The use of effective dose for medical procedures is inappropriate

The quantity “effective dose” was originally introduced as a way to quantify the potential detrimental stochastic (cancer and hereditary) effectiveness of nonuniform radiation exposures of populations of workers and the general public for radiation protection purposes. It was not intended to be used to represent patient exposures, yet over the past decade, it has become commonplace to specify doses to patients and patient populations undergoing imaging procedures in terms of effective dose. It has been proposed that this is not appropriate, and this is the premise debated in this month's Point/Counterpoint. In 1991, the International Commission on Radiological Protection (ICRP) published ICRP 601 to replace the Recommendations published in 1977 as ICRP 26. Among the changes, “equivalent effective dose” was called “effective dose,” a term retained in the latest ICRP Recommendations, published in 2007 as ICRP 103.2 Effective dose is a risk-related radiation protection quantity designed to take into account the radiobiological effectiveness of different types of radiation at low doses and dose rates and the contribution of these risks in individual organs and tissues to overall detriment from stochastic effects such as cancer and hereditary effects. Over the years, as the knowledge on radiation effects on humans improved, the number of organs/tissues considered and the numerical values of the tissue-weighting factors used in calculating effective dose changed. However, the concept and its intended use did not change. Effective dose is to be used in planned exposure situations to show regulatory compliance with dose limits and constraints for workers and the public. It is applied to a reference person and it was never intended to provide a measure of risk to individuals. ICRP 1053 clarified this by stating that effective dose is not appropriate for medical exposures because “The age distributions for workers and the general population (for which the effective dose is derived) can be quite different from that of the overall age distribution for the population undergoing medical procedures using ionizing radiation, and will also differ from one medical procedure to another depending on the age-and-sex-prevalence of the individuals for the medical condition being evaluated.” In spite of these caveats and published uncertainties of more than 40% for a reference patient population,4 the scientific literature, including AAPM reports, abounds in the use of effective dose for patients, regardless of patient age and whether deterministic effects may be present, such as after radiotherapy. Brenner5 reported that “less than 1/3 of the 2008 PubMed citations on radiation ‘effective dose’ refer to radiation protection, the rest are for clinical patient dosimetry.” Furthermore, many authors do not specify the set of values used, thus making intercomparisons across publications often meaningless. In recent years, the use (or misuse) of effective dose has been debated, and an alternative quantity called “effective risk” has been proposed.5 While disagreeing on which set of risk factors better express health detriment, all the authors debating this issue agreed that effective dose should not be used for medical exposures!5,6 What is wrong with just determining organ doses for which measurement methodologies7 and risk estimates8 are readily available? Deterministic effects in medical imaging are rare, with serious skin burns currently estimated to occur in only of all interventional radiological procedures.9 Accordingly, radiation risks to patients exposed to ionizing radiation in medical imaging examinations primarily relate to the stochastic processes of carcinogenesis and the induction of genetic effects.2 The effective dose quantifies the (approximate) amount of radiation that a patient receives in a radiological examination and is directly related to the stochastic risk.10,11 Effective doses can be obtained for any type of radiological examination including radiography, fluoroscopy, CT, and nuclear medicine.12,13 Effective doses are employed by the medical imaging community to understand the amount of radiation used in radiological examinations and appreciate the significance of this radiation exposure, as well as for optimizing protocols so that patient radiation risks are minimized. In radiological imaging, it is essential that practitioners understand how much radiation a patient may receive from any given procedure. A chest CT examination (effective dose 5 mSv) makes use of about 100 times more radiation than a chest x ray (effective dose 0.05 mSv). A ventilation perfusion scan performed to investigate a possible pulmonary embolism (effective dose 2.5 mSv) uses about half the radiation of a chest CT examination. No other radiation dose parameter (e.g., organ doses and energy imparted) comes close to conveying the information that is encapsulated by the effective dose. The effective dose is also used to quantify natural background radiation exposures and for regulatory purposes. In the United States,14 effective doses from any radiological examination can be compared to those from ubiquitous natural background radiation , average radon exposures , as well as regulatory dose limits for occupational exposure (50 mSv/yr) and members of the public (1 mSv/yr). Comparing natural background and regulatory effective doses with effective doses from diagnostic tests helps put medical exposures into an appropriate perspective. It is possible to attempt to convert effective doses into (approximate) radiation risks. A uniform whole body dose of 100 mGy, which corresponds to an effective dose of 100 mSv, has a cancer risk of in 30 yr old males and in 30 yr old females.15 Risks in young children would be higher, and in older individuals would be lower. Such radiation risk estimates in diagnostic radiology may be compared to other hazards in medical imaging (e.g., use of iodinated contrast agents), medicine (e.g., surgery), or everyday life (e.g., of dying in automobile accidents). Optimization of diagnostic imaging involves finding x-ray techniques that offer the lowest patient dose when image quality is kept constant. Plotting effective dose as a function of x-ray tube voltage (kV) at constant image quality in CT permits identification of the kV value that minimizes patient risks. Importantly, CT optimization using alternative dose metrics (e.g., or ) has been shown to be inappropriate.16 My experience in medical imaging convinces me that the effective dose is (by far) the most appropriate way to quantify the “amount” of radiation patients receive in any radiological examination, as well as explain the “significance” of such exposures.17 I also believe that there are no alternative metrics that could meet the current needs of the medical imaging community, as outlined in this Opening Statement, with the simplicity and succinctness of the effective dose. I agree with Dr. Huda that “in radiological imaging, it is essential that practitioners understand how much radiation a patient may receive from any given procedure.” Indeed, effective dose has been used to quantify stochastic risk in many radiological procedures, as Dr. Huda so aptly documented. But to use publications from the NAS/BEIR,8 the ICRP,2 and the NCRP14 to support his claim that “there are no alternative metrics that could meet the current needs of the medical imaging community,” is misleading. The BEIR VII Report8 calculates cancer risks from organ, not effective, doses. The ICRP Report 1053 recognizes the role effective dose may have, but advises caution regarding the referred population: “Effective dose can be of value for comparing doses from different diagnostic procedures and for comparing the use of similar technologies and procedures in different hospitals and countries as well as the use of different technologies for the same medical examination provided the reference patient or patient populations are similar with regard to age and sex.” Many calculations are done not only disregarding the latter consideration but also, if comparisons are performed using different values, significant differences may result. To illustrate this point, NCRP 16014 calculated the effective dose resulting from a mammogram consisting of two views of each breast and a mean glandular dose to the total breast tissue of 1.8 mGy per view. Using a for breast of 0.05 (ICRP 60),1 the effective dose was 0.18 mSv; using a of 0.12 (ICRP 103),3 it was 0.42 mSv! I fully agree with the ICRP that “for planning the exposure of patients and risk-benefit assessments, the equivalent dose or the absorbed dose to irradiated tissues is the relevant quantity.”3 This approach permits both deterministic and stochastic risks to be quantified. Dose thresholds for deterministic effects are known (mainly) from radiotherapy experience,3 and BEIR VII has calculated stochastic risks for many organs/tissues exposed to low doses of low LET radiations.8 Also, there is no reason why equivalent doses from internal and external exposure cannot be added for a given organ.18 Finally, if the purpose is to reduce patient dose, effective dose is not needed. Diagnostic reference levels are always expressed in machine parameters, such as incident air-kerma for radiography/fluoroscopy and CT air-kerma index and air-kerma length-product for CT.7 Consider a retrospectively gated coronary computed tomography angiogram (CTA) examination performed on the author of this Rebuttal, who is a 59 yr old male weighing 88 kg. This examination on a representative 64 slice CT scanner would likely use a of 60 mGy and be 17 cm long. Doses in 88 kg patients are lower than doses in the 70 kg phantom19 used by the ImPACT CT Patient Dosimetry Calculator.20 Accordingly, my organ doses from this examination would be 59 mGy to the lung, 13 mGy to the red bone marrow, and 16 mGy to the stomach, with dose values for 21 additional organs. I do not believe that such a list of a total of 24 “organ doses” is either required or helpful to medical imaging practitioners. Combining these 24 discrete organ doses in Walter Huda's coronary CTA according to the ICRP 103 rules shows the effective dose to be 23 mSv, which is very informative. This cardiac CTA, for example, results in an exposure that is three times higher than for a cine cardiac catheterization.13 One can easily convert this effective dose of 23 mSv into a cancer induction risk, which is independent of organ tissue weighting factors.20 My hypothetical cardiac CTA would have a cancer incidence risk estimate of 0.087%,20 with quantifiable contributions from the lungs (60%), red bone marrow (12%), stomach (4%), and liver (4%). In summary, effective doses are not risks per se, but a practical way of dealing with nonuniform doses in medical imaging. When necessary, effective doses for any type of radiological examination can easily be converted into radiation risk taking into account patient demographics. I therefore have little doubt that the effective dose will (rightly) continue to be one of the most important patient dose metrics in medical imaging.