Part 1. The NRPB's Effort to Protect Women from This Book
How It All Began
In mid-April 1995, Mark Lewis and Julian Bellamy of 20/20 Television in Britain began to investigate the thesis of this book. They had received a copy carried from San Francisco to London by one of their colleagues. Mark (producer) and Julian (correspondent) prepare 28-minute investigative reports for the well-watched British television program called "The Big Story."
Mark and Julian accepted none of the book "on faith," for they are hard-nosed skeptics whose first rule is never to be manipulated or fooled by anyone who might appear in one of their reports. I began getting questions from them in late April, after experts at the NRPB told them that I was "wrong by 100-fold." My impression, by telephone, was that Mark and Julian were highly suspicious of both me and NRPB.
The decision actually to build a report around the book was not made until later. During the intervening weeks, I was providing Mark and Julian with scientific papers and answers to every objection they heard on three continents. They were transmitting to me many scientific assertions which the makers declined to put into writing. They seemed exceedingly suspicious of experts who would not stand behind their claims.
Meanwhile, we were shipping books for peer-review. NRPB received one from us by Express Mail on May 25th.
1a. Should This Program Be Banned?
After a tentative decision to do a program, Mark and Julian asked the NRPB to participate and to present its own position. The NRPB declined in writing, with the following statement which was included prominently in the final program: "We feel Dr. Gofman's work is so poor that we do not want to be associated with it by appearing on the program. His track record has no impact on other professionals."
But Mark and Julian had independently checked on my "impact" (for example, see Chapter 43, Part 4). They were not deterred. They just continued to read the scientific literature, to interview experts everywhere, and to bombard me with scientific questions. After a while, they displayed an amazing grasp of the topic --- far, far better than most radiologists and greater even than many epidemiologists. The broadcast was planned for late July or early August. Mark and Julian continued to urge the NRPB to participate in the program.
But the NRPB had other plans. It tried to stop the broadcast, according to a news story in the London Express by Paul Crosbie, on August 3. The program was scheduled for broadcast at 7:30 pm on August 3. The huge headline on Crosbie's report is, "Doctors' outcry over X-ray link to breast cancer; Call for ban on TV show." According to Crosbie:
"The decision to show the programme has been condemned by scientists and doctors in Britain." And:
"The national radiation watchdog, the National Radiological Protection Board, urged ITV chiefs to halt the broadcast. It called the professor's claims `alarming, unsound, and misleading.' The NRPB said showing the programme could cause women unnecessary anxiety and disrupt screening for the early detection of breast cancer."
1b. An "Unprecedented Step" by the NRPB
All the "outcry" was set in motion by the Head of NRPB's Medical Department, Dr. Chris Sharp. On July 11, 1995, Dr. Sharp signed a "Dear Colleague" memo on NRPB letterhead. It begins:
PREVENTING BREAST CANCER - JOHN W. GOFMAN
"The Royal College of Radiologists has agreed, in principle, to join with the NRPB in briefing media journalists prior to transmission of the 20/20 TV programme on the risks of medical radiology relating to breast cancer and, if necessary, to arrange a Press Conference on the day after transmission - in which the Department of Health would participate." And:
"I attach the draft material on which NRPB would wish to base oral briefings to individual newspaper journalists. The material is in 2 parts: the first two pages are a fairly simple statement of the main points with the important sections bolded for the journalists to use as quotes. The remaining three pages are a more detailed brief to give background should they want it." The five pages for the press are in single-spaced rows. NRPB's draft press handout winds up as follows:
"The Board wishes to reassure women, in particular, that Professor Gofman's claims are unsound, inaccurate, misleading and unnecessarily alarmist and should not dissuade them from accepting X-ray procedures to detect or assist in the treatment of serious or significant disease. However, NRPB continues to encourage all medical practitioners to only use diagnostic X-ray procedures when justified and to keep exposures as low as reasonably practicable, even though the risks are small."
The NRPB's pre-broadcast campaign was acknowledged by Dr. Sharp and his assistant, Dr. John Harrison, in a joint letter published by the British Medical Journal, November 11, 1995. Defending NRPB's failure to participate in the 20/20 program, they wrote: "The board [NRPB] distributed a detailed scientific critique of Gofman's work to the media." (See Part 1c, below.)
Stories began appearing in the newspapers on July 27: "Cancer link with X-rays `alarmist'" by the science editor of the Daily Telegraph. "X-ray link to breast cancer `misleading'" by Liz Hunt, the medical correspondent of the Independent. Ms. Hunt wrote:
"In an unprecedented step, the National Radiological Protection Board took action to discredit the claims by Professor John Gofman of the University of California, Berkeley, who features in a television programme to be broadcast next week." On August 3, there were many additional stories, including "Doctors' outcry" about the attempted ban.
In the end, the NRPB's "war" to protect women from our work had one very educational effect: Often for the first time, many science and medical reporters, as well as the women who do not watch "The Big Story," heard a well-established fact --- medical x-rays are a proven cause of breast cancer. In addition, millions watched the program. (Its title: The X-Ray Effect).
1c. The Official NRPB Critique, and Our Request to Dr. Sharp
The NRPB publishes the Radiological Protection Bulletin. Its August 1995 issue devotes 2.5 pages to a critique of this book. The Bulletin critique is almost identical to the critique prepared for the press. The Bulletin critique is authored by Dr. Colin Muirhead, head of epidemiology at NRPB, and Dr. Chris Sharp, head of the medical department at NRPB, so their critique is Muirhead 1995 in our Reference list.
On August 23, 1995, I faxed Dr. Sharp to make arrangements for my response in the Bulletin. I reminded him that "It is customary for any Journal to provide a reasonable space for response to an article such as yours, which is an attack on many points in my book."
On August 23, 1995, Dr. Sharp faxed his reply: "The Bulletin is not a peer-reviewed publication, but primarily the means to express to a wider audience NRPB's interpretation of developments in the science of radiological protection. It is therefore not our normal practice to publish letters in discussion of our articles."
Our attitude is this: Women and their families deserve experts who care enough to push toward a resolution of this issue. When the issue is the main cause of breast cancer and its consequent prevention, attempts to suppress exchange among peers are especially appalling.
Some Very Consequential Errors by Muirhead and Sharp
Muirhead and Sharp offer one argument in their critique which confirms our estimate (see Part 8b, below). When they try to use the argument against our estimate, it's a surprise. And we think they make additional errors, three of which will be easier for readers to comprehend than the others (Parts 5b, 5c, and 7). Muirhead and Sharp present no information which was not already considered by us when we did the First Edition. We explained, especially in Chapter 40, our considerations and judgments on the points which are unknowable by anyone.
We take up the NRPB's points in their printed order, for the remainder of this chapter. Page-references to our First Edition are the same in the Second Edition.
Part 2. NRPB's Extensive Consultation for Its Critique
The introductory paragraph of the NRPB critique asserts that "Gofman's claims could cause widespread anxiety among women if publicised widely in the media and could dissuade them from accepting procedures beneficial to their continuing good health. The potential effects of the book are significant and it is felt that an article rather than the normal book review is appropriate." And:
"The purpose of this article is therefore to discuss the basis of Gofman's claims and highlight where his calculations of risk differ from those published by NRPB and others for female populations exposed to medical x-rays. In writing this article, we have conferred with colleagues in the Radiation Effects Research Foundation and the US National Cancer Institute and with members of the United Nations Scientific Committee on the Effects of Atomic Radiation (UNSCEAR)."
Comment on Female Anxiety
We too, like the NRPB, are eager to make it clear to women that our findings do not mean that they should reject all uses of medical x-rays. So, we made the point clear even on the rear cover of the First Edition! We quote from the rear cover:
We have often been asked the question: "Have Medical X-Rays Done More Harm Than Good?" That is the wrong question. The appropriate question is, "Can we have all the good things in medicine which x-rays can do with vastly less harm?" The answer is unequivocally, "Yes" ... We hope that all physicians will join with health scientists in a determination to reduce unnecessary x-radiation. As discussed in this book, this effort can succeed without ever interfering with a single essential x-ray examination. (End of quotation from the rear cover.) Also, we made the same points in Chapter 1 (on its first page) and in Chapter 42 (on its first page, followed by details).
Was it accurate for the NRPB to tell medical and science reporters (and thus the public) that our work was "unnecessarily alarmist?" Obviously not. We note that this phrase (see Part 1b, above) was edited out of the Bulletin version of NRPB's critique.
Comment about Worldwide Consultation
Muirhead and Sharp state that they conferred with various peers at RERF, NCI, and UNSCEAR in preparing their critique. In other words, they confirm our statement in Chapter 43 that this work is in the process of very wide peer-review indeed --- just as we intended.
Part 3. NRPB on Known Breast-Cancer Causes: What's on the List?
NRPB's introduction is followed by a section which names known causes of breast cancer. "Breast cancers occur more frequently with increasing age, being relatively rare below the age of 35 years and increasingly common over 50 years. A range of epidemiologic studies has positively established that the occurrence of breast cancer is related to [and here, for clarity, we will subdivide NRPB's single paragraph]
--- reproductive factors: the age of the woman at the time of birth of her first child and the age at which her periods start (menarche) and her age at menopause;
--- a family history of breast cancer;
--- the occurrence of previous benign breast disease;
--- exposure to ionising radiations (x-rays and gamma rays)."Comment about Non-Radiation Causes of Breast Cancer
We have no disagreement with NRPB about this list. It may be too short, however.
Many people who only hear of our 75-percent estimate for radiation (but who have not seen the book) assume that we have ruled out important roles for any other cause. Not so. Our finding is fully compatible with important roles for other factors --- as we emphasized in Chapter One (Part 5), Chapter 40 (Part 3), and Chapter 41 (Part 2) of the First Edition. Also, now, in Chapter 47 (Part 7).
Comment on Inclusion of Radiation
NRPB's inclusion of ionizing radiation, in a very short list of known causes of breast cancer, affirms our statements that radiation-causation is an undisputed fact among experts in the field.
Another 1995 affirmation, more readily available to readers in the USA, is the giant medical text entitled "Cancer of the Breast: 4th Edition," edited by William L. Donegan, M.D., and John S. Spratt, M.D. (Donegan 1995). In Chapter 8, which they wrote themselves, Spratt and Donegan state:
"The capacity of ionizing radiation to produce breast cancer has been repeatedly confirmed" (p.131). After this statement, they cite much of the same evidence flagged in our Reference list (with the symbol "#").
Part 4. NRPB Claim: Gofman Did Not Use the Latest Data
Now we arrive at NRPB's first science-based complaint about the First Edition of this book. The complaint comes with three parts, which we will label "a, b, c" (not so labeled in the original). Muirhead and Sharp say (p.12):
"Significantly, he [Gofman] (a) has not used the latest breast cancer incidence data from the A-bomb survivors. Instead he uses older mortality data for all solid cancers and (b) [he uses] the T65D radiation dose estimates, which have now been superseded. In transferring risks from the Japanese to the American population, (c) Gofman has also apparently not relied upon recent data on baseline breast cancer rates in the two countries, but upon data from some 30 years ago. This is particularly inappropriate as many of the breast cancers in the A-bomb study have occurred only in recent years."
Response to (a): Old data not specific to breast cancer
This claim would sound very different if Muirhead and Sharp had told the correct story. In addition to using mortality data for all cancers combined, we did employ the newest breast-cancer data from the A-Bomb Study in arriving at our risk-estimates. In fact, we used two separate reports. In Chapter 3, p.14, we cited both Thompson et al 1994, and Tokunaga and Land 1994, and we used information from both studies in our Chapter 40 (pp.278-279, and 282-283). Our decision to start our analysis with the all-cancer mortality data is related to part (b) of NRPB's complaint.
Response to (b): "Superseded" T65D Dose-Estimates
The comment by Muirhead and Sharp, about the "superseded T65D radiation dose estimates," identifies a fundamental difference between myself and NRPB about how properly to handle proposed revisions of dosimetric estimates. A large part of my 1990 book showed why the T65D dose-estimates for the A-Bomb Study can not just be jettisoned and superseded by an illegitimate, shuffled set of data.
With perpetual reshuffling of the A-bomb database, after results are in, "researchers" could find just about any answers which would please them or their sponsors --- which are the Japanese and United States governments. That is why a fundamental anti-bias rule in prospective biomedical research is that you can supplement your original study as it progresses, but you must also reveal what the follow-up shows in the unaltered, legitimate database. If researchers could conceal unwelcome results just by labeling them "superseded," we would really have reached Orwellian, non-credible, worthless science.
No one (myself included) wants to "ban" corrections and improvements in any study --- indeed, we warned about dosimetry errors in the A-Bomb Study (Gofman 1981). New dose-estimates were released by Japan and the United States in 1986 (so they are called "DS86" estimates). And in 1990, our book demonstrated the scientifically proper way to use them, in parallel analysis with the T65D estimates. Our method, called "constant-cohort, dual-dosimetry," rules out bias but includes the revised dose-estimates.
We demonstrated in Gofman 1990 that proper use of the new DS86 dose-estimates, without retroactive shuffling of the cases, gives virtually the same answer as the original T65D database. It is the unnecessary and improper case-shuffling, done by the radiation establishment, which has the effect of reducing risk-estimates from the database (and helping the sponsors).
The A-Bomb Study has been a unique biomedical treasure for all humanity, but the Orwellian handling of its "superseded" past is threatening to ruin its credibility. It is high time for analysts at the NRPB and other bodies to "stick their necks out" just a little to assure that the A-Bomb Study retains a permanent anchor in the T65D cohorts, with parallel analyses using both the T65D and the contemporaneous dose-estimates for direct comparison.
Response to (c): Baseline Data from 30 Years Ago
At issue is the difference in observed vulnerability of American women and Japanese women to breast cancer (see p.280). The gap has been decreasing over time (see p.281). By vulnerability, we mean the integrated tendency to develop breast cancer from any and all causes, as reflected in breast-cancer rates.
It is an amazing phenomenon for Muirhead and Sharp to discuss what we "apparently" did about the date of our baseline rates for breast-cancer in the USA vs. Japan. On page 280, we explicitly provided the date (1964-1965), the sources, and the reason: 1964-1965 is approximately the midpoint of the follow-up period of the A-Bomb Study.
Our choice of 1964-1965 ("30 years ago") is called "particularly inappropriate" by Muirhead and Sharp, and they hint that breast-cancer rates from "recent years" would be the appropriate choice. Did they back up either idea with any references to evidence? No.
Here is everyone's problem. The A-bomb survivors received their bomb-exposure in 1945. That is the year in which their radiation-induced cases of breast cancer were "put on the shelf for delivery later" (see Chapter 4). For women of equal age in 1945, is the delivery-response per rad of exposure likely to depend more heavily on the existing vulnerability to breast cancer at the time of their irradiation, or more heavily on their vulnerability at the time of diagnosis, or most of all on their vulnerability during their various latency periods? We can not answer the question. No one can.
We chose to apply the "midpoint" 1964-1965 ratio, as an approximation of the relative vulnerability to breast-cancer in the USA versus Japan, and we make the assumption that this relative vulnerability affects the response to radiation received in the 1920-1960 period of our analysis.
Muirhead and Sharp claim that they know better, when they denounce our choice of the 1964-1965 ratio as "particularly inappropriate." They hint that use of a "recent" ratio of USA/Japan breast-cancer death-rates would be appropriate. Later, they use the 1990 ratio. The Muirhead-Sharp assumption gives a lower risk-estimate than our assumption (see Part 6b).
In contrast to their inappropriate certainty about our assumption, we feel some humility about biomedical mysteries which remain to be solved, so we wrote in the First Edition, at page 281):
"At this time, we think the 5.67 ratio of background breast-cancer mortality-rates (USA / Japan, in the mid-1960s) is the most reasonable approximation to use in our conversion-factors. As additional evidence develops, it may point to other ways to handle country-to-country `transport' of observations. Meanwhile, we keep an open mind toward a variety of old and new hypotheses about the actual process of radiation carcinogenesis, including interaction between radiation and other factors."
Part 5. Claim: Gofman Used Assumptions "Not Accepted" by Most
"A number of assumptions are made in the book to calculate risk which are not accepted by most of the scientific community" (Muirhead 1995, p.12). This claim also comes in three parts. No part is based on assumption. We go with the existing evidence, whereas the radiation establishment makes some non-protective assumptions. A non-protective assumption is one which can operate to increase cancer in the population.
5a. The Shape of the Dose-Response
According to Muirhead and Sharp (p.12), our first "assumption" is that "low doses are twice as effective at inducing cancer when compared with high doses --- this is not consistent with analyses of the A-bomb data which show a linear trend in the risk of solid cancers with dose ..."
Response:
The meaning of a linear dose-response is that one rad has a fixed, unchanging carcinogenicity regardless of dose-level. By contrast, supra-linearity means that each incremental rad is less carcinogenic as dose rises; low-dose rads are more potent than high-dose rads.
The evidence from the A-Bomb Study for all cancer-sites combined, both in the legitimate T65D dosimetry and in the supplemental DS86 dosimetry, fits a supra-linear dose-response provably better than it fits a linear dose-response --- if the DS86 dose-estimates are properly used (without retroactive shuffling of the cancer cases). See Part 4b. Our claim, above, is demonstrated step-by-step in Gofman 1990, from raw-data to conclusion, with no mystery-gaps.
How do the NRPB and other government committees manage to miss the supra-linearity? They embrace the retroactive shuffling of the cancer mortality cases.
There is something additional for NRPB to explain, when it attacks supra-linearity as an "assumption" at variance with the evidence. In the A-Bomb Study, the latest incidence data specifically on breast cancer are depicting supra-linearity at page S26 of the 1994 Thompson study --- as we noted on page 277 of the First Edition. Muirhead and Sharp are silent on this.
The North American Studies
Muirhead and Sharp assert (p.12) that "women irradiated for medical reasons in North America" show a linear (not a supra-linear) dose-response. Fortunately, it is impossible to obscure the important contribution of those North American studies in helping to settle the threshold issue, as shown in Chapter 45, Part 2d. Beyond that function, however, the utility of those studies is very limited, for the reasons given in Part 9 of this chapter.
5b. The Issue of Low Dose-Rate
According to Muirhead and Sharp (p.12), our second "assumption" is shown when we use "no reduction in stochastic [cancer] risk for low dose rates - both NRPB and ICRP calculate the risks at both low doses and low dose rates to be half of that at high doses and high dose rates." Muirhead and Sharp want us to assume that each rad received from medical x-rays is only half as carcinogenic as each rad received by the A-bomb survivors, because of a low dose rate.
This is a really bizarre comment, since the medical patients in our study and the A-bomb survivors both received their rads at acute rates, not low rates. Why on earth would a reduction in risk "for low dose rates" even occur to Muirhead and Sharp when medical radiation overwhelmingly dominates our Master Table? We wondered if this allusion to rate was just an editorial error at NRPB. Apparently not. We find the same substance in the NRPB pre-broadcast memo where NRPB says Gofman makes the "assumption" that "radiation given at low dose rates is equally capable of inducing cancer as those [sic] given at high dose rates."
A Non-Protective Assumption by NRPB and Other Protective Bodies
We need to add a comment about "assumption" here. There is no human evidence to support any reduction in risk-values for low doses, delivered either fast or slow in the circumstances at issue here. Moreover, there is no need to extrapolate from high doses to low doses, when the A-Bomb Study provides statistically significant evidence of excess cancer all the way down to about 11-15 rads of exposure. All this is extensively shown, including the raw data themselves, in Gofman 1990.
NRPB claims repeatedly that the human evidence fits a linear dose-response (in NRPB 1993, NRPB 1995, Muirhead 1995).
By definition, linearity means that the risk per rad is identical at low doses and at high doses. Yet Muirhead and Sharp admit (p.12) that NRPB calculates risk "at both low doses and low dose rates to be half of that at high doses and high dose rates." Then overriding the NRPB's claim that the human evidence shows linearity, the NRPB and ICRP substitute an assumption --- an assumption which produces lower risk-estimates and thus poorer protection of the public, by a factor of two here. The BEIR-5 Committee does it too.
5c. The Potency of X-Rays vs. Gamma Rays
According to Muirhead and Sharp (p.12), our second "assumption" is that "x-rays are twice as effective as gamma radiation at inducing cancer - this is not supported by comparisons between North American studies (x-ray exposures) and the A-bomb study (mainly gamma)."
Response:
Muirhead and Sharp do not reveal that the North American studies differ among themselves too much to elucidate the issue at all. For example, the Nova Scotia study shows a risk per rad six times what the other North American studies show. Against such variation, no one can rationally try to settle the gamma / x-ray difference.
However, we did not just "assume" the two-fold difference. We used the existing evidence from very mainstream sources, listed below in chronological order. (The usual details are in our main list of References.)
o - 1978, Bond et al, in Health Physics.
o - 1985, Sinclair in Radiation Protection and Dosimetry.
o - 1986, ICRU (International Commission on Radiation Units and Measurements), in its Report 40.
o - 1988, Kerr in Health Physics.
o - 1990, BEIR-5 Report from the National Academy of Sciences, USA.And new, since the First Edition of this book went to press:
o - 1995, Lucas et al, in Health Physics (see Figure 3).
Surely, the explicit, easily readable statements in BEIR-5 (which NRPB quotes often in its own reports) cannot have escaped Muirhead and Sharp. We quote the statements on this issue from BEIR-5, p.218:
"The general applicability of the experience of the Japanese A-bomb survivors is uncertain on additional grounds. Most human exposures to low-LET ionizing radiation are to x rays, while the A-bomb survivors received low-LET radiation in the form of high energy gamma rays. These are reported to be only about half as effective as ortho-voltage x rays (ICRU 1986). While that is not a conclusion of this Committee, which did not consider the question in detail, it could be argued that since the risk estimates that are presented in this report are derived chiefly (or exclusively) from the Japanese experience, they should be doubled as they may be applied to medical, industrial, or other x ray exposures."
Another Non-Protective Assumption by the NRPB
The apparent scandal is that the NRPB makes no use of the listed evidence. The NRPB appears to assume that all the evidence cited above is inferior to the non-appropriate evidence on which it relies. By invoking this assumption, the NRPB arrives at lower risk-estimates and thus poorer protection of the public, by another factor of two here. While the NRPB proclaims its desire to prevent public anxiety, we are seeing in the same article how it actually arrives at soothing reassurances. The public might be fascinated by this amazing new facet of protection from radiation.
Part 6. Claim: Gofman's Risk-Factors Too High by at Least 10-Fold
"Overall, Gofman's risk estimates are at least 10 times higher than those calculated by NRPB for a UK population ..." (Muirhead and Sharp, p.12). We agree that our estimates per rad are indeed at least ten times higher than those of the NRPB and similar radiation bodies. They use retroactively shuffled data, they invoke non-protective assumptions (see also Part 9), and they indulge in scientifically questionable practices which are fully detailed in Gofman 1990.
Muirhead and Sharp try to quantify how our per-rad breast-cancer risk becomes higher than theirs: "The reasons, and scale of attributable difference, appear to be [and now we subdivide their paragraph, and we add letters to facilitate discussion]
(a) - older Japanese data, not specific for breast cancer --- risk increased 1.35 times;
(b) - old baseline rates for breast cancer in the USA and Japan --- risk increased by 1.6 times;
(c) - higher values for the effects of x-rays compared with gamma radiation --- risk increased 2 times;
(d) - higher values for the effects of low doses, low dose rates and fractionation - risk increased about 4 times." End of quotation.
Response to (a): Older data
Our reduction of the mortality data for all solid cancers, into the component for breast-cancer, was perfectly appropriate. Although the data are a few years older, they are valid forever, because dead people do not come back to life. Our decision to use the slightly older data allowed us to use the legitimate, unshuffled cohorts --- not the illegitimate, retroactively shuffled data used by NRPB and the other radiation committees. Properly handled, the old and new dose-estimates yield almost identical risk-estimates (see Part 4b). So NRPB's risk-estimates should rise by this factor of 1.35.
Response to (b): Transport Factor from Japan to USA
Where does this factor of 1.6 come from? Muirhead and Sharp derived it from our Chapter 40, as follows.
o - On page 280, readers will see that in 1964-1965, the breast-cancer death-rate per 100,000 females was 21.55 in the USA versus 3.80 in Japan. That ratio is (21.55 / 3.80), or 5.67.
o - On page 281, in the box, readers will see that in 1988-1991, those rates had changed to 22.4 in the USA and 6.3 in Japan. That ratio is (22.4 / 6.3), or 3.55.
o - Muirhead and Sharp claim that I should have used the smaller ratio (see Part 4c). Our use of the higher ratio yields a risk-value which is (5.67 / 3.55), or 1.6 times higher than use of the lower ratio. That is where the factor of 1.6 originates for Muirhead and Sharp.
Unlike Clark Heath (Chapter 44, "Part 3), Muirhead and Sharp do not challenge the use of some ratio, for transport of results between these two countries. Health claims that our factor of 5.67 introduces a 6-fold error, and that the right factor would be 1. Muirhead and Sharp clearly indicate, with their 1.6 factor, that they would have used 3.55. We used 5.67. We discussed the basis for our choice in Part 4c above. We would make the same choice again if we were starting today. But this is a point on which no one can be certain. Therefore, we would not claim that NRPB's preference for 3.55 is wrong. Conversely, NRPB would not be entitled to suggest that our choice of 5.67 is wrong.
Response to (c): X-Rays vs. Gamma Rays
NRPB should be embarrassed to have raised this issue, as explained in Part 5c. NRPB's risk-estimates for medical x-rays should rise by this factor of 2. Clark Heath (Heath 1995) did not make this error.
Response to (d): Low Doses and Low Dose-Rates
The factor of 4 consists of two factors of 2 each.
The first factor of 2 for low-doses represents (correctly) the difference in per-rad risk at low doses between the supra-linear dose-response demonstrated from the legitimate, unshuffled cohorts of the A-bomb survivors, versus the linear dose-response derived from the illegitimate, retroactively shuffled database of the A-bomb survivors. So NRPB's risk-estimates should rise by this factor of two.
What about the additional factor of 2 for low-dose rates? Again, the NRPB should be embarrassed to have raised this issue, as explained in Part 5b. Clark Heath (Heath 1995) did not make this error. The NRPB's own risk-estimates should be raised by this factor of 2.
Part 7. Claim: Gofman's Dose-Estimates for 1920-1960 Are Unreliable
Muirhead and Sharp say (p.12): "Gofman also makes his own estimates of medical doses to the female population of the USA in the period 1920 to 1960. These doses are based on extrapolations from local surveys in the USA to medical practices throughout the country. Such calculations will inevitably contain large uncertainties which make the calculated collective doses to patients unreliable. This considerably undermines the assessment of the total number of breast cancers induced by medical exposures ..."
Response:
Yes, we emphasize repeatedly in the book --- with a summary in Chapter 38 --- that the uncertainties are so large that the true past average annual doses may well have been two to four times higher than the estimates we used. but they can not have been lower! And we show amply why. Why did Muirhead and Sharp fail to mention that our calculations do not even include breast exposures from most of the diagnostic and interventional fluoroscopy, from mammography, from treatment of skin disorders, and from the treatment of dozens of inflammatory diseases?
Is it just an oversight for them to raise the dose-issue --- and then not to reveal that the uncertainties work very strongly in favor of our 75-percent estimate?
Part 8. "If Gofman Were Right"
We Should See Far More BC Than We Do
Muirhead and Sharp allege (p.13): "As well as being inconsistent with current knowledge, Gofman's risk estimates are inconsistent with the actual occurrence of breast cancer in the American and A-bomb study groups ... In essence, if Gofman was right, then far more women who were exposed to high medical doses in the USA would have developed breast cancer than actually have." In support of that assertion, Muirhead and Sharp (p.13) cite three pairs of figures and make two comparisons.
8a. Every Word of the Faulty NRPB Assertion
We will show below that one of their two comparisons would confirm our estimate, and that the second comparison obviously fails to meet a passing grade in basic epidemiology. At the outset, we set forth their three pairs of figures (doses and percentages), which they presented where we used three dots above.
(a) Fluoroscoped TB patients, USA:
Dose = ~ 79 rems. Share of Breast-Cancer due to radiation = 30 percent.
(b) A-Bomb Survivors, Japan:
Dose = ~ 25 rems. Share of Breast-Cancer due to radiation = 30 percent.
(c) Gofman Analysis (see p.267, Column T, and p.285 at the top):
Dose = ~ 28 rems. Share of Breast-Cancer due to radiation = 63 percent.
Now we will fill in the exact words of Muirhead and Sharp, where we had the dots above: "In the study of women who received high medical doses during fluoroscopic procedures for TB in the USA, the proportion of breast cancers attributable to radiation is about 30% (average radiation dose of about 0.79 Sv) and in the A-bomb survivors study also about 30% (average dose of about 0.25 Sv). However, Gofman attributes 75% of breast cancers in the American population to radiation, based on his calculated average dose of about 0.28 Sv." Then follows "In essence," as quoted above. We have omitted not a single word.
8b. NRPB's Unintended Confirmation of Our Result
We begin by explaining NRPB's third dose figure. The average dose of 28 rems (0.28 Sv) refers to this book at page 267, where we add up the annual breast doses in Column T of the Master Table --- in medical rads. This total is 28 medical rads for someone who received the listed exposures for 64 years, starting at newborn-age. It is the average dose accumulated in 64 years of life by all women which would produce 114,300 breast-cancers per year at equilibrium (p.268).
NRPB's second dose figure of 25 rems (0.25 Sv) is presented as the average dose in the A-bomb survivors --- the study which does cover all ages at exposure. Without endorsing that dose-value, we can say that it is not converted to medical rads --- because NRPB rejects that factor of 2, as discussed in Part 5c. And NRPB is mistaken on that.
Because medical x-rays are twice as injurious as bomb-radiation, it means that half the dose from medical x-rays would have produced the same results cited by Muirhead and Sharp. So, NRPB's average Japanese dose must be corrected to just 12.5 medical rads before it can be compared with the results of our analysis.
And NRPB says that this dose accounts for 30% of the breast cancers. If those cancers have been estimated for a full lifetime follow-up of all ages at bombing, then it is appropriate for Muirhead and Sharp to make a comparison with our estimate (c). Indeed, it would be absurd for them to offer the comparison at all, unless it includes a lifetime projection --- as ours clearly does.
The claim by Muirhead and Sharp is that our work is "inconsistent" with the "b" figures. Hardly! One could barely imagine a better match:
If a dose of 12.5 medical rads accounts for 30% of breast cancer in the A-Bomb Study, and if the radiation dose is 28 medical rads in our analysis, then this type of comparison means we would expect our dose to account for (28 rads / 12.5 rads) times (30 percent), or 67.2 percent of breast cancer. We estimate 63 percent (p.285, top). This is a fabulously close match. And without our appropriate adjustment for medical rads, the difference would be two-fold, not "at least ten-fold" --- as Muirhead and Sharp claim elsewhere (Part 6).
With the appropriate adjustment for medical rads, the NRPB's comparison supports our finding.
8c. NRPB's Second Comparison
We wonder how our peers at the NRPB could have imagined that the "a" figures above, from the fluoroscopy study, could possibly be compared with the "c" figures from this book, to back up their charge that this book is "inconsistent" with existing evidence. The comparison would be scientifically invalid and meaningless, even though both dose-estimates are in medical rads. Why?
It's a comparison of "apples with oranges," as we say. The first rule of epidemiology is that groups compared for radiation-response must be comparable --- most especially in follow-up time and age-distribution.
We do not know if Muirhead and Sharp adjusted the observations in the fluoroscopy study (USA) for the full lifespan. In any case, the bigger problem in this particular comparison is the age-distribution.
The "women who received high doses" did not include any infants or children, who make a large contribution to the total of our estimate (see p.268). Even for the limited age-band of fluoroscoped females, the frequencies of ages would have to be adjusted to be like the general population, before comparison with any part of our Master Table --- much less with the sum of the entire Master Table. How could peers at the NRPB offer such an inappropriate comparison?
Part 9. NRPB Preference: Better to Use American Data
"Risks of radiation-induced breast cancer calculated for the UK population by NRPB are based on studies in the American women with x-ray exposures for medical reasons, specifically those who underwent multiple fluoroscopies for TB and therapeutic irradiation for acute postpartum mastitis. These data were used by NRPB, not because of any doubts about the credibility of the data on the Japanese A-bomb survivors, but rather because there is uncertainty about how to transfer breast-cancer risks from a Japanese to a western population" (Muirhead 1995, p.13).
We agree that there is uncertainty on this point --- as we made amply clear in the First Edition (Chapter 40, Part 3).
Why do Muirhead and Sharp not admit that there is at least equal uncertainty inherent in using the studies which NRPB used?
The two biggest uncertainties inherent in the American studies come from the fact that (a) they lack information on many age-groups, including the young, and (b) they differ in risk-values among themselves by six-fold. This revelation requires just a single sentence, but none is present in the Muirhead-Sharp critique. Considering both (a) and (b), we do not agree with NRPB's preference for the American data over the Japanese data on breast cancer.
A Very Non-Protective Use of the North American Data
When Muirhead and Sharp refer to NRPB's use of "American" studies, they mean North American (see NRPB 1993, p.66/61). In the USA, the BEIR-5 Report (1990) also makes use of the North American studies --- selectively. The BEIR-5 Committee just discards the studies which show the six-fold higher risk. We quote from BEIR-5, p.255:
"Within the Canada-TB cohort, the estimated risk per Gray for women treated in Nova Scotia was about six times that for women treated in other provinces. This difference is highly significant (p<0.001) ... the higher risk observed among Nova Scotia women is not attributable to non-linearities in the dose response. Since there is currently no explanation for the difference within the Canadian-TB cohort and since the Committee was generally interested in low dose effects, it was decided to use the data on the Canadian-TB cohort without the Nova Scotia women, as the basis for risk estimates in the parallel analysis."
The excuse about "low dose effects" is unconvincing, to say the least, since the Nova Scotia Study was itself a low-dose study. The average dose per exposure was 7.5 rads (cGy) --- as discussed in our Chapter 45, Part 2d, paragraphs 14-17.
A Warning for the General Public
The effect of discarding the Nova Scotia Study is to underestimate the risk of breast cancer, per rad of radiation exposure, by a lot. Although NRPB 1993 does not reveal if it, too, discarded the Nova Scotia Study, it appears to embrace the same exclusion by not rejecting it (NRPB 1993, p.58/31). But we can not be absolutely sure what NRPB did. Our warning, therefore, may apply more to American than British people.
And our warning is this: Remember the Nova Scotia Study. Do not assume that the BEIR-5 committee --- or any other radiation committee or institute --- invariably puts protection of your health ahead of the needs of the radiation community from which its members so often come.
Part 10. NRPB on "Optimising" Doses: Talk vs. Action
Near the end of their critique, Muirhead and Sharp write (1995, p.13): "Professor Gofman's book is ingenuously written, persuasive, and easily readable by the non-expert. His main message, that all clinicians must justify exposures and optimise the doses, are those firmly held and preached by all in medical radiological protection."
We note that Muirhead and Sharp claim only that people in radiological protection preach the policy of preventing unnecessary radiation exposure of patients today. Muirhead and Sharp evade any claim about execution of the policy. They probably knew (they should have known) what Mark Lewis and Julian Bellamy learned as they prepared the 20/20 broadcast:
The radiation dose, for the same x-ray exam, varies from one British facility to another by up to 50-fold, according to an NRPB source. Yes, some variation is unavoidable. For instance, patients vary in their thickness. But such factors do not justify a variation as large as 50-fold. Such a great variation for the same common exam indicates that some patients are receiving higher radiation doses than necessary to obtain the medical information.
Any radiation above the amount necessary for a good x-ray picture is simply radiation going into unnecessary cancer-production. A risk of unnecessary radiation-induced cancer per patient, times millions of patients, is a rate of real and unnecessary cancers for a nation. We have already written about this deadly carelessness in Chapter 42. And in Chapter 48, we are adding even more information (from UNSCEAR 1993).
The PB in NRPB stands for "Protection Board." If the NRPB would replace its preaching with a verifiable program to eliminate unnecessarily high medical exposures, we would be happy to join publicly with NRPB in relieving any unwarranted anxiety about x-ray overdoses in Britain. But let's not issue pseudo-assurances at a time when some anxiety about overdosing would be a healthy phenomenon.
Part 11. Claim: The Book Offers a "False Hope"
Muirhead and Sharp wind up their critique (1995, p.13) as follows: "It is ... particularly lamentable that he [Gofman] should claim that medical radiology is the dominant cause of breast cancer, based on unjustifiable calculations, in order to promote these aims [optimised doses]. The most likely effects will be to discourage women from accepting clinically necessary exposures, increase anxiety in an already uncertain world, and hold out a false hope that cumulative dose reduction will dramatically reduce the incidence of breast cancer in the future."
Muirhead and Sharp claim that our 75-percent estimate is based on "unjustifiable calculations," and yet they have failed to discredit even one ingredient in those calculations. While trying, they revealed NRPB's use of clearly non-cautious assumptions for radiation protection (especially Part 5c), they revealed what looks like bias (Part 7), and they even managed to produce an argument which confirms our calculation (Part 8).
If this critique represents the expertise of Britain's National Radiological Protection Board, plus some input from the RERF, NCI, and UNSCEAR, then this peer-review should increase everyone's confidence that our First Edition is very reasonable --- which means that the dramatic hope offered by this work is also very reasonable.
# # # # #