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.