Document jykjKRYLdpk52XNe5q7GNMvDQ

ST00G4902 An Historical Prospective Mortality Study of the Sarnia Division of Dow Chemical Canada Inc., Sarnia, Ontario (1950-1984) R.D. EGEDAHL, M.D.', G.W. OLSEN, D.v.M., Ph.D.1,2E. COPPOCK, B.Sc.3, M.L. YOUNG4, I.M.F. ARNOLD, M.D.5 We examined the mortality experience of 3.479 male Dow Canada employees who were employed at Sarnia Division for at least 12 continuous months during the years 1945 through 1983, utilizing the Canadian Mortality Data Base maintained by Statistics Canada, covering 1950-1984. We analyzed cause-specific mortality using male, age and calendar-year-adjusted death ratesfor Canada and Ontario. Total mortality was significantly below expectation whether the entire follow-up period (240 observed vs. 366.9 expected) or a 15-year latency period (171 observed vs. 290.4 expected) was considered. Statistically significant fewer observed deaths werefoundfor all respiratory cancer, cancer ofthe bronchus and lung, circulatory disease, ischemic heart disease, cerebrovascular disease, digestive disease, cirrhosis and other liver disease and deaths due to accidents, poisonings and violence. The observation of three deaths due to mesothelioma, a rare cancer often associated with asbestos exposure, was a significant finding as was a statistically significant elevation of observed deaths in the category "other forms of heart disease ". On a examine l experience de mortalite de 3,479 employis males de Dow Canada, quifurent employes a Templacement Sarnia Division pour une periode d'au moms 12 mois consecutifs pendant les annees de 1945 a 1983. La constation de mortalitefut obtenue en utilisant la base de donnees sur mortalite canadienne par Statistiques Canada et comprenait les annees 1950 a 1984. Les analyses de mortalite, cause specifique, furent accomplies en se servant des taux ajustes pour male, age, et annec civile a travers le Canada et la province de iOntario. On remarque une mortalite totale considerablement au-dessous de la prevision que Ton examine la periode enliere de suite (240 remarques contre 366.9 prevus), ou que Ton considere une periode de latence de 15 ans (171 remarques contre 290.4 prevus). Le nombre de marts remarquees forme un abaissement statisliquement significatifpour tout cancers respiratoires, cancers des bronches et du poumon, maladies circulaioires, maladies ischemiques du coeur, maladies cerebrovasculaires. maladies digestives, cirrhoseset autres maladies du foie, ei les morts dues aux accidents, empoisonnements et violence. La constatation de trois morts causees par un mesotheliome. un cancer rare souvem associe a 1'exposition a Tamiante.fut une decouverte significative: ainsi que fut une elevation siatistiquement significative de morts remarquees dans la categorie "autres formes de cardiopathies". I Corporate Director ol Occupational Health. Dow Chemical Canada Inc 2. Department of Epidemiology, Health and Environmental Science*. The Do* Chemical Company. 3. Research Analyst, Occupational and Enuronmenul Health Research Lnil. Statistics Canada 4. Corporate Occupational Health, Do* Chemical Canada. Inc. 5 Regional Medical Dirrnor. Dow Chemical Canada. Inc Any correspondence or requests for reprints should be sent to: Ronald D F.gedahl. M.D.. Do* Chemical Canada Jnc.. P O Box 10IZ Sarnia. Ontario. N7) 7K7 1 el.: (519) J39-49M. Canadian Journal of Public Health Concerns about possible long-term effects of occupational exposures1-' have led to large-scale mortality analyses of petrochemical worker populations in recent years. Studies of this nature may. I) identify workplace hazardous exposures. 2) demonstrate the effectiveness of health and safety pro grams. and 3) delineate further areas of research. We studied the cause-specific mortality experience of 3,479 male employees of the Sarnia Division of Dow Chemical Vol. 80, November/December 1989 441 Canada Inc. who had at least one year of service from 1945 through 1983. We compared the mortality experience among this employee cohort and the Canadian and Ontario male populations. Integral to this study was the Canadian Mortality Data Base (CMDR) developed by Statistics Canada which allowed for the identification of deaths since 1950.4 STUDY METHODS Description of the Sarnia Division The Sarnia Division chemical manufacturing site began operations in 194S. The Sarnia Division facility is a major Canadian producer of chlorine, sodium hydroxide, chlorinated solvents, styrene, vinyl chloride monomer, epoxyresins, vinyl ester resins (Derakane"), propylene oxide, polyols (VoranoP). latex, plastics and glycols. Cohort Identification and Definition A Vital Status Registry (VSR) was established to assemble information on all active employees, retired pensioners, transferred and terminated workers who were employed for a minimum of one continuous 12-month period between 1945 and 1983. The completeness of the assembled cohort for employees since 1960, checked by using the annual departmental employee census lists as a second independent data source, was found to be 100%. Prior to I960, there were no available independent data sources that could be used to check for cohort completeness. The study cohort comprised all male Dow Chemical Canada Inc. employees of the Sarnia Division who worked for at least 12 continuous months between January' I, 1945 and December 31,1983 and, if employed prior to 1950, were still actively employed as of January 1, 1950. The mortality experience of the study cohort was examined during the period 1950-1984. Trademark of The Dow Chemical Company Currently, approximately 20%- of the workforce is employed in administrative or office positions. Plant workers comprise 80% including salaried engineers and chemists as well as hourly maintenance and operator personnel. Maintenance workers are employed as machinists, insulators, painters, pipefitters, boilermakers and carpenters. Histori cally, the percentage of plant workers has been at least as high as the current levels. Canadian Mortality Data Base The record linkage facilities of Statistics Canada were utilized to search the Canadian Mortality Data Base(CMDB) for deaths in the study cohort. The CMDB is a computer file maintained by Statistics Canada containing information derived from death certificates collected by the provincial vital statistics registries.4 It is frequently used for epidemiologic investigations. At the time of this study, death data were available from 1950 to 1984. The mortality * ascertainment procedure involved the computer-matching of ^ the 3,479 males in the study cohort with the CMDB file. For ^ those found deceased, the date and underlying cause of death were abstracted from the CMDB. For each study cohort death, a copy of the death certificate was made and assigned a (--t standardized cause-of-death by Statistics Canada nosotogists ^ based upon the Manual of the International Classification of Diseases Adapted for Use in the United States, Eighth Revision (ICDA-8).5 Analysis The analysis used for the study data involved the calculation of a Standardized Mortality Ratio (SMR) for the various causes of death. Monson's computer program6 has been expanded in accordance with the 174 codes of death developed by the Laboratory Centre for Disease Control. Health and Welfare Canada.7 Person-years for the cohort members were accumulated across five-year age and Age < 20 20-29 30-39 40-49 50-59 to* Total TABLE I Distribution of Employees by Age and Year of Onset of Person-Years Exposure With Person-Years Contribution to the Study in Parentheses 1950-1959 157 (4.7981 719 122.05K) 353 (10.644) 100 (2.697) 16 (311) 5 (10?) 1.350 (40.6111 Year of Onset of Person-Years Experience 1960-1969 1970-1979 136 (2.650) 670 (13.412) 129 (2.640) 36 (705) .1 (46) 0 (0) 974 (19,453) 206 (2,097) 621 (6.447) 96 1942) 19 (198) 4 (41) 0 (0) 446 (9.725) 1980-1983 24 (107) 144 1613) yy (146T ) (281 1) 10) 1 (4) 209 IMS) 442 Canadian Journal of Public Health Total 523 (9,652) 2154 (42.530) 61) 114.372) 162 (3.6281 23 (398) 6 (107) 3.479 (70.687) VoJ. 80 *106*19001 TABLE II Observed (N > 2) and Expected Deaths*, SMRs and 95% Confidence Intervals by Cause for the Sarnia Division Male Employees (1950-1984) Analyzed for the Entire Follow-Up Period Cause of Death (ICDA-Sth Revision Codes) Obs All Causes (000-999) Infective and Parasitic Disease (000*136) Neoplasms (140-239) Pharynx Cancer (146-149) Dijcsiive Cancer OJO-159) Esophagus(150) Siomach (ISO Large Intestine and Rectum (153.154) Liver and Gallbladder (155,156) Pancreas (157) Respiratory Cancer ()60* 163) Bronchus and Lung (162.1) Pleura (163.0) Malignant Melanoma (172) Genital Cancer (I8S-187) Prostate (185) (frinarv Cancer (188,189) Kidney 1189.0. 189.1, 1X9.2) Bladder(188) Brain and Other Nervous Svstcm (191.192) Brain (191) Lymphoid Malignancies (200*203) Non-Hodgkin's Lymphoma (200,202) Multiple Myeloma (203) Leukemias (204-207) Neoplasms. Secondary. Ill Defined and Benign (195-I99.2IU-239) Endocrine, Nutritional and Metabolic Diseases (240-279) Nervous System and Sense Organ Disease (320-3891 Multiple Sclerosis (3401 Circulatory' Disease (390-458) Hypertensive Heart Disease (400-404) Ischemic Heart Disease (4)0-413) Other Forms of Heart Disease (420-429| Cerebrovascular Disease (430-438) Diseases of Arteries, Arterioles and Capillaries (440-448) Respiratory Disease (460-519) Pneumonia (480-4X6) Chronic Bronchitis. Emphysema and Asthma (490-493.319.3) Digestive Disease (520-577) Cirrhosis and Other Liver Disease (570-573) Genitourinary Disease (580-629) Nephritis and Nephrosis (5R0-5R4) Musculoskeletal Disease (7)0-738) 111 Defined Symptoms (780-7%) Accidents, Poisoning and Violence (800-999) 240 2 63 2 21 3 A 6 2 6 15 It 3 3 2 2 4 2 2 4 4 5 % 2 3 4 4 3 2 HO 2 86 8 9 4 to 5 5 4 4 2 2 3 3 33 *Based on male Canadian and Ontario Mortality rates. Exp 366.9 3.8 80.9 1.2 24.3 1.8 5.9 9.8 1.4 4.4 26.9 24.8 0.1 J.l 4.2 3.4 3.9 2.1 1.7 3.7 3.2 5.2 2.X 1.0 3.4 3.4 b.l 4.9 0.7 148 2 14 107.8 27 17.2 58 17,4 5.8 8.6 18.2 11.2 3.9 2.1 0.7 3.1 74.0 Canada SMR 65 52 78 160 89 164 68 62 143 136 56 44 2073 270 48 59 102 94 117 108 126 97 mx 193 X7 102 66 6) 2X2 74 140 80 29X 52 68 57 87 5* 22 36 50 95 414 97 45 95% C.I. 57- 7* 6- 189 60- 100 19- 600 53- 132 33- 479 18- 174 22- 134 16- 516 50- 295 31- 92 22- 79 417-6058 54- 789 5- 173 7- 212 2X- 262 11- 341 13- 422 24- 277 34- 323 31- 226 22- 316 22- 698 IX- 255 27- 260 IK- I6K 12- 179 32-1019 61- 89 16- 506 64- 98 128- 587 24- 99 58- 175 27- 106 28- 203 19- 137 6- 56 10- 91 6- 182 It- 342 83-1210 20- 285 31- 63 Exp 357.7 2.5 79.7 1.3 23.9 2.0 5.1 10.7 1.5 4.2 26.4 24.3 0.1 1.4 4.2 3.3 3.9 2.0 1.8 3.8 3.4 5.0 2.7 1.0 3.5 3.4 s.x 4.6 0.6 153.7 1.2 1(5.7 2.0 17.3 6.2 16.8 6.3 8.0 18.1 11.6 3.8 2.1 0.7 2.4 63.5 Ontario SMR 67 80 79 155 88 148 78 59 135 142 57 45 2738 215 48 61 102 98 112 106 117 too 112 192 86 117 95% C.l. 59- 76 9- 288 6(- )0I 17- 559 54- 134 30- 434 21- 200 21- 128 15- <88 52- 308 32- 94 23- 81 550-8000 43- 628 5- 172 7- 2)9 * 27- 261 II- 353 13- <04 29- 272 32- 300 32- 234 22- 326 22- 693 17- 253 31- 299 69 19- I7X 65 13- 190 348 .19-1255 72 59- Rf. 163 18- 587 74 59- 92 397 171- 782 52 24- 99 65 17- 166 59 28- 109 80 26- 186 63 20- 147 22 6- 57 34 9- 88 52 6- 1X9 96 11- 346 4.U 87-1267 123 25- 366 52 36- 73 C.O calendar-year specific categories from the date of their entry into the cohort (one year after their date of hire) until their date of death or study termination, whichever occurred first. The expected number of deaths was computed by multiplying the study cohort's age, sex and calendar-year stratum specific person-year estimates and the corresponding Canadian or Ontario cause-specific mortality rates (1950-1984) together and summing the products across strata. SMRs were calculated and reported for all causc-of-death categories in which there were two or more observed deaths. Approximate 95% confidence intervals were calculated7-* and statistical significance was considered when the null value, 100, was not included in the interval. RESULTS Descriptive The Sarnia Division study cohort comprised 3,479 males who accumulated 70,687 person-years between 1950 and 1984. Table 1 is a distribution of the 3,479 employees by age, onset of person-years experience and distribution. November/December 1989 443 STOOGif 905 TABLE HI Observed (N > 2) and Expected Deaths *, SMRs and 95% Confidence Intervals by Cause for the Sarnia Division Male Employees Excluding Person-Years and Deaths Within 15 Years of Hire Cause of Death (ICDA-8ih Revision Codes) Obs All Causes (000-999) Infective and Parasitic Disease (OOCM36) Neoplasms (140-239) Pharvnx Cancer < 146-149) Digestive Cancer (150-159) Esophagus(150) Stomach (151) Large lntesiine and Rectum (153,154) Liver and Gallbladder (155.156) Pancreas (157) Respiraiorv Cancer (160-163) Bronchus and Lung (162.1) Pleura (16X0) Malignant Melanoma (172) Urinary Cancer (188-189) Kidncv (189.0. 189.1. 189.2; Brain and Other Nervous System (191.1921 Brain (191) Lymphoid Malignancies (200-203) Non-Hodgkin* Lvmphoma (200.202) Leukemias (204-207) Neoplasms. Secondary. Ill Defined and Benign (I95-I99.2IO-239) Endocrine. Nutritional and Metabolic Diseases (240-279) Nervous System and Sense Organ Disease (320-389) Circulatory Disease (390-458) Ischemic Heart Disease (410-413) Other Forms ol Heart Disease (420-429) Cerebrovascular Disease (430-438) Diseases of Arteries. Arterioles and Capillaries (440-448) Respiratory Disease (460-519) Pneumonia (4&CM86) Chronic Bronchitis, Emphvsema and Asthma (490493 .519.3) Digestive Disease (320-577) Cirrhosis and Other Liver Disease (570-573) Genitourinary Disease (580-629) Nephritis and Nephrosis (580-584) Musculoskeletal Disease (710-738) 111 Defined Symptoms (780-796) Accidents. Poisoning and Violence (800-999) |7| s 48 2 17 3 3 4 2 5 II 8 2 2 3 2 3 3 3 2 2 4 3 2 82 6b 7 6 2 9 5 4 3 3 2 2 3 2 12 Based on male Canadian and Ontario Mortality rates. Exp 1.8 70.0 l.l 20.9 1.7 4.8 8.5 1.2 3.9 249 23.3 0.1 09 3.5 19 2.9 2.6 4.0 2.3 2h 3.2 5.0 3.6 125 4 92.5 2.6 14 5 5.3 14.9 4.7 7.8 15.3 9.9 2.6 l.l 0.6 2.5 44.3 Canada SMR 59 110 69 186 81 179 63 47 162 127 44 34 1450 228 87 108 105 115 75 89 76 126 59 56 65 71 271 41 38 60 106 51 20 30 76 175 513 80 27 95% C.I. 50- 68 12- 398 51- 91 21- 673 47- 130 36- 524 13- 148 13- 121 18- 585 41- 297 22- 79 15- 68 163-5235 26- 823 18- 254 12- 390 21- 306 23- 335 15- 220 10- 321 9- 275 34- 322 12- 174 6- 204 52- 81 55- 91 109- 559 15- 90 4- 136 27- 114 34- 247 14- 131 4- 57 6- 88 9- 275 20- 633 103-1500 9- 290 14- 47 Exp 285.1 1.4 68.9 1.2 20.6 1.8 4.2 8.9 1.3 3.8 24.2 22.7 0.1 l.l 3.5 1.8 30 2.8 3.8 2.2 2.6 2.8 47 3.4 129.4 98.3 1.9 14.5 5.6 14.4 5.1 7.3 15.3 10.3 2.5 1.2 0.6 2.1 37.7 Ontario SMR 95(8 0. 60 141 70 173 82 163 72 45 153 133 46 35 1917 182 87 III 101 105 78 93 76 141 51- 70 16- 508 51- 92 19- 626 48- 132 33- 476 14- 210 12- 115 17- 552 43- 311 23- 82 15- 69 215-6920 20- 656 17- 253 12- 400 - 20- 296 21- 308 16- 228 10- 335 9- 274 36- 362 64 13- 186 58 7- 211 63 50- 79 67 52- 85 364 146- 749 41 15- 90 36 4- 129 63 29- 119 98 31- 226 55 15- 141 20 4- 57 29 6- 85 79 9- 285 175 20- 630 526 106-1536 96 11- 346 32 16- 56 The average age at hire was 26 years and the mean duration of follow-up was 20.3 years. Nearly 40% of the cohort was hired before I960 and accounted for 57% of the person-years experience. Comparison with the Canadian Male Population For all causes, there were 240 observed and 367 expected deaths which resulted in an SMR of 65 (Table 11). The 95% confidence interv al (95% C.I.) was 57-74. Three deaths known by the company from insurance records were not identified by Statistics Canada. If the three deaths were included among total deaths, the SMR would not differ appreciably (SMR 66, 95%C.l.. 58-75). Contributing to the reduced all-cause mortality were statistically significant fewer deaths in the following 444 Canadian Journal of Public Health subgroups: respiratory cancer, circulatory disease, digestive diseases, and accidents, poisonings and violence (Table 11). Deaths due to malignant neoplasms bordered on being a statistically significant deficit (SMR 78, 95% C.I. 60-100). Four specific causes of death, cancer of the lung and bronchus, ischemic heart disease, cerebrovascular disease, and cirrhosis of the liver showed statistically significant fewer observed deaths. Two causes of death were significantly elevated: pleural neoplasms and other forms of heart disease. The latter is an 1CDA-8 disease mortality classification that included cardiomyopathy, congestive heart failure, pulmonary heart disease, cardiac dysarrythmia and myocardial insufficiency. There were three causc-of-death categories that had nonstatistically significant SMRs greater than 200: malignant Vol. 80 ST 00GI.9S6 melanoma, multiple sclerosis and musculo-skeletal disease. AH three categories had only two or three observed deaths and less than one expected death. SMRs were also calculated after ignoring person-years and deaths which had accumulated within 15 years of hire (Table III). For all causes, there were 171 observed and 290.4 expected deaths (SMR = 59,959c C.l. 50-68) which was lower than the corresponding SMR for the entire follow-up period analysis (SMR = 65). There were statistically significant fewer deaths due to all neoplasms and cancer of the bronchus and lung. Cancer of the pleura remained the only cancer site that had a statistically significant excess with one of the three pleura deaths occurring within IS years of hire. Comparison with the Ontario Male Population Tables 11 and 1(1 also show the number of expected deaths. SMRs and 95<ic confidence intervals based on Ontario male mortality rates for the two follow-up periods. Cause-specific mortality experience was comparable to that based on the corresponding Canadian male mortality rates. Mortality Distribution by Age and Number of Years Since Entering the Cohort The distribution of observed and expected number of total deaths by age and onset of person-years experience for the cohort is presented in Table IV. Only two age and calendar-year stratified SMRs were above the null value of 100 and they were based on only one observed and one or less expected deaths. DISCUSSION The observation of lower-than-expected mortality among the Sarnia Division employed population may be partially attributed to the "healthy worker effect".* The "healthy worker effect "is greatest in younger working populations and has less effect as a study cohort ages. There is not a homogeneous effect across all causes of death as cardiovascular disease generally shows a greater deficit of deaths than cancer. Exclusion of the first 15 years after hire minimizes the healthy worker effect, and also considers the latency aspect of carcinogenesis. A latency analysis excludes tnc person-years at risk for a potential chemical-induced cancer for which the cancer's clinical appearance would be unlikely. It is therefore noteworthy that the SMR for total cancer mortality was lower with the 15-year latency analysis than with the entire follow-up period. A major reason for the total neoplasms deficit was the 14 fewer-ihan-expected brcnchus and lung cancer deaths. A contributing factor may have been the strict restriction of tobacco smoking in the operating units which has been observed elsewhere to have reduced lung cancer mortality in an occupational cohort.1,1 Wigle reported no increased risk lor lung cancer among the Sarnia population or of men whose usual lifetime occupation recorded on the death certificate was the Sarnia petrochemical industry." Furthermore, the population of Lambton county has been shown to have no elevated risk for total cancer mortality or ofany specific major cancer site.12 Our findings further indicate that there has not been an elevated risk of total cancer or lung cancer among Dow Chemical workers in Sarnia. Three mesothelioma of the pleura deaths were observed with 0.1 expected. Increased risk of developing mesothelioma has been demonstrated among workers in asbestos mines, asbestos mills and factories, as well as asbestos manufacture and installation.11 Of the three former Sarnia Division employees who died of mesothelioma of the pleura, one individual worked in various positions in the warehouse for 22 years beginning in the early 1950s. During this period, the warehouse was the focal point TABLE IV Distribution of Observed and Expected (O/E) Deaths by Age and Year of Onset of Person-Years Experience (SMR in Parentheses) Year of Onset of Person-Years Experience Age 1950-1959 1960-1969 1970-1979 < 20 20-29 30-19 40-49 50-59 601 otal 4; 12.1 (32.91 61-94.1 (64.XI 77 I0X.3 (71.11 56 57.2 (97.X 1 11-13 4 (K2.ll 4 10.1 (39.7) 213 295.2 (72.2) 2/4.4 (45.8) 10.26.7 (37.5) 3 !3.0 (23.1) 3 7.4 (40.K) 1 1.0 1 Ult.HI 0 00 19 52.5 (36.2) 2 3.3 (60 7) 4 9.9 (40 4) t 2.3 (41.4) 0 15 (0) 1 06 1157 7) II 0 0 X 17.6 145 4) SMR cannot he calculated becau>c of zero observed person-years 1980-1983 0 0.2 (0) U 0.9 (0) 0 0.3 (0) 001 (0) 0 0.0 0 0.1 (0.0) \J 1.6 10) Toial K 20.0 <40. Il 75 131.6 (57.0) Kl 121.9 (65.4)1 59 66.2 (89.2 n 150 (86.6, 4 10.2 (19.2) 240 166.9 (65.4) November/December 1989 445 L86'l900i for asbestos shipped to the plant site. The asbestos, in a loose, dry form, was transported by truck in boxes. Warehouse employees unloaded the asbestos shipments and stored this material or transported it to the end users on the plant site. Boxes would occasionally split open spilling the asbestos. Warehouse employees would then sweep up the asbestos and manually transfer this material to intact boxes. These workers also handled all asbestos gaskets shipped for use on the plant site. Consequently, it is evident that warehouse employees who worked at Sarnia Division in the 1950s and 1960s. including the individual who died of mesothelioma, routinelyhandled asbestos on a regular basis without the benefit of respiratory protection. Appropriate handling techniques were instituted in the early 1970s at Sarnia Division. This resulted in significant reduction to asbestos exposure. The second individual worked for three years as a pipefitter at Sarnia Division and had the potential of asbestos exposure during delagging operations. Asbestos exposure risk related to his non-Dow employment is unknown. The third man worked for 32 years as an operator in the solvents, ethylene and styrene plants. In some areas of these plants, asbestos was used as an insulating material around piping and in high temperature furnaces. Although this material was well sealed, necessary maintenance could have led to greater risk of exposure. In the residual category of "other forms of heart disease" there were eight observed versus only 2.8 expected deaths. It is unlikely that this observation is occupationally related. It may be due to differential medical acumen from rural to urban, and primary to tertiary medical care settings. Roberts el al have reported the sensitivity and specificity of the CMDB record linkage process.14 A separate random sample of 1.000 study subjects of unknown mortality status was traced by independent means to provide an accuracy check. A total of 925 subjects were tracked and 63 of them were deceased. The record linkage of the CMDB detected 58 or 92Vf of these deaths and did not wrongly identify any of the 862 study subjects found to be alive by the independent follow-up. In the present study, the CMDB identified 150 of the 153 deaths (9895-) known by the company. One of the three deaths not identified by CMDB died outside of Canada. It is possible that some past employees, who were presumed to be alive because they were not identified by Statistics Canada as deceased, may have died as citir.ens of othei Commonwealth countries or of the United States. It was beyond the scope of this study and its resources to address this issue, although it was felt there would be few additional deaths found if such a search were conducted. The findings of the present investigation, as well as the study by Roberts and his co-workers, support the conclusion that mortality ascertainment by use of the CMDB is both sensitive and specific for occupational cohort studies. Furthermore, the CMDB allows the investigator to obtain the cause-of-death codes which is a more efficient process than that done in the United States where the National Center for Health Statistic's National Death Index only provides information on whether death has occurred. Cause-of-death information must be subsequently requested from the appropriate state's vital status registrar. Further studies will update the mortality experience of this cohort. A major goal is to computerize the cohort's retrospective work histories which will allow for analyses by- CO occupation and duration of employment. Acknowledgements We thank <i Qucsnel, M Knir. and P l.alondt ot the Occupational and Environmental Health Research l*nii of the VnaJ Stances and Disease Registries Section ol Statistics Canada Also, wc thank the provincial vital *Uli>lii registrars and the Health Dms.un of Smthtics Canada (or the uc al the Canadian Mortality DiU Haw. Wc also thank S. Raman ano J Gentleman tor helpful comments REFERENCES 1. Bond GG. McLaren EA. Cartmill JB. Wymer KT, Sobel W. LippsTf, Cook RR. Cause-specific mortality among male chemical worker*. Am J ind Med I9K?; 12: 353-83. 2. Wen CP. Tsui SP. McClellan WA. Long-term mortality study ol oil refiners workers. I. Mortality of hourlv and salaried workers. Am J Epidemiol 198V MX: 526-42. V O'Berg Ml. Burke CA, Chen Jl. Cancer incidence and mortality in the DuPont Company: An t pdate. JOM 19X7; 29; 245-52. 4 Smith ME. Sewcombe HB. Use ol the Canadian mortality database lor epidemiologic lollow-up. CanJ Public Health 1982. 73: 39-46. 5. Manual of the International Statistical CUvnlication ol Diseases. Adapted lot Use in the United State* Eighth Revision US Department ol Education and Welfare. 1977 (%. Mnnson R R. Analysis of relative survival and proportional tnnrtalitv Comput Biumed ftn 1974;7:325-32. 7. Raman S Mortality Analyst* u*ing Mon*ons Computer Program Statistics Canada. I9h$ funpublished). X. Rothman K. I. Biec .ID Epidemiologic Analysts with a Programma ble Calculator 11 S Cmvt Printing Office. Washington. D.C .. 1979. 9. Wen CP. T>ai SP Anatomy ot iltc healthy worker eflect A critique of summary statistic* employed in occupational epidemiology Scorn! J Work Environ Health 19x2; X (Supplt: 4K-52 10. l.awlcr AH. Mandcl JS. Sehuman I.M A retrospective cohort mortality Mudv of iron ore f hematite) miners in Minnesota. JOSt 19X5. 27.507- IV 11. Wigle IVI 1 he distribution ot lunecancel tn tv. n Canadian cities Con J Puhln Health 1977. AX 46?X 12. Mmtahtv Atlas of < .inad.i ( anadtun Government Publishing C cone Hull. Quebec. 19*0 IV Sclikoll |l. Hammond 1C (Editors) Health hazard' n| asbestos exposuK* Ann .Vf >Uti</Sci 1979; 330. I-Xll. 14. Roheits RS. Julian .1 V Muir DCI and Shannon IIS Cancel ntnrinhiy associated with the high tempeutme osidation of nickel suhsuthde IAR( Ncicuniii' Publication 19x4,5? 2?-VV Received I ehrunrs 16. I9XX Accepted July 26. I9xx 446 Canadian Journal of Public Health Vol. 80