The effect of exposure to radiofrequency electromagnetic fields on cognitive performance in human

Abstract

The effect of exposure to radiofrequency electromagnetic fields on cognitive performance in human experimental studies: Systematic review and meta-analyses (SR 6) Pophof B, Kuhne J, Schmid G, Weiser E, Dorn H, Henschenmacher B, Burns J, Danker-Hopfe H, Sauter C, Pophof B. The effect of exposure to radiofrequency electromagnetic fields on cognitive performance in human experimental studies: Systematic review and meta-analyses. Environment International (2024). doi: 10.1016/j.envint.2024.108899. Abstract Background The objective of this review is to evaluate the associations between short-term exposure to radiofrequency electromagnetic fields (RF-EMF) and cognitive performance in human experimental studies. Methods Online databases (PubMed, Embase, Scopus, Web of Science and EMF-Portal) were searched for studies that evaluated effects of exposure to RF-EMF on seven domains of cognitive performance in human experimental studies. The assessment of study quality was based on the Risk of Bias (RoB) tool developed by the Office of Health Assessment and Translation (OHAT). Random effects meta-analyses of Hedges’s g were conducted separately for accuracy- and speed-related performance measures of various cognitive domains, for which data from at least two studies were available. Finally, the certainty of evidence for each identified outcome was assessed according to Grading of Recommendations Assessment, Development, and Evaluation (GRADE). Results 57,543 records were identified and 76 studies (80 reports) met the inclusion criteria. The included 76 studies with 3846 participants, consisting of humans of different age, sex and health status from 19 countries, were conducted between 1989 and 2021. Quantitative data from 50 studies (52 reports) with 2433 participants were included into the meta-analyses. These studies were performed in 15 countries between 2001 and 2021. The majority of the included studies used head exposure with GSM 900 uplink. None of the meta-analyses observed a statistically significant effect of RF-EMF exposure compared to sham on cognitive performance as measured by the confidence interval surrounding the Hedges’s g or the significance of the z-statistic. For the domain Orientation and Attention, subclass Attention – Attentional Capacity RF-EMF exposure results in little to no difference in accuracy (Hedges’s g 0.024, 95 % CI [-0.10; 0.15], I2 = 28 %, 473 participants). For the domain Orientation and Attention, subclass Attention – Concentration / Focused Attention RF- EMF exposure results in little to no difference in speed (Hedges’s g 0.005, 95 % CI [-0.17; 0.18], I2 = 7 %, 132 participants) and probably results in little to no difference in accuracy; it does not reduce accuracy (Hedges’s g 0.097, 95 % CI [-0.05; 0.24], I2 = 0 %, 217 participants). For the domain Orientation and Attention, subclass Attention – Vigilance RF-EMF exposure probably results in little to no difference in speed and does not reduce speed (Hedges’s g 0.118, 95 % CI [-0.04; 0.28], I2 = 41 %, 247 participants) and results in little to no difference in accuracy (Hedges’s g 0.042, 95 % CI, [-0.09; 0.18], I2 = 0 %, 199 participants). For the domain Orientation and Attention, subclass Attention – Selective Attention RF-EMF exposure probably results in little to no difference in speed and does not reduce speed (Hedges’s g 0.080, 95 % CI [-0.09; 0.25], I2 = 63 %, 452 participants); it may result in little to no difference in accuracy, but it probably does not reduce accuracy (Hedges’s g 0.178, 95 % CI [-0.02; 0.38], I2 = 68 %, 480 participants). For the domain Orientation and Attention, subclass Attention – Divided Attention RF-EMF exposure results in little to no difference in speed (Hedges’s g −0.010, 95 % CI [-0.14; 0.12], I2 = 5 %, 307 participants) and may result in little to no difference in accuracy (Hedges’s g −0.089, 95 % CI [-0.35; 0.18], I2 = 53 %, 167 participants). For the domain Orientation and Attention, subclass Processing Speed − Simple Reaction Time Task RF- EMF exposure results in little to no difference in speed (Hedges’s g 0.069, 95 % CI [-0.02; +0.16], I2 = 29 %, 820 participants). For the domain Orientation and Attention, subclass Processing Speed – 2-Choice Reaction Time Task RF- EMF exposure results in little to no difference in speed (Hedges’s g −0.023, 95 % CI [-0.13; 0.08], I2 = 0 %, 401 participants), and may result in little to no difference in accuracy (Hedges’s g −0.063, 95 % CI [-0.38; 0.25], I2 = 63 %, 117 participants). For the domain Orientation and Attention, subclass Processing Speed – >2-Choice Reaction Time Task RF-EMF exposure results in little to no difference in speed (Hedges’s g −0.054, 95 % CI [-0.14; 0.03], I2 = 0 %, 544 participants) and probably results in little to no difference in accuracy (Hedges’s g −0.129, 95 % CI [-0.30; 0.04], I2 = 0 %, 131 participants). For the domain Orientation and Attention, subclass Processing Speed – Other Tasks RF-EMF exposure probably results in little to no difference in speed and does not reduce speed (Hedges’s g 0.067, 95 % CI [-0.12; 0.26], I2 = 38 %, 249 participants); it results in little to no difference in accuracy (Hedges’s g 0.036, 95 % CI [-0.08; 0.15], I2 = 0 %, 354 participants). For the domain Orientation and Attention, subclass Working Memory – n-back Task (0–3-back) we found Hedges’s g ranging from −0.090, 95 % CI [-0.18; 0.01] to 0.060, 95 % CI [-0.06; 0.18], all I2 = 0 %, 237 to 474 participants, and conclude that RF-EMF exposure results in little to no difference in both speed and accuracy. For the domain Orientation and Attention, subclass Working Memory – Mental Tracking RF-EMF exposure results in little to no difference in accuracy (Hedges’s g −0.047, 95 % [CI −0.15; 0.05], I2 = 0 %, 438 participants). For the domain Perception, subclass Visual and Auditory Perception RF-EMF exposure may result in little to no difference in speed (Hedges’s g −0.015, 95 % CI [-0.23; 0.195], I2 = 0 %, 84 participants) and probably results in little to no difference in accuracy (Hedges’s g 0.035, 95 % CI [-0.13; 0.199], I2 = 0 %, 137 participants). For the domain Memory, subclass Verbal and Visual Memory RF-EMF exposure probably results in little to no difference in speed and does not reduce speed (Hedges’s g −0.042, 95 % CI [-0.15; 0.21], I2 = 0 %, 102 participants); it may result in little to no difference in accuracy (Hedges’s g −0.087, 95 % CI [-0.38; 0.20], I2 = 85 %, 625 participants). For the domain Verbal Functions and Language Skills, subclass Verbal Expression, a meta-analysis was not possible because one of the two included studies did not provide numerical values. Results of both studies did not indicate statistically significant effects of RF-EMF exposure on both speed and accuracy. For the domain Construction and Motor Performance, subclass Motor Skills RF-EMF exposure may reduce speed, but the evidence is very uncertain (Hedges’s g −0.919, 95 % CI [-3.09; 1.26], I2 = 96 %, 42 participants); it probably results in little to no difference in accuracy and does not reduce accuracy (Hedges’s g 0.228, 95 % CI [-0.01; 0.46], I2 = 0 %, 109 participants). For the domain Concept Formation and Reasoning, subclass Reasoning RF-EMF exposure results in little to no difference in speed (Hedges’s g 0.010, 95 % CI [-0.11; 0.13], I2 = 0 %, 263 participants) and probably results in little to no difference in accuracy and does not reduce accuracy (Hedges’s g 0.051, 95 % CI [-0.14; 0.25], I2 = 0 %, 100 participants). For the domain Concept Formation and Reasoning, subclass Mathematical Procedures RF-EMF exposure results in little to no difference in speed (Hedges’s g 0.033, 95 % CI [-0.12; 0.18], I2 = 0 %, 168 participants) and may result in little to no difference in accuracy but probably does not reduce accuracy (Hedges’s g 0.232, 95 % CI [-0.12; +0.59], I2 = 86 %, 253 participants). For the domain Executive Functions there were no studies. Discussion Overall, the results from all domains and subclasses across their speed- and accuracy-related outcome measures according to GRADE provide high to low certainty of evidence that short-term RF- EMF exposure does not reduce cognitive performance in human experimental studies. For 16 out of 35 subdomains some uncertainty remains, because of limitations in the study quality, inconsistency in the results or imprecision of the combined effect size estimate. Future research should focus on construction and motor performance, elderly, and consideration of both sexes. Other This review was partially funded by the WHO radioprotection programme. The protocol for this review was registered in Prospero reg. no. CRD42021236168 and published in Environment International (Pophof et al. 2021). 4.9. Implications of practice and policy The present systematic review provides mostly moderate to high certainty of evidence that short-term RF-EMF exposure at SAR levels within the recommended limits (ICNIRP 2010) does not negatively affect the investigated domains of cognitive function. 4.10. Implications for research Although the number of studies and participants is low for several domains / subclasses, the certainty of evidence for lack of an effect is very low only for the speed category of D5 Construction and Motor Performance. Only two small studies with a total number of 42 participants contribute to the result with substantial heterogeneity and the body of evidence suggests a large negative effect with very low certainty. Further studies are needed to confirm or refute this effect. In particular, all outcome measures used in both studies should be considered and reported numerically. The number of studies that provide age-specific information on RF-EMF effects on cognitive performance is small. In particular, only five of the included studies were performed in children and only one in the elderly. Since children are generally considered to be a sensitive group, and the cognitive performance tends to decline with age, studies in children /adolescents and in elderly subjects are recommended to investigate whether these both age groups are more affected by RF-EMF exposure than young adults. Since the subgroup analyses (Supplementary Data 11) provide weak indications for a possible effect of sex, analyses stratified for sex of the participants are recommended. In general, any new study should be designed in a way that any potential RoB is minimized. The criteria for definitely low RoB are summarized in Supplementary Data 7 of the protocol (Pophof et al. 2021). Furthermore, power calculations should be conducted prior to the final study design and study power should be sufficient for detecting even small effects. A comprehensive list of further issues that should be considered when conducting studies on possible effects of RF-EMF on cognitive performance, are published in Regel and Achermann (2011). Authors of future studies are encouraged to make the original individual data publicly available in appropriate data repositories to avoid missing data due to incomplete reporting. 5.2. Deviations from the protocol In the protocol, we did not describe how the ratings of the individual RoB questions for a given study would be used to define an overall quality rating for that study. This admission could be considered as a potential risk of bias for this systematic review if, for example, certain studies were selectively included/excluded from subgroup analyses based on the quality rating. However, given that the heterogeneity of effects across studies was low for most outcome measures, such selective analysis would in principle have little to no impact. The GRADE assessment, however, strongly depends on the overall study quality assessment. Thus, in order to be as transparent as possible, we have explained and justified the chosen method for allocating studies to quality tiers in detail in Methods − Risk of Bias assessment. We did not conduct separate analyses for individuals with and without IEI-EMF, because there were only seven studies including subjects with IEI. Only six of the seven studies provided numerical data (for only some domains / subclasses) to be used to calculate effect sizes. The data of these six studies do not allow a detailed analysis following the approach pursued in this paper. Because there was only one study investigating females, heterogeneity could not be assessed for this subgroup. Therefore, we performed subgroup analyses stratified by studies investigating only males, and studies investigating males and females together. However, the explanatory power of this approach should not be overestimated. We did not evaluate a potential dose–response relationship or perform subgroup analyses regarding exposure levels, because independently from the exposure level there was predominantly no effect of RF-EMF, and the exposure assessment in most studies did not provide spatially resolved exposure levels in the brain. For similar reasons we did not analyse the exposure category D separately from A, B and C (see Protocol, Pophof et al. 2021), but combined all exposure categories in analyses. Due to the huge amount of data available for the domains D1 and to a lesser extent for D6, and due to the fact that some studies reported results for a large number of different subclasses of domains, a problem with the unit of analysis would have resulted in a huge amount of original data that could not be used for a meta-analysis if we had kept the outcome categories defined in the protocol. Therefore, we defined more detailed outcome categories according to the domain subclasses (Table 2) as defined by Lezak et al. (2012, Part II The Compendium of Tests and Assessment Techniques, p. 391 ff.). This enables a better differentiation of different cognitive abilities, and increases the number of possible meta-analyses and a higher proportion of original data included in the meta-analyses. In order to explore possible sources of heterogeneity, we additionally included a stratification according to sample sizes (<30; ≥ 30). Studies with small sample sizes are likely to have a higher variability than studies with larger sample sizes. Deviating from the protocol, we did not conduct subgroup analyses for the exposure-related categories of signals with different modulation, time course of exposure, different frequencies, for the nine outcome measures of subclasses with results with at least moderate heterogeneity (Table 3). This is because a large number of studies falls in the category “continuous GSM-modulated near-field exposure with frequency lower than 1 GHz” (in six out of these nine outcome measures at least half of results include this exposure category) and most of the outlier studies that strongly contribute to heterogeneity of the results also fall into this category (five studies). 5.3. Support This review was partly funded by the World Health Organization (contracts 2020/1028371–0, 2022/1276784–0). WHO provided the basis for the protocol and methodological support throughout the review process. Declaration of competing interest The authors declare the following financial interests/personal relationships which may be considered as potential competing interests: [The authors declare that they have no known competing financial interests or personal relationships that could have appeared to influence the work reported in this paper. BP is member of the ICNIRP Scientific Expert Group (SEG) on environment, BfS observer in the working group SSK-A630 of the German Commission on Radiological Protection and was German delegate of European Cost Actions BM0704 and BM1309 “EMF-MED”. GS is member of the Committee “Non-Ionizing Radiation” (SSK-A6) and member of the working group SSK-A630 of the German Commission on Radiological Protection. GS is chair of the Austrian Standardization Sub-Committee TSK- EMV-EMF “Electromagnetic Fields”. HDHs research is entirely funded by public or not-for-profit foundations. She has served as advisor to a number of national and international public advisory groups concerning the potential health effects of exposure to non-ionizing radiation, including the World Health Organization, the German Commission on Radiological Protection (member of the committee “Non- Ionizing Radiation” (SSK-A6) and member of the working group 5G (SSK-A630)) and the Independent Expert Group of the Swedish Radiation Safety Authority. JK is member of the ICNIRP Scientific Expert Group (SEG) on ultrasound and BfS observer in the working group SSK-A630 of the German Commission on Radiological Protection]. Open access paper: sciencedirect.com

AI evidence extraction

Main findings

Across meta-analyses, the review reports no statistically significant effects of short-term RF-EMF exposure versus sham on cognitive performance, based on confidence intervals around Hedges’s g and z-statistics. GRADE assessments were described as mostly moderate to high certainty that short-term RF-EMF exposure at SAR levels within recommended limits (ICNIRP 2010) does not negatively affect investigated cognitive domains, with remaining uncertainty in some subdomains (notably construction/motor speed based on two small heterogeneous studies).

Outcomes measured

• Cognitive performance (accuracy and speed measures) across multiple domains, including orientation/attention, processing speed, working memory, perception, memory, verbal functions/language, construction and motor performance, concept formation/reasoning

Limitations

• Some subdomains had low numbers of studies/participants, leading to imprecision.
• Uncertainty remained for 16 of 35 subdomains due to study quality limitations, inconsistency, or imprecision.
• Construction and motor performance (speed) relied on only two small studies with substantial heterogeneity and very low certainty.
• Few studies in children (five) and elderly (one), limiting age-specific conclusions.
• Potential protocol deviation regarding how individual risk-of-bias ratings were combined into overall quality tiers (noted by authors).
• Meta-analysis not possible for verbal expression due to missing numerical reporting in one study.

Suggested hubs

• who-icnirp (0.62)
  The review discusses SAR levels within ICNIRP 2010 recommended limits and notes partial funding by the WHO radioprotection programme.

{
    "publication_year": 2024,
    "study_type": "systematic_review",
    "exposure": {
        "band": "RF",
        "source": "experimental (primarily head exposure; majority GSM 900 uplink)",
        "frequency_mhz": 900,
        "sar_wkg": null,
        "duration": "short-term"
    },
    "population": "Humans of different age, sex and health status (experimental studies; multiple countries).",
    "sample_size": 3846,
    "outcomes": [
        "Cognitive performance (accuracy and speed measures) across multiple domains, including orientation/attention, processing speed, working memory, perception, memory, verbal functions/language, construction and motor performance, concept formation/reasoning"
    ],
    "main_findings": "Across meta-analyses, the review reports no statistically significant effects of short-term RF-EMF exposure versus sham on cognitive performance, based on confidence intervals around Hedges’s g and z-statistics. GRADE assessments were described as mostly moderate to high certainty that short-term RF-EMF exposure at SAR levels within recommended limits (ICNIRP 2010) does not negatively affect investigated cognitive domains, with remaining uncertainty in some subdomains (notably construction/motor speed based on two small heterogeneous studies).",
    "effect_direction": "no_effect",
    "limitations": [
        "Some subdomains had low numbers of studies/participants, leading to imprecision.",
        "Uncertainty remained for 16 of 35 subdomains due to study quality limitations, inconsistency, or imprecision.",
        "Construction and motor performance (speed) relied on only two small studies with substantial heterogeneity and very low certainty.",
        "Few studies in children (five) and elderly (one), limiting age-specific conclusions.",
        "Potential protocol deviation regarding how individual risk-of-bias ratings were combined into overall quality tiers (noted by authors).",
        "Meta-analysis not possible for verbal expression due to missing numerical reporting in one study."
    ],
    "evidence_strength": "high",
    "confidence": 0.85999999999999998667732370449812151491641998291015625,
    "peer_reviewed_likely": "yes",
    "stance": "reassurance",
    "stance_confidence": 0.83999999999999996891375531049561686813831329345703125,
    "summary": "This systematic review and meta-analyses evaluated short-term RF-EMF exposure and cognitive performance in human experimental studies. Across meta-analyses, no statistically significant differences were observed between RF-EMF exposure and sham for accuracy or speed outcomes in the assessed cognitive domains. The authors conclude that, mostly with moderate to high certainty (GRADE), short-term RF-EMF exposure within recommended SAR limits does not reduce cognitive performance, while noting uncertainty for some subdomains (especially motor speed).",
    "key_points": [
        "The review searched multiple databases and included 76 studies (80 reports) conducted between 1989 and 2021.",
        "A total of 3846 participants were included across studies, with 2433 participants contributing quantitative data to meta-analyses.",
        "Most included studies used head exposure, with GSM 900 uplink being the most common exposure condition.",
        "None of the meta-analyses found statistically significant effects of RF-EMF exposure versus sham on cognitive performance outcomes.",
        "GRADE assessments were described as mostly moderate to high certainty that short-term RF-EMF exposure does not reduce cognitive performance within investigated domains.",
        "Evidence for construction and motor performance (speed) was described as very uncertain and based on two small, highly heterogeneous studies.",
        "The authors highlight limited evidence in children and elderly and recommend more age- and sex-stratified research with minimized risk of bias."
    ],
    "categories": [
        "RF-EMF",
        "Cognition & Neurobehavior",
        "Systematic Reviews & Meta-analyses",
        "Human Experimental Studies"
    ],
    "tags": [
        "Systematic Review",
        "Meta-Analysis",
        "Radiofrequency EMF",
        "Short-Term Exposure",
        "Cognitive Performance",
        "Attention",
        "Processing Speed",
        "Working Memory",
        "Reaction Time",
        "Sham Exposure",
        "GSM 900",
        "Head Exposure",
        "Risk Of Bias",
        "GRADE",
        "ICNIRP Limits"
    ],
    "keywords": [
        "RF-EMF",
        "radiofrequency electromagnetic fields",
        "cognitive performance",
        "human experimental studies",
        "systematic review",
        "meta-analysis",
        "Hedges's g",
        "OHAT risk of bias",
        "GRADE",
        "GSM 900",
        "sham"
    ],
    "suggested_hubs": [
        {
            "slug": "who-icnirp",
            "weight": 0.61999999999999999555910790149937383830547332763671875,
            "reason": "The review discusses SAR levels within ICNIRP 2010 recommended limits and notes partial funding by the WHO radioprotection programme."
        }
    ],
    "social": {
        "tweet": "Systematic review/meta-analyses (76 human experimental studies; 3,846 participants) found no statistically significant effects of short-term RF-EMF exposure vs sham on cognitive performance across multiple domains; most evidence rated moderate–high certainty, with uncertainty for motor speed outcomes.",
        "facebook": "A 2024 systematic review and meta-analyses of human experimental studies (76 studies; 3,846 participants) reports no statistically significant differences in cognitive performance between short-term RF-EMF exposure and sham across multiple cognitive domains. The authors rate most evidence as moderate to high certainty for no reduction in performance within recommended SAR limits, while noting uncertainty for some subdomains (especially motor speed).",
        "linkedin": "Environment International (2024) systematic review/meta-analyses of short-term RF-EMF exposure and cognition in human experimental studies (76 studies; 3,846 participants) reports no statistically significant effects vs sham across assessed cognitive domains. GRADE ratings were mostly moderate–high certainty for no reduction in cognitive performance within recommended SAR limits, with remaining uncertainty in select subdomains (notably construction/motor speed)."
    }
}

Comments

Log in to comment.

No comments yet.