The Gold Standard and ALARA

Hi,
I have an MSc in Nuclear Engineering but I have always struggled with the concept of statistical deaths, whether you’re using LNT or SNT, and I think it should be explained better. My understanding is that statistical deaths refers to individuals who have died prematurely, from cancer as a result of acute or long-term radiation. In other words the central estimate for the number of early deaths.

Section 5.5 “Sigmoid No Threshold and Chernobyl” is extremely consequential for people to authoritatively argue about the number of deaths rather than simply saying “2 people died in the explosion, 28 from ARS and 50 to 160 thyroid cancer deaths, but there is no statistically significant difference in the non-thyoid cancer rates post 1986, so there were no additional cancer deaths”. For example, Michael Shellenberg’s, in ‘Apocalyse Never’, ignores the question of what happened to the 128 ARS suvivors and the 200,000 liquidators with an average dose of >100mSV.

As you say in your book, ignoring them is neither defensible or believable because 100mSv is about the threshold at which a statistically significant dose-effect relationship can be inferred. We need the SNT model. But I don’t understand your implementation of it. Please could you clarify some things:

1. MOVING FROM STATISTICAL DEATHS TO LOST LIFE EXPECTANCY
The SNT model in Table 5.2 predicts 1562 induced cancers. You then say “As per Section 4.1, if we want to compare [the cancer incidence numbers] with sure deaths in terms of Lost Life Expectancy, we should divide these numbers by about a factor of two.” We need to have a more authoritative way of doing this

Why? That assumes all these individuals die exactly half way through their lives. Isn’t the average radiation cancer going to occur half way between the average age at which they are irradiated (median age of population) and the average age of death (life expentancy) meaning that each radiation induced cancer results in a a quarter of a life lost, rather than half a life lost in an average aged population. Surely you should say ‘divide by four’ here. Surely there is a more scientific way of doing this though?

2. ARE INDUCED CANCERS, CANCER DEATHS?
You variously use the terms: ‘cancers’, cancer deaths’, ‘delayed cancer deaths’, ‘statistical deaths’ and ‘early deaths’ and you appear to mean the same thing, i.e. “the central estimate for the number of individuals prematurely dying from cancer”. Please could you be consistent in all tables, graphs and text with a single term.
For example:
Table 5.2 compares their analysis of cancers due to Chernobyl with an SNT – then in the table title you present ‘cancer deaths’ – are all these ‘cancers’ cancer deaths?
Table 5.3 present Russian Liquidator Doses and the LNT & SNT cancers – Do these cancers result in deaths?
‘Kashcheev et al estimated that the early deaths…’ – Are ‘early deaths’ cancer deaths?

3.THE DOSE RATE AND NUMBER OF LIQUIDATORS YOU USE HERE IS TOO HIGH
The WHO document “Health effects of the Chernobyl accident and special health care programmes.” (2006) uses the figure of 200,000 Liquidators at a 100mSv dose. Even Cardis et al. 1996, used 240,000 Liquidators at 100mSv.

Why do you use 530,000 Liquidators at a 145mSv dose? I believe you are using the total figure of certified total workers who visited within the 30km Zone, rather than the high dose liquidators of in 1986-7.

I do see your footnote “Overall the UCS dose numbers appear inflated relative to other sources. Compare Table 4.3 with Table 5.2 but we use them for illustration anyway” . But for this reason, it is doubly important for you to make a clear summary table of your predicted deaths from Chernobyl as I suggest in point 5.

4. YOU DON’T PREDICT THE PREMATURE DEATHS FROM THE 128 FIREFIGHTERS WHO WERE ARS SURVIVORS
“Of the on-site personnel and fire-fighters, about 300 had to be hospitalized for burns and the diagnosis of possible radiation injuries. These individuals were observed and given care and, if necessary, specialized
treatment”. From UNSCEAR 1988.

Of these, 237 were originally diagnosed with ARS but ultimately, only 134 emergency workers were confirmed as ARS sufferers after detailed clinical analysis.

The 128 ARS survivors are not accounted for in your analysis even though they all received above 800mSv. Unfortunately I have not been able to find dose estimates for the approximately 150 firefights who weren’t clinically diagnosed with ARS. Its possible they also received high doses.

You should add a Table 5.4 where you include the following data on ARS survivors from Figure VII in UNSCEAR Annex D, 2008 and do an SNT analysis to predict their lost life expectancy too. None of these people died of reasons directly attributable to the radiation:
ARS Grades by Irradiation dose:
Mild (I) 0.8-2.1 Sv: 41 survivors
Moderate (II) 2.2-4.1: 49 survivors
Severe (III) 4.2-6.4: 15 survivors
Very Severe (IV) 6.5-16: 1 survivor

“Their doses were estimated mainly by means of clinical dosimetry methods, i.e., on the basis of blood formula and/or cytogenetic parameters of blood lymphocytes (UNSCEAR, 2000); these methods are applicable to small numbers of human subjects and not for large-scale epidemiological studies.”

5. USING SNT, HOW MANY PEOPLE ARE YOU ULTIMATELY SUGGESTING DIED FROM CHERNOBYL?
You summarise how SNT can be used to predict cancers, but we have to do some additional calculations to figure out the actual estimate you are claiming for the non-thyroid cancer deaths (in addition to the 30 explosion/ARS deaths and approximately 100 thyroid deaths. Your analysis in Table 5.2 makes it look like you’re predicting 1562 premature non-thyroid cancer deaths from Chernobyl using the average dose rates. You need to add a Table 5.5 which summarises your analysis.

Is this summary correct:
Central Estimate of Premature Deaths assuming SNT using Average Dose Rates – :
Evacuees 27 deaths
Liquidators: Not Applicable
SCZ Residents 7 deaths
Other contaminated 3 deaths
Other <1 deaths
Total: 38 deaths

Central Estimate of Premature Deaths assuming SNT using the Stratified Dose Estimates for Russian Liquidators:
Russian Liquidators (67,500): 213 deaths
Crudely Extrapolating to the (200,000) Total Liquidators: 631 deaths

The total lost life expectancy is almost entirely in the liquidator population of 200,000. Assuming an average age of irradiation of 25, and life expectancy of 75, with a mean death age of 50 an average third of each life is lost ie. 210 full lives. This equates to a lost life expectancy to an individual within this 200,000 strong group of 29 days. This contrasts very well with your earlier statement: "according to Cohen, the LLE associated with coal pollution is around 23 days for the average American".

6. ADDRESS THE ACCURACY OF THE LIQUIDATOR DOSE RATES
The level of occupational monitoring was very poor by modern standards and wearing of personal dosimeters was allegedly banned in Medvedev's account. First hand reports from the worst exposed known as ‘Bio-Robots’ assert that they were recorded with far lower doses than the actual conditions on the reactor building roof.

This is attested in:
G. Medvedev, Chernobyl Notebook. Moscow: Novy Mir, 1989.
and
Z. A. Medvedev, The Legacy of Chernobyl. Oxford: Basil Blackwell, 1990.

I think it is worth confronting how the dose estimates for the exemplar group of Russian liquidators was actually arrived at in Kashcheev 2015 because most people will correctly suggest that the Soviets had a incentive to lie about this and a track record of doing so about such things. Kashcheev 2015 gets its data from Pitkevitch et al. (1997). This data is from the Russian National Medical and Dosimetric Registry (RNMDR). It is based on the documents given to liquidators by the over 600 organizations that performed dosimetric monitoring in the zones of operation. It covers 119 416 russian liquidators (78.4% of the total number of the 152 325 russian liquidators).

Pitkevitch et al. performed a plausibility analysis using "different statistical characteristics were derived to assess the reliability of the information." In short they compared the known dose rates in 'recovery operation' locations against where they are recorded have lived during their time at Chernobyl and for what duration.

It notes that
– plenty of liquidators recorded doses which exceeded the official limits, implying a reliability in the data.
-"the distribution of absorbed doses of the liquidators in different calendar years is consistent with the radiation situation in the zones of radiation contamination and with the type of work carried out by the liquidators."and;
– "information of the RNMDR is, as a statistical aggregate, generally consistent with the data on the radiation situation in the RO zones"

It states: "In conclusion the statistical analysis of the doses registered for liquidators in the RNMDR does not bring out major evidence of unreliability.".

But in general, the data is seriously lacking because it doesn't have any information of individual ‘routes’ and duration for liquidators in the zone of radioactive contamination during the course of their work. However, the sheer volume of data does imply reliability as a data fabrication exercise on this scale, would be quite clearly visible in the data. Even when teachers self-report their student's grades, basic statistical methods are able to determine whether it contains statistically implausible data.