The Anatomy of a Rumor
Karolinska’s Maria Feychting Cites Pathology Bias To Discredit NTP RF Cancer Study
A few days ago, I received an urgent warning from a longtime contact in Sweden. An industry associate had told him that the U.S. National Toxicology Program’s study on cell phone cancer risks was screwed up and essentially “useless.”
I was tempted to disregard it as nothing more than a corporate delusion. But the original source was said to be Maria Feychting, a professor at the Karolinska Institute and the vice chair of the International Commission on Non-Ionizing Radiation Protection (ICNIRP). She had cast doubt on the landmark $25 million NTP RF–animal study in a talk presented at the Royal Swedish Academy of Sciences —the institute that awards the Nobel prizes in physics and chemistry every year.
I decided I had to check out the rumor.
The crux of Feychting’s argument, I was told, is that the pathology analyses were not properly blinded. That is, the pathologists were aware which samples had come from the exposed animals and which were from the controls. The diagnoses were therefore subject to bias and could not be trusted. The net result would be that the higher tumor rates reported by the NTP had, as the rumor put it, “no value.”
The evidence, I was told, is buried in Appendix C of the NTP’s report of its “partial findings,” issued in May 2016 (pp.21-22):
“All PWG [Pathology Working Group] reviews were conducted blinded with respect to treatment group and only identified the test article as “test agent A” or “test agent B.”
Feychting and others appear to have assumed that “A” and “B” were code for the exposed and controls rats.
They were wrong.
“The PWGs were carried out on slides that were blinded as to exposure group or control,” John Bucher, the study director and the associate director of the NTP wrote in an e-mail when asked about the Feychting rumor by Microwave News. He also confirmed that agents “A” & “B” referred to the different RF modulations.
This same concern had already been raised and addressed during the internal NTP review prior to the release of the interim results last year. That report states that “A” and “B” refer to the two types of cell phone signals under study, GSM and CDMA (p.69).
Feychting did not respond to a request for comment.
A Rumor is Born at an ICNIRP Workshop
ICNIRP invited Bucher to present the results of the NTP study at a meeting held in Munich, November 8-10. Most of what Bucher said had already been presented at the BioEM 2016 conference in Ghent last year. There was nothing new, said Martin Röösli of the Swiss Tropical and Public Health Institute in Basel. Röösli, a member of ICNIRP, attended both meetings.
Bucher recalls that the ICNIRP members asked a lot of questions about the NTP study, especially about the pathology review procedures. (These are spelled out in detail in Appendix C of last year’s interim report.)
I asked Eric van Rongen, the chair of ICNIRP, for a copy of the agenda of the ICNIRP meeting. He refused, explaining that it was a “closed meeting.”
Maria Feychting of the Karolinska Institute Cites Pathology Bias
John Bucher, the RF Study Director at the NTP
Less than two weeks later, on November 21, Feychting spoke at a seminar on EMF health risks held at the Royal Swedish Academy of Sciences. The meeting was organized by the Swedish Committee for Radio Science (SNRV). What she said about pathology bias is based on two reports from the meeting, which were relayed to me by my Swedish contact.
One was from Lars-Erik Larsson of TeliaSonera, the dominant telecom in the Nordic countries. In an e-mail exchange, he confirmed much of what I had heard. He told me that, in her talk, Feychting had pointed out that no tumors had been detected in the NTP control group, a finding that is inconsistent with the historical data. She attributed this to the fact that the “tumor pathology was not done blind,” Larsson wrote.
The other report was from Yngve Hamnerius of Chalmers University of Technology in Gothenburg, who organized the seminar. Hamnerius did not answer a request for comment; he declined to share the SNRV attendance list.
The Feychting rumor was greeted with disbelief in the U.S. “I see no basis for such statements,” said Ron Melnick, who designed the NTP study. (He has since retired.) “They followed standard operating procedures. So, if the basic NTP methodology is flawed you have to throw out all the NTP studies.” He stressed that, “The controls are handled the same way as the exposed.”
“This sounds like some conspiracy theory,” Melnick said.
The rumor also reached Joel Moskowitz at the University of California School of Public Health in Berkeley. “It is disconcerting that a few scientists are trying to dismiss this study which is considered the strongest toxicology study ever conducted on cell phone radiation and cancer,” he told me.
Some in Europe were equally skeptical about Feychting’s claim. “I have read all the reviews and never found a hint that the slides were read unblinded,” stated Michael Kundi in an e-mail from the Medical University of Vienna. Kundi is the head of the university’s Institute of Environmental Health.
Feychting’s Incomplete Conflict of Interest Statement
Telia helps pay for some of Feychting’s research and, if all goes according to plan, will continue to do so for many years to come. The company is one of the sponsors of the Swedish branch of the COSMOS Study, a multi-decade prospective epidemiological study of the health impacts of mobile phones led by Feychting. Two other telecoms, Telenor and Ericsson, also support the Swedish study.
Feychting’s relationship with the Swedish telecoms through COSMOS is not included in her 2017 “Declaration of Personal Interests,” filed with ICNIRP. The section on “research support from commercial entities” is blank.
A possible explanation for the omission is the mention on the COSMOS Web site of a “firewall” between the telecoms and the Karolinska Institute. What this means is hard to decipher since it is no secret that the telecoms are helping to pay for Feychting’s work. She knows it. Telia knows it. The Karolinska knows it. They all know the money tap is turned on and could be turned off. Yet, no one suggests that there may be a funding bias —or a rumor bias.
On its Web pages, ICNIRP states that it is free of “vested interests,” and that ICNIRP members “cannot be employed by industry.”
Feychting was a member of the IARC Interphone epidemiological study group and has consistently challenged the interpretation that it points to a brain tumor risk. She argues that the study is flawed due to selection and recall bias in the data collection. Now, Feychting appears to be claiming that the NTP finding of excess brain tumors among rats exposed to cell phone radiation is due to pathology bias.
The NTP answered these concerns long ago.