Do Cell Phones Cause Cancer? (part 2 the argument for)
In part 1, I gave the motivation for this examination and its starting point. I also gave a bit of background discussing radiation scares from my childhood and placing them into context for the current debate. I discussed a bit about my personal educational biases that you should consider in reading this text. We are now ready to look at the evidence for harm. Remember these are the studies with the burden of proof on them. If they are not strong enough you must not accept the conclusion of harm. However, you have the responsibility to determine if the strength of their conclusions is enough for you to take action.
Jeromy provides an extensive list of research articles claiming some sort of damage from EMR. I didn’t count them but he must list more than 10,000 citations; one site alone lists 6000. Given this, the very impressive list was much too long for me to examine, so I selected a few of the newer articles to report on here. I focused on cancer and brain tumors from cell phone radiation. However, my list also includes one article on damage from house electrical wiring for you to consider.
As a practicing statistician, I tend to discount meta-analyses that are not based on randomized control trials. These are studies where the subjects while not necessarily randomly selected are randomly assigned to a treatment. Joining data from studies where the subjects were not randomly assigned needlessly complicates the inference—and rarely is this considered by readers. But there is more to it than that. Meta-analyses tend to overly discount spatial and temporal effects.
Cell phones have significantly changed over the last 30 years. What was true of the early models may not be true of the latest, so we should expect researchers to look for a temporal effect. It may be that early adopters suffered more adverse effects than later adopters. In addition, the effects from the radiation are not instant, we expect them to accrue over time.
Also, the technology for measuring EMR has improved. At least according to advertising, early models of test equipment are in general less precise than the newer models, and may not be as carefully calibrated as newer models. This combination causes the statistical mathematics fits. At the very least these data are violating the assumption that the data (technically deviations from the selected model) are not related to each other and they come from the same distribution.
I also discount studies relying on p-values from specialized statistical tests without p-values from the standard tests. I understand, and agree, that there are reasons to use a specialized test. And I acknowledge that papers failing to find an effect are not published as frequently as papers that do. This results in a tendency for researchers who failed to get the results they wanted with the simple test to go off in search of tests supporting their contentions. They keep testing until they get the right answer. In my practice, I report to clients the results of all tests that I conducted on their data (all of them, everytime). This lets the user of the results decide.
From the many papers that Jeromy cites, I selected a few to read in detail by examining their abstracts.
I discarded the 2000 papers from the Naval bibliography, because they are old (see above), most are written in languages I don’t understand (personal failure), and most importantly there are plenty of modern studies supporting his contention. They simply aren’t needed.
Here are a few of the studies I read with my comments after the citation. I provided links so you can read them yourself if you desire.
Hardell, Lennart et al. (2015) Mobile phone and cordless phone use and the risk for glioma – Analysis of pooled case-control studies in Sweden, 1997–2003 and 2007–2009. Pathophysiology , Volume 22 , Issue 1 , 1 – 13.
This is a case-control study, where researchers take people with the malady of interest (the case) and match them with a person not having the malady but having similar demographics (the control). It should be clear that the quality of the matching will determine the quality of the results. Unfortunately, it’s very hard to determine how well the researchers did with this from a paper; any paper. I’ve seen both good and bad matching. The gold standard of case-control is based on identical twins, which is not the case here.
I want to point out that matching is very hard—maybe impossible. Cancer is a genetic disease so the researchers are trying to match environmental, behavioral, and genetic factors. How many people do you know who share nearly the same genetic values, and live in the same environment, and do the same things, as yourself?
The researchers used Swedish brain tumors data and matched a control by randomly selecting and individual of the same gender and 5-year age class from the Swedish population registry. They sent each control a survey asking about cell phone use. They ended up with 1498 case to explore. I think the matching criteria is a bit weak; what do you think, good enough for you?
One more thing, in this type of study the inference is a bit stilted. Since they start with a fixed amount of case data, inference on what fraction of the population has the disease is not possible. I think you can see this by envisioning drawing ten red M&M’s from the bag then matching them with ten green ones. How many red ones in the bag? Right, twelve less than there was when you opened it (the two you ate while doing the study). So, who knows? There is just not enough information.
So, when you read the conclusions they are a bit convoluted. The researchers found that if you had brain tumors you were 30% more likely to have used a cell phone than if you didn’t have tumor. However, this went up to 70% more likely to have used one, if that use was long-term (> 25-years). Since cell phone technology is continually evolving, the researchers wanted to correct for the differences in when the phones were used. Older phones required a stronger signal, different frequencies, and different modulation techniques, so were perhaps more or less dangerous.
These are fairly compelling results in spite of the weak matching.
The next paper is really out of the scope of our discussion as it is focused on 50 to 60 cycle radiation as might be encountered in power lines. Feel free to skip it if you are strictly concerned with microwave (cell phone/smart meter) radiation. I just found it interesting.
Martin L. Pall (2013) Electromagnetic fields act via activation of voltage-gated calcium channels to produce beneficial or adverse effects. J. Cell. Mol. Med. Vol 17, No 8, pp. 958-965.
This paper describes a mechanism by which radiation might create genetic damage, influencing calcium channels. The researchers All of the examples were in the lab in support of the calcium channel hypothesis.
Dr. Pell lists 24 lab experiments exposing various tissues, in test tubes, to radiation and noting the changes in the calcium channel function that resulted. This doesn’t mean that cancer did happen, only that it might have happened if the tissue were still in its host. The occurrence of cancer is dependent on the goodness of the model.
This is an example of the causal methodological approach I mentioned in part 1; the model (calcium channel function) matters. I don’t feel that I’m qualified to comment on his model; I’ll accept that it was good enough that it didn’t raise objections in the publication process.
I give the author special credit for including in his results studies of clinical methods designed to improve human health (stimulating bone growth) through the use of this radiation.
The next paper I included because it hearkened back to the papers of the 19th century; it’s observational. This is one of those that sends shivers up the spines of many academics. The author’s see an effect, watch a change, and observe another change.
Tetsuharu Shinjyo and Akemi Shinjyo (2014) Significant Decrease of Clinical Symptoms after Mobile Phone Base Station Removal –An Intervention Study. Umwelt-Medizin-Gesellschaft, 27(4), S.294-301.
A cell phone tower was installed on top of a Condominium building in Okinawa, some of the residents got sick, cell phone tower was removed, some residents got better. Purely observational, but none the less compelling in its simplicity and clarity. We have no idea if it was the EMR from the tower or some other emission, but who cares, it was very likely something associated with the tower.
I find it interesting that there is recognition that children are at a significantly higher risk of damage from cell phones than are adults. However, few studies have been done on children. I wouldn’t interpret this as a statement of relative risk as children are accorded special legal protections, so it’s much harder to gain approval to do research on them. There is currently a world-wide case-control study in progress: Mobi-Kids. Hopefully the results will be available soon.
Smart Meters
Given Jeromy’s history it’s not surprising that there is a whole section of his website devoted to anti-Smart Meter studies. In case you haven’t met them, smart meters are an invention allowing an electric utility to monitor your electric use without sending a meter reader to you home. Oh, and if you are using an incorrect amount of electricity a smart meter will allow the utility to adjust your usage (read as cutoff), or maybe just charge you more for peak time usage.
Smart meters use similar transmitters as cell phones to communicate with the office. They are regulated by the FCC. They have been implicated in starting a few fires, exploding under stress, as well as irradiating people.
There seems to be quite a bit of information to imply that these are not something you want to have on your home. EWEB in Eugene OR has installed these meters on some homes. I mention this because EWEB is a past client and my wife’s family lives in Eugene.
After reading these you are no doubt ready to throw your cell phone against the wall, declare war on electric utilities, and the politicians who support them.
Well let’s look at the other side. Perhaps you will as much, or more, compelling data and research failing to find harm in this radiation.