ID Number		540
Study Type		Human / Provocation
Model		450 MHz (AM 7Hz) exposure to humans and analysis of cognitive function and memory
Details		Human volunteers (n = 50 exposed, 50 sham exposed students [63 male, 37 female]) were exposed to 450 MHz (AM 7Hz) for 10-20 minutes at a peak brain SAR of 0.351 W/kg using a portable mobile phone transmitter with a 1 watt output. A statistically significant increase in the number of mistakes in a test to measure corrective attention was found. The authors also report a statistically significant increase in errors in a test designed to measure divided attention short term memory, although this was not statistically significant by standard t-test, only by using an F-test for variance (mean error score = 8.6 +/- 9.2 exposed, 5.2 +/- 4.5 sham). A third test measuring visual short term memory also showed no statistically significant change in mental speed or errors using t-test analysis, but did show an increase in errors using an F-test for variances (mean error score = 7.1 +/- 5.1 exposed, 6.5 +/- 2.7 sham). The speed with which the subjects performed the tests was not changed to a statistically significant extent. The authors reported significant differences from person to person, and interestingly a greater effect of RF exposure on women than on men. In subsequent studies using a procedure they term visual masking, subjects [10 male, 6 female] exposed as above were provided visual stimuli and assessed for memory ability. The authors report statistically significant decrease with RF exposure, and conclude a weak effect of exposure using a 7 Hz modulation on early stages of visual information processing.
Findings		Effects
Status		Completed With Publication
Principal Investigator		Tallinn Technical Univ, Tallinn Estonia
Funding Agency		Sci Foundation, Estonia
Country		ESTONIA
References		Rodina, A et al. Bioelectromagnetics, (2005) 26:571-577 Lass, J et al. Int. J. Radiat. Biol., (2002) 78:937-944
Comments		The statistical analysis of the data is clearly in question. Although the abstract declares statistically significant increases in errors in the first and third cognitive function tests, these are not significant by standard t-test to test for differences in the mean values, only when an F-test to compare the variance between exposed and sham groups. The F-test simply compares the variances between groups (or, the +/- SD) and does not compare the actual mean values between the two experimental conditions. Also importantly, an F-test must be done between 2 normally distributed populations, and in the methods section under 'statistical analysis' the authors suggest that the data may not have always been normally distributed. In the discussion, the authors state: "the results of task 1 showed clear tendencies produced by the effects of modulated microwave radiation, but the statistical values of differences between the exposed and sham groups were not significant: therefore, further investigations are necessary" (how convenient). An increase in errors is further not supported by previous reports of cognitive function by Koivisto et al and Preece et al. No error bars are associated with the figures, and in them the groups seem to be stratified into smaller groups taking even more power away from the statistical analysis. There was finally a considerable inter-indivudual variation ascknowledged by the investigators. The mean error scores of the first and third tests are: (test 1: 8.6 +/- 9.2 exposed, 5.2 +/- 4.5 sham) and (test 2: 7.1 +/- 5.1 exposed, 6.5 +/- 2.7 sham). The +/- SD overlap considerably, suggesting the differences may be largely due to the effects of a few outlying individual test scores. In support of this, the authors report a significant inter-individual variation, although no raw data is given to see exactly. It is also interesting that related tests measuring a) corrective attention and b) divided attention short term memory showed an increase in errors, while a related test measuring visual short term memory showed a decrease in errors. This does not seem intuitively obvious to me - does this suggest some mechanistically complex effect on higher cognitive brain function areas ? The discussion glosses over this seeming discrepancy and states "... the reason probably lies in the different levels of cognitive processes involved during the tasks rather than in the opposite character of the radiation effect". [How about maybe experimental variation ?]