Article Text
Abstract
Objectives As more children use mobile (cellular) telephones, public anxiety grows about the possible adverse health effects of radiofrequency (RF) exposure upon developing nervous systems. Most epidemiological studies investigating the health effects of mobile telephones have relied on self-reports from questionnaires. While there are some validation studies investigating the accuracy of self-reported mobile phone use in adults and adolescents, self-reported laterality of use has not been validated at any age. Although this study mainly sought to validate the accuracy of self-reported laterality of mobile telephone use in adolescents, investigation also covered number and duration of calls.
Methods We monitored 455 calls in 30 students, mean age (SD) 14 (0.4) years. For 1 week, participants used hardware modified phones (HMPs) which logged dosimetric parameters such as laterality (side of head), date, number and duration of calls. These ‘gold standard’ measurements were compared with questionnaire self-reported laterality and estimated typical weekly phone use.
Results Agreement between HMPs and self-reported laterality was modest (κ=0.3, 95% CI 0.0 to 0.6). Concordance between HMP measured and self-reported number of calls was fair (intraclass correlation coefficient (ICC)=0.38, 95% CI 0.07 to 0.69), but poor for duration (ICC=0.01, 95% CI 0.00 to 0.37) with wide limits of agreement for both.
Conclusions These results suggest that adolescent self-reported laterality was of limited validity. Adolescent self-reported phone use by number and duration of calls was generally inaccurate but comparable to recent adult studies. Epidemiological studies of mobile phone use based on self-reported information may underestimate true associations with health effects.
- Laterality determination
- mobile phones
- exposure assessment
- epidemiology, adolescents
- environmental exposures
- epidemiology
- public health
- exposure assessment
- electromagnetic fields
- non-ionizing radiation
Statistics from Altmetric.com
- Laterality determination
- mobile phones
- exposure assessment
- epidemiology, adolescents
- environmental exposures
- epidemiology
- public health
- exposure assessment
- electromagnetic fields
- non-ionizing radiation
What this paper adds
Concerns about the vulnerability of certain population subgroups to radiofrequency radiation emitted by mobile telephone handsets and associated base stations have been investigated in epidemiological studies which mostly relied on self-reports for exposure assessment.
This is the first study to investigate direct measures of laterality determination at any age using hardware modified phones.
Researchers should be aware of the importance of laterality determination and its implications for exposure assessment in epidemiological studies investigating the health effects of mobile telephone use.
Importantly, future studies should be designed and conducted taking into account the limited recall of mobile telephone use by adolescents in terms of laterality, number and duration of calls.
We suggest validation studies nested within larger mobile phone epidemiological studies, particularly where questionnaires are applied for exposure assessment.
More people including children and adolescents use mobile (cellular) telephones, particularly in the last decade. However, this increase in mobile telephone uptake has generated increased anxiety and fears of possible adverse health effects associated with exposure to radiofrequency (RF) energy emitted by the handsets and associated base stations. Particular concerns have been expressed regarding the vulnerability of children and adolescents to electromagnetic fields.1 As children are not occupationally exposed, the primary source of RF exposure may be their use of mobile telephones.2
Associations between mobile phone use and adverse health effects have been extensively reviewed. These reviews identified poor exposure assessment as a common weakness limiting interpretation of all the epidemiological studies. They independently called for improved exposure assessment methods and instruments3–7 in mobile telephone health effects studies.
It has been suggested that the side of the head on which a mobile telephone is usually used (laterality) may be related to the location of brain tumours. While laterality has engaged neuroscientists for over a century,8 it has been poorly addressed by RF researchers. Some studies have found increased risk of tumour associated with ipsilateral (same side of head) phone use.9–11 Others have demonstrated associations between phone use and tumours on the contralateral (opposite) side of the head11 12 or similar risk on both sides.9 While some studies have found a protective effect on the opposite side of the head, others have found no difference based on laterality of mobile phone use and tumour location.12 13 A meta-analysis found an overall protective effect of mobile phone use.14 However, there appears to be consensus that most RF energy during conventional voice calls with a mobile telephone is experienced by the side of the head on which the phone is used.15–18
Most studies investigating associations between brain tumours and mobile phone use have been limited in two ways: first to adults and second by reliance on participant self-report via administration of questionnaires. Questionnaires yield data of limited precision particularly in mobile phone health effects research,19–21 prompting calls for population based validation studies.22 Some studies of adverse health outcomes in children and adolescents have been conducted,23–25 but most have relied on questionnaire-based self-reported information of mobile phone use.
There is as yet no published study validating questionnaire self-reports of mobile phone laterality in adolescents. It has been a common practise to ask about handedness of participants as well as the usual side of head of phone use. However, assuming handedness indicates laterality of mobile phone use may be misleading, as the relationship between handedness and laterality of phone use may not be perfect. Furthermore, the presence of a tumour may modify behaviour. In a recent Japanese study of mobile phone use and acoustic neuroma, it was found that 95% of cases had subjective symptoms such as hearing difficulty before diagnosis.12 Such hearing impairment may have caused them to switch the side of phone use. Another study of mobile phone use and risk of glioma26 found that the concordance between reported side of phone use and handedness was only 59% among cases and 64% among controls.
The aims of the present study were to determine laterality using hardware modified phones (HMPs) and to apply these novel devices in a group of adolescents to validate self-reported use of mobile telephones in terms of laterality, number and duration of calls. To achieve these aims, we conducted a laboratory HMP calibration study using a specific anthropomorphic mannequin (SAM) and a convenience sample of humans (reported elsewhere),27 and a population-based study involving early secondary school students which is the subject of this paper.
Methods
HMP technology
The HMPs employed in this study were modified GSM type handsets (Motorola model T280i; figure 1). The phones logged date and time of calls, number and duration of calls, transmit power levels during calls as well as orientation of phone to the head every 2 s. These phones incorporated three tilt sensors which were placed in mutually orthogonal planes and gave a unique combination of angles in every possible orientation of the phone. The recorded orientation was relative to the vertical. With an assumption that the phone user was mostly in the vertical position during phone use, it was possible to determine the laterality of mobile phone use based on orientation of the tilt sensors.
Calibration
Calibration to determine the laterality of mobile telephone use was performed using the DASY3 measuring system (Schmidt and Partner Engineering, Zürich, Switzerland) which was IEEE 1528b approved for specific absorption rate (SAR) measurements.
The phone was linked to a study computer via a data cable. The laboratory component of this study extensively assessed the laterality of phone use with the SAM phantom as well as a convenience sample of humans and conclusively showed that the HMP was capable of discriminating the side of head of phone use.27
Recruitment of participants
This validation study with HMPs was nested within the Mobile Radiofrequency Phone Exposed Users Study (MoRPhEUS)28 between 1 August and 30 November 2007 inclusive.
We recruited volunteers who were year 8 students, aged 13–15 years from across the three school sectors (government, Catholic and independent) in Melbourne, Australia who owned and were the sole users of their mobile phones. Full demographic information for participants is described elsewhere.28. Permission for students to participate was obtained from the administrative offices of the three school sectors. Parents/guardians and the students themselves gave informed consent, as is recommended.29 Approval for the population study was obtained from the Standing Committee on Ethics in Research Involving Humans (SCERH) at Monash University.
Population study design
Self-reported mobile phone use was assessed using a questionnaire adapted from the Interphone questionnaire16 to suit the age of participants. Relevant exposure questions are presented in appendix 1. Participants were asked their preferred side of mobile phone use and requested to estimate the average number and duration of calls they would make in a typical week. To avoid collusion and other possible external influences, the questionnaire was completed in class under relaxed test conditions supervised by a teacher and an investigator. Validation of self-reported phone use was undertaken using HMPs. Participants had their own SIM cards transferred to HMPs by the researchers and then used the HMPs for the 7 consecutive days following completion of the exposure questionnaire. Data from the HMP were automatically transferred to the researchers via email during routine battery charging. At the end of the 7-day monitoring period, the researchers transferred SIM cards back to participants' phones. Where participants did not exhaust their battery charge sufficiently to warrant recharging during the monitored period, all data were retrieved on return of the HMP.
Statistical analysis
Agreement between self-reported and HMP records of laterality of mobile phone use was assessed using the κ statistic with a 95% CI obtained using 1000 bootstrap resamples. Regarding the self-reported duration and number of mobile phone voice calls per week, distributions were skewed and these data were subjected to logarithmic transformations prior to analysis. The levels of agreement between self-reported and HMP measured number of voice calls and call duration were then assessed using the graphical method of Bland and Altman.30 Limits of agreement were calculated as mean difference±1.96 SD of differences on the log scale. These limits were transformed back to the original scale where they are interpreted as limits for the ratio of a person's HMP measured to self-reported values. Agreement between self-reported and HMP measured number and duration of calls was further assessed using the intraclass correlation coefficient (ICC).31 The strength of agreement was interpreted from κ statistics and ICCs according to Altman.32 The statistical package used for all analyses was Stata v 9.0.33
Results
Altogether 455 calls were monitored in 30 students (20 boys, 10 girls) whose ages ranged from 13 to 15 years with a mean (SD) age of 14 (0.4) years. Agreement between self-reported laterality of mobile phone use and HMP (‘gold standard’) determined laterality was fair (κ=0.3, 95% CI 0.0 to 0.6).
Laterality as assessed by HMP was not fully determined by self-reported handedness. Among 24 right-handed participants, half were predominantly right-side phone users, nine (38%) demonstrated regular use of both sides of the head for calls and the remaining three (12%) used phones predominantly on the left side. The six left-handed users were split equally into predominantly right, predominantly left and regular users of both sides (table 1).
The mean numbers (SD) of self-reported calls per week were 22.7 (41.6) and HMP measured 15.2 (10.4). The corresponding values for duration of calls were 18.4 (65.5) and 23.1 (22.4) for self-report and HMP measured, respectively. More details are given in table 2.
The agreement between self-reported and HMP measured numbers of calls and call duration is shown in figure 2A,B with corresponding Bland and Altman plots in figure 2c,d. Agreement with number of calls was fair between the two exposure methods (limits of agreement for ratio of HMP measured to self-reported number of calls: 0.14–13.5; ICC=0.38, 95% CI 0.07 to 0.69). However, agreement was poor for duration of calls (limits of agreement for ratio of HMP measured to self-reported duration of calls in minutes: 0.002–54.6; ICC=0.01, 95% CI 0.00 to 0.37).
These analyses found greater variability in self-reported average weekly number and average duration of calls than in HMP measured number and average duration of calls over 1 week. This can be seen in the Bland and Altman plots as a downward trend in the scatter of data points and indicates that the assumption of equal variance underlying the calculation of limits of agreement was not satisfied, that is our results may over-estimate the width of the limits of agreement.
Discussion
This study has demonstrated the feasibility of using HMPs to determine the laterality of mobile telephone use in adolescents. Self-reported laterality by questionnaire has also been compared with the HMP determined laterality as the ‘gold standard’.
This investigation is novel in several ways. This is the first study to determine laterality of mobile telephone use independent of the user. Laterality information was automatically transferred from the telephone, thus avoiding the reporting bias that is often associated with questionnaires. Here is also presented for the first time, the validity of self-reported mobile telephone use in a school based sample of adolescents.
Most validation studies of mobile telephone use have used convenience samples of adults.34–37 These studies mostly used billing and network operator records as ‘gold standards’ for validation purposes of mobile phone use. However, some of these studies experienced technical difficulties ranging from failure to provide for incoming calls37 and misclassification of phone users due to inability to discriminate between shared and sole phone use.20 There is also a further concern that network operators may be unwilling to divulge customer information for business reasons38 or technically unable to provide such information.36 Furthermore, none of these studies could provide information on laterality. The HMPs transferred data without the involvement of telephone network operators or the user. This also avoided loss of data that might result if phone memory overflowed before phones were returned to investigators.
The relationship between handedness and HMP measured laterality was fair but far from perfect, for example only half of the right-handed adolescents used the phone predominantly on the right side. The agreement between self-reported and HMP measured number of calls was fair, but the corresponding agreement for duration of calls was poor. Similar results for number and duration of calls have been observed in validation studies in adults34 36 where software modified phones (SMPs) were used. However a similar study from the same group utilised network operator records and found participants recalled duration of calls better than number of calls.35
A further concern involves the integrity of adolescent self-reports. A recent study found that the truthfulness and accuracy of adolescent self-reports may be genuinely compromised by the difficulty of such recall or intentionally manipulated depending on perceived social acceptability or otherwise of the behaviour under investigation.39
However, our finding of greater variability in self-reported average weekly number and average duration of calls than in HMP measured number and average duration of calls over 1 specific week is surprising from a statistical point of view, as averages should have less variability than single assessments. This surprising finding may suggest a problem either with the nature of the adolescent self-report or the experience of using the HMP for a week.
It is also possible that behaviours may have changed as a consequence of study participation, a situation often described as the ‘Hawthorne effect’.40 This effect applies in situations where the process of measurement influences the measured behaviour. This would only be applicable to the present study if use of HMPs significantly altered participants' routine use of mobile phones. Although this phenomenon is real in most human population and experimental studies,41 42 such an effect was less likely in the main section of this study given the direct data transfer to the researchers. For example, participants had no control over HMP measured laterality and it would seem highly unlikely that people would change the side of head of mobile phone use as a consequence of acquiring a new one (or using an older unfashionable one). Thus, it is reasonable to assume a minimal Hawthorne effect on laterality of mobile phone use in this study.
The results suggest that adolescent self-reported mobile phone use in terms of laterality, number and duration of calls is of limited validity and epidemiological studies based on adolescent self-reported mobile phone use may underestimate true associations with health effects.
Our study has some limitations. Although practical steps such as recruitment of only phone owners and exclusion of shared phone users were taken to avoid selection bias, the possibility of shared phone use on the outcome of this study cannot be ruled out. The nature of the HMP effectively excluded third generation (3G) phone owners. Furthermore, as some potentially eligible participants found the HMPs ‘uncool’, it is conceivable that adolescents who volunteered to swap their better looking phones for the HMP for the 7 consecutive days may differ from non-volunteers in their self-reported accuracy. Overall, the physical appearance of the present generation of HMPs affected adolescent participation negatively and resulted in the small sample size of the present study. To improve participation, especially among adolescent girls, the HMPs would benefit from redesign. While the sample size was modest, it was sufficient for 95% CIs for ICCs and κ statistics to rule out the possibility of excellent agreement.
While HMPs do hold promise as exposure assessment instruments in future epidemiological studies of mobile telephone use, the more widespread application of this technology could be limited by cost. Some mobile telephone manufacturers do now include limited HMP equivalent features such as inclinometers in specific handsets. This would allow for easy determination of laterality. The inclusion of such features in most models would be more cost effective than individual research groups attempting to develop their own HMP equivalent mobile phones.
Conclusions
The findings of the overall fair agreement between self-reported laterality and numbers of calls and the HMP measured records of these exposure metrics suggest caution in the interpretation of epidemiological studies investigating the health effects of mobile phones in general, but also specifically in adolescents. These results suggest that the laterality of phone use can be determined with HMPs in epidemiological studies of health effects. Further studies with larger sample sizes and participants with wider age ranges more representative of adolescents43–45 and other subgroups of the general population are needed to replicate and extend these findings.
Acknowledgments
The authors thank the teachers from participating schools, parents and students for taking part in this study. Our gratitude also goes to Drs Ken Joyner, Antonio Faraone and Giorgi Bit-Babik for their technical assistance. We acknowledge Nick Jandiery and Gia Saparishvili of Tbilisi State University, Georgia for their contributions to the design of the HMP.
Appendix 1 Relevant questionnaire based exposure indices explored in this study
References
Footnotes
Funding The Australian Centre for Radiofrequency Bioeffects Research (ACRBR) and Dr Benke are supported by the National Health and Medical Research Council of Australia.
Competing interests Michael Abramson holds shares in Telstra and SingTel which operate mobile telephone networks in Australia. He has also received research support from the GSM Association. Ray McKenzie is an employee of Telstra Corporation Limited, which operates mobile telephone networks in Australia. The other authors declare no competing interest.
Ethics approval This study was conducted with the approval of the Standing Committee on Ethics in Research Involving Humans (SCERH) at Monash University.
Provenance and peer review Not commissioned; externally peer reviewed.