The policy Implications of SNT

This piece assumes some familiarity with two radiation harm models, Linear No Threshold (LNT), and Sigmoid No Threshold (SNT). At a minimum, you will need to be familiar with LNT is Nonsense and A Sigmoid No Threshold Primer.

SNT has been attacked from the pro-nuke side on the grounds that, in accepting the premise that there is no perfectly harmless dose, SNT concedes too much. This school holds nuclear's future depends on a threshold dose below which there is absolutely no harm, or better a benefit. This piece argues that combining standard EPA and NRC risk limits with SNT will result in radiation regulatory policies under which nuclear power could thrive, and with it the human race.

The NRC has something they call a Quantitative Health Objective (QHO). For cancer, the QHO is the annual risk of cancer fatality to an individual living within 10 miles of a nuclear plant should not exceed 2 in one million. If we assume an 80 year life, to first order this translates to 160 in one million (0.00016) lifetime risk.

The EPA has a somewhat similar standard which they call Maximum Individual Risk (MIR). The EPA's lifetime MIR for contacting cancer is 100 in one million. Since roughly half of cancers eventually prove mortal, the EPA's MIR is 50 in one million (0.00005) fatalities, about one-third the NRC's QHO.

Both these limits strike me as arbitrary and ad hoc. There's no attempt to balance risk versus benefit.¹ The NRC QHO is is not even operational. It is not at all clear how you would determine compliance. But let's assume like the EPA they really mean the most exposed person, and translate the limits into dose rate profiles.

Table 1 shows the allowable lifetime constant dose rate for a range of MIR's using LNT and SNT to convert dose rate profile into cancer incidence.

Both the EPA and the NRC are observant believers in LNT. In the words of John Rodgers, head of EPA's radiation division, EPA's support for LNT is ``set in stone". The EPA is quite prepared to openly reject evidence that contradicts this policy. Here's the official EPA rejection of the obviously non-linear radium dial painter data.

EPA policy is to assess cancer risks from ionizing radiation as a linear response. Therefore, use of the dial-painter data requires deriving a linear risk coefficient from significantly non-linear exposure data or abandoning EPA policy. \cite{epa-1991}

Abandoning EPA policy was not an option. But I digress.

The EPA's CERCLA superfund clean up level is 0.15 mSv/y. This is very roughly consistent with the MIR. NRC's limit of 1 mSv/y for the public is a bit lax under my interpretation of the QHO. If EPA is given full control of radiation regulation as Ramaswamy and others have recommended, we can expect considerable tightening.

Linear No Threshold (LNT) is the theory that all sieverts are the same. It does not matter if a large dose is received in a few seconds of spread over a few score years. This may be approximately true when it comes to DNA damage. But what we are interested in is cancer incidence, which is the result of unrepaired damage.² It turns out evolution has endowed us with remarkably effective DNA repair processes; but those processes can be overwhelmed if the dose rates are high enough.\cite{bissell-2011}

One way to do that is drop an atom bomb on people, creating a dose rate spike. What we've done is measure the response to that spike, and then assume we will get the same response from the same dose even if it is spread over a lifetime.

Sigmoid No Threshold (SNT) takes a more enlightened approach. Given our undisputed ability to repair radiation damage, SNT claims what counts is the dose in the DNA repair period, which is something less than a day. It then estimates the probability of cancer in each repair period as a function of the dose in that period --- currently set at a day --- using the same atom bomb survivor data on which LNT is based. It then combines all the repair period probabilities to come up with the overall cancer incidence associated with a dose rate profile.

Table 1 shows that, when you convert the EPA's MIR of 0.00005 to lifetime dose rate using SNT, the allowable dose rates are roughly 3000 times higher than the LNT allowable. The annual allowable goes from 0.01 mSv to more than 30. This is consistent with the fact that in high background dose areas such as Kerala people experience dose rates of up to 60 mSv/year with no detectable increase in cancer, Figure 1.

Figure 1. Kerala cancer incidence, 70,000 person study, reference \cite{nair-2009}

Personally, I have a problem thinking in terms of cancer incidence. I find a more meaningful number is Lost Life Expectancy (LLE). On average, contracting a fatal cancer costs an American about 12 years of life. Table 1 indicates that the EPA's MIR is equivalent to losing about a fifth of a day. Bernie Cohen estimated the LLE of a whole range of activities, Figure 2.\cite{cohen-1991} The number that caught my eye is coffee. According to Cohen, coffee drinking has an LLE of 6 days. That tells me I don't worry at all about risks that costs me 6 days or less. For me, the EPA's MIR is overkill. But for the sake of argument, let's accept their MIR as acceptable risk.

Figure 2. Cohen's Lost Life Expectancies (coffee drinking in yellow)

In a nuclear power plant release, the dose rate profiles will not be constant. For the public assuming no evacuation, they will look more like Figure 3, a step jump in dose rate, then a period of rapid decrease, followed by a long period of near constant, slightly elevated dose rate.

Figure 3 is a reconstruction of the dose profile for someone living 2 km NW of the Fukushima plant who did not evacuate and then lives for 40 years in her home.

Figure 3. Fukushima worst public air dose rate profile.

The cumulative dose rate over those 40 years is close to 500 mSv. According to LNT, this results in a lost life expectancy of 100 days, practically all of which is due to the dose received after the first 50 days. But the SNT lost life expectancy is less than an hour, one-quarter of which is due to dose received in the first 50 days. The SNT cancer incidence is 8 in a million, well below the EPA's MIR.

Under LNT the policy is evacuate slowly --- to avoid the social costs of rapid evacuation --- and don't come back. Under SNT, assuming reasonable buffer zones, evacuation is almost never beneficial. In the extreme cases where limited evacuation is indicated, you will be able to return in a matter of weeks.

EPA and NRC rules and policies are order of magnitude consistent with their view of acceptable risk and LNT. In fact, you can make an argument that under LNT the rules could be tightened. Under SNT and the same acceptable risk, these rules and policies are total nonsense, nonsense that has tragically ruinous implications for society and the planet. If we adopted SNT, we could regulate nuclear power very much like we regulate other beneficial, potentially hazardous activities. Nuclear power could realize its promethean promise.

1. The EPA appears to recognize this. They allow themselves the ability to adjust the MIR, presumably because the cost of abiding by the 100 in one million number is judged too high. The EPA has accepted 200 in one million in several cases.

2. For cancer, successful elimination of a damaged cell, either by cell suicide or the immune system counts as a repair.