)
关B, C兴
Reaction time (p⬍.05) 关A,
E兴, memory comparison
(p⬍.05) 关D,E兴, selective
attention (p⬍.05) 关B, E兴,
dual-tasking reaction time
(p⬍.05) 关D兴, filtering
irrelevant information (p
⬍.05) 关A兴
A, Comparison of inactive and 900 MHz for EHS group; B, comparison of inactive and 2100 MHz for EHS group; C, comparison of inactive and 900 MHz for control group; D, comparison of inactive
and 1800 MHz for control group; E, comparison of inactive and 2100 MHz for control group.
EHS ⫽electromagnetic hypersensitivity; UMTS ⫽universal mobile telecommunications system.
ELECTROMAGNETIC HYPERSENSITIVITY
227Psychosomatic Medicine 67:224–232 (2005)

(UMTS) mobile phone base station signal resulted in greater
levels of anxiety, somatic symptoms, inadequacy, and hostility
for EHS participants than exposure to a sham signal (p⬍.05).
A healthy control group was also significantly more affected
by the UMTS signal than the sham signal, experiencing an
increase in inadequacy only (p⬍.05). These effects were
smaller in the control group (overall SMD for well-being ⫽
0.22) than in the EHS group (SMD ⫽0.36), although the
authors of the study caution against direct comparison be-
tween the groups because of the demographic differences
observed between them. Some changes in objective cognitive
measures were also apparent for both samples as a result of
exposure to UMTS, 900-MHz and 1800-MHz signals, al-
though no consistent pattern was found in terms of which
cognitive variables were affected by which type of signal (see
Table 2). Furthermore, although some cognitive parameters
(ability to filter information, reaction time) showed significant
decrements as a result of exposure, others (memory, visual
attention, dual-tasking reaction time) showed significant im-
provements.
Of the 4 remaining blind or double-blind mobile phone-
related studies (25–28), none found any effect indicative of
biophysical hypersensitivity.
Only 1 EHS participant (0.8%) of the 125 for whom
information was available in these studies was consistently
able to discriminate active from inactive exposures. No com-
parable data were available for any control participants. Nev-
ertheless, the probability of 1 or more of 125 participants
identifying “on” from “off” 9 times out of 9, as this participant
did, but purely by chance, is p⫽.22.
General Electromagnetic Hypersensitivity-Related
Studies
Ten studies tested individuals reporting EHS using provo-
cations with other sources of weak EMF (Table 3). One of the
earliest, by Rea et al. (29), tested 100 patients with EHS and
comorbid “biological inhalant, food and chemical sensitivi-
ties.” Of these, 16 individuals were identified who repeatedly
responded to certain EMF frequencies with symptoms but
who did not respond to inactive challenges. A control group of
healthy volunteers showed no response to either type of chal-
lenge. Although the substantive parts of this study were de-
scribed as double-blind, the authors reported that the exposure
equipment and its operator were present in the testing room
during the experiment. A subsequent attempt by this group to
replicate their findings, but this time using a screen to prevent
participants from seeing the manipulation of the exposure
equipment, did not find any evidence of biophysical hyper-
sensitivity (30).
Aside from this study, only 1 other in this category found
any significant effect of EMF (31). This crossover experiment
exposed EHS participants to 4 hours of nighttime EMF or an
inactive sham condition over the course of 4 weeks. Measure-
ments of mood during the morning revealed significantly
higher levels of pleasure (p⫽.01) and arousal (p⫽.05)
during the EMF condition, a finding that runs contrary to the
self-reports of EHS sufferers. The explanation for these find-
ings is still unclear, but given their unexpected direction, they
do not seem to support the hypothesis that EHS sufferers are
adversely affected by EMF. Of the remaining 7 studies, none
identified any significant effect of EMF on EHS sufferers
(32–38).
Six of 95 EHS participants (6.3%) and 1 of 47 control
participants (2.1%) were reliably able to discriminate active
from inactive conditions in these experiments, a nonsignifi-
cant difference in proportions (chi-square ⫽1.18, df ⫽1, p⫽
.28).
Other Studies
One other study (39) was identified. This used a between-
participants design to test a group of 66 volunteers drawn from
student and military populations. Two of the 3 provocations
used are particularly relevant here: exposure to EMF and
noise, and exposure to noise only. Adjusting for performance
in a third condition in which noise and EMF were not present,
participants classified as hypersensitive were significantly
more affected by the presence of EMF than control partici-
pants in terms of their performance on visual processing and
visual attention tasks (p⬍.05). No such effects were seen for
3 other cognitive variables or for subjective discomfort (see
Table 4). However, interpretation of these results is compli-
cated by several methodologic factors, including the catego-
rization of participants into sensitive or control groups on the
basis of responses given after testing had been completed and
the inappropriate use of 1-tailed significance tests.
DISCUSSION
Current Evidence
To date, 7 blind or double-blind provocation studies have
found some effect of EMF provocation on people who report
EHS (12,13,23,24,29,31,39). However, even the original au-
thors have been unable to replicate the results of 2 of these
(14,30), the results of 3 more seem to be statistical artefacts
reflecting the large number of significance tests conducted
(12), the large number of participants tested (23); unpublished
Johansson study), or the inappropriate use of 1-tailed signifi-
cance tests (39), and the results of the remaining 2 are mutu-
ally inconsistent, with one showing improved mood as a result
of provocation (31), while the other shows worse mood (24).
Meanwhile the cognitive effects of this last study were appar-
ent in both the control group and the hypersensitive group and
appeared to reflect both improvements and impairments in
cognition (24). Twenty-four other blind or double-blind prov-
ocation studies have found no evidence that people with
apparent EHS are especially sensible or hypersensitive to
EMF. Our metaanalyses also confirmed these findings. In
summary, we have therefore been unable to find any robust
evidence to support the existence of EHS as a biologic entity.
On the other hand, several of the experiments reviewed
also examined the effects of a nonblind exposure to the
relevant stimulus (7,22,23). All found that when the partici-
pants were aware that the EMF source was switched on, they
G. J. RUBIN et al.
228 Psychosomatic Medicine 67:224–232 (2005)

TABLE 3. Provocation Studies for Generalized Electromagnetic Hypersensitivity.
Reference Sample Active Stimulus Number and Length of
Exposures
Results (all p⬎.05 unless otherwise indicated)
Total Number of Correct
Discriminations Between
Active and Sham (number of
participants apparently
demonstrating
electromagnetic sensibility)
Type of Self-Reported
Symptoms Measured and
Comparison Between
Active and Inactive
Conditions
Other Health
Outcomes
Measured and
Comparison
Between Active
and Inactive
Conditions
Rea, 1991
(29)
100 EHS, 25 healthy
controls
Exposure to EMFs of varying
frequencies
Three testing phases involving
repeated testing with 3-min
exposures to different
frequency EMFs and inactive
challenges
EHS: not measured; control:
not measured
Patients asked to describe
any symptoms; 16/100
EHS participants
consistently reported
symptoms in active, but
not inactive conditions,
compared with 0/25
control participants
None
Wennberg,
1994 (32)
25 EHS, 13 healthy
controls
ELF/VLF fields Multiple exposures of less than
10 min
EHS: unreported (0/25);
control: unreported (0/13)
Patient asked to report
onset of any symptoms
None
Johansson,
1995 (33)
7 EHS “Electric and/or magnetic
fields”
Unknown number of 60-min
provocations
Study abandoned; authors
reported that “several
milieu factors” may have
interfered with the study
Not reported Not reported
Wang, 1995
(30)
19 EHS, 34 healthy
controls
Exposure to EMFs of varying
frequencies
Nine 3-min provocations; 6
active, 3 inactive
Not measured Unspecified symptoms None
Bertoft, 1996
(34)
4 EHS Exposure to EMFs generated
by a dental chair and unit
Two 1-hr provocations; 1
active and 1 inactive
Not measured Neurologic,
musculoskeletal,
cardiovascular,
oral/respiratory,
gastrointestinal, ocular,
and dermal symptoms
None
Toomingas,
1996 (35)
1 EHS Whole-body Helmholtz coil
with 2 field intensities
Twenty-four 1 or 10-s
provocations to inactive or
active conditions
EHS: unreported (0/1) Patient asked to describe
any symptoms
None
Mueller,
2000 (31)
53 EHS Exposure to intermittent or
constant EMF
Exposure to active or inactive
provocations (1 4-hr period
per night for 25 nights)
conducted while participant
was sleeping
Not measured Sleep quality, pleasure (p⫽
.01), arousal (p⫽.05)
None
Rei
␤
enweber,
2000 (36)
37 EHS, 37 healthy
controls
50-Hz sinusoidal field Twenty 2-min exposures, half
active, half inactive
EHS: unreported (unreported);
control: unreported
(unreported); authors
report “no significant
differences in guess
probabilities of groups”
Patients asked to describe
any symptoms, but no
analyses reported
None
Lyskov, 2001
(37)
20 EHS, 20 healthy
controls
Weak magnetic fields, with
or without concurrent
mathematical task
Two testing days: day 1
entirely inactive, day 2 had
4 10-min exposures to
active and inactive fields,
with or without maths task
EHS: unreported (0/20);
control: unreported (0/20)
Tactile feelings and general
fatigue more prevalent
in inactive condition (no
statistical analysis)
None.
Mueller,
2003 (38)
49 EHS, 14 healthy
controls
Low-intensity EMF Twenty 2-min provocations: 10
active and 10 inactive
EHS: unreported (6/49);
control: unreported (1/14)
None None
EHS ⫽electromagnetic hypersensitivity; EMF ⫽electromagnetic field.
ELECTROMAGNETIC HYPERSENSITIVITY
229Psychosomatic Medicine 67:224–232 (2005)

reported being able to detect the EMF or experienced more
symptoms. Another study found that participants’ beliefs
about the status of a double-blind exposure significantly pre-
dicted symptom reporting regardless of whether these beliefs
were correct (21). Given that the actual presence of EMF did
not correlate with increased symptom severity in these studies,
these findings suggest that psychologic mechanisms may play
at least some role in causing or exacerbating EHS symptoms.
Review Methodology
Our results illustrate the need to conduct reviews system-
atically and without heed to the publication status, language,
or conclusions of the primary research. By doing this, we have
been able to identify a large number of studies that have gone
unreported in previous reviews. It is always possible, how-
ever, that additional provocation studies exist that we were not
able to find. Yet, given that publication bias makes studies
with significant results easier to locate, it is very unlikely that
any missing studies would alter our conclusions.
A subtler problem for this review concerns the homogene-
ity of the participants included in the original research. Al-
though some studies focused on participants with a single
complaint such as skin rashes caused by VDUs (17) or symp-
toms caused by mobile phones (28), others tested volunteers
with more generalized and typically more severe sensitivities
under the assumption that this would make any effect easier to
detect (16). In this review, we have made no distinction
between these categories. Given that there is no a priori reason
to assume that any biophysical mechanisms governing the
adverse effects of EMF will differ between these groups, we
believe we are justified in doing this. Nevertheless, arguments
might be made for assessing these groups independently.
Although poor reporting by the original studies makes subdi-
vision into these categories difficult (1), we do not believe that
doing this would alter our conclusions; the studies reviewed
provide no robust evidence for the existence of generalized
EHS, and there is currently no good evidence for the existence
of more specific sensitivities to mobile phone signals or VDU
emissions. However, as new technologies with different EMF
characteristics are developed, it is likely that new sensitivities
to them will be reported. As such, there will always be room
to argue that the latest form of EHS has yet to be fully
investigated.
Research Methodology
Why have so many provocation studies failed to produce
significant findings? A number of suggestions have been put
forward. For example, it has been suggested (24,29) that only
studies conducted in specially designed chambers, which
shield against all extraneous background EMF, would be able
to produce a significant result. However, although this would
presumably reduce any “noise” in symptom reporting and
make increased symptoms easier to identify, people with EHS
do usually report being able to detect the presence of electro-
magnetic triggers in everyday life. It should therefore also be
possible for scientists to detect this sensitivity against the
TABLE 4. Other Provocation Studies
Reference Sample Active Stimulus Number and Length
of Exposures
Results (all p⬎.05 unless otherwise indicated)
Total Number of Correct
Discriminations Between Active and
Sham (number of participants
apparently demonstrating
electromagnetic sensibility)
Type of Self-Reported
Symptoms Measured
and Comparison
Between Active and
Inactive Conditions
Other Health Outcomes
Measured and Comparison
Between Active and Inactive
Conditions
Trimmel,
1998
(39)
36 EHS, 30 healthy
controls
EMFs generated by a
transformer coil
Exposure to either 1
hr of EMF ⫹noise
or to 1 hr of just
noise
Not measured Discomfort Unsuccessful visual processing,
precise visual processing (p
⬍.05), visual attention (p
⬍.05), visual perception,
verbal memory
EMF ⫽electromagnetic field.
G. J. RUBIN et al.
230 Psychosomatic Medicine 67:224–232 (2005)

backdrop of normal EMF levels. Moreover, many of the
studies reviewed did use techniques to reduce or control for
background EMF but were still unable to identify an effect
(7,12,16,19,21,28,30,36,38).
Other arguments sometimes heard are that the participants
in an experiment were not sensitive enough, that the wrong
symptoms were measured, that the exposure used was the
wrong kind, or that the follow up was too short. Again, none
of these seem very plausible as an explanation for all the
negative results. In particular, we note that when provocation
studies have included nonblind arms or training sessions, then
it usually has been possible to measure an increase in symp-
toms (7,22,23). Other studies have found participants to be
confident in their assertions that a particular session was
active or inactive (15,21) or have identified increases in symp-
toms that are similar to those found after exposure to real-life
triggers (18,34,37). The conclusions of the 1997 European
Commission report thus still seem to be valid: participants do
experience “realistic” symptoms in these experiments, but this
is apparently not associated with the presence of EMF (1).
More of an issue are “hangover effects.” EHS sufferers
sometimes report symptoms that can last for several days (3).
It is therefore important that studies use an appropriate inter-
val between provocations to prevent symptoms experienced in
one from masking any effects of the next. It is difficult to
know how many studies were affected by this, but at least 2
reported that hangover effects did seem to have an impact on
their results (18,29).
Low statistical power might also explain some of the neg-
ative findings. This is not simply an issue of sample size;
testing a single participant numerous times can be a powerful
design, although it does reduce the generalizability of any
negative findings. Nevertheless, the majority of studies in-
cluded relatively small sample sizes and failed to justify this
by providing a power calculation. Such a calculation could be
based on participant reports of usual symptom severity after
real-life exposure to the stimulus and should now be given in
any future research.
Implications for Researchers and Clinicians
We are aware of at least 6 more double-blind provocation
studies into EHS that are currently ongoing, each of which is
examining hypersensitivity symptoms associated with mobile
phone-type signals. The results of these studies will determine
whether any more research in this area is needed. If further
studies are necessary, the following guidelines should be
considered: 1) Studies should include a healthy control group.
This will not only provide a standard against which to judge
heightened sensibility (38), but it will also provide a useful
indication of the adequacy of the blinding (16). 2) The nature
and severity of participants’ self-reported sensitivities should
be assessed and reported. This might then be used to produce
a power calculation, which should also be reported. 3) The
inclusion of open-blind provocation sessions should be con-
sidered as a useful way of ensuring the face validity of the
experiment. 4) Hangover effects should be controlled by the
use of lengthy intervals between exposures or else checked for
by the use of preexposure measures of symptom severity.
Our review suggests that treatment for EHS should not
simply focus on reducing EMF exposure because this is un-
likely to address the root causes of the problem (14). Identi-
fying what these causes are may require careful investigation.
For some, complaints of EHS may mask organic or psychiatric
pathology, whereas others may benefit from a course of cog-
nitive behavioral therapy (21).
CONCLUSIONS
This systematic review could find no robust evidence to
support the existence of a biophysical hypersensitivity to
EMF.
The authors thank Paul Philo, Carina von Schantz, Sophie Sedgwick,
Bella Stensans, and Julia Weingarten for their help with translation
and Gary Hahn for his help locating and organizing the literature.
This work was undertaken by James Rubin, Jayati Das Munshi, and
Simon Wessely, who received funding from the Mobile Telecommu-
nications and Health Research Programme. The views expressed in
the publication are those of the authors and not necessarily those of
the funders.
REFERENCES
1. Bergqvist U, Vogel E, Aringer L, Cunningham J, Gobba F, Leitgeb N,
Miro L, Neubauer G, Ruppe I, Vecchia P, Wadman C. Possible health
implications of subjective symptoms and electromagnetic fields; a report
prepared by a European group of experts for the European Commission,
DG V. Solna: Arbete och Halsa; 1997.
2. Hillert L, Berglind N, Arnetz BB, Bellander T. Prevalence of self-
reported hypersensitivity to electric or magnetic fields in a population-
based questionnaire survey. Scand J Work Environ Health 2002;28:
33– 41.
3. Ro¨o¨sli M, Moser M, Baldinini Y, Meier M, Braun-Fahrla¨nder C. Symp-
toms of ill health ascribed to electromagnetic field exposure—a ques-
tionnaire survey. Int J Hyg Environ Health 2004;207:141–50.
4. Independent Expert Group on Mobile Phones. Mobile phone and health.
Report of an independent expert group on mobile phones. Chilton:
National Radiological Protection Board; 2000.
5. Ziskin MC. Electromagnetic hypersensitivity: a COMAR technical infor-
mation statement. IEEE Eng Med Biol 2002;21:173–5.
6. Gangi S, Johansson O. A theoretical model based upon mast cells and
histamine to explain the recently proclaimed sensitivity to electric and/or
magnetic fields in humans. Med Hypotheses 2000;54:663–71.
7. Lonne-Rahm S, Andersson B, Melin L, Schultzberg M, Arnetz B, Berg
M. Provocation with stress and electricity of patients with ’sensitivity to
electricity.’ J Occup Environ Med 2000;42:512– 6.
8. Leitgeb N, Schro¨ttner J. Electrosensibility and electromagnetic hypersen-
sitivity. Bioelectromagnetics 2003;24:387–94.
9. Levallois P. Hypersensitivity of human subjects to environmental electric
and magnetic field exposure: a review of the literature. Environ Health
Perspect 2002;110(suppl 4):613– 8.
10. Leitgeb N, Fluhr H. Einflussfaktoren der Elektrosensibilitat. VEO J
1998;6:44 –50.
11. Skyberg K, Skulberg K, Eduard W, Vistnes AI, Levy F, Djupesland P.
Electric fields, dust and health problems among VDU-users in an office
environment. An intervention program. Advances Occup Ergonomics
Safety 1998;2:81–3.
12. Sjo¨berg P, Hamnerius Y. Study of provoked hypersensitivity reactions
from a VDU. In: Katjalainen, Knave B, eds. Electromagnetic Hypersen-
sitivity. 2nd Copenhagen Conference; 1995.
13. Oftedal G, Vistnes AI, Rygge K. Skin symptoms after the reduction of
electric fields from visual display units. Scand J Work Environ Health
1995;21:335– 44.
14. Oftedal G, Nyvang A, Moen BE. Long-term effects on symptoms by
reducing electric fields from visual display units. Scand J Work Environ
Health 1999;25:415–21.
ELECTROMAGNETIC HYPERSENSITIVITY
231Psychosomatic Medicine 67:224–232 (2005)

15. Keisu L. Successful treatment of ‘electricity hypersensitivity.’ The pa-
tient was assisted in curing himself. [Swedish] Lakartidningen 1996;93:
1753–5.
16. Flodin U, Seneby A, Tegenfeldt C. Provocation of electric hypersensi-
tivity under everyday conditions. Scand J Work Environ Health 2000;
26:93– 8.
17. Nilsen A. Facial rash in visual display unit operators. Contact Dermatitis
1982;8:25– 8.
18. Swanbeck G, Bleeker T. Skin problems from visual display units. prov-
ocation of skin symptoms under experimental conditions. Acta Derm
Venereol 1989;69:46 –51.
19. Hamnerius Y, Agrup G, Galt S, Nilsson R, Sandblom J, Lindgren R.
Double-blind provocation study of hypersensitivity reactions associated
with exposure from VDUs. Preliminary short version. R Swed Acad Sci
Rep 1993;2:67–72.
20. Sandstro¨m M, Stenberg B, Hansson-Mild K. Experiences of provocations
with electric and magnetic fields. R Swed Acad Sci Rep 1993;2:62– 6.
21. Andersson B, Berg M, Arnetz BB, Melin L, Langlet I, Lide´n S. A
cognitive-behavioral treatment of patients suffering from ’electric
hypersensitivity’: subjective effects and reactions in a double-blind prov-
ocation study. J Occup Envion Med 1996;38:752– 8.
22. Hellbom M. Provocation attempt with individuals hypersensitive to elec-
tricity. Effects of information about the presence of electric and magnetic
fields. [Swedish Dissertation] Uppsala: Department of Applied Psychol-
ogy, University of Uppsala; 1993.
23. Johansson O. Hypersensitivity to electricity and sensitivity to mobile
phones. Results from a double-blind provocation study of methodological
character [Swedish Report]. Stockholm: Department of Experimental
Dermatology, Karolinska Institute; 1995.
24. Zwamborn APM, Vossen SHJA, van Leersum BJAM, Ouwens MA,
Makel WN. Effects of global communication system radio-frequency
fields of well being and cognitive functions of human subjects with and
without subjective symptoms [Report]. The Hague: Netherlands Organi-
sation for Applied Scientific Research (TNO); 2003.
25. Radon K, Maschke C. Hypersensitivity to electricity. [German] Umwelt-
medizin in Forschung und Praxis 1998;3:125–9.
26. Barth A, Maritczak L, Valic E, Konnaris C, Wolf C. Pseudoangina
caused by exposure to electromagnetic fields (’electrosmog’). [German]
Dtsch Med Wschr 2000;125:830 –2.
27. Raczek J, Runow K, Oetzel H, Gailus T, Herget I. Investigations of
electrosensitivity to a GSM signal at 900MHz for a self-reported elec-
trosensitive target group. The Bioelectromagnetics Society 22nd Annual
Meeting; 2000:269 –70.
28. Hietanen M, Ha¨ma¨la¨inen AM, Husman T. Hypersensitivity symptoms
associated with exposure to cellular telephones: no causal link. Bioelec-
tromagnetics 2002;23:264 –70.
29. Rea WJ, Pan Y, Fenyves EJ, Sujisawa I, Samadi N, Ross GH. Electro-
magnetic field sensitivity. J Bioelectricity 1991;10:241–56.
30. Wang T. Sick building syndrome: a study of some contributing factors
[Dissertation]. University of Surrey; 1995.
31. Mueller Ch H, Krueger H, Schierz Ch. Project NEMESIS: double-blind
study on effects of 50Hz EMF on sleep quality and physiological param-
eters in people suffering from electrical hypersensitivity. The Bioelectro-
magnetics Society 22nd Annual Meeting; 2000:89 –90.
32. Wennberg A, Franze´n O, Paulsson L-E. Electromagnetic field provoca-
tions of subjects with ’electric hypersensitivity.’ In: Simunic D, ed.
COST 244 meeting on Electromagnetic Hypersensitivity. Graz: COST
244; 1994:133–9.
33. Johansson O, Liu P-Y. ’Electrosensitivity,’ ’electrosupersensitivity’ and
’screen dermatitis’: preliminary observations from on-going studies in the
human skin. In: Simunic D, ed. Proceedings of Cost 244: Biochemical
effects of electromagnetic fields—workshop on electromagnetic hyper-
sensitivity. Graz: EU (DG XIII); 1995:52–7.
34. Bertoft G. Patient reactions to some electromagnetic fields from dental
chair and unit: a pilot study. Swed Dent J 1996;20:107–12.
35. Toomingas A. Provocation of the electromagnetic distress syndrome.
Scand J Work Environ Health 1997;22:457– 8.
36. Rei
␤
enweber J, David E, Kentner S. Different aspects of electromagnetic
hypersensitivity. The Bioelectromagnetics Society 22nd Annual Meeting;
2000:270 –1.
37. Lyskov E, Sandstrom M, Hansson Mild K. Provocation study of persons
with perceived electrical hypersensitivity and controls using magnetic
field exposure and recording of electrophysiological characteristics. Bio-
electromagnetics 2001;22:457– 62.
38. Mueller CH, Krueger H, Schierz C. Project NEMESIS: perception of a 50
Hz electric and magnetic field at low intensities (laboratory experiment).
Bioelectromagnetics 2002;23:26 –36.
39. Trimmel M, Schweiger E. Effects of an ELF (50 Hz, 1 mT) electromag-
netic field (EMF) on concentration in visual attention, perception and
memory including effects of EMF sensitivity. Toxicol Lett 1998;96 –97:
377– 82.
G. J. RUBIN et al.
232 Psychosomatic Medicine 67:224–232 (2005)