Document 2J7EqDdVvJqw8B7JG2YjBzRO7

REYNOLDS ALUMINUM REYNOLDS METALS COMPANY RICHMOND. VIRGINIA 23261 1980 December 10 Mr. Thcnas B. Bcnney Aluminum Company of America 1501 Alcoa Building Pittsburgh, Pennsylvania 15219 Dear Torn: Please find my second draft of the section on contaminants in our write up of the sampling and analytical procedures in the aluminum reduction industry. After the recent Washington meeting of the task group, substantial revision of my initial draft was proposed. In addition, a new section dealing with the ingot plant as well as a table listing the contaminants and possible ranges of exposure was to be included. Please look at the section on the ingot plant and the list of contami nants especially regarding the range of exposures to be very preliminary and subject to change. As you know, other members of the task force were going to send me the range of exposures for their particular operation. Also, Mr. Martyn was to provide a detailed description of the processes. Since neither of these itans has been accomplished, I attempted to draw them together using Tan Walker's list of contaminants as indicated in the initial IPAI Health Protection Seninar in Copenhagen. Task group members should feel free to make whatever changes they deem necessary in this draft and I will try to have it reworked by the IPAI Health Caanittee meeting in January. Thank you very nudi. Sincerely, HMC:jet cc: Mr. Peter Martyn Dr. Alfred Steinegger Mr. Eirik Nordheim Mr. Ihcmas J. Walker Haner M. Cole Manager, Environmental and Industrial Hygiene TX TIMER RMC0021012 DRAFT - 1980, December 09 SAMPLING AH) ANALYTICAL PROCEDURES IN THE ALUMINUM REDUCTION INDUSTRY AILMINtM PRODUCTION PROCESSES Introduction Aluminum is produced by using the Hall Heroult process in which alumina is electrically reduced to aluminum. The alunina is dissolved in a molten bath (approximately 950 C.) of fused salt electrolytes consisting primarily of cryolite with minor additions of other salts, mainly fluorides. The entire process is carried out in a reduction cell consisting nainly of a carbon anode, a cathode, and the electrolyte contained in a carton-lined steel shell. After reduction, the metal, being more dense than cryolite, is deposited at the cathode. Oxidation of carbon is primarily to carbon dioxide, but with sane secondary production of carbon monoxide occurring at the anode. Conse quently, the anode carbon is continuously being consisted during the reduction of alunina. There are currently two types of processes (prebake and Sodesberg) being used in aluminum reduction. Both processes contain basically the same type of cathodes, i.e. a steel shell lined with an insulating material and a carbon layer approximately 12__inches thick. The main difference is in the design of the carton, anodes. Prebake Process In the pcebake process, anode blocks (previously baked in the carbon plant) are suspended along the entire length of the pot. Aluminun or copper electrical connector rods, used to support the anode blocks in the pot, are attached to the anodes in the rodding roan. Prebake pots may be of two types depending upon the method in which the alimina is added to the pots. The first type is the "center brake pot" in vhich the alunina is fed into the center of the pot. TX TINER RMC002101 -2- The seccnd type is the "side brake pot" in which the alunina is added in the interval between the outside of the anode blocks and the inside walls of the pots. In the prebake potroans, several emissions are produced during the reduction process. Particulate and gaseous fluorides are emitted from the elec trolyte? sulfur dioxide originates from the sulfur content of the coke used to make the anodes. Carbon monoxide and carbon dioxide emissions result from the oxidation of the carbon anodes. Nuisance dust (alumina and carbon dust) is released into the workplace during the normal operation and maintenance of the pots. As mentioned previously, the anodes are pure carbon and, consequently, contain no ENAs. There my, however, be other sources of PPCM such as the line cathode baking process (coal tar pitch used as a birder in the cathode mix) or pitch used as a binder to secure the anode to the electrical connector rod. In the prebake carbcn plant green mill, sized calcined petroleim coke is proportioned alcng with reclaimed anode material and approximately 25 percent ooal tar or petroleum pitch in a batch or continuous mode. The materials are then blended and heated in a mixer to a temperature of approximately 160 C. The mixture is then transported via a conveyor to an anode press (mixture temperature about 135 C.). Here, the mix is pressed into carbcn blocks by the hydraulic anode press, or vibrated. Lubricant may be sprayed onto the molds to prevent sticking of the anode paste mixture, The green blocks are then packed in layers in the carbcn baking pits, with coke car anthracite goal packed between the layers of blocks and the refractory flue wall, to provide mechanical support and heat transfer, as well as to prevent oxidation during the baking of the carbcn. The uppermost layer of blocks is also covered with a blanket of packing material. The blocks are then slowly heated to a temperature of approximately 1200 C. They are held at this temperature for approximately 40 hours and then slowly cooled to a torperature of about 160 C., before they are removed frcxa the carbcn furnace. During this baking period and cooling cycle, which may be extended TX TINER RMC00 21014 -3- over several weeks, pitch volatiles fran the baking blocks are drawn into the canfcusticn flue through open spaces in the refractory brickwork. During the entire baking and cooling cycles, the baking pits are operated under a negative pressure, to contain the pitch volatiles and bum thesn for the available fuel value. As a result, very little volatile hydrocarbons are released into the workplace in the carbon bake area. The baked blocks are then cleaned to remove minor amounts of the packing material adhering to the blocks. These prebaked blocks are then transported to the rodding roan, where the anode connector rod is attached. The blocks may also be spray-coated with aluminum, to prevent oxidation. The finished anode assembly is then ready for installation into a pot. Occupational exposure in the carbon bake area would be to carbon monoxide, sulfure dioxide, fluorides, and nuisance dust. Carbon mcrttxide results from the incomplete combustion of organic materials (e.g. natural gas or oil used to fire furnaces, as well as the coal tar pitch binder) Sulfur dioxide is produced fran the sulfur present in the coke or coal used to make the blocks. Fluorides are generated fran the fluoride contained in recycled anodes. Finally, nuisance dust, primarily carbon, is produced by the handling of coke and used anode blocks. Soderberg Process In the Soderberg process, the anode, which ccnsists of petroleum coke and coal tar pitch binder, is of the self-baking type. The anode mix is added to the top of the anode superstructure as a viscous paste car briquette. The anode is continuously lowered to amintain the desired distanoe from the cathode. As the paste mix gradually lowers into the pot, its temperature rises until it reaches that of the bath at which time it has been completely baked to pure carbon. Iron rods or steel pins are inserted into tte soft anode to carry electrical current to the pots. The pins may be arranged either vertically or horizontally. TX TINSR RMCOO 2101i -4In the Soderberg potroans, a variety of emissions are produced. Particulate and gaseous fluoride are emitted from the electrolyte? sulfur dioxide originates fran the sulfur found in the coke, which comprises the major portion of the anode. Carbon monoxide and carbon dioxide emissions result from the oxidation of the carbon anodes. Coal tar pitch volatiles are produced during the anode baking porcess, and nuisance dust (alunina and carbon dust) is released into the workplace during the normal operation and maintenance on the pot. In the Soderberg paste plant, calined petroleum coke is ground in ball mills to the proper sizes. Ibis raterial is mixed with 25 to 26 percent coal tar pitch (softening point is approxirately 120 C.) and blended in a heated mixer. After mixing, the anode paste is dunped into the anode paste bucket at a temperature of 160 to 190 C. The raterial is then transported to the potrocm, where it is added to the top of the pot by an overhead arane. Exposures in the Soderberg paste plant include coal tar pitch volatiles, coal tar pitch dust, and carbon dust. The coal tar pitch volatiles found in the Soderberg paste plant are generally of a lower molecular weight than those found in the Soderberg potrocm. Cathode Lining The cathode raterial is produced by mixing sized anthrocite coal with approximately 13% rrv*'[ tar pitch or petroleum pitch binder. The mixture is blended and heated in a TM'**** to a temperature of approxirately 135 C. The material is then transported- to the pot repair area where it is used for cathode relining. Exposures in the cathode lining operation are mainly low molecular weight hydrocarbons and nuisance dust. Minor amounts of PNA's ray be released during tte rarrming process. TX TINER RMC00 21C16 i - 5v Maintenance In this complex industry systanatic and effective maintenance of plant and equipment is essential. Pollution abatement systems such as fune collection hoods, burners, etc., nust be kept in good working order to insure that process workers are given the roaximm protection. Special attention should be paid to maintenance workers to minimize undue exposure to harmful vapors, gases and dust when carrying out repair or changes to process our fume control equipment. Although sene occupational exposures to maintenance personnel are unique to the particular process, many are quite similar {e.g. welding funes, coal tar pitch volatiles, dust, etc.). The following is a brief description of the major maintenance operation at an aluminum reduction plant. Remarking and/or replacement of cathode shells requires the removal of old bath and spent cathode from the shell. Water is normally aided to the pot at this stage for cooling purposes as well as loosening up spent pot lining material. Pot shells may then be dunped or dug out using a pneumatic pot digger. The shell is then cleaned by abrasive blasting so that welders may repair cracks and patch holes. Exposures during this process include anrnenia generated by the addition of water to the pots as well as nuisance dust and fluorides from the old bath and cathode. Abrasive cleaning of the shell usually results in exposure to either crystalline silica or alumina. Finally, welders are exposed to a variety of dust and welding fumes associated with mild steel welding. Electrical connections (aluninun and copper bus bars, flex, and prebake rods, etc.) are continually being replaced and repaired. Exposures during this procedure are mainly to copper dust and fiane. Cleaning and reworking of the used anode pins and channels involve abrasive cleaning and straightening. Consequently, workers may be esqaosed to dust of carbon alunina and fluorides. TX TINER RMC00 210 17 -6- Finally, repair of the fume control system results in a variety of exposures. Ducts must be continually cleaned to prevent accunulaticn of material which reduces the effectiveness of the ventilation system. In addition, continued cleaning and repair of scrubbers (Wet Electrostatic Precipitators, Dry Electrostatic Precipitators, etc.) oust be performed to maintain the necessary collection efficiency for the various contaminants. Employees involved in these jobs can be exposed to nuisance dust, fluorides, carbcn monoxide, and coal tar pitch volatiles. Ingot After the molten metal has been removed from the pot (tapped with a tapping crucible) it is taken to the casthouse where the metal is prepared for distribution to the customer. The raw pot metal may be sent directly to the customer in the molten state. The molten metal my also be pigged directly into 1,000 and 2,000 lbs. molds. I2he final possibility is for the metal to be added to a holding furnace where it is alloyed and fluxed to remove impurities and degas the metal. Ihe fluxing process, which uses C/2 or a mixture of chlorine and other gases, causes the inpurities to float to the top of the molten metal. Biis dross is removed prior to casting and stored in a dross house until it is oool. After the skinning has been ccnpleted the metal is cast into ingots, billets, or small pigs. Exposures in the casthouse would be to fluorides, chlorine, chlorites, axmmia, phosphine, nuisance dust, oil mist, and possibly alloying agents. Fluorides are produced from the cooling dross material due to the presence of pot bath material. Chlorine exposure results from the use of chlorine as a ocnyaMaiL in the fluxing gas. Consequently, chlorides, reaction by-products of the chlorine, are also emitted from the cooling dross. Armenia is produced from hot dross when it cones in contact with moisture. *Nuisance dust is also produced from the dross as well as scrap handling. Oil mist is emitted from TX TINER RMC0021018 -7- the billet, saws which utilize the oil as a cutting lubricant. *Metal exposures may occur fran the various alloying agents being used, e.g., magnesium, chranium, etc. . Phosphine may also be produced fran the hot wet dross if there is some arsenic present. `Asbestos exposure can occur due to the use of marinite in the molten metal transfer troughs. Ibe marinite must be machined and sized for each particular casting operation. Asbestos exposure may also occur in the siphon cleaning and repair shop where siphcns are cleaned with a pneumatic router, and asbestos gaskets removed and installed on the crucible lid and siphcn. TINer RWC0021Q19 AIRBORNE contaminant AUKBM4 AUMDXM CXHE AMCNIA ASBESTOS sefvllhw OCKZU4 DUST/RME CARBCN DZCDare CARBCN fCNCKTOE CHLORINE & CHL3UEES CXL TAR PITCH VCLA3ILES COPPER DUST/RME ELUCRIBES (Gaseous, liquid 4 particulate) HYDROGEN OHOUZE HYDROGEN FLUORIDE HBCK3ttRCETHANE LEAD MUGANESE mas* Nl'lRUjfcN DIOXIDE MXMESHM CDOCC RME NUISANCE DUST (Coke, fibrous glass, mineral wool fiber) OZONE ncsac HOSHCNE SILICA (Amorphous) SILICA (Crystalline) SUUTJR ECOCIDE WEEDING RME HUNCIPAL CEE CR SOURS OF MSITM By-produet froa droaa 4 H^O Insulation; cewerings; lagging rtaterlals; nerinite; molten metal ccntrol; brake linings Silver soldering; brazing Rotroane; anode baking pits; centusticn sources By-product of incomplete oorisusticn in: potrcoie, furnaces, internal acntousticn engines; inert gas furnaces; enclosed spaces, tanks and silos Fluting in casting opera tions? water traatiiait Binder for electrode* Anode rods and flex; electrical ocmectors; metal alloying; brazing cryolite plant; potroane: pot repair, solid fluxes, metal treatment; welding flux; dross recovery casting; (formed frcm chlorine in presence of water); dsninaralizers Alloying agent Metal alloying Rectifiers; laboratories Alloying egent By-product in electrical discharge end welding Qaifeustloi product of halides; Note: Ns welding in confined space in presence of chlorinated solvent or other halide Gaseous by-product of dross and water Sandblasting By-product of cartousticn of coke, coal, anodes, oil Bid gu; waste water treetsant By-product of welding? principally osane, cotides of nitrogen end natal (a) welded TYPICAL WPTM3RE RANO? <1-30 ngV <1.0-40 ppn <0.1-50.0 fibers/cc <0.01 mg^a* <5-1200 ppa* <0.05-10 ip <0.01-2.0 BQ/at <0.01-20.0 rtg/n? <0.01-50 mgrtrf <0.01-20.0 mg/ta* <0.01-5.0 ppm <0.01-200 ug/ta* <0.01-5.0 agM <0.01-0.30 mg/ta* <0.01-5.0 ppm <0.01-15 reg/B? <1.0-50.0 ng/m1 <0.01-5 ppm <0.01 pgm <0.01-1.0 pgsn <1-20 ngM <1.0-200 vg/m* <0.1-5 ppm <0.01-30 mgfa} KMC0021 o \PA\ International Primary Aluminium Institute Ntw Zasignd Houm (9th Floor) Hoymorktt London SW1Y4TE Unrt*d Kingdom TUdphont01*930 0528 Tttox 917837 IPAI London 2nd October 1987 To all Members of the IPAI Health Committee Copy to the Chairman, IPAI Environmental Committee Report: "Mortality of Aluminium Reduction Plant Workers in France" Please find enclosed a copy of the above report mentioned by Dr J-P Coulon at the recent Sixteenth IPAI Health Committee Meeting. Peter Martyn Secretary General Enc INDUSTRIAL HYGIENE RECEIVED TX TIN RMC0020 A comwny fcnwtaa Ov guaranca* Rogotorod London. no. 1052007 Rogaurod orffieo 9 CMinrii London 6C2V 6A0 UW Kingdom C*J CO internetionel Journal of Egidemioiogy $lnt*mitionl Epidemiological Association 1987 Voi 16. No- 2 Printed m Great Britain Mortality of Aluminium Reduction Plant Workers in France J M MUR*. J J MOUUN*. C MEYER-BISCH*. N MASSINV J P COUIONT AND 4 LOULERGUEt Mur j M (Institut National de Recherche at da S6curit6, Avenue de Bourgogne 54500--Vandoeuvre lee Nancy, France). Moulin j J. Mever-Biseh C. Masam N, Coulon J P and Loulergue J. Mortality of aluminium reduction plant workers in France. International Journal of Epidemiology 1987. 18; 257-264. The mortality between 1950 and 1976 of 6455 French aluminium plant workers was analysed in order to assess occupational risks (especially lung cancer! associated with electrolysis, particularly with the SoderOerg process. Mortality from all causes (SMR 0.85). was lower m this cohort than in the French male population ('healthy worker effect'), and cancer mortality (SMR - 1.09) was only slightly higher. There was an excess of mortality from accidents (mainly non>occupationai) m electrolysis workers (SMR 1.38) and from cirrhosis of the liver in maintenance workers (SMR - 1.63). Among electrolysis workers, only those who had worked less than 10 years had a relative excess mortality from lung cancer (SMR 1.94), but this did not seem to be associated with a particular electrolysis process. However a substantial underlying risk of lung cancer m Soderberg workers could not be excluded, although such a risk appeared unlikely for prebake workers. Several epidemiological surveys, analysed by Doll1. Enteriine; and Simonato'. have been earned out in order to evaluate occupational nsks in the aluminium industry. In particular the risk of iung cancer was suspeered because certain production techniques, such as the Soderberg electrolysis process, generate in the working environment polycyclic aromatic hydrocar bons."' some of which (benzo(a)pyrene) are known to be experimental carcinogens." The results of several of these surveys seemed to confirm this hypothesis, although the risk of lung cancer could not be firmly ascertained. Furthermore, some of these surveys sug gested that there might be other risks for the workers in this industry; other cancer sites (brain, pancreas, liver, lymphohaematopoietic cancers) and other non-neo* plastic diseases (emphysema, cirrhosis of the liver). In order to provide additional data about all these suspected risks, a mortality survey was initiated, for the first time, among workers at all aluminium reduc tion piants in France. MATERIAL AND METHODS Aluminium Processing Aluminium is produced by electrolytic reduction of alumina/ A tank, whose internal carbon-coated wall is the cathode, contains the electrolysis bath, made of alumina and of a fluxing agent (cryolite). Differences * Insmut National de Rcchcrcflc et de Sccuntc (INRS). Avenue dc Bourgogne--5J50O--Vandoeuvre les Nancy--France, r Aluminium Pcchincy. 22 nic Balzac--75008 Pans--France between the electrolysis processes essentially depend on the type of anode used. These anodes, made from a mixture of coke and of coai tar pitch (sometimes of petroleum pitch) can be: --either baked (in ovens) before being used: in this case, they are solid blocks immersed in the electroly tic bath ('prebake anodes'): --or raw (Soderberg process): a paste is continuously supplied and earned to the electrolytic bath: in con tact with the bath, which is at a high temperature (950C). the lower pan of the Soderberg electrode is baked (`self-baking anodes'). Survey Plants The survey was conducted in all 11 French aluminium plants (Figure 1). For historical reasons (hydroelectnc power generation), these are located in two moun tainous regions of France, the Alps and the Pyrenees. The plants notably differ from one another in age. manpower, technology, and industrial hygiene. More over. certain features have evolved with time, especi ally electrolysis techniques and industrial hygiene. The Soderberg process which was used onginaily. has par tially or totally been replaced by the prebake process. Pollution was gradually reduced by means of gas col lection hoods on potrooms and work in tight booths. Population We conducted an 'historical prospective study' with the cohort including all men who had worked for at least one year in one of these plants between 1950 and 1976: 257 TX TINER RMC00 20 3i> 258 INTERNATIONAL JOURNAL OF EPIDEMIOLOGY Plants Year of beginning Process PB . Prtbake S SoCerberg Size of the cohort Rioupcroux 1926 PB * S 452 St Jean 4c Maunenne 190? PB * S 1532 Venthon 1926 PB * S 511 La Saussaz 1950 PB 182 La Prai 1925 PB * S 205 L'Argentine 1925 PB 572 Ncgueres 1959 S 779 Lannemezan 1939 s 747 Mercus-Auzai-Saoarr 1954 PB 1163 (Several plants) (300 Persons 6455 Person-years 113 671 Ficuttt 1 Aluminium reduction plana where the study was conducted. ie 6455 men. The cut-off date was 31 December 1976. The cohort was formed in the following way: all the workers who had left these plants, whatever the reason (illness, death, retirement, change of occupation), between 1 January 1950 and 31 December 1976. were added to the manpower still working in the 11 plants at the beginning of 1977. These subjects were identified from the files of the company. The total man-years of the survey was 113 671. Survey Method --The occupational history of each employee was reconstructed from the administrative records of the plant, listing as accurately as possible the various workplaces, the length of rime during which the subject worked there and the type of electrolysis process used. In a few cases, workers had been employed in industrial sectors other than aluminium electrolysis but reliable information about these occupations could-not be obtained. --Tobacco consumption was investigated by directly questioning the workers still working on 31 December 1976. For the workers who had left the plant, information was obtained by consulting the plants' medical records, or by questioning acquain tances (family, colleagues, close relations). This information could be obtained in only 49.1% of the subjects. --The vital status of the employees included in the cohort who were not working in the plants on 31 December 1976 could be obtained, for 99% of Frenchmen, by systematically consulting the registry offices of their birthplaces. For foreigners (9.7% Italians. 7.6% North Africans, and 9.6% other nationalities), enquiries were made at the relevant consulates and embassies. Results were not as good: vital status was known only in about 95% of cases. In total. 996 deaths were ascertained in this way. --In France, death certificates are completed without personal identification and thus cannot be used for mortality surveys. In this survey, causes of death were obtained in the following way. Firstly, local sources of medical information were sought, ie iden tification of the attending physician at the death, or evidence ofhospitalization was obtained (hospital or cancer treatment centre...): then, the causes of deaths were traced at these sources, ie by question ing the attending physician and/or consulting hos pital records. Such investigations are long and complex, but currently, they are the only wav in France to identify individual causes of death. Using this method, causes of 71.3% of deaths could be determined. They were coded according to the International Classification of Diseases (ICD: 8th revision)11* by a physician who knew nothing of the workers' occupational history. TX TINER RMC00 20 814 MORTALITY OF ALUMINIUM REDUCTION PLANT WORKERS IN FRANCE 259 T*jl I Monalitv in ihe whole cohort Actuat number o&servsd SMR 95^ confidence interval All causes of aeath Malignant tumours, all sites (ICD: 140-209) Diseases of the circulatory system (ICD. 390--*58) Cirrhosis of the liver (ICD: 571) Accidents, poisonings, traumas (ICD: E800-E999) 996 199 190 $2 129 0.85 1 09 0.81 1.13 1.27 0.80-0.91 0.97-1.22 0.72-0 91 0.88-1.38 1.09-1 46 Data Processing Because the cause of death was not always known, the number of deaths for each cause was estimated by multiplying the number observed for this cause by a correcting factor (number of deaths/number of known causes). The estimated number of deaths from each cause was compared to the expected number calculated according to the French male population"-11 after adjustment for age and year of death, in order to obtain the SMR (Standardized Mortality Ratio). The confidence intervals of the SMRs were calculated assuming a Poisson distribution of the estimated number of deaths for each cause (actual observed number multiplied by the aforementioned correcting factor): because these numbers are fractional, their confidence intervals according to the Poisson distri bution were estimated from the nearest integer values. This analysis was done for total mortality (all causes) and for the most frequent causes of death in France, ie cancers, and particularly, lung cancers (ICD * 162). upper respiratory and alimentary tract cancers (ICD 140-149, 161). cardiovascular diseases (ICD * 390458), cirrhosis of the liver (ICD * 571). 'violent' deaths (accidents, poisonings, traumas--ICD * E800E999). The mortality ratios were calculated in the whole cohort and in some subgroups defined according to workplace namely electrolysis, smelting and mainten ance. For electrolysis workers, mortality ratios were calculated according to: --duration of service and time since first exposure, ie less than 10 years, between 10 and 19 years. 20 years or more. These categories are exclusive, ie each worker is in a single category. --electrolysis process, ie prebake or Soderberg pro cess. The Sbderberg group' was formed of all work ers who had ever used the Soderberg process (including those who had aiso used the prebake pro cess at some time); the 'prebake group' was formed of aH workers who had only used the prebake process. RESULTS About two thirds of the workers had been hired before 1960 (more than 16 years before). One third had been working less than 10 years, one third from 10 to 19 years, and the remainder for 20 years or more. Analysis of Mortality in the Whole Cohort (Tables l and 2) Mortality from all causes is significantly lower (SMR * 0.85) than in the French male population taken as a reference. This deficit is paniculariy obvious for deaths from cardiovascular diseases (SMR * 0.81). On the contrary 'violent' deaths (accidents, poisonings, trau- ICD 140-149*161 150 151 152*154 155-156 157 162 170-P4 185 188 19! 200* 202 * 203 201-207 T*lx 2 Deaths from malignant tumours in the whole cohort Deaths from cancers: sice Actual number observed Upper respiratory and alimentary tract Oesophagus Stomach Intestine * rectum Liver, gall-bladder and hepatic bile duets Pancreas Trachea, bronchus, lung Bone. connective tissue, skin, breast Prostate Bladder Brain Lymphosarcoma and rcttculum-ceil sarcoma * other tumours of Ivmphoid tissues * multiple mveloma Leukaemia n 14 20 16 5 9 37 5 9 7 6 6 9 SMR 1.04 0.92 1.01 087 2.*5 1.49 1.14 2-04 0.87 2.09 M3 1.18 ! .56 9S% confidence interval 0.75-1.43 0 58-1.44 0.67-1.44 0.46-1.47 093-4.80 0.78-2.49 085-1.48 0.78--.Q0 0.31-1.45 0 96-3.68 0.98-4.07 0.55-2.28 0.81-2.61 TX TIMER RMC00 20 8i 260 international journal of epidemiology Table 3 Mortality according to workplace Electrolysis No. - 3245 D 492 (69.9<*) All causes of death Malignant tumours All sites (ICD: !40-20)) Malignant tumoun of trachea, bronchus and lung (ICD: 162) Malignant tumoun of upper respirator* and alimentary tracts (ICD: 140-149*161) Diseases of the circulatory system (ICD: 390-458) Cirrhosis of the liver (ICD: 571) Accidents, poisonings, traumas (ICD: E800-E999) 0.87 [0.80-0.951 (492) 1.09 [0.92-1.291 (99) 1.16 [0.78-1.65] (19) 1.00 (0.58-1.53) (13) 0.89 [0.74-1.04) (101) i.ll [0.79-1.521 (26) 1 38 [l. 12-1.66) (71) No. " Size of subcohon D * Number of deaths (% of known causes) SMR In each cell: [959fc confidence interval) (Actual no. of deaths observed) Maintenance No. 1963 D - 274(74.8<*) 0 81 [0.71-0.91J (274) 1.03 [0.81-1.27] (60) 1.05 (0.57-1.7]] (ID 1.44 [0.82-2.27) (12) 0.83 [0.66-1.02] (62) 1.63 [t. 10--2.20) (24) 0.95 [0.66-1.27] (3D Smelting No. 929 D - 107 (7| o^) 0 84 [0.68-l.0l| (107) 0.80 [0 50-1.17[ (17) 1.29 [0.51-2.62) (5) 0.31 [0.01-1.24) (l) 0.90 (0.61-1.22) (23) 1.76 [0.96-2.24) (10) 1.01 [0.61-1.56) (13) mas) are more frequent (SMR * 1.27) than in the national population. Mortality from cancer does not significantly differ from the reference population, for all site cancers together (SMR * 1.09) and for the lung (SMR * 1.14) and upper respiratory and alimentary tract localizations (SMR 1.04) which were the most frequently reported ones. SMRs for cancers of other sites are also not significantly different from 1 although close to the limit of significance for bladder (SMR * 2.09) and brain (SMR 2.13) cancers: however there were few cases of these types. Analysis According to Workplace (Table 3) The Standardized Mortality Ratio for all causes is sig* mficantly lower than 1 among electrolysis (SMR 0.87) and maintenance workers (SMR 0.81). The difference is not significant for smelting workers (SMR 0.84) who are less numerous. In spite ofthis low overall mortality, there is a signifi cant excess of deaths from cirrhosis of the liver among maintenance workers (SMR 1.63) and from violent' deaths among electrolysis workers (SMR 1.38). Detailed Analysis of Mortality among Electrolysis Workers (Tables 4 and S) Analysis of mortality among those who had worked for some time in electrolysis, according to length of service as potroom workers, indicates a relative excess mor tality from cancers (SMR 1.36). particularly lung cancers (SMR * 1.94) among workers who had worked less than 10 years. For those who had worked longer, cancer mortality does not significantly differ from the national reference. When taking into account rime since first employment (Table 5). no trend is observed in cancer mortality (all or lung cancers) according to latency. The increase in cancer mortality (all and lung cancers) observed among potroom work ers who have worked less than 10 years particularly concerns those who were exposed 10 to 20 years ago. Deaths from violent' causes are significantly increased in electrolysis workers whatever the length of service as potroom worker may be. but especially for those who were employed more than 20 years ago. Because of the difficulty of determining exactly what electrolytic processes were used in the past by the potroom workers, the analysis of mortality according to electrolysis process concerns only some (31 %) of the 0TX tiner KMC00203L, MORTALITY OF ALUMINIUM REDUCTION FLANT WORKERS IN FKANCE Table 4 Mortality among electrolysis workers according to length of employment and process 261 <10 years No. 1529 D 151 (66.9*) All causes of death Malignant tumour* All site* (ICD: 140-209} Malignant tumour* of trachea, bronchus and lung (ICD: 162) Malignant tumour* of upper respiratory and alimentary tract* (ICD. 140-149*161) Disease* of the circulatory svstem (ICD: 390-456) Cirrhosis of the liver (ICD: 571) Accidents, poisonings, traumas (ICD: E8QO-E999) 0.98 [0.82-1.!5| (151) 1.36 [100-l.82[ (31) 1.94 [1.00-3.331 (8) 1.13 [042-2.461 (4) 0.90 [0.63-1.23! (24) 1.09 10.57-2.05) (7) 1.54 |1.12-2.12| (27) No Size of subcohort D * Number of deaths (% of known causes) SMR In each cell: )95% confidence interval! (No. of deaths observed) Length 10-20 yean No. 904 D 142 (78.2%) >20 years No. 592 D 164 (70.1%) 1.13 (0.94-1.33! (t40) 1.28 (0.90-1.70) (30) 0.70 [0.20-1.90} (3) 1.49 (0.51-3.04) (5) 1.30 (0.95-1.67) (39) 1.23 [0.56-2.45) (7) 1.87 [1.15-2.69) (19) 0.95 (0.80-1.10) (164) 1.12 (0.82-1.46) (34) 1.08 (0.52-2.23) (6) 0.72 (0.18-1.71) (3) 0.81 (0.58-1.07) (32) 1.43 (0.77-2.35) (10) 1.77 (1.07-2.50) (17) Process Prebake No. 225 D * 19 (84.2%) Sdderberg No. 166 D - 78 (75.6%) 1.16 [0.6^-1.81] (19) 1.26 (0.43-3.07) (4) 0.00 10-5.27] (0) 2.00 (0-7.80) (1) 0.30 10-1.17) (1) 1.09 lO-t.26) 0) 3.06 (2,47-7.58) (8) 1.14 [0.90-1.431 (78) 0.76 (0.39-1.33) (12) 1.36 10.39-3.46) (4) 0.79 (0.09-2.88) (2) 0.99 (0.58-1.58) (17) 1.30 (0.48-2.84) (6) 1.20 10.66-2.02) (14) potroom workers: it indicates a significant increase only in 'violent' deaths among workers who have used the prebake anode process (SMR 3.06). Although not significant the SMR for lung cancer mortality is higher among 'Soderberg workers' (1.36) than among 'prebake workers' (0.00). DISCUSSION In France, the determination of individual causes of deaths is hindered bjra major obstacle, ie the law protecting personal privacy in general, and medical privacy in particular. Hence the National Mortality Register is anonymous and the release of information about a deceased patient by the attending physician is illegal. Of course, these measures protect the confiden tiality of information concerning private life, but they are a considerable hindrance to epidemiological sur veys and especially to mortality surveys. It is only thanks to the goodwill of the medical profession and of most of the health authorities, that we were able, with great difficulty, to determine 71% of the causes of deaths. The lack of information on 29% of the deaths may introduce a bias when mortality by cause in this cohort is compared with that of the French population. For instance, if some causes of death in the cohort are better known than others (which is likely), the inci dence ofthe former causes will be overrated and that of the latter underrated. However, we have established that the rate of known causes of death did not differ statistically significantly according to the main occupa tional factors analysed: workplace, electrolysis pro cess. length of employment. Thus, internal comparisons between subgroups may be more infor mative than the comparison with national rates. This is why more attention should be given in this study to the relative values than to the absolute values of mortality ratios. Thus, as an example, the comparison of the mortality ratios among smokers (defined as ever smokers) and non-smokers (never smokers) of the cohort actually shows the well-known pathological TX TTNER RMC00 20 817 262 INTERNATIONAL JOURNAL Of EPIDEMIOLOGY Table 5 Mortality of potroom workers according to length of employment and latency sutce first exposure Causes of death All causes Malignant tumours All localizations (ICD: 140-209) Maiifnant tumours of trachea, bronchus and lung (ICD: 162) Malignant tumours of upper respiratory and alimentary tracts (ICD: 140-149*161) Diseases of the circulatory system (ICD: 390*458) Cirrhosis of the liver (ICD: 5?1) Accidents, poisonings, traumas (ICD: E800-6999) (m each cell: SMR (95% confidence interval]) Latency (years) Length of employment at electrolysis <10 years 1(W0 years >20 years <10 10-20 >20 <10 10-20 >20 <10 10-20 >20 <10 10-20 >20 <10 10-20 >20 <10 10-20 >20 <10 10-20 >20 0.64 (0.18-1.621 0.93 (0.60-1.37] 0.94 (0.79-1.12] 0.00 (0.00-3.36] 2.18 [t.00-^.16] 1.25 [0.82-1.83] 0.00 [0.00-18.5] 3.00 (0.25-10.75] 1.57 [0.57-3.40] 0.0010.00*18.50] 1.33 (0.00-7.44] 1.29 (0.35-3.31] l.H (0.00-6.22) 1.30 (0.41-3.02] 0.78 10.48-1.21) 0.00 (0.00-9.25] 0.00 (0.00-2.44] 1.28 (0.52-2.65] 0.40 (0.00-2.24] 0.81 (0.30*1.77] 1.80 (1.10-2.79] 0.70 (0.32-1.33] 1.10 (0.93-1.29) 1.33 (0.37-3.40] 1.11 (0.73*1.62] 0.00 (0.00-6.17] 0.72 (0.14-2.11] 2.86 [0.29-10.29] 0.95 [0.19-2.79] 0.42 (0.00-2.33] 1.29 (0.92-1.75] 0.91 (0.0D-5.09) 1.32 (0.52-2.70] 0.91 (0.18-2.67] 2.10 (1.22-3.35] 0.95 [0.81-1 02] 1.15 [0.80-1.60] 1.10 10.40-2.40] 0.73 [0.15-2.15] 0.83 [0.57-1.17] 1 46 [0.70-2.69] 1.82 (1.06-2.91) effects of tobacco consumption, in particular on the incidence of cancers of the respiratory tract, although the SMR for smokers (1.53) versus non-smokers (0.79) is smaller than usually observed. Mortality from All Causes The low overall mortality (all causes) cannot be attributed to a non-exhaustive deaths count because for the French workers in the cohort, whose vital status is well-known (>99%). mortality is even slightly lower (SMR 0.83) than that among the whole cohort (SMR m 0.85). This 'healthy worker effect'.0 is generally explained by the selection of workers hired in indus trial production units. Deaths from Cardiovascular Diseases These were less frequent than would be expected from national statistics, but this may be due to differences in the method of determination of causes of deaths com pared with the national deaths register. Only the sur vey carried out by EEH14 indicates a slight excess mortality from other hypertensive diseases (ICD 1955: 444-457) among aluminium electrolysis workers. Cirrhosis of the Liver This disease is frequent in France and is generally attributed to overconsumption of alcohol. In this sur vey, cirrhosis of the liver was more frequently a cause of death among maintenance workers than among electrolysis workers. We have no information about alcohol consumption among these workers. In contrast to our results, Giovanasi1* noticed a relatively higher mortality from this cause among potmen compared with other aluminium electrolysis plants workers. `Violent' deaths Most of these (88) were caused by road accidents: 22 deaths were due to occupational causes; the annua) incidence rate of fatal occupational accidents was higher among smelting workers (28.4 per 100 000) and maintenance staff (26.8 per 100 000) than among elec trolysis workers (10.7 per 100 000). Cancers Although not statistically significant in our study, there might be. as found by other authors, a slight increase in cancers of the liver.of the brain.1* of the pan creas.1of the skin.1'' of the bladder.*0 and in lymphomas.18 However, in our study cancers of these sites were not sufficiently numerous for a thorough analysis. Mortality from lung cancer was not increased when compared with the national reference although it was a little higher among electrolysis and smelting workers tx timer KMC0020313 MORTALITY OF ALUMINIUM REDUCTION PLANT WORKERS IN FRANCE 263 than in maintenance staff. According to available data about tobacco consumption, these differences in lung cancer mortality could not be explained by differences in tobacco consumption between groups (Table 6). Several authors*'7 l4_l!'-:, :` :j have reported an excess of lung cancers in aluminium industry workers; but in most of these surveys the risk was found for particular age groups, length of employment or involvement in the electrolysis process. As regards potmen, mortality by lung cancer is higher among those with the shortest length of employ* ment (<10 years) and an intermediate latency (10-20 years). So, lung cancer incidence does not increase with the length of exposure, nor with the time since first exposure as some authors have observed;2*15 instead it shows a minimum incidence for an intermedi* ate length of work (10 to 19 years), as observed by others.71 In our study, differences in lung cancer mor tality according to length of employment as potmen may be due partly to differences in tobacco consump tion according to length of service (Table 6). However, because of the number of statistical tests performed and the small size of subgroups, such variations may be accounted for by chance. In most of the surveys which have shown an excess of lung cancer in aluminium workers, the risk could be attributed mainly to use of Soderberg electrolysis process.6 7 u l,:* a In our survey, such a risk could not be significantly confirmed. However, it must be pointed out that no case of lung cancer was found among the 19 deceased electrolysis workers who had worked only with the prebake anode process. Because of the rela tively small number of potmen for whom the particular elecrrolysis process could be established, a substantial nsk of lung cancer in Soderberg workers cannot be excluded, although such a risk appears unlikely for prebake workers. CONCLUSION This analysis of mortality between 1950 and 1976 of aluminium reduction plant workers in France, does not TaAle 6 Tobacco consumption according to occupation ae of smokers All causes of death Electrolvsis Maintenance Smelting Elcctrolvsis: <10 vears Electrolysis: 10-20 vears Electrolysis: 520 years Elcctrolvsis: PrePakc anodes Electrolysis: Soderberg' process 65.1 6?3 67 4 61.8 64.1 548 75.4 62.2 78.9 reveal a major fatality risk. Indeed, a basic overall `undermortality' is observed, although there are excess deaths from `violent1 causes and cirrhosis of the liverAmong potmen, the suspected nsk of lung canceT only appears in workers with the shortest length of employ ment and does not seem to be associated with a particu lar electrolysis process. However the number of men in this study with sufficiently long exposure for a notice able risk to be likely is too small to strongly contradict the hypothesis of such a risk. Therefore the monitoring of the incidence of cancers among workers employed in this industry will be continued. ACKNOWLEDGEMENTS This research was achieved with the technical assist ance of Ms M Betz, F le Bot and D Faure. We gratefully acknowledge the administrative, medical and technical departments of the Aluminium Pechmey Company for their help. REFERENCES 1 DoU R. Risks of cancer in the pnmary aluminium industry. PAH-- Symposium. Oslo. 1984. Aluminium mdustnens Mil* josekretanar. 1984. Enterline P . Epidemiologic studies of aluminium reduction work ers. in: Hughes J P ed. HeaUM Protection in Primary Alumin ium Production. London. International Primary Aluminium Institute. 1977; 109-16. ' Simonaio L. Carcinogenic nsk in the aluminium production indus try: an epidemiological overview. Med Lavo'o 1981: 4:26676. 'Attenom H. Lafonume M. Limasset J C. Vincent R. Truy S Enqudte global* sur ies nuisances dans les ateliers d'6!ectrolyse de I'industne de I'alumimum: usine AluminiumPechiney de Samt-Jean-de-Maunenne. Teneurs atmospnenques des pnncipaux polluams chimiques presents sous forme gazeuse ou paroculaire. Rapport INRS. n" 983/PP. Juiilet 1981. * Bjorseth A. Bjorseth O. Fjeldstad P E. Polycyclic aromatic hydro carbons in the work atmosphere. 1. Determination tn an alu minium reduction plant. Scand J Work Environ <& Health 1978; 4: 212-23. 9 EPAA Occupational Health Committee- Report on an environmen tal and health study (197^1975). LochaPer Works. The British Aluminium Company. Fori William. Scotland. Feb ruary 1976. 'Konstantinov V G. Kuz'minykh A I. Tarry substances and 3.4benzpvrene in the air of electrolytic shops of aluminium works and ihetr carcinogenic significance. Oigtena i samian/a. USSR 1971:36- 39-42. ' [ARC. Evaluation of the carcinogenic nsk of chemicals to humans. IARC Monographs supplement 4. Lyon: France. 1982:222-8 - Ravicr E F. Technology of alumina reduction. In Hughes J P ed. Health Protection in Primary Aluminium Pnaucnon London. International Aluminium Institute. 1977: i"-22. "'O.M.S. Classification Internationale des Maladies: revision 1963. 2 vol. Genive. Organisation Mondiale de la Same. 1968 " INSEE. Stamtiques des causes de dects: aperpu ritrospecnf Annees 1950 a 1967 Pam: Impnmene Nationale. 19'0 TX TIMER RMCQ020819 264 INTERNATIONAL JOURNAL OF EPIDEMIOLOGY r: iNSERM. Sutistiques dcs causes medicates de ddcPs: rtsultats France. Fans: INSERM. 1968........ 1976. '> Fox A J. Collier P F. Low mortality rates m industnai cohort studies due to selection for wort and survival in the industry. Bnr J Prrv Sot Med. 1976: 3*: 225-30. 11 Equitable Environmental Health Inc. Monality of aluminium work ers. Final report (prepared for the Aluminium Association Inc). Berkeley. Calif.. February 1977. " Giovanazs A. D'andrea F. Cause di mono fra 1 lavoraton di un imptanto di nduaone elettrotiuca di allumina. Med Lavoro 1981:4: 277-82. '* Litvinov N N. Goldberg M S. Rimma S N. Morbidity and monality in man caused by pulmonary cancer and its relation to the pollution of the atmosphere m the areas of aluminium plants. Acio Unionu Intemauonala Contra Centrum 1963:19? 742-3. " Milham S. Monality in aluminium reduction plant workers. JOM 1979: 21: 475-80. ' Milham S. Cancer mortality patterns associated with exposure to metals. In: Sa/ftoti U and Wagoner i K (eds). Occupational carcinogenesis. Ann iV Y Acad So 1976; 271: 243-9. "* Konstantinov V G. Simakhma P G. Gotlib Y V. Kuz'mmykh A I Problem of the carcinogenic hazard in aluminium electrolysis hails. Rrofezstonnal'nyt Rak USSR 1974; 87-91. *' Rockette H E. Arena V C. Monality studies of aluminium reduction plant workers: potroom and carbon department. JOM 1983. 25: 549-57. :i Andersen A. Dahlberg B E. Magnus K. Wannag A. Risk of cancer m the Norwegian aluminium industry. Ini J Canter 1982; 29: 295-8. r Theriault G. Guire L De. Cordier S. Reducing aluminium: an occupation possibly associated with bladder cancer. CMA Journal 1981: 124: 419-25. a Theriault G. Tremblay C. Cordier $. Gingras S. Bladder cancer in the aluminium industry. Lancet 1984: 1: 947-so. Gibbs G w. Horowitz 1. Lung cancer mortality m aluminium reduc tion plant workers. JOM 1979:21: 347-53. r Gibbs G W. Mortality of aluminium reduction plant workers. 1950 through 1977. JOM 1985; 27: 761-70. {Revised version received August 1986) TX TIMER RMC00 20 8 20