Frequent mobile phone use was associated with current stress, sleep disturbances, and symptoms of depression among the young adult men and women in cross-sectional analysis. Prospective analysis indicated that high frequency of mobile phone use could be a risk factor (or marker) for developing sleep disturbances in the men, and symptoms of depression in both the men and women, at 1-year follow-up. The pattern of PRs larger than 1.0 was rather consistent (though not all statistically significant), suggesting a robustness of results, and there was even an indication towards a dose-response relationship between exposure and mental health outcomes (if looking only at PRs). It should be noted that the "high" category of mobile phone use in our study does not reflect an extreme part of the population, since almost 25% of the study group belonged to this category. The use of the Cox regression procedure for estimating PRs gives wider than adequate CIs , which was corrected for in the cross-sectional analysis by adding the robust variance option. However, in the prospective analysis the CIs are still conservative. The results are further supported by the finding of prospective associations between high frequency of mobile phone use and mental health outcomes in our previous study among young adult university students .
The majority of the young adults reported that they were expected to be reachable via the mobile phone all day or around the clock. One could expect that this would feel compelling and perhaps even stressful, but most respondents did not consider the accessibility to be stressful, and there was no association between the two variables. Yet, expected availability around the clock was associated with most mental health outcomes in cross-sectional analysis (no clear prospective associations). The risk for reporting mental health symptoms at follow-up was greatest among those respondents who had indicated that they perceived the accessibility to be rather or very stressful, and in cross-sectional analysis, it was even sufficient to consider the accessibility to be just a little stressful for higher prevalence of mental health outcomes. The over-all low associations between the mobile phone variables suggest that availability demands and accessibility stress not necessarily coincide with actual frequency of use.
Reports in the media claim nightly disturbances by mobile phone calls or messages to be a menace for today's adolescents. This may be the case among younger persons, but was not as obvious in our group of young adults, with only few being woken up regularly. However, there were cross-sectional associations between being awakened a few times or more during the past month and all mental health outcomes (no clear prospective effect).
It has been suggested that mobile phone use enhances social support [12, 37], but, in our study, high frequency of use had little or no association with perceived access to social support in private life.
Quite a few participants reported subjective overuse which could indicate possible addiction to the mobile phone or its functions. Addictions can consist of excessive behaviors of all types, and some factors can be argued to be present in all types of addictions (e.g., salience, tolerance, withdrawal, conflict, and relapse) . The most common symptom of problem mobile phone use among adolescents in a study by Yen et al  was "withdrawal symptoms without cellular phone use". Furthermore, impulsivity, especially urgency, has been related to mobile phone dependency, and feeling compelled to provide for needs as soon as possible has been suggested to increase the likelihood of using the mobile phone in a destructive way, for example when prohibited . There is also the risk for addiction through gambling on mobile phones , which could be detrimental since the mobile phone enables gambling without time or space restrictions.
We know little about what time span may be relevant when assessing possible effects of the exposure on mental health, and whether concurrent, short-term, or long-term exposure and effects are of interest. We have data from baseline and follow-up after 1 year, making it possible only to perform either cross-sectional analysis (so that causal inferences cannot be made) or prospective analysis with a 1-year latency period that could be considered rather long. The exposure during the latency period is not known, and the same applies to the mental health outcomes, concerning symptoms that are common in the population and that could appear and disappear in the latency period. Consequently, it is difficult to draw clear inferences about the effect of the exposure on the outcomes within the study design.
Using a questionnaire to collect information on exposure as well as health aspects poses several limitations. It is important to emphasize that the study concerns subjective symptom-reports and not actual mental disorders or diagnoses. The prevalence of reported depressive symptoms was alarmingly high in our study group. The suggested procedure that it is sufficient if one of the two PRIME-MD depressive items is confirmed in screening for depression [32, 33] proposes that about 20% of the study group would be clinically depressed (positive predictive value of 33% ). The prevalence of depression is most likely lower in our population than in primary care populations as, for example, the 1-month prevalence of depression among Finnish young adults (20-24 years of age) was 9.6% . Hence, the instrument seems too sensitive for our population, and we chose to analyze one-item and two-item responses as separate outcomes, with the expectation that the two-item outcome has higher specificity than the suggested procedure.
Recall bias and recall difficulties are most certainly present in the study, with, for example, difficulties to correctly specify the average number of calls and messages sent and received per day over the past month. Furthermore, when merging calls and SMS messages into one variable (mobile phone use) we lose information about specific exposure. Also, while the high and low categories are distinct from each other, the medium category overlaps to some extent with the high and low categories, which means that, in some instances, individuals in the medium category may in fact have had a higher exposure (number of calls and SMS messages) than some individuals in the high category, or lower than some in the low category. There is a risk that misclassifications obscure results.
We have limited our study to psychosocial aspects of mobile phone use. Possible biophysical pathways due to exposure to electromagnetic fields have not been considered. Furthermore, there might be factors, e.g. individual factors or personality traits, not accounted for in our study, which co-varies with exposure variables and are "true" pathways to mental health problems. This could particularly be the case concerning accessibility stress which had no association with availability demands and low association with actual frequency of use, but yet seemed to be the greatest risk factor among the mobile phone variables for developing mental health symptoms.
The study suffered from a high drop-out rate, which is fairly common when performing studies via questionnaires in the general population. The young adult population is probably especially difficult to recruit because more often than in another age group, their life situation undergoes drastic changes, including moving more often and therefore being more difficult to reach. The drop-out analysis shows that especially women and native-born Swedes are overrepresented in the data. Earlier studies, e.g. [13, 14, 21], have indicated gender differences in mobile phone usage, therefore gender-specific analyses were performed. However, the results of the analyses were strikingly similar for men and women in the present study. There is probably a healthy participant selection bias, and there is also an indication of bias towards lower mobile phone exposure, which could affect results in cross-sectional analyses but should have less influence in the prospective analyses. Even though the study group is more representative in comparison to studies among only college and university students, caution must be used when generalizing the results to the general population of young adults.
The place of mobile phones as a technology distinct from landline phones on the one hand, and from computers on the other, is declining, as mobile phones increasingly are taking the place of stationary phones and at the same time are approaching computers in function. Therefore, defining the exposure becomes difficult as technology and possible uses are developing and changing rather swiftly. The use of mobile phones puts high demands on the individual's own capacity to set limits for use and accessibility. Norms on how to use mobile phones are set in interaction with others. If a young person thinks that "all others" are available at all times, he/she might feel stress if not available. Attitudes are probably an important factor to focus in prevention strategies. This could include information to children, adolescents, and young adults about the importance of sleep and recovery, and the advice to set limits for accessibility (i.e., turn off the phone) at certain times such as at nighttime, when needing to focus or rest, or when others need to focus or rest. Furthermore, shifts in attitude could also include limiting your demands and expectations on others' availability, i.e., not expecting others to be available at all times. In our study, a clear risk factor for reporting mental health symptoms was to perceive the accessibility offered by mobile phones as stressful. Thus, actually perceiving something as a "problem" could indicate a more general problem, and could serve as a warning signal for taking measures to preclude constant accessibility and overuse.