Document yd4Ky2DZmRRBE2kmZ2G6xXK2

A Historical Prospective Mortality f Study of Workers in Copper and Zinc Refineries James N. Logue, Dr.P.H.; Michael D. Koontz, M.S.; and Michael A. W. Hattwick, M.D. Earlier health studies o f workers in electrolytic refinery operations have raised concern that workers may experience excess morbidity and premature death as a result of hmardous occupational exposures. This study was designed to determine if the refinery operation is associated with any excess mortality patterns. A cohort of 4,802 male workers exposed for at least one year during 1946-19 75 was assembled from nine U.S. zinc and copper refinerles. Vital status n m ascertained for 4,241 (88%) of the cohort. Death certificates were obtained for 84% (3551423) of h e deceased. Overall standard mortality ratios (SMRs) were calculated to be 92 for the cohort, 97 for the subgroup of copper refinery workers, and 83 for the subgroup of zinc refinery workers. Significantly high cause-specific SMRs \.cpereas follows: (1) cerebrovascular disease (CBVD) for the cohort; (2) all cancers, cancer o f the digestive tract, and CBVD for the copper subgroup; (3) all cancers, cancer o f the respiratory tract, and CBYO for one plant that demon- strated a significantly high overall SMR (133). The significant excess of cancer deaths among the study cohort is largely due to the plant that exhibited the significantly high overall mortality rate, but lack of smoking data qualifies this finding. The significant excess of CBVD seems to be consistent across plants and furt?x; iEjdurti) seems wrranted to verify the finding. A number of studies have been published on the health hazards of occupational exposure within the copper and zinc smelting industry. Employment in this industry has been statistically associated with increased mortality from cancer o f . the respiratory tract, including lung cancer,'" and cardiovascular disease,'n3 and i s possibly associated From GEOMET Technologies, Inc., 1801 Research Blvd., Rockville, M D 20850 (Dr. Logue, Senior EpidemioIogistlProject Manager; Mr. Koontz, Research Scientist), and Medical Information Management, Inc., Falls Church, V r . (Dr. Haltwick). Dr. Logue IS now Chief, Epidemiology Section, Epidemiologic Studies Branch (HFX- 130), Division of Biological Effects, Bureau of Radiological Health, US. Food and Drug Administration, Rockville, M D 20857. This work was supported by a contract from the Smelter Environmental Research Assoclatlon, Farmington, Mich. 398 with cirrhosis of the liver, leukemia and mveloma.3 Thew studies support the view that occupational exposure to arsenic in the smelter industry is particularly hazardous. A significant excess of deaths due to cancer of the respiratory tract was also noted in a group of chemical Worken exposed to arsenic associated with insecticide production.* Furthermore, an excess of cancer of the lymphatic and hematopoietic tissues, excluding leukemia, was noted, although the numbers involved in the calculations MR small. Excess lung cancer mortality has been observed In counties with copper, lead, or zinc smelting and refinin8 industries, thus leading to speculation that arsenicat air pollution may be hazardous to individuals living in & vicinity o f these smelters.6 However, this observation and speculations therefrom were subsequently challenged,' leading to an investigation8 that yielded no association between lung cancer mortality and residence near a Utah copper smelter. The electrolytic refining operations of major corpora. tions have not been studied as extensively as their smelting operations in terms of the health consequences of potential occupational exposures. Surveys of electrolytic copper and zinc refining tank houses have indicated that workers in these settings may be exposed to arsenic, acid mist, and various metals such as antimony, but such occupational exposures have appeared to be minimal as a rule. Epi. demiologic evidence to date concerning tank house workers has resulted from their use as a comparison group in r e t m spective studies of copper smelter employees, as will be dis. cussed. In 1977, Rencher et a19 reported the results of a retro spective epidemiologic study of mortality that involvcd comparisons of copper smelting, mine, concentrator and other (control) operations of a large copper corporation. The average age at death for the four employment talc. gories was estimated to be 63.9, 62.2, 61.9and 59.0 years for the smelter, mine, concentrator and control groups, respectively. Proportionate analyses of cause of death revealed a higher proportion of deaths due to cancer of the respiratory tract in the smelter group compared with the other three groups and with all residents of the state In which the corporation was located; however, the contrd i i ! Mortality in Copper and Zinc RefinerieslLogue at d J which consisted of employees in the electrolytic $,,tfIil\fIl(er r1 y, research center and conipany offices, demon- red a higher Proportion Of deaths for cancers other than ~3 dl [he respiratory tract compared with the other three ~ r c V l p 5and with state residents. Smoking irjormation was YfJ,,allled for a l l deceased smelter employees and for ran- ."C.. . sav+ of mine and concentrator decedents, but not ,nr control group. ,.--,A\rprond e.pidemlologic study by Milby and Hinelo was c,,,ducted on essentially the same group of employees as by Rencher et 21,' but electrolytic refinery $e wFrkerS were examined separately and compared with cnvI, , V ~ e ~in concentrator, mine, smelter, and reduction plana. study results included an estimated average age at for the refinery workers of 53.4 years, in contrast wirh estimates ranging from 57.4 to 70 years for other groups. h4,)re recently, Rencher e t all1 described the results of a epidemiologic study of morbidity as a followuc their earlier report. Group comparisons were made employees at the copper smelter, mine, concenfrator and refinery. The refinery subgroup demonstrated highest morbidity levels, both within the respiratory disfa$e classification and on an overall basis. Neither of these studies included any smoking information. smelters and refineries represent distinct but related environments, as discussed by Dungey.12 Carper smelting is almost entirely a pyrometallurgical prmersin which intense heat is used to extract copper from a CcjfnpIex mixture of other substances in the concentrate. The srnelting process produces a copper sulfide matte that is separated from slag and then oxidized in a converter to produce blister copper. Blister copper is generally fired in a refining furnace to increase its purity and then trans- ferred to a refinery as 98% to 99% pure copper anodes, which are further purified through an electrolytic refining process. Ouring electrolysis, copper travels from anodes to pure copper cathodes through an aqueous solution of copper sulfate and sulfuric acid, with impurities dropping to the bottom of tanks. Vast amounts of particulate matter and sulfur dioxide are emitted from smelting furnaces and converters. These emissions are passed through various pollutioncontrol devices prior to venting of the residual gases to a stack. Employee exposures may be associated with fugitive ernivions of these pollutants. In the refinery i t is beliwd that minimal ernployee exposures to dust or gases occur; noise caused by slapping sheets and stacking cathodes is considered the only significant industrial hygiene problem. Since to our knowledge refinery workers have not previously been treated as a primary study group, the implications of any health problems that may occur in this occupational setting are largely unknown. In March, 1977, arrangements were completed for a historical prospective mortality study of nonferrous metal Horkers involved in the electrolytic refining of copper and zinc at nine separate plants; the results of that study are the subject of this report. A tenth plant originally intended for inclusion in the study was excluded due to inadequate reporting. The purpose of the study was to determine i f employment in the zinc or copper refinery operations was associated with any health problems that might be detected Journal of Occupational MedicinelVol. 24, No. 5/May 1982 by comparing the mortality experience of a large, geographically dispersed cohort of refinery workers with that of a standard population. A cohort of 4,802 men who had been employed in the nine participating copper/zinc refineries between 1946 and 1975 was followed up to determine i f they were alive at the end of 1975. Overall, exposure-specific, plant-specific, and cause-specific standardized mortality ratios (SMRs) were constructed based on corresponding age-specific rates for the U.S. male population in 1970.' Methods Prior to the start of this investigationa study was under- taken in 1976 to determine the feasibility of a historical prospective mortality study. As part of this effort, walk. through surveys were conducted at two representative facilities to evaluate current and historical exposures in both electrolytic copper refining and electrolytic zinc refining. From these surveys i t appeared that zinc and copper tank houses represented different working environments with respect to potential exposure to acid mist and other potentially harmful substances such as arsenic. Therefore, the analysis of separate mortality experience for zinc tank house workers and copper tank house workers was planned as part of the study design. Observation at each study site revealed that there was little or no meaningful distinction between the exposures entailed by the different job categories. The various jobs conducted within the tank houses involved considerable mobility about the building so that average exposures, certainly by the respiratory route, to acidic aerorols ap- peared to be similar for all tank house workers. Therefore, i t was not planned to differentiate between different job descriptions within the tank house, but merely to ascertain total duration of employment in the tank house for each individual. Female employees were not included in this study because they were insufficient in number to allow statistically valid analyses. A worker was included in the study cohort i f he was found tohave been (1) employed in the participating refinery as a tank house worker as of Jan. 1, 1946, and worked for at least one year in the electrolytic department by Oec. 31, 1975; or (2) hired between Jan. 1, 1946, and Dec. 31, 1974, and worked for at least one year in the electrolytic department. A researcher visited each plant and microfilmed appropriate records of all past and current employees who worked in tank houses between Jan. 1, 1946, and Oec. 31, 1975. Microfilmed records were reviewed by the study staff and hard copies printed for all workers eligible for inclusion in the study cohort. The study population was divided into "actives," who worked through the end of the study period (Dec. 31, 1975), and "inactives," who terminated before the end of the study period and required follow-up. Vital status of the inactives was determined through use of death notifications filed at the employment locations together with submission of employee names and associated social security numbers to the Social Security Administration (SSA) for its review of earnings reports and death benefit applications. Copies of death certificates were obtained for members of the inactive known dead, both from the companies participat- 399 r Table 1 - Exposure by Plant ___Plant _T_o_ta_l _ Zinc _Co_ppe_r Both Unknown ~ Code No.(%) No.(%) No.(%) NO.(%) NO.(%) .- 1 856(100) 291 (34) 291 (34) 269(31) 5(1) 2 687(100) 687(100) 0 (0) 0 (0) O(0) 3.10 3,259(100) 0 (0) 3,259(100) 0 (0) O(0) Total 4,802(100) 978 (20) 3,550 (74) 269 (til 5(0) ing in the study and from bureaus of vital statistics of several jurisdictions. For each cohort member whose vital status could not be ascertained via the SSA submission or whose death certificate could not be obtained through the agency indicated in SSA files, final attempts were made to obtain the needed information through the efforts of individual plant personnel directors. The underlying (primary) cause of death for each decedent was coded by a trained nosologist according to the International Classification of Diseases, Eighth Revision14 and as recommended by the U.S. Public Health Service.I5 To compute SMRs for the analysis of the cohort's mortality experience, expected numbers of deaths were calculated by ( 1 ) grouping the person-years of observation for the entire cohort and subsets thereof by five-year intervals of age a t risk, (2) multiplying the person-years in each interval by the 1970 US. death rate for all males in that age group, and (3) summing the resultant products. A two-hiled statistical test was applied to each ratio of observed to expected deaths, assuming a Poisson distribution for the observed numbers of deaths. The BailarEderer test for the ratio of a Poisson variable to i t s expectation was used to test departure of the ratio from unityI6; all testing was performed at the 5% level of significance, Average age at death was also computed to provide a comparison with other estimates reported in the literature. ResuIts Description of Study Population - The distribution of the study population's occupational exposure (copper, zinc, combined) by plant is given in Table 1 . A majority of the men in the study were exposed tocopper only (74% [3,550/ 4,802)) and large subgroups of these men were found in all plants except plant 2. Of the 978 (20%) men exposed only --e- to zinc, all were clustered in plants 1 and 2, which r e p q 291 and 687 individuals, respectively. Table 2 lists selected characteristics of the entire +s,t cohort by plant. These figures indicate minimal variance across plants with respect to average age at hire of the study population. Indices reflecting year of hire also show mal variance, with the exception of plant 7. For this plant, the average year hired was nine years above the grand mean, and virtually no employees were hired before 1956. wi% the exception of workers at plan& 3, 7 and 10, the men in this cohort were almost exclusively white. The apparendy small number of nonwhite workers, coupled with differ. ential reporting of race across plant records (race unknow for 22.576 of the entire cohort), prohibited the applicati,,,, of any reliable race-specific analyses. Follow-up of Study Population - Table 3 shows the vital status of the study population by plant as of Dec. 31, 197.5. Of the 4,802 workers in the study, complete follow. up was achieved for 4,241 (88%) individuals. The unknom rate by plant varied between 506 (plant 6) and 17% (plant 10). Of the 423 (9%) men known to have died, death certificates were obtained for 3 5 5 (84?6). Eight of the2 certificates, however, had either missing or illegible cause of death information; men in this subgroup were necessarily eliminated from analyses of specific causes of death. The plant-specific success rates for obtaining death Certificates for study decedents varied between 63% (plant 7) and 97% (plant 8). Table 4 gives the vital status of the population by ex. posure as of Dec. 31, 1975. The unknown rate by exposure was 12% for the copper- and combined-exposure groups and 11% for the zincexposure group. Further details concerning the study's success vis-$4, ascertainment of vital status and procurement of death certificates are seen in the Fig. For 38 decedents whose certificates were in the custody of the New York City Health Department (and who are included in the count of 355 obtained certificates), causes of death were ab- stracted by that agency, forwarded to the study team, and coded by the study nosologist. Although causes of dcath were not associated with specific individuals for reasons of confidentiality, the identities of individuals for whom certificates could be located were revealed, and all such persons were found to be copper refinery workers employed at plant 10. Thus, these 38 deaths were included i r i all exposure-specific, plant-specific, age-specific and causespecific SMR analyses. Table 2 - Age, Race, and Him Data Characteristics by Plant PlanP All All All Characteristics 1 (Zinc) 1 (Copper) 1(Both) 2 3 4 6 7 8 9 10 Zinc Copper Plants No. in Cohort Average age at hire, yr Average year hired, yr %hired before 1956 %whiterace 29 1 24 1956 40 98 29 1 269 687 504 474 303 357 407 650 564 978 3,550 4,802 26 24 29 29 29 24 28 27 27 31 26 28 27 1957 1955 1959 1954 1953 1956 1965 1955 1957 1954 1958 1956 1956 43 46 43 59 53 38 0 7 43 47 42 46 45 98 49 100 68 97 97 43 100 92 48 99 79 84 individualsat plant 1 worked with either Zn or Cu or both; individuals at plant 2 worked with Zn;and individualsat plants 3 through 10 worked with Cu; five individualsat plant 1 with unknown exposure status are includedonly in summary figures 400 Mortality in Copper and Zinc Refineries/[-ogue et a1 / pant Code 1 2 3 4 6 1 8 9 10 Total -Table 3 Follow-up Status of 4,802 Copper and Zinc Refinery Worken by Plant 1 __T_ot_al_ No.(%) 856( 100) 687 ( 100) 504( 100) 474 (100) 303(1001 357( 100) 407(100) 650(100) 564( 100) 4,8U2(1UU) AliveNo.(%) 673(79) 575(84) 37404) 365(77) 266(88) 312181) 330(81) 532 (82 39 1(69 3,818(80) Oead No.(%) 58 (7) 49 (7) 64(13) 82417) 21 (7) 8 (2) 35 (9) 30 (5) 76(13) 423 (9) -U-nk_now_n _ No.(%) 125(15) 63 (91 66( 13) 27 (6) 16 (5) 37(101 42(10) 88(14) 97(17) 561(12) -_~ Death Cer-tificate - -Re-ce-ived _N_ot R-eceived No.(%) No.(%) - - 52(90) 43(88) 52\81) 72f88) 16(761 5631 34(97) 22(73) 59(78) 355(84) 6(10) 6(12) 12(19) lO(12) 5(24) 3(37) 1 (31 8(27) 17(22) 68(16) Exposure-Specific,Plant-Specific, and Age-Specific Anal- -ywI ~t was previously noted that comparisons between ,qnc and copper workers were planned as part of the study dc51BnC.omparisons by plant and by age group were also pcrforr,~edS. uch internal comparisons have been advocated In [he occupational health l i t e r a t ~ r e , p~a~rt~icu~la* rly when general reference population is used for SMR calculations. ficeden t characteristics used for these comparisons are by expcsure status and plant in Table 5. The ,hkaracteristics for each group of workers are number of decedents, average age a t death for the decedents, overall SMR and SMRs for five 10-year age groups within which but one study death occurred. The SMR for the entire cohort is 92. The highest SMR bv eyposure group is an SMR of 97 for copper workers; s\.lRs are equal to 83 for the zinc workers and 53 for the wrke:s exposed to both substances. Average age at death IS highest for the workers with combined zinclcopper exposure. Although the low SMR for combined exposure is Itatistically significant, this subgroup consists of only 269 uorkers who were all employed at plant 1. The average age at death across all plants for the de- ceased group was 58 years. The SMR of 133 for plant 4 is wgxficantly higher than unity and the SMR of 54 for plant 9 is significantly lower than unity. The respective average Jqes a t death (59 and 60) for these two plants, however, are consistent with the average age at death across all plmls. The markedly low average age a t death fe plaiit 7 tl4 years) seems to be associated with the relatively late average year of hire (1965) and low average age a t hire (28) for that plant (Table 2) and serves to illustrate the incomparability of age at death statistics across studies unless factors such as average age at hire, average year hired, and length of follow-up are also considered. The SMRs by age group in Table 5 are given to address the phenomenon known as the "healthy worker effect," which has been discussed in some detail by M~hlichael.'~ He has noted that such an effect, which presumably stems from an initial better-than-average health status persons entering most work environments, tends to disappear and perhaps reverse with increasing age. For the present exercise, observed and expected deaths in pairs of adjacent five-year age intervals were combined to lend some stability to the resultant SMRs. Despite aggregation, some SMRs are based on fewer than 100 person-years a t risk and there are no person-years at risk at plant 7 for the oldest age group (65 through 74 years). While the age-specific SMRs for the total cohort and for the copper and zinc subgroups are consistent with the healthy worker effect, this phenomenon clearly does not operate uniformly across individual plants. In particular, observed deaths exceed those expected for the four youngest age groups a t plant 4 and for the three middle age groups at plants 8 and 10. Excess deaths are also apparent among the youngest age group a t plants 6 and 7 and among the youngest zinc workers at plant 1. Some additional decedent characteristics not included in Table 5 are noteworthy. The average year of hire for the -Table 4 Followup Status of 4,802 Copper and Zinc Refinery Worken by Exposure Exposure Zinc Copper Both Unknown Tot81 JOIIV- _To_tal _ No.(%) 978(100) 3,550( 100) 269( 100) 5(100) -4,802(100) Alive No.(%) 793 (811 2,797 (79) 223 (83) 5(100) 3,818 (80) Dead No.(%) 73(7) 355(9) 15(6) 040) 423(91 ral MedicineNol. 24, No. SIMay 1982 Unknown No.(%) 112(11) 418(12) 31(12) 0 (0) 561412) - Death C e r t i ficat . e Received N__o_ t Rec_e_iv-ed- No.(%) ` No.(%) 62(85) 279(83) 14(93) 0 (0) 355(84) 11(151 56(17) 1 (7) 0 (0) 68(16) 40 1 plant Personnel Records vital Status Ascertalnmant Total Cohort Members 4,802 Social Security Administration Vital Statistics Bureaus n Plant Followup I _-Death Certificate Procurement Certificate Included in Record I 327 F]Certificate Obtained Certificate not I<bF] Certificate not Obtained 68 Total. All Sources Certificate Obtained 355 (83.9%) Certificate not Obtained 68 116.1%) Includes 38 deceased for whom New York City Health Department did not release death certificate but abstracted cause of death information Flow of record rrcertainmentin the study. deceased was 1942, with 89% having a hire date before 1956. The difference between average age a t death and at hire was 25 years, indicating that potential latency effects associated with zinc/copper can probably be discerned from the collective cohort experience. The interval between hire date and date of death was 20 years or more for 67% of the deceased group and ten years or more for an additional 2 1% of the group. Most researchers studying latency periods have chosen 20 years or more duration since first exposure as a latency definition.17 Cause-Specific Analyses -Table 6 presents cause-specific SMRs for major causes o f death among the 4,802 male refinery worken, based on 347 decedents for whom the underlying cause of death could be assigned by the study 402 nosologist. The observed numbers of deaths by cause generally are not significantly different from expectation, with the exception of deaths from major cardiovascular diseases (SMR, 84), particularly heart disease (SMR, 78), from cirrhosis of the liver (SMR, 24), and from accidents (SMR, 53), suicide (SMR, 41) and homicide (SMR, 44), all of which are significantly lower than expected. For the malignant neoplasms, SMRs are over 100 for cancer of the digestive system (105), cancer of the genitals (159), cancer of the urinary tract (103), and "other" lymphomas and neoplasms (104). None of these ratios, however, are significantly greater than 100. Also, although the SMR for major cardiovascular diseases (84) is significantlv lower than 100. i t is interesting to note the Mortality in Copper and Zinc AefinerieslLogue et ai -Tible 5 Decedent Chrrecttiktkr by Plant and Exporurc Plint. - All All All CharacteristiCS 1 (Zinc] 1 (Copper) 1 (Both) 2 3 4 6 7 8 9 10 Zinc Copper h n b Decedents 24 Average age at death, yr 59 SIfl 69 Ape SpecificSMRs, age group 25-34 118 35-44 92 4544 28 55-64 71 65.74 161 19 15 49 64 82 21 8 35 30 76 73 335 423 61 67 58 60 59 52 34 48 60 61 58 57 58 73 5 3 t 93 109 133t 74 88 104 54t 105 83 97 92 44 00 70 71 115 129 183 76 33 27 81 79 74 00 00 52 96 110 93 61 125 31 148 62 90 79 85 27 116 110 209 58 52 119 49 111 88 110 101 193 111 141 93 132 47 00$ 137 45 110 102 101 103 36$ 1934 162 217 92 140$ . . . 00$ 301$ 137 161 140 146 Y Individuals at plant 1 worked with either zinc or copper or both; individuals at plant 2 worked with zinc, and individualsat plants 3 through 10 worked with copper t Statistically Significant a t the .05 level Based on fewer than 100 person-years at risk cycess ( 124) for cerebrovascular disease (CBVD). This however, fails to achieve the 5% level of signifi- cance. For a supplemental analysis, deaths with unknown Causes were allocated to specific disease categories in pro- portion to the distribution of deaths with known causes. (Of the 423 deaths among the cohort, 68 had no available death certificates and eight had certificates with missing/ illegible cause of death.) Following this ZdjzskijW t, the SMRS for cirrhosis of the liver (29) and accidents (65) remain significantly lower than unity while the SMRs for hypertensive heart disease (339) and CBVD ( 1 5 l ) , particularly cerebral thrombosis (204),become significantly higher than unity. However, the SMRs for cirrhosis of the liver Jnd hypertensive heart disease are based on fewer than five observed deaths each. Since 74% of the study cohort were exposed only to copper and were employed in eight of the nine plants, it was considered important to present SMRs for selected causes of death for the 335 decedents in this major subgroup (Table 7). The only significant SMR prior to allocation of 64 deaths with unknown causes is for ischemic heart disease (78), with the SMR for CBVD (148) falling just short of i t s critical value (149). However, after allocation of deaths with unknown causes, the SMR for ischemic heart disease (97) i s no longer significant but SMRs for all cancers (128), the subgroup of cancers of the digestive tract (157), and CBVD (1 83) are significantly high. Since the overall SMR for plant 4 was found to be sig- nificantly high, an examination of mortality from selected causes was also performed for the 82 deaths associated with that plant (Table 7). Although no cause-specific SMRs based on the 72 deaths with cause-of-death information are significant, several causes have elevated SMRs, including all cancers (157), cancers of the digestive tract (199), cancers of the respiratory tract (191), major cardiovascular disease (122), and CBVD (221). On allocating the ten deaths with unknown causes in the manner described previously, the SMRs for all cancers (179), cancers of the respiratory tract (218), and CBVD (252) attain significance. Since the overall SMR for plant 4 was noted to be sig- Journalof Occupational MedicinelVol. 24, No. 5/May 1982 nificantly high, i t is natural to question whether this particular plant accounts for the excess cause-specific mortality previously detected for the copper refinery cohort following the allocation of deaths with unknown causes. This question can be addressed by comparing the SMRs in parentheses for all copper refinery workers, plant 4 workers, and other plant workers in Table 7. From this evidence, i t is apparent that plant 4 accounts for the majority and statistical significance of excess deaths due to cancer experienced by the copper cohort, since the SMR of 110 for malignant neoplasms among workers at other plants is not significant. Even though the SMR for cancers of the rewirx tory system among all copper workers is elevated ( 1 14) hut not statistically significant, plant 4 does account for a l l excess deaths in this category. The significant excess of deaths among all copper workers from cancers of the digestive system is not significant in either subcategory, but i s elevated in each. The only disease category for which the significant excess among all workers persists within both subcategories is CBVD. Discussion This investigation represents a historical prospective mortality study of 4,802 men who had been employed in one of nine participating copper and zinc refineries. rhrce refineries, corresponding to seven copper planti, one zinc plant, and one copper/zinc plant, are located in various geographic areas throughout the United States and represent five of the maior copper or zinc corporations. To our knowledge, this multicenter study i s one of the first of i t s kind since i t permitted internal comparisons of the entire cohort based on type of exposure and individual plant experiences, in addition to a "pooled" cause.specific mortality analysis. The cause-specific SMRs relating to the entire cohort showed no rates significantly higher than expectation, although the rates were significantly low for maior cardiovascular diseases, cirrhosis of the liver, and accidents, suicide and homicide. The SMRs for some cancer sites were somewhat elevated, but, again, none were significantly 403 Table 6 - Caus64pecifE SMRs for All Refinery Workers IC0 No.+ Disease Category No. of Observed Deaths 000.999 090-097 140.209 150-159 160-163 180.187 188-189 170-173, 190-199 204-207 200.203,208,209 210-239 250 390-448 390-398,402,404,410-429 402 410-413 410 411 412 430-438 431 433 430,432,435-438 440 441-448 470-474,480-486 490-493 571 580-584 600 800-949 950.959 960-978 All causes Syphilis Malignant neoplasms Cancer of the digestive system Cancer of the respiratory system Cancer of genitals Cancer of the urinary tract Other rnali~nintneoplasms Leukemia Other lymphomas, neoplasms Benign neoplasms Diabetes Major cardiovascular diseases Heart disease Hypertensiveheart disease Ischemic heart disease Acute myocardial infarction Other acute ischemic heart disease Chronic ischemic heart disease Cerebrovascular disease Cerebral hemorrhage Cerebral thrombosis Other cerebral diseases Arteriosclerosis Other diseases of the arteries Influenza and pneumonia Bronchitis, emphysema, asthma Cirrhosis of the liver Nephritis and nephrosis Hyperplasia of prostate Accidents Suicide Homicide 423 1 75 22 25 7 4 10 1 6 1 6 167 127 4 1 I4 66 3 45 33 7 9 17 1 6 5 5 4 3 1 29 6 6 All other known causes Unknown causes 38 76 ' ICD indicates lnternational Classification of Diseases, Eighth Revision t The SMRs in parentheseswere calculated following allocation of 76 deaths with unknown causes $ Statistically significant at the .05 level No. of Expected Deaths 461.905 0.103 81.058 20.883 28.605 4.411 3.900 10.982 3.461 5.788 1.292 6.161 199.162 163.359 1.437 146.834 99.5 10 1.396 45.869 26.59 1 7.717 5.385 13.364 1.993 5.28 1 11.349 8.216 16.951 2.581 0.323 54.514 14.716 13.591 51.288 ... ---r SMRt 92 . . . 969 (1,184) 93 (113) 105 (128) 81 (107) 159 (193) 103 (125) 91 (111) 29 (35) 104 (126) 77 (94) 89 IlOS) 84t(lo2) 78t 195) 278 (339t) 78 195) 66 (81) 215 (262) 98 (120) 124 (151$l 91 (111) 167 (204t) 127 (155) 50 (61) 114 (138) 44 154) 61 (74) 24 t (29t 1 116 (142) 301 1377) 53t (65t1 4 1t (50) 44$(54) 74 (90) ... higher than expected. Adjustments were also made t o the SMRs by proportionally allocating deaths w i t h unknown causes. Following this adjustment, only the SMRs for C B V D and hypertensive heart disease became significantly high and only the SMRs for cirrhosis of the liver and accidents remained significantly low, with hypertensive heart disease and cirrhosis of the liver each accounting for fewer than five observed deaths. Since a majority of the cohort was exposed only to copper, we also determined cause-specific SMRs for this subgroup. A significantly l o w SMR was noted for ischemic heart disease prior to alloca- tion o f diseases with unknown causes; however, following adjustment for these unknowns, a significant excess of deaths was detected for all cancers, cancers of the digestive 404 tract and CBVD. Finally, since the overall SMR for plant 4 was significantly high, we also investigated the causespecific SMRs relating to this particular plant, After alloca. tion of unknown causes, significantly high SMRs were noted for all cancers, cancers o f the respiratory tract, and CBVD. I t was further discerned that plant 4 accounted for the majority o f excess deaths due t o cancer among the copper refinery cohort, but that the significant excess of deaths due to C B V D following allocation o f unknowns persisted both among workers at plant 4 and among the cornbined workers at other study plants. It was n o t determined i f the increased mortality effect from CBVD identified in this study showed any gradient w i t h length of employment, but the deceased group did Mortality in Copper and Zinc RefinerieslLogue et a1 -c' -Table 7 No. of Observed and Expected Deaths for Selected Disease Categories Amon! All Copper Refinery Workers, Copper Refinery Workers at Plant 4, and Copper Rofinary Workan at Other Study Plants All Copper Refinery Workers Rant 4 Workers - --- Other Plant Workers 1 Obsened/ 0bsewed/ 0brewed/ Disease Category / Malignant neoplasms Expected Deaths 63/61.01 SMR" 130 (128t) Expected Deaths 18/11.48 SMR" 157(179t) Expected Deaths 45149.53 -SMR'- 91 (110) Oigestive organs and peritoneum 20115.71 127 (157t) 613.01 199(227) 14/12.70 110 (133) Respiratoq system 20/21.68 92 (114) 814.18 191(218tl 12117.50 69 (83) Genitourinary organs 816.16 130 (161) w.19 84(96) 7/4.97 111 (170) Lymphatic and hematopoietic tissue 616.93 87 (1071 311.21 248(2821 3i5.72 52 (63) Ma,or cardiovascular disease 1281148.65 86 (1061 34127.82 122(139) 94t120.83 78t(94l Ischemic heart disease 861109.92 78t(97l 22/20.32 108U23) 6W89.60 7 1t (86) Cerebrovascular disease 29/19.58 148 (183t) 8/3.62 221(252t) 21i15.96 132 (159t) + The SMAs in parentheses were calculated following proportional allocation of deaths with unknown causes; death with knownlun- k.n~own causes were 335164 for all cop. p. er workers, 72/10 for plant 4 workers, and 263/54 for other plant workers t Statistically significant a t the .05 level consist primarily of men who had been hired 20 years before the end of the study period (89K hired before 1956). Thus, the CBVD effect may be related to exposures in the or to other factors not considered herein. It is interesting to note, however, that an SMR of 450 for attributed to this disease was relatively small; the issue of contributory causes of death was not examined. Ostfeld's review did not implicate cigarette smoking as a major contributor to strokes. Thus, although cigarette smoking was not analyzed in this study, lack of data on this CBVO had been noted earlier among workers at a Swedish copper snielter3 which, when the data base was stratified lo eliminate the confounding effect of arsenic exposure, !till appeared as an S M R of 170 (pz.07 based on a onetailed test) for copper exposure only. A more recent study of Swedish copper smelter workers20 demonstrated a liniilar significant effect for CBVD (SMR=157; p<.O1 based on a two-tailed test). 4 recent review of the epidemiology of stroke by OstfelJ2' pointed out that high blood pressure (BP) is the most important risk factor for stroke and that heart disease and diabetes are also major disease contributors to the risk of stroke. It is possible that either high BP or diabetes or both were disproportionately high in occurrence among the 5tudy cohort and accounted for the high cerebrovascular ratio for the entire cohort, for the subgroup of workers exposed to copper, and for workers in plant 4. The S M R for hypertensive heart disease in this study was significantly -high for the entire cohort although the number of deaths factor would not seem to have had a confounding influence on the apparent CBVD effect. Since blacks are known to be a t greater risk of mortality from strokes than whites, the racial distribution of CBVD decedents was examined. The proportion of nonwhites among these decedents was lower than the corresponding proportion among the entire cohort, thereby ruling out the confounding effect of race on stroke mortality in this study. Finally, Ostfeld noted the difficulties in diagnosing stroke and the fact that stroke and heart disease represent competing risks for mortality. The finding of an excess of deaths due to CBVD and a shortfall of deaths due to heart disease in this study i s consistent with that notion. The finding of a significant excess of deaths due to cancer among the copper refinery cohort in this study i s subject to multiple interpretations. Wagner22 has indicated in a report on environmental conditions in smelters in the United States that arsenic-inducedcancers probably exist in this industry. Though it i s tempting to apply this specula- Table 8 - Percent of U.S. Male Oeaths Among Selected Ape Groups Caused by Selected Diseases" Age Group, yr 40-44 45-49 50-54 55.59 60.64 65.69 70.74 All Malignant Neoplasms 14.4 17.5 20.3 22.2 22.6 21.6 19.7 Cancers of DigestiveTract 3.1 4.2 5.2 5.9 6.3 6.2 6.2 Diseases -____ Cancers of Respiratory Tract -~ - _ _ 5.2 6.7 8.1 9.0 8.7 7.8 6.2 Source: U.S. Public Health Service, Vital Statisticsof the UnitedStates, 197013 _ _.___ .. ._ Cerebrovascular Disease 4.2 4.4 4.8 54 6.6 8.2 10.2 Journal of Occupational Medicine/Vol. 24, No. S/May 1982 405 tion to refinery workers, most refineries in this study are not geographically proximate to smelters and, as noted earlier, potential arsenic exposures in the refinery seem to e x i s t only a t low concentration levels. The major exception to the point made above is study plant 4, where the refinery is directly adjacent to the smelter. This plant was previously noted to account for the majority of excess deaths due to cancer in this study and, more specifically, for all excess deaths due to cancers of the respiratory system. Many of the employees at plant 4 are older workers who have bid into the refinery from the smelter after attaining sufficient seniority; a smaller proportion of workers are assigned to the less demanding work in the refinery hecause of a handicap, disability, or chronic health problem (Lowell D. White, Written Communication June 4, 1981). Thus, the analysis of cancer mortality a t this plant is confounded by prior occupational exposures that most likely are more severe than those experienced in the refinery together with a possible reversal of healthy worker effect. At plant 8, the smelter and refinery are about a mile apart and the refinery workers similarly may have transferred from the smelter and probably represent a higher proportion of disabling or chronic conditions than in most work settings. The SMRs greater than 100 were previously noted for the four youngest age groups at plant 4 and the three middle age groups at plant 8. Since all cancers of the respiratory tract detected among the study cohort were specifically cancers of the lung, it follows that the SMR for this anatomical site would be higher than that for cancers of the respiratory tract as a group. However, the lack of smoking data for this cohort qualifies any findings with respect to lung cancer. Sterling and Weinkam23 have used data from the 1970 Household Interview Survey (HIS) conducted by the National Center for Health Statistics to demonstrate that smoking prevalence tends to be uniformly highest among b!ue collar workers. A surveyz4 of workers at a Garfield, Utah, smelter yielded estimates that 69.9% of white male workers had ever smoked, a figure that does not differ appreciably from the corresponding HIS estimate of 69.2% for white males. However, the Garfield survey data further indicate that 33.6% of the smelter workers smoked a pack of cigarettes or more per day, in comparison to HIS estimates of 28%. Moreover, a survey5 of retirees from a Tacoma, Wash. smelter estimated that 50?6 of those individuals were current smokers, as contrasted with HIS estimates that 23% of males aged 65 and over currently smoked. I f the two surveys are indicative of smelter worker smoking habits, then these workers tend to smoke more intensively and abandon the smoking habit less frequently than U.S. males as a whole. However, the extent to which findings among smelter workers are applicable to the refinery setting cannot be discerned from available data. AI though there are p r e ~ e d e n t s i~n ~thie~ o~ccupational research literature for allocating deaths with unknown causes in the manner used in this study, potential biases associated with such an approach should be explored. Selected comparisons between deceased with death certifi- cates and those without indicate that such biases should be minimal in this study. The distributions of the two groups with respect to calendar year of death were virtually identical so that possible distortions attributable to 42 "<.changing customs for assigning cause of death shwld manifest themselves here. On the other hand, the a,,eral, age a t death for the deceased without certificate more than six years lower than that for the groups`klc certificates. i f the disease categories in which eycesr st'kith deaths occurred also claim an increasing proport\on9 ,deaths with increasing age, then the allocation process or ~ f have favored these diseases. Data in Table 8, presented f age groups in which nearly 90% of study deaths ~ c u r r c 4 show this to be the case; however, the aberage increase percentage intervals i s of lives claimed over successive five-year only about one percentage point for C R V m~Fd c.malignant neoplasms and i s a lesser amount for the subcategories of the process may interest. Thus, on the have "overallocated" a bt amsoissot of ntehed%eadth`4 any of the disease categories, since the total number study deaths with unknown causes was 76. Such a minlml distortion should not have an impact on study findings. Our use of U.S. male mortality fates as the Standard in calculating SMRs i s justified in the sense that the nine 5t"d) plants are geographically dispersed and their coIIectke racial distribution is similar to that among US. male. Though we may be open to criticism for using 1970 rate) in lieu of sets of rates specific to calendar years, deaths ament the study cohort tended to cluster in the ten-year perid between 1966 and 1975. According to the review of Silverberg and Hollebz7 of major trends in cancer over a recent 25-year period, age-adjusted cancer death raws generally remained constant over that period although a decrease was noted for stomach cancer and an increase was noted for lung cancer. A review2* of the leading caucej of death over the period 1950-1969 indicated increaw for malignant neoplasms and decreases for CBVD. In view of this evidence, our use of 1970 rates as a standard may haw inflated SMRs for CBVO and cancers of the digestive tract and deflated SMRs for cancer in general and the rubcet of cancers of the respiratory tract. However, such biases should again be minimal since most study deaths were proximate in time to the year 1970. The results of this study seem to shed additional light on conflicting "age at death" statistics reported from earlier studies. Rencher et ,I9 reported average ages at hire and at death of 32.3 and 59.0 years from their 1977 study of refinery workers at a Utah copper plant. Our corresponding figures are 32 and 48 years for a Utah copper plant and 33 and 58 years across all study plants. While their statisticc are based on deaths between 1959 and 1969 for one plant, our overall figures are based on deaths between 1946 and 1975 for nine plants. Thus, our figures have a more comprc- hensive base, both in terms of industry representiveness and coverage of time. Milby and Iline's1 average age at death estimate of 53.4 for Utah refinery workers had no accom panying age at hire statistics and was based on deaths between 1950 and 1972 among those who worked for ten years or more for one company and, furthermore, had a complete work history available. Since average-age-at-death statistics from this and prior related studies are not based on the complete follow- up experience of any cohort, they are not valid indicators of premature death and are not truly intercomparable. A suggested statistic for future comparisons is the expected age at death, as discussed by gaffe^.^^ This statistic may be 406 Mortality in Copper and Zinc RefinerieslLogttee t al -Table 9 Mortality Rates and Probability of Death by Ago for All RafineryWorkers and for U.S. Males in 1970 fige Group. Vr Person-Years a t Risk 1-yrMortality Rates _. Refinery 1970 Workers U.S. MdK SMR Probability of Death During 5-yr Interval. Refinery Workers 1970 U.S. Males Relativa Risk rc- 15.19t 20-24 25.29 30.34 35-39 40.44 669.3 5534.2 8809.9 9354.5 9278.2 8600.4 ... ... ... ... ... 0.000181 0.002257 8 0.000905 0.01 1222 0.001249 0.002027 62 0.006226 0.010084 0.001924 0.002301 84 0.009574 0.011439 0.00 1940 0.003142 62 0.009653 0.015587 0.0044 18 0.004849 91 0.02 1848 0.023954 45.49 50.54 55-59 7336.2 5565.9 3156.9 0 . 0 0 6 8 16 0.007567 90 0.033506 0.037 128 0.012936 0.01 1793 110 0.062633 0.057260 0.019679 0.018572 106 0.093791 0.088679 60-64 2047.5 0.027350 0.027863 98 0.127812 0.130046 65-69 10-14 - > 75t 768.4 415.5 313.1 0.071577 0.04 1179 174 0.300847 0.186082 0.064982 0.058934 110 0.277408 0.255223 ... ... ... ... ... Equal to I.e-(l-yearmortality rate x 5) for a given age group ... 8 62 84 62 91 90 109 106 98 162 109 . . .- t N o decedents from the cohort within these age groups c a ~ c ~ j l ~btye dlife-table methods with knowledge of the specific age. From the inputs in Tables 9 and IO, the ex- relationship between age-specific mortality rates and pected age at death for workers entering a t age 20 is prcbahilities of death during specific age intervals, as dis- estimated to be 6 9 . 3 5 years for all refinery workers and by Chiang.30 Inputs to the calculation of expected 69.15 years for the subgroup of copper refinery workers. Jqe a t death for all refinery workers and the subset of The entire cohort has an average age a t hire of 27 years and crtpper refinery workers in this study appear in Tables 9 an average year of hire of 1956. By comparison, expected and IO, respcctively. Since there were no cohort decedents age a t death for the U.S. white male population aged 20 lged 75 or older, it was assumed that workers reaching years during 1949-1951 was estimated to be 6 9 . 5 year5.j' age 75 would live to an average age of 83.3 years, based on Thus, from evidence in this study, employment a t the re- 1970 U.S. life-table values for this age group.13 finery does not seem to be associated with premature The expected age-at-death statistic utilizes the experi- death. However, use of a U.S. life-table value for workers ence of an entire worker cohort and may be interpreted as reaching age 75 tends to bias refinery workers' l i f e expect- life expectancy on entering a given work environment at a ancy toward that of the U.S. population. r -Table 10 Mortality Rates and Probability of Death by Age for Coppar RefineryWorkers and for U.S. Males in 1970 I - Age Group, yr 15-19t 20-24 25.29 30.34 35-39 40-44 45.49 50-54 55-59 60-64 65-69 70.74 3 75t Person-Years a t Risk 359.9 3751.9 642 1.5 7063.4 7003.3 6448.8 5561.6 42 10.0 2879.0 1556.7 598.2 322.4 199.4 1-yrMortality Rates Refinery Workers 1970 U.S. Males ... 0.000267 0.001402 0.001982 0.002285 0.004962 0.007552 0.0 140 14 0.018757 0.028265 0.066867 0.065137 ... ... 0.002257 0.002027 0.00230 1 0.003142 0.004849 0.007567 0.0 1 1793 0.018572 0.027863 0.04 1179 0.058934 ... SMR ... 12 69 86 73 102 100 119 101 101 162 111 ... Probability of Death During 5yr Interval. Refinery Workers 1970 U.S. Males ... 0.001334 0.006986 0.009861 0.01 1360 0.024505 0.031056 0.06767 1 0.089522 0.131793 0.284186 0.271967 ... ... 0.01 1222 0.010084 0.011439 0.015587 0.023954 0.037 128 0.057260 0.088679 0.130046 0.186082 0.255223 ... Relative Risk ... 12 69 86 73 102 100 118 101 101 153 109 ... Equal 10 l.e-(l.year mortality rate 5) for a given age group t No decedents from the cohort within these age groups Journal of Occupational MedicineNol. 24, No. 5/May 1982 407 Su n i m a r y The study results are essentially negative, with most positive findings subject to qualification. The significant excess of deaths due to cancer among the study cohort following allocations of deaths with unknown causes was largely attributable to excess deaths due to cancer at one study plant. Unlike other study plants, its refinery is directly adjacent to a smelter; a number of workers at this refinery have transferred from the adjacent smelter wherein more severe occupational exposures probably occurred arid these workers as a group are believed to possess a higher proportion of disabling or chronic conditions than in most work settings. Lack of smoking data further qualifies the study results with respect to cancer and, in particular, with respect to the significant excess of cancers of the respiratory tract at the refinery cited above. A significant excess o f study deaths from CBVD, however, could not be attributed to that plant alone. The findings with respect to CBVD are similar to those found in several earlier studies of smelter workers, although smelters and refineries probably do not represent similar exposure environments. A possible qualifier to the CBVD excess is the notion of competing risks, especially since the SMR for ischemic heart disease in this study was lower than unity. I f the CBVD effect is a real association, i t will be necessary to determine which occupational factor(s) contribute to the excess. For example, noise may be involved as a risk factorJ2 based on an understanding of the refinery environment. Potential biases associated with the allocation of deaths with unknown causes and with the use of 1970 mortality rates for SMR calculations were investigated and considered to be negligible. An exercise based on life-table methods indicated that employment in the study refineries did not appear to be associated with premature death. The study would benefit from an additional five years of follow-up information, coupled with more intensive efforts toward death certificate procurement, since any significant excesses of cause-specific mortality followed the allocation o f deaths with unknown causes. The 200 or more deaths expected between 1976 and 1980 for this cohort, together with increasing years-at-risk contributions from older cohort members, would tend to stabilize the age- and cause-specific SMRs reported at this time. Further research may be warranted to substantiate the positive findings of Rencher e t all1 with respect to refinery worker morbidity patterns, but studies that do not include ancillary data on smoking behavior are not recommended. Acknowledgment Regina Werner provided technical assistance; Katheryn Lyndon was the study nosologist; Ruth Prost and Betty Klein assisted in manuscript preparation; Bob Albrecht, M.D., Frank Lundin, M.D., and Lowell White, Ph.D., gave critical readings of the test. References 1. Lee AM, Fraumeni JF: Arsenic and respiratory cancer in man: An occupational study. JNCI 42:1045-1052, 1969. 2. Kwlatsune MI Tokudome S, Shlrakusa TI e t al: Occupational lung cancer among copper smelters. Inr J Cuncer 13552-558, 1974. 3. Axelson 0, Dahlgren E, Jansson DC, et al. Arsenic exposure and mortality: A case-referentstudy from a Swedish copper smelter. Br J IndMed 35:8-15, 1978. 4. Pinto SS, Henderson V, Enterline PE: Mortallty experit. of arsenic.exposed workers. Arch Environ Htalfh 33:325.331, l a 5 . Ott MG, Holden BE, Gordon IIL: Respiratory cancer occupational exposure to arsenicals. A r c h Envlron M d f h 29.2 255, 1974. 6. Blot WJ, Fraumeni JF: Arsenical air pollution and I, cancer. Lancer 2:142-144, 1975. 7. Nelson KW: Industrial contributions of arsenic to the vironment. Environ Health Perspccf 19:31-34, 1977. 8. Lyon JL, Fillniore JL, Klauber MR: Arsenical alr pollu! and lung cancer. Lancer 2:869, 1977. 9. Rencher AC, Carter MW, McKel OW: A retrospective , demiologlcal study of mortality at a large western copper snlcl /OM 19:754-758, 1977. 10. M i l b y TH, Hlne CFI: A survey of m o r t ~ l l t ydue to rc(r tory diseases among employeer o f the Kennecott Corp. Sari 1 , cisco: Dlvislon of Ambulatory 3nd Community Medicine, Unive:. of California, School of Medlcine. San Franclsco. 1974. 11. Rencher AC, Carter MW, McKee DW: A retriirprctiv* demiological study of morbidity at a large western copper me. Am IndHyg Assoc J 40:695-701, 1979. 12. Dungey Curtis E: Copper, in Cralley L V , Cralley L l (En Industrial Hygiene Aspects of Plant Operations: Procea Flc New York: MacMiilan Publishing Co., vol. 1, in press. 13. US. Public Health Service: Vital Statistics of the Un, States, 1970, vol. I I , Mortality, Part A. Rockvllle, Md.: 1974. 14. International Classificatlon of Dlseaw, Eighth Re*i<' VOl 1. Washington, D.C.: DHEW Public Health Service Publica; 1693, U.S. Government Printing Offlce, 1969. I S . US. Department of Health, Education, and Welfare: Statistics: Instructions for Classifying the Underlying C a w Death, 1976-1977. Rockville, Md.: Health Resources Admini. tlon, National Center for Health Statistics, June 1975. 16. Bailar JC, Ederer F: Significance factors for the ratio I Poison variable to its expectation. Biomctrks 20:639643, 1`1 17. Enterline PE: Pitfalls in epidemiological research. / 18:150-156, 1976. 18. Schottenfeld D, Haas JF: Carcinogens In the workplace. 29:144-168, 1979. 19. McMichael AJ: Standardized mortality ratios and "healthy worker effect": Scratching beneath the surface. i 18: 165-168, 1976. 20. Wall 5 : Survival and mortality pattern among S w r smelter workers. Int J &pidemio/9:73-87, 1980. 21. Ostfeld AM: A review of stroke epidemiology. p i & Rev 2:136-152, 1980. 22. Wagner WL: Environmental Conditions in U.S. Cor Smelters. Salt Lake Clty: US. Department of Health, ducal and Welfare, Public Health Service, Center for Disease Ccnt National Institute for Occupational Safety and Health, Oivlsior Technical Services, April 1975. 23. Sterling TD, Weinkam J J:Smoking characteristics by tvp employment JOM 18:743-754, 1976. 24. Archer VE, Gillam ID: Chronic sulfur dioxide exposurr smelter: II.Indices of chest disease. JOM 20:88-95, 1978. 25. Cooper WC, Gaffey WR: Mortality of lead workers. , 17: 100-107, 1975. 26. Wang HH, MacMahon B: Mortality of pesticlde applica! JOM 21:741-744, 1979. 27. Silverberg E, Holleb AL: Major trends in cancer: 25-q survey. CA 25:2-21, 1975. 28. US. Department of Health, Education, and Welfare: hlo ity trends for leading causes of death: United States .- 1950.1` DHEW Publication (HRA) 74-1853, series 20, No. 16. Hyatts~ Md.: National Center for Health Statistics, 1974. 29. Gaffey WR: A critique of the standardlied mortallty r . /OM 18~157-1601, 976. 30. Chiang CL: A stochastic study of the life table an+ applications: 111. The follow-up study with the conrideralioi competing risks. Blometrics 17:57-78, 1961. 31. U.S. Bureau of the Census: Statistical Abstract of the Un States: 1979 (100th ed). Washington, D.C.: U S . Government P ing Office, 1979. 32. Resnekov L: Noise, radio frequency radiation and cardiovascularsystem. Circulation 63:264A-266A, 1981. Mortality in C.opper and Zinc RefinerieslLoguee