The effect of exposure to radiofrequency electromagnetic fields on cognitive performance in human experimental studies: Systematic review and meta-analyses

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; 0.23], 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).