I’m a bit conflicted, Doc – in the last few months I’ve seen news articles about two papers from Harvard Medical School, one saying that living close to nuclear power plants can cause cancer and the other saying that flight crews are getting cancer from their flying. On the one hand, I know you’re always saying that these sorts of things are not as bad as what we might fear, and you give very good reasons why this is the case. On the other hand…Harvard. I mean, they’ve got a lot of very smart people there and it’s sort of hard to question them. Isn’t it?
I understand how tempting it can be to simply listen to the smart folks at Harvard (and they really do have a lot of smart folks there) and accept their pronouncements as proven fact. But even the smartest of us can make mistakes, and even scary-smart people can succumb to the temptation of working outside of their area(s) of expertise and make some, frankly, rookie mistakes. And to me, the biggest rookie mistake is to decide that radiation is responsible for cancers without ever bothering to measure the radiation exposure. But before we go there, let’s start with something called the Ecological Fallacy.
When I was in Cambodia in 2003 I was surprised to learn that, in the entire nation, there was only one radiation oncology clinic. By comparison, there are more than a dozen in Manhattan alone, and NYC has less than half the population of Cambodia. When I asked, the head of the clinic we visited said that there simply weren’t enough cancer patients to require a second clinic. It’s reasonable to ask why – and it wasn’t due to poverty (I asked).
Cambodia is a small, poor nation; the Cambodian diet is high in seafood rice, and vegetables; there’s not much heavy industry (or light industry, for that matter); not a ton of pollution outside of Phnom Penh and industrial areas…reduced exposure to environmental contaminants and processed and unhealthy foods must play a role in America’s higher rates of cancer. Right?
Here’s the thing – we can make a case that Cambodians’ lack of exposure to so many bad things that are present in the US must be responsible for their lower rates of cancer (or for our higher rates) – they certainly seem to be correlated. But then I remember my first day of Biostatistics and what the professor said was the cardinal rule of the field – “Correlation does not imply causation.” It’s not enough to show that Americans have both higher cancer rates and a higher exposure to processed foods and environmental risks – we need to be able to show that they’re linked, and that’s a bit harder to do.
To cut to the chase, the main reason for Cambodia’s low cancer rates seemed to have little to do with diet or of the other factors; the key statistic here was the average life expectancy – just over 40 years in Cambodia and about double that in the US – due largely to a combination of poverty, malnutrition, landmines, and infectious disease. Most Cambodians simply didn’t live long enough to develop cancer because most cancers strike late in life at ages that most Cambodians never reached, and that most Americans do. This mistaken attribution of (in this case) excess cancers in the US to environmental causes is the Ecological Fallacy.
And this takes us to the Harvard papers you asked about!
One study, the one about nuclear reactors, purported to show that people who live closer to nuclear reactors have higher rates of cancer than those more distant. But the authors never confirmed that proximity actually led to higher radiation doses, primarily because they never measured a single radiation dose or dose rate at a single location or for a single person. They didn’t even cite studies of dose rates…but, then, they never really showed that people in the “nearer” category actually lived closer to the reactors because they didn’t look up any addresses from which they could measure distances. Instead, they assumed that everyone in a county lived at the same distance from a nearby reactor, notwithstanding the odd, large, and irregular shapes of many counties. But, to my mind, the cardinal sin of this study is the failure to even try to show a correlation between cancer rates and actual radiation exposures. They just made a few ill-considered assumptions and started speculating. The failure to show that there was a real correlation between actual radiation exposures and rates of cancer incidence makes their speculation (possibly) intriguing, but unconvincing.
The other study, which claims that flight crew have elevated rates of cancer which the authors attribute to elevated levels of cosmic radiation is just as unconvincing. Even the authors acknowledged that the lack of dosimetric data weakens their claims that the incidence of several cancers are higher in flight crew and that this increase must be due to the elevated levels of cosmic radiation received in flight. But, unlike the nuclear reactor paper, the authors don’t seem to have even tried to find a proxy for exposure (e.g. flying hours or number of long-haul and/or polar flights). And, although they reported increased incidence for a number of cancers, only a few of these showed any risk ratios high enough to be likely due to something other than random chance: cancers of the skin, prostate, uterus, and breast with skin cancer being the highest risk ratio. The thing is, none of these tissues are highly sensitive to the levels of radiation to which even flight crew are exposed and the cancers most likely to appear (leukemia and other blood-related cancers) weren’t even mentioned.
Interestingly, a study of cancer in flight crew performed a quarter-century ago[1] also showed very low mortality ratios, with the exception of skin cancers. In this study, though, the authors attributed this increase to disruptions to crew members’ circadian rhythms due to time zone changes and work schedules as well as to a tendency to take extended layovers in sunny locations (e.g. Rio de Janeiro) rather than in cloudier or colder locales.
These findings fly I the face of conclusions reached by Naval Reactors, who actually did measure radiation exposures to their sailors and other workers, finding the risks (if any) over a 20-year career were no greater than the risks from driving. [2]
What really jumps out at me – and possibly explains the authors’ willingness to attribute cancer incidence to radiation without feeling the need to measure or calculate a single actual radiation exposure – could be that not a single author seems to have experience actually working with radiation; there are environmental scientists, physicians (who don’t appear to be in departments that use radiation regularly), and statisticians, all of whom decided to take a quick look in an area in which they lack a deep understanding. On top of that, both papers were published in journals that don’t specialize in radiation and its health effects, raising the possibility that neither reviewers nor editors understood the importance of having actual dose measurements (or at least calculations) when trying to claim that radiation must be causing cancers. The authors appear to have fallen prey to the Ecological Fallacy, and they should have known better.
The bottom line – I’m not going to change my travel plans, nor would I advise any of my family or friends to move away from a nuclear reactor based on anything in these papers. Even very smart people from prestigious institutions can make mistakes – especially when they’re tempted to stray outside their areas of expertise.
[1] Pukkala, E. Incidence of cancer among Nordic airline pilots over five decades: occupational cohort study. British Medical Journal, 325, p 567 (2002)
[2] Office of Naval Reactors. Occupational Radiation Exposure from Naval Reactors’ Department of Energy Facilities (Report NT-19-3). May 2019