The lack of knowledge regarding the use of wireless telephones among children, and recent trends that seem to show increasing frequency of mobile phone use, emphasize the need for more research on potential long-term health effects and the close monitoring of user habits. The main aim of this population-based study was to assess ownership and use of mobile phones and DECT. The results showed that mobile phone use started early in life and, as expected, the GSM phone was the most commonly-owned and used type. In total, 95.0% of the children reported access to either a mobile phone or a DECT, or both.
The results also showed a rapid increase in the number of regular users with age. Of those who reported access to a mobile phone among the 14-year-olds, 44.9% (n = 79) were regular users; they talked for 2 min or more per day. Overall, 26.7% (n = 292) of the children were regular users. The reported use of hands-free equipment varied across age groups but was generally very low; 14-year-olds reported the most frequent use.
We found age and gender differences: both mobile phone and DECT use increased with age and was more common among girls (Table 1). Girls also sent more SMS than boys did. Almost half the children aged 7–8 years reported access to a mobile phone, but fewer than 10% of those were regular users. Mobile phone access increased with age and almost all children aged 14 years (98.3%) had access, though as mentioned only 44.9% of those were regular users. This probably indicates that among the younger children, mobile phones were mainly used for purposes other than talking just for fun, such as expedient contact with parents in an emergency.
In Table 2, the unconditional logistic regression analysis adjusted for age, sex and household income revealed further statistically significant factors, which predicted mobile phone access and regular use. When we compared the factors relevant to access with regular use, some results differed. The most sparsely populated region (H6) gave OR = 2.0, 95% CI = 0.9–4.6 for mobile phone access, compared with OR = 0.5, 95% CI = 0.2–0.9 for regular use. There is no obvious explanation for these results, but perhaps parents' desire to be able to contact their children is greater in sparsely populated areas, for instance because distances between home and school are long. Increasing time spent playing computer games increased the OR for mobile phone access but not for regular use, which may be due to social factors. Insufficient sleep was not associated with access but gave increased OR for regular use. Some of these results, however, were based on low numbers.
Factors that predicted regular mobile phone use were watching TV, use of DECT, regular physical activity, sleeping less than 7 hours and insufficient sleep. Thus, these variables seemed to be correlated with each other as life-style factors. In contrast, living in a sparsely populated area, having siblings and playing computer games were factors that predicted less mobile phone use. As shown in Table 2, children whose guardians reported below-average income were less likely to have access to a mobile phone, but those in this group who did report access tended to use their mobile phones more regularly than those who reported average income. This finding was based on low numbers but similar results were obtained in a Finnish study that looked at reported mobile phone use among adolescents in relation to guardians' education . We also found in our study that children whose guardians reported above-average income tended to report more mobile phone access and ownership (data not shown) than those who reported average income. The reasons for these findings – if they are real and not due to chance – are unclear, but one hypothesis could be that it is easier for parents with above-average incomes to gain access to and provide their child with a mobile phone, while they are also more aware of potential health risks and therefore exert more control over the child's use. On the other hand, the latter is contradicted by that fact that so few used hands-free equipment.
Some of our results differ from those of the German study , one of the two existing similar studies [11, 12], in that the proportions of mobile phone owners and users were higher for socially disadvantaged children. However, we used below-average income as a marker of socioeconomic status while in the German study this was based on a schoolteacher's judgement , so the results might not be comparable. In contrast to the German study, our study and the Hungarian one  also found that the boys were less likely than girls to be regular mobile phone users.
There was some similarity in the results regarding factors that could explain regular mobile phone use, e.g. time spent watching TV, physical activity and sleep habits. However, our study is only partially comparable with the other two because of differences in design. For example, the German and Hungarian studies used the same questionnaire and surveyed mostly 9-11-year-old children, from primary schools in one city in Germany and three cities in Hungary. Our study was population-based and included a sample of children from the whole of Sweden.
The aim of this study was not to explore explanatory factors in detail. For that, in-depth interviews would be necessary. For example, regular physical activity of more than 14 hours/week was significantly associated with regular mobile phone use. However, because regular physical activity was defined by number of hours/week regardless of type of physical activity and the analysis was based on low numbers, this finding must be interpreted with caution.
There are several limitations to our study that might have affected the results. One potential bias could be a sex difference in the 7–14 year age group in Sweden, since equal numbers of boys and girls were drawn from the Population Registry. In 2005 there were 51.3% boys and 48.7% girls according to the Population Registry , compared to 49.8% boys and 50.2% girls among the respondents in our study. However, this difference is not statistically significant (P = 0.28). At the end of 2005, the number of children in Sweden was unequally distributed among the 7–14 year age groups: there were somewhat more children in the older than in the younger groups. Compared with the distribution among the respondents in our study, the difference was significant (P = 0.01). Since use of wireless phones increased with age, the overall results may therefore be underestimates. However, we also present age-specific results that are not biased in this respect.
The participation rate was 71.2%, so we cannot exclude the possibility that response bias might have influenced some of the findings. Therefore we compared reported mobile phone access and use by early responders with responders who were sent at least one reminder. While there was no difference with regard to access (P = 0.95), there was difference for use (P = 0.01): the responders who were sent at least one reminder used their mobile phone more than early responders. This could also mean that reported use was underestimated in this study.
We also saw differences between girls and boys in use of both mobile phone and DECT. If this was due to bias, then the girls either over-reported their use, and/or regular users among the boys under-reported their use or did not answer the questionnaire. However, there were no differences between girls and boys either in response rate or in the distinction between early responders and those who were sent at least one reminder (P = 0.94).
Another possible source of response bias is that several questions might have been difficult for young children to answer. We did address the letter of information to the guardian and stressed the importance of assisting the child in answering the questionnaire. Unfortunately, no question was included to show whether the child and the guardian together, or only one of them, had answered the questionnaire. This could also have altered the results. However, if the questions were too complicated for the younger children to answer, then we would expect there to be an age difference between early and late responders, so that more young children would only have answered after at least one reminder. We could see no such significant difference (P = 0.18).
Although we did assess use of hands-free equipment, we could also have asked questions regarding other means of reducing unnecessary exposure such as talking only when the mobile phone has a good communication with the base station. Another weakness of this study concerns the durability of the data. Since this was a prevalence study it only gave information about average mobile phone use (for example) during a limited time period. It did not cover changes in use over a longer time period. However, other findings of this study such as the differences between girls and boys in both mobile phone and DECT use are not as likely to be time-related.
Finally, some comments should be made about the likely validity and reliability of the estimates of wireless phone use in this study, and also about our choice of method. Other sources of information on mobile phone and DECT use would have been valuable for comparison. However, no such information about DECT use was available among the telephone operators, since no information on the type of telephone used was registered. Similar problems are related to mobile phone use, since use of pay-as-you-call cards is likely to be common. Several children do not have their own mobile phone, especially the younger ones, and billing records give no information about incoming calls. Contacts with telephone operators have shown that it is not easy to obtain information about mobile phone use over time owing to the structure of their data. It is therefore difficult to say how reliable the estimates of use in this study are.
Ideally, in future studies, comparing the estimates with the true cumulative times of incoming and outgoing calls should test their reliability. This can be done using specially software-modified phones handed out to a sample of the respondents. Preferably these phones should not only record the number of calls but also the output power during operation. Such devices exist and have been used e.g. for validating self-reported mobile phone use in the Interphone Study . Regarding validity, we checked for possible bias in the analysis and found that some results could have been somewhat underestimated, e.g. overall use of mobile phones. A strength of this study, however, was its population-based design which, if not flawed with systematic errors, makes the results representative of all Swedish children aged 7–14 years.