Document GKpV33DvG7pEyaZbbj2wDoN0v

R&S 114715 'fiio-HEDICALRESEARCH DOCUMENT DESCRIPTION FORM Duplicate .'in all cards; 63 - 68 69 76 L ff 1 year as-1961- File number [Right justify [Numeric only] Author (s), as Last Name FS (No Punctuation) and coden for journal as JAMA preceeded by one blank space 1 20 21 40 41 77 78 Sub-Index Code 60 61 62 11 12 13 Title of Report; end with space-hyphen-hyphen-space. Follow with Index Terms, separated from each other with comma-space. Avoid other punctuation; Brief Summary 12 10 SUMMARY: 31 32 61 62 61 62 63 64 VC 0000 VC ru<r Amimcan Journal or Epidemiology Copyright 1981 by The Johns Hopkins University School of Hygiene and Public Health All right** reserved Vol. 1X4, No. 5 Printed in USA POWER CONSIDERATIONS IN EPIDEMIOLOGIC STUDIES OF VINYL CHLORIDE WORKERS JAMES J. BEAUMONT1 and NORMAN E. BRESLOW1 Beaumont, J. J. (NIOSH, Cincinnati, OH 45226) and N. E. Breslow. Power considerations In epidemiologic studies of vinyl chloride workers. Am J Epidemiol 1981;114:725-34. Nine retrospective mortality studies of workers exposed to vinyl chloride were reviewed to determine whether differences in their hypothesis testing results might be due to differences in statistical power. Where possible, the power of each study was calculated for cancer of the lung, brain and liver. When power was taken into consideration, the results for liver and brain canc r were found to be consistent with an etiologic role for vinyl chloride. For lung cancer, the data were not consistent with an etiologic role, in that two studies with very high power yielded negative results. epidemiologic methods; respiratory tract neoplasms; brain neoplasms; liver neoplasms; vinyl chloride k The nine mortality studies of workers sites of cancer--liver,: brain and lung-- Exposed to vinyl chloride that have been was calculated, and their hypothesis test completed to date disagree in their hy ing results, positive or negative, were pothesis testing results for some causes of considered in the context of power. death, and for those causes it is difficult to draw conclusions about excess risk Methods (1-12). The studies that are negative for Power calculations a cause of death are especially difficult to Statistical power is the probability of interpret, since they may or may not have not overlooking an excess risk, i.e., of not had the statistical power to detect an ex making a Type II statistical error. To aid cess risk if, in fact, one existed. To aid in in the interpretation of this frequently the interpretation of the studies, their confusing subject, a brief review of Type I statistical power with respect to three- and Type II errors follows. Investigators are most familiar with Received for publication November 14,1980, and in final form April 13, 1981. 1 Industrywide Studies Branch F-8, National Institute for Occupational Safety and Health, 4676 Columbia Parkway, Cincinnati, OH 45226 (address for reprint requests). * Dept of Biostatistics, School of Public Health and Community Medicine, U. of Washington, Seattie, WA 98195. The authors thank Jay Lubin, Richard Waxweiler, Thomas Fischbach, Noel Weiss and William Halperin for their valuable advice. Their special thanks go to Philip Landrigan for encouraging this approach. . This paper was presented at the Conference to ReEvaluate the Toxicity of Vinyl Chloride, Polyvinyl Chloride and Structural Analogues, Rockville, MD, March 20-21,1980. Type I errors. Ifthe true state of nature is the null hypothesis (no increase in risk), then a Type I error is wrongly rejecting the null hypothesis and declaring that there is an increase in risk. The probabil ity of a Type I error is known as alpha, or more commonly, the "level of signifi cance." For example, when an inves tigator finds an excess and declares it "significant at the 0.05 level," in a sense what he or she is saying is that there is less than a 5 per cent chance of having made a Type I error. Type II errors are less familiar but, 725 726 BEAUMONT AND BRESLOW nevertheless, very important (13). Under lying this concept is the fact that when the true state of nature is the alternative hypothesis (increased risk), the research goal is to detect it (i.e., reject the null hy pothesis). Suppose one has specified an alpha level (0.05, for example) for testing a null hypothesis of no increased risk for a particular exposure. If an exposure is harmful with a relative risk ofR, then not rejecting the null hypothesis (i.e., wrongly accepting the exposure as harm less) is called a Type II error. The proba bility of a Type II error is usually denoted beta 03). Conversely, the probability of correctly rejecting the null hypothesis and therefore of detecting the excess risk is called the power and is equal to 1 - /3. Thus, power quantifies the ability of a particular study to detect an excess risk that truly exists. It is intuitively clear that with a fixed amount of data there is a greater likelihood of detecting a large ex cess risk. Similarly, an increase in the amount of data increases the chance of observing a given risk, i.e., increases the power. The approximate power 1 -- p of the vinyl chloride studies to detect a relative risk R at the alpha level of significance was calculated from the following for mula, which uses the fact that the square root transformation stabilizes the vari ance of the Poisson distribution (see the statistical appendix): Z,_, - Za - 2(Vj - lXVEj Here Za denotes the upper 100o per centile of the standard normal distribu tion and E the expected number of cancer deaths based on general population rates. This approximation agrees well with exact power calculations based on Poisson probabilities made by Cutler et al. (14, 15). When the discreteness of the exact test based upon Poisson probabilities is accounted for, the approximate and exact power curves are virtually identical. A family of power curves based upon the approximation is shown in figure 1, where each curve is for a different rela tive risk. It can be seen that as the as sumed relative risk increases for a given number of expected deaths, the power also increases. Similarly, as the expected number of deaths increases for a given relative risk, the power also increases. Application to vinyl chloride literature The assumed relative risks for the power calculations were the median (ap proximate) standardized mortality ratios reported for each cancer site in the vinyl chloride literature. Power calculations are usually performed with a somewhat arbitrarily chosen relative risk, but since relative risk information was available from the vinyl chloride studies, it was thought best to make use of these data. Separate assumptions were made for analyses considering all person-years at risk and for analyses considering only person-years after a minimum latency (time since first exposure), because in the former case the risk was diluted by the 10 to 20 years that are often required for cancer to develop after exposure to a car cinogen. The median reported mortality ratio was not used in one instance: for overall lung cancer, where the median was 1.03, essentially no excess risk. Since some risk needs to be assumed to calcu late power, the arithmetic mean (approx imately 1.5) was used. The statistical powers of mortality studies of workers exposed to vinyl chloride were calculated with respect to three sites of cancer: liver, brain and lung. These three sites were chosen be cause they have been the subject of most of the discussion of causality in the vinyl chloride literature. The powers of the studies were subsequently plotted on power curves for two purposes: to show the variability in powers, and to relate positive and negative findings to power. For the purposes of this review, a positive finding was defined as an excess risk for a R&S 114717 POWER CONSIDERATIONS IN VINYL CHLORIDE STUDIES 727 EXPECTED DEATHS Figure. 1. Approximate power curves for various assumed relative risks in standardized mortality ratio studies (for one-sided comparisons with alpha 0.05). particular cancer with statistical signifi cance at the 0.05 level, using a one-sided Poisson test. It was expected that if vinyl chloride was carcinogenic for an organ site, there should be a pattern: the studies with high power should, in general, be positive, and the studies with low power should, in general, be negative. This assumed that exposure would result in a constant value of R across the various populations. The assumption was somewhat tenuous be cause, as discussed below, the populations differed in a number of respects. Results The historical prospective studies of vinyl chloride exposed workers are listed table 1. The sizes of the study popula- tns varied considerably, from 255 in the idy by Nicholson et al. (3) to 9677 in the Equitable Environmental Health study (6, 7). The reports were not entirely in dependent; for example, many of the workers in the study by Ott et al. (10) were included in the Equitable Environ mental Health study, and there was an overlap of about 800 workers in the studies by Waxweiler et al. (4) and Wax- weiler (5). It should be noted that the studies were dissimilar in many other ways. As can be seen in table 1, a minimum exposure of 1-5 years was required in some studies, while others required only one day. Some studies reported findings based upon all person-years at risk, some included person-years at risk only after a mini mum time since first exposure, and some reported both types of analysis. The con trol (standard) populations used were dif ferent: four of the five American studies 728 BEAUMONT AND BRESLOW used United States rates, the fifth used Texas rates; and the British, Swedish and German studies used their respective na X XX X X tional rates. There were also minor differ ences in the International Classification of Diseases codes included for the specific X XX XX X cancers. Finally, there were probably dif ferences in age composition which could be important if an effect were dependent o X XX xxxxxx upon age. While it was felt that none of these dissimilarities was important enough to prevent comparison of statistical powers, it should be noted that factors other than I 8 1 .-Cs XXXX power could have had a bearing on the ability ofthe studies to detect excess risk. Liver cancer was reported in the four studies listed in table 2 and graphically presented in figure 2. When the results 3J were tested for statistical significance ft with a one-sided Poisson test, it was found C/> XXXXXX that three of the four studies showed sig 3 nificant excesses (4, 5, 8, 11). The powers were all very high, despite the fact that -J S *(O - -S H -a E_ II s3 | E O HI ID O g-.E >> A A we *-4 0 A 0 A AA A the expected numbers of deaths were in s all instances small. The small expected numbers would normally have led to low I0 statistical powers, but the high estimated E B S' e JS> E 1I r- m ' to ft~--< in 0m1 c t' tD *W--OC<"CV.NNiOOh NN^ relative risks had a large effect on the cal e 3 culations. The powers of the studies are shown in two columns in table 2 and on 8 two lines in figure 2 because the calcula tions assumed different relative risks for tt--' Ct'O analyses considering all person-years at risk and for analyses considering only .8 X> tfl (C t* P* t" t>* 0r-0 r- Irf*t tto- *33 SKa person-years after a minimum latency (relative risks of 5 and 10, respectively, were assumed for liver cancer). ~ The results for brain cancer (table 3) Sc si were more variable, in that three of five 04 _wv*r w-^w v 1Cg .E & studies had statistically significant find a e ings (6-7, 11), and that the powers ill'll -s=Sf . "3 ** ^W :C " & tto c> i4.> ua> aa> w 5 A " -HwJi o^W' *4*> **3` .! oe jHj jKj * S & ^ 33 4> jk -- tS e c=o> * 5o C e*> \9 CO Q 1 <i- Sw OK Si OJ ranged from 12 per cent to a maximum of approximately 80 per cent. In figure 3, the range of powers is presented graphically. It can be seen that, in general, the studies with high power had positive findings and * those with low power had negative find ings. Of the eight studies of lung cancer, only POWER CONSIDERATIONS IN VINYL CHLORIDE STUDIES 729 Table 2 Liver cancer: results and powers of epidemiologic studies of vinyl chloride Author deaths Expected deaths Mortality ratio Excess p < 0.05? Power (%) lfjilt = 5.0 URR - 10.0 Fox and Collier (1, 2) Waxweiler (6) (10+ latency) Waxweiler et al. (4) -_ (15+ latency, 6+ exposure) Byren et al. (11) (10+ latency) Reini et al. (8) 4 10 8 -7 .4 4 12 1.6 2.44 2.3 4.27 0.5 15.09 0.4 16.08 1.0 4.13 0.7 5.89 0.9 15.23 + + + 93 98 92 . 86 80 98 76 EXPECTED DEATHS Figure 2. Approximate powers ofvinyl chloride studies reporting results for livereancer: 10.0 " assumed relative risk after a minimum latency, 6.0 *= assumed relative risk overall. REINL = Rein] et al: (8); BYREN = Byren et al. (11); FOX = Fox and Collier (1, 2); WAX = Waxweiler et al. (4) and Waxweiler (6). two yielded statistically significant re sults (4,5,9) (table 4 and figure 4). Again, there was a wide range of powers, from 14 per cent to almost 100 per cent, but here the studies with very high power were gative, an unlikely occurrence if vinyl chloride is indeed a human lung carcin ogen. Discussion The data regarding liver cancer in vinyl chloride exposed workers suggested 730 BEAUMONT AND BRESLOW strongly that the chemical is carcinogenic for the liver. Three of the four studies re ported statistically significant excess Cf-M CosoJ o risks (4, 5, 8, 11), and the nonsignificant study (1, 2) reported a small excess (4 ob served, 1.6 expected). The statistical pow w 08 ers of the studies were high (all above 75 per cent) due to the high observed relative DC risk. Even without knowledge of the animal studies, and without knowledge Ift that angiosarcoma of the liver is ex tremely rare, one might conclude from these data that vinyl chloride is car T.oe cinogenic for the liver. The powers of the studies that reported & '3 BO WSK o1V4/. I ++ ++ I brain cancer ranged from 12 per cent to 80 per cent. The expected trend for a causa tive association was seen, in that the ~oa *Q Io 4&; ^o &"'*32 1lib0 Ooi t%co NN JBU wVc. V E a o rt t-t ci o> a o O f<N 0) fCAl studies with high power were statistically significant and the studies with low power were, with one exception, nonsig nificant. The most reasonable interpreta tion, therefore, is that the data are consis tent with an etiologic hypothesis for vinyl chloride and cancer of the brain. Only two of eight studies reporting lung cancer results showed significant excesses (4, 5, 9). While some of the negative re c c &s *8 B IS P4 Qi OO h cs ft S' e sEo C4 (4 M sults could be explained by low statistical power, two studies that were negative had very highj>ower (1, 2, 6, 7). The lack of trend is evidence that vinyl chloride may not be a lung carcinogen, or that the ac tual relative risks were much lower than the assumed values of 1.5 and 2.0, in which case the powers were also much lower. It should be noted that one of the U <N _ = E I 6r~ WS S<!-&CO *-s*u>>-..+2Xc(5w>ISa^+*aK ^dS" Z54 *S *| + *5 + S5 rX- *e *"g : s2 * w w `5o- <0* n3E> .<s2i two negative studies with high power (by Fox and Collier (1, 2)) was also negative for liver cancer. Since it is generally ac cepted that vinyl chloride is a liver car cinogen, it may be that the exposures in that study were minimal. The other nega tive study with high power (Equitable) did not report liver cancer results, but did find a significant excess of brain cancer, which may constitute evidence of sub stantive exposure. Examination of "positive" and "nega- POWER CONSIDERATIONS IN VINYL CHLORIDE STUDIES 731 EXPECTED DEATHS Figure 3. Approximate powers of vinyl chloride studies reporting results for brain cancer; 3.0 * assumed relative risk after a minimum latency, 2.0 = assumed relative risk overall. BYREN = Byren et al, (11); REINL = Reinl et al. (8); FOX = Fox and Collier (1, 2); WAX * Waxweiler et al. (4) and Waxweiler (5); EQUIT = Equitable Environmental Health (6, 7). tive" hypothesis testing results in the light of statistical power is only one of several approaches to reviewing the evi dence of carcinogenicity for a substance. For example, one can compute a combined relative risk for each cancer site (by com bining the observed and expected num bers from the studies), and then, with the help of homogeneity testing, determine which studies are compatible with the combined risk estimate (see the Appendix for the homogeneity test). Liver cancer in the present review has a combined overall relative risk of 5.17 (p < 0.00001), and also a significant (p = 0.002) result in homogeneity testing, largely due to the Reinl (8) relative risk of 15.23. The com bined relative risk for overall brain cancer is 1.74 (p < 0.01); here the results are more homogeneous (p = 0.10), al though Byren (11) is somewhat of an out lier with a relative risk of 6.12. Finally, the combined relative risk for overall lung cancer is 1.06 (nonsignificant), with some evidence (homogeneity p = 0.06) that the results from Waxweiler (4,5) and Buffler (9) are out of step with the others (relative risks of 1.49 and 2.89, respec tively). Close examination of the individual studies can also be helpful in searching for reasons for differing results. For example, the expected number of deaths from liver cancer in the Byren study (11) is unusually large relative to the expected numbers for brain and lung. Detailed examination of this anomaly might be informative. There are other consid- R&S 114723 732 BEAUMONT AND BRESLOW I Tabus 4 Lung cancer: results and powers of epidemiologic studies of oinyl chloride Study deaths Expected deaths Mortality ratio Excess p < 0.057 Power (%) If HR - 1.5 IfRR-2.0 Fox and Collier (1, 2) (15+ latency) Waxweiler (5) (10+ latency) 46 61.2 0.90 - 94 28 26.0 1.08 100 42 28.2 1.49 + 77 39 24.9 1.56 + 99 Waxweiler et at. (4) (15+ latency, 6+ exposure) i 1 5.7 1.94 + 63 Equitable Environmental Health (6, 7) (1+ exposure) 44.3 1.02 -- 91 (15+ latency, 1+ exposure) 41 39.2 1.05 -- 100 BufBer et al. (9) 5 1.7 2.89 + 14 (5+ latency, 2.29+ exposure) 4 1.0 3.81 + 21 Byren et al. (11) 3 1.8 1.68 -- 15 Duck et al. (12) 16 15.5 1.03 - 55 Reinl et al. (8) 22 24.6 0.95 - 75 OU et al. (10) 4 5.2 0.77 27 POWER CONSIDERATIONS IN VINYL CHLORIDE STUDIES 733 EXPECTED DEATHS Figure 4. Approximate powers of vinyl chloride studies reporting results for lung cancer: 2.0 = assumed relative risk after a minimum latency, 1.5 = assumed relative risk overall. BUFF = Buffler et al. (9); BYREN = Byren et al. (11); OTT = Ott et al. (10); DUCK = Duck et al. (12); REINL = Reinl et al. (8); WAX = Waxweiler et al. (4, 5); EQUIT = Equitable Environmental Health (6, 7); FOX = Fox and Collier (1, 2). erations, such as dose-response, degree of exposure, concomitant exposures, and confidence limits for risk ratios, that also need to be kept in mind. While consid eration of statistical power cannot pro vide a complete answer, it is one more useful way of looking at epidemiologic evidence. References 1. Fox AJ, Collier PF. Low mortality rates in in dustrial cohort studies due to selection for work and survival in the industry. Br J Prev Soc Med 1976;30:225-30. 2. Fox AJ, Collier PF. Mortality experience of workers exposed to vinyl chloride monomer in the manufacture of polyvinyl chloride in Great Britain. Br J Ind Med 1977;34:1-10. 3." Nicholson WJ, Hammond EC, Seidman H, et al. Mortality experience of a cohort of vinyl chloride-polyvinyl chloride workers. Ann NY Acad Sci 1975;246:225-30. 4. Waxweiler RJ, Stringer W, Wagoner JK, et al. Neoplastic risk among workers exposed to vinyl chloride. Ann NY Acad Sci 1976;271:40-8. 5. Waxweiler RJ. An epidemiologic investigation of lung cancer in a multixenobiotic environ ment. Ph.D. Dissertation. Chapel Hill, NC: University of North Carolina, 1978. 6. Tabershaw IR, Gaffey WR. Mortality study of workers in the manufacture of vinyl chloride and its polymers. J Occup Med 1974;16:509-18. 7. Equitable Environmental Health, Inc. Epidemi ologic study of vinyl chloride workers. Final re port submitted to Manufacturing Chemists As sociation, 1978. 8. Reinl W, Weber H, Greiser E. Diseases caused by vinyl chloride. Special report of the State In dustrial Physician, Dusseldorf, West Germany, 1978. 9. Bufiler PA, Wood S, Clayton E, et al. Mortality- experience of workers in a vinyl chloride monomer production plant. J Occup Med 1979;21:195-203. 10. Ott MG, Langner RR, Holder BB. Vinyl chloride exposure in a controlled industrial environ ment. Arch Environ Health 1975;30:333-9. 11. Byren D, Engholm G, Englund A, et al. Mortal ity and cancer morbidity in a group of Swedish R&S 114724 734 BEAUMONT AND BRESLOW VCM and PVC production workers. Environ Health Perspect 1976;17:167-70. 12. Duck BW, Carter JT, Coobes EJ. Mortality study of workers in a polyvinylchloride produc tion plant. Lancet 1975;2:1197-9. 13. Preiman JA, Chalmers TC, Smith H, et al. The importance of beta, the Type H error, and sam ple size in the design and interpretation of the randomized control trial. N Engl J Med 1978;299:690-94. 14. Cutler SJ, Schneiderman MA, Greenhouse SW. Some statistical considerations in the study of cancer in industry. Am J Public Health 1954;44:1159-66. 15. Molina EC. Poisson's exponential binomial limit. Huntington, NY: Robert E. Krieger Pub lishing Company, 1973. 16. Armitage P. Statistical methods in medical re search. Oxford: Blackwell Scientific Publica tions, 1971. R&S 114725 Appendix The approximate power formula assumes that the observed number of deaths from the cause of interest follows a Poisson distribution with mean RE, where E is the ex pected number based on standard population rates and R is the ratio of rates (relative risk) for the study as opposed to the standard population. It follows that for large values of RE, the square root of the observed deaths is approximately normally distributed with mean \/RE and variance V* (16). Now, if a random quantity has a normal distri bution with mean /z and variance a2, its power 1 -- /3 to reject the null hypothesis /t = /t0 at the a level of significance is given by rr _ n `-M-- A^O ------ a Substitution of /z = \/RE, /t0 = VjE, and a -- Vi into this expression yields the re quired result. The homogeneity test for the combined relative risk uses the chi-square distribu tion with n - 1 degrees of freedom. The equation is y (O, - `6Etf Xoi-i) i-1 SEj where n = the number of studies, 0( = the observed deaths from the ith study, Et = the expected deaths from the tth study, and = the combined relative risk estimate. v, VC The Role of Human Genetic Monitoring in the Workplace Betty ). Dabney, Ph.D. R&S 114726 The history and current state of some newer short-term tests for occupational genetic monitoring are reviewed. These are: cytogenetics, sister chromatid exchange, body fluid analysis, tests utilizing sperm, and detection of somatic cell variants. Occupational studies on benzene, vinyl chloride monomer, and epichlorohydrin are critically discussed from the standpoints of design and interpreta tion. It is concluded that these tests are not appropriate for risk assessment at the present time. Their clinical rele vance, if any, is unknown. Proper validation and standard ization have not been done^and design problems have often clouded the results of previous studies. There is a critical need for further research in the area of occupa tional genetic monitoring. Future applications should inMude integration with prospective morbidity and mortality ptdies, standardization of design and statistical methods, and development of new tests with genetically relevant endpoints. The question of whether or not occupational expo sure to mutagens may pose a genetic risk to future gener ations is one.of both public and scientific concern. Reli able short-term tests are desirable to define potential genetic problems and to assist in preventing their occur rence in the work force. Several recent publications have reviewed the subject of estimating induced genetic ef fects in man.1'5 All of these reviews convey the same message: that no chemical or physical agents are known to cause mutations in man. Even the extensive genetic studies on survivors of the atomic bomb attacks in Japan and their offspring have failed to-demonstrate any in duced mutations in that population.* However, one can not assume that exposure to mutagens in the workplace would have a similar lack of effects. In the light of growing interest in occupational genetic monitoring, it is timely to review the history and current state of this field. These comments are based only upon evaluation of the scientific merit of the proposed tests, and do not attempt to resolve ethical, moral, or legal questions arising from human genetiiTmohitbrmgj For the purpose of this discussion, several terms need )be defined. Genetic "screening" refers to a one-time At the time this manuscript was prepared, the author was a Consultant in Oc cupational Health, West Columbia, lea. Present address*. IBM Corporation, P.O. Bo* 1900. Boulder. CO 80302. 626 test for known heritable diseases, such as sickle cell anemia. Genetic "monitoring" will describe a program of repeated short-term tests on an individual or group to determine possible genetic damage in somatic or germ cells, damage which is not necessarily heritable. A "muta tion" is defined as a heritable change in a cell or individ ual, in distinction to "genetic damage," which shall refer to any structural or functional disruption of the genetic apparatus. Genetic damage is not necessarily equivalent to a mutation, since it may either be correctly repaired, or be lethal to a cell, instead of giving rise to a heritable change. This discussion is concerned with genetic monitoring. In this context, genetic monitoring is designed to detect effects brought on by occupational exposures. It has been proposed that genetic monitoring techniques should be used as a biological dosimeter for employees working with, or exposed to, suspected mutagens. The proposals also suggest that monitoring techniques could be used to identify individuals in that group who may be unusually susceptible to those agents. As an "early warning system/' genetic monitoring would be used to identify groups or in dividuals at risk for possible clinical effects before other clinical signs are apparent It is extremely important to understand that all of the genetic monitoring tests to be discussed here are still in the developmental stage and are not vet useful as predic tive tools. Consequently, the tests should not be consid ered part of a routine industrial medical surveillance pro gram, nor should they be used for risk assessment at this time. These concerns are substantiated for the following reasons: (11 the clinical implications, if any, of abnormal results in these tests are not vet clear: and (2) the intrinsic variability of the tests and the multiplicity of external fac tors affecting their results, and most importantly, previous experiences from occupational studies utilizing cyto genetic monitoring bear out these conclusions. These points will be expanded liter in this paper. Description of Tests Cytogenetics, the study of numerical and structural chromosome aberrations, is distinguished here from clas sical clinical cytogenetics, which detects heritable chro mosome abnormalities present in every somatic cell of an individual. In occupational cytogenetics, a positive test refers to detection of a significant increase in the frequen- cy of damaged chromosomes in samples from a group of individuals over a detectable frequency in a properly selected control group. Generally, cultured lymphocytes' are usedTFrom five- to tenfold variations in the normal or control group lymphocyte chromosome breakage fre quencies, with unexplained seasonal peaks, have been re ported in different studies.7 * * In most of the occupation al studies, less than a twofold elevation in chromosome breakage frequencies usually is found. One of the great difficulties in interpreting results of cytogenetics studies arises from the wide variation in chromosome aberration frequencies found within replicate studies of an in dividual and among individuals within a study group. Some of this variation can be attributed to the use of different culture media,1811 to recent viral infection in th<T subjects,'1 to the staining procedure used to visualize the chromosomes.11 toThe use o? whole blood vs. washed lymphocytes.14 or to smoking or recent alcohol consump tion.11 Intra- and inter-scorer and inter-laboratory varia tions in scoring criteria may also come into play, particu larly when dealing with the problem of whether or not to count achromatic lesions (gaps). The scoring criteria recommended by the World Health Organization,11 which do not include gaps, are used by many labora tories, but there is an ongoing dialogue among scientists concerning the most relevant criteria for scoring chromo some aberrations. The major strength of cytogenetics for occupational studies is that there is a far larger data base for this test than for any of the others discussed here, in addition to the occupational studies, there is considerable literature on in vitro and on animal studies. Major limitations for cytogenetics are the high degree of skill required ofTthe part of the scorer and the limited capacity tor handling additional .cases in existing laboratories, suchThat any industry-wide general monitoring program would be im possible at the present time. Sister chromatid exchange (SCE) is a related test in which dividing cells incorporate 5-bromodeoxyuridine into DNA during two consecutive rounds of replication, causing the two chromatids in each chromosome to take up stain differentially. This procedure allows easy visual detection of crossover events and exchanges between the sister chromatids. The number of these SCEs per cell in creases in whole animal and in vitro cell culture studies after sufficient exposure to mutagens. Details of this pro cedure as it relates to testing chemicals for mutagenic potential have recently been the subject of a number of review articles, such as the one by Latt17 The SCE techniques have also had limited use in genetic monitoring programs. Usually lymphocyte cul tures obtained from the study groups are processed to reveal chromatid exchanges in the same manner used for the experimental cell culture systems. An excellent exam ple of using the technique is a study conducted to evaluate chronic effects in patients treated with arsenic.1* Elevated frequencies of SCEs were reported to persist for many years in the arsenic-treated patients, but it is not known if this is generally true. At the present time the use of SCE for occupational monitoring should be regarded as highly exp rimeritall ~~Some occupational studies describing the'useof 5CE will be discussed further in this paper. Body fluid analysis is a microbial cell mutagenicity assay used to detect mutagenic activity in urine, blood or feces of exposed individuals. This test has been used on individuals involved in acute exposures, with somewhat varying results.1' Samples from normal unexposed indi viduals are sometimes positive, because of cigarette smoke metabolites10 or possibly of metabolites of certain foods.11 Other tests utilizing genetic endpoints in the host's cells may be easier to interpret from the standpoint of risk assessment However, thoroughly validated body fluid analysis techniques might some day prove to be use ful for determining whether individuals are at risk from occupational exposures to certain types of chemicals. Further research and comparison with other tests need to be completed before any judgment can be made on the general utility of body fluid analysis. Other tests are being developed which may ultimately be useful for occupational studies. Several of these newer systems utilize sperm, thus providing an advantage over the tests using somatic cells when searching for potential ly heritable effects. Animal models suggest that mutageninduced heritable changes in sperm shape may prove use ful in the future.1111 The animal tests include studies on the loss of sperm enzymes as detected by cytological staining procedures.1411 The human YFF sperm cell test has been used by Kapp et al11 in an occupational study. In this test a positive ef fect is indicated by a significant elevation in the frequency of sperm containing two or more fluorescent bodies. These bodies are presumed to be revealing the presence of multiple Y-chromosomes produced by nondisjunctions occurring during meiosis. The YFF test needs further development, particularly to minimize false positives due to fluorescence of non-Y chromosomes. None of these tests on sperm have endpoints proven to be of genetic significance (i.e., associated with heritable disease states). Until validation is accomplished, these tests should be regarded purely as experimental systems- Other tests under development examine the frequency, of somatic cells lacking certain enzymes, the best of these taking advantage of the fact that cultured lympho cytes deficient in the enzyme hypoxanthine guanine phosphoribosyl transferase are not killed by 6-thioguanine.17 This test is particularly promising because it studies a recognized genetic endpoint in the individual's cells, and is relatively simple and rapid to perform. Its use, though still quite limited, has been compared to both chromosome aberrations and SCE. In untreated psoriasis patients, there was a significant elevation in the frequen cy of 6-thioguanine-resistant lymphocytes, while SCE and chromosome aberration frequencies were normal.1*11 " Other potentially promising single-cell variant assays are also being intensely studied.14 Further developments in genetic monitoring may include automation and flow cytometry techniques to facilitate chromosome and mu tant cell analyses, but at the present time these methods are not practical for general use.11 History of Genetic Monitoring in the Workplace The use of short-term tests for genetic damage in occo- pational studies is a fairly recent practice, with the majority of studies having been published by European laboratories over the past 15 years. Most of the occupational Journal of Occupational Medicine/Vol. 23, No. 9/September 1981 627 ,, t : 20 CO ,* studies have looked for structural chromosome damage (e.g., breaks, dicentrics, exchanges) or for aneuploidy (ab normal number of chromosomes). A description of an in- jstrial monitoring program is given by Kilian and PicFiano." The subject of chromosome studies on industrial populations has been reviewed recently by Purchase.31 A relationship between relatively high occupational ex posures and increased frequencies of chromosome aber rations has been implicated for several chemicals in previ ously published studies. Those which are most represen tative have been chosen for further elaboration in this paper. Benzene has the largest literature of occupational ex posure associated with chromosome aberrations. In their early studies Forni et al selected subjects with histories of high occupational exposures and clinical hemopathies. A representative study by Forni et al3* included subjects with a variety of work and clinical histories. Approximate ly 100 cells from each culture were scored according to the "Cu-Cs" system of Buckton et al.3S "Cu cells" consist of cells with chromosome breaks, dicentrics, and rings, and "Cs cells" are those containing chromosome ex changes and inversions. The 25 patients with past benzene hemopathy and with 1 to 18 years since occupa tional exposure to benzene averaged 1.89% Cu cells and 1.22% Cs cells. The matched controls averaged 0.49% and 0.04%, respectively. Statistical analysis (x1) revealed a significant difference between workers and controls, but no trends were seen with respect to severity of hemopathy or to age. (However, other studies, such as hose of Tough et al,3* have identified an age-related lend among exposed benzene workers.) Two of three objects with documented acute exposures to high con centrations of benzene exhibited elevated frequencies of chromosome aberrations three to four years after expo sure, while a third subject was within the control range. This study has several merits, including the practice of repeated samplings. It also attempts to document expo sures to benzene and other chemicals, and uses age- and sex-matched controls. While the studies of Forni et al are commendable in the sense that they followed the pre cepts of epidemiology in their design and established the association between chromosome aberrations and ben zene-induced hemopathies, they cannot be used to ad dress the question of the utility of genetic monitoring as a dosimeter or as a predictor of clinical risk. The use of sub jects only with previously established clinical findings precluded the possibility of establishing any temporal or causal relationship between chromosome aberrations and disease. Unfortunately, in the case of benzene the persis tence of and increase of chromosome aberrations cor related with time after of exposure make it impossible to establish a relationship between the duration of exposure and induction of chromosome aberrations. The most recent occupational cytogenetics study on benzene is that of Picciano.37 This study included two groups of workers totaling 52 individuals and 44 controls selected during preemployment examinations and atched as closely as possible for age, sex, and date of sting. Exposures to benzene for the worker groups aver aged about 2 ppm (time-weighted average) over the pre ceding four-year period. No information on possible peak exposures was given. From an evaluation of 200 cells per person, no significant difference was found between worker and control groups with respect to abnormal cells (1.6% vs. 1.4%) or to chromatid breaks (1.0% vs. 1.1 %i but frequencies of chromosome breaks (0.67% vs. 0.35%) and of "marker" chromosomes comprised of rings, dicen trics, and exchanges (0.19% vs. 0.06%) were shown by x3 analysis to be significantly elevated in the worker groups. In contrast to the findings of other studies,34 no trend was seen with age, but there was an unavoidably large age dif ference between the worker and the control groups (39 vs. 27 years average). The Picciano study illustrates several recurrent prob lems in structuring and interpreting occupational cyto genetic studies. Firstly, it is not clear if external variables which might have influenced the results (smoking, recent viral infection or drug use, nonoccupational exposure to chemicals and radiation, occupational exposure to chem icals other than benzene, presence of interfering disease states) were adequately controlled in structuring both the worker and the control groups. Secondly, while the differ ence in chromosome aberration frequencies for some cat egories was statistically significant, the overall frequen cies were quite low and would be considered in the nor mal range in most laboratories. Thus, the biological or clinical significance of the elevated frequencies is debat able. Repeat studies were not reported, making it unclear if the small differences between groups were sustained with time. Thirdly, the failure to include information on peak exposures makes it impossible to determine if the small elevations could be explained by chronic exposures to low concentrations, or by high acute exposures. Finally, there is still some question of the most meaningful para meter for expressing the chromosome aberration frequen cies found in the study. The current trend seems to be toward using the percentage of abnormal cells, rather than isolating individual categories of aberrations for analysis, but this issue is far from being resolved, ,_Elevated frequencies of chromosome aberrations have also been reported to be associated with occupational ex posure to vinyl chloride monomer (VCM). Since the origi nal study by Ducatman et al3* in 1975, several labora tories have followed chromosomal changes in occupa tional groups exposed to VCM. Illustrative of this are two ongoing studies of recent reports. In the first of these studies, one by Kucerova et al,3* nine workers exposed to relatively high concentrations of VCM (20 to 150 ppm) were evaluated over a two-year period for chromosome aberrations and also for SCE at the time of the final sampling. A control group of eight age- and sex-matched individuals was sampled concur rently with the final worker study. Smoking, drinking, di agnostic radiation, and drug usage patterns were docu mented for both the controls and the third sampling of the workers. Statistical analysis (t-test) was done only on the third sampling, where control values were available. The results showed significant elevations in the percent of cells containing chromosome aberrations in the worker group compared with the controls (5.2% vs. 1.8%) and also for SCEs (13.8/cell vs. 9.4/cell). .There, was no ap parent effect of smoking or of lifestyle. Evident in the study by Kucerova et al are several prob- 628 Human Genetic Monitoring in the Workplace/Dabney R&S 114728 i 114729 lems in interpretation. Firstly, there was a wider range in In this study are presented several additional problems values within individuals sampled at different times (11 % with occupational genetic monitoring. The first problem maximum) than there was between controls sampled at is in choosing a proper control group, there being a signifi the same time (3%). It is unclear if this temporal variation cant difference between control groups. Since the signifi was due to laboratory artifacts or to true biological differ cance of findings in a worker group is dependent upon ences. Thus it is difficult to determine the significance of findings in an appropriate control group, the definition of the findings in the first two worker samplings, where no an appropriate control group needs to be clear-cut before concurrent controls were used. This illustrates that such the study is undertaken. In view of the lack of concurrent group studies are best done as cross-sectional studies, controls for the first two samplings, and the temporal vari utilizing concurrent matched controls, rather than longi ability problem discussed previously, one cannot con tudinal ones, utilizing previous values from an individual clude with a high degree of certainty that the increasing as his own control. Another problem with this study is the frequencies of chromosome aberrations were due to oc small group size of only nine persons. It is important to cupational exposure to ECHH. Another problem illustrat consult with a statistician at the design stage in order to ed in this study is the relatively small distinction between ascertain if the group and sample sizes are sufficiently groups: This type of difficulty tends to cloud any biologi large to permit desired limits of sensitivity to be detected. cal relevance of the results. The second ongoing occupational study on VCM is A second occupational study where elevated frequen that of Hansteen et al,40 in which 39 VCM workers "and 16 cies of chromosome aberrations were attributed to occu community controls were originally sampled in 1974. All pational exposure to ECHH is that of Picciano.4' In this but two of the workers and 32 matched controls were study 93 epoxy resin workers were compared with 75 pre- studied three years later. Initially chromosome breakage employment individuals. The durations and actual or rela 30 frequencies were elevated in the worker group compared tive exposures levels of the workers were not reported. (/) with the controls (3.4% vs. 1.8%). The difference was sig nificant by the Wilcoxon two-sample ranking test. Values Chromosome aberration frequencies were shown to be significantly greater in the worker group compared to the from heavily exposed workers were slightly higher than preemployment group by x3 analysis (4.25% vs. 2.38%). those of the other workers (3.9% vs. 3.15%). Documented This study has serious flaws which are not readily ap exposures to VCM were reduced considerably during the parent Considerations of epidemiology similar to those interval between the two samplings, with final exposures which were discussed in relation to the study by Picciano averaging 1 ppm. Samples from the follow-up study should apply to this study design. Secondly, and more im showed no significant difference between workers and portantly, unreported subsequent sampling of the abnor controls for chromosome breakage frequencies (1.25% to mally high cases showed significant reductions in fre 1.91% for workers vs. 2.33% for matched office worker quencies of chromosome aberrations.4i Several types of controls). Frequencies of SCE for 16 of the workers and culture media had been used over a period of months. their matched controls at the time of the final sampling Most of the"fifgh frequencies in the initial sampling coufd were similar, averaging 7.6 vs. 7.5 SCE/cell. Chromosome be attributed to one type of culture medium in bothJfc breakage frequencies were not significantly different be worker and control groups.4' failure to adequately con-^^ tween the two control groups composed of different in p'ol the many variables in this study rendered the results dividuals (1.79% in 1974 vs. 2.33% in 1977). uninterpretable. This study by Hansteen et al is valuable in that it il These studies on workers exposed to ECHH illustrate lustrates the importance of continued exposure monitor the importance of carefully controlling laboratory vari ing for the duration of any such study. It also suggests ables and applying the precepts of epidemiology in any that chromosome aberrations associated with exposure to "such-group study. As a result the question of whether or VCM may be less persistent than those found with ben not occupational exposure to ECHH can induce chromo zene exposure. some breaks is still unresolved. A third chemical implicated in elevating chromosome Other studies of chromosome aberrations in lympho breakage frequencies as a result of occupational expo cytes have been reported for groups of workers exposed sure is epichlorohydrin (ECHH). Two relevant studies to pesticides, arsenic, styrene, mercury, lead, and cad which follow are presented in detail in this paper. The mium. These studies generally suffer from defects in their first, an ongoing study by Sram et al,4' follows 28 workers, experimental design and interpretation similar to those 23 of whom were first reported in 1977. This study is of in discussed earlier. In these cases the information available terest because it utilized two control groups, one age-, is too sparse to allow for any general conclusions on the sex-, and lifestyle-matched from within the same or near validity of the current observations. by factories, and the second from the general population. Only one occupational study, that by Kapp et al,M has Over a period of four years the frequencies of abnormal been published, in which germ cell effects were examined cells among the worker group increased with duration of in a monitoring system. In this study occupational expo employment (1.37%, 1.91 %, 2.69%, and 3.02% with 0,1, sure to 1,2-dibromo-3-ch!oropropane was associated with 2, and 4 years of employment, respectively). Matched fac increased frequencies of nondisjunction in sperm, as tory worker controls for the final sampling averaged determined by the YFF test Eighteen persons comprised 2.06% abnormal cells, compared with 1.33% from a the worker group, and 15, the control group. However, no general population sample also taken at the time of the details were given concerning the latter group. Frequen final sampling. There were significant differences be cies of sperm containing two fluorescent bodies were tween the final worker samples and the matched controls, found to be 3.8% for the workers and 1.2% for the co^^ and between control groups as well. trots. These values were found to be significantly differa^^ Journal of Occupational Medicine/Vol. 23, No. 9/September 1981 629 -* ^ I' R&S 114730 by x1 analysis. This group of workers needs to be followed in a reproductive epidemiology study, and further studies need to be conducted with the YFF test, before it can be etermined if abnormal results in th s test are associated 'ith an increased risk for siring aneuploid offspring. It is evident from these occupational studies that there exist many problems confounding the interpretation of their results. As has been discussed in this paper, different laboratories have used different statistical analyses. It is concluded, therefore, that appropriate and consistent statistical methods need to be developed for these types of studies. By necessity, many studies are conducted over a period of months or years. Civen the great influence of laboratory variables on the results, and the impossibility of controlling them over the long run, one wonders if the small group differences reported in these studies were real. THe~prd5Iem of variability in chromosome aberration frequencies is so great that often two laboratories cannot agree on a definition of the normal range of values. Compounded with these problems is the unan swered question of whether or not the small differences seen, and even the endpoints themselves in chromosome aberrations or SCE, are genetically or clinically relevant. Of the studies performed on clinically normal subjects, there has often been no apparent purpose in the studies other than for determination of the chromosome aberra tion or SCE frequencies. Unless these results are related to other clinical findings and the individual cases are fol lowed in prospective epidemiology studies (morbidity, mortality, and reproductive effects), the clinical signifi cance of the short-term tests may never be known. ^ Indeed, with the interpretive and technical problems of Bytogenetics discussed here, one wonders how useful this test can ever be for risk assessment Fifteen years of its development Th occupational monitoring have so far not contributed much to understanding genetic risk in the workplace. Perhaps the researcher's time would be better spent developing other tests which are not subject to these problems. Common Considerations There are some important considerations which apply to all short-term tests for genetic monitoring. A major dif ficulty lies in the lack of experience in designing and per forming the studies. A second concern is that the majority of the tests are using readily accessible somatic cells. It is often impractical or impossible to determine if com parable genetic damage or mutations occur in other body tissues, because testing of those tissues cannot be ade quately performed with existing techniques. A third and practical concern is cost of testing and the limited capaci ty of existing laboratories. Therefore, tests should be per formed on exposed occupational groups only where there are prior positive findings in some animal system such as those described by Brewen.44 Finally, all of these tests are relatively new, indicating that not enough time has elapsed since their inception to allow for determination of what long-term effects, if any, are associated with posi tive findings. Thus, all of these short-term tests must be nsidered research tools. They are not established cliniI tests. When used in an occupational group study, they should be part of a well-structured research program, which includes ongoing epidemiological studies. 630 Guidelines for Genetic Monitoring As research tools, these tests should be part of a well- designed protocol for occupational study. Assistance of epidemiologists and statisticians, as well as of laboratory scientists, industrial hygienists and physicians, should be enlisted at the design stage. To facilitate comparisons, different facilities conducting these studies should use a consistent experimental design. The core of this design should include a basic questionnaire for recording family and personal work, health, lifestyle and reproductive histories and for the updating of forms to follow up changes in health or work status as the study progresses. Population size should be sufficient to allow for detec tion of predetermined limits of sensitivity. Before any workers are sampled, a stable unexposed comparison group should be developed initially, with repeated samplings done to establish the expected range of normal values, and to directly examine possible confounding var iables. Repeated samplings should be taken on the work er group at specified intervals, with comparison control group samplings performed at the same times. Workers and controls with known significant exposure to potential mutagens outside of the exposures expected to be en countered in the normal course of the study should be excluded. Reliable exposure estimates prior to and during the study are extremely important in establishing cause and effect Longitudinal studies are not recommended until factors influencing the variability of these tests can be identified and controlled in the design of the study. As a rule, recognizable factors with unknown but potentially confounding. influences should be controlled in the design of the study. Samples should be coded and scored blind to minimize scoring bias. Clearly positive findings in an occupational group, con firmed by repeated samplings at more than one time, should be an indication for further clinical studies which might include formal reproductive epidemiology and pro spective morbidity and mortality studies. Such popula tions should be carefully monitored for any clinical ef fects. Further actions on an individual basis should lie in determining more precisely any potentially mutagenic ex posures either on or off the job, and to reduce such expo sures where appropriate. Subsequent genetic monitoring should then be conducted. Individuals with persistently abnormal findings should be referred to a medical specialist in genetics for further evaluation. Summary and Conclusions There are still many unknown factors Jn human genetic monitoring. Because few adequate studies have been per formed on workers in well-defined industrial settings, and because these have not always been performed in con junction with other clinical studies, the predictive value of short-term genetic tests has not been established, and these tests should be regarded as research tools still in the development stage. More occupational group studies with follow-up clinical evaluations and proper epidemio logical design need to be performed before the clinical value of these tests can be established. It is imperative, therefore, that future studies of this type be carefully planned to produce meaningful results. The greatest need is in developing and validating rapid tests with clearly genetic endpoints indicating unequivocal clinical rele- Human Genetic Monitoring in ihe Workplace/Dabney vance. Until that goal is accomplished, genetic monitor ing should not be used to set standards for safe exposures, or to establish conditions for the prevention of occupa tional disease. The author acknowledges and appreciates the support of the American Indus* trial Health Council in the preparation of this paper. References 1. Bloom AD: Induced chromosomal aberrations in man. Adv Hum Cener 3:99-172,1972. . 2. Symposium: Long-term hazards to man from man-made chemi- cals. Proc R $oc Lond 205:1-159,1979. 3. McElheny VK and Abrahamson S (Eds.): Banbury Report #1: Assessing Chemical Mutagens: The Risk to Humans. Cold Spring Har- bor Laboratory, 1979, 367 pp. 4. Evans HJ and Lloyd DC (Eds.): Mutagen-Induced Chromosome Damage in Man. Edinburgh: University Press, 1978. 5. Berg K (Ed.): Genetic Damage in Man Caused by Environmental Agents. New York: Academic Press, 1979. 6. Neel JV and Schull WJ: The effect of exposure to the atomic bomb on pregnancy termination in Hiroshima and Nagasaki. National Academy of Scienees-National Research Council, Publ, #461,1956. 7. Lubs HA and Samuelson J: Chromosome abnormalities in lym- phocytes from normal human subjects. Cytogenet Cell Cenet 6:402-411,1967. 8. Littlefield LC and Goh K-O: Cytogenetic studies in control men and women, I. Variations in aberrations frequencies in 29,709 meta- phases from 305 cultures obtained over a three-year period. Cytogenet Cell Cenet 12:17-34,1973. 9. Mattei MC, Ayme S, Mattel JF, et al: Distribution of spontaneous chromosome breaks in man. Cytogenet Cell Cenet 23:95-102,1979. 10. Keck M and Emerit I: The influence of culture medium compose tion on the incidence of chromosomal breakage. Hum Cenet 50:277-283,1979. 11. Sutherland GR: Fragile sites on human chromosomes: Demon- stration of their dependence on the type of tissue culture medium. Science 197:265-266, 1977. 12. Harnden DC: Viruses, chromosomes and tumors: The interac tion between viruses and chromosomes, in Chromosomes and Cancer, ). Cerman (Ed.). New York: John Wiley and Sons, 1974, pp 151-190. . 13. Sofuni T, Shimba H, Ohtaki K, and Awa AA; A cytogenetic study of Hiroshima atomic-bomb survivors, in Mutagen-Induced Chromo- some Damage in Man, H. J. Evans and D. C. Lloyd (Eds.). Edinburgh: University Press, 1978, pp 108-114. 14. Dufrain R), Littlefield LC, and Wilmer |L: The effect of washing lymphocytes after in vivo treatment with streptonigrin on the yield of chromosome and chromatid aberrations in blood cultures. Mutat Res 69:101-105,1980. 15. Funes-Cravioto E, Lambert B, Lindsten J, et al. Chromosome aberrations in workers exposed to vinyl chloride. Lancet 1:459,1976. 16. Buckton KE and Evans JH: Methods for the Analysis of Human Chromosome Aberrations. Geneva: World Health Organization, 1973. 17. Latt SA: Sister-chromatid exchanges. Cenetics 92:S83-S95,1979. 18. Burgdorf W, Kurvink K and Cervenka J: Elevated sister chro- matid exchange rate in lymphocytes of subjects treated with arsenic. Hum Cenet 36:69-72,1977. _ 19. Legator M, Truong L, and Connor TH: Analyses of body fluids in cluding alkylation of macromolecules for determination of mutagenic agents, in Chemical Mutagens: Principles and Methods for their Detection, A. Hollaender and F, J. de Serres (Eds.). New York: Plenum Press, 1978. 20. Yamasaki E and Ames BN: Concentration of mutagens from urine by adsorption with the nonpolar resin XAD-2: Cigarette smokers have mutagenic urine. Proc Natl Acad Sci USA 74:3555-3559, 1977. 21. Bruce WR, Varghese A), Furrer R, and Land PC: A mutagen in the feces of normal humans, in Origins of Human Cancer, H. H. Hiatt, /. D. Watson, and J. A. Winsten (Eds.). Cold Spring Harbor Laboratory, 1977, pp 1641-1646. 22. Wyrobek AJ and Bruce WR: The induction of sperm-shape ab- normalities in mice and humans by chemical mutagens, in Chemical Mutagens: Principles and Methods for Their Detection. A. Hollaender and F. J. de Serres (Eds.). New York: Plenum Press, 1978, pp 257-285. 23. Topham JC: Chemically-induced transmissable abnormalities in sperm-head shape. Mutat Res 70:109-114,1980. 24. Mailing HV: In vivo point-mutation systems. Cenetics 92:S67-S71,1979. 25. Mailing HV, Ray CP, and Burkhart JC: Variation in the level of r-glycerolphosphate dehydrogenase in single sperm after treatment with procarbazine. Environ Mutagen 2:264-265,1980. 26. Kapp RW Jr, Picciano DP, and Jacobson CB; Y-chromosomal inondisjunction in dibromochloropropane-exposed workmen. Mutat 1Res 64:47-51, 1979, 27. Strauss GH and Albertini RJ: Enumeration of 6-thioguanineiresistant peripheral blood lymphocytes in man as a potential test for :somatic cell mutations arising in vivo. Mutat Res 61:353-379,1979. 28. Strauss CH, Albertini RJ, Krusinski PA, and Baughman RD: (6-Thioguanine resistant peripheral blood lymphocytes in humans ifollowing psoralen, long-wave ultraviolet light (PUVA) therapy. / In 'vest Dermatol 73:211-216. 1979. 29. Brogger A, Waksvik H, and Thune P: No evidence for chromo isome damage in psoriasis patients treated with psoralen and long 'wave ultraviolet light, in Mutagen-Induced Chromosome Damage in IMan, H. J. Evans and D. C. Lloyd (Eds.). Edinburgh: University Press, 1978. pp 221-226. 30. Faed M and Mourelatos D: Sister chromatid exchanges in lym |phocytes treated with 8-methoxypsoralen and exposed to long-wave ui ltraviolet light in Mutagen-Induced Chromosome Damage in Man, HI . J. Evans and D. C. Lloyd (Eds.). Edinburgh: University Press, 1978, pp 2i 16-220. 31. Carrano AV, Gray JW, and Van Dilla MA: Flow cytogenetics: IProgress towards chromosomal aberration detection, in MutagenInduced Chromosome Damage in Man, H. J. Evans and D. C. Lloyd (Eds.), Edinburgh: University Press, 1978, pp 326-338. 32. Kilian DJ and Picciano D: Cytogenetic surveillance of industrial pi opulations,.in Chemical Mutagens: Principles and Methods for their IDetection, Al Hollaender (Ed.). New York: Plenum Press, 1976, pp 321-339. 33. Purchase IFH: Chromosomal analysis of exposed populations: A ri eview of industrial problems, in Mutagen-Induced Chromosome IDamage in Man, H. J. Evans and D. C. Lloyd (Eds.). Edinburgh: Univer s! ity Press, 1978, pp 258-267. 34. Forni A, Cappellini A, Pacifido E, and Vigliana E: Chromosome ichanges and their evolution in subjects with past exposure to benzene. -Arch Environ Health 23:385-391,1971. 35. Buckton KE, Jacobs PA, Court Brown WM, and Doll R: A study ""o f- the-chromosome damage persisting after x-ray therapy for anky Ilosing spondylitis. Lancet 2:676-682,1962. " 36. Tough IM, Smith PC, and Court Brown WM: Chromosome :studies on workers exposed to atmospheric benzene: The possible in Ifluence of age. Aur I Cancer 6:49-55, 1970. 37. Picciano D: Cytogenetic study of workers exposed to benzene. iEnviron Res 19:33-38, 1979. 36. Ducatman A, Hirschhorn K, and Selikoff IJ: Vinyl chloride expo sure and human chromosome aberrations. Mutat Res 31:163-168, 1975. 39. Kucerova M, Polivkova Z, and Batora J: Comparative evalua tion of the frequency of chromosomal aberrations and the sister chro imatid exchange numbers in peripheral lymphocytes of workers oc cupationally exposed to vinyl chloride monomer. Mutat Res 67:97-100,1979. 40. Hansteen l-L, Hillestad L, Thiis-Evensen E, and Heldaas SS: Ef fects of vinyl chloride in man. A cytogenetic follow-up study. Mutat Res 51:271-278,1978. 41. Sram RJ. Zudova Z, and Kuleshov NP: Cytogenetic analysis of peripheral lymphocytes in workers occupationally exposed to epichlorohydrin. Mutat Res 70:115-120,1980. 42. Picciano D: Cytogenetic investigation of occupational exposure to epichlorohydrin. Mutat Res 66:169-173,1979. 43. Dabney B, Barna-Lloyd G, Daniel R, et al: Cytogenetic Findings in Epoxy Resins Employees. Report submitted to the Occupational Safety and Health Administration, 1979. 44. Brewen JG: Human risk assessment Ecotoxicol Environ Safety. In press. R&S 114731 Journal of Occupational Medicine/Vol. 23, No. 9/September 1981 631 R&S 114732 {s 321 LEGISLATION DENMARK FOOD ADDITIVES Promulgation order no.65 of 20 February 1981 contains general regulations concerning the use of additives in foods. It also contains new regulations on the labelling of food products with information on the additives used. PESTICIDES Promulgation order no.410 of 17 September 1980 lays down rules concerning the classification, packaging, labelling and registration of pesticides. DANGEROUS SUBSTANCES* Notice no.408 of 17 September 1980 from the Danish Ministry of the Environment lays down rules concerning the classification, packaging, labelling, sale and storage of dangerous substances and products. The alphabetical list of chemicals that have already been classified as dangerous substances which is referred to in Annex VI of the notice, has since been replaced by a new list (in Danish) published in Notice no.147 of 16 March 1981. EUROPEAN ECONOMIC COMMUNITY ANALYSIS OF VCM IN FOOD* The official method of analysis for the control of vinyl chloride released by materials and articles into foodstuffs has now been published as Commission Directive 81/432/EEC of 29 April 1981 (Off. J. Europ. Commun. 1981, 24 (L167), 6). Member States are required to enforce this method of analysis by 1 October 1982. Further details of this document and of other items of legislation similarly marked may be obtained from the BIBRA Information Section. October 1981 Vol. 23 No. 10 Cancer Mortality of a Group of Canadian Workers Exposed to Vinyl Chloride Monomer C. Theriault, M.D., Dr.P.H., and P. Allard, M.D., MSc. R&S 114733 The present study was undertaken to find out whether their first exposure more than 15 years before their death. there was an excess of cancer mortality from causes other than angiosarcoma of the liver among a group of workers While most authors have not been able to document a significant excess of any other cancer in their studies, they heavily exposed to vinyl chloride monomer (VCM). The mortality of 451 workers exposed to VCM for more than five years was compared with that of 870 workers from the same company who had not been exposed to VCM. The have noted that such an excess may exist. In order to better document the cancer mortality asso ciated with exposure to VCM, the present study of the^^ workers of the Shawinigan vinyl chloride polymerizatior^B relative risk for digestive cancer was significantly higher than 1 (6.25, confidence interval 2.69 to 14.52) in the ex posed group. The standardized mortality ratio (SMR) for plant was undertaken. The main objectives of the study were (1) to compare mortality among vinyl chloride workers with the mortality digestive cancer was also higher'(SMR 259.26 p < 0.07) of workers from a nearby industrial complex; and (2) to than that of the general population. No other cancer was in compare the mortality of the VCM workers with that of excess. Since the exposed workers are known to have had a the general population. The Shawinigan vinyl chloride cigarette smoking experience similar to that of those who - -- plantjopened in 1943 as an independent company which jvere not exposed, it is concluded that the association be produced both VCM and PVC (polyvinyl chloride). In tween lung cancer and VCM exposure, if present, is indeed 1958, it was,merged into an industrial complex (to be used rather small. as the comparison group) of which it became the Cana dian Resins Division. In the late 1960s, the VCM produc tion ceased and only the polymerization process con tinued operation. In 1972, this division was sold and again several studies have demonstrated that vinyl chloride became an independent plant Over the years, the VCM monomer (VCM) produces angiosarcoma of the liver among people exposed at work,1-7 There is, however, much controversy about the capability of VCM to gener and PVC productions fluctuated. The average annual work force at the plant was 225 men. In 1965, VCM pro duction was estimated at 90 million pounds and PVC at ate other types of cancer in man. One author,* using a 60 million pounds.10 In 1977, PVC production was 22.3 proportional mortality ratio technique, described ex cesses of lung cancer (observed, 13; expected, 7.9; obs/exp 1.6) and brain cancer (observed, 5; expected, 1.2; obs/exp 4.2) in a group of 161 workers from two vinyl chloride plants.' Another author,' who studied the mortality of a group of 1,294 American workers exposed to VCM, re ported excesses of lung cancer (observed, 11 expected, million pounds." Since several episodes of unconscious ness among workers were reported, the level of VCM in the air at the workplace is believed to have been high dur ing this period. Ten cases of angiosarcoma of the liver have been reported since then.7 The industrial complex from which the comparison group was drawn comprises several divisions: a carbide 7.5; SMR 194, p < 0.05) and brain cancer (observed, 3; ex division which produces calcium carbide and acetylene pected, 0.6; SMR 498, p < 0.05) among men who had had black; a chemicals division which produces several chem^^ icals derived from acetylene: acetone, chloral, butanol^r From the Department of Social and Preventive Medicine School of Medicine. Laval University, Ste-Foy, Quebec. Canada. C1K 7Pa This research was financed partly under Crant No. 780398 from the Conseil de la Recherche en Sant* du Quebec. acetic acid, acetaldehyde, several acetates, solvents and resins; a stainless steel division which consisted of a small foundry and a division formed in 1957, which produced Journal of Occupational Medicine/Vol. 23, No. 10/October 1981 671 V R&S 114734 Fig 1. -- Population under study. sulfuric acid, chlorine and cyanide. This complex started The men whose number of years of work for these operations irl 1927; some of its divisions opened at a later companies, as established through occupational histories, date; others closed in the late 1960s or early 1970s. The did not exceed five were removed from the study. The re annual number of male workers averaged I.OOO.'1 maining population was then subdivided into three groups. The first group (known as the unexposed cohort) Methods consisted of men who stated they had not been involved Population Under Study. -- All production workers with the production of VCM or PVC, or both, for a period whose names appeared on the unions' lists or the com greater than five months. The second group (constituting panies payrolls of the VC plant and the industrial complex the exposed cohort) comprised men who declared they between January 1,1948, and December 31,1972, consti had worked on the production of VCM-PVC for more tuted the population under study. Several methods (old than five years. Finally, the third group consisted of the addresses, telephone books, Quebec motor vehicles men who had worked on the production of VCM-PVC for bureau, inquiries with fellow workers and families, a period of six months to five years. This group was ex searches in homes for retired persons) were used to trace cluded from analysis because it neither fulfilled the these men, 18% of whom were still working for the same definition of the exposed nor the unexposed cohort companies. Once identified, the worker himself or his Data Sources. -- The vital status of these men was next of kin (spouse, child or close relative) was questioned established as of December 31, 1977, Causes of death at home or at work by a trained interviewer. The ques were determined from death certificates. They were tions permitted the reconstruction of a detailed occupa- coded according to the 8th revision (ICDA) by one of the nal history within and outside the companies under authors, who was unaware at the time of the coding as to dy. Questions were also asked regarding smoking which group the man belonged. For cancer cases, a histohabits. For each deceased person believed to have been pathological confirmation was sought on autopsy, biopsy exposed to VCM, the occupational history was confirmed or cytology reports contained in the medical files at the with the help of a fellow worker. hospital. Information on Quebec's population and mor* 672 Mortality of Workers Exposed to Vinyl Chloride Monomer/Th6riault and Allard V R&S 114735 ___ Table 1. -- Age Distribution of the Men in the VCMExposed and Unexposed Cohorts as of December 31,1977 (Deceased Excluded). Age 25-29 30-34 35-39 40-44 45-49 50-54 55-59 60-64 65-69 70-74 75 + Total Exposed No. % 0-- 0-- 1 0.3 28 7.1 75 19.1 96 24.5 102 26.0 53 13.5 30 ----- 7.7 6 1.5 1 0.3 392 100.0 Unexposed No. % 1 0.2 3 0.5 6 0.9 45 7.1 55 8.6 84 - 13.2 112 17.6 110 17.3 113 17.7 65 10.2 43 6.7 637 100.0 tality by cause, age and sex was secured from Statistics Canada for the year 1971. In order to ensure the validity of the results, only the causes of death which appeared on the death certificates were used in the comparisons. Statistical Analysis. -- The results were analyzed using the "man-years at risk" method as described by Hill.11 The number of person-years of observation was established for each five-year age group and for each five-year period between 1948 and 1977. The mortality in the exposed cohort was compared with that in the unexposed cohort according to the method of relative risk. The tests of significance used were those of Mantel, HaenszeP4 " and Miettinen.'*17 The observed mortality was also compared with the mor tality expected from the population of the Province of Quebec according to the method of the standardized mortality ratio (SMR)." The test of significance used was the one described by Bailar and Ederer." Quality Control Procedures. -- A thorough occupa tional history was developed for workers who were ex posed within the VC plant The questionnaires presented a detailed profile of the plant while specifying possible jobs within its major units. Before the beginning of the study, it was pre-tested with former employees. Interviewers who contacted the workers had received training by the same person and were supervised during the entire study. The completed questionnaires were re viewed weekly by one of the authors and missing infor mation was sought by telephone. Data entered in the computer were systematically checked against the origi nal questionnaires. Results As shown in Fig 1, the population under study originally comprised 1,659 workers. Of these, 48 men were untrace able, leaving the number in the study population at 1,611 (or 97.1 % of the original number). After the removal of the 156 men who had worked at the companies for a period not exceeding five years and the 134 men who had been exposed to VCM for a perod extending from six months to five years, the exposed cohort comprised 451 men and the unexposed one, 870 men. In the exposed cohort, the proportion of men who responded to the questionnaire for themselves was 84.3%. In the unex posed cohort, this proportion was somewhat lower at 68.2%. The age distributions of the workers in the exposed and unexposed cohorts as of December 31,1977, reveals that workers in the unexposed cohort (Table 1) were somewhat older. Since the industrial complex had been in operation for several years when the VC plant opened, the finding of older people in the latter cohort was predictable. Table 2 shows the distribution of the exposed workers according to their length of exposure and their observa tion period (time between first employment and end of follow-up). Among these, 81 % (365 out of 451) had an ob servation period of 15 years or more. For 23% of them, the time elapsed between first exposure and the end of the follow-up period exceeded 29 years. The mean obser vation period for the exposed cohort was 22.6 years. Table 3 gives the diagnosis at death and the diagnosis after histopathological review of all the cancer cases found in the exposed group. On the death certificates, eight cancers of the liver were reported. The histopatho logical review rejected one of them as being a cancer of the sigmoidal flexure of the colon, but added another as an incorrectly diagnosed cancer of the peritoneal cavity. All eight cancers of the liver turned out to be angiosar coma of the liver, it is noted that two more angiosar comas of the liver were certified as cirrhosis on the death certificates and were considered as such throughout this -Study. Tabled compares the mortality of the exposed workers with the mortality of the unexposed workers. The con fidence interval (95%) of the relative risk indicates that there is a significant excess of mortality due to cancers of the digestive system in the exposed group (RR 6.25, con fidence interval 2.69 to 14.52). There is no excess in the other causes of death. Table 5 shows the standardized mortality ratios (SMRs) for the exposed workers in comparison with the entire population of the Province of Quebec. The SMR for all causes is 83.02, thereby showing no excess of death in the exposed cohort Although the SMRs for all cancers, respi ratory system diseases and digestive system diseases ex- Table 2. -- Distribution of Exposed Workers by Length of Exposure and Observation Period. Exposure (Yurt) 5- 9 10-14 15-19 20-24 25-29 30+ Total Observation Period (Yurt) 5-9 10-14 15-19 20-24 25-29 30 + 30 28 7 40 2 4 28 26 19 17 5 -- -- 17 35 19 8 -- -- ___ 22 24 15 34-- -- --- -- 33 -- ___ ___ ___ -- 38 30 56 50 116 96 103 Journal of Occupational Medicine/Vol. 23, No. 10/0ctober 1981 Total 111 95 79 61 67 38 451 673 Table 3. -- Age at Death, Diagnosis on Death Certificate and Diagnosis on Histological Review of the 20 VCM-Exposed Workers Who Died of Cancer. II Case Number Death Certificate Histological Review Age at Death (Years) 1 Cancer of the liver Cancer of the sigmoid flexure of the colon 2 Cancer of the intestine Metastatic cancer of the colon 3 Carcinoma of the trachea Carcinoma of the trachea 4 Angiosarcoma oi the Angiosarcoma of the liver peritoneal cavity 5 Cancer of.the lung (no review) 6 Cancer of the liver Angiosarcoma of the liver 7 Cancer of the liver Angiosarcoma Of the liver 8 Cancer of the prostate Cancer of the prostate 9 Angiosarcoma Of the liver Angiosarcoma of the liver 10 Cancer of the pancreas Epithelioma ol the head of the pancreas 11 Adenosarcoma of the intestine Adenosarcoma of the intestine 12 Hepatoma Angiosarcoma of the liver 13 Cancer ot the liver Angiosarcoma of the liver 14 Cancer of the pancreas Anaplastic epithelioma of the pancreas 15 Chondrosarcoma Chondrosarcoma 16 Hepatoma Angiosarcoma of the liver 17 Primary cancer of the liver Angiosarcoma of the liver 18 Cancer of the digestive track Cancer ot Ihe cecum 19 Cancer of the biliary ducts Cancer of the head of the pancreas 20 Lymphoid leukemia Lymphoid leukemia 73 56 50 41 54 42 57 64 62 40 53 53 53 59 43 48 53 52 54 48 Total Exposure (Years) 15 16 15 11 16 19 17 26 18 8 . 25 24 22 29 7 22 5 26 12 21 Latency Period (Years) 28 17 15 11 16 19 23 29 23 13 28 24 23 30 10 22 12 26 12 21 ceed 100, they are not statistically significant On the other hand, deaths by accident and violence are signifi cantly lower (p < 0.01). As for specific cancer deaths, a significant excess is noted for cancer of the digestive system only. This excess is accounted for exclusively by fcancer of the liver (8 cases observed versus 0.14 expected, 5,714.29). No other cancers are in excess. The cancer mortality among the workers who were still alive 15 years after their first exposure shows the same results as the mortality in the entire group, namely that only cancers of the digestive system are in excess in the exposed cohort (Table 6). Discussion Difficulties encountered in comparing the mortality of a group of workers with the mortality of a whole popula tion are well documented. They are termed: "healthy population selection effect," "survivor population effect,"10 "ethnic/cultural/social characteristics." They are believed to underestimate the actual mortality of a group of workers and therefore to underscore the effects of the risks to which these individuals are being exposed at work. There are no easy ways to avoid such difficulties. One method would be to compare a group of workers with another group of workers with the same characteris tics except for the exposure to the risk under study. The problems encountered in so doing are no easier. Although the group most similar to industrial workers is probably anothergroup of industrial workers from the same town, it is almost impossible to think of a group of industrial R&S 114736 ^ ^ P 674 Table 4. -- Deaths Observed in the VCM-Exposed and Unexposed Cohorts, Age-Adjusted Relative Risks and Confidence intervals of the Relative Risk. Cause of Death All cancers (140-209) Mouth and pharynx (140-149) Digestive (150-159) Respiratory (160-163) Bone, skin, connective (170-173) Urogenital (185-189) Eye, CNS (190-192) Leukemia (200-209) Other cancers Cardiovascular (390-458) Respiratory (460-519) Digestive (520-577) Accidents and violence (800-899) Others Total Significant according to Miettinen's test11 tOne cause unknown No. of Cases Exposed Unexposod 20 0 14 2 2 1 0 1 0 25 6 4 2 2t 59 52 2 9 20 2 7 2 4 6 124 15 12 15 15 233 Relative Risk 1.48 -- 6.25 0.36 1.67 0.83 -- 1.15 _ 0.80 0.81 1.44 0.51 0.57 1.07 95% Confidence Interval of the Relative Risk 0.84- 2.61 ___ 2.69-14.52* 0.08- 1.58 0.21-12.99 0.10- 7.10 _ 0.07-17.80 _. 0.47- 1.36 0.44- 9.35 0.49- 4.25 0.21- 1.24 0.14- 2.39 0.79- 1.45 Mortality of Workers Exposed to Vinyl Chloride Monomer/Th6riault and Allard k Table 5. - Observed and Expected* Numbers of Deaths in the VCM-Exposed Cohort and Standardized Mortality Ratios (SMRs). Cause of Death All cancers Mouth and pharynx Digestive Respiratory Bone, skin, connective Urogenital Eye. CNS Leukemia Other cancers Cardiovascular Respiratory Digestive . _ Accidents and violence Others Total (140-209) (140-149) (150-159) (160-163) (170-173) (185-189) (190-192) (200-209) (390-458) (460-519) (520-577) (800-999) Observed 20 0 14 2 2 1 0 1 0 25 6 4 2 2% 59 Expected 16.37 0.64 5.40 5.78 0.38 1.33 0.60 1.67 0.54 31.67 3.21 3.85 10.58 5.40 71.07 SMR 122.17 -- 259.26) 34.60 526.32 75.19 -- 59.88 -- 78.94 186.91 103 SO 18.90) 37.03 83.02 'Expected numbers obtained from 1971 males Quebec death rates applied to person-years in each age group tpCO.OITM $One cause unknown workers not exposed to some kind of carcinogenic risks. In the present study, the authors decided to compare the exposed workers with both the general population and a group of unexposed workers. It must be noted that among the unexposed workers, no causes of death were found to be in excess when their mortality was compared with that of the general population. The comparison of the mortality of the exposed and the unexposed vyorkers gave results similar to the one reached by the comparison with the general population. The nuhnber of cancer of the digestive system was signifi cantly higher in the exposed cohort No other cancer was in excess and there even seemed to be a deficit (not statis tically significant) in respiratory cancers. Considering the fact that animals experimentally ex posed to VCM have developed not only angiosarcomas of the liver but also malignant tumors of several organs (skin, lung, bone," Zimbal glands, kidneys, mammary glands)" 33 33 and since several epidemiological studies on VCM-exposed workers have found an excess of lung cancer,34 the results of this study of heavily exposed men are surprising. Three questions can be raised. One con cerns the ascertainment of the population. In retrospec tive studies conducted from old company or union rec ords it sometimes happens that records of people who die or leave get lost These losses could account for the low cancer rates observed. But this is a remote possibility, since the approach followed was a historical-prospective retrospective one, which began with old lists of workers composed in 1948 and 1952 and followed almost every one through time, adding new workers as they first ap peared on more recent lists. This was done without knowl edge of exposure to VCM. A second reason for low lung cancer rates could be a bias in the classification of persons as exposed or not ex posed. The xlassification was made essentially on the basis of personal recollection for living persons and by family members with confirmation by co-workers for deceased persons. To minimize such bias, much attention had been given to the occupational history (with indica tion of the buildings the subject had worked in) and workers with VCM exposure shorter than five years were rejected. There remains the possibility that some of the .men who died of liver cancer, had they lived longer, mighTTiave developed lung cancer. The small size of the population under study increased the probability of not detecting an excess that was actual ly present. This was a case where "the number of person/ years at risk may be too small to detect an increased risk of cancer."33 It was particularly true for rare cancers such as cancer of the brain and cancer of the lymphatic system. Inasmuch as cancer of the lung was concerned. Table 6. -- Relative Risks and SMRs for Cancer Deaths Among Men * Having an Observation Period of 15 Years and More. Cancers All cancers Mouth and pharynx Digestive Respiratory Bone, skin, connective Urogenital Eye. CNS Leukemia Other cancers *p<0.01' (140-209) (140-149) (150-159) (160-163) (170-173) - (185-189) (190-192) (200-209) Relative No. Risk 15 1.30 0-- 11 5.68 2 0.35 0-- 1 0.82 0-- 1 0.59 0-- 95% Confidence Interval of the Relative Risk 0.97- 1.74 -- 2.72-11.58 0.25- 4.73 -- -- -- -- -- Journal of Occupational Medicine/Vol. 23, No. 10/October 1981 SMR 129 -- 281* 47 -- -- J- 675 R&S 114737 R&S 114738 the situation was somewhat different. When one applies the Quebec lung cancer death rate after adjustment for age to the exposed cohort the probability of finding an Bicess twice as high as in Quebec was 73%. This proba bility would have been 99% had the excess been three times as great in the exposed group.15 In view of the fact that the smoking experience of the exposed workers was similar to that of the unexposed and considering that the subjects in this study are believed to have been heavily exposed to VCM, it seems reasonable to conclude that the excess of lung cancer associated with VCM exposure, if present is indeed rather small. The authors wish to express their gratitude to the following who contributed to this study' workers, union representatives and management personnel of the companies Canadian Resins and Chemicals, Shawinigan Chemicals. B.F Goodrich and Culf Canada, Drs H Hurra and J Fabia, epidemiologists, and Dr. F Delorme, pathologist, for their help and advice; the several hospitals con tacted. and the several persons who assisted with the collection of information and the writing of this article References 1. Creech JL Jr and Johnson MN: Angiosarcoma of liver in the man ufacture of polyvinyl chloride. / Occup Med 16:150-151, 1974, 2. Tabershaw IR and Caffey WR: Mortality study of workers in the manufacture of vinyl chloride and its polymers. I Occup Med 16 509-518, 1974. 3. Nicholson WJ. Hammond EC, Seidman H, et al: Mortality experi ence of a cohort of vinyl chloride polyvinyl workers. Ann NY Acad Sci 246.225*230. 1975. 4 Duck BW. Carter JT. and Coombes E): Mortality Study of workers in a polyvinyl-chloride production plant. Lancet 2:1197-1.199. 1975. ^5 Byren D. Engholm C, Englund A, and Westerholm P: Mortality cancer morbidity in a group of Swedish VCM and PVC production Bakers Environ Health Perspect 17167-170, 1976. 6 Fox Al and Collier PF: Mortality experience of workers exposed to vinyl chloride monomer in the manufacture of polyvinyl chloride in Great Britain. Br/ Ind Med 34:1-10. 1977. 7. Delorme F and Theriault G: Ten cases of angiosarcoma of the liver in Shawinigan, Quebec. I Occup Med 20:338-340. 1978, 8. Monson R, Peters J, and lohnson M: Proportional mortality among vinyl chloride workers. Lancet 2:397*398, 1974 9. Waxweiler JR, Stringer W. and Wagner JJ: Neoplastic risk among workers exposed to vinyl chloride, Ann NY Acad Sci 270:40-48, 1976. 10. Special Report by the Editors: PVC aims for a bigger share in plastics. Can Chem Process 49:41-44.1965. 11.Conseil des Sciences du Canada: Vue d'ensemble des dangers de la contamination par le chlorure de vinyle au Canada. Quebec Science (Suppl) :1-24. 1977. __ 12. LaRochelle F; Shawinigan depuis 75 ans. Ateliers de Timprimerie Pdblicite PSquet Enr., 1976, pp 497-525. 13. Hill ID: Computing man years at risk. Br I Prev Med 26:132-134 1972. 14. Mantel N and Haenszel W: Statistical aspects of the analysis of data from retrospective studies of disease. I Natl Cancer Inst 22:710-748,1959. 15. Mantel N: Evaluation of survival data and two rank order statis tics arising in its consideration. Cancer Chemother Rep 50:163-170, 1966. 16. Miettinen OS: Standardization of risk ratios. Am J Epidemiol 96:383-388, 1972. 17. Miettinen OS: Simple interval-estimation of risk ratio. Am j Epidemiol 100:515-516.1974. 18. Caffey WR: A critique of the standardized mortality-ratio. I Oc cup Med 3:157-160.1976. 19 Bailar JC and Ederer F: Significance factors for the ratio of a Poisson variable to its expectation. Biometrics 20:639-642.1964. 20. Fox A| and Collier PF: Low mortality rates in industrial cohort studies due to selection for work and survival in the industry. Br / Prev Soc Med 30:225-230. 1976. 21. Viola PL: Cancerogenic Effect of Vinyl Chloride, Vol. 29, X Inter national Cancer Congress, Houston, 1970. 22. Viola PL, Bigotti A, and Caputo A: Oncogenic response of rat skin, lungs and bones to vinyl chloride. Cancer Res 31:516-519.1971. 23. Maltoni C and Lefemine G: Cancerogenicity bio assays of vinyl chloride: Current results. Ann NY Acad Sci 246:195-218,1975. 24. Wagoner IK: Vinyl chloride and pulmonary cancer. J Env Pathol Toxicol 1:361-362. 1978. 25. Pell S, O'Berg MT. and Karrh B: Cancer epidemiologic surveil lance in the DuPont Company. J Occup Med 20.725-740.1978. Reducing Financial Risks There is too little evidence that the nuclear industry, which was never very good at calculating its own interest, is seriously contemplating the effect of another Three Mile Island on its future -- or its fortunes. The industry has already returned to complaints that the safety bureaucracy is nitpicking it to death, running up construction costs and delaying licenses. The latest complaint is that, in the aftermath of the TM1 accident, NRC diverted safety reviewers to deal with operating reactors and fell behind in ap proving new licenses. There is a kind of tunnel vision in the failure to see the connection between safety and the special financial risk involved in nuclear investment To use the current energy crisis to avoid safety requirements is to ignore the fact that safety is essential to protec ting the heavy investment in nuclear power. Some hope lies in the fact that, although the industry thinks safety requirements are a pair, in the neck, they may be forced to look at them as an investment in financial public relations to alleviate the bankers' skepticism. -- Victor Gilinsky, Commissioner, the Nuclear Regulatory Commission, as quoted in "Notable and Quotable," in The Wall Street Journal. May 21.1981. il vM 676 Mortality of Workers Exposed to Vinyl Chloride Monomer/Th6riault and Allard Original Articles October 1981 Vol. 23 No. Cancer Mortality of a Group of Canadian Workers Exposed to Vinyl Chloride Monomer G. Theriault, M.D., Dr.P.H., and P. Allard, M.D., M.Sc. fi&S 114739 The present study was undertaken to find out whether there was an excess of cancer mortality from causes other than angiosarcoma of the liver among a group of workers heavily exposed to vinyl chloride monomer (VCM). The mortality of 451 workers exposed to VCM for more than five years was compared with that of 870 workers from the same company who had not been exposed to VCM. The relative risk for digestive cancer was significantly higher than 1 (6.25, confidence interval 2.69 to 14.52) in the ex posed group. The standardized mortality ratio (SMR) for digestive cancer was also higher (SMR 259.26 p < 0.01) than that of the general population. No other cancer was in excess. Since the exposed workers are known to have had a cigarette smoking experience similar to that of those who were not exposed, it is concluded that the association Be^ t1ween lIung c_ a_ ncer an. d./ iV/C"Mi a expos_ure, iiftpres_entm, is i`ndeedJ rather small. Several studies have demonstrated that vinyl chloride monomer (VCM) produces angiosarcoma of the liver among people exposed at work.1*7 There is, however, much controversy about the capability of VCM to gener ate other types of cancer in man. One author,* using a proportional mortality ratio technique, described ex cesses of lung cancer (observed, 13; expected, 7.9; obs/exp 1.6) and brain cancer (observed, 5; expected, 1.2; obs/exp 4.2) in a group of 161 workers from two vinyl chloride plants.* Another author,* who studied the mortality of a group of 1,294 American workers exposed to VCM, re ported excesses of lung cancer (observed, 11 expected, 7.5; SMR 194, p < 0.05) and brain cancer (observed, 3; ex pected, 0.6; SMR 498, p < 0.05) among men who had had From the Department of Social and Preventive Medicine School of Medicine, Laval University. Ste-Foy, Quebec, Canada, C1K 7P4 This research was financed partly under Grant No. 780398 from the Conseil de la Recherche en Santfc du Quebec. their first exposure more than 15 years before their death. While most authors have not been able to document a significant excess of any other cancer in their studies, they have noted that such an excess may exist. In order to better document the cancer mortality asso ciated with exposure to VCM, the present study of the workers of the Shawinigan vinyl chloride polymeriza^k plant was undertaken. The main objectives of the study were (1) to compare mortality among vinyl chloride workers with the mortality of workers from a nearby industrial complex; and (2) to compare the mortality of the VCM workers with that of the general population. The Shawinigan vinyl'chloride plant opened in 1943 as an independent company which produced both VCM and PVC (polyvinyl chloride). In 1n95rr8t, uit was _m___e__r_g__e_dJ Iin.lto* an industrial complexf(>t_oLbAe. used as the comparison group) of which it became the Cana dian Resins Division. In the late 1960s, the VCM produc tion ceased and only the polymerization process con tinued operation. In 1972, this division was sold and again became an independent plant. Over the years, the VCM and PVC productions fluctuated. The average annual work force at the plant was 225 men. In 1965, VCM pro duction was estimated at 90 million pounds and PVC at 60 million pounds.'0 In 1977, PVC production was 22.3 million pounds." Since several episodes of unconscious ness among workers were reported, the level of VCM in the air at the workplace is believed to have been high dur ing this period. Ten cases of angiosarcoma of the liver have been reported since then.7 The industrial complex from which the comparison group was drawn comprises several divisions; a carbide division which produces calcium carbide and acetylene black; a chemicals division which produces several chem icals derived from acetylene: acetone, chloral, buMbl, acetic acid, acetaldehyde, several acetates, solvent^^d resins; a stainless steel division which consisted of a small foundry and a division formed in 1957, which produced Journal of Occupational Medicine/Vol. 23, No. 10/October 1981 671 33 So CO Fig 1. -- Population under study. sulfuric acid, chlorine and cyanide. This complex started operations in 1927; some of its divisions opened at a later date; others closed in the late 1960s or early 1970s. The annual number of male workers averaged 1,000.'* Methods Population Under Study. -- All production workers whose names appeared on the unions' lists or the com panies payrolls of the VC plant and the industrial complex between January 1,1948, and December 31,1972, consti tuted the population under study. Several methods (old addresses, telephone books, Quebec motor vehicles bureau, inquiries with fellow workers and families, searches in homes for retired persons) were used to trace these men, 18% of whom were still working for the same companies. Once identified, the worker himself or his next of kin (spouse, child or close relative) was questioned at home or at work by a trained interviewer. The ques tions permitted the reconstruction of a detailed occupa tional history within and outside the companies under study. Questions were also asked regarding smoking habits. For each deceased person believed to have been exposed to VCM, the occupational history was confirmed with the help of a fellow worker. "The* men whose number of years of work for these companies, as established through occupational histories, did not exceed five were removed from the study. The re maining population was then subdivided into three groups. The first group (known as the unexposed cohort) consisted of men who stated they had not been involved with the production of VCM or PVC, or both, for a period greater than five months. The second group (constituting the exposed cohort) comprised men who declared they had worked on the production of VCM-PVC for more than five years. Finally, the third group consisted of the men who had worked on the production of VCM-PVC for a period of six months to five years. This group was ex cluded from analysis because it neither fulfilled the definition of the exposed nor the unexposed cohort. Data Sources. -- The vital status of these men was established as of December 31, 1977. Causes of death were determined from death certificates. They were coded according to the 8th revision (ICDA) by one of the authors, who was unaware at the time of the coding as to which group the man belonged. For cancer cases, a histopathological confirmation was sought on autopsy, biopsy or cytology reports contained in the medical files at the hospital. Information on Quebec's population and mor- 672 Mortality of Workers Exposed to Vinyl Chloride Monomer/Thdriault and Allard R&S 114741 Table 1. -- Age Distribution of the Men in the VCMExposed and Unexposed Cohorts as of December 31, 1977 (Deceased Excluded). Age 25-29 30-34 35-39 40-44 45-49 50-54 55-59 60-64 65-69 70-74 ''' 75 + Total Exposed No. % 0-- 0-- 1 0.3 28 7.1 75 19.1 96 24.5 102 26.0 53 13.5 _30 -------- 7.7 6 1.5 1 0.3 392 100.0 Unexposed No. % 1 0.2 3 0.5 6 0.9 45 7.1 55 8.6 84 . 13.2 112 17.6 110 17.3 113 17.7 65 10.2 43 6.7 637 100.0 tality by cause, age and sex was secured from Statistics Canada for the year 1971. In order to ensure the validity of the results, only the causes of death which appeared on the death certificates were used in the comparisons. Statistical Analysis. -- The results were analyzed using the "man-years at risk" method as described by Hill.11 The number of person-years of observation was established for each five-year age group and for each five-year period between 1948 and 1977. The mortality in the exposed cohort was compared with that in the unexposed cohort according to the method of relative risk. The tests of significance used were those of Mantel, Haenszel14 '* and Miettinen.1417 The observed mortality was also compared with the mor tality expected from the population of the Province of Quebec according to the method of the standardized mortality ratio (SMR).1* The test of significance used was the one described by Bailar and Ederer.19 Quality Control Procedures. -- A thorough occupa tional history was developed for workers who were ex posed within the VC plant. The questionnaires presented a detailed profile of the plant while specifying possible' Jobs within its major units. Before the beginning of the study, it was pre-tested with former employees. Interviewers who contacted the workers had received training by the same person and were supervised during the entire study. The completed questionnaires were re viewed weekly by one of the authors and missing infor mation was sought by telephone. Data entered in the computer were systematically checked against the origi nal questionnaires. Results As shown in Fig 1, the population under study originally comprised 1,659 workers. Of these, 48 men were untraceable, leaving the number in the study population at 1,611 (or 97.1 % of the original number). After the removal the 156 men who had worked at the companies fo period not exceeding five years and the 134 men who had been exposed to VCM for a perod extending from six months to five years, the exposed cohort comprised 451 men and the unexposed one, 870 men. In the exposed cohort, the proportion of men who responded to the questionnaire for themselves was 84.3%. In the unex posed cohort, this proportion was somewhat lower at 68.2%. The age distributions of the workers in the exposed and unexposed cohorts as of December 31,1977, reveals that workers in the unexposed cohort (Table 1) were somewhat older. Since the industrial complex had been in operation for several years when the VC plant opened, the finding of older people in the latter cohort was predictable. Table 2 shows the distribution of the exposed workers according to their length of exposure and their observa tion period (time between first employment and end of follow-up). Among these, 81 % (365 out of 451) had an ob servation period of 15 years or more. For 23% of them, the time elapsed between first exposure and the end of the follow-up period exceeded 29 years. The mean obser vation period for the exposed cohort was 22.6 years. Table 3 gives the diagnosis at death and the diagnosis after histopathological review of all the cancer cases found in the exposed group. On the death certificates, eight cancers of the liver were reported. The histopatho logical review rejected one of them as being a cance the sigmoidal flexure of the colon, but added anothe an incorrectly diagnosed cancer of the peritoneal cavity. All eight cancers of the liver turned out to be angiosar coma of the liver. It is noted that two more angiosar comas of the liver were certified as cirrhosis on the death certificates and were considered as such throughout this study. ---Table 4 compares the mortality of the exposed workers with the mortality of the unexposed workers. The con fidence interval (95%) of the relative risk indicates that there is a significant excess of mortality due to cancers of the digestive system in the exposed group (RR 6.25, con fidence interval 2.69 to 14.52). There is no excess in the other causes of death. Table 5 shows the standardized mortality ratios (SMRs) for the exposed workers in comparison with the entire population of the Province of Quebec. The SMR for all causes is 83.02, thereby showing no excess of death in the exposed cohort. Although the SMRs for all cancers, respi ratory system diseases and digestive system diseases ex- Table 2. -- Distribution of Exposed Workers by Length of Exposure and Observation Period. Exposure (Years) 5- 9 10-14 15-19 20-24 25-29 30+ ______ Total Observation Period (Years) 5-9 10-14 15-19 20-24 25-29 30 + 30 28 7 40 2 4 -- 28 26 19 17 5 -- -- 17 35 19 8 ------ 22 24 15 ------ -- 34 33 _ ------ -- 38 30 56 50 116 96 103 Journal of Occupational Medicine/Vol. 23, No. 10/0ctober 1981 Total 111 95 79 61 67 38 451 673 Table 3. -- Age at Death, Diagnosis on Death Certificate and Diagnosis on Histological Review of the 20 VCM-Exposed Workers Who Died of Cancer. Case Number 1 Death Certificate Cancer of the liver 2 Cancer of the intestine 3 Carcinoma of the trachea 4 Angiosarcoma of the peritoneal cavity 5 Cancer of the lung 6 Cancer of the liver 7 Cancer of the liver 8 Cancer of the prostate 9 Angiosarcoma of the liver 10 Cancer of the pancreas 11 Adenosarcoma of the intestine 12 Hepatoma 13 Cancer of the liver 14 Cancer of the pancreas 15 Chondrosarcoma 16 Hepatoma 17 Primary cancer of the liver 18 Cancer of the digestive track 19 Cancer of the biliary ducts 20 Lymphoid leukemia Histological Review Cancer of the sigmoid flexure of the colon Metastatic cancer of the colon Carcinoma of the trachea Angiosarcoma of the liver (no review) Angiosarcoma of the liver Angiosarcoma of the liver Cancer of the prostate Angiosarcoma of the liver Epithelioma of the head of the pancreas Adenosarcoma of the intestine Angiosarcoma of the liver Angiosarcoma of the liver Anaplastic epithelioma of the pancreas Chondrosarcoma Angiosarcoma of the liver Angiosarcoma of the liver Cancer of the cecum Cancer of the head of the pancreas Lymphoid leukemia Age at Death (Years) 73 56 50 41 54 42 57 64 62 40 53 53 53 59 43 48 53 52 54 48 Total Exposure (Years) 15 16 15 11 16 19. 17 26 18 8 25 24 22 29 7 22 5 26 12 21 Latency Period (Years) 28 17 15 11 16 19 23 29 23 13 28 24 23 30 10 22 12 26 12 21 ceed 100, they are not statistically significant On the other hand, deaths by accident and violence are signifi cantly lower (p < 0.01). As for specific cancer deaths, a significant excess is noted for cancer of the digestive system only. This excess is accounted for exclusively by cancer of the liver (8 cases observed versus 0.14 expected, 5MR 5,714.29). No other cancers are in excess. The cancer mortality among the workers who were still alive 15 years after their first exposure shows the same results as the mortality in the entire group, namely that only cancers of the digestive system are in excess in the exposed cohort (Table 6). Discussion Difficulties encountered in comparing the mortality of a group of workers with the mortality of a whole popula tion are well documented. They are termed: "healthy population selection effect," "survivor population effect,"20 "ethnic/cultural/social characteristics." They are believed to underestimate the actual mortality of a group of workers and therefore to underscore the effects of the risks to which these individuals are being exposed at work. There are no easy ways to avoid such difficulties. One method would be to compare a group of workers with another group of workers with the same characteris tics except for the exposure to the risk under study. The problems encountered in so doing are no easier. Although the group most similar to industrial workers is probably another group of industrial workers from the same town, it is almost impossible to think of a group of industrial Table 4. -- Deaths Observed in the VCM-Exposed and Unexposed Cohorts, Age-Adjusted Relative Risks and Confidence Intervals of the Relative Risk. Cause of Death All cancers (140-209) Mouth and pharynx (140-149) Digestive (150-159) Respiratory (160-163) Bone, skin, connective (170-173) Urogenital (185-189) Eye, CNS (190-192) Leukemia (200-209) Other cancers Cardiovascular (390-458) Respiratory (460-519) Digestive (520-577) Accidents and violence (800-899) Others Total Significant according to Miettinen's test17 tone cause unknown 674 No. of Casos Exposed Unexposed 20 52 02 14 9 2 20 22 17 02 14 06 25 124 6 15 4 12 2 15 2t 15 59 233 Relative Risk 1.48 -- 6.25 0.36 1.67 0.83 -- 1.15 0.80 0.81 1.44 0.51 0.57 1.07 95% Confidence Interval of the Relative Risk 0.84- 2.61 -- 2.69-14.52* 0.08- 1.58 0.21-12.99 0.10- 7.10 0.07-17.80 _ 0.47- 1.36 0.44- 9.35 0.49- 4.25 0.21- 1.24 0.14- 2.39 Q.79- 1.45 Mortality of Workers Exposed to Vinyl Chloride Monomer/Th6riault and Allard R&S 114742 V R&S 114743 ,\r A Table 5. -- Observed and Expected* Numbers of Deaths in the VCM-Exposed Cohort and Standardized Mortality Ratios (SMRs). Cause of Death All cancers Mouth and pharynx Digestive Respiratory Bone, skin, connective Urogenital Eye, CNS Leukemia Other cancers Cardiovascular Respiratory Digestive Accidents and violence Others Total (140-209) (140-149) (150-159) (160-163) (170-173) (185-189) (190-192) (200-209) (390-458) (460-519) (520-577) (800-999) Observed 20 0 14 2 2 1 0 1 0 25 6 4 2 n 59 Expected 16.37 0.64 5.40 5.78 0.38 1.33 0.60 1.67 0.54 31.67 3.21 3.85 10.58 5.40 71.07 $MR 122.17 -- 259.261 34.60 526.32 75.19 -- 59.88 -- 78.94 186.91 103.90 18.901 37.03 83.02 Expected numbers obtained from 1971 males Quebec death rates applied to person-years in each age group tp < 0.0119 $One cause unknown workers not exposed to some kind of carcinogenic risks, in the present study, the authors decided to compare the exposed workers with both the general population and a group of unexposed workers. It must be noted that among the unexposed workers, no causes of death were found to be in excess when their mortality was compared with that of the general population. The comparison of the mortality of the exposed and the unexposed workers gave results similar to the'one reached by the comparison with the general population. The number of cancer of the digestive system was signifi cantly higher in the exposed cohort. No other cancer was in excess and there even seemed to be a deficit (not statis tically significant) in respiratory cancers. Considering the fact that animals experimentally ex posed to VCM have developed not only angiosarcomas of the liver but also malignant tumors of several organs (skin, lung, bone, Zimbal glands, kidneys, mammary glands)" 1313 and since several epidemiological studies on VCM-exposed workers have found an excess of lung cancer,34 the results of this study of heavily exposed men are surprising. Three questions can be raised. One con cerns the ascertainment of the population. In retrospec tive studies conducted from old company or union rec ords it sometimes happens that records of people who die or leave get lost. These losses could account for the low cancer rates observed. But this is a remote possibility, since the approach followed was a historical-prospective retrospective one, which began with old lists of workers composed in 1948 and 1952 and followed almost every one through time, adding new workers as they first ap peared on more recent lists. This was done without knowl edge of exposure to VCM. A second reason for low lung cancer rates could be a bias in the classification of persons as exposed or not ex posed. The classification was made essentially on the basis of personal recollection for living persons and b'i family members with confirmation by co-workers for deceased persons. To minimize such bias, much attention had been given to the occupational history (with indica tion of the buildings the subject had worked in) and workers with VCM exposure shorter than five years were rejected. There remains the possibility that some of the men who died of liver cancer, had they lived longer, might have developed lung cancer. The small size of the population under study increased the probability of not detecting an excess that was actual ly present. This was a case where "the number of person/ years at risk may be too small to detect an increased risk of cancer."23 It was particularly true for rare cancers such as cancer of the brain and cancer of the lymphatic system. Inasmuch as cancer of the lung was concerned, Table 6. -- Relative Risks and SMRs for Cancer Deaths Among Men Having an Observation Period of 15 Years and More. Cancers All cancers Mouth and pharynx Digestive Respiratory Bone, skin, connective Urogenital Eye, CNS Leukemia Other cancers *p< 0.0119 (140-209) (140-149) (150-159) (160-163) (170-173) (185-189) (190-192) (200-209) Relative NO. Risk 15 1.30 0-- 11 5.68 2 0.35 0-- 1 0.82 0-- 1 0.59 0 95% Confidence Interval of the Relative Risk 0.97- 1.74 -- 2.72-11.58 0.25- 4.73 -- -- -- -- Journal of Occupational Medicine/Vol. 23, No. 10/0ctober 1981 SMR 129 -- 281* 47 -- A fl 1 675 # the situation was somewhat different. When one applies the Quebec lung cancer death rate after adjustment for age to the exposed cohort, the probability of finding an excess twice as high as in Quebec was 73%. This proba bility would have been 99% had the excess been three times as great in the exposed group.11 In view of the fact that the smoking experience of the exposed workers was similar to that of the unexposed and considering that the subjects in this study are believed to have been heavily exposed to VCM, it seems reasonable to conclude that the excess of lung cancer associated with VCM exposure, if present is indeed rather small. The authors wish to express their gratitude to the following who contributed to this study: workers, union representatives and management personnel of the companies Canadian Resins and Chemicals, Shawimgan Chemicals. B F Coodrich and Gulf Canada. Drs. H. Iturra and 1 Fabia, epidemiologists, and Dr F Delorme, pathologist, for their help and advice; the several hospitals con tacted, and the several persons who assisted with the collection of information and the writing of this article. References 1. Creech |L Jr and Johnson MN- Angiosarcoma of liver in the man ufacture of polyvinyl chloride. / Occup Med 16:150-151,1974 2. Tabershaw IR and Caffey WR: Mortality study of workers in the manufacture of vinyl chloride and its polymers. I Occup Med 16:509-518, 1974. 3. Nicholson WJ, Hammond EC, Seidman H. et al: Mortality experi ence of a cohort of vinyl chloride polyvinyl workers. Ann NY Acad Sci 246:225-230. 1975 4. Duck BW, Carter JT, and Coombes EJ: Mortality study of workers in a polyvinyl-chloride production plant Lancet 2:1197-1199,1975. 5. Byren D, Engholm G, Englund A, and Westerholm P: Mortality and cancer morbidity in a group of Swedish VCM and PVC production workers. Environ Health Perspect 17:167-170,1976. 6. Fox AJ and Collier PF: Mortality experience of workers exposed to vinyl chloride monomer in the manufacture of polyvinyl chloride in Great Britain. Br) Ind Med 34:1-10,1977. 7. Delorme F and Theriault G: Ten cases of angiosarcoma of the liver in Shawimgan, Quebec. / Occup Med 20.338-340, 1978, 8. Monson R, Peters J, and Johnson M. Proportional mortality among vinyl chloride workers Lancet 2.397-398, 1974 9. Waxweiler JR, Stringer W, and Wagner JJ. Neoplastic risk among workers exposed to vinyl chloride. Ann NY Acad Sci 270 40-48, 1976. 10. Special Report by the Editors: PVC aims for a bigger share in plastics Can Chem Process 49 41-44. 1965. 11. Conseil des Sciences du Canada. Vue d'ensemble des dangers de la contamination par le chlorure de vinyle au Canada. Quebec Science (Suppl) :1-24, 1977 12. LaRochelle F: Shawinigan depuis 75 ans. Ateliers de I'impfimerie Publicity PSquet Enr,, 1976, pp 497-525. 13. Hill ID: Computing man years at risk. Br I Prev Med 26:132-134, 1972. 14. Mantel N and Haenszel W: Statistical aspects of the analysis of data from retrospective studies of disease. I Natl Cancer Inst 22:710-748, 1959. 15. Mantel N: Evaluation of survival data and two rank order statis tics arising in its consideration. Cancer Chemother Rep 50:163-170. 1966. 16. Miettinen OS: Standardization of risk ratios. Am / Epidemiol 96:383-388, 1972. 17. Miettinen OS. Simple interval-estimation of risk ratio. Am I Epidemiol 100:515-516, 1974. 18. Caffey WR. A critique of the standardized mortality ratio. I Oc cup Med 3:157-160, 1976. 19. Bailar JC and Ederer F: Significance factors for the ratio of a Poisson variable to its expectation. Biometrics 20:639-642,1964. 20. Fox A) and Collier PF: Low mortality rates in industrial cohort studies due to selection for work and survival in the industry. Br I Prev Soc Med 30:225-230. 1976 21. Viola PL: Cancerogenic Effect of Vinyl Chloride, Vol. 29, X Inter national Cancer Congress, Houston, 1970. 22. Viola PL, Bigotti A, and Caputo A: Oncogenic response of rat skin, lungs, and bones to vinyl chloride. Cancer Res 31:516-519,1971. 23. Maltoni C and Lefemine G: Cancerogenicity bio assays of vinyl chloride: Current results. Ann NY Acad Sci 246:195-218,1975. 24. Wagoner JK: Vinyl chloride and pulmonary cancer. / Env Pathol Toxicol 1:361-362. 1978 25. Pell S, O'Berg MT, and Karrh B: Cancer epidemiologic surveil lance in the DuPont Company. ) Occup Med 20:725-740,1978. R&S 114744 Reducing Financial Risks There is too little evidence that the nuclear industry, which was never very good at calculating its own interest, is seriously contemplating the effect of another Three Mile Island on its future -- or its fortunes. The industry has already returned to complaints that the safety bureaucracy is nitpicking it to death, running up construction costs and delaying licenses. The latest complaint is that, in the aftermath of the TMI accident, NRC diverted safety reviewers to deal with operating reactors and fell behind in ap proving new licenses. There is a kind of tunnel vision in the failure to see the connection between safety and the special financial risk involved in nuclear investment. To use the current energy crisis to avoid safety requirements is to ignore the fact that safety is essential to protec ting the heavy investment in nuclear power. Some hope lies in the fact that, although the industry thinks safety requirements are a pain in the neck, they may be forced to look at them as an investment in financial public relations to alleviate the bankers' skepticism. -- Victor Cilmsky, Commissioner, the Nuclear Regulatory Commission, as quoted in "Notable and Quotable," in The Wall Street Journal, May 21,1981. 676 Mortality of Workers Exposed to Vinyl Chloride Monomer/Thgriault and Allard u/*'**- ! Exposure Indices for Epidemiological - Surveillance of Carcinogenic Agents * in an Industrial Chemical Environment t _- Richard A. Greenberg, Ph.D., and Carlo H. Tamburro, M.D. A prospective system for establishing chemical expo sure indices was developed and implemented for 22 chemicals used at a Louisville chemical plant Validation of the indices was done statistically using industry-related cancer (liver angiosarcoma) and worker-matched controls. A rank ordered system for exposures was used to identify a relationship between the occurrence of disease and the presence of a suspect chemical used in the industrial en vironment A major difficulty in the epidemiology of occupa tional carcinogenesis is obtaining accurate exposure data, especially if the data must be procured after cancer develops. While epidemiological investigations of out breaks of disease, infectious or chronic, are always retro spective, the long latent period between exposure to a causative factor and the occurrence of cancer compli cates this problem. Routine continuous recording of expo sure to possible carcinogens is an ideal goal. Unfortunate ly, this is not practical for most chemicals in a modern in dustrial setting. What is eminently practical, however, is a system utilizing rank ordering of exposures for highly suspect chemicals. The B. F. Goodrich Louisville Chemical plant devel oped such a system in 1974 in response to the discovery of cases of hepatic angiosarcoma.15 This system was the basis of the initial reports. It was extensively modified during a prospective medical screening program estab lished by the University of Louisville under contract NOlCN-55212 with the cancer control program of the Nation al Cancer Institute.* 7 The authors are unaware of other similar existing data sets. Occupational studies are usual ly based on group, rather than individual, exposures. An excellent review of the literature is given by Gamble et al.` A recent discussion of a computerized system is given by Kerr.* From the University of Louisville, Dept, of Community Health (Dr Greenberg) and Dept of Medkme (Dr. Tamburro). P,0. Box 352tO, Louisville, KV 40232 Journal of Occupational Medicine/Vol. 23. No. 5/May 1981 The system for establishing exposure indices on an on going basis was developed through employee work histo ries and rank ordered job exposure categories. All jobs are classified uniquely by both area location and work de scription by means of area-description (A-D) codes which identify employment occurring in a particular building or area, independent of job; in a particular job, independent of building or. area; or by both area and description. The exposure index combines two components, i.e^ work history and job exposure category, by utilizing th^ A-D code. The chemical exposure rating is an ordered, six-category ranking assigned to each A-D number for each calendar year, as follows: Rating 0 1 2 3 4 5 6 Level of Exposure Absent from Environment (on leave, furlough, layoff, etc.) Lowest Exposure (includes exposure up to somewhere near one hour per day) Minimal Exposure to Low Levels (chemical in building -- not handled; low vapor pressure and dust level; individual prob ably works on different floor) Moderate Exposure (works around the chemical, but exposure is minimal; individ ual is frequently exposed to little spills or leaks and infrequently -- less than once per month -- to large spills or leaks) Works in Area Subject to High Occupation al Exposures (normally exposure is mini mal, but large spills or leaks occur once per month or more) Works in Areas Where Level is High (exposure levels in area are frequently high; might consider that some risk is involved if the chemical is very toxic) Intimate Contact -- Skin or High Inhala^ tion (includes individuals with daily aril direct contact with the chemicals, such aP poly cleaner in the old days and those who handled slurry) 353 I I Table 1. -- Detailed Work and Exposure history. 1A IB Work History Exposure Rank for Each Chemical AD Year Building Job No. Months Vinyl Chloride 2* 3* 1944 1945 1945 1946 1947 1948 1948 1949 1949 1949 1949 1949 1957 1957 1957 1957 1972 1972 1973 1974 1974 000 000 111 111 111 111 121 121 112 _ 112 112 112 121 121 121 121 117 000 000 000 Terminated ' 576 576 194 194 194 194 192 192 253 253 253 253 235 235 235 235 573 574 574 574 6 5 7 12 12 8 4 4 8 8 8 8 12 12 12 12 5 7 12 8 2 1 1 ........................................ 2 1 1 ........................................ 5 1 1 ........................................ 6 4 1 ........................................ 6 4 3 ........................................ 5 4 3 .............................................. 4 3 1 ........................................ 4 3 i ................................... r.. 2 6 3 ........................................ * 4 3 1 ........................................ 1 2 1 ........................................ 2 1 1 ........................................ 2 1 1 ........................................ 2 1 1 ........................................ Other chemical 22* 4 4 1 1 1 1 1 1 2 1 1 4 4 4 R&S 114746 Twenty-two chemicals from two distinct manufacturing processes, one related to synthetic rubber and the other to plastics, were selected for rating. The ratings were assigned by panels of chemical exposure judges. The full six-point scale was not utilized for all chemicals. Al though it would have been ideal to have a quantitative, continuous scale of actual exposure to parts per million, this was not available for all B. F. Goodrich employees for all chemicals (nor is it likely to be available in any other industry). In order to overcome the subjectivity inherent in this exposure rating system, knowledgeable people from each area of the plant were selected to serve as that area's chemcal exposure judges. The primary criterion for their selection was experience in the given area. As a rule, production foremen and technical people (i.e., chemical engineers) were in the majority among each area's panel of judges. The area production foremen were especially invaluable as they are experienced in and knowledgeable of all jobs performed in their areas. Most valuable, how ever, were those individuals who had been at the plant since its inception, or shortly thereafter. The chemical ex posure judges met as a group and agreement was by con sensus. Table 1A gives the work history for a hypothetical in dividual. Note that each job is identified for each year in dicating both the A-D number and the number of months worked during the year. This individual, therefore, began work in 1944 and worked for six months on A-D number 000576. He worked an additional five months on this number during 1945. He completed the last seven months in 1945 working at A-D number 111194. Table IB iden tifies the exposure rating assigned to each of the 22 monitored chemicals for each A-D number for each year. For instance, the first row indicates the exposure ratings assigned in 1944 to A-D number 000576. In 1945 these re mained unchanged. The third row indicates the exposure ratings for A-D number 111194 during 1945. The informa tion in Tables 1A and IB is combined to give cumulative exposure rank months (CERM) and average exposure ranks (AER). In this example, this individual in 1944 would have had 12 CERM for vinyl chloride in 1944, having worked six months at an exposure rank of 2 during the year. During 1945, this individual would have had CERM to vinyl chloride of (5 x 2) plus (7 x 5) or 45. He would have worked a total of 12 months and would have had an AER (45 h- 12) equal to 3.75. The assumption is made that the CERM can be used as a rank order statistic. The assump tion is empirically validated later in this paper. The CERM "cannot be considered as representing interval data and no such use of it is intended. The results given here are based on revised work histories and A-D codes developed dur ing the NCI contract period. These same codes are now applied to the individual monitoring of all employees. The data are accurately obtained in a prospective man ner. Work histories are obtained on an ongoing basis. Each time an employee changes jobs, the change is iden tified by a payroll change card indicating the last A-D number, the new A-D number and the date of change. In addition, the rating of exposures by A-D number is now done on an annual basis using auxiliary information avail able about the A-D number including individual and area monitoring pertinent to the particular A-D number. Thus far this is available for only three chemicals and only since 1974. The rating of exposures is now obtained from chemical exposure judges consisting of representatives of both labor and management. This process takes about two weeks annually. In order to determine whether such a system could work retrospectively, work histories were abstracted by A-D numbers from payroll records for all employees who worked on or after January 1,1974, from records dating back to the opening of the plant in 1942. Payroll records 354 Exposure Indices for Epidemiological Surveillance/Greenberg and Tamburro i Table 2. -- Standard Normal Deviates Comparing Exposure Ranks of Angiosarcoma Cases with the Average Exposure of Matched* Controls (by Selected Chemicals). Chemical 1 (23 Controls) 2 (29 Conlrols) Group 3 (36 Controls) (4 Controls) 1 -1.43 2.32 -0.66 -2.20 2 -2.13 -0.93 -2.00 -0.53 __ 3 -3.54 -2.16 -3.16 -1.50 4 -3.31 -1.79 -0.06 -1.76 5 3.97 0.84 5.08 1.11 -- 6 -2.01 -2.58 -1.59 -- 7 -2.33 -2.59 -1.93 -1.50 8 0.29 -0.47 1.56 -- 9 -3.92 -0.27 6.44 -2.28 10 -0.76 3.64 0.17 -0.71 11 11.27 7.59 0.93 5.08 12 -1.69 -1.35 0.45 -1.00 13 -1.78' -1.41 5.95 0.00 14 -2.41 -2.91 -2.35 - -0.85 15 -1.87 -1.63 0,40 -1.21 16 5.13 3.12 5.77 2.55 17 -0.77 -3.43 5.53 -1.50 18 1.55 -1.45 5.68 1.07 10 2.56 -4.36 4.64 -0.83 20 9.11 2.65 3.72 2.60 21 -3.87 -2.39 -2.91 -1.31 22 6.55 2.34 2.36 2.85 Matched by age, sex, race, year of employment and survival as a B. F. Goodrich employee to January 1, 1974 did not have the A-D numbers that were subsequently developed; these older records referred to many jobs which no longer existed, and to some which were per formed in buildings long since torn down. However, a staff of knowledgeable employees matched the jobs list ed on these payroll records with current A-D numbers. Final determination for controversial work records was made, wherever possible, by an individual employee's review. The chemical exposure judges, who were assign ing ordered rating exposures to A-D numbers for each year, faced these same problems. They used, in addition to their memory, whatever records of chemical processes and procedures that were available. (The company main tains a file on all products ever produced and all pro cesses ever used at the plant.) This system of determining work exposure data was validated empirically in the following manner: The in cidence of hepatic angiosarcoma at the B. F. Goodrich plant since January 1, 1974, presented in four subjects. Each of the four subjects was matched by exact year of birth, by sex and race (white or non-white), by exact year of employment, and by continuing employment at the B. F. Goodrich plant through January 1,1974. All matches are included in the subsequent analysis. Each of the angiosarcoma subjects and the corresponding matched group Were compared on CERM separately for each calj^ endar year. The results were then ranked within each c^^p endar year and the ranks were summed over the calenda^^ years. If the work histories and the ordered rating expo sures were no better than random assignments, one would expect a uniform distribution of ranks within each year and independence from year to year. Otherwise, rank order exposures should demonstrate higher exposure to vinyl chloride in those individuals with hepatic angio sarcoma. Table 2, which gives standard normal deviates, was cal culated from the observed sum of ranks, the expected sum of ranks, and the theoretical standard error (condi tional on the observed pattern of tied ranks) of the expect ed sum of ranks. Sums were over years. The results show a very clear pattern of high exposure to chemical number 16, vinyl chloride. This is to be expected a priori if the work history data and the exposure rank data are valid. A second finding of importance is that all four angio- Table 3. -- Comparison of Frequency of Observed and Expected Excess Exposure to Selected Chemicals. Number of Angiosarcomas Showing Excess Exposure 0 1 2 3 4 Total Expected* Proportion 1/16 4/16 6/16 4/16 1/16 Calculated Irom the binomial distribution with n-4 and pi/2 Observed Frequency 6 6 2 1 5 20 Expected Frequency 1.25 5.00 7.50 5.00 1.25 20.00 Contribution to Chi Square 18.05 0.20 4.03 3.20 11.25 36.73 D < 0.001 - 1 I Journal of Occupational Medicine/Vol. 23, No. 5/May 1981 R&S 114747 '/ / / Fig. 1 -- Observed and expected exposure rank to vinyl chloride for each full year of employment matched by age, sex, race, year of employment and survival as a B. F. Good rich employee to January 1, 1974. ** DO (/> -si 00 siosarcoma Case II C29 controls) Fig 2. -- Observed and expected exposure rank to vinyl chloride for each full year of employment matched by age, sex, race, year of employment and survival as a B. F. Good rich employee to January 1, 1974. ngiosarcoma Case III SI 51 SI SI 51 SI SI SI SI SS SI SI 17 SI SI 71 71 71 7] Year of Employment (36 controls) R A N K 0 F T o T A L rvE X expected rank (19.0) p 0 5 U R Fig 3. -- Observed and expected exposure rank to vinyl chloride for each full year of employment matched by age, sex, race, year of employment and survival as a B. F. Good rich employee to January 1, 1974. $4 17 II 4) *1 *. As . , k i tc fi t/I Year of Employment 5 Exposure Indices for Epidemiological Surveillance/Greenberg and Tamburro r I f t Angiosarcoma Case IV \ I R&S 114749 R A N K 0 F T 0 T A L _^E- x p 0 s u R E o (Chemical Humber) W controls) t 'expected rank C3.0) Fig 4. -- Observed and expected exposure rank to vinyl chloride for each full year of employment matched by age, sex, race, year of employment and survival as a B. F. Good rich employee to January 1,1974. > (t 67 66 6? il *3 7; '2 7 .1 Year of Employment Fig 5. -- Observed and expected cumulative exposure rank sums for four angiosarcoma cases compared with matched sets of controls. (Matched by age, sex, race, year of employment and survival as a B. F. Goodrich Chemical Company employee to January 1, 1974. The numbers of controls for the four angiosarcoma patients were 4, 23, 29 and 36. The maximum possible rank sum is 96 and the minimum is 4. Expect^ rank sum is 50. A rank sum of 80 or would occur by chance 2.6% of the time^r rank sum of 20 or less would occur 2.6% of the time.) sarcoma subjects also showed significantly high expo sures to chemicals number 20 and 22. Chemical number 20 is a group of catalysts used only with vinyl chloride and chemical number 22 is hexane, used as a solvent for the group of vinyl chloride catalysts. In addition, the data also strongly suggest a high exposure to chemical number 11, diethyl maleate, also a specialized catalyst used only for a specialized process with vinyl chloride. A clear pattern of exposure is apparent in Table 2. All four angiosarcoma subjects showed excess exposure to 5 of 20 chemicals (numbers 5,11,16, 20, 22). By chance, this would be expected to occur for only 1.25 chemicals. All four angiosarcoma subjects also had less than expected exposure to six additional chemicals (numbers 2, 3, 4, 7, 14,21). The expectation again is only 1.25. When the bino mial distribution is inspected (i.e., the distribution of the number of successes in four trials with the probability of success at each trial being one-half), the results are statis tically significant (xf = 36.7: p < 0.001). These are summa rized in Table 3. Once more the null hypothesis of ran dom assignment of exposures is rejected For chemicals number 6 and 8, the exposure was tied for all members of group 4 and they were not included in the analysis. Figs 1 to 4 show the observed to the expected rank exposure for each employee with angiosarcoma by each full year of exposure. The pattern is one of consistent high exposure Journal of Occupational Medicine/Vol. 23, No. 5/May 1981 as compared to the matched controls. It is evident that the system does reflect the autocorrelation in jobs over Time-as'well as the exposure to vinyl chloride among the subjects with angiosarcoma. To remove the assumption of independence from year to year (as would be required in new field studies), the CERM were summed over years for each angiosarcoma subject and the associated controls. These then provided a single ranking for each matched group. The observed in dependent ranks of the four angiosarcoma subjects were then summed and compared to the expected rank sums in 5. The exact distribution of the rank sums was obtained by direct enumeration. The angiosarcoma subjects again show significantly high exposure to chemicals number 16 (vinyl chloride), number 20 (catalysts), number 22 (hexane used as a catalyst solvent), and number 11 (diethyl maleate). Fig 6 displays the vinyl chloride rank of the angiosarcoma subject in each matched group; this is the chemical through which empirical validation of the pro cedure is achieved. Prior knowledge of the etiology of angiosarcoma was used by the authors to validate these exposure indices. It is questionable what the situation would have beeadt such prior knowledge had not existed.-This study wc^B have shown that employees who subsequently developed angiosarcoma had had high exposure as compared to 357 1 <" I II ANGIOSARCOMAS III IV Fig 6. -- Vinyl chloride exposure rank of matched individuals with angiosarcoma. (Matched by year of birth, sex, race, year of employ ment and survival as a B. F. Goodrich Chemical Company employee to January 1, 1974.) matched controls for a set of four highly correlated chem icals selected from a larger set of 22 studied. These 22 had been selected because of potential toxicity or carcin ogenicity by the following criteria: known hepatotoxin, suspected carcinogen, degree of toxicity, degree of con tact, and location relative to vinyl monomer polymeriza tion areas. The chemicals would certainly be suspect, but because of correlated exposures, a determination of an exact cause-effect relationship would not be possible. It is certain that further studies would be initiated to inter pret the observed events. . The total cost incurred in setting up this system retro spectively (i.e,, the cost of going back into records cover ing plant operation from the beginning up to the present), including the standardization of 321 job classifications ($450), the application of A-D codes to work histories for 1400 employees ($4,760), the assignment of exposure rat ing to each job classification for each of 35 years for each of 22 chemicals ($880), the extraction of medical history data ($6,000), and computerization ($7,920) was $20,010, or $16 to $17 per employee. The prospective cost, i.e., the cost of work that is recorded regularly forward in time (not including work and exposure history from past times) averaged $3,281 per year, or $2 to $3 per employee, with a breakdown as follows: job classification (1 to 3 new jobs per year), $60; work history (240 new employees per year), $136; exposure indices (20 new chemicals per year), $160; medical data $1,425; and computer costs $1,500. Costs were determined on the basis of actual man-hours worked using industrial hourly wages of the individuals who performed the work. These data demonstrate empirically that a system of rank ordered individual exposure indices (CERM) for highly suspect chemicals can be implemented in an in dustrial environment and can identify a known causative relationship between exposure and the development of disease. This system can provide a means of monitoring industrial environments of any size at minimum costs. It can also allow for the introduction of sophisticated monitoring methods and prospective surveillance of an environment, taking into account co-factors which in fluence biological outcome. References 1. Creech JL and Johnson MN: Angiosarcoma of liver in the manu facture of polyvinyl chloride. / Occup Med 16:150.1974. 2. Creech II and Makk L: Liver disease among polyvinyl chloride production workers. Ann NY Acad Sci 246:88-94, 1975. 3. Greenberg RA, Tamburro CH, and Kupchella CE: Prospective medical surveillance program for detection and prevention of indus trially related cancer,' in Prevention and Detection of Cancer. Parts 1 and 2. H- Nieburgs (Ed.). New York: Marcel, Dekker, 1978. 4. Tamburro CH: Texas Reports, Environmental Cancer: A Report to the Public, 1978. 5. Tamburro CH: Chemical hepatitis, pathogenesis, detection and management. Med Clin North Am 63:545-566,1979. 6. Tamburro CH, Creenberg RA, Newby LG. and Turns DM: Im plementation and Assessment of a Demonstration Cancer Control Detection and Prevention Program in a Cohort of Industrial Workers. Final Report, NCI Contract #N01-CN-55212, November, 1978. 7. Tamburro CH: Medical surveillance for chemical hepatotoxicity, in Guidelines for Detectior. of Hepatotoxicity Due to Drugs and Chemicals, C. S. Davidson, C. M. Leevy, and E. C. Chamberlayne (Eds ). (DHEW-NIH Publication No. 79-313. October, 1979). pp 60-80. 8. Gamble | and Spirtas R: Job classification and utilization of com plete work histories in occupational epidemiology. I Occup Med 18.299-404. 1976. 9. Kerr PS: Recording occupational health data for future analysis. / Occup Med 20:197-203,1978. R&S 114750 358 Exposure Indices for Epidemiological Surveillance/Greenberg and Tamburro v