User Feedback (from 18/09/2006):

Interesting article - but Hardell et al always seem to have results that are significantly different from others. The FDA state that "the results reported by Hardell are inconsistent with the results obtained in other long term studies. The use of a mailed questionnaire for exposure assessment and lack of adjustments for confounding factors other than time of diagnosis make the A Hardell et al study design significantly different from other studies. These factors along with the lack of an established mechanism for action in supporting animal data make the Hardell et al findings difficult to interpret". Dont you think it would be balanced to include a link or reference to this? Perhaps give the opportunity for feedback to be displayed on the page?

Powerwatch Response

Dear {Name Withheld},

Thanks for the feedback. We don't have an automated display of feedback due to past experience where people posted random insulting comments that were no use to anybody. We are more than happy to publish sensible feedback, and if you would like to have our correspondence on this issue on the site we would be more than happy to do so.

I agree entirely with the importance of holding as balanced and impartial point of view as possible, and I will concede that at times we don't treat "independent" papers with the same level of analysis as the industry related ones.

Some of your questions I can answer, some I will need you to clarify exactly what you mean. The mailed questionnaire is a genuine problem, and whilst it is open to inaccuracy, I see no evidence that the inaccuracy would be in favour of increased risk - decreased risk due to people overestimating their phone use I see as just as reasonable. For your interest, Lennart Hardell asked multiple times for the phone records from the cellphone companies and was bluntly refused access to any data on each occasion.

It most likely there are other confounding factors not addressed in the paper, as it is with pretty much all papers - however, without more information it is hard to say whether such factors would have much effect on the results, and it would be interesting to hear specifically which ones you are thinking of that may have an important impact on the study outcome.

I also am more than happy to accept that there is no established mechanism, but I don't think that is any reason in itself to dismiss results that are found to be statistically significant. If reasons can be identified that would suggest the results are incorrectly elevated to the point of making insignificant results significant, then that would be cause to treat the results with caution. I have yet to see sound reasoning that would suggest this is likely. To simply summarise that they are difficult to interpret is to ignore the potential significance of the result, and I believe that this would be scientifically irresponsible when so much could be it stake if the claims are found to be true. Kjell Mild, an author on most of the recent Hardell research is also fully supportive of the quality of the research, and is a member of MTHR so would have a reasonable basis of comparison between this research and some of the MTHR related industry papers.

If you look at some of Lloyd Morgan's graphical analyses of recent data (View Graphs), there are two things that I find particularly interesting. Firstly there is the clear negative, or protective, trend in industry funded papers. In some of the results this actually depicts a statistically significant protective effect from phone use, and similarly without some mechanism to support this nor any claim from the study authors that the protective effect is real, this is equally suspicious with regards to scientific accuracy. Of course, it could be partly explained by the other point I find interesting, which is related to the number of cases in the study points. Most of the significant Hardell et al points have at least 50 cases whereas few of the industry points have over 15. The sample of people used for the industry studies are astonishingly small to be able to find particularly significant results.

On the issue of study replication, the latest Interphone studies have made an enormous omission (if we are to take any of the Hardell studies seriously) by not taking into account cordless phone use as a confounding factor. A regular cordless phone user who is not a regular cellphone user finds themselves in the "unexposed" category. Considering the potential increase of relevant risk found by Hardell for cordless phone usage, if the increased risk is real it will go some way to nullifying the results in the interphone studies, where the "unexposed" group could have a significant increase in risk. This may go some way to explaining a number of the papers whose results are different to the Hardell research.

Finally, the real difficulty with brain cancer is tumours typically develop after between 15 and 25 years after the exposure to the causative agent (this is taken from now generally accepted research connecting ionising radiation with brain tumours, and the exposure times are generally known in the cases used). With mobile phones, usage started to increase significantly about 1997, and the real explosion of usage was about the year 2000. If we are seeing anything significant regarding tumours after 6 to 9 years, then this should be real potential cause for concern, as any increase that is genuinely associated to the phones will be far greater in 10 years time.

I would not be willing to make the summary of results being "difficult to interpret", because under the basis of the argument you are using, the same could be said about a lot of research papers not judged in that manner. I am more than willing to comment in the main article on confounding factors which have not been taken into consideration, and possible reasons why increased risk could be incorrectly elevated, but such things will need to be specifically mentioned and explained.

Best Regards,
- Graham Philips
Powerwatch Technical Manager

Extra Note:

In further correspondence, the user and Powerwatch agree that it is very important to maintain balanced sound arguments on both sides, with documentation where possible on why and where judgements have been made beyond current scientific understanding.