Comments on ICRP Draft Report for Consultation:
The opportunity provided by ICRP to comment on this draft report is much appreciated. The draft reflects the mammoth effort by Committee 2 task groups INDOS and DOCAL in its preparation and an extraordinary depth of expertise and detail. The comments below fall only partially within the remit of the dosimetric modelling exercise and have more to do with fundamental policy matters for the Main Commission.
The comments relate to Section 12 dealing with radon (pp.204-245). This section is different from the others, as noted in the Abstract, in that it includes proposed numerical values for dose coefficients – in this case, in terms of effective dose per unit exposure to radon progeny. For other nuclides (and for elemental radon) dose coefficient values are yet to be provided. As a consequence, some of the information on which the preliminary radon progeny dose coefficients are based is not yet available: specifically, the dose coefficients for each of the individual short-lived radon progeny in terms of inhaled activity, which are needed to perform the calculation set out in Eq.12-10 (para.718). But this is only a temporary difficulty. There is a much more important issue.
The proposed dose coefficients for radon progeny suffer from a fatal flaw because the treatment presented in this draft does not deal properly with the complicating carcinogenic effect of tobacco smoke. Para.725 states: “…ICRP does not take account of smoking statistics. Thus, it should be recognised that ICRP nominal risk coefficients and dose coefficients apply to a mixed population of smokers and non-smokers”. This astonishingly lax treatment of smoking is out of place in this otherwise meticulously prepared report.
On the one hand, to “not take account of smoking statistics” implies that ICRP regards smoking status as unimportant in estimating a value for the dose coefficient. If that were truly the case, an effective dose derived from application of the dose coefficient would reflect the risk from radon progeny alone. It would be valid for non-smokers and smokers alike.
On the other hand, stating that the dose coefficient applies “to a mixed population of smokers and non-smokers” implies that its value does depend on smoking status and that it reflects the pattern of smoking in a population.
The above alternatives cannot both be true. If, as it seems, the latter is intended, what mixture of non-smokers and smokers does ICRP have in mind? The risk of fatal lung cancer varies by a factor of about 3000% between non-smokers and continuing regular smokers. The pattern of smoking in the population would have a huge effect on such a dose coefficient. Specifically, if the recommended values (11 mSv/WLM for mines and 21 mSv/WLM for indoor workplaces) are intended to apply to some typical mixed population of smokers and non-smokers, they would be several times smaller (in the region of 2 to 3 mSv/WLM) in the absence of tobacco smoke, based on epidemiological studies.
Because effective dose is being used here as an indicator of risk that includes the carcinogenic effect of tobacco smoke, the underlying dose coefficient varies from about 2 mSv/WLM to over 50 mSv/WLM (ball-park figures only) depending on smoking status. In the current ICRP approach, this underlying variation is ‘fixed’ by selecting a particular pattern of smoking in a population: "a mixed population of smokers and non-smokers".
It makes no sense to quote a dose coefficient to two significant figures if it is, in reality, unknown by up to several hundred percent. And it would be unacceptable to use such an ill‑defined dose coefficient to assign individual doses in a regulatory and legal context, which happens routinely in occupational exposure situations.
The problem arises from a fundamental difficulty with the ICRP’s use of the quantity effective dose when applied to exposure to radon and radon progeny. Again, there is a dichotomy, depending on whether the emphasis is on ‘effective’ or ‘dose’.
On the one hand, ICRP wishes effective dose to be regarded as a dosimetric quantity. ICRP emphasises (in Publication 103 and elsewhere) that it is a protection quantity, rather than a metrological quantity, but it is nevertheless intended to embody the characteristics of a ‘dose’: an insult to the body from an external agent, with the quantum of the insult delivered being an inherent property of that external agent
On the other hand, ICRP also uses effective dose as an indicator of risk of harm. The transition from ‘dose as insult’ (absorbed dose) to ‘dose as an indicator of risk’ (effective dose) is made through application of radiation weighting factors and tissue weighting factors. The problem in the case of inhalation of radon and radon progeny is that the risk (predominantly, risk of fatal lung cancer) is not an inherent property of the quantum of radiation, except for non-smokers. It depends on, and is dominated by, another carcinogenic agent: tobacco smoke.
If effective dose from exposure to radon progeny were to be used as an indicator of risk for a ‘mixed population of smokers and non-smokers’, it would need to be redefined to include the risk from tobacco smoke, and the pattern of smoking in the population would need to be specified. This would mean that it could no longer be regarded as an indicator of radiation risk alone, and that it ought not to be used to assign individual radiation doses to workers. Alternatively, if its use as an indicator of radiation risk is to be retained, then the dose coefficients for radon and radon progeny should be derived for a non-smoking reference individual.
The latter option is the more desirable. Effective dose (radiation dose) should reflect the risk of harm from radiation. Where there is a need for protection purposes to consider the joint effect of radiation and smoking, the combined risk (not ‘dose’) from radiation and tobacco smoke should be estimated and used in establishing numerical protection criteria such as reference levels. This achieves the same end as would a redefinition of effective dose and it avoids turning the risk arising from the carcinogenic effect of tobacco smoke into millisieverts.
The alternative – redefining effective dose to include tobacco-smoke risk – is a highly undesirable path to follow. Effective dose for a given exposure to radon progeny would then become inherently variable over more than an order of magnitude depending on smoking status, fixed only by an arbitrary device: selecting a particular pattern of smoking. Further, the risk would be attributed in the public mind to radiation when it is in fact mostly attributable to tobacco smoke. Such a misperception of the origin of risk would significantly distort the optimization of protection.
While the proposed dose coefficients are presented as radiation dose per unit exposure to radiation, they in fact represent a combined tobacco-smoke and radiation ‘dose’ per unit exposure to radiation and per population-average exposure to tobacco smoke. Attributing all of the implied risk to radiation (radon/radon progeny) when it is known to be predominantly attributable to tobacco smoke is unacceptable.
A consequence of taking the path preferred here: estimating the dose coefficient for radon and radon progeny in terms of radiation risk – that is, in the absence of tobacco-smoke risk – is that the proposed values in this draft report need to be reassessed. If the proposed values are interpreted as applying to non-smokers, they are far too high (by several hundred percent) in comparison with estimates based on epidemiology. Until this issue is resolved, there can be no confidence in the validity of the currently-proposed dose coefficients for radon progeny.
For validation of the dosimetric modelling, there should be a reasonable level of agreement with epidemiological studies based on mortality data. While the modelling of absorbed dose to lung tissues may well be accurate, conversion to effective dose using nominal values for weighting factors may need review. Unless some fundamental fault is identified with the available epidemiological studies, they should merit greater reliance than a computer model that includes an unproven conversion to effective dose for alpha particles deposited in lung tissues. In fact, comparing the dosimetric modelling with the epidemiology – if both are assumed to be free of error – provides one way of estimating appropriate weighting factors for this circumstance of exposure.
A different but likely lesser problem with the proposed dose coefficients is that some of the input parameters to the calculation on which they are based – such as the unattached fraction of radon progeny in mines – are now quite old and may not reflect current circumstances. This is acknowledged in para.655, and it is to be hoped that the ultimately-adopted dose coefficients will take into account more up-to-date information as and when it becomes available.
In summary, despite the extensive and sophisticated modelling work evident in the draft, the proposal for dealing with radon progeny remains a step short of satisfactory completion. Before a dose coefficient for radon progeny (or for elemental radon) is adopted, ICRP should consider the following process for resolution:
(a) ICRP should confirm that the quantity ‘effective dose’ is a protection quantity that reflects the risk of harm from ionizing radiation and that it does not knowingly include the amplification of that risk caused by the presence of other carcinogenic agents. To include tobacco smoke risk would result in a misuse of the term ‘dose’ which is universally taken in this context to mean radiation dose.
(b) The dose coefficient for radon progeny should be derived using data applicable to non-smokers. This will avoid the complication introduced by smoking and result in a dose coefficient that reflects harm from radiation only. (The harm arising from the combined effect of tobacco smoke and radiation can be controlled by optimizing protection through the use of reference levels and the monitoring and control of exposure (see (f) below)).
(c) As far as possible, the dose coefficient derived from dosimetric modelling should be based on up-to-date input parameters for quantities such as unattached fraction and equilibrium ratio. (There is nothing ICRP can do about this until new and reliable information becomes available.)
(d) There should be consistency between the dose coefficient derived as in (b) and the dose conversion convention approach (Publication 65) updated to include recent epidemiological data. If there is not (as is currently the case), the dose conversion convention should be preferred in the meantime, as it is based on actual cancer incidence and mortality data. A shift to adopt a dose coefficient could be made subsequently as and when the modelling becomes reasonably supported by the epidemiology. (If ICRP believes that the current modelling is more reliable than the epidemiological approach, it should suggest how the dose conversion convention can be several hundred percent in error.)
(e) ICRP should make clear to its readership the respective magnitudes of harm from tobacco smoke and radon progeny in order that optimization of protection can be based on valid information.
(f) Reference levels for radon concentration and for radon progeny concentration should be derived taking into account the combined risk from radon/radon progeny and tobacco smoke. In addition to ensuring a high level of protection for non-smokers, this will also lead to a reasonable degree of protection for smokers from the incremental additional risk they incur from exposure to radon progeny.
The approach described above will preserve both the objective and the integrity of the ICRP system for protection from radiation.