Abstract

Mobile phone use, predominantly smartphones, is almost ubiquitous amongst both adults and children. However
adults and children have different usage patterns. A major challenge with research on mobile phone use is the
reliability of self-reported phone activity for accurate exposure assessment. We investigated the agreement between
self-reported mobile phone use data and objective mobile operator traffic data in a subset of adolescents
aged 11–12 years participating in the Study of Cognition, Adolescents and Mobile Phones (SCAMP) cohort. We
examined self-reported mobile phone use, including call frequency, cumulative call time duration and text messages
sent among adolescents from SCAMP and matched these data with records provided by mobile network
operators (n = 350). The extent of agreement between self-reported mobile phone use and mobile operator traffic
data use was evaluated using Cohen's weighted Kappa (ĸ) statistics. Sensitivity and specificity of self-reported
low (< 1 call/day, ≤ 5 min of call/day or ≤ 5 text messages sent/day) and high (≥ 11 calls/day, > 30 min of
call/day or ≥ 11 text messages sent /day) use were estimated.

Agreement between self-reported mobile phone use and mobile operator traffic data was highest for the duration
spent talking on mobile phones per day on weekdays (38.9%) and weekends (29.4%) compared to frequency
of calls and number of text messages sent. Adolescents overestimated their mobile phone use during weekends
compared to weekdays. Analysis of agreement showed little difference overall between the sexes and socio-economic
groups. Weighted kappa between self-reported and mobile operator traffic data for call frequency during
weekdays was κ = 0.12, 95% CI 0.06–0.18. Of the three modes of mobile phone use measured in the questionnaire,
call frequency was the most sensitive for low mobile phone users on weekdays and weekends (77.1, 95%
CI: 69.3—83.7 and 72.0, 95% CI: 65.0–78.4, respectively). Specificity was moderate to high for high users with
the highest for call frequency during weekdays (98.4, 95% CI: 96.4–99.5).

Despite differential agreement between adolescents’ self-reported mobile phone use and mobile operator traffic
data, our findings demonstrate that self-reported usage adequately distinguishes between high and low use.
The greater use of mobile smartphones over Wi-Fi networks by adolescents, as opposed to mobile phone networks,
means operator data are not the gold standard for exposure assessment in this age group. This has important
implications for epidemiologic research on the health effects of mobile phone use in adolescents.