Page 33 - Contributed Paper Session (CPS) - Volume 6

CPS1483 Ken Karipidis et. al
incidence, with corresponding 95% confidence intervals (CI) over three time-
periods: 1982-1992 (representing increased CT and MRI use), 1993-2002
(representing advances in MRI) and 2003-2013 (representing substantial and
increasing mobile phone use; more than 65% of the population).(9) Lowess
smoothing was used in the graphical representation of the time trends.
Mobile Phone Use Data Sources: Mobile phone use was estimated using
information on mobile phone accounts and survey data on actual use. Data
on the annual number of mobile phone accounts from 1987, when mobile
telephony first commenced in Australia, to 2013 was obtained from the
national telecommunications regulator, the Australian Communications and
Media Authority (ACMA). The number of mobile phone accounts per capita
for each year was calculated by dividing the number of accounts by the total
Australian population in that year (obtained from the Australian Bureau of
Statistics), noting that since 2008 the annual number of accounts has been
exceeding the number of people in the population. This data is not a true
indication of mobile phone use as some users may have had more than one
account and other users no account. A consumer survey conducted by ACMA
reported that approximately 90% of the population used mobile phones in the
years 2009 to 2013.(10) We estimated the annual prevalence of mobile phone
use by multiplying the annual number of accounts per capita by a factor of
0.9.(10) It was not possible to stratify prevalence of use by age or gender; thus
an overall estimate of prevalence is provided equally for all ages across the
20-59 age range and for both males and females.
Statistical Analysis of Predicted Incidence: With the assumption that
mobile phone use is associated with glioma in adults as reported by the
Interphone and Swedish studies, we calculated predicted incidence rates and
time trends by applying various relative risks (RRs, 1.5, 2, 2.5, 3) and latency
periods (1, 5, 10, 15, 20 years) for three different mobile phone use scenarios:
a) All users – RRs were applied to all mobile phone users
b) Heavy users – RRs were applied to heavy mobile phone users (defined as
19% of mobile phone users by the Interphone study)
c) Regular users and heavy users - RR of 1.5 applied to regular users (81%
of all users) and RRs of 2, 2.5 and 3 applied to heavy users (19% of all
users)
The annual predicted incidence rates were calculated for the period 1987-2013
using the formula:
Predicted Incidence = (P × RR × IB) + ((1 - P) × IB)
where P denotes the annual prevalence of mobile phone use, RR the relative
risk and IB the pre-mobile phone baseline incidence from 1982-1987.
Confidence intervals and statistical significance of observed and expected
incidence rates were calculated using Poisson confidence intervals.(11)
Analyses of predicted incidence time trends were carried out by estimating the

22 | I S I W S C 2 0 1 9