The nature of this subject is such that new data, and new analyses, and new claims made by other analysts, appear all the time. Therefore, it is our intention to up-date this book whenever it would be appropriate -- either because of new evidence, or because of new claims made by others.
While we were writing this book, the United Nations' radiation committee, UNSCEAR, released its new report in early 1989 (the reference is Un88). And on December 19, 1989, the (U.S.) National Academy of Sciences's radiation committee, BEIR-5, released its new report too (Beir90).
After studying those two updates from the radiation community, I find the need for an independent analysis to be as great as when I began. Perhaps greater. The UNSCEAR and BEIR reports do not examine some of the central issues in this field.
Comparisons Made :
For the convenience of readers, we have set forth the key differences between this book and those two reports in Chapter 25. Although the differences are mentioned in Chapter 1 also, very few readers can expect to understand the origin or consequences of the differences until Chapter 25. After all, if such matters were self-evident, the intervening chapters could have been omitted.
The topic of radiation-induced cancer from low-dose exposure is obviously not closed. In the most valuable low-dose study of all -- the Atomic Bomb Survivors -- more than half the people are still alive. The full radiation-response of the most important group -- those under 20 years of age, at the time of bombing -- is far from established in terms of excess fatal cancers.
This edition of our book carries its analysis through the most recent epidemiologically valid data from the A-Bomb Study (1950-1982). We have addressed the issue of why the data between 1982-1985, as currently available, cannot be regarded as valid by the common standards of epidemiological science.
Indeed, one purpose of this first edition is to help re-direct the reporting of data from the A-Bomb Study into sound epidemiological lines, for otherwise, this uniquely valuable study can become scientifically meaningless. Our position on this fundamental issue constitutes a profound difference with the 1988 UNSCEAR and the 1990 BEIR reports, both of which accept the practices to which we object. In our judgment, such practices can lead to the death of epidemiology as a science.
"Questionable Practices" Examined :
Partly because radiation research has been so well funded, there are far more data about ionizing radiation as a potentially toxic agent than there are about many other agents to which entire populations are exposed. Thus the field can be regarded as the "canary" which can warn humanity about practices which mean "trouble ahead," if adopted in other fields of toxicology.
The hazard which society needs to recognize is that there are ways in which preconceived notions can enter into scientific health matters, and that, in all research concerned with toxic assaults on health, it is possible to stand medical knowledge on its head. Indeed, medical "unknowledge" can become dominant, if certain practices are handled in scientifically inappropriate ways.
Among the practices questioned by this book are the retroactive alteration of databases, the replacement of actual observations by preferred hypotheses, the artificial constraint of equations to rule out certain dose-responses, the subdivision of data until even the largest database becomes inconclusive, and more (see Index, "Scientifically Questionable Practices").
It is possible to mishandle the evidence concerning radiation injury in such a manner that the next 100 years of human history will be characterized by total medical "unknowledge" in this field, instead of knowledge.
In this first edition, we have endeavored to suggest differences between some sound and unsound practices in epidemiological research.
In subsequent editions of this book, we intend to add an examination of whatever new evidence and new reports warrant consideration, and to state our evaluation of whether the field is moving in the direction of reality-based estimates of hazard, or in the direction of estimates based on unfounded conjecture.
The period ahead may be particularly difficult for keeping the record straight, with regard to radiation risk. Those seriously concerned with realistic estimates will need to be careful in separating speculation from real evidence -- a topic examined in Chapter 35. I predict the pitfalls here are going to grow, rather than diminish.
No Retreat from Human Evidence :
In this book, we have not glossed over uncertainties which presently exist in quantifying the cancer-hazard from exposure to ionizing radiation. Only time -- and the preservation of scientific legitimacy for important studies like the A-Bomb Study -- can resolve them.
However, the fact that there is a range of uncertainty in current risk-estimates cannot justify a retreat to wishful thinking, and cannot justify the replacement of the disturbing human data by more "optimistic" (less disturbing) data from other species. When human evidence of good quality is available -- and it is -- then human evidence is the substrate from which we will derive our estimates of risk. Readers will see those estimates emerge from the current evidence, step-by-step.
By contrast, the 1988 UNSCEAR and 1990 BEIR-5 reports rely heavily on non-human evidence to arrive at their recommended risk-estimates for low dose-rates and (UNSCEAR only) for low total doses.
Semi-Prudence versus Prudence :
The human evidence has developed over time and will continue to do so. At each new stage in its development, the growing body of evidence yields a better and better estimate of risk. The estimate which "falls out of the data" at any stage is the so-called "central value" and it is also called "the best estimate" because it is the value most likely to be right, within the evidence. So this "best estimate" is the appropriate one for use, of course. And it has a range of uncertainty, indicated by upper and lower confidence limits.
Nonetheless, there are people who try to focus the public's attention on only the lower limit of the current estimate. Therefore, a reminder is needed: If a central estimate is "off," the health risk may not be lower than the central estimate. The health risk may be greater than the central value which is being used.
Thus, from the point of view of health, the scientifically best estimate is only semi-prudent.
John W. Gofman, January 1990.