Document 855aV1q45pV60YeY5qEQYKnom

T.427 TRANSLATION . MORTALITY AND CANCER MORBIDITY IN WORKERS IN THE SWEDISH JVC-PROCESS INC INIinSTHY Bo Booberg, Stig Elofsson, Lars Holmlund, Rein Massing, Gustavo Molina and ?eter Westerholm* * Vinyl chloride has proved to he the cause of sclerodermia, Raynaud'b disease acro-osteolysis, liver damage and liver cancer (haemangiosarcomaX7) in workers exposed to vinyl chloride monomer (VCM). This has been proved in investigations carried out in businesses producing polyvinyl chloride (JVC) (5,8). Studies using animal.experiments have resulted in reports that inhalation of VCM has given rise to malignant tumours in different organs in rodents (6,9,18). In Sweden, in 1974, two cases of haemangiosarcoma of the liver were confirms in employees of a VCM and PVCTproducing company (2). Subsequently, two further cases occurred in the same factory. Studies (15,16) of different forms of 'cancer indicate that workers exposed to VCM in JVC-producing industries may run an increased risk of dying from forms of cancer other than haemangiosarcoma of the liver. Earlier, an increased rate of.mortality from heart/vessel diseases(l) was also observed in employees.of the JVC-producing industry. This retrospective cohort study was carried out with the purpose of surveying the pattern of morbidity and death in the FVC-proeessing industry. In general there is a lower level of exposure to VCM in the FVCprocessing industry than in the production industry. There are ca. 5,000 people employed in production in the FVC-proeessing industry. * MATERIAL AND METHODS For the investigation, information was collected from four FVC-proeessing enterprises. The four enterprises all use JVC which, following the addition of different chemicals, undergoes heat-treatment to produce, amongst other things, flooring, FVC-coated fabrics, pipes, and food packaging. e Collection of data The following information was collected from lists of personnel employed with the enterprises : t Staff number Name Start and end of exposure (year and month) Class of exposure. UCC 067074 -2- . The requirement for a pers n to be Included in the riginal' cohort 1b at least 3 m nths' employment, between 1945 and 31.12.74* Exposures are classified as follows : Class 3 (high) : working in the blending department; Class 2(medium) : heat-treatment machines; Class 1 (low) : other production departments. The collected data was transferred to punched cards and magnetic tape for statistical processing. The magnetic tape was compared, at the Central Office for Statistics (SCB),vith the country's total population and with the "death tape" tot the period 1961-76, and checked against the cancer register. The number of persons who could not be found vas checked at the State Tax Dept. In all, 2075 persons were included in the original cohort. 103 persona had dropped out (5 96) of whom 70 have gone abroad, 5 appear in the State Tax Dept.'s register of missing persons and 28 could not be traced. Study cohortb For statistical purposes, the resulting oohort of 1970 persons vas . divided into a number of sub-cohe ort#s* (study cohorts) : 1. All persons withat least 0monthsexposure.Follow-up period frombeginning ofexposure up to 1976 inclusive. 2. All persons with at least 6 monthsr exposure, excluding those who left before 1961. Follow-up period from beginning of exposure, but not earlier than 1961, up to 1976 inclusive. 3> All persons with at least 6 months'- exposure, beginning not earli r than 1961. Follow-up period from beginning of exposure up to 1976 inclusive. 4. All persons with at least 2 years' exposure. Follow-up period from two years after beginning of exposure, but not earlier than 1961, and up to 1976 inclusive, but not more than 10 years after end of exposure. 5. All persons with at least 2 years' exposure. Follow-up period from 10 years after beginning of'exposure, but not earlier than 1961, and up to 1976 inclusive. The last two study cohorts were selected in order to investigate whether there are any differences in the cause-of-death pattern with regard to how long after the beginning of exposure death occurs. The object of the first of the study cohorts is to shed light on the cause of death occurring relatively early - e.g. accident caused by the work. The object of the others is to shed light on any causes of death occurring after a fairly long period. For example, tumours caused by exposure at work often have a long incubation period, as a rule 5-10 years or longer. Statistical methods The final analysis vas carried out on the basis of two methods. In Method A the observed and anticipated number of deaths/cascs of cancer (cancer incidence) was calculated on the basis of the calendar year, and in Method B. the 'observed/anticipated annual risk of death vas calculated on the basis of the risk year. i tI ! j UCC 067075 - 3- Method A 'i For each person, a calculation'is carried out of the "annual contribution to the anticipated number of deaths" by dividing each individual's death expectation by a small part of each calendar year, in accordance with the formula : exy * ^xy px-1 ,y-1 ........ - Pa^ where e^ is the contribution to the anticipated number of deaths during year x at the age of X snd is the risk of death throughout the country for those who were X years old during year 3. The survival probability is p^ 1 - 0^. However, for the first risk year, the survival probability is calculated as pa^ = 1 - a ** can seasonably assumed that on average exposure began in the middle of the year. The individual er_y has been added up to E where B_x_y is the anticipated number of deaths during year x amongst those who are (or would have been) aged 3. Corresponding sums have been carried out for the observed number of deaths. By taking all. ages into account for a certain year, the expected/observed number of deaths that year is obtained. m As each annual total consists of a relatively small number, as time goes on, the years have been added to each other (accumulated) to give a better picture of the development. ' The accumulated annual sums show the developm n up to and including the year to which the sum relates. In order to facilitate comparison of different sub-groups, the accumulated sums have been converted to percentages of the number of persons included in the follow-up up to the present. The expected number of deaths from a specific cause, e.g. heart infarction, has been calculated by multiplying by the proportion of deaths from heart infarction in relation to all deaths throughout the country during year 3 at age 3. The numbers thus obtained have been added up for two periods, 196l--1968 and 1.969"1976. The reason for division into these particular periods 1b a thorough revision of the cause-of-death classification which was first applied in 1969. ucc 067076 To examine whether the total number of deaths deviates from the expected O figure to an extent too great to be attributed to chance, an X test has been used. The observed distribution of deaths amongst different causes was checked against that expected by means of an I? test. Calculations were carried out in the same way for the incidence of cancer, except that the following formula was -used t eiy * *x-1 ,y-1...........* * b * x-1,y I* * pb where ray, constitutes the risk of a tumour being diagnosed at age --y. during year x, * 1-r^ and p^'denotes survival probability. This somewhat more complicated formula takes into accountthe fact that a person's risk of cancer at a certain time is calculated with reference to the probability of his surviving up to that time. Method B The beginning of exposure or the time when the risk is seen as starting constitutes the starting point (time = 0) and time is calculated in risk years. The observed risk of death during a risk year (x,x + 1) is calculated in accordance with the formula 1 - p^, where px is the probability of surviving the current risk year, provided that the person concerned survived the previous year (ref M ; V - i/8d; il 4 a;2 * 4(x - Va-X) (.x + i/z v.) l2 In the formula, K is the number of persons alive at the beginning of risk year z, s^ is the number of persons who survived the risk year (= + 1), whilst w^ is the number of persons who dropped out of the follow-up during a risk year, as the closing date (31.12.76) occurred e during their risk year, d 1b the number of persons who would have dropped out of the cohort but died, and nX_ is the sum of wz and dx' . ucc 067077 -.5 - The expected risks for each risk year, have been calculated as the average value of the risks of deathmentioned in Method A above* since account vas taken of when the follow-up started. For a person whose follow-up Btarted in year a and who was years old, the risk during the first follow-up year was calculated as Qq - l/2 <1^ + l/2qa 1 ^h average risk during the first risk year vas calculated as the average value of all such risks q^. The development over the period is illustrated by calculating the accumulated probabilities of survival on the basis of the observed and expected annual probabilities. The formula * Pox = p0 * Pi*............*Px - 1 ** * expresses the probability of surviving from the beginning of the follow-up until a point in time x. If no discrepancy is * found betv, een observed and expected risk, the observed and expected risks for each risk year .should be equally great'. In order to illustrate whether there are changes over the period, the observed risk series haB been established by a 3-part sliding average, i.e. the risk for year 2 is the average value for the years 1-3* the risk for year 3 Is the average value of the risks for years 2-4 etc. The series established in this way has been divided by. the expected risks, year for year, thus obtaining a risk quota. If no 'discrepancies are found, thiB risk quota will be ca. 1, or, expressed as a percentage, 100. An increasing series indicates a risk increasing with exposure, and a series which is constant at .first, but after some years beginB to increase, indicates that the exposure must reach a certain level before there can be a question of an increase in risk. RESULT The original cohort was relatively young at the beginning of exposure. Age distribution into different classes of exposure is shown by Table 1. There are certain differences between the classes of exposure. In Class 1 (low), 41.7 % were younger than 55 years at the beginning of exposure, in Class 2 (medium), 47.7 9S. and in Class 3 (high), 50.6 %. The duration of exposure also differs according to the Class of exposure (Table 2). ucc 067078 It should however he noted that the table Includes cases still subject to exposure on the last, date Tor inclusion in the cohort (31.12.74)* which explains a certain tendency towards short periods of exposure. Notwithstanding this* Exposure Class 3 includes longer periods of exposure on average. The cohort as a whole shows up no noteworthy increase in the total risk of death compared with the average for the country, nor was there any indication of this in the sub-groups comprising the study cohorts. Study cohort 1, which includes all'those with at least 6 months' exposure, and with the risk calculated from the beginning of exposure, is somewhat remarkable insofar sb the expected number of deaths is significantly higher than that observed in 1964 (Fig. l). Further comments on this are contained in the discussion. Study cohort 2 (Tabl s 3 and 4* Fig. 2) includes persons with at least 6 months' exposure, excludin. those who left before 1961. The calculation of the risk is carried out from the beginning of exposure, but not earlier than 1961, and up to the end of the follow-up period (*1976). The number of deaths observed Is somewhat lower than expected - in Exposure Class 2 much lover. Classes 2 and 3 are relatively small' and with this type of analysis are subject to random developments. To prevent chance developments affecting the result, the classes have been combined; this applies to all the study cohorts. Table 4 shows the distribution amongst different causes of death. The observed and expected number of deaths during the period 1961-68 is relatively small - only a few cases, and as the cause-of-death classification was changed in 1969* as mentioned previously* the period I96I-68 iB not shown separately. The picture is broadly the same as for the period shown - 1969-76. Table 4*shows that the observed number of deaths, in particular from tumours of the digestive organs, heart infarction and accidents etc. is somewhat higher than expected. However the discrepancies are not significant. Study cohort 3 (Tables 5 and 6* Figs. 3-5) which apply to those who started in 1961 or later but who otherwise fulfil the same criteria as study cohort 2, shows a similar picture. UCC 067079 -.7 - An analysis of Study Cohort 3 in accordance with method B (Figs 4-5) indicates that the annual risk during -the first years of exposure is somewhat lower than expected , but after ca. 10 years there is an increase in risk, so that the observed risk is higher than expected. In study cohort 4 (Table 7) which relates to the period of continuous exposure or a relatively short time after the end of the exposure, i.e. the "short-term perspective", there is an Increased risk of death from heart infarction. In this case the risk from other causes is somewhat lower than expected. Finally, in study cohort 5 (Table 8) there is an indication of an increase in the riBk of death as regards tumours, but also heart infarction. Discrepancies between the observed and expected number is however n t statistically establishedat the 5 % level. The result with regard to mortality can be summarised as follows ; Taken as a whole, there is'no apparent, increase in mortality in the study cohorts. On the other hand, there are indications of a shift in the cause-of-death pattern as compared with the average for the country. This shift is chiefly expressed by the fact that the number of heart infarctions is evidently higher during continuous exposure or within a relatively short time after exposure ended. There is also an indication that the risk of death from tumours can be increased amongst persons, with a long incubation period (Tables 7 and 8). With regard to cancer morbidity there is no definite increase in study cohort 2 (Table 9 and Fig. 6). With regard to tumours of the digestive organs, 11 cases were observed, as against 8.3 expected. The differences are not statistically established. One of the 11 cases of tumour was cancer of the liver. * DISCUSSION A noteworthy fin* ding of `the analysis of the total mortality for the cohort (Fig. 1) is the fact that the number of deaths at the beginning of the observation period (1947-64) is significantly lower than that to be expected in relation to the average for the country. This IJCC * 067080 discrepancy is bo great that it eafmot simply be attributed to chance, nor can it be wholly ascribed to the so-called "healthy worker effect". Theoretically it is of course possible*that with regard to mortality and connected factors, the selected cohort differs considerably from the population of the country as a whole. However a more likely explanation ia that the list of persona available in the company concerned at the time of this investigation was incomplete with * regard to persons employed during an earlier period. A list of persons from which for example in the mid-1960s persons who left its employ before i960 were removed, may result in the above-mentioned discrepancy*. The company concerned has stated that, to its knowledge, no such "thinning out" was carried out. Should this "thinning out" have .taken place after all, this would have resulted in a loss of persons with a long observation period at the time of the follow-up. In this investigation the risk calculations have been restricted so that they begin no earlier than 1961. This means limiting the analysis to the group of surviving employees, beginning in 1961 so that the risk of a loss can certainly be eliminated. However this restriction means that the analysis is weakened, as sections of the cohort with long follow-up periods are excluded. This principally means that there is Icbs possibility of* detecting an increased incidence of cancer, should one exist. Heart-infarction mortality (ICD 410.90)'is increased in the cohort. This increase appears most clearly in the category of the total cohort with at least 2 yean* employment, where the analysis is concerned with the period following two years from the beginning of employment- and extending to a maximum of five years after termination of employment. It thus deals with those deaths from heart infarction which, with regard to time, occur relatively close to the period of employment. It is not possible on the basis of such observations to draw the conclusion that the increase iB caused by exposure to vinyl chloride. The observed increase in mortality from heart infarction * is however so narked, that in combination with known facts concerning vinyl chloride's toxic properties, it must be taken into account. ucc 087081 -9 - ucc 067082 There 1b no reason to assume that varying diagnostics, standards or practice In the filling-in of death certificates alone can provide an explanation. Disregarding the possibility of a chance local phenomenon, a natural conclusion is therefore that the increased frequency can be ascribed to either a selection of individuals susceptible to risk, or an outbreak of risk factors in the employee's immediate environment. In theory, a combination of these is, of course, also possible. In thiB connection it Bhould be noted that many risk factors in the case of heart infarction are connected with environment in the sense that they constitute part of the life style in the present-day social environment of an industrialised country. Cigarette smoking, physical * inactivity, overweight, high blood^lipid content thus constitute environmental factors which are`connected with social behaviour. It is a well-known fact that the risk of coronary heart disease varies according to the total riBk factors amongst other things. Other risk factors which say be mentioned include hereditary characteristics and e high blood pressure. In this connection it should be remembered that the causes of coronary disease include many factors, and in this respect the disease is connected with environment. It should also be borne in mind that the total risk increases when several risk factors, known or unknown, are combined C*5*17)- It has not been possible to establish -the distribution of such already know risk factors for coronary disease in the cohort examined in comparison with the population of the country in general. No continued analysis of the question of causal relationship between immediate environment and heart-disease morbidity can therefore be carried out within the context of this study. In this study, exposure classes 2 and 3 constitute sub-cohorts which are too small to give rise to meaningful discussion of heart-infarction risks in relation to different levels of exposure within the processing industry. In this connection account should also be taken of the fact that the exposure classes in the Btudy are based on interviews with employees concerned with the work environment between ten and fifteen years ago. It' thus deals with an environment which has since undergone changes. There are no obj ctive classification criteria with regard to exposure, e.g. in the form of environmental measurements. The . distribution into exposur classes is therefore impaired by uncertainty. % Animal testa have shown that the toxicity picture in rodents with . chronic exposure to VCM involves blood'vessels. In addition to hesmangiosarcoma of the. liver and other organs (6,9), inhalation of VCM also seems to lead to telangiectasis (6) of the liver in mice, n which may lead to death in haaoocoeleChanges in sinus cells have been observed in liver biopsies of VCM-exposed workers (14) Capillary changeb in the skin of the fingers have also been observed (10,11,12) both in VCM-exposed workers with other diseases involving vessels, such as acro-osteolysis, Raynaud's disease and selerodermia and in VCM-exposed workers without such diseases. An over-repreBentation of cases of death from heart/vessel diseases has also been observed in a study of the. FVCmanufacturing industry (l). Animal* experiments and earlier medical studies of people exposed to VCM thus support the assumption that the increased risk of heart infarction `Observed in this study could be ascribed to VCM exposure. With regard to tumour mortality and morbidity, the result is unclear. There are certain indications of an increase but the differences have not been established by statistics. Two possibilities can be considered here t 1. There is in reality no increase in the risk of tumours. 2. There is an increased rfsk of tumours in the works itself. The result neither confirms nor denies this. Tumours only appear after a long incubation period. The majority of the persons included in the study did not begin their exposure until the 60s or 70s, and could therefore not be followed up for a sufficiently long time. A careful follow-up of this cohort during the coming 5-year period should considerably clarify this question. In this connection it is interesting that in a recently published mortality study (4) of over 4300 deaths in the American FVC-processing industry there is an over^representation of cancer mortality (all cancer), and in particular cancer of the stomach/intestines in both sexes. ucc h 067083 SUMMARY Personnel lists were collected from four FVC-processing enterprises for production workers with at least 3 months* employment between 1945 and 31.12.1974. Of 2073 persons* 103 could not be followed up* e.g. because they had gone abroad.' The remaining cohort of 1970 individuals was analysed and compared with the population of the country with regard to mortality in different diseases and to cancer morbidity. The risk of death from heart infarction was higher in the cohort. This increase is shown most clearly in the sub-cohort with at least'2 years' exposure* where the analysis was concerned with events occurring during exposure or shortly after. The.risk of heart infarction in connection with vinyl-chloride exposure is discussed in relation to earlier studies of the effect of vinyl chloride on vessels. There is also an indication of an increased risk of morbidity and mortality in connection with tumours of the digestive Bys'tem. This has not been established by statistics. A future follow-up of this study is necessary to clarify whether there is any increased risk of tumours in the FVC-processing industry. Key words : Retrospective cohort study, vinyl chloride, FVC-processing industry* cancer, tumours of the digestive system, heart infarction. References : ucc 067084 Exposure Class Age _ -1? 20-24 25-2? 30-34 35-39 40-44 45-49 50-54 55-5? 60-44 65- No. Of persons 1 1.6 7.5 13.5 19.1 17.0 13.7 11.1 8.0 5.1 2.6 0.7 * 2 2.0 14.6 15.7 15.4 13.2 f2.6 12.6 7.8 3.9 2.0 0.3 3 8.9 14-.'3 21.4 17.0 10.7 9.8 8.9 5*. 4 1.8 1.8 0.0 100 X (1501 *t) ioo k 100 X (357 it) (112 t) 1-3 2.1 9.2 14.4 18.3 15.9 13.3 11.3 7.8 4.7 2.4 0.6 100 X (1970 t) 1 Table 1. Age distribution in the original cohort at the beginning of exposure (%). e % Jionths -5 6-23 24-5? 40-11? 120- Exposure Class 1 13.1 38.4 25.0 15.3 8.2 100 X '2 0.3' 8.4 *17.4 45.7 28.3 100 X 3" 0.0 8.9 10.7 15.2 65.2 100 X 1-3 10.1 31.3 22.8 20.8 15.1 100 X Table 2. Exposure-period distribution in'the original cohort (%). ucc 067085 Cohort Exp. kloss 1 Exp. klcss 2 Exp. kloss 3 Ho. No. of deaths Quota Approx. 95 % conf Obs. Em. interval 1303 356 112 53 w. .6 55.5 21.9 10.3 0.95 0.64 0.70 i 0.26 - 0.34 * 0.47 Exp. kloss 1-3 1771 73 87.8 0.84 - 0.19 Table 3. Expected and observed number of deaths as at J1.12.76 amongst those with at least 6 months' exposure, excluding those who left before 1961. Calculation of risk from the beginning of exposure, but not before 1961. Study cohort 2 (1771 persons). Malignant tumours 140-209 Tumours of the digestive system 1 JO-159 Heart and vessel diseases VII Heart infarction 410.90 * Accidents, suicide etc. XVII Observed 17 e 22 15 15 Exoected 14.0 4.9 24.3 10.0 9.2 Quota 1.21 1.63 0.91 1.49 1.42 Table 4. Expected and observed number of deaths from certain causes during the period 1969-1976 in those with at least 6 months' exposure excluding those who left before 1961. Calculation of risk from the beginning of exposure, but not earlier than 1961. Study oohort 2 (1771 persons). ucc 067086 Cohort Exp. Class 1 Exp. Class 2 Exp. Class 3 Ho. 1159 247 42 Ho. of deaths Obs. Exp. 45 41.2 . 4 11.7* 1 1.8 Quota 1.04 0.34 not calcul ated Approx. 95 % eorvf. interval + 0.31 + 0.34 not calculated Exp. Class 1-3 1428 . 48 54.7 0.88 + 0.25 Table 5. Expected and observed number of deaths as at 31*12.76 in those with at least 6 months' exposure beginning not earlier than 1961. Calculation of risk from beginning of exposure. Study cohort 3 (1428 persons). Malignant tumours 140-209 e Tumours of the digestive system 150-159 Heart and vessel diseases VII Heart infarction 410.90 Accidents, suicide etc. XVII 11 iO0 bserved 9 4 16 \A 11 Expected 9.7 5.3 16.2 11.2 7.3 Quota 0.93 1.20 0.99 1.25 1.51 Table 6 . Observed and expected cumber of deaths from certain causes during the period 1969-1978 amongst those with at least 6 months' exposure beginning 1981 or earlier. Calculation f risk from the beginning of exposure. Study cohort 3 (1428 persons). i. . Observed Expected Quota Malignant tumours 140-209 Tumours of the digestive system 150-159 Heart and vessel disease Heart infarction Accidents, suicide etc. XVII 5 (9) 7-4 (8.9) 0.68 (1.01) 2 (4) 15(16) 11(12) 4 (5) 2.6 (3.2) 12.7(15.8) 5.4 (6.6) 4.6 (5.1) 0.78 (1.27) 1.18 (1.01) 2.03*0 .82*) 0.87 (0.97) * p <0.05 Table 7. Observed and expected number of deaths from certain causes during the period 1969-76 amongst those with at least 2 years' exposure beginning not earlier than 1961. Calculation of risk from beginning of exposure. Study cohort 3 (1428 persons). UCC 067037 Malignant tumours 140-209 Tumours of the digestive system 150-159 Heart and vessel diseases VII Heart infarction 410.90 Accidents* suicide etc. X7II Observed 9. Exoeeted 6.0 4 2.2 12 11.1 8 4-5 2 2.5 Quota 1.51 1.85 1.08 1.77 0.79 Table 8. Observed and expected number of deaths from certain causes during the period 1969-76 in those with at least 2 years' exposure. The calculation of risk is carried out from ten years after the beginning of exposure. Study cohort 5 (680 persons). Observed Expected Quota Malignant tumours (total) Tumours of the digestive system (150-159) 51 44.6 11 8.5 1.14 1.29 * Table 9. Observed and expected number -of cancer cases during the period 1961-76 in those with at least 6 months' exposure excluding those who left before 1961. Risk*calculation from the beginning of exposure, but not earlier thap 1961. Study cohort 2 C**771 persons) A riff- 1, Accumulated proportion of deaths in %. Expected value calculated from beginning-of exposure. Study cohort 1 (1970 persons). The percentage for a given year has been calculated as 10q number of persons dying up to the given year incluslv number of persons beginning exposure up to the given year inclusive UCC 067038 1 rig._2 Accumulated proportion of deaths in %. Expected value calculated from 1961 inclusive. Study cohort 2 (l771 persons). The percentage for a given year,is'Calculated as in Fig. 1. (This also applies to Fig. 3 and Fig. 6, which relates to cases of cancer). I '* Fig. 3 - ----------- Accumulated proportion of deaths in % of those whose exposure began in 1961 or later. Study cohort 3 (1428 persons) * ucc 067089 7. ISO 100 ,--V/ so O t 7 3 4 S * 7 . I 10 11 II 1} 14 IS V.Vf| Fig. 4 The observed risk of death per risk year at different times after the beginning of expreseed as a percentage of the corresponding expected risk in those whose exposure began in 1961 or later. Study cohort 3 (1428 persons). --------- fcpOlM 100 t Fig. 5 Accumulated survival probability in % for those whose exposure began in 1$6l or later and who have at least 6 months* exposure. Study cohort 3 (1428 persons): Fig. 6 Accumulated proportion of cancer cases in %. Study cohort 2 (1771persons). Transl. RQ 12.7.79 ucc 067090