Heinrich et al recently published a meta-analysis on effects of static magnetic fields on cognition, vital signs, and sensory perception in this journal (1). Having been involved in the majority of studies on neurobehavioral effects considered in this meta-analysis, we would like to address a few issues.
An important basis for conducting a meta-analysis is that the individual studies included share an underlying common risk factor and endpoint, and that variations that result from clinical or methodological differences among studies are due to chance (2). However, Heinrich et al attempt to combine effects from exposure to (ultra)high, but spatially homogeneous static magnetic fields (SMFs) encountered inside the magnet bore with an order of magnitude lower, but spatially heterogeneous SMF surrounding magnetic resonance imaging (MRI) systems (stray fields); or with time-varying (electro)magnetic fields induced by movements in these stray fields. This issue has been discussed specifically for these studies previously (3). They have also included a nonexperimental study that assessed exposure during normal work practices of MR engineers (4). This study investigated whether neurobehavioral performance differed after a work shift compared to the start of the work shift to evaluate whether any measurable effects remained after exposure to the MRI-generated (electro)magnetic fields had ended. This is a distinctly different study design and hypothesis than for the other studies included by Heinrich et al, in which the effect of these fields were tested while the individuals were present in these fields. In other words, in this meta-analysis different exposures have been combined, while also effects experienced in the bore (by patients) are taken together with those that may be experienced by personnel working around MRI scanners.
By combining these studies without taking into account the different types, levels, and patterns of exposure, Heinrich et al assume that the magnitude of any effect is independent of type and level of exposure; a hypothesis that directly contradicts the conclusion from one of the studies also included in this meta-analysis (5), but also from a pooled analysis published in this journal in 2007 (6). This analysis pooled data from three case-crossover studies with comparable positioning of test subjects in the stray fields, but with MRI systems of different strengths. These analyses indicated exposure-effect associations for visuomotor speed and visual contrast sensitivity during exposure, with data from other sources indicating that these effects disappear quickly after exposure has ended (4, 7, 8).
Heinrich et al further present a list of recommendations that will be a useful framework for future studies in this area. However, to our surprise two important shortcomings identified in our work: 1) the use of a single-blind study design in which test administrators could not be blinded to the magnet status, and 2) the absence of personal exposure measurements during the studies to improve exposure assessment in such a heterogeneous SMF, were not mentioned and translated into recommendations for future studies. Furthermore, in contrast to their assessment which rightly suggests that exposure to SMF consistently affects tests of visual contrast sensitivity, they do not recommend a more thorough investigation of these visual impairments using appropriately sensitive, non-introspective, visual tests. Instead, they suggest new unguided explorations of more cognitive functions (even though they conclude data do not indicate consistent effects on such functions) and use of questionnaires on sensory perception, which is not likely to provide new information on neurobehavioral mechanisms not already known.
In conclusion, we do not believe these studies should have been combined in one meta-analysis, while in addition we recommend that Heinrich et al update their list of recommendations to incorporate the issues discussed here.