In 1945, a profoundly sad experiment in public health began when U.S. forces dropped a 13-kiloton nuclear fission bomb on Hiroshima, Japan. Three years later, President Harry Truman ordered the National Academy of Sciences to study the long-term health effects of radiation on roughly 100,000 survivors. (A hundred thousand more perished in the blast and its immediate aftermath.) As the most rigorous research of its kind (no longitudinal study of the Chernobyl disaster's survivors was ever done), the Life Span Study of the Hiroshima cohort now guides almost all responses to major radiation disasters, including the recent near-meltdown at the Fukushima reactor in Japan. Yet its findings seem to have been ignored completely in the breathless reporting, over the past few weeks, of radiation contamination across the United States.
Within days of the tsunami, the nation's potassium iodide pills—which counteract the effects of radioactive iodine—sold out. The Food and Drug Administration banned vegetable and milk imports from provinces near the reactor. Just the other day, the Environmental Protection Agency reported that traces of cesium-137 had been found in milk in Vermont, while elevated levels of other radioactive isotopes were showing up in samples from Phoenix and Los Angeles. And more than a dozen cities have detected radiation in their drinking water. Despite reassurances that elevated levels of other radioactive isotopes in milk and drinking water are not dangerous, some health departments are still advising cautionary measures, like a blanket avoidance of drinking rainwater.
This contradictory advice—don't worry! OK, worry a little …—arises from a fundamental scientific problem: The true health effects of low-level radiation exposure are unknowable, since any study that could identify them would require an impossibly large sample size—in the millions, not the thousands. To understand why requires a simple lesson in epidemiology.
While very high-dose radiation causes immediate illness and burns, the cancer-causing effects of smaller amounts don't appear for many years. Perhaps the best-known example was Sadako Sasaki, who was 2 years old when the atomic bomb code-named "Little Boy" detonated roughly one mile away from her house in Hiroshima. Sadako survived, but after turning 12, she developed unusual purple bruises on her legs and was soon diagnosed with leukemia. (As popularized in the children's book Sadako and the Thousand Paper Cranes, she believed she would be granted a wish if she folded 1,000 cranes but died after folding only 644.)
How do we know whether Sadako's death should be blamed on radiation exposure? The Life Span Study researchers took a very simple approach: Each victim's radiation exposure was estimated based on his or her location during the blast and compared to his or her disease outcome over the ensuing decades. Did more radiation produce more cancers? It turned out there was a pretty clear dose-response for radiation and leukemia. (The data had less to say about whether the bomb caused other kinds of cancer.) For the victims with the worst exposure—those who received a radiation dose of 4 sieverts, or the amount you'd get from having 500 CT scans at once—there was a 20- to 40-fold increase in the risk of leukemia. At 2 sieverts, cases of leukemia were still elevated, but the risk was somewhat lower—a four- to eightfold increase over the baseline. All told, the researchers estimated that the "Little Boy" bomb accounted for 75 additional cases of leukemia among the 100,000 survivors in the cohort, and most correlated with the highest exposure levels.
But the study had little to say about the people who received the lowest doses of radiation. Among survivors who had been exposed to something on the order of 0.15 sieverts, there was no detectable increase of risk. Does that mean that these quantities of radiation, which amount to what you'd get from two dozen CT scans, are actually harmless? Not exactly.
The perils of absorbing such doses might exist, but they could be small enough that you'd need a huge sample of victims to see them show up in the statistics. When dealing with something as uncommon as, say, leukemia (which, under normal conditions, affects less than 0.1 percent of the population over a lifetime), a small increase in risk would add just a handful of cases per decade. If you wanted to prove the risk is real, you'd need a lot of people and a very long study—far more than the 100,000 in the Hiroshima group. When it comes to measuring the effects of low doses of radiation, it's safe to say that the logistics are just too daunting, and we'll never get a satisfactory answer.
In the absence of data, the debate over low-level radiation is largely one of faith. Some researchers believe in the "linear no-threshold relationship." This mouthful assumes that if high doses of radiation impose high risks, low doses must bring their own, albeit lower, risks. On the other hand, some nuclear experts believe that low doses are harmless, and there is indeed a "threshold" below which radiation exposure can be ignored. By analogy, they might say, smoking three packs a day is bad news, but smoking one cigarette a month won't kill you.
Despite the lack of data on radiation exposure, most regulatory bodies conservatively adopt the no-threshold approach. That is why, for example, the Nuclear Regulatory Commission * caps occupational exposure at 0.05 sieverts per year. Such fears have also driven considerable hand-wringing among doctors, who point to the four-fold increase in CT scans since the 1990s and claim that 2 percent of all cancers now are actually caused by the scans. But that's only true if there's a linear relationship between radiation exposure and cancer risk. In other words, it's assumed without any evidence that exposing a million people to 0.00001 sieverts is just as bad as exposing one person to 10 sieverts.
Is there any downside to being so conservative? Consider what happens when people think they've crossed over a "safe" limit. According to a review from the National Institutes of Health, inchoate fears and misinformation in the wake of the Chernobyl disaster led to roughly 100,000 additional abortions among pregnant women and over 1,250 suicides. Meanwhile, outsize fears of radiation risks may lead patients and physicians to avoid necessary X-rays. Last month, the International Commission on Radiation Protection recommended that Japan temporarily raise the annual radiation limit from 0.001 sieverts to 0.02 sieverts per person, and the Tokyo Electric Power Company suddenly raised radiation worker limits to 0.15 sieverts per year. Though the changes are scientifically defensible—since no data exist showing that 0.15 sieverts are dangerous—those who believe in the no-threshold model may assume the safety of citizens and workers has been sacrificed for convenience.
Ultimately, the debate over the presence or absence of a safe threshold is the most basic divide in our society's approach to environmental regulations. Whether we argue over the safety of BPA in infant bottles, lead in old houses, radiation from nuclear accidents, or trace amounts of radon in homes, we're talking about the same thing. Everyone knows lots of lead or radiation or radon is bad; we'll just never know for sure whether the relationship holds at lower doses.
Back in 1972, the nuclear physicist Alvin Weinberg presciently wrote that the dilemmas of low-dose radiation or toxic chemical exposures are "trans-scientific." To decide whether low dose radiation is dangerous would require a research project of monumental, impossible scope. Such questions, he wrote, "can be asked of science and yet cannot be answered by science. [This] elementary point has been lost in much of the public discussion of environmental hazards." The only solution was for policymakers and the public to learn to think in probabilities instead of absolutes. Instead of imposing a black-white dichotomy on suspected toxins, a graded approach—perhaps a gradient with shades of gray?—would make more sense.
Taken another way, Weinberg encouraged debates over public health regulations to acknowledge, and even embrace, uncertainty. He identified the one substance that certainly can be effective even in the smallest possible concentration: humility.
Correction, April 15, 2011:The article originally referred to the Nuclear Regulatory Commission as the Nuclear Regulatory Agency. (Return to the corrected sentence.)