EHT letter in the BMJ commenting on the Danish mobile phone study
Re:Not enough data excluding cellphones’ morbidity
Devra L Davis, Director
Ronald B. Herberman, Yael Stein
Environmental Health Trust
Re: Use of mobile phones and risk of brain tumours: update of Danish cohort study. Frei, et al. 343:doi:10.1136/bmj.d6387
(BMJ. 2011 Oct 19;343:d6387 Frei et al studied tumours in the central nervous system among Danish mobile phone subscribers aged 30 and over born in Denmark after 1925, subdivided into subscribers and non-subscribers of mobile phones before 1995, which included 358,403 subscription holders with 3.8 million person years and excluded business users. The authors acknowledge some design limitations and concluded that there were no increased risks of tumors of the central nervous system, associated with duration of mobile phone subscription.
We note a number of additional serious errors, faulty study design and major contradictions that bias the analysis toward Type II errors (falsely negative results) as detailed below.
1. In view of the study aim to investigate how mobile phone use impacts risk of brain tumours, the size of the cohort was too small to find a significant change in underlying risk of relatively rare brain tumours, and accordingly lacked statistical power.
More specifically: The background rate of brain cancer in the general population is about 7 per 100,000 persons and the rate of benign brain tumors is about 12 per 100,000. Therefore, studying brain tumors in a prospective cohort of less than several million persons has a very low chance of finding any increase of a risk.
The authors note that study had no information on actual cellphone use, nor on side of the head commonly used, but only on subscriptions that began in 1995 or earlier. These parameters are among the least reliable ones to estimate exposure and find an effect.
As others have noted in rapid response here, subscribers who began to use phones after 1995 and those who were business users before 1995 – are considered nonsubscribers, (since no data were available on subscriptions after 1995). This effectively puts those who began to use phones later than 1995 into the nonuser (control) group, although we all know that pricing, etc. stimulated both wider use and longer call times (exposures) among these later users who were also more socially diverse.
2. The authors lacked information about other sources of microwave radiation (cordless phones, routers, and base stations). Because “user” was defined as someone who made one call a week for six months, this combines slight and heavy users into one group. This could lead to a serious, systematic underestimation of the relationship between actual phone exposure and brain tumors risk.
3. From the onset: The cohort is too small, the average exposure time is too short, and the design is therefore likely to find no effect.
Brain tumors are known to have a latency of at least ten years (or more) in adults. Thus, the time period observed here, together with the relatively small size of the cohort monitored, at the outset was unlikely to find any effect attributable to cellphone use.
Regarding latency covered in studies of brain tumours and mobile phone radiation, a recent comment from respected epidemiologists, Rudolf Saracci and Jonathan Samet, notes, that none of the today’s established carcinogens, including tobacco, could have been firmly identified as increasing risk in the first 10 years or so since first exposures began. Ionizing radiation is a recognized cause of brain tumours. With the exceptions of rare instances where exposure began in childhood or was very high, radiation induced cases occur on average after 10-20 years since the time of first exposure.
The INTERPHONE study shares with all studies previously carried out on mobile phones and cancer the inherent limitation that it can investigate only a short period of observation since first exposure; the distribution of exposure is brief and truncated, leaving limited exposure time for an exposure-related cancer to develop.
4. The study confuses “statistical significance” with “public health importance”.
In studies of rare events, it is difficult to have sufficient numbers to produce statistically significant results within the conventional 95 % confidence interval, where the chance that the finding is wrong is .05 or less. In fact, this study actually finds increased risks in several subgroups, but fails to report these effects because the numbers do not achieve statistical significance (as to be expected in underpowered studies). These borderline significant results are consistent with other reports from well-designed case control studies that have found significant risks of glioma after a decade of regular use.
A weight of the evidence approach that examines the totality of studies on this topic finds that most well-designed cohort studies have found a doubled risk, according to a recent meta-analysis that differentiated quality of studies (5). Thus, the lack of statistical significance in this study reflects the lack of statistical power. Lack of an association in an underpowered study never can be: proof of no association between cellphone use and brain cancer.
5. Significant data are “diluted” and obscured among other findings.
For example: the authors report that “The highest estimate was found for cerebral ventricle based on eight cases (2.58 (1.08-6,15) . And even though this result achieves statistical significance, it is dismissed.
6. Contrary to Frei et al’s assertion that there is no secular increase in brain tumour incidence, incidence has increased significantly in some Nordic countries that have been heavy and persistent mobile phone users, while Swedish incidence data are underreported.
The study claims that there is no underlying secular trend in brain tumors in Sweden and therefore cellphone use does not cause brain cancer. Comparisons of tumours reported to the Swedish National Board of Health with those reported to the cancer registry indicate that brain tumour incidence is significantly under-reported as is incidence of soft tissue tumours, leukemia and lymphoma. (6) In other Nordic countries, there is clear evidence of recently increased incidence in brain tumours with reduced mortality. From 2005-09, brain tumour incidence annually increased in Norway in males and females 1.7% and 2.8 % iand .5 % and .1 % annually in Finland. Mortality declined substantially during this same time period in Norway -.8% and -1.7% in males and females in Norway and -11.4% and -.5% in Finland. (see appendix below).
7. Unacknowledged conflict of interest
The authors declare no conflict of interest in support for this analysis. In fact, the original cohort was first established partly with Danish phone company funding.
(accessed 12/30/09) According to Bloomberg Financial News (Mobile Phones Don’t Cause Brain Cancer of Leukemia, Study Finds; 2/26/02), the International Institute for Epidemiology completed a study that cost $373,000 and was funded in part by Denmark’s largest phone company, Tele Danmark A/S, which is partially owned by SBC Communications, and the second-largest mobile phone service in Denmark, Sonofon A/S, owned by Telenor AS and BellSouth Corp.
Based on the evolving science at hand, in May of this year the IARC expert committee (2) determined that cellphones and other wireless radiation should be classified as possible carcinogens. In reaching that conclusion, IARC expressly rejected earlier reports from the same Danish cohort used by Frei et al, noting a number of methodological flaws in the original design which also afflict this updated analysis. Reflecting the emerging evidence, a growing number of the world’s leading experts are issuing calls for protective policies.
Two leaders of the World Health Organization Interphone study Elisabeth Cardis, PhD, of Barcelona’s Centre for Research in Environmental Epidemiology and Siegal Sadetzki, M.D., M.P.H., of Tel Aviv’s Gertner Institute for Epidemiology and Health Policy , advise the following (7):
” While more studies are needed to confirm or refute these results, indications of an increased risk (of brain cancer) in high and long -term users from interphone and other studies are of concern… even a small risk at the individual level could eventually result in a considerable number of tumors and become an important public-health issue. Simple and low cost measures, such as the use of text messages, hands free kits, and/or the loudspeaker mode of the phone could substantially reduce exposure to the brain from mobile phones. Therefore, until definitive scientific answers are available, the adoption of such precautions, particularly among young people, is advisable.”
NOTE: clicK here for the full letter, including graphs.
1. Schuz J, et al. (2006) Cellular telephone use and cancer risk: update of a nationwide Danish cohort. JNCI 2006. 98:1707-13.
2. WHO IARC Monograph Working Group, Carcinogenicity of radiofrequency electromagnetic fields. Lancet Oncol. 2011 Jul;12(7):624-6.
3. Frei P, et al, (2010) Classification of personal exposure to radio frequency electromagnetic fields (RF-EMF) for epidemiological research: Evaluation of different exposure assessment methods., Environ Int 2010 Oct;36(7):714-20
4. Gandhi O, et al (2011) Exposure Limits: the underestimation of absorbed cellphone radiation, especially in children Electromag Biol & Med. Early Online: 1-18.2011.
5. Levis et al (2011) Mobile phones and head tumours. The discrepancies in cause-effect relationships in the epidemiological studies–how do they arise? Env Helth, 2011, 10:59
6. Barlow et al (1998) The completeness of the Swedish Cancer Register
Acta Oncologica, 2009; 48: 27_33
7. Cardis E and Sadetzki S, Indicationsn of Possible Brain-Tumour Risk in Mobile-Phone Studies: Should we be Concerned? Occup Environ Med 2011;68:169-171 doi:10.1136/oem.2010.061358
INCREASED AGE-ADJUSTED INCIDENCE IN BRAIN TUMOURS
MALES AND FEMALES IN FINLAND AND NORWAY
Finnish National Cancer Registry, as reported in NORDCAN, Association of the Nordic Cancer Registries ? All Rights Reserved. 27.10.2011
Finland: Brain Tumour Incidence
Norway: Brain Tumour Incidence
Competing interests: None declaredLeave a reply →