Thurman Wenzl

How difficult to have to follow Tony Mazzocchi at the end of the afternoon. He is a hero of mine, for having played an important role in encouraging cooperation between public health scientists and labor activists in the late 1970s. In several cities, local Committees on Occupational Safety and Health were begun then, which made it possible to start new worker education efforts and to encourage dialog on the connections between regulatory concerns and the scientific uncertainties concerning workplace exposures. Many of us took a lot of encouragement from Tony's organizing work, both then and now. I don't think that there need be any conflict between continued health effects research and public pressure for improved worker protection; I think that the research can help us best set priorities so that we use our energies to control those exposures which we best understand to be hazardous.

I'm going to shorten my talk a bit, since several earlier speakers have made the same points that I was going to make. Hal Morgenstern, Gregg Wilkinson, and Steve Wing have already covered some of my points. My conclusion will be that recent studies of radiation exposed workers strongly suggest that exposures allowed by current standards and guidelines, may still confer excess risk of cancer death.

Before I cover these recent scientific findings, I'd like to summarize how my agency, the National Institute for Occupational Safety and Health (NIOSH) came to have responsibility for directing workplace epidemiology at Department of Energy (DoE) sites. In the mid-1980s public concerns were expressed about potential conflict of interest in the ways that the DoE was directly funding epidemiologists to study the possible health effects at their plants. Following congressional testimony on this possible conflict, the Secretary of Energy at that time, Admiral Watkins appointed a 'Secretarial Committee' of independent scientists. They recommended separating the health investigations from the agency that directed the sites themselves. Through negotiations that followed that recommendation, the Department of Health and Human Services (HHS) agreed with the DoE that HHS would take control of analytic epidemiology associated with past worker and community exposures. So, I'm from NIOSH - one part of HHS which carries out workplace health research. We have a sister agency at the Centers for Disease Control in Atlanta, called the National Center for Environmental Health, which carries out community environmental epidemiology, among residents near the DoE sites but not inside them.

We make independent scientific decisions at NIOSH. We are indirectly funded by DoE, via the HHS and CDC channels, but we make our own decisions about how studies are conducted and interpreted. We also listen very much to workers, to a scientific advisory committee and to community groups for their input on research priorities. And we make every effort to involve a wider range of scientists among the researchers working on these health questions. Most basically, we are studying what lessons can be learned from these past radiation exposures that will help us better protect workers who may still be exposed to radiation at work.

Radiation protection standards and guidelines are both based on the past exposure patterns of the survivors of the bombings at Hiroshima and Nagasaki, even though those patterns are very different from those to which current workers are exposed. Workers generally receive their exposure spread out over an entire work-life, while the bomb survivors got all their exposure 'at once.' It is not at all clear that the biological effects are the same.

Overall, previous epidemiologic studies of workers at single DoE sites have been inconsistent, with some showing excess cancer risk associated with workplace exposures, and others not doing so. Hanford and Savannah River are 2 of these sites which are primarily devoted to the production of nuclear materials, with reactors and the facilities for separating plutonium from the irradiated rods. On the other hand. Los Alamos and Oak Ridge are laboratories with the associated facilities for actually making the first atomic bombs.

One thing to keep in mind when interpreting these studies is that worker mortality rates are often lower than those of the population as a whole. That was also noted by Dr. Morgenstern in his presentation on Rocketdyne earlier today. One reason why we focus on mortality studies is that there are nationwide registries to determine who has died - for example, long after they have left work at a DoE site. It is more difficult to study non-fatal disease, since there do not yet exist comprehensive registries of disease diagnoses. As a result we carry out mortality studies to better learn the health consequences of past exposures at the DoE sites. One difficulty with these single workplace studies is that approximately 20% of past workers have died to date, and that for rare cancers there is often inadequate sample size to detect an effect, even if the effect is there. Also, the percent of highly exposed workers is not huge. For example, in many of these worker groups, only about 15-20% of the entire cohort has a cumulative exposure above five rem. Thus the range of past exposures is not large, hindering the ability to detect harmful effects even if they are present.

What is being done about improving study quality to resolve these inconsistencies? One important strategy has been to pooling sites to increase sample size, to improve our abiltiy to detect increased risks for rare outcomes like leukemia - which only makes up about 1% of all deaths. Let me summarize two well- done recent pooled epidemiologic studies of workplace risk at nuclear facilities. What is challenging about this work, is that despite the improved methods, one of these studies showed increased lung cancer but the other showed increased leukemia risk.

Ed Frome is a statistician in Tennessee who has studied a combined Oak Ridge cohort of 106,000 workers. For 28,000 of these workers had exposure monitoring data he carried out dose-response analyses, to learn if cancer risk increased with increasing cumulative dose. He found that there was a dose-response relation for lung cancer mortaility, while a comparable pooled study by Elizabeth Cardis showed a dose-response effect for leukemia, other than the chronic lymphocytic type (CLL). Leukemia is actually several different diseases, and CLL is thought to not be related to radiation exposure. Despite this inconsistency, these studies suggest that there well may be excess risk from workplace radiation exposures.

Let me now illustrate how we might interpret the Cardis study on leukemia, with a realistic example of a radiation exposure pattern that could happen today. This will be slightly different than the way that some agencies present their risk estimates. One traditional way of estimating risk, from the International Commission for Radiation Protection (ICRP), is to combine deaths from all types of cancer, and then compute how many cancers might result, on average for 2500 workers exposed to one rem of cumulative exposure.

I think that it is more informative to look at one well-done recent study, that by Cardis above, and use its results to illustrate how we might estimate excess risk at currently allowable exposures. For this example, I will choose a group of workers who might be exposed to 750 millirem per year for each of their working years. Such an exposure is significantly below current standards and guidelines, which are 5000 or 2000 millirem per year, respectively. According to an exposure summary for 1995 for the combined DoE sites, approximately 300 workers were exposed to at least 750 millirem per year. Now suppose that these workers are exposed to that level for each year of a working lifetime, say 40 years. We compute cumulative dose by summing each of the annual doses over all the years worked. In this case, this total comes out to 30,000 millirem, for a cumulative lifetime dose of 30 rem. Now we can look at the dose-response curve for leukemia mortality from the Cardis study, to estimate the excess risk for these hypothetical workers.

On the left of Figure 1, the vertical axis represents relative risk (RR) for leukemia mortality; the horizontal axis represents cumulative radiation exposure in rem. The line sloping upwards to the right shows the increase in the RR for leukemia mortality as the cumulative workplace radiation dose increases, based on the findings of the Cardis study. Where the line intersects the vertical axis at RR=1, that represents the background risk of leukemia mortality for those with no workplace radiation exposure. Now if we move over to 30 rem cumulative and go up to the dose-response Risk line and over to the vertical axis, you can see that the relative risk is 1.6 for this cumulative dose of 30 rem. That is the same as saying that such a group of exposed workers would have an estimated 60% excess relative risk of leukemia mortality. In this way we have estimated risk associated with an annual exposure of 750 millirem, well within the current guidelines.

This is just one example of how we can conclude that there is moderately strong evidence that there is excess risk of leukemia mortality at currently allowable exposures. At this point, not all recent studies of leukemia are consistent with these results from Cardis.

Briefly I'd also like to summarize how we might compare worker studies to community studies of possible radiation risk. Dick Clapp earlier gave a very interesting example a community study that made every effort to figure out whether there was a connection between powerplant radiation emissions and community cancer incidence in nearby towns. Because workers are usually closer to the radiation sources, their exposure is often higher than the exposure of nearby community residents. That exposure is also much better documented. So, if we are looking for a population to detect health effects if they are present, then we would rather study workers. This is not to claim that all community studies are bad - but I am suggesting that if we are seeking to clarify uncertain health effects, the worker studies is more likely to find it. Also, list of potentially exposed workers is usually available. For example at the Savannah River Site, the DuPont personnel department can tell us everybody they've hired during the years that they operated that plant for the DoE. On the other hand, there are no comparable of residents who have ever lived next door to a given plant, so it's difficult to define the group under study for possible health effects. For most of these chronic disease studies, it is the distant past exposures which are the relevant. We are not so interested in recent years' exposures for cancer studies, because of long disease process which we believe is involved, with lung cancer for example. These past exposures for both community and worker groups are often very difficult to estimate, but at least for radiation, most of the exposed workers were regularly measured for external radiation exposure, starting in the mid-1940s.

Let me conclude with a few examples of some the research which is now underway at NIOSH to resolve these inconsistencies. We now have underway combined investigations of leukemia and lung cancer, where we are pooling DoE sites and evaluating exposures more carefully. I'm directing a leukemia case-control study that has twice as many cases as the Cardis study to which I've been referrring. We are trying to clarify these leukemia uncertainties with a larger study, which has more careful exposure assessment. We are also initiating a lung cancer study at NIOSH, since Frome and others have found indications that lung cancer mortality is associated with external radiation exposure.

Cardis, who I've mentioned above, works at the International Agency for Research in Cancer (IARC) in Lyon, France, where NIOSH is among the many sources of research funds. Following up on the earlier study, they are now pooling groups of exposed nuclear workers from 14 countries to clarify the inconsistencies which I've been mentioning. With a combined cohort size of more than 500,000 workers, they will be able to clarify whether rare cancers like leukemia are actually associated with cumulative radiation dose which results from gradual exposure over a working lifetime.

One of the other initiatives is to better evaluate women's mortality from radiation exposure at the DoE sites. In most of the past individual site studies, not many women have been included, so we know little of their possible risks. To correct this problem, Gregg Wilkinson (who spoke earlier) is now completing a combined study of women's mortality across 14 sites; he hopes to have an adequate population size to detect any adverse effects if they are present. At Rocky Flats, outside Denver, a cancer incidence study among workers is underway - where the goal is to detect whether there are non-fatal cancers which might be associated with past exposures to plutonium or beryllium. At Idaho National Engineering Lab we are conducting a large mortality study, to learn whether workers who have designed reactors and separated the isotopes from the rods following their irradiation are at any increased risk from their exposures at work.

In addition to past exposures at DoE during the Cold War, many workers may now be exposed in the many cleanup operations that are underway. We could investigate possible acute effects of current exposures, but we anticipate that five to ten years in the future we are going to be asked whether there are any chronic diseases associated with exposures to current cleanup workers. Right now we don't believe that the exposure and work history records are being retained in a way that will allow such a health study to take place. We expect to be making recommendations to the DoE about improved record keeping for their subcontractors and contractors.

One further point that I would like to highlight here, is that there are many radiation-exposed workers outside the DoE complex - in hospitals, dental labs and in nuclear power plants - who could logically ask if they are at increased risk. I can't offer details on their range of exposures at work, but the National Cancer Institute has already published one study which suggests that X-ray technologists who were first employed before 1940, have excess breast cancer risk that is associated with their duration of work. Is this breast cancer effect due to their external radiation at work? We really don't know yet. But this is one example of a non-DoE group of workers which may very well be at excess risk. In summary, then, there may well be excess risk at currently allowable radiation exposures, and many workers outside the DoE complex are exposed to radiation but their risk potential remains poorly studied and uncertain.