A statistical evaluation of medical data in terms of cancerous diseases
in the vicinity of a mobile base station was carried out by German
physicians. This survey was published in a German journal. The judgement of Federal
Office for Radiation Protection (BfS) is described below.
BACKGROUND
The study was published by the end of 2004 in a German journal: Eger H,
Hagen KU, Lucas B, Vogel P, Voigt H. Einfluß der räumlichen Nähe von
Mobilfunksendeanlagen auf die Krebsinzidenz.
Umwelt-Medizin-Gesellschaft 2004 (4): 326-332.
A PDF-file of the article can be downloaded under
http://www.hese-project.org/de/emf/Studien/StudienDiskussion/NailaStudie/20050226_naila-studie.pdf
METHODS AND MATERIALS
Study region
In June 1993 a GSM base station was installed in close proximity to the
residential part of the small town Naila in southern Germany. In September
1993 it became operational. The transmitter has a nominal power of 15
watts per channel with a typical frequency of 935 MHz.
Data collection
The study was conducted by 5 medical doctors. The
study population consists of their patients. Two regions were defined, the
inner region (distance less than 400m from the base station) and the outer
region (more than 400 m distance). A random sample of comparable streets was
chosen in both areas. A street
including a large nursing home in the inner region was excluded. The leader of
each medical practice examined lists of the streets under study and identified
patients from his practice and the cases of tumours since 1994 to 2004. It was
a condition that patients had remained in the same area for the full ten-year
study period. The authors argue that the coverage with those living in the
study areas was about 90%. All malignant tumours were included with the
exception of skin cancers but including malignant melanoma. The diagnoses were
confirmed by data of health insurance funds and an independent route such as
hospital discharge reports and histological diagnoses.
There was no individual exposure assessment. Persons living in the
inner area were defined as “exposed” and those in the outer area were
defined as the comparison group. Information on other potential
confounders was not collected. The study included a total of 933
patients, among them 302 from the inner area and 631 from the outer
area.
Statistical methods
The odds ratio (OR) was calculated by means of a simple two by two
table using a Chi2 test with 95% confidence limits. In addition the
relative risk (RR) was determined. This was done for the 10 year period
1994 to 2004 and for the 5 year period 1999 to 2004, assuming a 5 year cancer
latency period. Age and sex or other variables were not included in the
statistical analyses. The authors argue that the mean age in the inner
and outer area was similar for women (41 years) and for men (38 Years)
as well as the percentage of women (55% in both areas).
RESULTS
- In the follow-up period 1994 to 2004 a total of 34 new cancer cases
occurred. Among them 18 cases in the inner area and 16 cases in the
outer area (see table 1). The main sites of cancers had been breast
cancer (5 inner/3 outer), pancreas (3 inner/2 outer), prostate (2
inner/ 3 outer), colon (2 inner/2 outer), lung (2 inner/1 outer), and
kidney (2 inner/ 1 outer). There was a 2.35fold higher risk to develop
cancer in the inner area compared to the outer area (OR=2.35, 95%
confidence interval: 1.18-4.67). The calculated relative risk was 2.27.
No adjustment was made for age, sex or any other variable. The mean age
of getting cancer was 64.1 years in the inner area and 72.6 years in
the outer area. The authors argue that over the 10 year period
malignant diseases developed at younger ages in the inner area.
However, there was no formal test. The risk of breast cancer was
3.5fold higher in the inner area compared to the outer area, yet this
increase in risk was not statistically significant.
- In the follow-up period 1994 to 1999 a total of 13 new cases occurred
(see table 1). Among them 5 cases in the inner area and 8 cases in the
outer area. There was no statistically significant difference in risk
between the inner and the outer area.
- In the follow-up period 1999 to 2004 a total of 21 new cancer cases
occurred (see table 1). Among them 13 cases in the inner area and 8
cases in the outer area, yielding a 3,38fold increased odds ratio (95%
CI: 1.39-8.25) for cancer in the inner area compared to the outer area.
The corresponding relative risk was 3.29. Again there was no adjustment
for age, sex, or any other variable. For this follow-up period no
information on type of cancer or age at disease was provided.
Table 1: Number of newly diagnosed cancer cases in the inner and outer area
by follow-up period
EVALUATION OF THE RESULTS BY THE BFS
The study shows a twofold higher probability to develop cancer for
people living near the base station compared to those living in the
outer area in the 10 year follow-up period. In the first five years
there was no increase in risk, while in the time period 1999 to 2004 –
five years after operation of the base station – there was a three
times increased risk for cancer.
One of the strengths of the study is that a rural study region with a
stable population was chosen, and that there was a small density of
base stations (installation of the first base station 1993, a second in
1997). On the other hand there are a series of methodological
weaknesses.
Age and sex of the patients was not considered in the statistical
analyses. It is not sufficient to refer to an equal average age and
ratio of sex in both regions, because differences in the common age and
sex distribution may exist. Not considering such differences can
produce a bias in risk estimate, particularly if specific sites of
cancers were considered.
Further on, it cannot be excluded that the number of cancers in the
outer area is underestimated. A rough comparison done by the authors
with the incidence rates from the cancer registry in the Saarland,
showed that substantially less cancers were observed than expected. An
underestimation in the outer area can induce an overestimation of the
risk. The authors argue that they would know if one of their patients
had become cancer. It is, however, unclear whether there was a complete
follow-up of the patients over the 10 year period by the doctors or
whether patients went to other doctors or hospitals outside or inside
the study area during the follow-up period.
Information on potential confounders for cancer is not available.
Since all sites of cancer are considered all potential confounders may
play a role (smoking, diet, obesity, alcohol, etc.). If the
distribution of these factors are different in the inner and outer
area, then bias due to confounding cannot be excluded.
There was no individual exposure assessment. On average, exposure in
the inner area might have been higher compared to the outer area, but
due to the complex dependency of wave propagation on environmental
factors (e.g. screening, reflection, absorption) there will be
substantial individual differences in exposure. The authors state in
the paper that measurements done by the Bayerischen Landesamt für
Umweltschutz showed that radiation intensities in the inner area are
about 100 times higher than in the outer area and significantly higher
than other emitting electromagnetic waves, such as radio, television
and radar. According to a communication from the Bayerischen Landesamt
für Umweltschutz (by 15.12.2004) this statement is not true.
A further limitation concerns the low sample size. The statistical
analyses are based on 34 cancer cases in the 10 year period and after
considering a latency period of 5 years only 21 cancers. Analyses on
specific types of cancer are therefore not possible. The authors
hypothesize that breast cancer is a „marker-carcinoma“ for high
exposure to radiofrequency. This is highly speculative, since in total
only 8 breast cancers occurred and the observed risk was not
statistically significant. Also, they argue that cancer patients near
base stations develop cancer at a much younger age. Again this is
speculative. The age at disease depends on site of cancer, age and sex
of patients. All these variables were not considered in the analyses.
Altogether this study has low significance and conclusion that can be
drawn from this study are very limited. The group of doctors themselves
characterize the study as a pilot study, which was conducted without
any financial resources in a small area with very simple methods.
They request from the responsible authorities in Germany, to repeat the
study in a similar manner in other regions in order to reproduce the
results. Studies of this type are in principle only meaningful, if the
above mentioned methodological problems can be resolved. This means, that for example data from complete cancer
registries can be used instead of data of patients, the study is
population based, a sufficient sample size is included, information on
other main risk factors can be obtained, and in particular an
individual exposure assessment is possible.
The evidence of an exposure-response-relationship is a major point. It
is not sufficient to compare a group of people which is “on average
more exposed” than a group which is “on average less exposed”, because a
single observed elevated risk may be the result of bias (selection
bias, information bias, confounding). The observation of an increase in
risk over several categories of exposure would be more supportive for a
potential causal relation. A pre-condition for allocation of study
subjects into different exposure categories is a retrospective
individual exposure assessment. The current scientific level of
knowledge does not allow such a retrospective exposure assessment.
CONCLUSION
Despite existent major limitationsin the “Naila Mobilfunk” study, the reported singular evidence of a potentially threefold increased cancer risk was considered seriously by the BfS. With respect to observed elevated health risks from other singular studies, the BfS conducted a large research program on telecommunication with a financial budget of 17 Mio EUR. About 54 research projects in the area of biology, dosimetry, risk communication
and epidemiology were funded. An overview on these projects is given in the
internet http://www.emf-forschungsprogramm.de/forschung. Among them there are
four large epidemiological research projects that directly investigate the association between cancer and mobile telecommunication, as well as dosimetric projects aiming to determine the exposure of the population by fields of base stations.
An evaluation summing up the results of the German Mobile
Telecommunication Research Programme can be found in the brochure (in German language only) and in the final report of the programme.
It is the main result of the DMF that the previous
indications of possible health effects due to high frequency electromagnetic
fields have not been reproduced. Regarding the use of mobile phones which
results in a much higher exposure to high frequency fields than that from base
stations an increase in cancer risk could not be proven when using them for up
to 10 years.
|
