| HOME | HELP | FEEDBACK | SUBSCRIPTIONS | ARCHIVE | SEARCH | TABLE OF CONTENTS |
| ||||||||||||||||||||||||||||||||||||||||||||||||
There appears to be no disagreement that the concept of effective dose lacks "scientific rigour"; nor is there any disagreement that the concept of effective dose is "often misused". Given that it is agreed that some quantity is needed in order to compare different partial-body exposure scenarios, the issue is what to do about this.
The distinguished members of International Commission on Radiological Protection (ICRP) Committee 2 argue cogently to keep the status quo, suggesting that these problems are "inevitable". By contrast, a counter suggestion [1] is to replace "effective dose" (i.e. summed organ doses, each weighted with a set of committee-generated numbers) with effective risk (i.e. summed organ doses, each weighted with actual epidemiologically based cancer risks). The logic is that effective risk would perform all of the comparative functions that we agree are needed, but (i) would eliminate the subjectivity associated with committee-generated weighting factors, (ii) would provide a more intuitively interpretable quantity relating to risk, leading in turn to (iii) less potential for misuse.
Specifically, the currently used effective dose is defined as
|
|
where HT are the tissue-specific equivalent doses in tissues T and wT are committee-defined dimensionless tissue-specific weighting factors. The proposal is simply to replace the subjective committee-generated tissue weighting factors, wT, with objective epidemiologically based organ-specific radiation-induced cancer risk estimates. The resulting "effective risk" is thus:
|
|
where rT are lifetime radiation-attributable organ-specific cancer risk estimates (per unit equivalent dose to tissue T). The effective risk, R, is thus a generic lifetime radiation-attributable cancer risk.
It can be seen that the two equations are structurally the same, so that effective risk would fulfil all of the same comparative functions as effective dose. But it would have the major advantages of objectivity, interpretability and less potential for misuse:
Objectivity
The suggestion [1] is that the tissue-weighting factors should no longer reflect subjective committee-generated judgements, but would be objective epidemiologically based quantities. That the ICRP tissue weighting factors are highly subjective can hardly be disputed. For example, it is not really the whole story to suggest, as do Menzel and colleagues, that the changes in ICRP tissue factors every decade or so reflect "relevant scientific advances". To take one of the major changes in tissue-weighting factors adopted by the ICRP in 2005 [2] — the increase in the weighting factor for the breast — this change was not so much made because our knowledge of radiation-induced breast disease advanced in the intervening 10 years, but in significant part because of a committee-based change in policy, now relying more on cancer incidence rather than cancer mortality.
This is not to suggest that the use of cancer incidence, or cancer incidence adjusted for mortality, is not reasonable, but rather to illustrate that the changes in tissue-weighting factors every decade or so reflect, to a considerable extent, different groups of people making different subjective judgements.
Another example of the subjective nature of present tissue-weighting factors relates to the fact that they currently represent a balance between radiation-induced cancer and hereditary effects. Prima facie, as Menzel and colleagues suggest, this is not an unreasonable goal. But any methodology for combining the risks of cancer and hereditary effects into a single number must be inherently subjective, again resulting in weighting factors that change not so much because of the science changes but because of committee changes.
This is the basis for the suggestion that objective epidemiologically based radiation-induced cancer risk estimates be used for tissue-weighting factors — the point being to remove the inherent subjectivity in their determination. Such objective weighting factors might still change over time, but in such a case it would really be because of "relevant scientific advances".
Interpretability
The goal is to have a generic quantity reflecting radiobiological detriment or risk, so why use a quantity that has units of Sieverts? A major advantage of the use of Equation 2 is the desire to have a quantity that is more directly interpretable as a risk. It is surely true that an effective risk of (say) 4 per 100 000 individuals is intuitively interpretable to the user in a way that an effective dose of (say) 1 mSv is not. As we struggle with the rapidly increasing man-made contribution to the overall population exposure [3], it is surely advantageous to have a measure of the radiological detriment that actually means something to most users.
Less potential for confusion
The confusion between organ dose and effective dose is widespread in the field of radiology, and it is probably true, as Menzel and colleagues suggest, that this confusion is "inevitable" if we stick with a quantity which (i) has dose in its name, (ii) has units of dose, but (iii) is actually a measure of radiological detriment. The confusion would be entirely avoided if measures of radiobiological detriment were in units of (for example) "per 10 000 individuals" (as in effective risk), rather than in Sieverts (as in effective dose).
Radiation protection vs patient dosimetry
Menzel and colleagues correctly point out that the original motivation for the effective dose concept was for external radiation protection, not for medical patient dosimetry. Perhaps one could make the case for the continued use of effective dose in non-medical radiation protection situations: here, the relevant population is often over 18 years of age and under 70 years, and so ignoring the differing age sensitivities of different organs may be acceptable. But the predominant use of effective dose is now for medical patient dosimetry: for example, less than one-third of the 2008 PubMed citations on radiation "effective dose" refer to radiation protection. The rest are for clinical patient dosimetry. For clinical applications, where we are particularly concerned about paediatric exposures, one cannot justify the use of age-independent weighting factors.
Effective dose, effective risk and linearity
Menzel and colleagues suggest that effective risk (Equation 2) somehow presupposes a linear no-threshold (LNT) risk model, in a way that effective dose (Equation 1) does not. This is not the case. In fact, both quantities are equally based on LNT, as can easily be seen from the equations. The confusion here lies with the fact that, in the effective dose framework, a radiobiological detriment or risk is expressed as a dose — and how could a dose contain hidden assumptions about LNT? But a risk by any other name is still a risk and, as the dose goes down, the effective dose goes down too, linearly and with no threshold. The fact that even such distinguished colleagues can sometimes get confused regarding effective dose is probably as strong an argument as one could muster that the concept is deeply flawed.
In summary, for radiation protection, one could perhaps make an argument for the continued use of effective dose, flawed and confusing as it may be. In practice, however, effective dose is now largely used for patient dosimetry, and there its use cannot be justified.
Yours etc.,
Center for Radiological Research, Columbia University Medical Center, 630 West 168th Street, New York NY 10032, USA. E-mail: djb3{at}columbia.edu
Received for publication December 3, 2008. Accepted for publication December 12, 2008.
References
Related articles in BJR:
| ||||||||||||||||||||||||||||||||||||||||||||||||
| HOME | HELP | FEEDBACK | SUBSCRIPTIONS | ARCHIVE | SEARCH | TABLE OF CONTENTS |
| BJR | DMFR | IMAGING | ALL BIR JOURNALS |
