Salivary Melatonin levels were unaffected by the short term use of a GSM cellular phone. 1hls result supported findings reported by Mann et al. (1998), de Seze et al. (1998) and Radon (2001) in human studies and Vollrath et al. (1997)
who studied the effects of 900 Wiz EJ\1Fs on rats. The studies by Radon and
Mann differed from the current study in that both used antennae located a distance from the subject to negate a possible heating effect (10 cm behind a seated subject, and 40 cm from the vertex of a supine subject, respectively). Such a placing of the antennae would result in a different signal intensity than that experienced by a person using a cellular phone placed against the ear. The experimenters also used a shielded chamber which prevented the reflection of signals that commonly occur and thus the conditions did not model normal cell phone use. The present study more closely reflected normal cellphone usage and had a larger number of subjects. However, the study may not have been large enough. Analysis of the results suggested a sample of 67 subjects was required to reach a t-value of 2 at p=O.OS. Although the study had to be truncated at 43 , it was still the largest study so far.
In the present study, there was no field effect found in the subgroup identified as 'naturally low excretors'. However, there were only 13 subjects in the group. This may be too small to detect an effect so the possibility of a type 2 error is high. This is the first study to assess this in RF EJ\1Fs. In power frequency
studies, two groups (Graham et al., 1996 and Crasson et al, 2001) had reported suppressed melatonin levels on exposure to EMFs in subjects with naturally low melatonin levels.
Analysis of the results by gender produced an interesting result. Females had been excluded from prior published research on the basis that variations in melatonin with changing phases of the menstrual cycle may add a confounding variable. In the present study, there was no significant difference between control and exposure sessions for the female subjects but there was a significant decline in melatonin levels in the males. However, there were only 16 males in
the study, compared with 27 females. Females showed almost no difference between control and exposure sessions (mean difference= -0.50 pg/ ml)
whereas male subjects had a -1 .77 pg / rnl mean difference. The results suggest there is no reason to exclude females from being subjects. This was also the conclusion for the 50 Hz studies in experiments one and two. The significant difference for males was probably due to one higher score distorting the results. A similar score occurred in the females but the effect was reduced by the larger number of subjects (Appendix 3.5).
Four subjects had zero or close to zero levels of melatonin. A prior screening would have seen these subjects eliminated from the study but would have increased the cost. Inclusion of the subjects did not alter the results. As
explained in experiment two, such results are not uncommon and studies often report excluding subjects in the pre-screening selection on the basis of non standard profiles, or undetectable melatonin levels. For example, Radon et al.
(2001) exposed 8 males to a 900 MHz signal pulsed at 217 Hz. They observed the effects on salivary melatonin levels. Subjects were preselected on the basis of having melatonin rhythms that showed the least intra-individual variation. This may help in removing possible coniounders but may also select out the very subjects who may be most likely to show an effect as they are likely to be the most sensitive to environmental stimuli affecting melatonin.
An exposure time of 15 minutes may not be long enough to effect melatonin
levels. Lewy et al. (1980) found exposure to bright light caused a suppression of melatonin well within 30 minutes. If an EMF effect used the same
mechanism then it could take longer than the 15 minutes used in the present study.
The present study measured melatonin levels in saliva. The concentration of melatonin in saliva is around 30% that of serum. Thus a study using this
method may not be as sensitive to small changes in melatonin levels as a study using serum. Radon (2001) also measured melatonin levels in saliva. Using a specially designed shielded chamber , in seated subjects, the melatonin levels
reported were very similar to the present study, rising from 12 pg/ml at 2200
hours to about 28 pg I ml at midnight.
Stevens (1987) proposed a melatonin hypothesis in which an EMF-induced
Research on humans, including the present study, has not supported this hypothesis. However, the number of studies published is very small and all looked at acute effects. Research is required on the possible long term effects of cellular phone use on melatonin. Digital phones have been in common use for only a short period of time. Most of the subjects used in this study had little prior exposure to cellular phone frequencies. It should now be possible for studies to begin looking for effects from longer term use.
Cognitive Parameters
Attention
The use of a digital cellular phone produced a significant drop in levels of attention. The result was due to an increase in time taken, rather than an increase in errors as the mean number of errors was 1 .4 per subject for both the exposure and control conditions. This could have implications for activities such as the use of cellular phones whilst driving or operating machinery. However, research in this area is sparse and the study would need to be replicated before any firm conclusions could be drawn.
The only studies using a similar test in the literature were in the 50 Hz
frequency range (Beale et al., 1997; Crasson et al., 1999). Both reported null results. No effect was found for this task in experiments one and two but the frequency was different and the exposure time was shorter as the task was done at the commencement of exposure. In experiment 3, the test was done at the end of the experiment after 15 minutes of exposure.
Attention was the only parameter, apart from temperature, to produce a strongly significant result. It is interesting to note that the pattern of the output of the phone during this task, was different to that for other cognitive tasks as this task was done in silence (Figures 7.2, 7.3). The difference in wave pattern may be significant, in that the brain may be more susceptible to that particular wave pattern, or it may be a coincidence. The statistically significant result for attention may need to be treated with caution as it would be expected to get 1
in 20 results sign by chance at p=O.OS. However, it would still be significant if a
Bonferroni correction was applied. As the cognitive parameters and
physiological measurements are likely to be independent of each other it is debateable whether this correction is necessary.
The results found in the present study could be due to heating effects as the phone heats up considerably in use. It could also be a result of direct simulation by the EMF of the orbito-frontal areas, inferior parietal cortex or medial
temporal cortex. A further study could be designed to answer this question with the cell phone replaced during the control sessions with an identical case containing a heating system that did not require an EMF. If the difference still
remained then it could be reasonably assumed that the EMF was not the cause
of the decline in attention and it was simply due to the application of heat to the head.