This is part of my 30 minutes--right?

Good morning. As stated in the introduction, I'm going to be talking about the ah, [Rocketdyne?] study which we recently completed part of at UCLA with my colleagues in this study I should mention are Dr. Ritz Freunds and ah, Young, ah, all of the department epidemiology environmental health sciences UCLA and the ah, the study which began in ah, 19--1993--the major purpose of the study as we outlined it was to estimate the effects of ah, ah, radiation--occupational exposure to radiation on the risk of dying from selected cancers among employees of Rocketdyne and slash Atomics International.

Now, I should say that Atomics International merged with Rocketdyne which is ah, the ah, Rocket propulsion division of North American Aviation in 1984. That later became Rockwell International and recently, in 1996, was purchased by Boeing North America. So this is really now a division of Boeing. So, throughout the ah, presentation, we usually refer to it as ah, Rocketdyne slash AI for Atomics International.

Now the workers that we were studying were primarily located at the Santa Suzanna field Laboratory ah, and related facilities, between Ventura and Los Angeles County in the Simi Hills near Simi Valley. These workers were involved in a number of ah, nuclear operations including the operation of ah, 10 to 12 nuclear reactors between ah, the early-mid 1950s and the early 1980's.

Ah, what motivates me as a--as an epidemiologist, as a scientist to do this study or propose doing this thing and getting involved, was things that I'm sure that you're already, ah, very familiar with--the intriguing [science/signs of the?] controversy that ah, we know that radiation causes cancer in high doses in humans, but what's so controversial is the magnitude of the effects of low doses that are received over many years. What happens to the effects of low doses that we receive over perhaps decades as opposed to ah, a single dose such as the A-bomb survivor studies.

Ah, the results as I--as I see it are somewhat variable across epidemiologic studies of nuclear cohorts that have been conducted for this purpose. Ah, but the major question is, is there an effect for workers who receive less than 50 [milliceverts?] or five rems, if you like ah, per year over many years. And just as important, or just as perhaps controversial is exactly what cancers are caused, or potentially caused by this type of radiation? And here the result are even more inconsistent, if you look at the studies that have been done. Although there is a good deal of theoretical speculation about certain tissues and organs being sensitive to radiation--radio-sensitive, if you like--ah, the results of previous studies are sort of across the board. And there's ah, probably most agreement about the effect of low level radiation on leukemia as you all know.

Just to ah, illustrate that, let me show you the findings from the recent 1995 analysis of [C...?]--the official ah, findings I'm told, about the effects of external radiation and humans. Ah, this is my version of it, it's not entirely their version. Ah, at the bott--look at the ah--[there we go]--the bottom of this, ah, graph shows the results--ah, the bottom of this graph shows the results of what was referred to by these authors as a "pooled analysis." And this is the only positive finding they had. This is ah, essentially, a rate ratio of relative risk that epidemiologist call it, corresponding to 100 [millicebrets?] of external radiation. Ah, what these authors did is divided all cancers into two types, leukemias and everything else. And for all the other cancers, they found essentially, a relative risk of one, meaning that the ah, rate of dying from other cancers is the same for people who have more radiation exposure than less, differing by 100 [milliceverets?] or 10 rem. But for leukemia, the result that was reported was sort of a modest, relatively small effect, a relative risk of 1.22, meaning that the rate of dying from leukemias is--ah, was 22% higher in people who were exposed to say, 100 [milliceverts?] ah, of radiation than those who were exposed to none. That's a relatively, ah, small effect, and you'll notice that the 95% [...?...] crosses the no-value of 1 which some people would refer to as non significant.

Ah, but what struck me about these results is that there's was great deal of heterogeneity in the results for the different studies, ah, which they pooled. The studies was--basically took data from seven different studies. This one is actually two studies right here and if one looks at the studies individually, you see quite a bit of variation. Ah, the one big study at Hanford showed actually a somewhat protective effect, which rated highly in this pooled estimate. But you'll see some of the smaller studies actually do show a rather strong ah, but very imprecise, nevertheless, estimates of the effect.

Well, this was my reason for getting involved in this study. I should say that this wasn't--not necessarily the reason why this study was funded. This--the--our study was funded in large part through the, ah, activities of community activists in this area, who ah, got a study conducted by the State of California o--um, ah, the residents who lived near the Santa Suzanna Field Laboratory where we did most of this work. The purpose of that study which was ah, finalized and I guess it was released in 1992, was equivocal. They--they thought there might be an increase in bladder cancer incidents among the people who lived close to Suzanna--Santa Suzanna, ah, but they weren't really sure and the results were sort of equivocal. Well, the community activists decided that what should be done now ah, to get better estimates of the effects of radiation is to study the workers. So, with the help of um, several local legislators, they secured money from the department of energy which is the basis for the funding of our study.

Now, what I'm going to show you today are results from ah, really two parts of the study ah, examining the effects of both internal and external radiation. The, ah, external radiation, penetration of the body by ah, gamma and Xrays primarily, was due to exposures from ah, the reactor experiments and various other nuclear operations at Rocketdyne over the years. The internal radiation, the in--ah, the inhalation and to some extent the ingestion of alpha emitting radionucleis primarily at Rocketdyne was due to the uranium in some mixed fission products that was about 95% of the exposures and ah, these exposures came primarily from fuel fabrication operations at Rocketdyne.

The type of study that we did is a type of epidemiologic study that we call a retrospective cohort study. Retrospective means historical, which means that ah, we went back in time and identified workers who were exposed to radiation based on the records available to us at Rocketime between the 1950's and the end of 1994. Ah, and the study that we conducted--the analyses that I'm going to show you today were done entirely among workers who were monitored for radiation. And part of the reason for that actually, is connected with ah, Dr. Stewart's talk. Ah, we felt that the workers who were monitored for radiation ah, and the ver--workers who were not monitored for radiation might have very different risks of cancer, independently of any radiation exposure, so we limited our study to those who were monitored.

All subjects in our study were followed from the f--first time they were monitored for radiation, until the end of 1994, ah, and the main outcome was death from various causes, primarily cancer, and we obtained information about the cause of death from company records and primarily three different sources of ah, vital records which were used to identify and locate the death certificates of all the deceased workers.

Ah, I should point out that this study was all based on records. No subjects were examined as part of this study. And we didn't conduct any surveys of workers, either.

The ah, study population is actually two parts shown by this ah, "then" diagram. We studied a total of about 4600 workers. This is one of the smallest, ah, nuclear cohorts to be studied epidemiologically, and ah, those workers were studied between 1950 and ah, th--those people were employed at Rocketdign between 1950 and 1993. They were all enrolled in the health physics monitoring program and they were monitored for external and/or internal radiation, be--ah, during that period and in addition, everybody here had their vital status known by us by the end of 1994. That was the end of the follow-up period. And as you can see from the ah, the "then" diagram, the ah, externally monitored cohort was used to estimate the effects of external radiation, and the internally monitored cohort was used to assess the effects of internal radiation, and ah, the the co--the larger of the two cohorts was externally monitored. That's 4563 workers. The internally monitored cohort was 22 hundred ah, and 97. Notice that the ah, the latter cohort is primarily a subset of the first.

We had several ah, sources of data that we used for this study--the ah, primary one for establishing information about the workers were the personnel records that were kept by Rocketdyne over time. From that we obtained information about employment histories, job titles, gender, ah, age and their pay type which was essentially a proxy for us for socioeconomic status. We also looked extensively at the medical records, but we only got one piece of useful information from the medical records that ah, we--that I'm going to show you some results from and that is about smoking. Ah, one of the things we're going to look at is the effects of ah, radiation on lung cancer, and since smoking is such an important risk factor for lung cancer, to ignore smoking is to leave open the possibility that your results are biased because of the effects of smoking--confounded by smoking. So, although we weren't able to obtain smoking information on the entire cohort--either the internally or externally monitored, what we were able to do is to find information about smoking status from the medical records in a subset of our workers--about 26% of them, actually, ah, who were actually surveyed about smoking habits and other things, ah, three times--once during the 1960's and twice between the 19, ah, 80's and early '90's.

Now the radiation records ah, that were collected assiduously by Rocketsign over time were used to construct cumulative doses of both external and internal radiation. I'm going to express those results in milliceverts, but I will convert them every once in a while to rems. Yeah, do rems? OK. Ah, every time I--OK, I'll--I'll do it as I go along. Ah, I don't know why--I guess I'm--I'm new at this--I like the milliceverts, but...Ah, the external radiation was measured by--primarily by film badges and a little bit by docimiters that were, ah, worn by the workers. Ah, it excluded--our estimates will exclude superficial skin doses and doses to the hands and feet alone. They also exclude ah, neutron exposures for somewhat more controversial reasons. And the internal radiation was measured by urinalysis and ah, whole body and lung counts.

I should add that the ah, docimitry for the ah, this part of the study was actually done by somebody who was ah, a consultant to us from the University of North Carolina, Doug Grawford Brown, and ah, his recommendation was to limit our doses to lung doses, even though I'm going to show you results for other types of cancers.

The last, ah, source of information we had were the death certificates for all deceased workers and we ah, coded using International Classification of Diseases, ninth revision. The underlying, as well as the contributing causes of death. But the results that I'm going to show you today are--are all based, I believe, on underlying cause of death. The results didn't really change appreciably when we look at ah, associated or contributing cause of death.

Now, one of the tricks of this study, in terms of doing statistical analysis that would make some sense, is to deal with the fact that ah, 4600 workers might sound like a relatively large number, but when it comes to studying cancer, it's really not. It's--it's--is--in fact one of the smallest cohorts, even though I will show these workers were studied for a long time, ah, even when every--eventually everybody dies--but most people don't die from cancers, they die from cardiovascular diseases. So we had the ah, the difficult task of trying to ah, choose what outcomes we would use. Would we look at all cancers? Would we look at specific sites, or wh-would we do a combination. But we had to ah, sort of play off several constraints in doing this and we decided, for a variety of reasons to choose a priori--we did this before we saw any of our outcome results, to look at certain outcomes that were based, with the exception of lung cancer, on grouping cancer sites. We couldn't look at most cancer sites specifically, because there simply weren't enough of those types of ah, cancer deaths to warrant the kinds of statistical analysis that ah, we did.

So for the externally monitored, ah, cohort, we examined, ah, as outcomes, all cancers together, since a number of other studies had done the same. We examined cancers of the blood and lymph system which included leukemias, but did exclude, ah, chronic lymphocytic leukemias. We ah, grouped all the solid cancers that were regarded as involving ah, radiosensitive tissues and organs according to [Bier-5?] which are the ah, lung, esophagus, colon, stomach, brain, breast, etcetera, and then we looked at lung cancer alone because that was the most common type of cancer death and we could--we had the numbers to do that, and then as a check we also examined the association between radiation dose and other non-radio sensitive cancers according to [Bier 5].

For the internally monitored cohort we were examining the effects of internal radiation. We--we grouped the ah, the cancers a little differently because our a-priori knowledge of what we should expect is a little different and we didn't think that the [Bear-5?] ah, theoretical basis would necessarily apply to ah, the effects of radio-nuclei. So we grouped them a little differently. First we looked again at all cancers. Again, looked at cancers of the blood and lymph system, excluding CLL, we looked at lung cancer specifically, and ah, the next group is actually a little different--and as far as I know hasn't been grouped this way in other ah, radio-epidemiologic studies--we looked at cancers of the upper air or digestive track collectively, grouped as one. That includes cancers of the oral cavity and pharynx, esophagus and stomach. And we also looked at cancers of the urinary tract--exit organs, if you like, including the bladder and kidneys.

Now, I don't want to ah, bore you of--with--with our statistical methods, at least not at this point, unless you ask me a question that forces me to, but I would want to just comment on the basic strategies we used to analyze the data, because it's very important for understanding how our results got distributed and understood by the public, if nothing else.

We used two approaches to analyze the data which are ah, sort of listed on this slide. External compar--what we call external comparisons and internal comparisons. An external comparison is where we take the entire cohort of workers and compare it, in this case, to the U.S. population using published data. And ah, the results of this are expressed in terms of what we call SMLR, STandardized Morbidity--ah, Mortality ratios rather and [c...symbols?] that are done by gender separately for men and women and adjusted for age and calendar time. An SMR is essentially a--a--a ratio, sort of like I had mentioned before. It's the weight, ah, in ah--in the worker group compared to the rate in the U.S. population, adjusted for those other variables.

Now the main reason for doing these external comparisons was not to assess the effect of radiation. On the contrary, all we did this for was for descriptive purposes and we want to understand, for example, how the workers at Rockezite might be--and we expected them to be--healthier than the general population, in part because they were, ah, working and because they had been s--been selected and retained in this, ah, radiation monitoring program over the years, so we expected the workers at Rocketdign to be healthier in--in general respect than the general population, because they were selected to be healthier.

Ah, in order to ah, test the hypothesis, the main approach we used were call internal comparisons, that is, we ignored the U.S. population and we contrasted different levels of radiation--ah, cumulative radiation within the worker cohorts. So we compared workers who were exposed to say, more than 200 ah, millicevers (20 rem) to those who were ah, exposed to less than one rem. That sort of thing. And we did it using statistical techniques called conditional--just a progression--and again, the results were expressed in the forms of rate ratios or what was often referred to as relative risk, the ratio of the rate in one group, divided by the rate of--of dying from cancer in another group, with 95% conference intervals, and we also were able to adjust for a number of variables that might confound our results. The results that I'm going to show you are all adjusted for age, ah, time since they were first monitored, the ah, pay type variable which is a proxy for socio-economic status, and the cumulative dose of the other type of radiation. So when we looked at the effect of external radiation, we controlled for internal, and vice versa.

And um, the cumulative radiation doses in our analysis were all treated as time-dependent, meaning that every time a death occurred, we reconstructed the doses for everybody in the cohort who was still alive at that time, and we [lagged?] the exposure measurements by anywhere from zero to 20 years, in other words, we deleted the exposure between 0 and 20 years prior to the time of each cancer death to take into consideration among other things the fact that ah, dying from cancer has a certain induction and latency period.

What we found as I show you in this slide, is ah--this is--this is a brief description of the two, ah, populations, the externally monitored and the internally monitored population. You notice that almost all of them are male. Um, they were followed for a fairly long period of time including those who actually died--it was 26 or 25 years in each of those cohorts. Ah, there were a total of 875 deaths, of which 258 were from cancer, and the cancer mortality rate was about the same in the internally and externally monitored cohorts and was also approximately the same as the ah--Well, let me--I'll get to that in a second--the ah, pay type distribution was also sort of distinguished. Notice that there's a large proportion--about 45 or 41 percent of these two cohorts were ah, relatively high ah, socioeconomic status. They were salaried workers in the managerial or professional track or technical administrative people. So only a relatively small number were, ah, [...?...] were--were--were hourly workers, which would be more typical of the U.S. population.

Ah, the radiation doses themselves were quite low in our cohort, which I acknowledge was quite disturbing, and--and [certainly?] pessimi--quite pessimistic about this after I saw what we had. Ah, we found that the doses were even lower than we had expected. This is sort of a brief distribution on the left for external ah, cumulative radiation and on the right for internal--these are cumulative doses broken down into categories and these percentages represented a portion of each of the ah, cohorts that are in that category. Notice that about 60% of the, ah--these two--60% of the external monitored cohort had, ah, doses less than 5 millceverts--half a rem.

And in the internally monitored cohort, the same thing is true--slightly different categories, but notice that nearly 60% had a zero cumulative dose--total cumulative dose, so that the mean doses, as you can see, are really quite small for ah--relative to other nuclear cohorts that have been studied epidemiologically.

Now, let me show you some of the results. Ah, the first results I'll show you just so you can appreciate um, how this study has been recognized in the--in the media, are the results of our external comparisons. This is comparing the Rocketsign population to the U.S. population in the form of SMRs. Now an SMR of 1 means that the rate of dying from the disease listed on the left is the same in the U.S. as it is in the Rocketsign workers, controlling for age and--and calendar time, and you'll notice if one looks at all deaths ah, together, that this, ah, SMR is much less than one for the Rocketime workers, meaning that the Rocketime workers had a much lower all cause mortality rate than the general U.S. population, and that was particularly true for cardiovascular disease and a little less so for cancers. Nevertheless, ah, ah, in lung cancer in particular it was only 75% so the rate of dying for the U.S. population. The only, ah--the only outcome, the only cancer or any other cause of death which had ah, an SMR greater than one was leukemias, in other words, despite the fact that there was a tendency for a much lower rate of death due to cancer in this cohort--79%--still leukemia was higher--the rate of dying from leukemia was higher in the Rocketdign cohort than in the U.S. population.

Um, does that take out of the 30 minutes? Oh, OK.

All right, now for the primary results our dose response analysis, the internal comparisons, let me just, ah, tell you a little bit about what's in this table. These are the results from all of the ah--all the major outcomes. Total cancer, ah, blood and lymph systems cancers, [radiosensitive?] solid cancers, and lung cancer. And for each ah, outcome here, you see the cumulative dose broken down into four categories, less than 10 milliceverts, less than one rem is the referent group. In other words we're going to compare the rates of dying from cancers for the other groups to that group. So, it--compared to itself the relative risk or rate ratio is one.

Ah, these are the ni--ah, these are the estimates of the relative risks and these are the 95% [conference?] intervals, this is the two sided [P..?...] that corresponds to the test [...?...] hypothesis that there's no association. Ah, you'll notice that with ah, cancers of the blood and lymph system and canc--and lung cancer specifically, the rate is ah, fairly much higher in those, ah, workers who were exposed to more than two--two hundred milliceverts--20 rems. Ah, both for blood and lymph and lung cancer. Notice--notice that the [conference?] intervals exclude the [null?] value of 1, suggesting that there's an excess rate of dying from these cancers in persons--in workers who were exposed more--to more than 200 milliceverts.

But there isn't exactly a dose response relationship. You don't actually see an excess in the next lowest category, for example, and very little here for lung. Ah, the test--therefore the test of the null hypothesis yields a P-value, which is, ah, small, but not necessarily as small as it might be because this is not actually a--a dose response relationship.

Ah, for total cancers, the results are actually similar, reflecting the effects of radiosensitive solid cancers and lung cancers. Notice that there is sort of a dose responsive relationship. The higher the dose, the higher the relative risk of dying from cancers, so that there's a three-fold increase of ah, dying from all cancers among workers who were exposed to more than 200 milliceverts, relative th--those who were, ah, exposed to, ah, less than ten--one rem, and that ah, P-value is .036, suggesting that the likelihood that this result is due to chance is rather small.

Ah, the results for the radiosensitive solid cancers are actually fairly similar to lung cancer. Ah, but the fact that that's true suggests that the observation, the association we observe for solid cancers isn't apparently limited to lung. It may also correspond to other cancers that are involved in that grouping as well. But again, we didn't have enough of those cancers to actually sort it out.

Um, in addition to looking at the overall effect of ah, cancer, we also wanted to consider the effect of--of the age at which you were exposed, and ah, to be brief, because I've got five minutes, ah, this here--this slide shows how the effect of ah, radiation depends on the age at which you were exposed. Ah, subject to the fact again these are very small numbers, we see ah, sort of a striking trend for all of the solid cancers and for total cancers. For example take a look at one. The relative risk of 100 milliceverts now expressed as a--as a ah, actual exact amount, 100 milliceverts, gets larger as you get older. In other words people who ah--workers who were expol--exposed after the age of 50, had a higher relative rate of dying from those cancers than workers who were exposed at younger ages. That doesn't appear to be true of cancers of the blood and lymph system. In fact, it may even be the opposite although the numbers are rather small to conclude that.

Ah, let me just ah, show you the key results for the internal, ah, monitored group--the effects of internal radiation. This re--the results here are set up sort of similarly to the ah--the ones I showed you for external radiation. Here each of the ah, major outcomes--oh, I should add that when we looked--in the external radiation, when we looked at non-radiosensitive cancers, the cancers that were supposedly not sensitive to radiation, we found no association between radiation and those types of cancers, which is what we would expect. Ah, here we looked at the effects of internal radiation--again, the categories are slightly different, but there are four of them, again, and again, the results are expressed in terms of relative risks or rate ratios in 95% conference intervals and P-values, and here the results are ah, rather striking, because for cancers of the blood and lymph system, as well as for this, ah, grouping called upper aerodigestive tract cancers, you see a rather striking dose responsive relationship between ah, the amount--the cumulative dose of ah, radiation, and the rate of dying from those cancers. So this is a ra--this is something you don't usually see this strongly in an epidemiologic study. Ah, this means, for example, that the ah, people who were exposed to 0 to 5 millicevers, zero to a half of rem ah, had more than twice the rate of dying from ah, blood and lymph system cancers as ah, workers who were not exposed to any internal radiation according to our records.

And similarly for the upper aero digestive tract cancers there's even a, again, a very striking dose-response relationship. Ah, same idea. interestingly, though, for lung cancer, we found no association. In fact, there's even a suggestion that the results sort of go in the opposite direction, although there's certainly not significant. Ah, so, let me, ah, summarize, or complete some ah, what we found at this point.

There are a number of limitations of this study, which I, unfortunately don't have time to ah, talk about, but I want to mention one. Ah, there's of course the always possibility that the results that you--we obtain in a study like this are confounded by other--other types of exposures and that's particularly relevant for lung cancer because smoking is such an important risk factor for lung cancer and we didn't actually adjust for smoking, but we were able to examine the relationship in a--in a sub cohort of our workers--the relationship between radiation and smoking and we found none. We found that among those workers where we did have available smoking information, there was no association between how much radiation--internal or external radiation they were exposed to, and their ah, their smoking status--whether they were a current smoker or former smoker or even how much they smoked. That suggests that ah, smoking didn't confound our results. That our results aren't due to ah, bias as a result of other factors such as smoking but they are more likely then to be the results of the exposures that ah, we attributed them to--radiation. So, the conclusions of the ah, of our study, based on--despite all the, ah, limitations and the problems that we had in--in putting together the data, ah, the limita--the conclusions are essentially that all available information from this study suggests that ah, prolonged exposure to low-level radiation--ionizing radiation in the work place, even at levels generally considered safe, has increased the risk of dying from certain cancers. And perhaps more importantly, that the cancers that are affected, or appear to be affected in our study by low level radiation, aren't limited to leukemias or even cancers of the ah, blood and lymph system, but also seem to ah, reflect the effects on solid cancers, specifically external radiation can affect the risk of dying from lung cancer and possibly other radiosensitive cancers, and we also found in internal radiation may affect the risk of dying from cancers of the upper aero-digestive tract, which is a--actually a new finding and will have to be replicated admittedly in other populations.

Ah, we also found that the effects of external radiation depend on the age at which you're exposed and that that pattern of effects due to age might vary by the type of cancer, ah, that the older you are, the more--the worse the exposure is, with respect to getting solid cancers like lung cancer. And finally, ah, since our study was based on deaths, only 20% of the workers in our cohort had died by the end of the study. That meant 80% are still alive at the end of our study. So we recommend that those workers be followed in the future. Thank you.