You all are right, the problem with small doses is the main statistical problem. But if you want to calculate risk with LNT model,  you just have to calculate effective dose from the data of your CT (CTDI, pitch factor, DLP). It´s not so easy, therefore you should use tables with typical effective dose factors. After that you can use the standard 6% cancer induction per Sievert. There is a lot of discussion eg in BEIR, ICRP .. about this model, if  matching or not. You can find some questions here in RG

I attach the complete BEIR VII report.

It depends on whether you want to estimate risks for individuals or for populations. Effective dose is not suitable for individualised risk estimates. The weighting factors used to calculated effective dose are age and sex averaged. Moreover, the effective dose for a male patient would assume he had breasts. If you want to estimate risks to populations then effective dose is fine and you can use the ICRP nominal risk of cancer of 5.5% per Sv. So for a CT scan giving an effective dose of 10 mSv, that would mean a risk of about 1 in 2000. This assumes the LNT relationship is correct.

If you want to estimate risks to individuals the best bet would be to use RadRat: (http://dceg.cancer.gov/tools/risk-assessment/radrat). This was developed by the NCI and uses BEIR VII risk models, thus assumes the LNT model. You can estimate risks for individual cancer sites if you have organ doses, or can apply doses to all organs to provide a total cancer risk estimate. You also need to take background rates into account. The rates used for RadRat (I assume) are for a US population, so if you're estimating risks for, for example, an Asian population, the results may be unreliable.

The LNT model is not scientifically validated at the low end of the dose range. It is used for radiation protection purposes, lacking a superior alternative.

Knowing that it is poor science, but lacking good alternatives, the CT-data as mentioned by Hanno can be used to calculate an estimate for the effective dose [here too there are uncertainties].

We already know a lot about other factors that seriously affect the individual risk for cancer due to radiation. So the figure calculated as described above is a risk estimate that is valid at the population level only [with a lot of uncertainties], but that may grossly underestimate the risks if one such other riskfactor actually is present.

For example already having a BRCA I or BRCA II mutation may very strongly affect the individualized risk estimate.

In fact the population risk estimate may very wel be dominated by the highly increased risks for the unhappy few with other risk factors. And the radiation risk for the majority without those other riskfactors may be very much lower than the population figure,

Therefore, any communication to individuals [even if not yet as a patient] should bear this uncertainty in mind. These aspects should be part of any risk communication at the individual level.

And especially for those unrecognised with such riskfactors, running a very tight ship with respect to radiation protection is of the utmoste importance.

This way one can explain why intrinsically we [doctors and physicists alike] can NOT give reliable truely individualised risk factors for low dose radiation.

The risk for stochastic effects due to radiation strongly depend on the presence or abcense of one of several such riskfactors [and we keep finding more of them every year!!]  

These effects are likely to be multiplicative and not additive.

So if you hear someone telling:  my mother died early of breast cancer, 2 aunts did so too, my brother died of pancreatic cancer, you know there is a high likelyhood that one such riskfactor is involved. For someone with such a history, population risk estimates are incorrect and are better not communicated without a very big CAVEAT.

This is not a very rare situation. In my hospital I attend various meetings on patient care for cancer patients on a weekly basis and we routinely encounter several cases every week.

This type of information must enter any discussion about individualized risk estimates.

In fact several of such riskfactors have a much, much larger impact on the individual risk than any diagnostic radiation dose.

@Richard,

you are right, effective dose describes average risks of a population. But a short remark to breast and males. The new effective dose concept of course takes into account the differences between males and females. Both can develop breast cancer. IRRP 103 shows a distinct algorithm. And you are right, typical asian populations are overestimated for the risks to develop breast cancer because of the now finally matched risk factors in effectve dose calculation.

Regarding the individual risk is much more complicated, not only because of age but also because of different repair ablities, gender, etc. It´s also complicated to calculate realistic organ doses. And last but not least, determing some risk factors what ever model you use (LNT, more realistic models), you cannot calculate numerically the benefit ot the CT-examination or any other radiological procedure.

The Beir report does not yet take into account that the mortality for breast cancer in males is dramatically higher than for females [after adjustment for stage at the time of diagnosis] [Their data does take into account the much lower incidence][outcome in women has significantly improved in the past 10 years, so riske stimates should be adjusted [lowered]].

With regard to the benefits of diagnostic procedures: as of yet the only thing we have is for treatment procedures: the number of years 'won' adjusted for a subjective measure of quality [i.e. unconscious at a respirator = quality 0, actively playing with your grandchild not hindred by any  handicap = quality 1].

When making judgements about diagnostic procedures, one has to weight grossly dissimilar risks and outcomes. For that subjective judgement cannot be avoided.

For example: 18F-FDG PET/CT is frequently used in lungcancer staging. It involves somewhere between 5 and 15 mSv of dose.

But if you have lungcancer, you are highly unlikely to survive for more than 5 years.

If I was that patient: the weightingfactor for any risks for stochastic effects would be near zero.

In fact: Thanks to FDG-PET/CT one has a [typically] 30% chance of not having to undergo mediastinoscopy [compared to prevous diagnostic schemes] [which means risk of narcosis / surgery / infection] and in addition: A lowered probability of subsequently undergoing major surgery without benefits [missing less metastases before surgery]. So from the quality of life side of things it is superior.

These are benefits that so far outweigh any risk of radiation from diagnostic procedures, that it can safely be stated that the introduction of FDG PET/CT has resulted in a major risk reduction for the vast majority of these patients.

This effect is so large that the diagnostic radiation dose does not in any way plays a role in the decisionmaking, as long as you apply this type of scanning in the right patient group.

The continuous emphasizing of radiation risks IN ISOLATION proves [to my opinion]  that the various pressure groups doing so, seem to forget the purpose of medical imaging: we need it to aply powerful therapeutic techniques that are themselfs also not risk free, to achieve a significant net benefit.

Surprisingly, no one questions whether doctors request diagnostic procedures for the right reasons. No one questions whether the information on the procedure requests are optimal for the performance of such diagnostic procedures.

The risk figures abstracted from assessing radiodiagostic procedures are just plain noise compared to most other medical risks involved.

To me it is o.k. to demand optimization of medical usages of radiation, or to ask for further insight, but it is not ok, and to my opinion unethical, to overemphasize radiation risks from medical diagnostic procedures and discuss them in isolation, without mentioning the direct medical benefits.

Before the x-rays were invented a large proportion of the population limped, as a result of poor healing of fractures and wounds.

If you ask the population: what is better: now an X-ray and a very very tiny increase in cancer 40 years from now, or having to walk with a limp for the rest of your life, over 99% of the population will have that X-ray gladly.

So I would want to change the subject into:

Why do we put this much emphasis on various scientifically strongly debatable risk estimates for a human activity that is clearly and far beyond discussion beneficial to mankind ? 

We do attempt to optimize it, both technically and with respect to referral policies. The beneficial effects far, far outweight the risks. The beneficial effects are usually seen on the short term the rare sideeffects are virtually always seen after a many year delay.

Why do we keep on overemphasizing an unproven and highly debated theory as the LNT?

This asks for an answer from disciplines other than medicine or physics.

And to my opinion it is an answer we, as a society with generalised fear for radiation, desparately need good answers for.

The cost / benefit ratio of medical use of radiation in diagnosis is already so good that the question on this sociological mechanism is far, far more important than yet another debatable risk estimate for a diagnostic procedure involving radiation.

I am not in the least opposed to a critical view and I do know there is room for improvement, but this over emphasis on radiation risks evaluated in isolation has to end.

Dear Judocus,

thanks for the clear arguments about benefit. You are right, and my personal is similarview  to yours. If I need help to survive I prefer X-Ray examination, eg within a cardiac intervention or a polytrauma accident.

These arguments about the cost benefit ratio are perfectly valid providing there really is a disease or injury that needs treating. But what about the screening of asymptomatic individuals or  situations in which 'there's probably nothing wrong but we'll just do a CT scan to rule anything out'? Is there a danger in these cases that we're creating an unacceptable future cancer burden on society?

Dear Richard,

thats exactly the point. We should differentiate between screening techniques like mammography screening, TB etc and the indicated examination in a special case. In Germany screening must be approved. And the concerned circle is the population. Effective dose could be right even with all limitation of risk models.  Abusus of X-ray examination because of laziness of the examiners must be avoided and restricted.

At my department [Nuclear Medicine] I have not seen a single request that even vagely resembled the "there is probably nothing wrong ........."line of argumentation, during the past 5 years.

What we do see is:

Patient has complained C and based on that [and usually some other data and a physical examination] a study S is requested [increasingly often according to a protocol or a guideline]

With protocols and guidelines, the choice of doing or not doing a certain diagnostic test is based on the severity of all possible diseases and the probability of finding them.

At the time of ordering the referring physician does not know the diagnosis for certain and thus accepts the probability of a negative result.

To the patient this is explained as: "I think you have disease A, but I am not certain enough, so I must rule out diseases B, C and D, using tests X, Y and Z and especially for disease C and test X it is better to do that now, better safe than sorry."

This may sound similar to you as 'there's probably nothing wrong but we'll just do a CT scan to rule anything out', but it is a marketly differnt use:

Test X is used for risk stratification, which is perfectly legitimate here.

We know a number of alarm symptoms that necessitate imaging studies to rule out a serious albeit rare condition. Then it is acceptable to perform such test even though in a large proportion of cases no therapeutic response will follow.

What we do ask or verify is that any action will be taken in case of serious finding.

Example: A large mass is seen on the chest X-ray of cachectic 90 year old man in the left lung. He is already oxygen dependent and extremily unlikely to be operated upon, or to receive radiotherapy or chemotherapy, because of very poor general condition. The protocol prescribes staging with FDG PET/CT [In Holland at least 85% should undergo PET within 3 weeks after the chest X-ray].

Such a patient will be refused FDG PET/CT because staging is considered meaningless, as there are no therapeutic options left.

@Hanno

Screening is an issue here too. In fact a considerable number of dutch people get screened in germany for reasons, that are illegal in Holland and the insurance companies [Krankenkasse] complain that they have to pay for follow-up diagnostic procedures for innocent findings [the so called incidentaloma's].

I see my share every year.

And the sad thing is: even the literature on screening shows considerable publication bias. There is under reporting of negative study outcomes. Even today in many western countries breast cancer screening for certain age groups is continued because of political pressure, not because of scientific evidence.

Commercial enterprises do their best to influence screening policies in their interests, as does all of the pharmaceutical industry.

In Holland there are good reasons to reform screening for cervical cancer and to vaccinate young women against HPV, but the journalists and politicions make it almost impossible to do the wise thing.

Thee people turn a screening 'thing' into a civil rigth instead of a difficult and carefully balanced weighting of pro's and con's. Such balancing is not a static thing but things change. I.e. for cervical cancer we have now got the vaccination against human papilloma virus, so the cost/benefit relation has changed.

I agree with the previous answers. The LNT is simple to apply, but you must calculate effective dose. The calculation may be required for some given purpose, but there is no basis for accepting the calculated risk as valid, because the LNT is not valid. The LNT is assumed for radiation protection purposes. The model assumptions are for individuals if the individual is assumed the average member of the general population. There is no method for applying the model to a specific individual.

The LNT model is not appropriate for risk benefit analysis. The LNT deliberately overestimates the risk and the uncertainty in the model is many times larger than the projected risk.

I agree on most Things that have been stated previously here. I suggest that one use the dose calculators available, but use the information wisely according to what other have stated here regarding the LNT model and individual risk. It is always good to keep track on the dose Levels you are giving Your patients during CT exams (and other exams using ionizing radiation). That can be a good fundament for Optimizing Your radiation Levels in Your hospital. But in the end, it is the justification of the examination that is the most important factor.

It may be better if risks can be more directly derived from CT studies. In this regard, there are several large CT studies such as Australian and British, but with limited information on predisposing conditions (the need to do CT scans and preexisting cancer), the number of examination repeats, individual dosimetry, etc.

Australian study

https://www.ncbi.nlm.nih.gov/pubmed/23694687   (PDF freely downloadable)

British study

https://www.ncbi.nlm.nih.gov/pubmed/22681860    (PDF freely downloadable)

For more details, please see

http://www.bmj.com/content/346/bmj.f2360/rr/648506

https://www.ncbi.nlm.nih.gov/pubmed/25816281   (PDF freely downloadable)

An update of the British study by Berrington de Gonzalez et al in the British Journal of Cancer attempted to address the concerns about predisposing conditions and pre-existing cancers (also freely available)

https://www.ncbi.nlm.nih.gov/pubmed/26882064

The US National Council on Radiation Protection and Measurements (NCRP) has just published Commentary No. 27 “Implications of recent epidemiologic studies for the linear nonthreshold model and radiation protection”.

Commentary No. 27 prepared by Scientific Committee 1–25 updates NCRP Report No. 136 “Evaluation of the linear-nonthreshold dose-response model for ionizing radiation” published in 2001. Commentary No. 27 provides a comprehensive review of recent relevant epidemiologic studies with quantitative dose-response analyses, and concludes that based on current epidemiologic data, the LNT model should continue to be used for radiation protection purposes.

https://www.ncrppublications.org/Commentaries/27

Interestingly the risk is not static buy dynamic and depends on age. For children, younger females and even younger males, it is 2-4 times more than aged individuals. So it can be as high as (15-20)%/Sv rather than 5 %/Sv.

The use of ICRP’s detriment-adjusted nominal risk (in Publication 103: 4.1%/Sv for the adult population at age 18-64 years; and 5.5%/Sv for a whole population at age 0-85 years) in estimating risks in individuals or subgroups is not intended. Estimates of lifetime cancer risk following early childhood exposure (and prenatal exposure) may be a factor of 2 to 3 times higher than those for a whole population, but UNSCEAR recommended in the Annex B of the 2013 Report that generalizations on the risks of effects of radiation exposure during childhood be avoided.

The following two summary articles for NCRP Commentary No. 27 “Implications of recent epidemiologic studies for the linear nonthreshold model and radiation protection” have been published

[In English]

https://www.researchgate.net/publication/326387649_Implications_of_recent_epidemiologic_studies_for_the_linear_nonthreshold_model_and_radiation_protection

[In Japanese (with the abstract and a table both in English)]

https://www.researchgate.net/publication/325036776_Outline_of_NCRP_Commentary_No_27_Implications_of_recent_epidemiologic_studies_for_the_linear_nonthreshold_model_and_radiation_protection