Document j7GeN4MVgpLan8npD25yM339

/f-'f British Journal of Industrial Medicine, 1977, 34, 1-10 Mortality experience of workers exposed to vinyl chloride monomer in the manufacture of polyvinyl chloride in Great Britain A. J. FOX* AND P. F. COLLIER Office of Population Censuses and Surveys, St Catherines House, IQ Kingsway, London WC2B 6JP and Health and Safety Executive,Employment Medical Advisory Service (EMAS) abstract Identification particulars were obtained for over 7000 men who were at some time between 1940 and 1974 exposed to vinyl chloride monome'r" in the manufacture of polyvinyl chloride. `Approximately 99% of these men have been traced and their mortality experience studied. The overall standardised mortality ratio, 75-4, shows a significant reduction compared with the national rates. Four cases of liver cancer were found. Two of these have been confirmed by a panel of liver pathologists as angiosarcoma and two as not angiosarcoma. There is no evidence to support the hypothesis that cancers other than those of the liver are associated with exposure to vinyl chloride monomer. The two cases of angiosarcoma were found in men who had been exposed to high con centrations of the monomer although the second man died only eight years after first exposure. The industry in Great Britain has expanded considerably since the second world war with over 50% of men having entered within the last decade. Conclusions drawn about the effect of vinyl chloride monomer on the mortality experience of men in this industry must consequently be tempered by the reservation that the full impact may not yet be in evidence. In 1974 three cases of angiosarcoma of the liver in workers at a plant manufacturing polyvinyl chloride (PVQ in the United States of America were re ported (Creech and Johnson, 1974). Further in vestigation by the firm led to the identification of two further cases who had died some years pre viously (Block, 1974). A programme of medical surveillance at the plant (Makk et al., 1974) resulted in the identification of two cases of angiosarcoma of the liver among current workers (Falk et al,, 1974). The rarity of this tumour may be judged from the incidence of reported cases: for the USA 20 to 25 cases a year (British Medical Journal, 1974) and for England and Wales some three to four cases a year (Baxter and Fox, 1975). Previously the only agents reported as associated with the development of angiosarcoma of the liver were arsenic (Regelson et al., 1968) and thorotrast (MacMahon et al., 1947). The first case of angiosarcoma of the liver in a man known to have been exposed to vinyl chloride monomer (VCM) in the manufacture of PVC in Formerly with EMAS Received for publication 19 April 1976 Accepted for publication 16 June 1976 Great Britain was reported by Lee and Harry (1974). The man had been a process worker for 20 years and he died in 1972 aged 71 years. At this time 19 cases of angiosarcoma in workers who had been exposed to VCM in the manufacture of PVC had been reported throughout the world. All these men had been ex posed to extremely high concentrations of VCM usually as autoclave `pot' cleaners. By 1975 some 32 cases had been reported all of which were in men who had been associated with the manufacture of PVC (Lloyd, 1975). These included a second British patient who had been exposed for four years and died aged 37 years only eight years after first expo sure. Early reports of epidemiological studies in the USA have been given by Monson et al. (1974), Tabershaw and Gaffrey (1974), Falk et al. (1974), Makk et al. (1974), and Nicholson et al. (1975). The Employment Medical Advisory Service (EMAS), with the assistance of company medical representa tives and the support of the unions, co-ordinated a prospective mortality study covering the manu facturing industry in Great Britain, The objectives of this study were to present a general picture of the mortality experience of workers engaged in the in- OCC 6154 2 A. J. Fox and P. F. Cottier dustry in Great Britain and to relate this to factors such as sex, age, year of entry into the industry, length of time exposed, and exposure levels. Methods The study covers all workers who may have been exposed to VCM through having been employed in this industry in Great Britain. All new entrants are being added to the population annually. Only four companies manufacture PVC and they have kept most of their personnel records at least since these processes were introduced.' The companies were ' asked to supply EMAS with'the names, sex, address (date at which pertinent), date of birth, year of entry into the industry, year of departure, job resulting in highest exposure, and maximum level of exposure-for 'each person who had entered their employment. As full job histories were not readily available for the population as a whole, it was decided to record the activity in which the maximum exposure was likely to have occurred and to follow-up individual cases, such as those dying with the mention of liver disease on their death certificate, to obtain a more complete job history. No measurement of the levels of exposure experienced by people employed before the mid-1960s was available. Estimates were made by the companies using the definitions: High Medium -- greater than 200 parts per million - between 25 and 200 parts per million Time - weighted average Low -- below 25 parts per million Constant -- most of the j time Intermittent -- occasionally. All data provided by the companies were sub sequently re-checked by them once they had been computerised. Throughout this analysis and dis cussion it is assumed that the information given to EMAS is accurate. It is also necessary to assume that persons not included in this analysis, either because they were not identified or because ade quate data were not available, do not constitute a biased sample. The number of deaths observed is compared with that expected using sex and age standardised death rates for England and Wales. Five-year age groups have been used for ages between 15 and 74 years. The Office of Population Censuses and Surveys (OPCS) provided death rates in accordance with the International Classification of Diseases (8th Revi sion) for the certified causes of interest and re-coded all death certificates for people in the study popula tion who died before 1968 (the year the 8th Revision was introduced). Death rates calculated for 1951, 1961, and annually after 1966 were applied to the person-years-at-risk for 1940-55, 1956-65, and annually after 1966. As 1974 death rates were not available, 1973 death rates were applied to 1974 person-years-at-risk. A comparison between ob served and expected values is made by assuming a Poisson distribution with mean equal to the ex pected value. Pearson and Hartley (1962) give the probabilities of individual terms of the Poisson dis tribution. For a mean greater than 15-0 the normal (gaussian) approximation to the Poisson distribution has been assumed. Results THE POPULATION These results cover the period between 1940 and 1974. Information relating to the deaths in people who entered in 1975 is not included nor is the fact that some people left during 1975. This information was made available during 1975 and will therefore be added to future analyses. Complete information on 7717 workers was obtained from the companies participating in the study. This was forwarded to the OPCS who notified EMAS of those whom they had traced together with the findings. Table 1 shows that 85 workers, only 11 % of the initial population, had not been traced at the time of analysis. This figure includes 59 workers at one factory for whom identi fying information was scanty. Another 72 workers (0-9%) who were traced were excluded because of inadequate information on their work history in the industry. A total of 409 deaths were recorded, 393 men and 16 women, and 148 workers emigrated before 1975. Two who emigrated returned during 1975 to live in Great Britain and will be included in future analyses. Male deaths have been studied in detail, but because so few women are involved in this industry their deaths are summarised in Appendix 1. Appendix 2 gives the deaths in 1940-74 notified since the analysis. Figure 1 shows the number of men employed in the industry each year from 1940 onwards and the number who had been previously employed. The work force in 1974 represents some 53% of men ever employed in the industry, reflecting the rapid growth since its inception. Table 2 shows that approximately 12% of men were exposed to high levels, almost all to constant high levels. The newness of the industry is reflected by the fact that only 8% of the workers have been employed in the industry for more than 20 years. Of these, only 34 men had been exposed to constant high levels. OCC 6155 i 1. d >t \ t> 1 ai ie | 5i Mortality experience of workers exposed to vinyl chloride monomer Tabic 1 Summary of the number ofpersons identified, traced, and flagged by the Office of Population Censuses and Surveys according to factory Factory (date of initial Humber identified Men U'omen Sot traced 2 (1944) 3 (1948) 4 (1940) 5 t--> 10 (1953) 11 (1948) 12 (1969) 13 (1970) Total population 397 1100 1540 41 440 2149 326 1568 m *h7561 67 -- 34 -- 5 22 7 21 V1 156 ___ -- 3 -- 59 ut 10 S3 (1-1%) * Deaths in men end women. tlncluda one penon who died abroad who has been excluded from the study. JFirst year in which more than 50 people were in current employment. Missing information 2 5 27 36 2 -- -- 72 (0-9 %) Deaths 1940-74 88 13 142 4 2 153 2 5 409 Emigrations 1940*74 13 34 _36 54 1 10 148 3 id le ct I >n ne II >n es le id at id l re ti- rs of he 93 ed I in in iis 1. sd in he he er 1h :ly all O' :rs 20 I to Fig. 1 Number of men in the industry and the total number who had left the industry for each year since 1940. OVERALL MORTALITY PATTERN Table 3 shows that only 393 deaths have been re corded as compared with 521-22 expected and the standardised mortality ratio (SMR) is 75-4. The SMR for cancers as a group is higher than this whereas the SMR for circulatory disease is approxi mately the same and that for respiratory disease, digestive disease, accidents, and suicides is below the overall SMR. It is difficult to identify angiosarcoma of the liver from death certificates (Baxter and Fox, 1975). Some cases will be classified as primary liver cancers {ICD no. 155), some as secondary or un specified liver cancer (ICD no. 197.7 or 197.8), and others may be in other categories (for example JCD no. 227). Only one death was certified as being caused by primary liver cancer compared with 0-71 expected. There were however three deaths, nearly 2i times the expected number, in these other categories. Histories for these four people are sum marised in Table 4. Slides for microscopical analysis obtained from these cases were referred with no Table 2 Levels of exposure and lengths of exposure for men included in the survey Length of exposure <years) Levels of exposure Constant Intermittent Total no. (%) High Medium Low High Medium Low 0-9 10-19 20 + Total (%) 605 142 34 781 (10-5%) 1202 117 92 1412(19-0%) 890 263 114 1266 07 1 %) 60 35 11 106 (1-4 %) 1857 210 130 2197 (29-7%) 1094 351 202 1647 (22-2%) 5708 (77-0%) 1118(151%) 383 (7-9%) 7409 (100%) Comtiot Intermittent High Medium Low Molt of the time. " Occasionally. -- > 200 pans per million. -- 200 parts per million > x > 25 pans per million. -- < 25 pans per million. } Time weighted average. l f ! i i j j OCC 6156 4 A. J. Fox andP. F. Collier Table 3 Observed and expected deaths, for 1940 to 1974, for all men covered by the survey, according to certified cause of death Couse of death (1CD* no.) 1940-1974 Observed Expected All cancers (MO-239) Cancer of stomach (121) Primary cancer of liver (125) Cancer of lung (162-163) Cancer of brain (191) Other liver cancers (197.7, 197.8, etc) Lymphatic and haematopoietic ti<ues (200-207) Endocrine, nutrition.l. nd met.boltc disuses <240-279) nitram of circulatory systems (39(M58) Ischaemic heart disease (48ffi4M) Discus of respiratory system (460-519) > Bronchitis (49<M91) Diseases of digestive system (220-277) Liver disease (270-573) Accidents (800-949) Suicide. (E930-939) All de.ths (1-999) 115 14 1 46 2 3 9 4 180 123 42 23 8 I -- 20 7 393 126-77 13 33 071 51-23 3-66 0-93 9-01 4-22 234-18 152*14 67-09 32-57 13-36 2-68 28-95 10-65 321-22 All dulhs hve been coded .ccording to the International Classification of Diseases, Sth Revision. SMX 907 91-3 140-8 898 34-6 322-6 99-9 94-8 76-9 80-8 62-6 70-6 57-4 37-3 69*1 63-7 75-4 Table 4 Deaths due to liver cancer Factory no* Date of birth Date of death Certified cause of death 2 27/10/1912 10/2/1970 Ascites; carcinoma of liverf and omentum 2 20/4/1901 23/12/1972 Haemangioendolhelioma of liver*; verdict, natural causes 2 15/6/1913 4/5/1973 Undifferentiated carcinoma of liverf* primary 11 2/6/1937 24/12/1974 Carcinomatosis; angiosarcoma of liver*; hepatic cirrhosis "Confirmed by panel of hisiopathologtsu u due to angiosarcoma of liver, tConfirmed by panel of hisiopathologists as not due to angiosarcoma of liver. Year joined 1961 1946 1921 1966 Year I'ft 1967 1966 1969 1974 Exposure levels Job description Low/inter* Fitter mittest High/constant Autoclave operator (1948-61) Medium/inter- Process roittent operator High/constam Polycleaner Table 5 Observed and expected deaths and standardised mortality ratio for all deaths between 1940 and f974 according to factory Factory no. Observed deaths Expected deaths SMR 2 3 4 5 10 11 12 13 Total population 77 13 137 4 2 153 2 5 393 86-67 55-63 163-17 5-19 16-29 176-90 4 35 13-02 521-22 88-8 23-4 84-0 77-1 12-3 86-5 45-0 38-4 75-4 details of their work history to a panel of liver pathologists (Baxter, 1976). A diagnosis of primary angiosarcoma of the liver was confirmed in two of them and the remaining two were confirmed as not being due to angiosarcoma. INDIVIDUAL FACTORY EXPERIENCE Table 5 shows that significantly fewer deaths were observed for men employed in factories 3,10, and 13 than would be expected from the numbers employed and their ages. In factory 10, 59 workers could not be traced and a further 36 were excluded from the analysis because of inadequate exposure information. Of these 36, three have died. The proportion of persons from this factory not traced is so much higher than for the other factories that it may include a disproportionately high number of deaths. Some early records for factory 3 were not accessible and these may also include a high proportion of deaths. The factories with the highest number of expected deaths (factories 2, 4, and 11) had similar SMRs (88-8, 84-0, and 86-5 respectively). The low number of expected deaths for factory 13 relative to the number employed is a reflection of its expansion during recent years. Three of the liver cancer deaths occurred in OCC 6157 Mortality experience of workers exposed to vinyl chloride monomer 5 factory 2 after 1966 (Table 4). This is significantly in excess of the expected 013 (p < 0 01). The number of deaths attributed to cancer, 21 observed, is marginally in excess of the expected 20-71, whereas the number from circulatory disease is below that expected (37 observed, 40-27 expected). Factory 4 shows a similar pattern with 41 observed cancers compared with 39-26 expected and 60 deaths at tributed to circulatory disease compared with 73-58 expected, as does factory 11 with 44 observed can cers, 43-14 expected, and 71 circulatory disease deaths compared with 80-08scxpected. .While these differences are not statistically significant they indi cate that in the three factories with the highest ex pected number of deaths, the SMR for cancer deaths was in excess of the SMR for all deaths and that for deaths associated with circulatory disease. Little further information was gained by analysing deaths by certified cause separately for each factory. Therefore in the following analysis all data have been combined. MORTALITY BY YEAR ENTERING THE INDUSTRY Table 6, by comparing cohorts who entered the in dustry in different quinquennia, shows that the ratio of observed to expected deaths is lower for those men who entered the industry recently. Figure 2 shows this to be the case for cancer and circulatory disease. The opposite trend is shown for accidents. Although the downward slope with succeeding year of entry is seen in Fig. 3 and Table 7 to be due to more recent cohorts having lower SMRs than earlier cohorts, the SMR for each cohort is seen to rise the longer the cohort is followed-up (Table 8). The SMRs increase from 40-5 for the first five years of follow-up, to 85-1 for between 15 and 20 years of follow-up, to 106,5 for 30 to 35 years of follow-up. MORTALITY BY EXPOSURE LEVEL The overall SMR in Table 9 is seen to be lowest in the group with the highest exposure, 67-8, and highest in the group with the lowest exposure, 78-0. The SMR for liver cancer was however highest in the group exposed to the highest concentration of VCM. The two liver cancers in this group were both angiosarcomas whereas the liver cancers in the low and medium exposure groups were not (Table 4). The only other high SMR in Table 9 is for brain cancer in the high exposure group. This results from one case being observed whereas 0-36 were ex pected. There is no indication of a dose-response relationship for cancer as a group, lung cancer, or diseases of the circulatory system and only a slight indication for diseases of the respiratory system where the SMR is increased from 59-7 to 64-4 to 69-1 with successive rises in exposure group; and only a slight and probably meaningless increase for diseases of the respiratory system. The SMR (Table 10) for those men who experi enced-constant exposures to VCM, 55-3, is signifi cantly below that for men who experienced only intermittent exposures, 84-1, The SMR for liver cancer is however higher in the former group. The two angiosarcoma deaths were men who had experi enced constant exposure (Table 4). This table indi cates an excess of lung cancers in men constantly exposed (17 observed, 15-72 expected, p > 0-05) which arises from men exposed constantly to medium concentrations (13 observed, 7-75 expected, p > 0 05). This group also had a slightly raised bron chitis rate (6 observed, 4-56 expected, p > 0-05). In Table 11 cohorts based on men not in the in dustry but who were alive 15 years after entry are subdivided according to the length of time spent in the industry. Fifteen-year survivors are considered in order to eliminate the biases that occur in a simple comparison with length of exposure. These arise because a man exposed for 15 years cannot have died before he completed this exposure, whereas a man who was exposed for less than five years may have died within five years of entering the industry. The mortality experience of the four cohorts con sidered in Table 11, men who entered between 1940 and 1944, between 1945 and 1949, between 1950 and 1954, and between 1955 and 1959 have been studied for 16 years, 11 years, six years, and one year respectively. Although there is a slight drop in the SMR in relation to the length of exposure for the 1940-44 cohort, there is an overall tendency for the Table 6 Analysts of observed and expected deaths by year of entry All deaths Year of entry 1940-44 1945-49 1950-54 1955-59 Observed Expected SMR 89 101*85 87*4 103 J 22-87 83-8 87 125*81 69-2 53 76*64 69*2 Six deaths in men who entered before 1940 are not included in this table. 1960-64 26 40-69 63*9 1965-69 19 30*84 . 61*6 1970-74 9 16 96 53-1 OCC 6158 6 A. J. Fox and P. F. Collier Table 7 Observed and expected deaths from all causes and SMRs according to year of death for cohorts based on year of entry to the industry Year of Year of death 1940-49 1950-59 1960-69 1970-74 194(M9 1950-59 1960-69 19W74 1 51-2 11*71 / 37 173-5 50-34 / 8 I 21'5 2912 /X'3 86 3 90 0 95-56 / 70 1- 72-6 96-39 / 13 ' 46 9 27-73 S 63 \ 34.g 74 26 / 61 \ 80-5 76-98 / 3437-78 /17133.,1 ,6 96 }33'' Six deaths in men who entered before 1940 are not included in this Ublc. Fig. 2 Standardised mortality ratios for all deaths, alt cancers, diseases of circulatory system, and accidents related to the year of entry into industry. Table 8 Observed and expected deaths in relation to length of time each cohort has been followed-up Length cf time cohort folionfd-up (ytars) Observed deaths Expected deaths SMR 0-4 5-9 10-14 15-19 20-24 25-29 30-34 20 49-32 40-5 so 86-44 57* 65 93-57 69-5 87 102-22 85-1 80 92-69 86-3 54 70-41 767 30 2118 106-5 Six deaths in men who entered before 1940 are not included in this table. Discussion Fig. 3 Standardised mortality ratios for deaths from all causes according to year of death for cohorts based on year of entry into industry. SMR to increase with increased exposure. The total SMR increases from 100-6 for those exposed for four years or less, to 104-7 for those exposed for five to nine years, to 113-3 for those exposed for 10 to 14 years. This industry has grown approximately tenfold since the second world war (Fig. 1). It is nearly twice as large as it was only 10 years ago. The environ mental conditions in the manufacturing processes are also known to have altered since men were first employed in the industry in the early 1940s. The main factors affecting the exposure levels were firstly, in the mid-sixties the discovery that exposure to high concentrations of VCM can cause acro-osteolysis and Raynaud-like phenomena which have since been classified under the heading vinyl chloride disease; and secondly, in 1974 the discovery that men ex posed to high concentrations of VCM had an in creased risk of dying from angiosarcoma of the liver. These changes in the size of industry and exposure mean that more than half of the people who have ever been exposed to VCM in the manufacture of PVC in Great Britain are currently employed in the industry; approximately 75% of men who have been employed in the industry have been employed for less than 10 years; only 8% have been employed for 20 years or longer; approximately half the persons who have been exposed to VCM have been intermit- I I i occ 6159 Mortality experience of workers exposed to vinyl chloride monomer Tabic 9 Observed and expected deaths according to whether exposure was thought to be high, medium, or low High Observed Expected SMR Medium Observed Expected SMR Low Observed Expected SMR All cancers 9 9-59 93 8 53 59*66 88-8 53 57*52 921 Liver cancer 2 0-13 1538*5 1 077 129*9 I 0*75 133*3 Lung ancer 2 3 67 $4 5 23 24*42 94*2 21 23*15 90*7 Brain cancer 1 0*36 277*8 --> 1*60 1 1*69 39-2 Disease* of circulatory system 12 16*79 71*5 85 110 59 76-9 83 106*81 77*7 Disease of ropiraiory system 3 4*34 69*1 21 32*62 64*4 18 30*13 59*7 All deaths 27 39*85 67*8 180 244-18 73*7 185 237-16 78 0 Exclude* one death in a man for whom an invalid measure of exposure was recorded. * -r> Table 10 Observed and expected deaths according to whether exposure was constant or intermittent Cause of death Constant Observed Expected SMR Intermittent Observed Expected SMR Alt canctn Liver cancer Lung Brain cancer Diseases of circulatory system Discasm of respiratory system All deaths 28 39*70 70*5 87 87-07 99-9 2 0*54 370*4 2 Ml 180-2 17 15-72 108*1 29 35*42 81-9 1 1-41 70*4 I 2-23 -- 36 69*82 51-6 144 164*37 87*6 n 17-82 61*7 31 49-27 62-9 89 160-94 55-3 303 360-25 S4-1 Excludes one death in a man for whom an invalid measure of exposure was recorded. 7 Table 11 Observed and expected deaths in relation to the length of exposure and year of entry for men who were alive 15 years after they entered Year ofentry Length of exposure (L4 years Observed SMR 5-9 years Expected Observed SMR J0-14 years Expected Obstnti SMR Expected 1940-44 {Deaths occurring in 1959-74) 1945-49 (Deaths occurring in 1964-74) 1950-54 (Deaths occurring in 1969-74) 1935-59 (Deaths occurring in 1974) Total 33 32-08 10 10-30 2 348 103 28 24*47 7 97 6*45 13 57*5 11*59 114 18 91 --* 19 83 2 13 4 2 109 l'04 3-85 4 1-37 8 2 112 136 587 M2 146 179 79 78*51 23 21*97 25 22-06 100-6 104-7 113-3 lently exposed; and only 10% of those whohave been constantly exposed, have been exposed to high con centrations. Thus relatively few persons have been ex posed for a long time to high levels of vinyl chloride monomer and even in cases in which men have completed 20 years of service, because their service has only recently been completed, the follow-up period is too short to evaluate the carcinogenic effect ofVCM. SELECTION AND SURVIVAL IN THE INDUSTRY This population comprises men who at one timesiDce 1940 were fit enough to obtain a job in the plastics industry. Because in general people entering em ployment are healthy their expected mortality rate should be lower than that observed for the popula tion as a whole with whom comparison is made. Table 6, in which the population was subdivided into cohorts according to year of entry into the industry. OCC 6160 8 A. J. Fox and P. F. Collier shows that the cohorts of men who entered more recently had lower overall mortality ratios than those men who entered long ago. Figure 2 shows that similar associations are observed for cancers and for circulatory disease. Figure 3 confirms that this association between SMR and year of entry into the industry is not a function of the cohort but is related to the length of time that it has been pursued and therefore to the length of time since the individual entered the industry. Table 8 shows that it may take 20 to 30 years for this effect to disappear. As this population comprises all men who have Jjccn identified as* Staving worked with VCM since the industry began in Great Britain in the early 1940s, jt includes men who have spent short periods in the industry as well as those who have had longer exposures. In any evaluation of the role duration of exposure plays in a model of cause and effect it is necessary to account for differences in character istics between those subjects who stay a long time in the industry and those who leave. Most of the latter leave for reasons unrelated to health, but there are some who will leave because they are not fit enough to continue working or because they consider the job to be affecting their health. It is difficult to separate this effect, the survivor effect, from the healthy population selection effect described above. Theywere both clearly described by Ogle (1885) but have not been measured. The 15-year survival effect for this population has been estimated by separating those persons alive 15 years after entering the industry into those working and those not working in the in dustry after the 15 years. The mortality patterns after these 15 years were then compared. The SMR for men not in the industry, 108-4. was approximately 50% higher than for those still in the industry, 74 0. In a separate report we have considered at greater length tfje implications of these effects and the effect of the period of follow-up on the interpretation of occupational mortality studies (Fox and Collier, 1976): they are of major importance in the interpreta tion of these data. Other studies Some hypotheses concerning the effects on man of exposure to VCM have been proposed as a result of the animal experiments conducted by Maltoni and Lefemini (1975) and the epidemiological studies con ducted by, for example, Tabershaw and Gaffrey (1974) and Nicholson el al. (1975). Maltoni and Lefemini (1975) have suggested that vinyl chloride monomer produced tumours in the three animal species studied: rats, mice, and hamsters. Liver angiosarcomas were observed in all three species. In addition other tumours that occurred included cerebral neuroblastomas and pulmonary adenomas. Tabershaw and Gaffrey (1974) confirmed the clinical suggestions by Creech and Johnson (1974) that there was strong epidemiological evidence that exposure to VCM caused angiosarcoma of the liver in humans. Their study also indicated an excess of cancer of the respiratory system and brain although this was not statistically significant. Nicholson el al. (1975) also confirmed that exposures to vinyl chloride monomer can cause liver cancer in humans. Although they observed three angiosarcomas of the liver and one brain cancer no deaths were due to lung cancer. Comparisons between the findings in Great Britain and those in the USA are limited because of the differing experiences in the two countries. If we apply the same population definition as employed by Tabershaw and Gaffrey (1974), we find 331 deaths observed and 462-29 expected. This gives an SMR of 72 compared with one of 75 found by Tabershaw and Gaffrey (1974). Men in the high exposure group have an SMR of 59 4 compared with one of 71-7 in the medium exposure group and 73-5 in the low exposure group. An unexplained trend similar to this was noted by Tabershaw and Gaffrey (1974). We believe the trend is due to selection and survival in the in dustry and to the length of follow-up. Using the same population definition as Tabershaw and Gaffrey we would have found two liver cancer deaths, one of which would have been due to angio sarcoma. Nicholson and co-workers in their study of all men in one plant with greater than five years' exposure followed-up 10 years from first employment, ob served 24 deaths compared with 19-0 expected, and nine cancers compared with 3-9 expected. Included in these nine cancers were three angiosarcomas. The last two columns in Table 11 show what might be termed a high risk group in that these are people who have experienced between five and 14 years' expo sure and have left the industry. They are studied here 15 years after first employment. In this group 48 deaths were observed compared with 44-03 ex pected. For cancer 13 were observed compared with 11-48 expected and for lung cancer six observed compared with 4-66 expected. It should be em phasised that none of the cases of liver cancer occur red in this high risk group. From Table 4 it can be deduced that a study in Great Britain similar to that conducted by Nicholson el al. (1975) would have identified one angiosarcoma and one other primary liver cancer. General assumptions and limitations Certain environmental factors are known to affect man's longevity. Some of these have been associated OCC 6161 ..oilier there ure to mans, of the is not ) also lomer i they d one :r. tritain of the apply ^d by leaths vIRof w and 3 have in the >osure s was 'dieve he in>g the and rancer angio- 11 nu^ aos^H t, OD^ 1, and ded in The ;ht be ewho expod here group )3 ex1 with .erved : em- 3CCUT- an be o that have imary affect ciated Mortality experience of workers exposed to vinyl chloride monomer 9 with particular causes of death such as cigarette smoking with lung cancer, heart disease, and bron chitis. Other associations such as those between social class and area of residence with cancers and bronchitis have not been fully explained. Epidemi ology does not provide the immediate answer be cause the cause and effect models are complex and individual susceptibility confuses the issue. In this particular study no account has been taken of these other factors. We have assumed that the social class, smoking, and geographical distributions are similar ,lo those for the*'hational population with which comparison is made. The mortality study provides a relatively crude epidemiological technique for evaluating the risk to health in the group under study. It cannot account for all the factors that may be relevant to the hypo theses that have been tested and its power is limited with respect to the detection of hazards which pro duce small excesses of mortality. Such studies can confirm relationships between exposures to specific chemicals and the subsequent development of un common tumours such as mesotheliomas, bladder cancers, and liver tumours. Unless the chemical con sidered is a powerful carcinogen these studies are less sensitive when studying common causes of death. Conclusions It is not yet possible to evaluate the overall impact of exposures to VCM on this population of workers but bearing in mind reservations discussed above two conclusions can be drawn at this stage. The first relates to whether or not there is evidence that exposure of men to VCM in Great Britain in the manufacture of PVC caused cancer. We have observed four liver cancers, two of which were sub sequently confirmed as angiosarcomas, significantly more than would have been expected. All these cancers occurred between 1970 and 1974 and similar experiences have been reported in the USA and in animal experiments. We therefore conclude that VCM is probably a carcinogen causing liver cancer in workers exposed to it. Although in this study there is little relationship with length of exposure, the angiosarcomas were associated with exposures to very high concentrations of VCM. The second conclusion relates to whether VCM causes cancers other than those of the liver. We find no evidence to support this suggestion. Only two brain cancers were observed and there is little indica tion of an excess of lung cancers. The SMR for cancer as a group is consistently higher than for all deaths but because of population selection factors this is difficult to evaluate. The effect of VCM on the mortality experience of workers exposed will only become clear after con tinuing this study for a further 10 years at least. Although the recently adopted hygiene standards may in future be shown to have a residual risk asso ciated with them, men who entered the industry in recent years should be at lower risk than those who were exposed at a time when conditions were Jess controlled. No study such as this could have been conducted without the assistance and co-operation of the in dustry and unions involved. Dr S. Gauvain was in volved with the planning of this project and an early draft of this report was discussed with Dr M. Greenberg, Dr A. M. Adelstein, and Dr P. J. Baxter to all of whom we are grateful. The EMAS Survey Control Section administered the survey, individual companies identified people who had worked in this industry, and the Office of Population Censuses and Surveys obtained and coded all death certificates. References Baxter, P. J. (1976). Epidemiological studies of PVC manu facturers and fabricators, and primary angiosarcoma of the liver. Proceedings of the Royal Society of Medicine, 69, 297-299. Baxter. P. J., and Fox, A. J. (197J), Angiosarcoma of the liver as the certified cause of death 1963-73. Lancet, 1, 2728. Block, J. B. (1974). Angiosarcoma of the liver following vinyl chloride exposure. Journal of the American Medical Association, 229, 53-54. British Medical Journal (1974). Editorial. British Medical Journal, X, 590-591. Creech. J. L., and Johnson, M. N. (1974). Angiosarcoma of the liver in the manufacture of polyvinyl chloride. Journal of Occupational Medicine, 16, 150-151. Falk, H., Creech, J. L., Heath, C. W,, Johnson, M. N-, and Key, M. (1974). Hepatic diseases among workers at a vinyl chloride polimerisation plant. Journal of the American Medical Association, 230, 59-63. Fox, A. J., and Collier, P. F. (1976). Low mortality rates in industrial cohort studies due to selection for work and survival in the industry. British Journal of Preventive and Social Medicine, 30, 225-230. Lee, F. I., and Harry, D. S. (1974). Angiosarcoma of the liver in a vinyl chloride worker. Lancet, 1, 1316-1318. Lloyd, J, W. (1975). Angiosarcoma of the liver in vinyl chloride/polyvinyl chloride workers. Journal of Occupa tional Medicine, 17, 333-334. MacMahon, H. E., Murphy, A. S., and Bates, M. I. (1947). Endothelial cell sarcoma of liver following tborotrast in jections. American Journal ofPathology, 23, 585-611, Makk, L., Creech, J. L., Whelan, J. G., and Johnson, M. N, (1974). Liver damage and angiosarcoma in vinyl chloride workers. Journal of the American Medical Association, 230, 64-68. Maltoni, C., and Lefemini, G. (1975). Carcinogenicity bioassaysjofvinyl chloride, current results. Annals ofNew York Academy of Sciences, 246, 195-218. Monson, R. IL, Peters, J. M., and Johnson, M. N. (1974). Proportional mortality among vinyl-chloride workers. Lancet, 2, 397-398. OCC 6162 10 A. J. Fox and P. F. Collier Nicholson, W. J. E-, Hammond, E. C., Seidman, H., and SelikofT, I. J. (1975). Mortality experience of a cohort of vinyl chloride--polyvinyl chloride workers. Annals of New York Academy of Sciences, 246, 225-230. Ogle, W. (1885). Letter to the Registrar General on the mortality in the registration districts of England and Wales during the ten years 1871-80, Supplement to the 145th Annual Report of the Registrar General of Births, Deaths and Marriages in England, p. xxiii. Pearson, E. S., and Hartley, H. O. (1962). Biometriea Tables for Statisticians, vol. I. Cambridge University Press: London. Regelson. W., Kim, U., Ospina, J., and Holland, J. F. (1968). Haemangio-endothelial sarcoma of liver from chronic arsenic intoxication by Fowler's solution. Cancer, 21, JI4-J22. Tabershaw, 1. R., and Gaffrey, W. R. (1974), Mortality study of workers in the manufacture of vinyl chloride and its polymers. Journal of Occupational Medicine, 16, 509-518. Appendix 1 Deaths in wqfiien Factory mr. Date Of birth Date of death a-- 2 2 2 2 2 2 2 2 2 2 4 4 4 (.12.1905 21-2.1899 29.1893 109.1894 17.10.1879 1X7.1897 13.11.1902 233.1921 1X73 894 1.7,1894 193.1909 8.103 890 22.83914 33 03 894 2X33965 4.7.1965 28.X1972 12303968 22,5.1950 833971 3.73974 5.8.1968 1130 1962 5.83952 15.12.1967 7.X1971 28.6.1949 M.ll.1972 Certified* cause of death 1CD code 5900.2901 4129 4129, 4271 9S00 4319 4409, 7824 4109, 2500 3949, 4500, 4530 4409, 4270 1579 1319 4340, 0389 4280. 4109, 5140 5810, 5932 Year joined 1944 1942 1944 1943 1940 1943 1944 1944 1942 1942 1942 1943 1941 1941 4 2X13895 73 23 948 5180. 7824, 4280 1940 4 73 3894 26.5.1966 5900. 4271, 4110 1943 Pximuy cause in italics. Year left 1945 1945 1945 1946 1946 1945 1945 1945 1945 1945 1945 1943 1941 1946 1940 1944 Exposure levels Job description Medium/imer* Process worker millent Medium/imer* Process worker mitteni Mcdium/imer- Process worker mitlenl Process worker Process worker Process worker Process worker Process worker Process worker Process worker Medium/imer* Process worker mitient Low/imermittcnt Mesi room attendant Worker with solvents Analytical laboratory attendant Analytical sampler Worker with solvents Appendix 2"-. Deaths in 1940-74 notified since analysis Factory no. Dote of birth Dote of death Certified* cause of death ICD code 4 15313904 28.8.1963 2500, 4850, 4339 4 213X1892 30.83955 9510 4 13,7.1901 31.33962 4000, 7920 10 7.73 906 15,12.1973 4109,4272 10 18,8,1925 7.7.1972 4309 11 533907 13,4.1972 5151. 4270, 4260. 5170 11 15,10,1900 30303960 1621, 4500. 4530, 4109 11 25.23892 10.93968 1500, 4500 11 9.73 902 8.83969 4360 Primary cause in italics. Year joined Year left 1953 1954 1954 1955 1940 1959 Not known 1971 Not known 1972 1955 1956 1951 1953 1951 1960 1957 1967 Exposure levels Job description Low/intermitient, Maintenance labourer Medium/imer- Maintenance mitient assistant Mediusn/consunt Electrician Low/jmermiitent Securuy/fixeman Medium/corutam Recovery/auto cleaner Medium/imer- Fitter mitient Low/imermiuent Grinder Medium/imer- Finer mitient Low/ifiiermitient Premia operator OCC 6163