Control Selection in the study of Childhood cancer in relation to distance from high voltage power lines in England and Wales
Leeka Kheifets1, Maria Feychting2, Joachim Schuz3
1. Department of Epidemiology, School of Public Health, UCLA, CA, USA
2. Institute of Environmental Medicine, Karolinska Institutet, Sweden
3. Institute of Cancer Epidemiology, Danish Cancer Society, Denmark
We have read with interest the paper from Draper and colleagues .
Given its large size the risk estimates in the paper should be stable.
Furthermore, because contact with the subject was not necessary selection
bias due to the differential participation among cases and controls, which
plagued some of the previous studies , has been avoided. Thus we were
particularly surprised by the dependence of the results on the chosen
control group noted by the authors, (who used CNS and other cancer
controls for leukaemia cases in one of the comparisons). To explore this
further we combined all controls into one group and used it for
comparison. We felt this is justified based on both theoretical and
empirical grounds: exposure at birth among controls chosen for leukaemia,
brain tumours and other cancers should not depend on the cancer subtype;
crude odds ratios calculated by us did not differ (beyond first decimal)
from the matched results presented by authors (data not shown).
Use of the combined control group revealed a pattern different than
the one presented in the original paper (Table 1). As would be expected,
results for all cancers combined show no relation to the distance. For
both leukaemia and brain cancer results at two distances are noteworthy:
for the 50-100 meters category an excess of leukaemia and a deficit for
brain tumours is observed. For the 500-600 meters category we observed a
modest excess for both leukaemia and brain tumours. Of note is that the
trend reported in the original paper is not present when the combined
control group is used, thus indicating that the trend depended on the
leukaemia controls rather than on the leukaemia cases. We agree with the
authors that the results of this study do not support a possible magnetic
field association, as has been reported by the IARC monograph .
However, distance is known to be a very poor predictor of magnetic field
exposure, and therefore, results of this material based on calculated
magnetic fields, when completed, should be much more informative.
Further insight might be gained by details on the methods used for
the control selection and sensitivity analyses by age, sex and time
1. Draper G, Vincent T, Kroll ME, Swanson J. Childhood cancer in
relation to distance from high voltage power lines in England and Wales: a
case- control study. BMJ 2005; 330:1290-2.
2. Ahlbom A, Day N, Feychting M, et al. A pooled analysis of
magnetic fields and childhood leukaemia. Br J Cancer, 83, 692-8 (2000).
3. IARC Monographs on the Evaluation of Carcinogenic Risks to
Humans. Vol 80 Non Ionizing radiation, Part 1: Static and Extremely Low ¨C
Frequency Electric and Magnetic Fields. 2002
For LK work with EPRI and consulting for utilities.
Competing interests: No competing interests