Oh no! My cell phone’s going to kill me!

I’m very puzzled.

Now, I know that my being puzzled isn’t particularly unusual. I’m frequently puzzled. I can’t figure out how, for example, anyone with the slightest bit of reasoning ability can do anything other than laugh when informed what homeopathy is and how it supposedly “works.” I can’t figure out why American Idol or Survivor is so amazingly popular.

And I can’t figure out why the University of Pittsburgh Cancer Center released this warning about cell phones:

PITTSBURGH July 24, 2008, 07:13 am ET · The head of a prominent cancer research institute issued an unprecedented warning to his faculty and staff Wednesday: Limit cell phone use because of the possible risk of cancer.

The warning from Dr. Ronald B. Herberman, director of the University of Pittsburgh Cancer Institute, is contrary to numerous studies that don’t find a link between cancer and cell phone use, and a public lack of worry by the U.S. Food and Drug Administration.

Herberman is basing his alarm on early unpublished data. He says it takes too long to get answers from science and he believes people should take action now — especially when it comes to children.

“Really at the heart of my concern is that we shouldn’t wait for a definitive study to come out, but err on the side of being safe rather than sorry later,” Herberman said.

Dr. Herberman is a highly respected cancer center director, which is why I can’t help but wonder just what on earth he was smoking when he decided to do this. It strikes me as being rash in the extreme; the announcement even admits that the published data do not support a link between cell phone use and brain tumors. This is alarmism that, I suspect, even Revere would have a hard time supporting, because it goes far beyond the published evidence and is based on “early unpublished data.” Scaring the nation based on “early unpublished data” is irresponsible in the extreme. Why did Dr. Herberman do it?

The question of whether cell phones cause or contribute to the development of brain tumors is not as easy a question to answer as one might think. First, there is the issue of biological plausibility. Radiowave energy at the power level used by most cell phones, is not ionizing, and our understanding of cancer is that, in general, ionizing radiation is what is required for radiation to cause or contribute to cancer. That does not mean that there isn’t a potential mechanism by which non-ionizing radiation can cause cancer that we don’t know about yet, but it makes hypothesis that cell phone radiation causes brain cancer less plausible. Too, we can actually test radiofrequency radiation in the same power range used in cell phones on cells in cell culture in order to determine whether exposure to such radiation can cause changes associated with malignant transformation. There is one confounding effect that has to be controlled for in such experiments (but is not always), namely that radiofrequency radiation interacts with water in order to heat it. Still, there are no compelling studies showing any specific effect of radiofrequency radiation on cells to induce changes associated with malignant transformation, at least none that I’m aware of. Animal studies are prone to the same sorts of problems as cell culture studies, but even so there is no good quality animal data that I’m aware of implicating cell phone radiation in the formation of cancer. On a basic science basis, there doesn’t appear to be strong evidence of a plausible mechanism or effect.

That brings us to epidemiological studies. For us to consider any epidemiological to be support for the hypothesis that cell phones cause brain cancer, there must be a few key results. First, there must be an increased incidence of brain cancer in cell phone users. It’s even more convincing if there is some sort of dose-response phenomenon. In other words, there should be an increasing risk of cancer with increasing cell phone use. Other results that also support the hypothesis would be tumors correlated with proximity. In other words, do people who primarily use their left hand to hold their phones to their ears tend to get tumors primarily on the left and people who primarily hold their phones with their right hand tend to get tumors primarily on the right? Finally, there should be a plausible lag time between exposure and tumor development consistent with known lag times for cancer, say 10-20 years, and some specificity. In other words, does exposure to cell phone radiation correlate with certain types of tumors and not others? There are other aspects of the results of a study that can more strongly support the hypothesis that cell phones cause brain cancer, but these are the main ones.

In general, however, getting “clean” data from an epidemiologic study of cell phone use that can support a strong enough correlation to suggest causation is very difficult. In order to correlate cell phone use with an increased incidence of brain tumors, it’s necessary somehow to be able to reliably quantify cell phone usage. This presents a big problem. It’s generally not possible to continuously observe people with their cell phones for years on end and obtain objective measurements. Another way is to ask people how much they use their cell phones, but memories are unreliable, and such methods are very prone to recall bias in the form of people with brain tumors being more likely to remember their cell phone use as having been heavy. That’s not even counting trying to control of the number of potentially confounding factors, such as heavy cell phone use being associated with certain jobs or, especially for 10-20 years ago when cell phones were far less common, with higher socioeconomic status. Then there’s the shift in technology from analog to digital in the early 2000s, which changed the power and frequencies used.

Of course, there are ways to overcome the limitations of retrospective studies. For example, an investigator can try to look at cell phone bills and see the number of minutes used per month, but who keeps their bills for 10 or 20 years, which is the time of exposure most relevant to the development of cancer? Alternatively, one can abandon the retrospective study altogether and follow people prospectively and have them report their cell phone usage from their bills every month. However, because brain tumors are a fairly uncommon form of cancer, thousands upon thousands of subjects would have to be followed this way, and it would take at least a decade or two to start to see any results. Also, it would be very hard to enforce a control group, given the ubiquity of cell phones.

Despite the difficulties, several epidemiological studies have been done, with largely negative results. Surprisingly, the NPR report actually summarizes them pretty well:

The issue that concerns some scientists — though nowhere near a consensus — is electromagnetic radiation, especially its possible effects on children. It is not a major topic in conferences of brain specialists.

A 2008 University of Utah analysis looked at nine studies — including some Herberman cites — with thousands of brain tumor patients and concludes “we found no overall increased risk of brain tumors among cellular phone users. The potential elevated risk of brain tumors after long-term cellular phone use awaits confirmation by future studies.”

Studies last year in France and Norway concluded the same thing.

“If there is a risk from these products — and at this point we do not know that there is — it is probably very small,” the Food and Drug Administration says on an agency Web site.

And:

A French study based on Interphone research and published in 2007 concluded that regular cell phone users had “no significant increased risk” for three major types of nervous system tumors. It did note, however, that there was “the possibility of an increased risk among the heaviest users” for one type of brain tumor, but that needs to be verified in future research.

Earlier research also has found no connection.

Steve Novella also summarized the state of the evidence regarding cell phone use and the risk of cancer and concluded that we can probably rule out a strong correlation between cell phone radiation and cancer, but we do not yet have enough data to rule out a small increased risk of brain tumors due to cell phones, particularly in children. I mostly agree with this characterization of the state of the evidence at present, which is why I find statements like this to be overblown fear-mongering:

A driving force behind the memo was Devra Lee Davis, the director of the university’s center for environmental oncology.

“The question is do you want to play Russian roulette with your brain,” she said in an interview from her cell phone while using the hands-free speaker phone as recommended. “I don’t know that cell phones are dangerous. But I don’t know that they are safe.”

I think PalMD got it right when he sarcastically retorted:

Hey, I don’t know for an absolute certainty that my popcorn won’t spontaneously combust, but I’m not yelling fire either.

To which I’d add: Hey, I don’t know with absolute certainty that vaccines don’t cause autism, but I’m not joining the anti-vaccinationists.

The actual warning is even worse, delving into some truly dubious comparisons:

In the early 1980’s, the owners of asbestos mines were reduced to bankruptcy as a result of lawsuits brought by the families of deceased exposed workers. A few years later, a key executive of Johns Manville, the most prominent company, drew lessons from the years of struggle of his industry against medical data and the scientists who were drawing attention to the risks of asbestos. He concluded with regret that greater warnings for the public, the establishment of more effective precautions, and more extensive medical research “could have saved lives, and probably also shareholders, the industry, and the benefits of its product.” [13, 14]

We call on the cell phone companies to provide independent access to records of use so that appropriate studies can be carried out.

Earlier in the report there was even a comparison with tobacco! Here’s the difference that makes these comparisons specious. Tobacco smoke is a known carcinogen. We have many in vitro and animal studies showing the mechanisms by which it induces cancer. Similarly, although the mechanism of how asbestos induces cancer remains unclear, we similarly have animal studies that show that it can. We have no such studies showing biological plausibility for cell phone radiation.

Particularly strange is Davis’s statement:

She said 20 different groups have endorsed the advice the Pittsburgh cancer institute gave, and authorities in England, France and India have cautioned children’s use of cell phones.

As PalMD astutely pointed out, these 20 groups did not actually endorse UPCC’s report; rather they endorsed recommendations similar to the recommendations in the report.

Having looked over this report and the news coverage it engendered, I still can’t figure out what on earth possessed Dr. Herberman to allow it to be issued. Even if the unpublished data are as alarming as implied (unlikely, given the number of previous studies that found either no risk or a questionable very slight risk) and the National Research Council in the U.S. is wrong about its being highly skewed due to selection bias, as a responsible leader in public health and cancer, you don’t pull something like this. You just don’t. It’s irresponsible as hell. A few months to get the data published in a peer-reviewed journal or presented at a major meeting are not going to make a difference, given the long lag time of cancer, and waiting until it is published will at least allow scientists and physicians to vet the data and decide if, on balance taking into account its limitations, it warrants such a strong warning. Or, if you really believe you can’t wait a few months because the risk is so horrible, then release the data with the warning, so that scientists can judge whether the warning is warranted.

Fortunately, many Americans seem to have more common sense than our cancer leaders:

“I think if they gave me specific numbers and specific information and it was scary enough, I would be concerned,” Loughran said, planning to call her mother again in a matter of minutes. “Without specific numbers, it’s too vague to get me worked up.”

Exactly.