Low-Dose Radiation Exposure and Risk of Cancer
by William R. Ware, Ph.D.Emeritus Professor of Chemistry, University of Western Ontario
The conventional wisdom holds that there is no threshold associated with the dangers of radiation. Any radiation above the ambient or background levels is considered to increase the risk of cancer and the relationship is regarded as linear. The risk is in direct proportion to the dose, starting at zero excess exposure. Thus this view involves the belief that there is no level below which exposure is safe. This is in fact called the Linear-No-Threshold Model or LNT model, a model of great utility since it allows extrapolation from data collected for high dose exposure way down to modern levels of diagnostic radiation. The high dose data is primarily derived from extensive studies of the survivors of the atomic bomb exposure in Japan with doses estimated according to the distance from the epicenters of the explosions. The bomb radiation was primarily from, x-ray and much higher energy gamma-rays. Both are so-called electromagnetic radiation which includes microwave and infrared radiation and visible and ultraviolet light. Gamma rays and x-rays also described as ionizing radiation.
Why is this important? Modern medicine makes ever increasing use of imaging with x-rays and radiation from infused radioactive isotopes, both for primary diagnosis, whole-body CT screening and for guiding invasive procedures such as angiograms, stent placement, etc. In addition, there are concerns regarding radiation associated with several aspects of the nuclear power industry and even worry about terrorists using so-called dirty bombs. If it is indeed true that any level of exposure exceeding natural background radiation increases the risk of cancer, then there is reason to limit all exposure no matter what the source. If the LNT model is in fact not correct and there is a threshold, then the conventional wisdom may be creating unnecessary concern and unjustified avoidance of diagnostic and screening procedures. It may also cause unjustified anxiety among patients who have had or are facing radiologic diagnostic procedures. If the LNT model is incorrect, then studies which use this model to predict future cancer incidence due to, for example, CT scans, are providing misleading guidance. While we have all been led to believe in the LNT model and its implications (although we may not have realized what the model was called), as will be discussed in this review the conventional wisdom has been repeatedly challenged. In fact, there are growing indications regarding the lack of evidence for and validity of the LNT model and the predicted dangers of low-dose radiation.
Radiation doses are generally expressed in the units of the Sievert (Sv) or the Gray (Gy). While the details of these units need not concern us, it is necessary to refer to dose levels and both are used in the context of diagnostic exposure. To keep this discussion simple, the millisievert (mSv, one thousandth of a Sievert) will be used and the Gy will be taken as equivalent to the Sv for x- ray and gamma ray radiation. To provide perspective, the following are the typical modern exposures per procedure for medical diagnostic application, given in units of mSv. Chest x-ray: 0.02-0.04; thoracic spine x-ray: 0.4; barium swallow x-ray: 1.5; barium enema x-ray: 7; mammography x-ray: 0.7; dental examination x-ray: 0.02-0.09; abdominal x-ray: 1.2; CT head scan: 2; CT chest or abdominal scan: 8-10; angioplasty procedures: 8-57; coronary angiogram: 5- 16; technetium-99 nuclear medicine scan: 2-7 (Source--Health Physics Society Website, www.hps.org, fact sheets). Incidentally, so-called electron beam tomography used for coronary calcium scans does not irradiate the patient with high-energy electrons but is an x-ray based CT procedure. The name is derived from the manipulation of the electron beam inside the x-ray tube in order to create a scan.
The above are approximate values and are dependent on the equipment used. But the general range is what is important for understanding the subsequent discussion. Radiation therapy involves much higher doses, generally well above 1 Sv and frequently above 10 Sv with multiple exposures. These numbers can be put in perspective by recognizing that in the U.S. people are exposed on average to background radiation level of about 3 mSv annually. However, most of this natural exposure comes from radon gas which deposits radioactive decay products in the lungs which then emit alpha-particles and gamma rays. Alpha particles have a somewhat different mechanism for inflicting radiation damage than ionizing radiation. Dose rates or accumulated doses associated with radiation therapy, in particular for cancer, are in a range where there is evidence of radiation carcinogenesis. This review is concerned with much lower doses.
Two recent reports from respected sources relate to the issues raised above. One originated as a joint effort of the French Academy of Medicine and the French Academy of Sciences and was published in 2005. A summary is available.1 The other report, also published in 2005, was from the U.S. National Academy of Sciences and is referred to as the BEIR VII report. Both address the question of the cancer risks associated with low dose ionizing radiation. The authors of the French report conclude that epidemiological studies have been unable to detect a significant increase in cancer incidence in humans for doses below about 100 mSV. This dose level is clearly above all common diagnostic, screening, and intervention associated radiation exposure. The French also addressed the question of the validity of the LNT model. The report points out that the studies used to justify the LNT model involved A-bomb survivors and individuals exposed to radiation in the workplace and that the levels of exposure were in the range of 125 to 500 mSv. If the studies are restricted to those receiving less than 100 mSv, they maintain that the excess cancer rates can not be determined, either because of the problems in statistical significance associated with the fact that the number of individuals included is too small, or because there is a lack of a carcinogenic effect at low doses. In taking this position they reject as not significant several studies that are frequently quoted in support of the LNT model in investigations attempting to estimate risk of developing cancer from medical diagnostic exposure to radiation.2
The conclusions of the French Report were challenged by Brenner and Sachs.3 They present essentially two arguments. One is based on studies of the cancer in offspring associated with in- utero exposure. The French Report considers the evidence of in-utero carcinogenesis produced by low doses to be doubtful. It is interesting in this context that an increased risk of leukemia was not found in the offspring of Japanese women pregnant at the time of exposure to the bomb radiation. Also, the principal concern is with adult radiation exposure and most critics of the conventional wisdom are willing to concede that there are potential risks for unborn and young children. Their second argument is based on what they regard as the absence of extensive understanding of the mechanisms whereby low-dose radiation could offer protection. This effect is called hormesis or the hormetic effect. Do we dismiss epidemiologic evidence where radiation hormesis seems indicated simply because the protective mechanisms are not fully understood? This second objection also ignored considerable evidence already in the literature regarding hormesis,4,5 especially cell culture and animal studies, and the fact that hormesis is a widespread phenomenon in biology and pharmacology. In fact, the French report devotes considerable space to a discussion of potential mechanisms for the hormetic effect and its impact on low-dose risks.
The position taken in the joint report of the two French Academies is echoed in position statements from the Australasian Radiation Protection Society6 and by seven participants of the 15th Pacific Basin Nuclear Conference which was held in Sydney, Australia in October 2006.7
The BEIR VII report does not agree with the French view. Rather, the investigators believe that a comprehensive review of the available biological and biophysical data supports the LNT model as the best representation of risk at low doses although they also appear to agree that there are no data demonstrating a carcinogenic effect below 100 mSv.8 Both groups had access to the same literature and such disagreements generally arise from differences in interpretation of results, different views of statistical significance and data quality or simply differences in opinion. It is well beyond the scope of this review to compare in detail these two reports and their differences, but there are studies that relate to the issues involved that post-date these reports that are of interest, and in addition, there has been growing criticism of the LNT model. Also, some studies that predate the French Report seem sufficiently relevant to merit review.
Finally, in the summary of the French Report,1 they urge great caution when considering studies that use the LNT model for risk prediction and cite a paper in the Lancet9 as an example. This same cautionary advice would no doubt apply to a recent paper by Brenner and Hall in the New England Journal of Medicine, where it was argued on the basis of the LNT model and data not accepted as adequate by the French investigators that in a few decades about 1.5 to 2% of all cancers may be the result of CT scan usage.10 However, the French Report admits that a significant although small risk associated with low-dose radiation may exist for pregnant women, infants and young children, and that caution should influence practice for these groups.
Toxicology mainly uses two models for exposure or dose dependence. One has a threshold followed by an increase in risk or incidence (the threshold model) whereas the other has no threshold and views the risk or incidence as increasing from a zero dose level. When the increase is assumed linear with dose, one has the Linear-No-Threshold model (LNT). The risks associated with non-carcinogens are generally described by the threshold model whereas for carcinogens it is assumed that there is no threshold. Neither model includes the possibility of a U- or J-shaped curve where there is a dose range that confers benefit (the risk ratio falls below 1.00, the reference for no exposure) followed by adverse effects at higher doses. Calabrese and Baldwin in 2003 stated that in their view the toxicological community made an error of historic proportions in its formative years (the 1930-1940s) by buying into only these two models, which once accepted, became dogma and provided the basis for subsequent progress as well as confusion.11 This was in spite of individuals such as radiation biologists, pharmacologists and even some toxicologists who pointed out unmistakable exceptions.
The U- or J-shaped dose response curve represents hormesis or a hormetic effect. Hormesis appears to be very commonly encountered in pharmacology. Calabrese and Baldwin list more than 30 receptor systems that show hormesis.12 Closer to everyday life, readers of this Newsletter have repeatedly encountered the J-shaped dose–risk relationship when reading about studies of alcohol consumption and heart disease. Parsons13 as well as Tubina et al8 have in fact described the hormesis model as an evolutionary imperative reflecting adaptation to environmental levels of toxins and other hazards such as radiation. As Caroline Hadley puts it in a commentary in the journal European Molecular Biology Organization Reports, "what doesn't kill you makes you stronger."14
The principal issue with the LNT model is its use in estimating risk at low doses by assuming a linear dose relationship. This application of the LNT model continues to be widely employed in studies aimed at assessing cancer risks associated with medical diagnostic procedures. The literature used to justify this procedure predates the French Report and was presumably considered when the French investigators took the position that the evidence was inadequate. Many scientists and even economists regard extrapolations as fraught with danger because the procedure assumes a regularity such as linearity or exponentially which has not been observed and may not actually exist.
RADIATION AND BREAST CANCER RISKS
As the French Report indicates, studies that attempt to address the question of risk of radiation- induced cancer at low doses, taken as a whole, in their opinion either yields statistically insignificant indications or no evidence of risk at all. But some actually show evidence of hormesis and in some cases the odds ratios indicating a protective effect are statistically significant as judged by whether or not the 95% confidence interval includes the null result, i.e. 1.00. Studies that examine the association of breast cancer and radiation exposure are interesting in this context, especially since breast tissue is regarded by some as being particularly susceptible to this risk. There are also studies that predate 2005 where it is hard to ignore the fact that the results appear to falsify the LNT model.
The use of fluoroscopic x-ray monitoring during the treatment of tuberculosis was common between 1920 and 1960 and has provided some interesting data on cumulative exposures that are both above and below the 100 mSv "threshold of uncertainty." Most of the data are derived from patients treated during the second quarter of the 20th century. Follow-up studies for breast cancer incidence and mortality after treatment for tuberculosis are thus interesting for what they reveal at the low end of the exposure range and as well, relate to the question of hormesis. Typically, each dose was in the range of 10 to 100 mSv and exposure occurred as frequently as every 2-3 weeks for 3-5 years. Studies have looked at the relationship between cumulative dose and breast cancer. One study by Howe et al15 which examined the association between breast cancer mortality and radiation exposure is of particular interest because it provided good dose stratification and also compared the results with an atom bomb survivor cohort. When a comparison among tuberculosis patients was made between those with doses < 10 mSv and those with cumulative doses between 10 and 490 mSv, the latter had a relative risk of 1.09 which statistically was indistinguishable from 1.00, i.e. not statistically significant. For the atom bomb survivors, the same comparison yielded a relative risk of 1.05 which again was not statistically significant. This dose range starts below 100 mSv and then exceeds it by a factor of 5, although it is unlikely that many patients received doses as low as 10 or 20 mSv. Thus, if x-ray radiation is dangerous for diagnostic x-rays that are either well below the reference of < 10 mSv or even at most twice this figure, why were not these patients found to be at enhanced breast cancer mortality risk? Furthermore, for the tuberculosis patients, even the dose range of 500 mSv to 1 Sv still failed to reveal a statistically significant risk, although it did appear in the atom bomb survivor cohort.
The results reported by Howe et al are consistent with those found by Boice et al16 where, compared to untreated patients, those tuberculosis patients exposed to 10 to 1000 mSv had a relative risk of 1.2 which was found to be not statistically different from 1.00. In this study, the observed to expected ratio for breast cancer incidence ranged from 0.59 to 0.75 for doses in the range of 470 to 700 mSv for women of age > 25. This suggests a hormetic effect below 700 mSv. In both of these studies, significant risk appeared above 1000 mSv and appeared dose dependent which is consistent with high dose studies and not in dispute.
Today, multiple and frequent exposures to x-rays at single doses from 10 to 100 mSv for the purposes of imaging would be unusual, although occasional exposure during some interventions would fall in this range. But in the above studies even when the accumulated dose ranged from 500 to 1000 mSv, no evidence was found of enhanced breast cancer risk. It would seem that these results, which involved a large number of individuals and a follow-up of 30 or more years, would suggest at least a threshold model for breast cancer below which the risk becomes insignificant. This is contrary to the LNT model.
Since the French and BEIR VII reports there have been several studies reported concerning diagnostic x-rays and breast cancer. The most recent by Ma et al17 involved a case-control study of 1742 cases and 411 controls resident in Los Angeles County. They found an elevated risk for chest x-rays which showed a trend for frequencies of 1-2 to equal to greater than 9 exams as compared to those who never had such an x-ray. For mammography the association failed to achieve statistical significance even though the doses were higher. Also, most of the results for dental x-rays provided results that were not statistically significant. All risk factors were adjusted for confounding. Recall that these procedures employ doses below 1 mSv. The authors point out that the median age at first exposure was 15 years for chest x-rays and 35 years for mammography exposure and that early exposure to chest x-rays might enhance the effect of radiation in spite of the lower dose. The chest x-ray results of Ma et al are inconsistent with the results for exposure associated with tuberculosis treatment as discussed above, where the exposure was frequent and at much higher. The results of Ma et al are also inconsistent with a recent study by Lie et al who found no clear association between ionizing radiation and cancer of the breast for a large cohort of Norwegian nurses who experienced work-related exposure.18
In 2006 Redpath reviewed work in his laboratory at the University of California, Irvine (Department of Radiation Oncology, School of Medicine) which demonstrated J-shaped dose- response curves for radiation-induced neoplastic (potentially malignant cellular transformations) in vitro (cell culture studies) for a variety of radiations including those used in low energy imaging. He then made a comparison with the variation of the incidence of breast cancer for doses from about 20 to over 700 mSv based on calculations involving data of Preston et al.19 Below approximately 200 mSv, for Preston's epidemiologic data (human studies) the relative risk dropped below 1.00 and over the dose range of 20 to 140 the relative risk ranged from about 0.75 to 0.9, i.e. protective. This behaviour mirrored quite well the relative risk of neoplastic transformation observed in the in vitro studies which also showed a dose response curve with a minimum at around 20 mSv. But the point is that relative to patients exposed only to natural background radiation, the relative risk Redpath derived from the data of Preston et al suggested the presence of hormesis and yielded a J-shaped curve, actually a sort of "lazy J" in the parlance of ranchers and their cattle brands. Even if one ignores the hormetic effect, no excess incidence was found over the dose range below about 200 mSv.
DATA FROM WORK-RELATED RADIATION EXPOSURE
Here the literature is extensive and consistently exhibits protection from cancer associated with low-dose radiation exposure. An excellent review is to be found in a very recent paper in the journal Dose-Response (full-text articles are free) by Sanders and Scott.20 They examine a large number of studies which have addressed this issue, which have found either no effect or a protective effect from radiation exposure. Nineteen studies are listed showing radiation hormesis where the percentage of lung cancer cases or mortality avoided ranges from 7% to 100% with typical results in the range of 25-50%. For all cancers, 13 studies showed evidence of avoided cases or mortality ranging from 7% to 49%.20 Exposures involved both ionizing radiation and alpha particle radiation from the decay products of radon.
Work-related epidemiology is confounded by what is called The Healthy Worker Effect (HWE). When examining the question of expected cases in unexposed individuals, it is common to used large population databases. The HWE arises when workers are potentially healthier or less susceptible due to selection at employment, company health benefits, early detection by company medical personnel, etc. Some of the studies discussed by Sanders and Scott avoided this source of confounding by using unexposed but otherwise more or less identical company employees as controls. For example, in a study of workers in the nuclear industry, the protocol compared workers required to wear film badges (radiation monitors) vs. those who were not considered at risk (unbadged workers). It was found for women working in 12 U.S. nuclear weapons facilities there were 25% fewer deaths from all causes and 17% fewer deaths from cancer among the badged workers compared to the unbadged workers. The relative risk for lung cancer mortality in badged workers was 49% lower than unbadged workers.21
Another study of interest in this context involved examining 100 years of data from British radiologists. It was found that for those professionally registered after 1920, there was no indication of enhanced risk. Expected mortality rates were calculated based on three different data sets: all men in England and Wales, all males of a social class equal to physicians, and all male practitioners. The results for the relative risk of cancer mortality were 0.63, 0.82 and 1.04 for comparisons with these three control groups, respectively. The first two numbers were highly statistically significant and suggest hormesis, whereas the comparison with male practitioners (RR = 1.04) was not statistically significant.22 Also there was no statistical evidence of decreased longevity compared to the control groups. Cameron has suggested that radiation provided protection by stimulated their immune systems.23
Thus some studies of work-related exposure may be confounded by the HWE but others are not. The majority of these studies consistently exhibit no enhanced mortality or enhanced incidence of either specific cancers or cancer in general, and the levels of exposure are high compared to modern diagnostic radiation. Indeed, hormesis appears to be the rule rather than the exception. This is exactly the opposite of what we have all been led to believe. As long as the dose is not too high, workplace radiation exposure appears to present no enhanced risk and may even be protective in terms of cancer incidence and mortality. These results appear to falsify the LNT model, and in addition the doses where no enhanced risk was found were well above those normally encountered in diagnostic procedures.
RADON AND LUNG CANCER
The notion that exposure to the radioactive emissions associated with radon and its decay products increase the risk of lung cancer is part of the conventional wisdom. As mentioned above, radon is a major source of natural background radiation. In 1995 Cohen24 found a highly significant negative correlation between radon exposure and lung cancer mortality in the U.S., even after adjusting for smoking and other socioeconomic factors. The study involved about 300,000 radon measurements in over 1600 counties in the U.S. This result held up under extensive adjustment (over 500 factors) for confounding. The author concluded that there was no evidence from this study that low-level radiation causes lung cancer, that there is at least some evidence that it may be protective, and the LNT model failed completely. In addition, Colorado which has the highest residential radon concentrations in the U.S., the average rate of lung cancer is well below that for the U.S. overall. Also, there are villages in Japan that have radon levels 3 times the national average, but have half the cancer death rates compared to the country as a whole. Radon is involved in therapy in Europe at a number of spas and other locations, has a long history, and is being actively studied as a therapeutic agent.25
However, the Biological Effects of Ionizing Radiation (BEIR) committee of the U.S. National Research Council took a totally different view. Luckey and Lawrence26 point out that this position is actually contradicted by the data in the BEIR reports which showed conclusively that for 68,000 miners and 2,700 cases of lung cancer, there was no statistical significant association with the extent of exposure to radon. Luckey and Lawrence conclude that carcinogenic particulates and/or noxious gases, not radon and its decay products, must be the cause of lung cancer in miners. In addition, the data presented by the BEIR VII committee on lung cancer from radon in homes showed that the relative risk did not rise significantly with increased radon exposure.
OTHER EVIDENCE OF HORMESIS OR PROBLEMS WITH THE LNT MODEL
There are areas around the world where the background radiation is very high compared to the normal levels in the US.27 Extreme levels of background radiation are found for example in Guarapari (Brazil), southwest France, Ramsar (Iran), parts of China, and the Kerala coast (India). Black sand beaches in Brazil have radiation levels 400 times normal. In India, the 570-km coastline of Kerala has very high natural radiation similar to the beaches in Brazil. In some locations in the city of Ramsar in Iran, the radiation levels are 55 to 200 times higher than normal background levels. What is interesting about these areas is that the individuals continually exposed to high background radiation do not appear to suffer any adverse health effects and in fact, in some cases, they appear healthier and live longer than those living in areas used for control.28 But these are anecdotal observations and must be treated with caution.
There is also the interesting case of the radioactive apartment houses in Taiwan.29 Accidental exposure allows a unique opportunity to examine the validity of the LNT model since obviously, given the views regarding the danger of any radiation, no one is going to volunteer to participate in meaningful studies. Such an exposure occurred in Taiwan where about 20 years ago, steel contaminated with cobalt-60, a radioactive isotope emitting penetrating gamma rays, was used in the construction of a number of apartment houses which were then occupied by about 10,000 persons for from 9 to 20 years. These residents unknowingly received doses that ranged from a high of over 500 mSv to a low of 18 mSv with an average of around 50 mSv. The half-life of cobalt-60 is 5.3 years. Subsequent to the discovery of this situation, the occupants were followed- up to ascertain the incidence of cancer. The predicted cancer deaths were based on two models, the natural rate expected in this population and the predicted rate from the LNT model, and were 238 and 302 respectively. The observed number of cancer deaths was 7. When residents younger than 20 were excluded, the predicted cancers were 186 and 242 according to the two models. Only 5 cancers cases were actually found. It was estimated that when the buildings were completed, the maximum annual dose rate was as high as 500 mSv in some apartments.
A second study regarding the radioactive apartment houses in Taiwan was published in 2006.30 This study criticized the earlier study for underestimating the cancer mortality rate. Also, they looked at the incidence of cancer rather than mortality. For all cancers except leukemia and for solid cancers, the standardized incidence ratios were 0.8 and 0.7, respectively with both results statistically significant. This lower rate of observed vs. expected cases again suggests a protective effect.
While the Taiwan apartment houses offered a unique opportunity to examine the carcinogenic effects of low-dose radiation, data analysis was complicated by a very wide range of exposures, both with regard to level and duration, given variable residency times, the differences in levels in various apartments and rooms, and the natural decay of intensity due to the 5.3 year half-life of colbalt-60. However, independent of the possibility of a protective effect, it seems hard to argue against the point that no statistically significant increase in mortality or incidence over that expected from unexposed individuals was found, even though the exposure was relatively high.
HYPOTHESES CONCERNING THE PROTECTIVE MECHANISM
Mechanistic investigations of the protective effect of low-dose radiation have involved both animal and cell-culture studies. This subject has been discussed by Scott4 and by Scott et al.5 Low doses and low dose-rates of gamma rays and x-rays appear to stimulate the body's natural defenses, an effect that has been called radiation activated natural protection (ANP). This protective mechanism involves selective removal of aberrant cells such as those that are precancerous via apoptosis (cell death). The selective removal of precancerous cells via apoptosis it thought to involve intercellular signaling involving reactive oxygen and nitrogen species and certain cytokines. In addition, there is considerable evidence supporting the action of low-dose radiation in connection with stimulating immunity against cancer cells. These protective effects would operate for both sporadic and hereditary cancers. As mentioned above, there are evolutionary arguments suggesting that such protective mechanisms may have been essential to survival.
The thesis developed in this mini-review is summarized by the warning: Beware of the conventional wisdom. A more comprehensive review would have provided more extensive support for the lack of carcinogenic effects and the possible protective effects of low-dose radiation. Some of the papers cited provide this additional information along with documentation. The question of the validity of the LNT hypothesis, especially in connection with radiation exposure, is very important since the LNT model has been the basis of environmental and public health policy for several decades. It is responsible for the fear of any radiation common among the general public, the reluctance seen among some individuals regarding diagnostic or screening procedures involving exposure to radiation, the fear of contamination from nuclear plant accidents or negligence and concerns about dirty bombs employed for terrorism. While there is no doubt regarding the risks of exposure starting somewhere between 100 and 1000 mSV, the result discussed above suggest at the very least the existence of a threshold and probably beneficial effects below this threshold such that many of the concerns enumerated are unjustified. Those who believe in the LNT model regard such statements as heresy, reckless and dangerous. But it appears that scientific research is chipping away at the foundations of the LNT theory and its use in extrapolating to low doses. The growing evidence of biologically plausible mechanisms for hormesis, its widespread presence in many biological systems, animal models and response to pharmacological interventions suggests that in the context of the interaction of radiation with human cells, it must now be taken seriously.
The two alternatives to the LNT model are a threshold model and hormesis. It would be surprising if the latter two models were not common since in order to survive organisms had to evolve defence mechanisms, but these mechanisms presumably could always be overwhelmed, thus providing a threshold for morbidity and mortality. Likewise, defence mechanisms, once activated, could provide a mechanism for a hormetic effect.
Articles such as the one mentioned above by Brenner and Hall in the New England Journal of Medicine,10 which predicted that in a few decades up to 2% of all cancers may be attributable to CT scans, have a profound impact on opinion and medical practice and yet the study in question was based on a model which more and more is being questioned and considered by some experts as invalid. But it is highly unlikely that readers of that paper will take the time to inform themselves regarding the pros and cons of the model even though the whole thesis of the paper critically hinges on its validity. In addition, the journal in question has a high profile and is highly respected world-wide. While the study no doubt correctly applied the model and within that framework obtained a significant result, if the model is wrong, then so is the result, and the readers and the media that track papers in these journals and duly report the results are misled into believing in risks that may not exist.
The issues raised in this review will probably not be resolved any time soon. The conventional wisdom generally blocks alternative views, discourages research and limits research funding, and makes publication of contrary results in major journals difficult. A substantial amount of the recent literature in this field is reported in journals that are not even covered by Medline (PubMed), the National Library of Medicine search engine. The purpose of this review has been to acquaint the reader with an alternative view which does not appear to be well known, not only among the general public but also among many medical professionals. Readers must recognize the uncertainties associated with this subject. However, when two distinguished Academies in France, one concerned with medicine and one with basic science, jointly tell us that there is no evidence of risk and possible indications of benefit for exposures to ionizing radiation below 100 mSv, the risk vs. benefit for medical diagnostic procedures appears to be shifting strongly in the direction of benefit. But their report is a monograph in French, and the summary, while in English, is in a specialized journal with a limited audience. Also, the absence of evidence of risk does not prove it is not there, but given the amount of research that has already been done and what it suggests, it is not unreasonable to have justified concerns regarding the conventional wisdom.