Document 99DmmBZyK6DgYnL5eynmKozq7

Causes of Death Among Employees of a Synthetic Abrasive Product Manufacturing Company David H. We#man, M.D., and Ellen A. Eneit, M.S. Analysts of the causes of death among workers employed in the manufacture of synthetic abrasive prod ucts was performed using data from death certificates and employment records. The subjects were 1,030 white male workers who had been employed by a single manufacturer and who were identified through a review of death certifi cates issued during a 20-year period. Proportional mortal ity analysis revealed excess digestive cancer and respira tory disease deaths. Personnel records permitted refine ment of the study population to 968 with classification of each individual into one of seven employment categories. Case-control analysis of workers ever employed in the indi vidual categories revealed elevated odds ratios for respira tory disease deaths among those most exposed to the syn thetic abrasive dust The excesses were greatest in those ex posed 20 or more years. Excesses of esophageal and rectal cancers were noted in two manufacturing areas although the numbers were small. Review of the literature further supports the finding that synthetic abrasive exposures may be associated with elevated respiratory morbidity and mor tality. I he manufacture of synthetic abrasive materials has been commercially successful for more than 50 years. These materials have largely replaced natural abrasives (such as sandstone) because they have better properties as abrasives and because studies of short-term health ef fects have shown them to be less hazardous However, there is little published information evaluating the effects of long-term exposure to the components of synthetic abrasives A cause-of-death analysis of former employees of one large producer of synthetic abrasive products was undertaken to investigate possible health effects of this work process From fh#* Mats*! Division of Qcrupdtionjl Hygiene ind the Harvard `x hoof o Public Health hbS Huntington Ave Boston MA 0211$ (Dr Wegmtn, As%<xidf* Ptottpwtyr o Orcupational Health and My Eiten) Supported by Grant* T 1S OH*O70-6 (ram the National Institute (or Of cup** tionat Satetv and Health and SSS 4 P10 ES0QQ2 from the National Imtitute a* Environmental Health Science* 748 Methods Study Croup. -- In Massachusetts death records are maintained by city and by year of death. Because the manufacturer of interest is a large employer in a small city, it was possible to select a study population directly from the death records. Records from the city where the manufacturer was located were examined for a 20-year period beginning with 1954 A record was chosen only if this employer's name was noted in the location marked "industry or business." Date of death, age at death, sex, usual occupation, underlying cause of death, name and ethnic origin of the individual were abstracted. Since few non-whites or females were found, this report will discuss white males only. After results of an initial proportional mortality analysis were presented to the company, they granted ac cess to their personnel records. A search of these records provided adequate work histories for most of the sub jects. These files also contained records of subjects who had not been included in the original study group, but who had died between 1954 and 1973. They had not been identified in the search of state death records because they had died in other than the study aty, or had died out of state, or because the company name did not appear on the death certificate Work Histories. -- The plant manufactures abrasives, beginning with the synthetic abrasive grains, m the follow ing production groups: (1) Abrasive Preparation -- prepa ration of the synthetic abrasive grains, including crushing, washing, sizing, and grading; (2) Production -- manufac ture of aluminum oxide or silicon carbide abrasive wheels, using a clay bonding material (vitreous bonding) or an organic bonding material (organic bonding) The process involves mixing of grains and bonding material, wheel molding kiln firing truing (to assure parallel sides), grading (for hardness and soundness) and inspection Pro duction of synthetic diamond abrasive wheels using either an organic or a metal bonding material is done similarly but on a much smaller scale, (3) Refractories Pro cess -- manufacture of light and heavy refractory materi als; and (4) Packing and Shipping -- packing and shipping UCC 096866 areas are located m each production department but are substantially removed from the production area. Each personnel record was abstracted without knowledge of cause of death, recording date of first employment and first job, date of entry into each subse quent job (up to a total of six jobs), total years worked in each job and year terminated. A job had to be worked for at least six months to be tabulated. Persons who worked more than 50% of the time in clerical or office jobs were considered clerical workers and excluded frond* most analyses The employer records of work history were excellent and reasonably detailed. The company has been under the same management for over 50 years and has kept records for that entire period. Records for those still work ing after 1945 showed specific job titles. The work history for persons who retired before 1945 was limited to major categories, generally departments. Since the number of subjects studied would not permit subdivision into more than five to ten groups, the limited data available before 1945 were considered detailed enough. A walk-through survey of the entire plant provided the information necessary to identify 26 unique jobs which were grouped into seven "departments". (1) Abrasive Preparation; (2) Vitreous Bonding; (3) Organic Bonding; (4) Packing and Shipping; (5) Nonabrasive (includes Machine Tool Division, SmaH Machine Manufacture and Maintenance); (6) Service, and (7) Miscellaneous (includes Diamond and Refractories Abrasive Manufacture as well as other unclassifiable production jobs). Analysis Data were analyzed first in aggregate and then within each department by proportional mortality analysis (PMR) using the computer program of Monson' updated in 1978. This program uses cause-, time-, sex-, and racespecific data for the population of the United States from 1925 to 1975 In this analysis the 7th revision of the Inter national Classification of Diseases1 was used. The proportional mortality analysis approach is recog nized to have shortcomings. It examines proportions, not rates, of deaths so it does not examine the actual force of mortality It also, as in this case, commonly employs the national mortality proportions as the standard and thus cannot consider regional environmental or socioeconom ic differences. In response to some of these shortcomings an odds ratio approach was also utilized in this investigation With this approach the expected ratios are derived from the death experience of members of the same work force Furthermore, a reasonable assumption can be made that, properly selected, the referent diseases have the same rates in the exposed group as in the unexposed group To the extent that this is valid, the odds ratios are the same as rate ratios, which do indicate the relative force of mortali ty for the diseases under study.1 Those work areas showing unusual proportional mor tality therefore were also examined by case-referent companson using as comparison diseases all diseases except cancer and nonmalignant respiratory disease. Subjects and controls were grouped according to whether they had never or ever.worked in the department under study Those who had worked in the department were also strati fied by duration of work. Odds ratios were calculated be tween the groups of never and ever exposed Age and ethnicity were specifically considered. With respect to age, odds ratios were adjusted using three age categories (younger than 65,65 to 74, 75 years and older) and associations were estimated and tested by the Mantel-Haenzel method.4 Since Scandinavians, predominantly Swedes, make up 40% of the work population, an effort was made to deter mine whether this ethnic group was overrepresented in any disease category. A log linear model' was used to describe the data in terms of department of longest job, ethnicity, cause of death, and the interactions between these three variables. Results Death records for 1,030 males were identified in the state records for the years 1954 through 1973 Digestive cancer (five out of six sites), Hodgkin's disease and non malignant respiratory disease categories were notably elevated in a proportional mortality analysis. Over 170 different occupation titles were included on the 1,030 death records. These were grouped and exam ined for group-specific excesses but none was found Expanded Study Croup. -- lob histories were available for 82% of the 1,030 decedents Of these, 90 had worked as clerical or office workers and were excluded Employ ment and death records were obtained for an additonal 229 employees who were identified in the personnel file as having died in Massachusetts but who had not been identified through the city's records of death All had either died in other cities or had death certificates on which the company's name did not appear Sixteen had worked as clerical or office workers and were excluded Since these two sets of deaths had been identified from different sources, PMR analyses of the two sets were com pared. The differences noted were that decedents iden tified through company records had a deficit of non malignant respiratory disease (8 observed, 120 expected) and a younger mean age at death (60.3 vs 69 6) PMR analysis of the clerical workers revealed a normal pattern The one excess observed was for digestive cancer, but the ratio was not significant and the elevation was accounted for by an excess of liver cancer (2 observed, 0 47 ex pected) There was no excess of nonmalignant respiratory disease and no death due to pneumonia The two groups, excluding clerical workers, were combined to provide 968 final study subjects Table 1 shows distribution by department of the nnal study subjects and the numbers who worked ten or more years Of the decedents who had worked in departments engaged with abrasives, the largest group had worked in the Vitreous Bonding department Proportional mortality analyses for the final study group and for those who had worked in each of the three departments in wNch significant PMR excesses were seen (Abrasive Preparation, Vitreous Bonding, Organic Bond ing) are shown in Tables 2 to 5. Since the PMR analysis uses the US. population as a standard, each of these de partment groups was also examined for selected diseases by the case referent method. The referents come from the Journal of Occupational Modiane/Voi. 23, No. 11/November 1981 749 UCC 096867 Tabte i. - Ntiteber if Parians Ever Empteysdand Empteyed ter Tan or Mare Yaare Accenting it Department. * died with diseases at special interest One died of pneu moconiosis, one of chronic interstitial pneumonitis, and two with silicosis (in one case as the underlying cause and 'i "I * Empteyed 10 Ever EtepMyed r Mere Yean in the other as an associated cause of death). If those who died of silicosis were excluded, the elevated odds ratio for Abrasive preparation Vitreous Ponding 155 98 those with 20 or more years experience was still signifi 322 188 cant Organic bonding Packing/snipping Factory (nonabrasive) Service Miscellaneous 140 56 Five of the 26 jobs identified were placed in the 117 59 Abrasive Preparation category. These were crushing 495 270 332 114 silicon carbide, crushing aluminum oxide, sizing abrasive 171 67 * grains, storing and inspection of grains and miscellaneous. The two workers with silicosis had worked for 28 and 29 Sublets are counted more man once if they worked In more than one derailment years, respectively, in the crushing of silicon carbide The silicon carbide was known to contain a low level of tree same group as the cases thus avoiding the use of the U S. silica but never more than 1 % population as the comparison group, as in the PMR Since Vitreous Bonding. -- Focus was next directed to those nonmalignant respiratory disease or malignancies oc who had been employed in Vitreous Bonding and had curred in excess in one or more of these departments, the died from nonmalignant respiratory disease or esopha referents selected for all three were those who died from geal cancer (ICD/7 = 150). There is an elevated odds ratio causes other than nonmalignant respiratory disease for nonmalignant respiratory disease for those 17 subjects (ICD/7 460-519) or malignancy (ICD/7 140-209). employed 20 or more years (OR = 1.7; p < 01). In con Abrasive Preparation. -- The authors were informed by trast to the findings for Abrasive Preparation, no trend a plant representative that the greatest dust exposures was apparent Employment in specific jobs within this had occurred in the Abrasive Preparation department department did not distinguish those who died of non Those who had worked in that department and who died malignant respiratory disease from those who died of alt with nonmalignant respiratory disease (ICD/7 460-519) as other causes. the underlying cause were compared with those who died As in Abrasive Preparation, for most of those with non from all other nonmalignant causes (fable 6). The odds malignant respiratory disease pneumonia was the cause ratio was found to be elevated and especially so for the of death. There were 6 other causes of death of interest (3 group which had worked in Abrasive Preparation for 20 or with pneumoconiosis unspecified and 3 with silicosis). more years. A trend in the ratio is apparent with increas Those with pneumoconiosis unspecified had worked 17, ing duration of employment in the department 22 and 25 years, respectively, m those jobs which pre Cause of nonmalignant respiratory disease death for pared the vitreous bonded wheels for firing in the kiln those in this department was predominantly pneumonia (mixing, molding, and shaving). One of those with silicosis or bronchopneumonia (11 deaths). There were four who had worked ten years firing the kilns. The second had Tabto 2. - Freparttonal MortsMy Analysts an Ftete Brenp. Observed IxpttM MM 98% Cl* All infection All cancer Digestive organs Esophagus Stomach Large intestine Rectum Liver Pancreas Respiratory system Lymphopoietic and hematopoietic Lymphosarcoma and rmcutaearcama Hodgkin's Leukemia and aleukemia All nervous system An cardiovascular All respiratory Asthma Pneumonia Emphysema Ail digestive system Cirrhosis Alt external I Accidents Residual 13 10.9 119 193 173.7 111 92 55.5 166 136 202 11 4 2 264 150 465 19 12 0 158 101 246 27 16.8 160 111 232 16 6.8 234 146 376 9 43 209 111 395 & 10.1 79 42 51 2 82 21 16.0 131 3 39 77 8 1.7 348 163 732 8 7 0 114 83 96.3 87 494 476.2 104 73 56.7 129 103 161 4 2 7 147 45 24 0 188 142 249 13 18.9 77 37 40.7 91 16 16 7 96 44 59.t 74 53 103 31 41 9 74 31 55.4 *95% confidence intervale ` For brevity oMy mow wgmflcantty elevated or Depressed wt Mod m fhte tatte i* * 968 wnrte mates with work tfetents ! 750 Causes of Death Among Employees of a Synthetic Abrasive Product Manufacturing Company/Wegman and Eisen ! UCC 096868 TaMt 3. -- PrepertiMat Mortality Analysis on Those Ever Employed In Abreeiw Preparation. * an cancer Oigestrve organs Esophagus Stomach Largo mtestine Rectum Liver Pancreas Lymphopoietic and hematopoietic Lymphosarcoma and reticulosarcoma Hodgkin's Leukemia and aleukemia Ail cardiovascular All respiratory Asthma Pneumonia Emphysema Remainder Observed 20 12 2 3 3 2 J "1 0 0 0 0 81 20 1 11 2 34 Exported 28.1 90 0.7 1.9 2.7 1.1 0.7 1.6 2.6 0.6 0.3 1.1 76,6 94 04 3.7 2,8 40.9 nut 71 134 291 156 111 181 143 61 0 0 0 0 106 212 225 294 72 99% Cl* 139 324 169 - 512 `95% confidence intervals * For brevity only those significantly elevated or depressed are listed on this table it = 155 white males worked 30 years firing, as well as 8 years in other jobs in this department. The third had worked only one year in this department but had worked 29 years in Abrasive Preparation The odds ratio analysis for esophageal cancer was elevated (OR = 2.5; p < 01). Two of the six who died of esophageal cancer and had been employed in this depart ment had worked predominantly in mixing, molding, and shaving of vitreous wheels (22 and 36 years) and a third had worked one year at that location and 36 years at crushing and sizing of silicon carbide. The other three had worked predominantly at the kilns. Two of the three had worked exclusively at the vitreous wheel kilns (28 and 35 years, respectively) and one had worked eight years at those kilns and 22 years at the refractory material kilns (a separate department). Organic Bonding. -- Lastly, subiects were distributed according to employment in the Organic Bonding depart ment and deaths due to nonmalignant respiratory disease or digestive cancer (ICD/7 = 150-159). There is an elevated odds ratio for the 17 who died with nonmalig nant respiratory disease, which is significant (OR = 1.9; p < 0.05). No trend, however, was seen for increasing length of employment with 13 employed less than 10 years. When digestive cancer as a whole was examined, an ex cess was noted for those 8 subjects employed 10 or more years (OR = 1.6; p < 0.05). The only specific type of digestive cancer appearing in more than two subjects was rectal cancer, with five cases The odds ratio is significant ly elevated for those four employed more than ten years (OR = 8.4; p < 0.01) Two of the five subjects who died of rectal cancer and had worked in this department had worked predominantly in mixing, molding, and shaving of organic bonded wheels (10 and 44 years, respectively) one had worked predominantly in truing and post-firing inspection and two had worked in unspecified jobs in this Tabid 4. - Puparttaeii Mortality Analysis an Thnan Ever Employed in Vitreous Banding. All cancer Digestive organs Esophagus Stomach Large intestine Rectum Liver Pancreas Lymphopoietic and hematopoietic Lymphosarcoma and rwiariosarcoma Hodgkin s Leukemia and aleukemia All cardiovascular All respiratory Asthma Pneumonia Emphysema Remainder Observed 64 28 6 6 7 2 2 5 8 1 3 2 152 30 2 15 6 76 Expodftd 58 2 18.8 14 41 57 23 15 3.4 5,1 13 0.5 2.3 1612 204 09 8.1 5.9 87 2 rant 110 149 424 148 123 86 137 147 16 80 816 88 94 147 223 185 102 '95% confident interval*.4 For brevity only most HgreftcaMtyottvatM or dapressad are Nttod on tins tabu n - 322 wtnta males Journal of Occupational Medicine/Voi. 23, No. Tl/November 1981 98% Cl* 104 213 204 883 102 - 207 113 302 751 UCC 096869 TaUa S. - Ptapartaml Martatty Analyst on Ttwsa Em Emptayed in Organic Banding. All cancer Digestive organs Esophagus Stomach Large intestine Rectum Liver Pancreas Lymphopoietic and hematopoietic Lymphosarcoma ana reticulosarcoma Hodgkin's Leukemia and aleukemia All cardiovascular All respiratory Asthma Pneumonia Emphysema Remainder Obtaread 35 16 3 1 3 5 > 2 3 1 4 2 59 17 2 9 5 29 EapaeM 25.1 81 o.e 18 2.4 1.0 0.7 15 23 05 0,2 1.0 698 8.5 0.4 3.5 2.4 36.6 HM 140 196 488 56 124494 309 137 353 180 1 710 210 85 200 460 259 207 86% Ci* 104 - 188 123 - 313 - 225 - 1.086 185 - 671 114-300 139 - 481 *95% confidence intervals.* For brevity only those significantly elevated or depressed are listed on this table n = 140 white males department Four of the 968 subjects had rectal cancer as a secondary cause of death All had worked exclusively in jobs involving nonabrasive materials. The proportional mortality analysis (Table 2) indicated an excess of Hodgkin's disease. Four of the six with Hodgkin's had worked in Organic Bonding, three in the same job category (mixing and molding of the abrasive products) for 1, 29 and 33 years, respectively. The fourth had worked three years in the truing and inspection of the kiln-fired products. Three of these four had worked also in their same |Ob categories in Vitreous Bonding. The two who had never worked in Organic Bonding had worked in maintenance or machining. Ethnicity. -- Since the incidence of certain diseases differs by ethnic group, the proportion of Scandinavians was examined by work group and cause of death. Jobs were grouped into the seven departments, ethnicity was grouped as Scandinavian or other and cause of death was grouped into digestive cancer, other cancer, nonmalig- nant respiratory disease and ail other Log linear analysis was performed to evaluate the significance of the interac tion between cause of death and ethnicity. The interac tion was found to be not significant (p = 0651 which suggests that Scandinavians were represented proportion ately within each of these major categories of death. The excess of nonmalignant respiratory disease or digestive cancer is unlikely to be explained by overrepresentation of Scandinavians. DnaMHon Natural abrasives for sharpening, grinding, and polish ing have been used for over 3,500 years These abrasives, mostly derived from siliceous rock, unfortunately have also been responsible for pulmonary disease, especially silicosis. At the end of the 19th century, the first commer cial synthetic abrasive, silicon carbide, was made. Shortly afterward, aluminum oxide, which had been available naturally, was first prepared synthetically The greater hardness of these two synthetic abrasives compared to sandstone, as well as the belief that they were physiologically inert, led to their replacement of sandstone by the 1920s Review of the literature concerning health effects of silicon carbide and aluminum oxide indicates that only a minor effort has been made to evaluate long-term health effects of these two materials The earliest report comes from a government study in Great Britain published in 1923.' Those performing metal grinding with abrasive wheels containing silicon carbide or aluminum oxide had much less pulmonary disease then those exposed to silica However, there was "clinical" fibrosis in 18 subjects Investigations among aluminum producers and china *$M (Ml (P 8) tOdds ratios are compared now tworkad group t95H cwtidanca Intarvals* 752 Causes of Death Among Employees of a Synthetic Abrasive Product Manufactunng Company/Wegman and Eisen UCC 096870 woritw, both exposed to finely powdered alumina, showed no x-ray or clinical evidence of pneumoconio sis * * ' However, significant fibrogemc activity of rela tively heayy exposures to gamma aluminum oxide has been reported from animal studies." " In a study which examined workers exposed to silicon carbide during its production, early x-ray changes and mild symptoms, accompanied by a rather marked loss in vital capacity (mean of S20 mis over 6 years), were found in 32 workers." Animal experiments with silicon carbide suggested it might promote tuberculosis but that it pro duced little, if any, fibrosis.1415 A report in 1925'* from the same plant examined in the current study summarized the experience of 137 persons exposed from 10 to 42 years. Forty-two showed no x-ray evidence of silicosis, 12 showed slight changes, 77 showed first-stage and 6 (four with no prior silica exposure) showed second-stage silicosis. An earlier report from this same plant'4 included causes of death reported by the mutual benefit associa tion between 1892 and 1924. Tuberculosis was not elevated as might have been expected if silicosis were prevalent. However, 20 out of 91 deaths were due to pneumonia, excluding epidemic influenza. A review of Massachusetts mortality records (between the years 1890 and 1925) for males 20 years or older showed that 9 to 12% of deaths were due to pneumonia The proportion of deaths due to pneumonia reported for employees there fore is twice the expected number. The later report" up dated the study of deaths and found that three of the seven subsequent deaths were due to pneumonia. Another artificial abrasive plant" received aluminum oxide and silicon carbide to dry, crush, size and bag for shipment Of the 52 employees who were x-rayed, 15 showed some pulmonary fibrosis, which in seven cases was classified as 'the nodular type resembling silicosis." Five of the seven had had no other inorganic dust ex posure Dust exposures were extremely high (in process ing, 12 of the 13 samples were over 50 mppcf, 7 were over 100 mppcf) This review and results of the current study suggest that exposure to synthetic abrasives may be associated with adverse effects on the lung Although cases of pneumoconiosis were reported in the current study, the predominant cause of death under All Respiratory Dis ease was pneumonia This is not a very reliable death cer tificate diagnosisReview of the individual death cer tificates did not reveal any miscoding but pertinent clini cal information which may have changed the underlying cause of death may not have been recorded. In addition, death assigned to pneumonia as the underlying cause may be over-reported in Massachusetts,'4 resulting irran elevated PMR compared to national proportions Regard less, if the pneumonia excess were simply a matter of misclassification of cause of death, the excess odds ratio in Abrasive Preparation should not have shown a trend with increasing employment Furthermore, the excess pneumonia reported in the literature for this company supports a possible work association. No prior published report of cancer excess in a work population was found. The digestive cancer excesses identified in this study may have resulted from possible confounding factors Cigarette smoking and alcohol consumption are poten tial confounders for which almost no data were available. Smoking might contribute to pneumonia susceptibility and is a known risk factor for esopfiageal cancer Alcohol intake is also a risk factor in esophageal cancer No sys tematic information on smoking habits of the deceased is available. Cancer of the respiratory system, however, is not excessive, which suggests against differential smoking habits as an explanation of the excesses. Mortality rates for esophageal and rectal cancer are 20% and 14% higher, respectively, than U S rates in the county where this manufacturing plant is located. This difference may account for some of the excess revealed in the PMR analysis. The use of the internal control m the case referent study, however, uses persons from the same area to estimate an expected proportion of deaths Using cancer mortality experience in a city near the one under study to calculate an expected value similarly did not alter the findings of an excess. In summary, the potential confounding factors con sidered do not appear sufficient alone to explain the ex cesses of observed cases of cancer of the esophagus and rectum. The excesses are based on small numbers and further comment is not indicated. The excess of nonmalignant respiratory disease found in this study and the association of the excess with dura tion of employment suggests a work-related health effect on the lung. This is supported by the few reports on health effects in those exposed to similar synthetic abrasives. It should be noted that the excesses in this study population reflect exposures of many years ago when levels may have been less well controlled. Of particular concern, however, is the fact that these abrasives have many and varied uses throughout modern industry Abrasive grind ing and polishing, for example, are found in all metal machinery and foundry operations, in many parts of transportation equipment fabrication and in maintenance and mechanical repair occupations Therefore, if present, such a risk could be widespread. The cancer excesses noted are based on numbers too small to permit interpretation of their relationship to this work environment The limitation of the proportional mortality analysis approach and the assumptions necessary to accept the odds ratio analysis have been noted. Further evaluation by respiratory morbidity studies, as well as mortality studies which expand the study population in time or size and use x standardized mortality analysis, should be informative. We with to thank Kim Schmidt Quit. Robbt* l*uter looile Pothier Derma and Susan Stood** for their assistance in carrying out this study References 1 Moruon RR. Analysis of relative survival and proportional mor tality Comput Biol Med 7 325-332, 1974 2 World Health Association Manual of the International Statisti cal Classification of Diseases. Iniuries and Cause of Death. 19S5, Revision, 1957 3 Miettinen OS and Wang ID. An alternative to the PMR Am / Epidemiol (m press). 4 Mantel N and Haeruel W Statistical aspects of the analyses of data from retrospective studies of disease I Natl Cancer Inst 22 719-74*. 1959. 5 Bishop YMM. Ftenberg SF, and Holland PW Discrete Multivariate Analyst* Theory and Practice. Boston: MIT Press. 1975 Journal of Occupational Medicine/Vol. 21 No. 11/November 1981 753 UCC 096871