Document 10wKyDZJm85ONqyNZpnkVJXzE

ischbein, Blum Chntcal g work- terphenvh, naphunds. Amsterdam, fH'o capacitor-nal Intitule for Oklahoma upland (atfsttcai Analysis itomous response mg stable labora1057-67, ncer research Vo] on. International Sci publ No 32 ) British Journal of Industrial Medicine 1987;44:711-715 Exposure to vinyl chloride monomer: report on a cohort study AGNfiS LAPLANCHE,1 FRANQOISE CLAVEL,2 J-C CONTASSOT,3 GLAUDINE LANOUZIERE* From the Departement de Statistique Medicals,1 and Unite de Recherche 287 INSERM,2 Institut Gustave Roussy, 94805 Villejuif, Alochem,3 69190 St Forts, and Solvay et Cief 39500 Tavaux, France abstract In 1980 a prospective exposed/non-exposed cohort study was initiated in France by the Institut National de la Sante et de la Recherche Medicale (INSERM U 287) to evaluate the associ ation between mortality and cancer morbidity and occupational exposure to vinyl chloride mono mer (VCM). Eleven hundred VCM exposed subjects and 1100 VCM non-exposed controls matched for age (2 years), plant, and physician were included and followed up over a five year period for vital, health, and occupational status. The percentages of deaths observed among the exposed (1-8%) and non-exposed subjects (16%) did not differ. Eighteen (1-6%) and 15 (1-4%) cases of cancer were reported among exposed and non-exposed subjects, respectively (NS). One case of angiosarcoma of the liver occurred among the exposed group; six cases of lung cancer occurred among exposed subjects and two among non-exposed subjects (NS). The percentage of diseases of the circulatory system was higher (p < 0-02) in the exposed group than in the non-exposed group; this difference was explained mainly by the high incidence of Raynaud's disease (p < 0 006). The percentages of diseases of the respiratory system did not differ between the two groups. Several epidemiological studies of subjects occupationally exposed to vinyl chloride monomer (VCM) have been conducted.1"19 Although these studies differ in size and design, and cover different periods, they all conclude that the risk of some diseases, angiosarcoma of the liver (ASL) and acrosteolysis, for example, is increased by exposure to VCM. In 1980 a prospective exposed/non-exposed cohort study was initiated by the Institut National de la Sante et de la Recherche Medicale (INSERM U 287) in France. The main purpose of this study was to evaluate the association between mortality and mor bidity (especially from malignant tumours) and occupational exposure to VCM. Materials and methods The present study included most of the French VCM polymerisation plants. The exposed group consisted of 40-55 year old employees, either presently exposed or having been exposed to VCM. Foreigners were Accepted 13 October 1986 included if they had been living in France with their family for at least five years. Controls were employees never exposed to VCM. Each control was matched to one exposed subject for age (+ 2 years), plant, and physician. Interviews for initial data collection were conducted by each plant physician over a year. Data collected included identification, employment history (the entire work history was coded), occupational cat egory, and amount of exposure to VCM (defined for polymerisation functions as low since 1976. moderate between 1970 and 1976, and high before 1970), and medical history in addition to smoking and drinking habits. Each year until December 1985 the subjects were followed up for vital, health, and occupational status. Causes of death and specific abnormalities were coded according to the International Classification of Diseases (ICD, 9th revision). Sub jects who had terminated their employment (resigned, transferred, or retired) were followed up by each phy sician by post. The data were analysed using the PIGAS system.20 Control for confounding factors, such as cigarette smoking, alcohol consumption, and socioeconomic status, was performed using the Mantel-Haensze! procedure.21 711 712 Table I Distribution ofsubjects by plant Tavaux (Solvay) Saint'Fons (Alochem) Roussillon (Rhone-Poulenc) Aubervilliers (Rhone-Poulenc) Lavera (Atochem) Saint-Auban (Alochem) Bully-les*Mines (CDF Chimie) Ribecourt (Rhone-Poulenc) Port-de-Bouc (Atochem) Commentry (AEC Rhone-Poulenc) Montluel (Atochem) Antony (Rhone-Poulenc) Total No oj subjects ( %) 886(40) 656(30) 170(8) 124(6) 110(5) 84(4) 70(3) 30(1) 22(1) 18(1) 18(1) 12(1) 2200 Results From June 1980 to May 1981, 2200 employees were included: 1100 exposed and 1100 controls in 12 plants (table 1). Two plants recorded over 70% of the entire cohort. DESCRIPTION OF THE CHARACTERISTICS OF THE COHORT Most initial characteristics, smoking, and drinking habits did not differ among exposed and non-exposed subjects (table 2). A socioeconomic characteristic was defined. Laplanche, Clavel, Contassot, Lanouziere Employees whose occupational activity was coded as worker for more than 75% of their entire work history were considered as blue collar workers; the remainder were classified as white collar workers. These percentages did not differ between the two groups. DESCRIPTION OF EXPOSURE TO VUM At the time of the interview, only 36% of the exposed group was exposed to VCM, the rest having been exposed in the past. They represented the following occupations: CV synthesis (17%). polymerisation (54%), copolymerisation (7%), and compoundage and pharmacy (22%). The mean ( + SD) total duration of exposure to VCM was 14 years (8) and the mean time between the first exposure and inter view was 18 years ( 8). The amount of exposure hav ing been coded as low, moderate, or high, there were seven different combinations of exposure level during the entire work history of exposed employees. Tabic 3 shows the distribution of exposed subjects according to these combinations, the mean exposure duration at each level, and the time between the first exposure and the interview. FOLLOW UP OF THE COHORT The follow up of the two groups was similar. By 1985, 15% of the subjects initially included had retired, 2% had resigned or been transferred, and 4% were on sick leave. The percentage of subjects lost to follow up Table 2 Initial characteristics ofexposed and control groups Exposed 1n = 1100) Age (mean SD, years) Foreigners Urban residence* Smoking habits: Non-smokers Smokers Ex-smokers Duration since stopping smoking (mean + SD, years) Daily consumption (mean + SD)' Cigarettes Pipes Cigars Duration smoking (mean + SD, years): Cigarettes Pipes Cigars Inhalation Drinking habits: Non dnnkerst Weekly consumption (mean + SD)J Clinical signs of alcoholism Socioeconomic status: Blue collar workers!) 47 + 4 14% 46% 27% 42% 31% 8+7 16 + 8 5+ 5 4+4 23 + 8 16 10 8+8 62% 13% 32 + 23 15% 75% City of 15 000 inhabitants or more. tSubjects not dnnking wine, beer, fortified wines, or hard liquors. JSum of glasses of wine, beer, fortified wines, and hard liquors. Workers for more than 75% of their entire work history. Controls (n = U00) 48 + 4 14% 45% 24% 46% 30% 8+7 16 + 9 5+6 4+4 24 + 8 18+11 9+9 64% 12% 31 + 20 11% 72% P NS NS NS NS NS NS NS NS NS NS NS NS NS NS 0 003 NS it, Lanouciere wa|^^led as B work worKers; the lar workers, ecn the two ' the exposed having been he following lymerisation impoundage + SD) total irs (+ 8) and re and intcrxposure havl, there were level during yccs. Table 3 ts according e duration at xposurc and iar. By 1985. retired, 2% t% were on to follow up NS NS NS NS NS NS NS NS NS NS NS NS NS NS 1003 NS 1 Vinyl chloride monomer, cohort study 713 Table 3 Distribution of exposed subjects according to levels ofexposure to VCM Level* Duration of exposure* % mean SD (years) Time between l.\t exposure and interview mean + / \ean i 1 2 3 1 +2 1 +3 2+3 1+2 + 3 37 D = Dl = 11 + 7 13 D = D2 = 9 + 8 8 D = D3 * 9 + 7 14 D = 16 + 7 Dl =8 + 5 D2 = 8 + 6 10 D = 18 + 8 Dl = 10 + 7 D3 = S + 6 3 D = 20 + 9 D2 = 11+7 D3 = 9 + 6 16 D = 21 + 4 Dl =7 + 3 D2 = 7 + 4 D3 = 6 + 5 14 + 8 17 + 7 18 + 8 19 + 6 22 + 7 24-7 22 + 5 *1 = Low, 2 = moderate, 3 - high, tD = Tola! exposure duration, D! = level I exposure duration; D2 = level 2 exposure duration, D3 = level 3 exposure duration during the five years--that is, whose vital and health status was unknown--was 3%. These percentages did not differ between exposed and non-exposed groups (table 4), MORTALITY Oh THE COHORT During the five year follow up period, 20 (18%) exposed and 17 (1-6%) non-exposed subjects died. These percentages did not differ significantly. Table 5 shows the causes of death. Raynaud's disease, and other vascular diseases The only association found concerned Raynaud's disease, and other vascular diseases. The only association found concerned Raynaud's disease (p < 0 007). The percentage of diseases of the respiratory system (ICD: 460-519) did not differ between the two groups and no case of pulmonary fibrosis was recorded. These results were not materially altered by adjustment for tobacco smoking, alcohol status, and socioeconomic category. MORBIDITY OF THE COHORT During the five year follow up period, 18 (1-6%) and 15 (T4%) cases of cancer were reported among exposed and non-exposed subjects, respectively. These percentages did not differ significantly. The dis tribution of cancer sites is presented in table 6. One case of angiosarcoma of the liver occurred among the exposed group. Eight cases of lung cancer were reported, six among the exposed subjects (0-5%) and only two among the non-exposed subjects (0-2%); the percentages of lung cancer did not differ between the two groups. The percentage of diseases of the circu latory system (ICD: 390-459) was higher (p < 0 02) among the exposed group (10%) than among the control group (7%) (table 7). Six categories were considered: hypertension, coronary insufficiency, cerebrovascular disease, arteriosclerosis of the limbs, Discussion Subjects were included in the study if they were under 55 in order to allow a five year minimal follow up without too many withdrawals, as retirement usually takes place around age 55-60. Moreover, the subjects had to be over 40 for two reasons: firstly, to include only subjects who had been exposed to VCM for a sufficiently long time when levels of exposure were appreciably higher than at present and, secondly, because the incidence of malignant disease increases with age. Considering the percentage of retired, resigned, or transferred subjects, the percentage of subjects lost to follow up (3%) is low. This results from each physi cian's personal acquaintance with the employees under his care In this type of occupational study the Table 4 Distribution of subjects according to follow up status ________________________________________________ '_________________ Lost to follow up Resigned or transferred On sick leave Exposed (l 100) Controls (1100) ip 24(2%) 39(4%) NS 27(3%) 25(2%) NS 44(4%) 33(3%) NS Retired 155(14%) 163(15%) NS AS! 000019208 714 Laplanche. Clave!, Contassot, Lanouztere Table 5 Causes of death observed during five yearfollow up period* Exposed t 1100, Controls < 11001 All causes Cancer Circulatory system Cirrhosis Trauma Others 20 9 5 0 3 3 17 il 3 2 1 0 * No differences were statistically significant exposed group is exposed not only to VCM but also to various other agents with possible pathological effects. To control for these different exposures, the control group included workers whose exposure was similar to that of the exposed subjects, except for VCM. Exposed and non-exposed subjects were not matched by socioeconomic characteristics but the percentage of blue collar workers did not differ between the two groups; moreover an adjustment for this factor was made. The mean time between the first exposure to VCM and the interview was 18 years. This is important, because a long gap is necessary between exposure and the onset of disease, especially in the case of cancer, which may take eight to ten years to develop. Most surveys have shown a significant reduction in the mortality of VCM workers as compared with mortality in the general population; this is called the "healthy worker effect."171114 15 This study, how ever, as well as other epidemiological studies com paring exposed and non-exposed workers,4 5 8 found no difference in mortality between the two groups. Most studies clearly conclude that the target organs for VCM include the liver, the brain, the lung, and probably the lymphohaematopoietic system, the buc cal cavity, and the skin (malignant melanoma). Our results are not consistent with those previously reported. In our study no significant association between cancer morbidity and exposure to VCM was observed, and tumours at the following sites were Table 6 Causes of cancer observed during five year follow up period* (ICD No 140-209) Exposed f 1100) Controls II100) All cancers. Lung Digestive organs 18 6 5 Buccal cavity-pharynx Angiosarcoma of liver Pancreas Lymphohaematopoietic Bone Kidney Unknown 1 1 0 1 1 1 l No differences were statistically signfiicanl 15 2 6 3 0 1 0 0 0 2 Table 7 Percentages of circulatory and non-neoplaslic respiratory diseases observed duringfive year follow up period Respiratory system (460-519) Circulatory system (390-459) Hypertension (401-405) Coronary insufficiency (410-414) Cerebral vascular disease (430-438) Arteriosclerosis of limbs <4476. 4479,4402) Raynaud's disease (4430) Other circulatory disorders Exposal i noth Controls t 1100, P 3" o 10*', 4% 2% 1% 1% 1% 2% _0 7% 3*o SB' ()*', 1% 0% 2"'. NS 0`02 NS NS NS NS 0 007 NS recorded: angiosarcoma of the liver (one exposed), skin (one exposed, one control, both epitheliomas), buccal cavity (one exposed, three controls), lympho haematopoietic system (one exposed), and lung (six exposed, two controls). Although the number of cases of lung cancer was three times greater in the exposed than in the non-exposed group, the difference was not significant. In our study a higher percentage of diseases of the circulatory system was observed among the exposed group compared with the controls. This result is consistent with others.14711 The difference was explained mainly by the high incidence of Raynaud's disease but also by the addition of small differences observed in most of the vascular categories consid ered. whose physiopathology may be the same as for Raynaud's disease. A possible relation between exposure to VCM and the occurrence of hypertension or coronary insufficiency has been reported.1218 Although suggested by some authors,316 we found no association between exposure to VCM and respi ratory disease. We are indebted to. Drs Berrod and Aubrun. RhonePoulenc, 92408 Courbevoie, Dr Pierre: Solvay, 39500 Tavaux, Dr Rety: Atochem, 69190 St Fons, Drs Quelin and Richard: Rhone-Poulenc Petrochimtc. 38150 Roussillon, Drs Bourrtchon. Rambaud, and Bellec: Rhone-Poulenc Centre de Recherchcs, 93308 Aubemlliers, Drs Colonnc, Cestin, and Bennech: Atochem, 13117 Lavcra. Drs Ruty and Lemoine: Atochem. 04600 St Auban, Dr Lcveque: CDF Chimie-L'sines du Nord. 62160 Bully-lcs-Mines, Dr Baylac: Rhone-Poulenc Specialites chimiques. 60170 Ribccourt, Dr Lussato: Atochem, 13110 Port-deBouc. Dr Barrat- Rhone-Poulenc. 03600 Commentry. Drs Pirot and Favre: Atochem-Balan, 01120 Montluel, Dr Puech: Atochem, 38190 Brignoud, and Dr hazard for their contributions of data. AS1 000019209 Lanouziere pl<d '(Mil ,,A P NS oo: NS NS NS NS 0007 NS ; exposed), theliomas), ,). lympho id lung (six her of cases he exposed ice was not ases of the he exposed s result is rence was Raynaud's differences ies considani^^for 1 pert^Rion d18 a we found and respi- in: Rhonevay, 39500 ,, Drs Quemie, 38150 tnd Bellec: :s. 93308 Benncch: Lemoine: ^ue: CDF Mines, Dr ues, 60170 0 Port-deommentry, in, 01120 :noud, and Vinvl chloride monomer, cohort study Requests for reprints to: Dr Agnes Laplanche, Departement de Statistique Medicale, Institut Gustave Roussy, Rue Camille Desmoulins, 94805 Villejuif Cedex, France. References 1 Tahershuw 1R, Guffey WR, Mortality study of workers in the manufacture of vinyl chloride and its polymers. J Ouup Med 1974,16:509-18 2 Lihs R, Anderson HA Nicholson WJ, Daum S. Fivchbein AS, Selikoff IJ Prevalence of disease among vmvt chloride and poly vinyl chloride workers Ann \) AutdSa 19^5,246 22-41 3 Miller A, Teirstein AS, Chuang M, Selikoff IJ Changes in pul monary function in workers exposed to vinyl chloride and polyvinylchloride, drin VK,4ai(/5t< 1975,246 42-52, 4 Nicholson WJ. Cuyler Hammond E. Setdmam H. Selikoff IJ Mortality experience of a cohort ol vinyl chloride polyvinyl chloride workers, Ann S') Acad Sa 1976 246 225-30 5 Oil MG, Langner RR. Holder BB Vinyl chloride exposure in a controlled industrial environment a long-term mortality experience m 594 employees irch Health 1975.30.333-9 6 Wyatt RH. Kotchen JM. Hochstrasser Dl , viol, An epi demiologic study of blood screening tests and illness histories among chemical workers involved m the manufacture ol poly vinyl chloride Ann NY Acad Sa 1975.246 K0-7 7 Fox AJ. Collier PF Low mortality rates in industrial cohort stud ies due to selection for work and survival in the industry Br J Prcv 5o( Med 1976;30 225-30, X Waxweiler RJ, Stringer W Wjgoner JK. Jones J Neoplastic risk among workers exposed to vinvl chloride Ann \ ) Acad Su 1976,271 40-8. 9 Brady J, Liberatore F. Harper P, et ol, Angiosarcoma ot the liver an epidemiologic survey J\CI 1977;59`1383-5 10 Chiazze L, Nichols W'E. Wong D Mortality among employees of PVC fabricators J Ouup \ted 1977.19:623-8 715 11 Fox AJ. Collier PF Mortality experience of workers exposed to vinyl chloride monomer in the manufacture of polyvinyl chlo ride in Great Britain Br J Ind Med 1977:34-1 -- 10 12 Waxweiler RJ, Falk H, McMichael A. Mallov JS, Scott Grivas A. Stringer WT A cross-sectional epidemiological survey of vinyl chloride workers, Cincinnati Department of Health, Edu cation. and Welfare, 1977 (NIOSH publ No 77-177,) 13 Anderson HA. Snyder J, Lewinson T. Wood C, Lilis R, Selikoff IJ, Levels of CEA among vinyl chloride and polyvinyl chloride exposed workers. Cancer 1978;42:1560-7, 14 Equitable Environmental Health, Inc. Epidemiological study of vinyl chloride workers, (Final report submitted to Manu facturing Chemists Association ) Washington: EEH, 1978 15 Pierre C, Tasstgnon JP. Etude de la moriahle chez des iravailleurs exposes au chlorure de vmyle. Archives des Maladies Professtonnclles, de Medecine du Travail el de Secunte Sociale 1979,40 1131-45, 16 Theriault G. Allard P, Cancer mortality of a group of Canadian workers exposed to vinyl chloride monomer. J Occup Med 1981,23:671-6 17 Forman D, Bennett B, Stafford J. Doll R, Exposure to vinyl chlo ride and angiosarcoma of the liver, a report of register of cases, Brjlnd Med 1985:42 750-3 18 Greiser E, Reml W, Weber H, Vmylchlond-exposiiion und mortahtat Deutscher chemiearbeiter 1m vergleich 7ur mortahtai nichlexpomencrchemiearbeiter und PVC-verarbeiler. Prophvlaxe und Ergonomic 1982,32 44-62, 19 Storetvedt Heldaas S, Langard SL, Andersen A, Incidence of can cer among vinyl chloride and polyvinyl chlonde workers Br J Ind Med 1984;41:25-30 20 Wartelle M, Kramar A, Jan P, etal. PIGAS, an interactive statis tical data base management system. In- Proceedings of {he second international workshop on statistical database management. Los Altos- Laurence Berkely Laboratory, Statistics Canada. 1983,124-32 21 Mantel N, Chi-square tests with one degree of freedom, extension of the Mantel-Haenszel procedure. Journal of die American Statistical Association 1963:58 690-700 i AS I 00001 9210 British Journal of Industrial Medicine 1987;44:716-720 Correspondence Secondly, Soutar postulates that there is "an ele ment of restrictive lung defect in coalminers" since, in Update on lung disease in coal miners sir--In his editorial (1987;44:145-8) Soutar fortifies some of his arguments by quoting two papers that describe certain investigations carried out by my col leagues and me when I was associated with the Na tional Institute of Occupational Safety and Health. In doing so, he either quotes out of context or misin terprets several of our findings. Firstly, in his remarks concerning coalworkers' pneumoconiosis, he states that "irregular opacities are related to dust exposure ... and to an impairment of lung function." He supports this comment by quot ing, among other references, a paper by Amandus et al.' In reality, Amandus and his coworkers found that the two main factors associated with the presence of irregular opacities in coalminers were cigarette smoking and increasing age. A surrogate measure of exposure to dust--namely, years worked underground--was also associated to a far lesser de gree. Moreover, in non-smoking coal miners irregular opacities were not associated with impaired lung function. In virtually every investigation of the fre quency and presence of irregular opacities, whether in coalminers or other dust exposed populations, there has been an association between prevalence of irregu lar opacities and cigarette smoking. It is abundantly apparent that irregular opacities are found in many occupations other than coalmining, including work ers exposed to kaolin,2 man made mineral fibres,3 and silica.* In addition, they have been noted to occur in non-dust exposed groups including women and here again, cigarette smoking played a pre-eminent part.5 6 Certainly, one cannot attribute the presence of irregu lar opacities in non-smoking granite shed or man made mineral fibre workers to emphysema and fibrosis. No one has yet suggested that exposure to silica in the absence of conglomerate silicosis leads to emphysema.7 Surely coalminers are susceptible to the same influences and agents that lead to the presence of irregular opacities in other dust and non-dust ex posed workers. The power and validity of the associ the dust exposed groups, both the FEV, and FVC are reduced. Contrary to what he states, obstruction is often associated with a decrease in both the FEV, and FVC, but by contrast, the residual volume (RV) is always increased and the total lung capacity (TLC) is often increased.9 10 In true restrictive impairment, all lung volumes are decreased. An increased RV is an early sign of obstructive impairment and the increase occurs at the expense of the FVC.9'11 Later, the TLC also increases but to a lesser extent; when it does so it partially and temporarily masks the effects of the in crease in the RV. In the absence of a knowledge of all lung volumes the mere presence of a reduction in the FEV, and FVC cannot be taken as an indication that restrictive impairment is present. If Soutar rereads our paper he will note that one of the points we made was that the RV was appreciably increased in smok ers whether they had bronchitis or not.12 There was, in addition, a slight but significantly increased RV in non-smokers who had industrial bronchitis or what he terms occupational bronchitis (fig 4). Secondly, he will note that we did not suggest that the airflow lim itation occurs in the upper airways, although we be lieve that most of the obstruction is located in central airways. Indeed our exact words were "flows at low lung volumes were unchanged unless expressed either at absolute lung volumes or as a percentage of TLC, this being a reflection of an increase in RV." Finally, although he maintains that some coalminers showed severe airflow obstruction due to dust exposure, he concedes that most are smokers and yet attributes the airflow obstruction mainly to dust exposure. This statement is difficult to reconcile with the postmortem study carried out by Femie et al at the institute in which they were unable to find evidence of pulmonary hypertension or cor pulmonale in the absence of a history of cigarette smoking or PMF.13 Chest Diseases Unit, University Hospital, PO Box 5339, Postal Stn A. London. Ontario N6A 5A5, W K C MORGAN Canada. ation between coaldust and irregular opacities are relatively weak and multiple regression analysis be comes less reliable when there is collinearity between such factors as age, cigarette smoking, and exposure to dust. The association between them becomes even more tenuous when the interobserver variation in re gard to the reading of irregular opacities is taken into consideration.8 Although the paper quoted to indi cate the problem with wide interobserver variation originates from the United States, I have similar un-, published data from Britain. References 1 Amandus HE, Lapp NL,, Jacobsen G, Reger RB Significance ot irregular small opacities in radiographs of coalminers in the USA BrJIndMed 1976;33:13-7, 2 Wagner JC. Pooley FD, Gibbs A, et al Inhalation of china stone and china clay dusts: relationship between mineralogy of dust retained in the lungs and pathological changes Thorax 1986,41 190-6 3 Weill H, Hughes JM* Hammad YY. et al. Respiratory health in x workers exposed to man-made mineral fibers Am Rev Respir Dts 1983;128:104-12, " 716 lerjg^'an elelin^Aince, m 1, a^rFVC are obstruction i$ h theFEV, and olume (RV) is pacity (TLC) is impairment, all ,*ased RV is an nd the increase Later, the TLC ten it does so it ffects of the innowledge of all eduction in the indication that Soutar rereads mints we made 'ased in smok..12 There was, icreased RV in xhitis or what I, Secondly, he he airflow lim~ though we he ated in central "flows at low xpressed either ntage of TLC, RV." Finally, miners showed t exposure, he t attn^tes the This te p^Kortem he institute in ' of pulmonary absence of a V K C MORGAN tB Significance of coalminer' in the tion of china stone nineralogy of dust change* Thorax >piraiory health in rs. Am Rev Respir Correspondence 4 Theriault GP, Peters JM, Johnson WM. Pulmonary function and roentgenographic changes in granite dust exposure Arch Envi ron Health 1974,28.23-7. 5 Carilh AD. Kotzen LM, Fischer MJ. The chest roentgenogram in smoking females Am Rev Respir Dis 1973.107 133-6. 6 Weiss W. Cigarette smoking, asbestos and small irregular opac ities. Am Rev Respir Dis 1984:130:293-301 7 Sussman NB, Townsend MC. Mazumdar S Effects of multicollinearuy on inference in regression analysis of occupational respiratory epidemiologic data Am Rev Respir Dis 1986;(supp!) AI56 8 Amandus HE, Pendergrass EP. Dennis JM, Morgan WKC. Pneumoconiosis: interreader variabhty m the classification of the type of small opacities in the chest roentgenogram. Am J Roent Rad Therap\ Nuc Med 1974,122:740-3. 9 Levine G. Housley E, MacLeod P, Macklem PT Gas exchange abnormalities in mild bronchitis and asymptomatic asthma. N **/ J Med 1970:282-1277-80 10 Meisner P, Hugh-Jones P Pulmonary function in bronchial asthma, Br Med J 1968,i 470-2. 11 Thurlbeck WM, Henderson JA. Fraser RG. Bales DV, Chronic obstructive lung disease Medicine 1970:49 81-145. 12 Hankinson JL, Reger RB, Morgan WKC Maximal expiratory flows in coalminers. Am Rev Respir Dis 1977:116175-80 13 Ferme JM, Douglas AN. Lamb D, Rucklcy VA. Right ventricu lar hypertrophy in a group of coalminers, Thorax 1983.38 436**42, Dr Soutar replies: Morgan takes issue with some of my necessarily brief remarks on small irregular opacities in the chest radiographs of coalminers, and the type of lung functional impairment related to exposure to coal mine dust. It is not disputed that irregular opacities in coalminers are related to age and may be related to smoking. I quoted evidence that irregular opacities are related also to a surrogate index of dust exposure (years worked underground)1 2 and to direct mea surements of cumulative exposures to dust.3 Despite his doubts. Morgan associates himself with the paper by Amandus et at' in which the authors state "smok ing, age, and years underground were independent factors which each contributed significantly to the prevalence of irregular lesions." Inter-reader differences in the interpretation of the appearances of the chest radiographs are well recog nised,3 * but agreement between readers may still be sufficient under some circumstances to enable the demonstration of relations between certain radiographic appearances and other variables. Amandus et al (Morgan among the authors) studied inter-reader differences in recognising irregular opacities, and rec ommended that "epidemiological studies that depend on the differentiation of these types of opacity should involve a majority opinion from several interpreters rather than relying on one reader only. In addition the rigid training of readers periodically by the same stan dards should help to keep this variation within ac ceptable limits,"4 Observing these precautions Dick ei at confirmed a positive relation between irregular 717 opacities and lifetime cumulative dust exposure which is unlikely to be due to chance.3 Referring to the functional implications of irregu lar opacities, Morgan correctly states that Amandus et al showed impairment of function related to irregu lar opacities in smokers but not in non-smokers.' In addition to the other evidence I quoted of functional impairment related to irregular opacities,156 a new study confirms impairments of function related to ir regular opacities in coalminers.1 Most (86%) of the coalminers were smokers or ex-smokers, but results from 104 non-smokers included in that study were consistent with an opacity related impairment in these men also. Secondly, (Morgan's notation), of course re strictive functional defects cannot be identified with complete confidence on the basis of spirometric mea surements alone, though some patterns of abnormal ity can be suggestive. Our epidemiological studies have shown that dust exposure is related on average to a parallel reduction of FEV, than FVC, a pattern contrasting with the effects of smoking, which is re lated to a much greater reduction of FEV! than FVC, and corresponding reduction of the FEV,/FVC ra tio.8 The former pattern is also shown by the group of non-smoking miners with bronchitis (assumed to be "industrial bronchitis") described by Hankinson et al," Table 1 in their paper shows that FEV,, FVC, and FEV,/FVC ratio in these men were on average 0171, 0*181, and 0-3% respectively lower than in an age and height matched group of non-smoking min ers without bronchitis. By contrast, the FEV,, FVC, and FEV,/FVC ratio in a group of smoking miners with bronchitis were 0-431, 0-231, and 5-5% re spectively lower than in the non-smoking miners without bronchitis. Trying to understand these differences in conventional terms, I suggested in re lation to dust induced defects "an element of a re strictive lung defect as well as airflow obstruction." Morgan for his part suggested on the basis of these and additional measurements "that dust induced bronchitis is primarily affecting the larger or upper airways, and is not associated with concomitant de struction of lung parenchyma" (page 178, 3rd para, lines 14-17), while later in the article suggesting addi tional changes in small airways.9 To resolve these differences, further character isation of the functional effects of dust exposure is needed, and this will be available shortly. This new information may enable the disabling pathological le sions eventually to be identified by helping to direct the pathologist's attention to the relevant parts of the lung. Finally Morgan refers to my mention of a small group of coalminers who showed a severe inverse re lation between dust exposure and FEV,. Our reasons AS I 000019212 718 for attributing the decrement in FEV! to exposure to dust are explained clearly in the source paper10; and just as we are not yet clear on the functionally im portant pathological changes related to dust, we also have little basis on which to judge the relevance of Morgan's point on pulmonary hypertension or cor pulmonale. Incidentally, the study by Fernie et al in cluded only eight non-smoking miners without progressive massive fibrosis,11 too small a group to provide substantial support for Morgan's argument. References 1 Amandus HE, Lapp NL, Jacobson G, Reger RB Significance of irregular small opacities in radiographs of coalminers in the USA. Br J Ind Med 1976:33 13-7. 2 Cockcroft A. Lyons JP, Anderson N, Saunders MJA Prevalence and relation to underground exposure of radiological irregular opacities in south Wales coal workers with pneumoconiosis. Br J Ind Med 1983;40:169-72 3 Dick JA, Jacobsen M, Gauld S, Pern PO The significance of irregular opacities in the chest radiographs of British coal min ers. In: Proceedings of the VI International Pneumoconiosis Conference. Bochum, 1983. Geneva International Labour Office, 1984:283-99 4 Amandus HE, Pendergrass EP. Dennis JM, Morgan WKC, Pneumoconiosis: inter-reader variability in the classification of the type of small opacities in the chest roentgenogram Am J Roent Rad Therapy Nuc Med 1974,122:740-3 5 Cockcroft A, Berry G. Cotes JE, Lyons JP Shape of small opac ities and lung function in coal-workers. Thorax 1982,37:765-9 6 Musk AW, Cotes JE, Bevan C, Campbell MJ Relationship be tween types of simple coalworkers' pneumoconiosis and lung function. A nine year follow-up study of subjects with small rounded opacities. Br J Ind Med 1981:38.313-20. 7 Collins HPR, Dick JA, Bennett JG, et al. Irregularly shaped small shadows on chest radiographs, dust exposure, and lung func tion in coalworkers* pneumoconiosis. Br J Ind Med (in press). 8 Soutar CA, Hurley JF. Relation between dust exposure and lung function in miners and ex-miners Br J Ind Med 1986;43 307-20. 9 Hankinson L. Reger RB, Morgan WK.C Maximal expiratory flows in coalminers. Am Rev Resptr Dts 1977,116:175-80. 10 Hurley JF, Soutar CA. Can exposure to coalmine dust cause a severe impairment of lung function? Br J Ind Med 1986:43.150-7. 11 Fernie JM, Douglas AN, Lamb D. Ruckley VA, Right ventricu lar hypertrophy in a group of coalworkers Thorax 1983,38.436-42 Incidence of cancer among vinylchloride and poly vinylchloride workers: further evidence for an associ ation with malignant melanoma sir--Recently Heldaas et al presented the results of a cancer morbidity study in vinylchloride exposed workers (1987;44:278-80). They are suggesting a re lation between exposure to VC and the incidence of malignant melanomas. This suggestion is based on visual comparison of observed/expected cases in sub groups, stratified on levels of exposure or on years from first exposure (resp tables 1 and 2). A statistically significant trend between duration or Correspondence level of exposure and specific cancer morbidity may be tested according to Breslow et al. This test on significant trend was applied on the observed/expected cases of melanoma in relation to level of exposure (table I) or in relation to years from first exposure (table 2). The probability, that the dis tribution of observed/expected cases of melanoma in the subgroups of table I and 2 was due to chance, was respectively 65% and 22%. The only significant finding is the increase in the number of cases of melanoma in the total group of workers. Such an increase, however, was also found in cohort studies of workers with widely differing oc cupations. Moreover, in all other cohort studies of workers exposed to vinylchloride no increase of mor tality from melanoma was found.2 Therefore, the present epidemiologic evidence does not support a causative relation between the incidence of melanoma and exposure to occupational vinylchloride. w F TEN BERGE DSM. Central Department for Safety and Environmental Matters. PO Box 603, 6160 MH Geleen, The Netherlands. References 1 Breslow NE, Lubin JH, Marck P. Langholz B Multiplicative models and cohort analysis. Journal oj the American Statistical Association I983;78 1-12. 2 Nicholson WJ, Henneberger PK. Seiderman H Occupational hazards in the VC-PVC industry. Prog Clin Biot Res 1984.141.155-75 Drs Heldaas, Laugard and Andersen reply: Ten Berge seems to put a great deal of emphasis on the statistical inference of the results in our study.1 We are less concerned about the statistics; in our view the emphasis should be put more on the design of the study. We think that the quality of epidemiological studies in occupational medicine should be judged primarily by the characterisation of the level and du ration of exposure to the potentially harmful agent, and to what extent selection has been avoided, and whether confounding has been dealt with in a proper way. Occupational diseases with presumed long latency periods have also to be dealt with in a manner which takes care of this phenomenon. A small sample size is a common weakness in the design phase of small studies. Ten Berge, however, seems to interpret this problem of small sample size and failure to obtain "statistical significance" as being evidence for non-causality. It is not obvious from ten Berge's letter how he has tested statistical significance. ASI 000019213 rrespondence u/^B, tesi 1 on the i relation to 0 years from that the disnelanoma in chance, was rease in the tal group of s also found differing oct studies of ease of morerefore, the t support a f melanoma de. 1 TEN BERGE nmemal Multiplicative rtcan Statistual ly: tmphasis on our study.1 , in our view lesign of the iemiological i be judged evel and dumful agent, voided, and in a proper umed long in a manner kness in the e, however, sample size cc" as being >us from ten significance. Correspondence He has, however, avoided discussing the six cases of melanoma versus one expected, which we consider to be the most important result. As pointed out in the previous study2 studies of cancer incidence have ad vantages compared with studies of mortality, particu larly for tumour sites that are successfully cured. We feel confident that the methodological aspects have been properly considered in our study.1 Con sequently, we are less concerned as to whether the results are "statistically significant" or not. We con sider the fact that two cases of melanoma developed in the a priori high risk group selected from the same VCM exposed population before the performance of the previous study2 as being of more significance for the interpretation of possible causality than the pres ence of statistical significance. Development of new cases in a preselected risk group may be of greater significance in a causality discussion than is the pres ence of "statistically significant" results. As opposed to "statistical significance," a previously selected a priori high risk group cannot be manipulated. A recent study from Du Pont. USA, seems to sup port our finding, but apparently our results have not attracted the author's attention.12 The current lack of adequate animal and in vitro models for studying the development of malignant melanomas makes the study of the possible causal re lation between VCM exposure and this tumour com plicated. Even so, Maltoni el al succeeded in producing malignant melanomas of the skin in Syrian golden hamsters exposed to a wide range of VCM levels,4 and they concluded that the tumours were re lated to the exposure. References 1 Heldaas SS, Andersen A. Langard S. Incidence of cancer among vinyl chloride and polyvinyl chloride workers: a further evi dence for an association with malignant melanoma. Br J ind Med 1987;44278-80, 2 Heldaas SS, LangArd S. Andersen A Incidence of cancer among vinyl chloride and polyvinvl chloride workers. Br J Ind Med 1984.41,25-30. 3 O'Berg MT. Burke CA, Chen JL Walralh J, Pell S. Gallie CR Cancer incidence and mortality in the Du Pont Company an update, J Occup Med 1987.29.245-51 4 Maltoni C. Lefemine G. Gilbert! A. Cotu G. Carretti D. Carcinogenicity bioassays of vinyl chloride monomer, a model of risk assessment on an experimental basis Environ Health Perspect 1981:413-29 Pulmonary fibrosis in asbestos insulation workers with lung cancer SIR--Rudd's letter (1987;44:428-9) points to several interesting questions. (1) Histopathological pulmonary findings in asbes tos insulation workers with long exposure to the dust. I reviewed lung parenchyma sections submitted to me 719 by pathologists in various institutions in the United States and Canada. As expected, the material was taken from various parts of the lung, often unspecified. It is widely accepted that pulmonary asbestosis is characterised microscopically by diffuse in terstitial fibrosis caused by inhaled asbestos fibres. This was sought in my examinations in available parenchyma; Rudd argues that the subpleura] areas be excluded, a rather arbitrary preference in view of the absence of data that parenchyma here differs from parenchyma elsewhere; particularly since this part of the lung often has a considerable concentration of asbestos fibres, the result of "pleural drift" of inhaled dust. Parenthetically, lung cancer after asbestos in halation is not infrequently peripheral in origin. (2) I reported the presence or absence of micro scopically evident diffuse interstitial fibrosis. I also re ported that in 130 of 138 one or more asbestos bodies were seen. It is commonplace that in single thin (5/t) sections, asbestos bodies may be absent. Does this mean that they are not present (sometimes in large numbers) in the lung? Of course not. The likelihood of showing asbestos bodies varies directly with the amount of tissue studied. They are more readily seen in thick (30/t) sections than thin, more when bulk tis sue is digested and filtered than in sections, more when exposure has been to amphiboles than chrysotile, and more when long and thick fibres were inhaled than short and thin. It is thus not enough to say that so many or so many asbestos bodies are seen; one should immediately add comment regarding how much tissue was available for examination and how they were sought. Indeed, in our study, we did not expect to find so many cases (since, occasionally, there was only a single thin section available for re view) in which at least one asbestos body was to be seen. My colleagues consider that perhaps this was related to the diligent, prolonged examination to which each slide was subjected. (3) I was rather surprised to learn from Rudd that in the United Kingdom when there are "too few" as bestos bodies in slides of asbestos workers' lungs the diffuse interstitial fibrosis is not infrequently catego rised as cryptogenic fibrosing alveolitis. There must be a great deal of this disease nowadays in the United Kingdom among asbestos workers. And many asbes tos workers with diffuse interstitial fibrosis arc not eligible for disability compensation. (4) Asbestos bodies again. I am sure that Rudd is as familiar as we are with the multitude of observations that in lung tissue of asbestos workers with or without lung cancer there tend to be myriads of asbestos fibres uncoated by the iron protein material that allows them to be visible by (relatively insensitive) optical microscopy and thereby seen as asbestos bodies. Gen erally, the uncoated fibres are demonstrable only by AS I 00001 921 4 720 electron microscopy, after appropriate special prepa ration of the tissues. It could be--indeed, it is likely-- that in the few cases in which we did not observe asbestos bodies, uncoated asbestos fibres would have been seen if we had had an opportunity to examine an adequate tissue sample by electron microscopy. It is perhaps worth noting that we have yet to study an asbestos worker's lung in which we do not find un coated asbestos fibres (in much larger numbers than in the unexposed general population). (5) Rudd comments that if our histopathoiogical observations were to be considered evidence of asbestosis in asbestos workers with lung cancer in the United Kingdom many more would be eligible for compensation for lung cancer. Far be it for us to ad vise how and whether United Kingdom agencies should compensate asbestos workers who suffer lung cancer. We did not proffer such opinions when, until recently, lung cancer was not considered adequate reason for compensation and we will not do it now. Parenthetically, can Rudd tell us what proportion of workers with lung cancer, after important oc cupational exposure to asbestos, have been denied compensation because a board's histological/asbestos body criteria were not met? We appreciate Rudd's comments. They reflect im portant issues in the pathogenesis of lung cancer, not addressed in the bare data reported in our paper but useful as we nowadays grapple with concomitant is sues of prevention, compensation, and public health Correspondence control. Perhaps the letter reflects these: "I do not intend to suggest that the fibrosis seen in these cases was not caused by asbestos exposure." This acknowl edgement goes to the heart of the matter. Besides the theoretical pathogenetic connotations, which can be argued about, there are the practical questions of whether and how we might provide help for those found injured after inadequate protection in the past. Mount Sinai Medical Center, I Gustave L Lev\< Place, New York. NY 10029, USA. Y SUZUKI H KIPFN RUTH LIUS I J SELIKOFF Notice The Health and Safety Executive Library and Infor mation Service has indexed all articles relevant to health and safety at work in the British Journal of Industrial Medicine from 1944 to date. Approximately 1600 items are available in the HSELINE database, information on which is available on Pergamon Infoline Computer Services. European Space Agency Information Retrieval Services, Datastar, Compact Disc Read Only Memory. AS I 00001 9215