Background

Although the issue of idiopathic intolerances attributed to environmental exposures (IEI) first appeared in the scientific literature more than five decades ago [1], the possible underlying causes, as the term “idiopathic” suggests, remain unclear [2] and there is no widely accepted protocol for the identification of patients and treatment [3]. A representative example is the variety of physical symptoms without a clear pathological basis that are attributed by the patients to relatively low-level exposure to non-ionizing electromagnetic fields (EMF), emitted by sources such as mobile phone devices and base stations, high-voltage overhead powerlines, computer equipment and domestic appliances [4]. This phenomenon is better known within the public and scientific context as "electromagnetic hypersensitivity"(EHS), although since 2005 the term “Idiopathic Environmental Intolerance Attributed to EMF" (IEI-EMF) has been proposed by the World Health Organization (WHO) as an etiologically neutral description [5]. In this paper, the descriptive term “IEI-EMF” is used.

According to the WHO [5], people with IEI-EMF are mainly characterized by the report of non-specific physical symptoms (NSPS), without a consistent pattern [6], such as redness, tingling, burning sensations in the facial area, fatigue, tiredness, lack of concentration, dizziness, nausea, heart palpitation and digestive disturbances. IEI-EMF is often accompanied by occupational, social and mental impairment [4, 7] and its estimated prevalence varies considerably, probably due to different methodological approaches; 1.5% in Sweden [6], 3.2% in California [8], 3.5% in Austria [9], 5% in Switzerland [10] and 13.4% in Taiwan [11]. Demographic characteristics such as age, gender and occupational status have repeatedly been associated with IEI-EMF [6, 10].

The experience and belief of IEI-EMF patients is in contrast with the scientific state of the art; results from systematic assessment of experimental and epidemiological evidence are consistent, concluding that a causal association of EMF exposure with symptomatic and other physiologic or cognitive reactions cannot be adequately supported [12–17]. IEI-EMF has been associated with psychological components [18–23] but their exact role is not clear. Although a possible effect of exposure cannot yet be ruled out because of methodological obstacles in research primarily regarding exposure assessment and study design [14, 16], more recent approaches stress the importance of looking into the interaction of environmental, biological, psychological and social pathways [24].

However, it is still controversial who should be categorised as having IEI-EMF. The lack of a validated, mutually accepted case definition and diagnostic instrument affects the quality of the research outcomes and increases the methodological heterogeneity, resulting in limited comparability between the studies. That stands in the way of a reliable estimation of the prevalence of IEI-EMF in the general population, proper meta-analysis of etiological evidence, the identification of health outcome patterns/profiles and contributes to a great deal of uncertainty regarding the characteristics, identification and management of this sensitivity by health care providers [25–27].

No systematic review has been performed yet focusing on the existing definitions and criteria for the identification of people with IEI-EMF. In light of the need to inform health care profesionals about relevant aspects of IEI-EMF and prepare the ground for discussion and consensus in the research community on widely supported case definition criteria, the present paper identified the relevant studies on IEI-EMF published to date, in order to summarize:

  • The descriptive terms used to define IEI-EMF.

  • The inclusion criteria and procedure for the identification of individuals with IEI-EMF.

Methods

Search strategy for the identification of studies

Initially, the following electronic databases were searched to detect relevant studies that were published from inception to April 2010: Embase (Elsevier B.V., Amsterdam, The Netherlands), Medline (US National Library of Medicine, Bethesda, Maryland), PsychInfo (American Psychological Association, Washington, DC). Web of Knowledge (Institute for Scientific Information, The Thomson Corporation, Stamford, Connecticut) and Scopus (Elsevier B.V., Amsterdam, The Netherlands). A wide range of (combined) keywords was used with regards to EMF exposure, sensitivity and related health outcomes, which is presented in Table 1. In addition to the electronic database searches, the reference sections of previous systematic reviews, key papers, international reports on EMF and health and research databases of websites focused on the issue of EMF such as the “EMF Portal” and the WHO webpage were checked for potentially relevant articles. A wide literature database held by the Mobile Phones Research Unit of King’s College London was also consulted. A second literature search was carried out in order to update our review with studies published from May 2010 to June 2011.

Table 1 Key search terms
Full size table

Inclusion criteria

Only primary studies written in English and published in the peer-reviewed literature were considered as suitable for inclusion in the current review. Conference presentations, brief communications and reviews were excluded. The primary condition to include a study was the report of use of at least one criterion to identify individuals with IEI-EMF. Studies focusing on health effects from wider environmental exposures (such as chemicals) were eligible as long as they attempted to identify sensitivity to EMF in their investigation. Studies recruiting exclusively “healthy” individuals without any attempt to assess IEI-EMF or identify relevant individuals were excluded. Since the “attribution” of health complaints to EMF is not necessarily synonymous with IEI-EMF and it is not an established prerequisite for its existence, studies relying solely on “attribution” without any mention of and explicit conceptual link with IEI-EMF or synonymous terms were not considered eligible for this review. Among papers based on the same sample and identifying criteria of IEI-EMF, the first publication was included.

Data extraction

For each included study, the following data were abstracted: reference and country, study design, methods and source of sample recruitment, IEI-EMF sample characteristics (such as sample size, age mean or range and gender distribution), type of sensitivity based on the triggering EMF source(s), the criteria used to identify individuals with IEI-EMF, exclusion criteria (based on self-report/interview or clinical examination) and the case definition procedure followed for the identification of IEI-EMF (such as self-report and/or medical examination to exclude the possibility that a diagnosed disorder was responsible for the reported health complaints) (Tables 2 and 3). The data provided in the tables were derived from the information that was given or could be inferred from the original publications. However, in some cases (part of) the necessary information was not provided in the reviewed articles.

Table 2 Experimental studies on IEI-EMF
Full size table
Table 3 Observational studies on IEI-EMF
Full size table

Review Process

The literature search was performed by the first author and the evaluation of inclusion criteria by CB, IVK and GJR, with uncertainties resolved through consultation among all the authors. The initial screening was based on the titles and/or abstracts. Next, the hard copies of the potentially eligible publications were examined to assess whether they met the inclusion criteria.

Results

Search results

Figure 1 illustrates the literature search process. We examined 5328 citations in total and identified 35 experimental and 28 observational studies that met our inclusion criteria.

Figure 1
figure 1

Flow diagram outlining the study selection process.

Full size image

Study characteristics

When reported, sample sizes of subjects with IEI-EMF ranged between 1 to 100 in the experimental studies and from 2 to 2748 in the observational studies. The percentage of female participants (exempting case-studies) ranged between 0 to 81.3% and 50% to 100% respectively. In all studies, the reported mean age of IEI-EMF individuals varied between 26.1 and 55.5 years. IEI-EMF triggered by several different EMF sources (“general”) was the sensitivity type of primary focus in the included investigations (n = 48), while 14 studies concentrated exclusively on “source-specific” IEI-EMF and three on both “general” and “source-specific” IEI-EMF. There was a variety of synonyms of IEI-EMF in the literature such as "hypersensitivity (HS) to EMF", "electromagnetic Hypersensitivity (EHS)", "electrohypersensitivity", "environmental annoyance attributed to EMF", "electromagnetic distress syndrome" and "environmental illness". “Hypersensitivity to EMF” (and its variants) was by far the most frequently used definition/descriptive term (Figure 2). In 35 studies the case definition procedure was solely based on the subjective report of the respondents. In 28 studies it was mentioned that objective assessment (e.g. medical and/or psychological assessment) was additionally taken into account.

Figure 2
figure 2

Distribution (%) of terms used to describe IEI-EMF in the reviewed literature. Abbreviations: IEI-EMF, Idiopathic environmental intolerance attributed to EMF; EHS, Electrohypersensitivity; HS, Hypersensitivity.

Full size image

The principal method of sample recruitment was via study description in advertisements and/or local or national media (22 studies). The vast majority of the reviewed studies were conducted in Europe (58 studies).

Experimental studies

The major inclusion criteria used by experimental studies to identify individuals with IEI-EMF were:

  • Attribution of NSPS to either various or specific sources of EMF (being reported 13 times).

  • Self-reported IEI-EMF (or synonymous terms) (n = 14).

  • Experience of symptoms during or soon (from 20 minutes to 24 hours) after the individual perception or actual presence or use of an EMF exposure source (n = 10).

  • High score on a symptom scale (n = 6).

In addition, two studies used limitation in daily functioning of the individual due to the attributed health effects as an inclusion criterion.

The main exclusion criterion was the existence of a medical and/or psychiatric or psychological condition that could account for the reported health complaints (n = 20).

Other exclusion criteria included undergoing treatment for somatic or psychiatric conditions (n = 8), pregnancy (n = 5), history of severe injuries (n = 3) and regular smoking (n = 2).

In 16 studies the case definition procedure did not only rely on subjective report, but also on medical and/or psychiatric and/or psychological examination. In eight studies, the sample recruitment was based on participants who were already referred or registered to a health care institution (such as a university hospital) for their health complaints. All extracted data from the experimental studies are presented in Figure 2.

Observational Studies

The major inclusion criteria used by observational studies to identify individuals with IEI-EMF were:

  • Self-reported IEI-EMF (or synonymous terms) (n = 16).

  • Attribution of NSPS to either various or specific EMF sources (n = 12).

  • Experience of symptoms during or soon (from 20 minutes to 24 hours) after the individual perception or actual presence or use of an EMF exposure source (n = 3).

  • Limitation in daily functioning of the individual due to the attributed health effects (n = 2).

The main exclusion criteria were a medical and/or psychiatric or psychological condition that could account for the reported health complaints and undergoing treatment for somatic or psychiatric condition (n = 4).

Eleven studies included medical and/or psychiatric and/or psychological examination to assess whether a pathological condition was responsible for patients’ complaints. In nine studies the sample was based on participants who were already referred or registered to health care institutions for their complaints. All extracted data from the observational studies are listed in Table 3.

The prevalence of IEI-EMF in randomly selected samples of population-based epidemiological studies varied and seemed to be influenced by the number and degree of strictness of the applied identification criteria. This is illustrated in Figure 3. These differences could also be due to the population under study, year of study and sample stratification (e.g. age range).

Figure 3
figure 3

Prevalence (%) of IEI-EMF based on the identifying criteria employed by population-based observational studies.

Full size image

Discussion

The present systematic review based on an extensive literature search, summarized the case definition criteria and methods that have been used in the published literature to date for the identification of subjects with IEI-EMF.

It is noteworthy that only 1% of the reviewed studies used the term “IEI-EMF” as a descriptive term, despite the fact that it has been proposed by WHO since 2005 [5]. Sixty-five percent of the studies used the description “Hypersensitivity to EMF” which seems to be mainly characterized by the following aspects: Self-reported sensitivity to one or more sources of EMF, attribution of NSPS to either several or specific EMF sources (such as mobile phones and VDUs), experience of symptoms during or soon after (from 20 minutes to 24 hours) the individual perception or actual presence or use of an EMF source and absence of a (psycho)pathological condition accounting for the reported health complaints. In the majority of the studies the case definition procedure was based exclusively on self-report. In a smaller number of investigations, medical and/or psychiatric and/or psychological assessment was included.

In most of these studies participants were recruited from registries to a health care institution for their symptoms and for whom medical data were available. Although there were no important differences between observational and experimental studies in the most frequently employed criteria, experimental studies used a larger number of criteria per investigation compared to observational studies. Moreover, the demographic profile of the recruited individuals with IEI-EMF in terms of age and gender was quite consistent; the frequency of female gender and age over 40 years were considerably higher in most of the studies.

Despite previous attempts to bring order to this field [6, 53, 70], as it appears in the literature, IEI-EMF is still predominantly a self-reported sensitivity without a widely accepted and validated case definition tool. This could be due to the absence of a bioelectromagnetic mechanism [17] or because of the varying patterns regarding the symptom type, frequency and severity [6, 41]. The other way around could also be the case: The lack of validated case definition criteria could have hindered the identification of homogeneous patient groups and consequently the recognition of symptom profiles and a physiologic mechanism. Furthermore, the application of very broad criteria could dilute the power of the studies and make difficult the detection of those individuals that really suffer from IEI-EMF. For example, although “Attribution” of NSPS to EMF could be considered as a first indication of suffering from IEI-EMF, it is questionable whether it comprises a sufficient identifying criterion when used alone.

Possible subgroups

Several subdivisions may exist within IEI-EMF that may be of relevance to clinicians and researchers.

One such division is that between patients for whom an alternative diagnosis exists, which might account for their symptoms and those for whom it does not. The absence of screening for pathological conditions which might underlie the symptoms reported by participants in many studies was notable. Previous studies have identified occasionally high levels of other diagnoses in such patients, such as somatoform and anxiety disorders which might account for their ill-health [89, 90]. Including these individuals in the same sample as those for whom there is no clear explanation for their symptoms may reduce our ability to identify causal factors for IEI-EMF.

An additional distinction that we may need to take into account is between patients who attribute symptoms to short-term exposure to EMF and those for whom longer-term exposure is relevant. Furthermore, it remains unclear whether generalized and source-specific IEI-EMF should be assessed separately or not. Exposure from far-field sources such as high-voltage overhead powerlines and mobile phone base stations is mostly continuous and people often perceive it as less controllable compared to near-field sources such as mobile phones [10] but there is still no convincing evidence for source-specific sensitivities [13]. As differences may exist between IEI-EMF patients in terms of their psychological and health-related characteristics, division into subgroups for the purposes of research may be of use [22, 23]. Perhaps the most complicated issue is to figure out whether self-reported-NSPS and objectively assessed physiologic reactions are preceded by events of the relevant (EMF) exposures, distinguishable from other random exposure events experienced during the day. Use of a prediction model based on modelled exposure from various sources [91, 92] could be a solution; however it is questionable whether and how it could be systematically incorporated in a case definition tool.

Table 4 illustrates a number of proposed aspects for IEI-EMF, based on a synthesis of the existing identifying criteria in the reviewed literature. Considering the fact that the reported symptoms are quite common in the general population and also the lack of symptom patterns [6, 53] and etiology, the only parameter that clearly distinguishes sensitive from control individuals is the causal attribution of symptomatology to EMF exposure. Therefore, the attribution of health outcomes and self-reported sensitivity to EMF inevitably constitute, at the moment, the cornerstone of IEI-EMF case definition in research and clinical practice. Additional aspects such as medical examination/history would elucidate whether the reported health outcomes can be explained by underlying pathology. Cognitive and behavioral aspects could be complementarily included in the case definition, since evidence on their role in IEI-EMF is promising [18] but not yet established. Moreover, taking into account potentially harmful environmental agents other than EMF would be an important addition for research.

Table 4 Proposed case definition aspects for IEI-EMF
Full size table

This is the first time that a systematic review is conducted on definitions and identifying criteria for IEI-EMF. Given the large number of included articles, it is unlikely that any missing (or even excluded) studies would alter the results or increase any publication bias, especially since the aim of the current paper was not to assess etiologic associations.

It is a challenge how all the different case definition parameters for IEI-EMF can be concisely embodied in one international operational tool which could be used in research and clinical practice, and how this instrument could be adjusted to the possible cultural differences (e.g. in terms of wording/phrasing questions on health outcomes). Nevertheless, without the harmonization of the conceptual framework and validation of identifying criteria, the value of the case definition standards for IEI-EMF will remain insufficient and possibly unreliable. Apart from research, this has an important impact also in primary care; physicians, who are often the first to be contacted by the sufferers, are usually not adequately informed about IEI-EMF, which can affect the patient-doctor interaction and the management of the patient [26].

In order to properly construct an operational tool, a proposed two-phase approach can be briefly described as follows: In the first phase, a case definition and case selection tool should be developed, taking into account sources such as the published literature, expert opinions (e.g. based on a Delphi procedure [93]) and information on IEI-EMF patient characteristics from available datasets/ongoing research. At this stage, EMF measurements or provocation tests should not be a priority since a provocation study will only have added value after the formulation of a proper case definition and participant selection. Additionally, if the aim of a “case selection tool” is to routinely test cases where symptoms occur without a clear underlying pathology, then that tool should be concise, inexpensive and easy to implement, such as a short questionnaire or checklist. In the second phase, the case definition tool should be validated in terms of practical usability and the ability to differentiate between subgroups of IEI-EMF and patients with other conditions (e.g. chronic fatigue) who report similar symptoms. Based on the findings, the requirements for a follow up study could be outlined.

Conclusions

IEI-EMF is a poorly defined sensitivity. Heterogeneity and ambiguity of the existing definitions and criteria for IEI-EMF show the necessity to develop uniform criteria that will be applicable both in research and clinical practice. Broader criteria identified in the published literature such as attribution of NSPS to EMF and subjective report of being EMF sensitive could be used as a working definition for IEI-EMF which will serve as a basis for the development of a case selection tool. However, further optimization is required, testing its reliability and validity in several different patient groups, leading to an international multidisciplinary protocol with the active involvement of health care providers. This could also be a stepping stone for the harmonization of concepts and case definition for the broader condition of IEI.