Effects of Weldingon Health V

Key Words — Welding cancer, disease,exposure, fumes, gases, health, literaturereview, noise, radiation, toxicology

Effects of Welding

on Health - V

Research performed by SRI International, Menlo Park, California, under contract with the American Welding Societyand supported by industry contributions.

An updated (July 1982-June 1984) literature survey and evaluation of the data recorded since the publication of the firstreport (1979). Some references to papers at an international conference held in February 1985 are also included. Thisseries of reports is intended to aid in the understanding of the health effects of welding.

Performed by:

Samual D. Kaplan

January, 1986

Abstract

This literature review with 105 citations has been prepared for the Safety and Health Committee of the AmericanWelding Society to provide an accessment of current knowledge of the effects of welding on health, as well as to aid inthe formation of research projects in this area, as part of an ongoing program sponsored by the Society. Previous workhas included studies of the fumes, gases, radiation, and noise generated during various arc welding processes.Referenced materials are available from SRI International.

Prepared forSAFETY AND HEALTH COMMITTEE

AMERICAN WELDING SOCIETY550 N.W. LeJeune Road, P. O. Box 351040

Miami, Florida 33135

International Standard Book Number: 0-87171-275-X

American Welding Society,550 N.W. LeJeune Road, P. O. Box 351040, Miami, Florida 33135

® 1987 by American Welding Society. All rights reservedPrinted in the United States of America

This report is published as a service and convenience to the welding industry and is the product of an independentcontractor (SRI International) which is solely responsible for its contents. The materials in this report have not beenindependently reviewed or verified and are only offered as information. AWS assumes no responsibility for any claimsthat may arise from the use of this information. Users must make independent investigations to determine theapplicability of this information for their purposes.

Personnel

AWS Safety and Health Research Committee

Dr. A. N. Ward, ChairmanK. L. Brown, Vice Chairman

M. E. Kennebeck, Jr., SecretaryW. T. Delong*

G. Hancock*W. S. Howes

E. Mastromatteo*R. A. Parent*

R. J. Simonton* Advisory Members

Caterpillar Tractor CompanyLincoln Electric CompanyAmerican Welding SocietyTeledyne-McKayBethlehem Steel CorporationNEMAConsultantConsultox, LimitedINCO Alloys International

in

Foreword

(This Foreword is not is part of the Effects on Welding on Health- V but is included for information purposes only.)

The health of workers in the welding environment is a major concern of the American Welding Society. To stayabreast of this subject, the health literature is periodically reviewed and published in reports titled Effects of Welding onHealth. Four volumes have been published to date. The first covered data published prior to 1978, while the next twocovered the periods 1978 to May 1979 and June 1979 to December 1980, respectively. The last of the four includedinformation that was published between December 1980 and June 1982. All should be read in conjuction with thecurrent volume for a comprehensive treatment of the literature on the effects of welding on health.

Included in this volume are studies of the characteristics of welding emissions that may have impact on the controltechnologies necessary to protect the welding worker (the Exposure). Much recent research has focused on chromiumand nickel, since exposure to certain chemical forms of these metals may cause serious chronic health problems.

In keeping with previous volumes, the health studies are organized according to the organ system affected (Investiga-tions in Humans). The respiratory tract, the primary route of entry of welding fumes and gases into the body, also is amajor target organ of a number of components of these emissions. Acute (e.g., metal fume fever, cadmium poisoning) aswell as potential chronic respiratory effects (e.g., emphysema, cancer) of welding emissions are of concern. The latter arefar less well understood and whether or not there is an excess risk of cancer from these exposures has not beenestablished. Continued research in the form of epidemiologic studies, investigations with laboratory animals, and invitro genotoxicity studies will help to resolve this question (Experimental Investigations).

IV

Acknowledgments

Funds for this project were provided by American Welding Society.The American Welding Society gratefully acknowledges the financial support of the program by industry

contributions.

Supporting OrganizationsAir Products and Chemicals, IncorporatedAirco Welding ProductsAllis-ChalmersAlloy Rods Division, The Chemetron CorporationAWS Detroit SectionAWS New Orleans SectionArcos CorporationThe Binkley CompanyCaterpillar Tractor CompanyChicago Bridge and Iron CompanyGrove Manufacturing Company, Division of Kidde, IncorporatedGeneral Electric CompanyThe Heil CompanyHobart Brothers CompanyINCO Alloys InternationalLincoln Electric CompanyMiller Electric Manufacturing CompanyNational-Standard CompanyA. O. Smith CorporationTeledyne-McKay, IncorporatedTrinity Industries, IncorporatedTruck Trailer Manufacturers AssociationWalker Stainless Equipment CompanyWeld Tooling Corporation

Many other organizations have made contributions to support the ongoing program from May 1979 to the present.

Comparative Listing — Welding Processes

Explanatory Note: Terms used in the technical literature do not correspond to those recommended by AWS in itspublication AWS A3.0, Standard Welding Terms and Definitions.

Accordingly, the following list will aid the reader in identifying the process in use.

EWH - VI

Gas or Flame Cutting

Gas Welding

MAGMIG, GMAMMA, SMATIGWire

(OC)

(OFW)

(GMAW)(GMAW)(SMAW)(GTAW)

Preferred AWS

Oxygen Cuttingor (OFC) Oxyfuel Gas Cutting

Oxyfuel Gas Weldingor (OAW) Oxyacetylene Welding

— (with specified shielding gas)Gas Metal Arc WeldingShielded Metal Arc WeldingGas Tungsten Arc WeldingElectrode

VI

Table of Contents

page no.

Personnel iiiForeword ivAcknowledgments v

Comparative Listing — Welding Processes vi

Technical Summary 1

1. The Exposure 51.1 Fumes 51.2 Gases 91.3 Electromagnetic Radiation 101.4 Noise 101.5 References 10

2. Investigations in Humans 112.1 Diseases of the Respiratory System 112.2 Metal Fume Fever 202.3 Effects on the Cardiovascular System 202.4 Effects on Kidney Function 202.5 Biological Monitoring 212.6 General Morbidity 232.7 Cancer of the Nasopharynx 242.8 Nasal Cancer 242.9 Cancer of the Larynx 252.10 Lung Cancer 272.11 Mesothelioma 302.12 Bladder Cancer 312.13 Kidney Cancer 312.14 Brain Cancer 312.15 Neuroblastoma 332.16 Hodgkin's Disease 332.17 References 34

3. Experimental Investigations 373.1 Animal Studies 373.2 In Vitro Studies 403.3 References 42

vn

Technical Summary

The ExposureFumes. A substantial amount of work has beenreported on the factors influencing fume formation,including voltage, current, shielding gas, and electrodecomposition. In a series of reports, Gray et al., havepresented data on both total fume formation, fraction-ated (elemental fractions) fume formation rates, andfactors that influence those rates. Other authors (Tandonetal., 1983,1984; and Kobayashietal., 1981,1983) havealso presented substantial bodies of data. The effects ofcurrent and voltage appear not to be linear —maxima and minima appear as the voltage and currentare increased from minimum acceptable levels throughthe working range. Tandon also confirmed the empiricalobservation that inexperienced welders produce sub-stantially more fume than experienced workers do.

Work has continued on the evaluation of methods forchromium analysis. Moreton et al. (1983) presented theresults of a major interlaboratory/intermethod round-robin exercise in which several methods for Cr(VI) andCr(III) analysis were compared. This topic has also beendiscussed by others, notably Gray et al. (1983), Sternet al. (1984), and Mcllwain and Nuemeier (1983). Thereasons for preferential formation of Cr(VI) in fumesand the sampling/analytical artifacts introduced bysome commonly used methods are beginning to be morethoroughly understood, although significant differencesof opinion among the active workers in the field are stillapparent.

Gases. Although gases have not received the recentattention that the fumes have, a fair amount of valuablework has been reported. Press and Florian (1982),Ditschun (1983); Downey et al. (1983), and Faggetter etal. (1983) have reported on mechanisms of gas forma-

tion, engineering methods for exposure control, andsampling and analytical methods. Both oxides ofnitrogen and ozone have been addressed; the controlmeasures found to reduce ozone have been addressed;the control measures found to reduce ozone levels havesometimes increased the levels of nitrogen oxides,although promising results in terms of reducing bothcompounds have been obtained by increasing turbulencein the outer layers of the shield envelope.

Electromagnetic Radiation. An interesting finding onthe time-course of UV exposure was presented byEriksen (1985). He found that the initial period of arcignition may be characterized by an overshoot (as muchas ten times the steady-state value) of UV irradiance,leading to eye risks for welders even though the steady-state values were not excessive. The implications for tackwelders and their assistants were pointed out.

Noise. Eichbauer and Schmidt (1982) have studied theeffects of various welding parameters on noise levels.There are substantial differences between manual metalarc (MMA), GMAW, and gas welding in the effects ofsuch variables as workpiece thickness and form of theweld.

Effects of Welding on Human HealthCancer Incidence. Despite the advances made in recentyears in studies of the relationship between weldingenvironment and health, gaps in knowledge remain, pos-sibly because there are few unique or specific diseasescaused by the chemical agents associated with welding.Many of the diseases can be related to factors other thanchemical in the welding environment; this can confound

the relationship between exposure and the disease thatdevelops. For example, in this review, one importantconfounding factor for respiratory tract diseases, thesmoking history of welders, has not been consistentlyavailable to researchers evaluating welding effects. Thisdifficulty can be overcome if the magnitude of theobserved risk is very great or if a dose-response isobserved, either as a function of years of welding or as afunction of the intensity of the welding tasks. Unfor-tunately, as discussed below, some of the associationsobserved are very weak, leading some investigators toconclude that some other underlying factor has beenresponsible for the increased incidence of cancerobserved. Studies that clearly define working conditions,exposure levels, welding materials, and general health ofworkers, including medical and smoking history, andhistory of their exposure to asbestos are necessary inorder to separate the possible synergistic effects.

An examination of the occupational exposure of weld-ers to chromium and nickel is under way (Waddington etal., 1984). If this study redresses the limitations of pre-vious studies, it may be possible to determine the extentof the health hazard posed by welding fumes.

Cancer of the Nasal Passage. Hernberg et al. (1983aand 1983 b) observed that patients with nasal cancer weremore than twice as likely to have an occupational historyof welding as patients with cancer of the colon or rectum.Whether this represents a decreased risk of colon/rectalcancers or a greatly increased risk of nasal cancer couldnot be determined from these studies.

Cancer of the Larynx. A weak association of weldingand all cancers of the larynx was found both by Olsenet al. (1984) and Sjogren and Carstensen (1985). Olsenet al., also found that the risk of cancer of the larynxincreased with the number of years of employment as awelder, particularly with the number of hours per weekspent welding. Olsen et al., also found a strong associa-tion of welding exposure and the rare subglottic laryn-geal cancer; the increased risk with exposure to weldingdust was seen only in smokers.

Lung Cancer. A number of studies have reported eitherno increase risk of lung cancer or a slight to moderateincrease in welders.

In a case-control study in Tidewater, Virginia, whereshipbuilding had been a major source of employment,Blot et al. (1980), found that a small excess of cases hadworked in a shipyard as a welder or burner. Similarfindings of slightly elevated risks among welders weremade by Sjogren and Carstensen (1985), Sjogren et al.,(1985), and Gallagher and Threlfall (1983). In contrast,McMillan and Pethybridge (1983) found no excess riskof lung cancer among welders, and only a slight excess

risk was observed by Newhouse et al. (1985). Results ofsome studies in progress may help clarify the relationshipof welding and lung cancer and determine whether theincreased risk is due to exposure to asbestos, which haspreviously been proposed as contributing to lung cancer.

Mesothelioma. Barnes (1983) and McMillan (1983)published case reports of five and seven welders,respectively. McMillan and Pethybridge (1983) found anincreased proportional mortality among welders in ship-yards compared with that in other shipyard workers whoare known to be at increased risk of mesothelioma.However, in each case, asbestos exposure was shown tobe responsible for the increased risk.

Kidney Cancer. Sjogren and Carstensen (1985) founda weak association with kidney cancer among weldersidentified during the 1960 Swedish census and sub-sequently followed for cancer. Because the association isweak, it may be explained by a common underlyingfactor, such as smoking.

Brain Cancer. Englundetal. (1982) found a 35 percentexcess of observed to expected brain cancer in Swedishwelders and gas cutters. In an apparent later follow-upby Sjogren and Carstensen (1985), this excess haddecreased to 20 percent. Because this relatively weakassociation could be explained by a common underlyingfactor, further work needs to be done to confirm thisfinding.

Hodgkin's Disease. In a proportional mortality analysisof deaths in British Columbia between 1950 and 1978,the deaths among welders from Hodgkin's disease werenearly 2.5 times the expected number (Gallagher andThrelfall, 1983). The authors recommended confir-mation of this finding through better designed studies.

Diseases of the Respiratory System. McMillan andPethybridge (1984) found a higher prevalence of dyspneaand mixed plumonary disease in shipyard welders thanin controls. All of the men had been MMA welders,usually with mild steel, for an average of 33 years. How-ever, a higher incidence of pleural plaques in the weldersindicated that there had also been higher exposure toasbestos. Results of a clinical study planned to assess theacute and chronic effects of welding on the respiratorysystem (Waddington et al., 1984) may help to eliminatethe confounding effects of asbestos.

Results of several studies were available only inabstract form. In a study of mild-steel welders, Newhouseet al. (1985) reported excess mortality from pneumonia.Mur et al. (1985) reported slight, but nonsignificant,excess in bronchitis and specific radiological abnor-malities and significant impairments of pulmonarycarbon monoxide transport in welders, particularly

those who smoked. On the basis of this study, theauthors recommended that a carbon monoxide transporttest be used to monitor pulmonary function in welders.

Fawer et al. (1982) reported that welders had moresick days, sickness absences, and length of absence perepisode of illness because of repiratory disease than didcontrol workers. It is not clear how clinically significantthese increased absences are in terms of long-rangehealth effects. As a result of an extensive study of respira-tory symptoms and pulmonary function in MMAwelders, Kalliomaki et al. (1982) concluded that welderscould not be considered a homogenous group withrespect to exposure and possible health-associatedhazards. Pulmonary function of mild-steel (MS) welderswas apparently more seriously affected than that ofstainless-steel (SS) welders. However, the study relied onself-administered questionnaires and clinical and labo-ratory results from various sources that may or may nothave been comparable. In a follow-up of this study,pulmonary function values were significantly belownormal in retired welders, whereas values in youngerwelders were within normal limits.

Boylen et al. (1984) determined the acute effects ofdifferent welding-generated fume during welding onaluminum, stainless-steel, and mild-steel shipyard weld-ers. From the information presented in the abstract, itappears that symptoms of pulmonary irritation andmetal fume fever were more associated with aluminumwelding than with SS or MS welding.Effects on the Cardiovascular System. The chemicalcomposition of aerosols generated by high alloy steelwas implicated in metabolic changes in the cardiovas-cular system and neurocirculatory history of welders.The differences found in the pathology were attributedto the high chromium and manganese content (Krasnyuket al., 1985).

Metal Fume Fever. Metal fume fever has been reportedin welders exposed both to zinc fumes (Ulvik, 1983) andaluminum fumes (Boylen et al., 1984). In one case report(Ulvik, 1983), a welder was occupationally exposed forfive hours to fumes containing 35 percent zinc, accordingto analysis. The fact that the welder's serum zinc levelswere found to be lower during the acute phase of thefever supported the hypothesis that increased serum zincis not part of the pathogenic mechanism in zinc fever.

Diseases of the Kidney. Although epidemiologicalevidence of association between kidney disease andexposure to metal welding fumes is still generally lack-ing, a recent study noted a correlation between exposureto cadmium-oxide exposure and chronic renal tubulardisfunction in welders. The role of SS welding and, morespecifically, chromium in kidney disfunction remains tobe resolved.

Biological Monitoring. Ever since it has been knownthat the working conditions of welders are variable fromworkplace to workplace, there has been a continuingsearch for methods to establish more exact dose-relationships between welding fume exposure and healtheffects. From the results of the studies reviewed here, it isapparent that further baseline work is required.

The kinetics of chromium excretion was investigatedby measuring urinary chromium levels. Welinder et al.,1983 demonstrated a two-compartment model forelimination of chromium: a fast-elimination compart-ment consisting of a chromium half-time that correlatedwith chromium in the personal air and a slow-elimi-nation compartment that was more approximate anddepended on the degree of previous exposure. From thiswork and the work of Rahkonen et al. (1984), it wasapparent that a single urinary measurement was notsufficient to estimate chromium exposure. Mutti et al.,(1984) determined that urinalysis was most suitable forthe estimation of water-soluble, hexavalent chromium.Correlations were found between chromium and hex-avalent chromium in personal air samples and bloodlevels, but more work in this area is needed.

Nickel concentrations in the urine correlated well withpersonal air concentrations, but nickel concentrations inwhole blood did not differ between welders and controls(Rahkonen et al., 1983).

Sjogren et al. (1983b) found that aluminum levels inurine, but not in blood, were correlated with years ofwelding experience. Measurements of urinary levels offluoride also proved a useful way to measure exposure tofluorides (Sjogren et al., 1984). The possibility of moni-toring individual maganese exposures from concen-trations in biological fluids was found to be complicatedby the wide interpersonal and intrapersonal variations inobserved concentrations.

Because previous studies have found that cadmiumexposure is not reflected in urine or blood levels, aneutron activation technique was investigated for itspotential to monitor cadmium levels (Baddely et al.,1983). This method shows some improvement over theuse of either urine or blood for monitoring subtoxicexposures to cadium.

Estimation of Retained Particles in the Lung. Studiesusing magnetopneumography have shown that increas-ed magnetic retention in lungs occurs in welders, andthat the amount of retention is related to the number ofyears of welding (Naslund and Hogstedt, 1982; Hogstedtet al., 1983; Freedman, 1982, 1984). As previous studieshave shown, smokers were observed to have lower levelsof remaining feromagnetic material in their lungs(Naslund and Hogstedt, 1982). To derive a medianinferred-lung burden for welding fumes, Sten et al.

(1985) used a new magnetopneumographic field tech-nique based on an alternating-current susceptibilitybridge to measure magnetic retention. Even though theyfound a high degree of correlation between medianinferred-lung burden of welding fumes and relative life-time occupational exposure in nonsmoking welders, nodifference was observed in chronic bronchitis incidenceor some lung-function measurements. This, in effect,precludes its use in detecting clinical abnormalities untilfurther baseline studies can be conducted.

Experimental InvestigationsAnimal Studies. The possible carcinogenic effects ofwelding fume particles have been examined in hamstersand rats. After repeated weekly intratracheal instillationof fumes generated from manual metal arc (MMA) ormetal inert gas (MG) welding, interstitial pneumonia,alveolar bronchiolization, and emphysema were observedafter welding fume particles containing chromium wereimplanted in the bronchi for 34 months. The relevance ofnegative results in this study may be questionablebecause the implantation technique may not adequatelymimic true human exposure conditions.

Pneumonitis or lung damage was observed when ratswere treated with the constituents of welding fumesgenerated from various kinds of welding (Hicks et al.,1984; Al-Shammaet al., 1983; Kennebeck, 1985). Stain-less steel welding particulates appeared to induce greaterand longer-lasting effects in rat lungs than did mild-steelwelding particulates, although each fume constituenthad distinctly toxic effects when it was administeredeither intratracheally or by inhalation.

Hexavalent chromium compounds appeared to be themost acutely toxic of a variety of trivalent and hexaval-ent chromium compounds that were administered intra-peritoneally to male and female mice. However, insubchronic studies in mice by the same route of adminis-tration, trivalent compounds were retained in the bodyat a rate of 6.5 times higher than that for the hexavalentchromium compound at 21 days. Excretion of hexaval-ent chromium was much faster in both the urine andfeces of mice. The low initial retention of hexavalentchromium is attributed to is low ability to form coor-dination compounds at physiological pH (Cryson andGoodall, 1983).

Several studies reported on the comparison of reten-tion in the lungs of rats after their exposure, by inhala-tion, to MMA/SS and to MMA/MS welding fumes.Alveolar retention of welding fume particulates wasmuch higher in rats after exposure to MMA/SS fumes.This probably reflects increased clearance of M M A / M Sfume particles, a finding that agrees with earlier observa-tions on welding fume particle retention in welders. Ironconcentration in the lungs was significantly higher in ratsexposed to MMA/MS fumes than in controls, butMMA/SS welding fumes did not significantly affecttotal iron concentration.

In vitro Studies. A mutagenic effect due to MMG/SSwelding fumes was demonstrated in a modification of theAmes Salmonella typhimurium assay that used freshlygenerated welding fume collected in an impinger (Stern,1983). Nickel in M1G / SS welding fumes and hexavalentchromium in MMA/SS fume were hypothesized to bethe active components in transforming BHK-21 andSyrian Hamster Embryo (SHE) cells in culture (Hansenand Stern, 1983, 1985). Neither trivalent chromium norMMA/MS fume was active in these mammalian celltransformation assays.

No significant increase in the number of cells withchromosome aberrations or sister chromatid exchanges(SCE) was observed in the peripheral lymphocytes oftwo different groups of MMA/SS welders (Littorin etal., 1983; Hugsafvel-Pursiaien et al., 1982, 1983).Although negative results in this assay system may not berelevant to the question of genotoxicity of welding fumesin man, it is worth noting that in both studies there wasa significant positive association between smoking andthe frequency of SCE within both the welders and con-trol groups.

MMA/SS, but not MIG/SS, welding fumes didinduce a higher frequency of SCE in an in vitro cellsystem using Chinese hamster ovary cells (Reuzel et al.,1985).

Welding fume particles, especially fumes from SSwelding, were toxic to pulmonary alveolar macrophageswhen tested in an in vitro system (White et al., 1984).Hexavalent chromium in concentrations equimolar tothe stainless steel was less toxic than the whole particu-late, but trivalent chromium was nontoxic at equimolarconcentrations. These observations were confirmed byHooftman et al. (1985).

Effects of Welding on Health V

1. The Exposure1.1 Fumes

1.1.1 Effects of Voltage and Current. In an inves-tigation of the mechanisms of fume formation in MM Awelding by Gray et al. (1983), it was found that increasesin arc voltage increased the formation of the flux andmetal fractions of welding fume. With increasing volt-age, the proportion of flux fume increased slightly, andproportions of the less volatile elements such as Fe, Ni,and Mo increased at the expense of the more volatileelements such as Cr and Mn. The mechanism for theeffect of voltage on the fume formation rate was partlyexplained in terms of arc temperatures and exposedsurface areas of the flux and metal, oxygen partialpressures near the metal surfaces, and the frequency andseverity of physical processes. However, it should benoted that Kobayashi and Suga (1981) observed norelationship between voltage and arc temperature. Achange of current at fixed voltage was found by Gray etal., to have " . . .little or no effect..." (on fume for-mation rate), while Kobayashi found that there was aminimal, although positive and consistent, increase intemperature with current and a strong effect of arctemperature on fume formation rate.

In the MIG welding process, as described by Gray andHewitt (1982), explosive evaporation and fume fromspatter result in unfractionated fume formation. Thus,relatively nonvolatile metals, such as nickel and molyb-denum, are found in the fume in concentrations farabove those expected on the basis of their vapor pres-sures. The effect of voltage in MIG welding was found tobe primarily on the overall fume formation rate, not onthe fume composition [although the comments below onCr(VI) should be seen]. The fume formation rate (FFR)was described as exhibiting an "N" shaped relationshipin which the FFR reached a maximum as the voltage was

increased to 30 volts, followed by a drop to a minimumat about 43 volts which was then followed by a rapid riseas the voltage was increased to above 50 volts. The effectof voltage increase on Cr(VI) percentage of the fumewhen MIG welding stainless steel was to increase theCr(VI) content to a maximum at about 20V, followed bya sharp decline as the voltage approached 40V.

Malmqvist et al. (1985) confirmed previous findingsthat the highest fumeemissions are found in SMAW andGMAW of aluminum. The fume emission rate wasdetermined primarily by welding current. Welding volt-age and current were both reported to have dramaticeffects on the elemental composition of the fume.

Tandon et al. (1983) studied the effects of current andvoltage over a wide range (80-120 A, 20-40 V) in MM Awelding and determined that an optimum arc lengthminimized fume generation. The rate of fume generationvaried so critically with arc length that even an expe-rienced welder might choose to operate under conditionsthat would result in substantially more than the min-imum rate of fume generation. An inexperienced welderusing the same machine setting might easily producefume at twice the minimum rate. Fume generation ratesfor welding from a single type of stainless steel electrodewere found to be 4 times greater under high voltageconditions and 1.5 times greater under low voltage (shortarc) conditions than the minimum fume generationobtained under optimum arc length.

Tandon et al. (1984) studied the effect of electricalvariables on fume generation rates and electrode meltingrates for hardfacing chromium and manganese and high-strength low alloy steel electrodes. In their work, theyfound that fume generation rates (FGR's) varied by afactor of two, while acceptable arc length was main-tained. FGR's increased almost linearly with voltage andwith power and decreased almost linearly with current.Electrode melting rates (EMR's) displayed opposite

behavior to the FGR's in their variation with electricalconditions. EMR's increased almost linearly with cur-rent and decreased almost linearly with voltage andpower. EMR's obtained using ac and deep (dc electrodepositive) were generally similar, but the EMR for elec-trode E-ll using dcen (dc electrode negative) wasapproximately 40% greater than that using deep. Incontrast to a previous referenced study, no systematicdifference was observed in FGR's with ac compared withdc welding.

Fumes generated at power levels of 100A and 120Awere found to have similar elemental composition withthe exception of sodium and potassium, where potassiumwas found to be more than nine times as abundant at120 A. In general, there was a marked increase in theabundance of most elements in fume produced under lowvoltage conditions; however, chromium was found inlower abundance under low voltage. The chromium con-tent in fume at 100 A was found to double as the arc volt-age was increased from 20 to 30V. Similar conclusionswere reached by Gray and Hewitt (1982) that increasingthe arc voltage reduced the hexavalent chromium con-tent of fume from MIG welding of stainless steel.

1.1.2 Effects of Shielding Gas. Fume generation rateswere found to be dependent on the oxidizing potential ofthe gas surrounding the electric arc (Malmqvist et al.,1985 and Gray and Hewitt, 1982). As the oxidizingpotential of the protective gas increased, so did the fumeproduction. Carbon dioxide in MIG welding resulted inmuch higher fume formation rates than argon- orhelium-based shield gases used in GMAW welding.Malmqvist found the oxidizing potential of the protec-tive gas to be as significant as welding current and volt-age in the production of welding fume.

1.1.3 Effects of Electrode Composition. The mecha-nisms of fume formation have been reported by Grayet al. (1983) and Gray and Hewitt (1982). Gray et al.(1983) studied the difference between TIG, MIG, andMM A welding and investigated the mechanisms of fumeformation. They studied the various sources of fume inMMA welding arising from the filler wire, flux, andworkpiece and concluded that to minimize fume genera-tion from the process, alteration of the composition ofthe electrode was an important factor. Welding fumefractions were defined as "flux fume", arising from theflux, and "metal fume" originating from the workpiece,filler metal, and metal powder in the flux mixture. Thesources of metal fume were determined as filler wire 70 to90%, metal powder in the flux less than 15%, and thework piece up to 10%. The flux is itself a large source offume, and despite a low metal fume formation rate, theoverall fume formation in MMA welding can be as largeas for the MIG process.

Formulation changes in the composition of the fillerwire and flux may reduce the fume formation rate orrender the fume less toxic (Gray et al., 1983 and Gray andHewitt, 1982). Metal added as a powder in flux makes alarger contribution to the fume than a similar weight asan alloying element in the wire. It was concluded thatincorporation of the metals in the wire would result inlower exposures to the metals in the fumes (Gray andHewitt, 1982). Gray et al. (1983) reported that flux thatcontains sodium or potassium silicates can introducelarge amounts of hexavalent chromium when weldingchromium alloys such as stainless steel. High tempera-ture of the arc produces trivalent chromium as mixedoxide spinel. They suggested that using lithium silicate oran organic binder causes significant reduction in thehexavalent chromium in fume.

Constituents of electrode coverings were always foundin welding fume, but sometimes in different percentagesof an element. Concentrations of Fe, Mg, and Ca inwelding fume were found in nearly identical percentagesas in the covering. Fluorine, Mn, Cu, Zn and lead werefound in higher concentrations in the fume and Cr andNi in lower concentrations than in the electrode covering(Gambaretto and Rausa, 1985). Kobayashiet al. (1983)studied the relationship between the contents of elementsin the coverings and in the fumes. The mechanisms offume generation were considered, and observations weremade by high speed camera. The contents of the ele-ments in the fume increased in accordance with theincrease of elements in the covering. The effects of limetype and non-lime type electrodes showed varying per-centages of elemental composition and alkaline contentin the fume. Iron oxide content in fumes from non-limetype electrodes was about 50%, regardless of the ironoxide content in the covering; whereas iron oxide con-tent in the fumes from lime type electrodes did not exceedone-half that of the non-lime type. The increase in alka-line content was found to be a result of the Ca content inthe covering, especially when present as CaF2. Fumecomposition with non-lime type coverings is regulatedby the vaporization from the molten metal surface at thetip of the electrode; that of lime type coverings is regu-lated by the vaporization from the molten slag surface.

The relationship between arc temperatures and fumeemissions generated using covered electrodes was de-scribed by Kobayashi and Suga(1981). The presence of or-ganic material and fluorite were found to raise arc temper-ature, while the presence of limestone, rutile, and alkalimetals such as Li, Na, and K lowered arc temperature.Fume emissions decreased as arc temperature decreased.Considering that a reasonably direct relationship betweenarc temperature and fume emission rate was found, theyconcluded that as the arc temperature drops, so does theamount of high temperature that evaporates within the arc.

1.1.4 Particles. Fasiska et al. (1983) conducted astudy of welding fume to provide a data base of chemical,crystallogic, and physical data for representative weldingfume types to aid in the understanding of the interactionsof these particles with the human respiratory system.Welding fumes from carbon steel, stainless steel, andaluminum were analyzed for particle size and chemistryby x-ray diffraction, energy dispersive x-ray analysis,scanning transmission electron microscope, and auto-mated electron beam analysis. Particle size diameterswere between 0.1 and 1.0 /xm. Less than 1% were foundto be greater than 1 yum, but STEM pictures revealedmany particles as small as 0.01 /xm. There was littlevariation in the size distribution for the six fumes testedand little variation with the type of welding consumableor with individual particle chemistry.

The morphology of welding fume particles appearedto be spheres or clusters of spheres. No crystallinefeatures were observed, though all particles examinedproduced electron diffraction patterns indicating theycontained crystalline material.

If fluorides were present in the consumable, thensodium, potassium, and calcium fluorides were found inthe fume. No transitional metal fluorides were detected.Silicon was found as an amorphous silica or was in solidsolution with the iron oxide. The analysis of fume froman aluminum wire indicated Al present primarily asalumina with small concentrations of aluminates.

Additional scanning electron microscopic studies byMinni et al. (1984) showed the average size of weldingfume of MMA of stainless steel to be 0.14 jum, MIGof stainless steel, 0.06 /um, and MMA of mild steel0.12 yum.

Minni et al. (1984) also used electron spectroscopy toanalyze MMA and MIG stainless steel fume and MMAof mild steel for the oxidation states of the main metalconstituents with electron spectroscopy. Following re-moval of the surface layers of the particles by ion sputter-ing, changes in the oxidation states and elemental con-centrations beneath the outermost surface layer werestudied. Iron in the fume of MMA welding of stainlesssteel exists primarily as K3FeF6 and chromium as hexa-valent K2CrO4 . Manganese was found to occur in theoxidation states 2+ and 3+ possibly as Mn3O4. Iron inMIG welding fume of stainless steel was found mainly asFe(III) in the form of NiFe2O4 and chromium as spinelsin the trivalent form as (Fe, Ni) Cr2O4. Manganese wasfound in the same oxidation state as MMA of SS withthe probability that it was Mn3O4. Nickel was present asa mixture of Ni(II) and Ni(III), together with metallicnickel possibly as NiO and Ni2O3, as well as Ni(II)spinels. Some particles were found to be coated apartfrom the normal contamination layer by a thin layer ofsilicon contining marterial. Iron in MMA welding fume

of mild steel was found as K3 FeF6 on the surface and asMnFe4 in the core of the particle.

The particle chemistry of ferrous fume was found to bedependent on the iron content of the fume (Fasiska et al.,1983). When large amounts are present, the main crystal-line phase is magnetite and other transition metals existas (Fe,X)3O4. Only when iron concentrations are rela-tively low is there a possibility that pure oxides of othertransition metals are present. Other oxidation states foriron may also be present, such as K(Fe,Cr)O4 found inSS fume.

1.1.5 Chromium. An "interlaboratory/intermethodround robin" study organized by The Welding Instituteand reported by Moreton et al. (1983) examined thespecific problems surrounding the analysis of solubleand insoluble Cr(VI) and total chromium in fume fromMIG and MMA welding of low alloy stainless steelconsumables. The principal method employed was theBlakeley and Zataka s-diphenyl carbazide color methodfor Cr(VI), modified by using differential alkaline leach-ing techniques to prevent reduction of Cr (VI) to Cr (III).Total chromium was determined using atomic absorp-tion spectrophotometry, x-ray fluorescence (XRF) orthe Blakeley-Zataka method.

The participating laboratories analyzed 1 to 10 milli-gram bulk samples of fume collected on PVC filters ofvarious types of fume generated by a selection of elec-trode types. Their results were in good agreement for thedetermination of the CR(VI) soluble and insoluble spe-cies using the modified Blakeley-Zatka procedure. TheBlakeley-Zatka method for total chromium, the per-chloric acid fuming technique, and the AAS procedureswere in good agreement for total chrome determina-tions. The effects of storage of the samples for periods oftwo weeks to three months prior to analysis for Cr(VI)indicated no significant differences in the results obtainedby the participating laboratories. Similar findings forPVC filters were reported by Blomquist et al. (1983).

Blomquist et al. (1983) reported the development offilter leaching techniques that resulted in higher Cr(VI)values than the Blakeley-Zatka procedure. The use of asodium acetate rather than sulfuric acid solution in theleaching of polyvinyl chloride filters resulted in the re-covery of higher amounts of Cr(VI), together withincreased stability of the solution as a function of time.The modified leaching procedure was shown to give thesame results as the more laborious carbonate method.The use of Rinehart's solution to extract Cr(VI) resultedin higher values than the INCO method utilizing a basicleach, though the INCO method resulted in higher totalchrome values (Mcllwain and Neumeier, 1983). Thetechniques used to determine the CR(VI) content ofMIG and MMA stainless steel welding fume were dis-

8

cussed by Hewitt et al. (1983) with emphasis on the needfor the establishment of a correct dissolution procedureto accurately determine welding fume constituents. Theyindicated that hexavalent chromium can be extractedefficiently by leaching filters with a sodium hydroxide/sodium carbonate solution compared to other solventsthat may give lower Cr(VI) results due to redox reactions.

Stern et al. (1984) studied the fume matrices of MM Aand MIG fume and found that the water soluble fractionof CR(VI) of MM A fume is stable, while that of MIGfume is not, and that the wide range for this speciesreflects the extreme sensitivity of the measurements tocollection, storage, and analysis techniques. Analysis offume samples collected on standard membrane filtersand those collected and fixed in impinger solutionsshowed wide variation. It was found that a maximum of30% of total chromium as Cr(VI) was found when col-lected and directly fixed in an impinger as opposed to 3%when first collected on a filter and analyzed afterwardsusing standard AAS techniques. Similar findings forwide variations in the amount of CR(VI) found usingimpinger and filters for fume collection were alsoreported by Gray, Goldstone et al. (1983). The reductioneffects of glass fiber and PVC filters were found to beminimal compared to that of cellulose ester filters whichresulted in a 50% reduction of Cr(VI).

The study of the chemical evolution of hexavalentchromium in MIG and MMA stainless steel weldingfume in both experimental and field conditions indicatedthat in the case of MIG fume, the hexavalent chromiumcontent appears to rise to a maximum some time afterformation of the aerosol, and then to partly decay again(Gray, Goldstone et al., 1983). The effects of aging ofMIG fumes showed the measured ratio of hexavalent tototal chromium to rise rapidly at first, reaching a maxi-mum within 20 seconds, and then to fall reaching asteady value within a few minutes with the final value forthis ratio the same as that for fume samples collected onglass fiber filters. In the case of MMA welding fume, theratio was not found to vary significantly over the periodfrom 1 to 90 seconds. The mechanisms of fume aging arediscussed as possible explanations for the differencesfound between the impinger and filter collection methods.The aging mechanism is a result of physical processessuch as cooling, aggregation and settling, and chemicalprocesses described as homogeneous, involving inter-actions between the constituents and atmospheric gasessuch as oxygen; and heterogeneous, involving both fumeconstituents and atmospheric gases such as oxygen.Aged MIG fumes consist mostly of metal oxides andmixed metal oxides with the spinel structure such asmagnetite (Fe(II)Fe(III)2O4) and (Fe(II)OCr(III)203)which are probably produced very rapidly at hightemperature in the early stages of fume formation. After

condensation and initial cooling of the fume particles,heterogeneous oxidation could lead to formation ofhexavalent chromium compounds by reactions such as:

Fe(II)O-Cr(III)2O3 + 2O2 + 2M - Fe(II)O

+ M(I)2Cr(VI)2O7

— where M could be any one of a variety of metalspresent as an element or compound. Such a reactioncould be inhibited by rapidly cooling the fume particle orby removing the particle from the oxidizing atmosphere,both of which could occur if the particle was collected inaqueous suspension.

Hexavalent chromium can be reduced by reactionwith metals of low oxidation state, for example:

3Fe(II) + Cr(VI) - 3Fe(III) + Cr(III)

This reaction proceeds rapidly in solution, but can alsooccur in the solid phase at temperatures of severalhundred degrees centigrade, and presumably couldoccur in fume particles before they finally cooled. Thesubmicron fume particles might be expected to cool sorapidly that temperature-related aging would be unob-servable. However, many of the chemical reactions thatcould occur in the aerosol, such as metal oxidation, areexothermic and might help to maintain a high particletemperature. Furthermore, reactions involving the solidphase can be relatively slow; proceeding at rates deter-mined by mass-transfer processes such as diffusion of thereactive species. These possible physical and chemicalprocesses suggest that the observed evolution of MIGwelding fume could be explained by an initial rapid risein the amount of hexavalent chromium in the fumeparticles as a result of oxidation reactions, and then fallas a result of homogeneous solid-phase redox reactions.In this scheme, it is envisaged that MIG fumes collectedon filter papers pass through these stages on the filter,and that by the time they can be analyzed, the process issubstantially complete.

This scheme could explain other properties of weldingfumes. It has been found that all or most of the Cr(VI) inMIG and other welding fumes is water soluble, althoughthe bulk of MIG fumes including chromium of otheroxidation states is not. If Cr(VI) were formed by homo-geneous reactions before or after fume condensation,then much of it might be expected to be found within theinsoluble fume particles. If it were formed by hetero-geneous oxidation reactions, then it would lie on or nearthe surface of the fume particles and would be moreeasily dissolved.

The large amount of Cr(VI) present in most fumesfrom stainless steel MMA welding has been shown to beassociated with the sodium and potassium compoundsin MMA welding fluxes, and it is almost certain that theCr(VI) is present in such fumes as chromates of these

metals. It is possible that the stability of these compoundscompared to most other metal chromates prevents orminimizes the processes leading to reduction of Cr(VI)in MIG fumes. Sodium and potassium chomates are alsostable at higher temperatures than most other chro-mates, and this could explain the much faster formationof Cr(VI) in MMA fume than in MIG fume. If thereaction can occur at a higher temperature during thecooling of the aerosol particles, then it will commencesooner and probably proceed at a higher rate.

A second possible explanation of the observed resultsis that some reactive component of the aerosol, producedup to 20 seconds after fume formation, might lead toCr(VI) formation in the presence of water. The reactivespecies might be a component of the fume particles or ofthe gas phase, for example, ozone. The eventual declinein the amount of hexavalent chromium collected in theimpinger train might be the result of the decay of thereactive species, or some other property of the aerosol,such as the particle temperature. In this scheme, it isenvisaged that reactions in the aerosol produce theamount of Cr(VI) found in filter samples, and an addi-tional amount is produced by collection in water whilethe aerosol is in a "reactive" condition.

It should be noted that the experimental conditions ofGray, Goldstone et al. (1983) included the use of cellu-lose ester filters as backups to the impinger samplingtrains. These filters have been found to result in lowerCr(VI) values due to the reduction effects of the filter,while glass fiber filters have shown negligible reductioneffects (Blomquist et al., 1983). Thus, the observed excessof Cr(VI) in the impinger train in the Gray, Goldstone etal. study may understate the true conditions.

Stern et al. (1984) studied the distribution of metallicspecies into various solubility fractions and oxidationstates and concluded that this is a function of the collec-tion, fixation, and analytical procedures employed. Todemonstrate this effect, impingers were used to collectfume containing various solvent mixtures which showedvariations of recovery of Cr(VI) of 0 to 8% to totalchrome. The observations reported raise serious questionsconcerning the interpretation of either positive or nega-tive bioassay results for metals or metallic compounds aswell as employee exposure determinations to the variousspecies of chromium (Gray, Goldstone et al., 1983).

1.2 Gases

1.2.1 Nitrogen Oxides. Press and Florian (1982) re-ported that when shielding gas was introduced through a16 mm diameter nozzle, nitrogen oxides generation in-creased with increased current in TIG welding with nitro-gen oxides levels increasing above 150A and in greaterproportions from 250 - 300 A. In the presence of ozone, itwas not possible to detect any nitrous oxide, but only

nitrogen dioxide; this confirmed the findings of one oftwo previous studies. The generation of nitrous oxide wasshown to be related to nozzle diameter. Larger amountsof nitrogen oxides were formed with smaller nozzles ascompared with nozzles of larger diameter. The increasewas attributed to the turbulent mixing between the streamof shielding gas and the ambient air due to the high exitvelocity of the shielding gas from the smaller nozzles.

1.2.2 Ozone. Factors that affect the production ofozone in TIG welding, such as arc voltage (length of arc),and parameters relating to the shielding gas were alsoreported by Press and Florian (1982). Ozone concentra-tions were found to increase with current to approxi-mately 175 A and drop with increasing current, a result ofthe increased concentrations of nitrogen oxides gener-ated above 150 A. The increased nitrogen oxide produc-tion in TIG welding is a result of the shielding gas'sinability to screen off the longer arc from the surround-ing air. At 250-300 A, the concentrations of ozone werefound to be very small due to the increased generation ofnitrogen oxides. Increases in voltage with increasing arclength resulted in higher levels of nitrogen oxides andlower levels of ozone.

Ozone levels were found to be higher with increasedflow of shielding gas due to the reduction in the amountof nitrogen oxides produced. Concentrations of ozonewere also found to be much lower when the rate of flowof shielding gas was reduced to below that normally usedin welding operations. It was concluded that low levels ofozone can be achieved by using the smallest amount ofshielding gas possible.

Fouling of the nozzle with spatter which created moreturbulence and mixing of ambient air resulted in increasednitrogen oxide levels and lower levels of ozone. Theeffects of different size nozzle diameters when used withidentical shield gas flow rates indicated varying concen-trations of ozone. Smaller diameter nozzles which createmore turbulence resulted in lower levels of ozone whencompared to larger nozzles producing less turbulence.

Using various proportions of hydrogen and helium ascompared to pure argon is known to affect arc character-istics. No conclusions were reached as to whether ozonelevels can be reduced by using various inert gas mixturesin the shielding gas. However, the introduction of nitro-gen oxides to the shielding gas was found to appreciablyreduce ozone levels while negligibly increasing the nitro-gen oxides levels.

Testing was conducted on the effects of increased airturbulence by blowing air onto the gas shield. It wasshown that as the rate of flow increased so did the levelsof nitrogen oxides with subsequent reduction of ozonelevels. The results of testing nozzles designed to createturbulence in the outer layers of the shield envelope werepromising and will require further investigative work.

10

Levels of ozone generated during GMA welding ofCu-Ni and Ni-Al bronze (NAB) alloys and ozone con-trol measures were reported by Ditschun et al. (1983).Concentrations of ozone using both alloys indicatedhigher concentrations of ozone with increasing voltage(26 to 29 V) by factors up to four; and decreasing current.The effects of the incorporation of helium in the argonshielding gas indicated suppressed ozone production.However, the effects were overridden with increasingvoltage, which caused the ozone levels to increase signifi-cantly. Various types of opaque arc shrouds positionedclosely around the welding arc or crossdrafts sufficientlystrong to remove ozonated air were found to dramati-cally reduce ozone levels. The use of shrouds reducedozone concentrations at given points by factors ofgreater than 5 to 10. Faggetter et al. (1983) reportedsuppression of ozone levels using stainless steel meshshrouds and ventilation shielding cabinets in GMAWwelding of aluminum with reduction in exposure tolevels well within the ACGIH TLV.

Downey et al. (1983) compared ozone measurementsmade with the conventional NIOSH Method [samplingwith impingers containing alkaline potassium iodide(AKI) followed by colorimetric analysis] with measure-ments made with a direct-reading chemiluminescentmonitor. They found that both methods were acceptablyaccurate and precise when tested with known pure ozoneconcentrations in the laboratory. When measuring ozoneconcentrations from MIG welding, however, the AKImethod gave much lower results than the direct-readinginstrument. For sample periods less than 10 minutes, theAKI results were only 10 to 15% of those found from thedirect device. For a longer sampling period (44 minutes),the AKI results were closer to, but still substantially lessthan, those of the direct device, falling in the 53 to 62%range. The authors suggest that previous measurementsmade with the AKI method probably under-representtrue ozone exposures of welders.

1.3 Electromagnetic Radiation

1.3.1 Ultraviolet Radiation. Measurements of ultra-violet (UV) radiation from MIG (GMAW) welding arcswere made by Eriksen (1985). The use of a Diode ArrayRapidused Scan Spectrometer indicated that an over-shoot of radiation can occur during the first 50 ms of arcignition. At 280 nm, the UV irradiance can overshoot (bymore than 10 times) the steady state average value. It wasimplied that the effective UV dose can be reached withone single ignition at 300A for an unprotected eye at adistance of 0.5 m. Tack welders and nearby assistantsshould use eye protection for any welding arc exposure.

1.4 Noise. The effects of various welding parameters onthe equivalent continuous sound levels was studied byEichbauer and Schmidt (1982). In MMA welding,

neither the thickness of the workpiece, the form of theweld, nor operators' welding positions had any effect onthe continuous sound level measured. The sound levelwas found to increase with current during the welding ofa root pass. During the welding of the final pass, thecontinuous sound level was found to be slightly lowerthan that for the root pass. At some point, a slightdecrease in the continuous sound levels were found withincreasing current. The continuous sound levels werefound to be between 70 and 80 dB(A).

Studies of gas welding showed that all weldingparameters had an effect on the increase of continuousnoise levels. Flames fitted with blowpipes were approx-imately 15 dB(A) quieter than those not so fitted.Increases in continuous noise levels were found withincreasing nozzle diameters and rates of gas flow.

Continuous noise levels during GMAW welding werefound to increase during the welding of thin-gauge sheetsand decrease on thick-gauge plates as current increased.MIG welding of thin-gauge sheets also indicated in-creased continuous noise levels with increasing current.The continuous noise level of MAG welding of thin-gauge sheets was approximately 80 to 93 dB(A) andthick-gauge sheets between 75 and 87 dB(A). MIG weld-ing of thin-gauged sheets indicated continuous noiselevels of 87 to 97 dB(A).

Octave band analysis of welding noise was used for theevaluation of subjectively perceived loudness of variouswelding processes. MIG welding was perceived to betwice as loud as TIG welding.

1.5 References

Blomquist G., Nilsson C , and Nygren O. "Sampling andanalysis of hexavalent chromium during exposure tochromic acid mist and welding fumes." Scand. J. WorkEnviron., Health. 9:489-495 (1983).

Ditschun A. and Sahoo M. "Production and control ofozone during welding of copper-base alloys." WeldingJournal, pp 44-46; August (1983).

Downey E. B., Buchan R. M., Blehm K. D., and GunterB. J. "A comparison of two ozone sampling methods."Am. Ind. Hyg. Assoc. J. 44(5)333-335 (1983).

Eichbauer H. and Schmidt R. "Noise and fume nui-sances affecting the welder." Schweissen und Schneiden.34(ll):535-539 (1982).

Eriksen P. "Time resolved optical spectra from MIG weld-ing arc ignitions.'Mm. Ind. Hyg. J. 46(3): 101-104 (1985).

Faggetter A. K., Freeman V. E., and Hosein H. R."Novel engineering methods for ozone reduction in gasmetal arc welding of aluminum." Am. Ind. Hyg. Assoc.J. 44(5):316-320(1983).

11

Fasiska E. J., Wagenblast H. W., and Nasta M.Characterization of arc welding fume. Miami, Florida:American Welding Society, pp 55 (1983).

Gambaretto G. P. and Rausa G. "Pollution caused bymanual arc-welding: quantity and quality variations ofthe welding fumes in function of electrodes characteris-tics (diameter, weight, and coat composition). "Abstracts,International Conference on Health Hazards and Bio-logical Effects of Welding Fumes and Gases. February18-21, 1985; Copenhagen (1985).

Gray C. N. and Hewitt P. J. "Control of particulateemission from electric-arc welding by process modifica-tion." Ann. Occup. Hyg. 25 (4): 431-437 (1982).

Gray C. N., Hewitt P. J., and Dare P. R. M. "Newapproach would help control weld fumes at source —part three: MAA fumes." Welding and Metal Fabrica-tion, pp 52-55; January/February (1983).

Gray C. N., Goldstone A., Dare P., and Hewitt P. J."The evolution of hexavalent chromium in metallic aero-sols.'Mm. Ind. Hyg. Assoc. J. 44(6):384-388 (1983).

Hewitt P. J. and Gray C. N. "Some difficulties in theassessment of electric arc welding fume. "Am. Ind. Hyg.Assoc. J. 44(10):727-732 (1983).

*Kobayashi M. and Suga T. "Measurement of arctemperature in covered arc welding." Kobe Steel Tech-nical Report. 31(3):76-80 (1981).

Kobayashi M., Maki S., Hashimoto Y., and Suga T."Investigations on chemical composition of weldingfumes." Welding Research Supplement, Welding Jour-nal, pp 190s-196s; July (1983).

Mcllwain J. F. and Neumeier L. A. "The generation,collection, and analysis of welding fumes." Bureau ofMines Report of Investigations. RI-8793; 18 pp (1983).

Malmqvist K., Johansson G., Bohgard M., andAkselsson K. "Process-dependent characteristics ofwelding fume particles." Abstracts, International Con-ference on Health Hazards and Biological Effects ofWelding Fumes and Gases. February 18-21; 1985,Copenhagen (1985).

Minni E., Gustafsson T. E., Koponen M., and Kallio-maki P., "A study of the chemical structure of particles inthe welding fumes of mild and stainless steel." J. AerosolSci. 15(l):57-68 (1984).

Moreton J., Bettelley J., Mathers H., Nicholls A., PerryR. W., Ratcliffe D. B., and Svensson L. "Investigations

* Complete documentation not available — need additionalinformation.

of techniques for analysis of hexavalent chromium, totalchromium and total nickel in welding fume: a co-opera-tive study.'Mnn. Occup. Hyg. 27(2): 137-156 (1983).

Press H. and Florian W. "An investigation of the pro-duction of foreign substances during shielded-arc weld-ing — TIG welding of alloy steel and aluminum alloys."Schweissen und Schneiden. 34(9):437-440 (1982).

Stern R. M., Thomsen E., and Furst A. "Cr(VI) and othermetallic mutagens in fly ash and welding fumes." Toxico-logical and Environmental Chemistry. 8:95-108 (1984).

Tandon R. K.. Ellis J., Crisp P. T., and Baker R. S."Fume generation and melting rates of shielded metalarc welding electrodes." Welding Journal. 63(8):263s-266s (1984).

Tandon R. K., Crisp P. T., Ellis J., and Baker R. S."Variations in the chemical composition and generationrates of fume from stainless steel electrodes under differ-ent ac arc welding conditions." Australian Welding J.28(l):27-30(1983).

2. Investigations in Humans2.1 Diseases of the Respiratory System. A very thor-ough evaluation of pulmonary function in welders (mis-labeled a case control study) was carried out by McMillanand Pethybridge (1984). The investigators invited all368 welders employed at Her Majesty's dockyards inDevonport, Portsmouth, and Chatham in England tohave clinical examinations, pulmonary function tests,and full-sized posterior and anterior chest x-rays. Of the328 (89.13%) who participated, a random sample of 135of these volunteers selected from the welders aged 45years and over formed the exposed group in the analysis.(The authors do not indicate denominators for all volun-teers 45 years and over or all welders 45 years and over.)The number of exposed welders was limited to 135because of time and funds constraints. Each of theexposed was individually matched — on sex, smokinghabit, age to one year, and an estimate of the potential toasbestos dust, with a skilled craftsman employed in theyard who had never been a welder but, because of hisskill, was considered similar in social class. Unfortu-nately, the matching was not retained in the analysis,resulting in a decrease in efficiency. An interview basedon a standard questionnaire and a clinical examinationwere conducted. Spirometry was carried out, includingforced expiratory volume in 1 second, forced vital capac-ity, and residual volume. Chest x-rays were read blindlyby three expert readers, with a consensus of at least twobeing needed to score the x-ray for pleural fibrosis, sec-ondary to asbestos exposure and small, regular opacities.

12

The welders had been welding for an average of 33years. All the men had been employed on manual metalarc welding, usually of mild steel, but a few had donemore metal inert gas welding or tungsten inert gas weld-ing. The proportions of welders and controls who gave ahistory of having pleurisy, tuberculosis, bronchitis, orpneumonia were similar in the welders and controls,although one wishes that only the discordant pairs hadbeen reported because a true association may have beenmissed through large numbers of concordant pairs. Inboth the welders and the controls, the period prevalenceof all respiratory disease was much higher in smokersand exsmokers than in nonsmokers, a finding suggestingthat an association that strong would be seen even withinthe limitations of the authors analysis.

The proportion of the 135 welders reporting mostrespiratory and cardiovascular symptoms was similar tothat in the 135 controls. The proportion of welders wassomewhat higher for chronic cough (defined as a coughon most mornings for at least three consecutive monthsof the year for two years or more) and hemoptysis (cough-ing up blood). There was a marked difference regardingdyspnea (shortness of breath), which was reported bynearly four times as many welders as controls. Physicalexaminations revealed that approximately twice asmany controls as welders had rales (abnormal sounds atthe base of the lung), but the number in each group withanother type of abnormal sound known as rhonchi wassimilar. Without further information, interpretation ofthe clinical significance of these findings is difficult.

With regard to the chest x-rays, one reader diagnosednearly twice the number of small opacities as did theother two readers who were in excellent agreement. How-ever, even that one reader found approximately 6% withsmall opacities, so that in reality, the overall agreementwas quite good. According to the authors, all these abnor-malities occurred in either present or former smokers.Although the data were not presented, the authors statedthat there was virtually complete agreement among thereaders regarding presence or absence of pleural fibrosis.Thirty-five of the welders and only nineteen of thecontrols had pleural fibrosis, with the difference beingstatistically significant in present smokers.

According to the spirometry, approximately twice asmany welders as controls had mixed obstructive andrestrictive respiratory disease with the differences beinggreater among the exsmokers and nonsmokers. Therewas slightly more pure restrictive disease among thewelders and slightly more pure obstructive diseaseamong the controls. However, among the 23 nonsmok-ers in each category, 6 of the welders and none of thecontrols had pure obstructive disease.

With regard to specific pulmonary function tests, theforced expiratory volume in 1 second decreased faster

with age among welders, although this difference was notstatistically significant. Residual volume also increasedover time faster in welders than in controls. The totallung capacity in welders decreased more slowly withincreasing age than in controls.

In summary, the authors found only a few differencesbetween welders and controls and surprisingly littledisease, considering how long the individuals had beenworking. Welders did have a substantially higher preva-lence of dyspnea and of mixed (in terms of restrictive andobstructive) pulmonary disease. However, the higherprevalence of pleural plaques in the welders suggested tothe authors that the welders apparently had higher asbes-tos exposure than did the controls, and this could haveexplained the differences found. Only four of the weldershad "radiographic evidence consistent with the diagnosisof siderosis", which also suggests that the differencesfound are due to asbestos exposure rather than weldingfume. An earlier study by McMillan and Pethybridge(1983) also suggested that it is difficult to study the effectof welding fume in shipyard welders because there was aclear gradient of mesothelioma and asbestosis with weld-ing fume exposure. This suggests that welders, withinthese shipyards at least, were more heavily exposed toasbestos than the other shipyard workers.

In spite of the fact that the authors did not recognizethe correct category into which their epidemiologic studyfell, this is a reasonably well-conducted and well-interpreted study. The only flaw is the failure to take thematching into account in the analysis, with the result thatcorrect relative risk estimates cannot be made and effi-ciency is lost. If the authors have retained their raw data,there is still an opportunity to have the date reanalyzedproperly, taking into account the effect of matching.

Because McMillan and Pethybridge (1984) were un-able to eliminate the confounding effect of asbestosexposure in such an otherwise well-designed study inshipyard welders, it is indeed welcome that Waddingtonet al. (1983) have announced that a clinical study isplanned to define acute and chronic effects of welding onthe respiratory system, with special respect to exposureto hexavalent chromium and nickel and stainless steelwelding fumes. Waddington et al. are also planning touse mild-steel welders as well as nonwelding cohorts asinternal controls and apparently are also proposingexternal controls. The exposed cohort will comprise aminimum of 500 stainless steel welders employed during1960 or earlier, with particular emphasis on finding thewelders employed during 1950 or earlier. The mild-steelwelding cohort is proposed to be larger, but numbers arenot stated. Because part of this study, assessment ofcurrent exposure to welding fumes will be based onchemical analysis of fumes collected on filters for iron,total chromium, total and water-soluble hexavalent

13

chromium, and total and acid soluble nickel, the authorsalso proposed to obtain a lifetime history of occupa-tional exposure to welding fumes and to hexavalentchromium by means of a questionnaire emphasizingwelding methods, relative ranks of exposure, use of elec-trodes or rods, types of steel, and use of primers andcoatings. They also propose, through the analysis ofblood and urine, to evaluate and validate the methods ofassessment of current chromium and nickel exposure inthe welders. They will use an informal questionnaire todetermine current health status of welders, with particu-lar emphasis not only on chronic bronchitis but also ondermatitis and metal fume fever. Pulmonary functionassessment will include at least forced vital capacity,forced expiratory volume in 1 minute, the ratio of forcedexpiratory volume to forced vital capacity and FEF25 75

and FEF75 75, with additional pulmonary function testslikely as well. They are also planning to carry out sputumcytology on the subjects. The results of this study, if it iscarried out as planned, should be of great interest.

McMillan (1983) summarized all the shipyard weldingstudies but did not discuss the methods or present theresults in detail. Fortunately, almost all of the discussionin this paper summarizes work of papers that have beenreviewed in detail either in this report or in its predeces-sors. However, a small amount of the discussion in thispaper (McMillan, 1983) has not appeared elsewhereand will be discussed here. As described earlier in thissection, there were 328 participating welders. Of these328, 18 were diagnosed as having the pneumoconiosis,siderosis. According to McMillan, "in the 18 men withsiderosis, there was no constant relationship betweenradiographic abnormality and length of employment asa welder, clinical abnormality of the chest, or lung func-tion impairment". McMillan interpreted the results asshowing low exposure to welding fumes among thedockyard welders, but in this summary paper there isinsufficient information to determine the correctness ofthe interpretation.

Also according to McMillan (1983), "a high (other-wise unspecified) proportion of the welders had sufferedfrom metal fume fever, usually due to welding on galvan-ized surfaces." Because McMillan (1983) also reportedfrequent eye injuries and burns and little use of protectiveequipment, the metal fume fever may also have resultedfrom inadequate work-protection practices, as McMil-lan implies in the summary.

McMillan (1985) recently presented a paper about riskof death from respiratory disease among welders by typeof exposure. Unfortunately, only an abstract is availablefor review, and a full discussion of the study will need toawait publication of the entire proceedings. McMillandiscusses the fact that the Registrar General for Englandand Wales had published decennial supplements of occu-

pational mortality, which had been frequently quoted bythose arguing that welders were at special risk of deathfrom respiratory disease. During the period 1970 to1972, there were 149 deaths from pneumonia, bronchitis,emphysema, or asthma among the Registrar General'sgrouping, gas and electric welders, steel cutters, andbraziers. Twelve of these were in burners or cutters, who— as McMillan points out — have different exposures.However, they make up only about 8% of the deaths;therefore, even if their mortality experience is different, itshould not affect the results very much. Since potentialhealth hazards can vary dramatically from one weldingprocess to another, it is unfortunate that only 39 of the137 death certificates for the welders gave any indicationof the process. (This abstract also supports the need forthe studies proposed by Waddington and colleagues.)McMillan then goes on to state appropriately that theRegistrar General's statistics should be thought of asgenerating hypotheses rather than testing them.

At the same symposium, Newhouse et al. (1985)reported on mortality of mild-steel welders at a shipyardin northeast England. According to the abstract, which isall that was available for this review, the welders showedan excess mortality from pneumonia, but the magnitudeof the SMR was not given.

In a study available for review only as an abstract,Mur et al. (1985) conducted surveys on the pulmonaryfunction of welders from one electromechanical plantand two factories producing industrial vehicles. A totalof 536 welders from the three plants and 558 controlsfrom the same plants were given a standard bronchitisquestionnaire, a clinical examination, chest radiophoto-graphy, and the following pulmonary function tests:flow-volume curves taken before and after acetylcholine-induced bronchial constriction, carbon monoxide lungtransport tests by the single-breath and steady-statemethods, and an N2 washout test.

Bronchitis and some specific radiological abnormali-ties such as nonsiderotic reticulomicronodules occurredslightly more often, but not significantly more often(statistics not reported), in welders than in controls. Nosignificant differences were observed in any of the pul-monary function tests between welders and controlsexcept the carbon monoxide transport test. Pulmonarytransport of carbon monoxide was significantly impaired(statistics not reported) among welders, especiallywelders who were smokers. Based on these results, theauthors requested that the carbon monoxide transporttest be used to monitor pulmonary function in welders.

A brief report of a cross-sectional study of welders fromEast Germany was published in English (Schneider et al.,1981). Perhaps because the paper is only two pages long,there are not enough details on the methodology to knowwhether the controls were selected appropriately. A

14

higher proportion of the welders complained of "drynessof the upper airways" and chronic sputum productionwithin each category of nonsmokers, exsmokers, andpresent smokers. According to the authors, there wereno differences in pulmonary function tests between weld-ers and controls, although the specific results were notpresented. However, in the absence of information on thequality of the study design, no conclusions can be made.

Fawer et al. (1982) carried out a small-scale investiga-tion of absences due to sickness among all 50 welderswho had worked between January 1, 1970, and Decem-ber 31, 1979, in a petrochemical plant. Fourteen wereexcluded from the study, thirteen because they hadworked less than one year or because of insufficientinformation. The remaining 36 welders were comparedwith 36 controls from the same social class within thesame plant. Each control was individually matched bydate of birth (within five years), duration of employmentduring that decade, date of hire (within three years),degree of fitness at the pre-employment examination(not specified otherwise), and smoking habits (appar-ently only in terms of whether they were nonsmokers,former smokers, or present smokers). Unfortunately, thestudy is weakened because after going to all the troubleto match controls individually, the authors ignoredmatching in the analysis. This is unfortunate because theauthors handled the often tricky problems of studydesign reasonably well. The controls are selected muchbetter than is usually the case, and the problems withascertainment of sickness absence were well handled asfollows. Although diagnoses for sickness absences are notsystematically validated, each employee is reviewed bythe company medical center if they are absent for morethan three days. The authors therefore confined theiranalysis to absences of more than three days for illness.

For all illnesses combined, the average number of dayssick, average length of each individual illness absence (bydefinition, over three days), the total number of spells,and the proportion of workers having one or more newspells were each similar in the welders and the controls.However, for respiratory diseases, the average numberof sick days for welders was 2.5 times that of the controlsand the total number of new spells was 80% higher forthe welders. The average length of the spell was about40% longer for the welders, and the proportion ofworkers with one or more new spells was 28% higheramong the welders. When the respiratory diseases weredivided into upper and lower respiratory tract, differenttrends emerged. For the average number of new spellsand the proportion of workers with more than one spell,very little difference between the welders and the controlswas seen for upper respiratory tract diseases. For lowerrespiratory tract diseases, however, the rate of new spellswas 5 times as high and the proportion of workers with

more than one spell was approximately 2.5 times as highamong welders. For the average number of sick dayslost, the trend in difference between welders and controlswas more marked for lower respiratory tract diseases,but was nearly twice as large for the upper respiratorytract diseases as well. For the average duration of aparticular sickness absence, the only differences seenwere for upper respiratory tract diseases, with thewelders' absences being an average of over 40% longer.For lower respiratory tract diseases, an interaction wasseen with smoking, with the differences between weldersand controls — at least in terms of average duration ofeach absence and the proportion of employees with oneor more new absences — being more striking for smok-ers in welders than in the controls. Surprisingly, thesmokers among the controls had slightly better indicesthan the nonsmokers. This may indicate that the smok-ers among the controls were younger, because smokerswere not matched to nonsmokers, or it may indicate thatthe smokers in the controls were relatively light smokers.If the latter were true, it would be hard to interpret thegreater differences between welders and controls seen inthese statistics among smokers; in that hypotheticalsituation, the welders might be heavier smokers. In anyevent, since the nonsmokers are comparable in theamount smoked and are matched, and therefore com-parable on other factors (even though the matching wasnot properly handled in the analysis), the differencesbetween welders and controls cannot be explained solelyby smoking habits.

Although the authors presented no data, they statedthat there were no differences between welders and con-trols in absences attributed to digestive diseases. Diges-tive diseases were analyzed because they were thought tobe unrelated to welding exposures and would be anindicator of bias in ascertaining sickness absence unre-lated to welding exposures.

In spite of the less-than-perfect analysis, the gooddesign of the study and the magnitudes of the differencesindicate that welders are more likely to be absent becauseof respiratory diseases, particularly lower respiratorydiseases, and to be absent longer from work with theseconditions than similar workers — at least in the com-pany studies (to the extent that the relatively smallnumbers permit generalization). It is not clear how clini-cally significant these increased absences are in terms oflong-range health effects. The authors did note fourabsences from pneumonia in the welders and none in thecontrols and pointed out that earlier investigationsreviewed in earlier editions of the Effects of Welding onHealth had found elevated SMRs for pneumonia inwelders. The authors specifically note that even thoughthey were able to control for social class, age, smokinghabits (at least in terms of present, former, and non-

15

smokers), and duration of employment, the results mighthave been influenced by other factors not controlled inthe study; e.g., attitude of supervisors, personality of theworker, and job satisfaction. The authors also point outthat the welders may consider themselves at higher riskof respiratory disease, and this belief may influence notonly their behavior but also their physician's diagnosis.

There remains another finding implied in the resultspresented by Fawer et al. (1982). For all diseases —other than respiratory — combined, the welders appar-ently have fewer absences and shorter absences than thecontrols; otherwise, the welders would have an excess inthese indices for all absences. This is shown by the find-ing that 56% of the days lost in welders and 24% of thedays lost in the controls were due to respiratory diseases,and 65% of the absences in the welders and 40% of theabsences in the controls were due to respiratory diseases.Therefore, even in the controls, respiratory diseasesmade up a large enough proportion of the sick daysabsences that something had to compensate for the sim-ilarity of findings in welders and controls for all sick daysand sick absences. This could mean that the controls areat increased risk for other diseases, or it could mean thatthe welders are in general healthier than the controls, inwhich case the absences pattern for respiratory diseasesis even more striking.

In summary, at least in this facility, Fawer et al. haveshown that for respiratory disease alone, and primarilylower respiratory disease, welders have more sick days,more sick absences, and greater severity in terms ofnumber of days absent from work for each sicknessabsence than do comparable workers. Although theimmediate clinical consequences of this finding are clear,the long-range implication for the welders' health isunclear from this study alone.

In a study published so far only as an abstract, DiMaria et al. (1984) examined 139 white male workers(those who had been exposed for at least five years) of169 white male workers at a metal factory where electricarc welding was a common practice. However, ninesubjects were excluded because they were unable to pro-duce satisfactory pulmonary function tests. Thirty-sevenmale employees were included as controls, but theabstract does not state how the controls were selected, socomparability of controls with the exposed individuals isnot clear. The authors obtained pulmonary functiontests and applied a questionnaire, presumably the Medi-cal Research Council standardized questionnaire on res-piratory diseases (but the abstract does not so state).Smokers and nonsmokers were analyzed separately. Theexposed workers, including the welders, had an averageof 14 pack years among their smokers in comparisonwith an average of 10.6 pack years among the smokingcontrols. This suggests that the two groups of smokers

were not comparable. Among the nonsmokers, only theexposed group had symptoms of cough, sputum, and ahistory compatible with the standard definition ofchronic bronchitis. In the smokers, the controls weremore likely to report cough, but the sputum-productionhistory and chronic-bronchitis-symptom history had ahigher prevalence among the smoking welders. In bothsmokers and nonsmokers, the welders were more likelyto have reduced Forced Expiratory Volume,, but thecontrols were more likely to have a reduced MaximumExpiratory Flow75. In both welders and controls, asmight be expected, the smokers were more likely to haverespiratory symptoms and abnormal pulmonary functiontests. The interpretation of this study will have to awaitpublication of a full report because the comparability ofthe controls cannot be evaluated from the abstract.

Keimig et al. (1983) reported the results of a cross-sectional study that was particularly designed to exam-ine mild-steel welders without asbestos exposure. Theinvestigation was carried out in a John Deere factorythat manufactured bulldozers, road graders, and heavyconstruction equipment of other types. The plant con-sisted of a series of interconnected buildings, whichhoused different parts of the production process. In twoof these buildings, gas metal arc welding and flux coredarc welding on mild steel were carried out by welderswho were not concurrently exposed to asbestos. Thecontrols were hourly paid workers in other buildingswho had minimal exposure to known respiratory irri-tants and who had never been employed as welders. Theparticipants were white males between the ages of 25 and49 who had been employed at the plant for at least fouryears, who were employed on the first shift at the time ofthe study, and who had minimal past employment infoundry, diesel, or solvent fume exposure type jobs.Nonsmokers who had smoked less than 20 packs ofcigarettes or 12 ounces of tobacco in a lifetime and hadnever smoked more than one cigar a week for one yearwere compared separately from smokers who weredefined as smoking at least one cigarette per day for onemonth before the pulmonary testing. Mean ages ofwelders and controls were comparable, as were meanweights except for the welders who smoked, who wereover 10 pounds lighter than all other groups, includingthe nonsmoking welders. The welders were an average ofapproximately 1.5 inches shorter than the controls,among both the nonsmokers and the smokers. The con-trols had been employed an average of one to two yearslonger at the plant, and the welders had an average ofapproximately nine years of welding exposure. Amongthe smokers, the controls had approximately 25% morepack years of smoking.

Welders and controls were scheduled for question-naire and examination together throughout each week to

16

avoid the possibility of any daily or weekly variability inpulmonary function measurements. On their scheduledday, participants reported to the pulmonary testing siteat the beginning of the first shift, before any workplaceexposures. A standardized, self-administered AmericanThoracic Society questionnaire was completed beforepulmonary testing, which included Forced Vital Capac-ity, Forced Expiratory Volume, and Forced ExpiratoryFlow measurements taken by one investigator on one pul-monary function instrument, under standard conditions.

The study population consisted of 46 nonsmokingwelders, 45 smoking welders, 35 nonsmoking controls,and 45 smoking controls. No welder had welded morethan 214 months. Twenty-two of the ninety-one weldershad welded outside of the plant; according to theauthors, they had primarily been production manual arcwelders of mild steel in apprenticeship programs in smallmanufacturing companies. Fifty-five of the controls wereemployed in machining departments, eighteen in assem-bly departments, and seven in shipping departments.

The authors presented questionnaire results for cough,sputum, number of episodes of cough or sputum, wheez-ing with colds and without colds, shortness of breath,whether colds went to the chest, and history of bronchi-tis, pneumonia, and asthma. Among both welders andcontrols, smokers had a higher prevalence of respiratorysymptoms or history of disease with one exception: therewas considerably less asthma among smokers than non-smokers. Cough sputum were more frequent in thewelders in both smokers and nonsmokers. The numberof episodes of cough or sputum and wheezing with a coldwere more prevalent in welders among the nonsmokers,with essentially no difference between welders and con-trols among the smokers. Several symptoms or diseasehistories, including wheezing in the absence of a cold,whether colds went to the chest, and a history of bronchi-tis, were more prevalent among welders in nonsmokersand more prevalent among the controls in the smokers.(As noted above, the smoking controls smoked morepack-years than the smoking welders.) Welders whosmoked, reported a higher prevalence of pneumoniathan did control smokers, but the results for the non-smokers were very similar for welders and controls.Essentially no difference was seen for prevalence ofshortness of breath symptoms or asthma; if anything,these were more prevalent among the controls.

For the pulmonary function testing, the authorsunfortunately only reported mean values. As is wellknown to epidemiologists (see, for example, the discus-sion on page 267 of MacMahon and Pugh, 1970) acomparison of distributions gives a clearer picture of riskassociated with an exposure than does a comparison interms of means or other summary statistics. For theForced Expiratory Flow statistics, the means for the

welders are consistently lower (by 3 to 7%) than for thecontrols — both for smokers and nonsmokers. Report-ing these findings as means may obscure a substantialdifference in risk, but without the distributions, we donot know whether this is the case. For Forced VitalCapacity, the welders had lower values than the controlsamong the nonsmokers but higher values than the con-trols for the smokers. In fact, the smoking welders hadhigher values than the nonsmoking welders and meansthat were practically identical to the correspondingmeans for the nonsmoking controls. For Forced Expira-tory Volume,, the means for the welders were loweramong the nonsmokers but were higher than the meansfor the controls among the smokers who (as notedabove) had greater average pack-years. Once again, forboth of these pulmonary function tests, because distribu-tions were not given, it is not clear whether the smalldifferences seen reflect substantial differences or virtu-ally no differences in risk.

The authors point out that the lower pulmonary func-tion measurements in welders relative to controls mightbe interpreted as a true welding effect that was observed,perhaps because the individuals most susceptible to theirritating effects of welding fumes could have transferredto other jobs, leaving less susceptible welders to bestudied, or because the welding exposures, which wererelatively short, may have been too brief for the effects toshow up. On the other hand, the authors note that theirresults are compatible with welding exposures not beingassociated with any pulmonary impairment. It is indeedunfortunate that they did not present distributions,which would help resolve these two possibilities. As theauthors note, their results are equivocal, probablybecause of the way that the data are presented. Differ-ences in the peripheral airway measurements could betrivial or they could represent the initial and possiblyreversible stages of chronic obstructive pulmonary dis-ease. The authors suggest that long-term follow-up pro-vided by a large cohort study would be needed. Such astudy of course could determine the fate of welders wholeft employment. Several such studies were reviewed inEffects of Welding on Health IV, and if there were anincreased risk of respiratory diseases from welding,increased risk of death from respiratory diseases, emphy-sema, or other chronic respiratory diseases they wouldhave been noted in these studies. The absence of such in-creased risks could be taken as evidence that there are noserious long-term respiratory health effects from welding.

In general, the Keimig et al. (1983) study is welldesigned, particularly in terms of control selection, butthe data are presented in such a way that the results aredifficult to interpret. The authors themselves have notedthat cohort studies offer better means of answering thequestion of whether there are serious health effects from

17

respiratory diseases as a result of welding exposure.Because these studies have already been done and werereviewed previously, their findings, if the studies weredesigned properly, should be taken as the best indicatorsof whether there is increased risk of serious respiratorydisease from welding exposures.

Lyngenbo et al. (1985) examined pulmonary functionand respiratory symptoms in highly exposed Danishelectric arc mild-steel welders, but this is only available asan abstract for this review. After 74 nonsmoking weldersand 31 matched controls completed the Medical ResearchCouncil standardized questionnaire on respiratory symp-toms, the welders had a statistically significant (p -0.001) higher prevalence of upper respiratory symptoms,such as colds, hoarseness, and inflammation of thethroat, and lower respiratory symptoms, such as cough,chronic bronchitis, dyspnea, and wheezing. The abstractprovides no information on how much higher the pre-valence was in welders or how the controls werematched. According to the authors, a dose-effectrelationship between welding fume exposure and upperrespiratory symptoms was observed. The lung functiontests showed that the welders have a statistically signifi-cant reduced TLC (total lung capacity), VC (vital capac-ity) (P - 0.008), FEV1 (forced expiratory volume in1 second) (p - 0.025), FVC (forced vital capacity)(p = 0.001), PEF (p = 0.0013), MEF75 (p = 0.03), TCO(p = 0.02), and slope of alveolar plateau (p = 0.02). Noinformation on the magnitude of the decrease was pro-vided in the abstract. The investigators concluded thatwelding fumes bring about upper and lower respiratorysymptoms and reduced lung function parameters. Evalu-ation of this investigation cannot be made until the fullaccount of the methods and results is available.

Kalliomaki et al. (1982) investigated respiratory symp-toms and pulmonary function of manual metal arc(MMA) stainless steel welders. The authors noted thatalthough MMA welders of mild steel had been exten-sively investigated, there had been very little reported oninvestigations of respiratory effects of stainless steelMMA welders. Because the fume emission of the MMAprocess is much higher than that of the tungsten inert gas(TIG) process (at least 100 times as high), the authors feltthat stainless steel TIG welders could be, in effect, anonexposed comparison group. The authors investi-gated 83 stainless steel welders from a shipyard withminimal exposure to mild steel welding fumes and otherindustrial metal aerosols (MMA mild-steel weldingapproximately 11 % of their work time during the last 10years, according to the authors) who have been weldingstainless steel for at least 5 years. They also investigated20 mild-steel MMA welders from one of the shipyards,presumably one of those from which they had obtainedsome stainless steel welders. The stainless steel welders

were divided into four groups according to the percent oftime spent on MMA welding, with the remainder of thetime being spent on tungsten inert gas welding. Appar-ently (the word "apparently" is used frequently in thisdescription because the descriptions are less completethan ideal and are occasionally ambiguous), the percent-ages of time that the four groups spent on MMA are asfollows: first group of 20, less than 10% (average of 3%);second group of 16, 11 to 40% (average of 30%); thirdgroup of 18, 50 to 79% (assuming a typographical errorin the table) (average of 58% approximately); and fourthgroup of 29, over 80% (average 96%). In one shipyard,presumably one of several shipyards from which data onstainless steel welders were obtained, 29 mild-steelwelders were also investigated. Apparently, this was donebecause the authors wanted to directly compare stainless-steel MMA welding with mild-steel MMA welding usingthe same methods. In comparing the groups on otherparameters, the proportion of smokers was higher in thestainless steel manual arc welders greater than 80%MMA and the 11 to 40% MMA stainless steel welderswith the proportion of smokers being lowest in the 50 to79% MMA in stainless steel welding group. Regardingmean ages, all that the authors presented was that the 50to 79% MMA stainless steel welders were somewhatyounger and the greater-than-80% stainless steel welders,and particularly the mild-steel welders, were somewhatolder than the other welders studied. The mean durationof exposure was somewhat shorter for the two groupswith 11 to 79% MMA and was considerably longer forthe greater-than-80% MMA and for the mild steelwelders. In making their comparisons, the authorsapparently did not adjust for age and exposure differ-ences but did attempt to measure the smoking parameterin an unspecified manner, but presumably by comparingthe nonsmokers with the smokers regardless of amountsmoked.

Data collected included the Medical Research Coun-cil standard questionnaire on chronic bronchitis, adetailed questionnaire (otherwise unspecified) on sub-jective respiratory symptoms "associated with welding",and a detailed occupational history. Pulmonary functiontests included maximum expiratory flow, forced expira-tory volume in 1 second, maximum expiratory flow at50% and 25% of the forced vital capacity, and the closingvolume. The authors apparently regressed these dataagainst the MMA welding in the stainless steel welders,using each of the four groups as a separate data point.The slope of the regression line was used to estimate theeffect of the MMA welding exposure. The effect ofsmoking apparently was estimated by comparing smok-ers with nonsmokers, but other than stating that theeffect of smoking was determined, the authors providedno information on how it was done.

18

Dust retained in the lungs increased with increasingpercentage of time spent in MMA welding among thestainless steel welders. The mean amount of lung dustwas approximately 20 times as high in the over-80%group than for the less-than-10% group. Interestingly,this value for the over-80% group among the stainlesssteel welders was approximately 4 times that seen in themild steel welders. Urinary chromium concentrationincreased among the stainless steel welders with increas-ing proportion of time spent in MMA welding, particu-larly for the over 80% MMA group which had a urinarychromium concentration approximately 6.5 times thatof the less than 10% MMA group. Cough, sputum, grade2-4 dyspnea, and chronic bronchitis increased withincreasing proportion at times spent on manual metalarc welding. According to the figures provided by theauthors, the magnitude of these effects was similar to themagnitude of the effect of smoking. For irritation of theupper respiratory tract and attacks of dyspnea asso-ciated with welding, determined as subjective symptoms,the increase with increased time spent on MMA weldingwas considerably more marked than the effect of smok-ing. In fact, attacks of dyspnea associated with weldingdecreased with smoking. Among pulmonary functions,vital capacity, forced expiratory volume, and maximumexpiratory flow decreased with increasing MMA weld-ing exposure and with smoking. The effects of the weld-ing exposure were comparable to or slightly greater thanthe effects of smoking. The increase in closing volumedifference was more marked in smoking than the corre-sponding increase with increased MMA welding. Accord-ing to the authors, the prevalence of symptoms ofchronic bronchitis was dependent only on smoking; themost prevalent symptom was chronic rhinitis. Althoughthe mean vital capacity of all groups was within normallimits, it decreased faster than predicted as a function ofexposure time. The estimated effects of welding exposure,as determined by the authors, seem to be more markedin the mild-steel than in the stainless steel welders.

The authors concluded that welders cannot be consid-ered a homogenous group in respect to exposure andpossible associated health hazards. They specificallyconcluded, as implied above, that the pulmonary func-tion of the mild-steel MMA welders was more seriouslyaffected than that of the stainless-steel welders. They alsonoted that "according to our preliminary study," pulmo-nary function tests indicate that pathological changesoccur mainly in the peripheral airways and alveoli. Thepollutants from welding mainly accumulated in thelower parts of respiratory organs. They noted that it isnow recognized that vital capacity forced expiratoryvolume in 1 second, percentage of forced expiratoryvolume in 1 second, and percentage of forced vital capac-ity are not sensitive enough to detect changes in small

peripheral airways. They suggested that results of pre-vious spirometric studies that compared welders andcomparison groups "conflicted" because these indiceswere not sensitive enough.

The most common symptoms of vanadium pentoxideexposure are respiratory tract irritation (i.e., coughing,wheezing), chest pain or discomfort, dyspnea, and eyeand throat irritation. These health problems occurredin at least 74/100 (all who could be examined) boiler-makers (mainly welders) who were exposed to high levels(0.05 to 5.3 mg/m3) of vanadium pentoxide (a volatilefuel oil constituent) fumes during an oil-to-coal conver-sion at a Massachusetts utility power plant company(Levy et al., 1984). The vanadium pentoxide levels mea-sured in the air samples were at or above the OSHApermissible exposure limit (PEL) of 0.05 mg/ m3, thoughchromium, nickel, copper, and iron oxide fumes werewithin acceptable limits. The most frequent symptomreported from the questionnaire survey was a cough withsputum (85%): wheezing was the most frequent signfound during the physical examination. Mild hypoxemiawas noted in several workers. The chest x-rays werenormal for 72 of the subjects. The pulmonary functiontests revealed that FVC was 87%, FEV1 was 93% ofpredicted value, FEV1/FVC was 79%, and the expira-tory flow rate over the middle 50% of the forced vitalcapacity was the most affected (median, 57% of pre-dicted).

Limitations of the study recognized by the authorsincluded: (1) it relied on self-administered question-naires; (2) its inability to obtain health status of 26workers; (3) clinical and laboratory information wasobtained from various sources and so may not have beencomparable; and (4) the investigation was performedafter exposure had ceased.

El-Gamal et al. (1985), in a presentation at a sympo-sium reported only as an abstract, determined the rela-tive sensitivities of single breath nitrogen and flowvolume indices as methods for detection of early respira-tory impairments in young men. Five hundred thirteenyoung shipyard workers (aged 18 to 47 years) were exam-ined, and the following tests were administered: an-thropometry, ventilatory capacity, flow-volume curves,closing volume and Fowler nitrogen test, and respiratoryunspecified and occupational questionnaires. The ship-yard workers were divided into three categories: welders,caulker/ burners, and other tradesmen. According to theauthors, occupational fumes were directly related todeterioration of lung function and accentuated theeffects of age and smoking. In this study, the single-breath nitrogen test (including closing volume) was morediscriminatory than the flow-volume curve. It was un-clear from the abstract which group of workers the inves-tigators were referring to when discussing the results.

19

A follow-up study of 47 shipyard arc welders from1976 to 1982 was presented by Kalliomaki et al. (1985) inthe same symposium, so only the abstract was available.Though this six year follow-up period was too short toexamine changes in the lung function parameters, pre-liminary tests performed included spirometry, closingvolume, the carbon monoxide transfer factor, andmagnetopneumography. The results from the magne-topneumography indicated a mean clearance rate of dustin the lungs of 20% per year (range 10 to 30%). Theresults from the pulmonary function tests reveal that theretired welders were significantly below normal values,whereas the younger welders were within normal limits.Further evaluation of this study cannot be done in theabsence of the complete presentation.

Unlike in the previous studies, Boylen et al. (1984)determined the acute effects of different fumes generatedduring welding on shipyard welders; 45 aluminumwelders, 27 welders of stainless steel, and 74 mild steelwelders were questioned and spirometry tests were per-formed on Monday before the usual workshift. Theauthors stated that 50% of nonsmoking aluminumwelders complained of chest tightness and 25% of allother welders experienced chest tightness. Results fromspirometry tests revealed that 36% of mild and stainlesssteel welders who smoked showed a decrease in flowrates, and 36% of stainless steel welders who smokedshowed reduced vital capacity. Twenty-five percent of allother welders had a decrease in flow rates and sixteenpercent had a decrease in FVC > 5%. Therefore, mild-steel welding reduces flow rates and stainless-steel weld-ing reduces flow rates as well as vital capacity. Fromthese preliminary results, the authors stated that itappears that symptoms of pulmonary irritation andmetal fume fever were associated more with aluminumwelding than with stainless steel and mild steel welding.Because the paper was available only as an abstract,complete evaluation of the work cannot be made.

Estimation of Retained Particles in the Lung. Mag-netopneumography was developed over a decade ago forthe measurement of ferromagnetic particles (primarilymagnetite) in the lungs and has been used by severalresearch groups to estimate retention of particles in thelungs of welders. This effort is relevant since ferromag-netic particles constitute as much as 30% of weldingfumes and dust (Lippman, 1985). Basically, in thisprocedure, a strong magnetic field is applied to the chestfor a few seconds, thereby magnetizing the lungs and anyparticles in them by creating dipoles within the particlesto rotate the magnetized particles into common align-ment. The magnetic field is then removed and theremaining magnetic field, or remanent field, producedby the magnetized particles within the lungs and the

relaxation times are measured with a magnetometer. Thestrength of the remanent field at the time the magnetizingfield is removed is related to the amount of dust in thelungs. The relaxation time is thought to be due to aprocess of random reorientation of the particles in thelungs. Relaxation times have been hypothesized to bemediated by such factors as age, degree of retention,smoking habits, airway diseases, the length of time thatthe particles have been retained in the respiratoryapparatus, particle size, and magnetopneumographictechnique. To date, three different magnetopneumo-graphic techniques have been used: the uniform fieldmethod, in which the entire thorax is magnetized forabout 15 seconds; the localized field method; and thealternating current bridge.

Naslund and Hogstedt (1982) measured ferromag-netic levels in 187 welders with at least five years expe-rience and homogenous exposure using the uniform fieldmethod and related it to frequency of chronic bronchitis,as determined by the BMRC questionnaire for chronicbronchitis. Welders showed an increased magnetitedeposition compared with two age-matched controlgroups. Interestingly, as previously reported {Effects ofWelding on Health IV) smokers had lower levels ofremaining ferromagnetic particles than did nonsmokers,an observation that the authors attributed to fastertracheobronchial clearance due to increased mucousproduction. The authors observed a dose-response rela-tionship in the incidence of chronic bronchitis in smokersand nonsmokers by magnetopneumography, an effectconfirmed in further studies by the same researchers,Hogstedt et al. (1983) and by Freedman et al. (1982,1984), who used a localized field technique. Freedman etal. also found that ferromagnetic content correlated wellwith the number of years of welding (p < 0.01, Spear-man rank correlation). It is important to note thatbecause ferromagnetic retention is not occupation-specific, an extensive job and hobby history should beobtained from all individuals studied.

Pilot studies have recently been conducted using a newmagnetopneumography field technique based on an alter-nating current susceptibility bridge. The measurementderived is theoretically less sensitive to the relaxationmechanisms, which introduce a time-dependence to theremanent field after magnetization and therefore mustbe measured by uniform field and localized field tech-niques. Susceptibility bridge measurements are sensitiveto all forms of magnetic material: diamagnetic, paramag-netic, and ferromagnetic substances of both endogeneousand exogenous origin (Stern and Drenck, 1982). Usingthis method, Stern et al. (1985) have found a high degreeof correlation (R2 = 0.49, p = 0.0001) between medianinferred lung burden of welding fumes and relative life-time occupational exposure in 59 nonsmoking shipyard

20

welders with more than five years of work experience.Net thoracic magnetic moment was measured to deter-mine retention of welding fumes. The change in thoraxmagnetic moment observed in a welding cohort, com-pared with controls who were electricians, is equivalentto a median of 110 mg magnetite. This corresponds to amedian lung burden of 220 mg of magnetite. No differ-ence was observed in chronic bronchitis incidence oraverage lung function parameters. The researchers con-cluded that this field method would not be useful fordetecting clinical abnormalities until further baselinedata were collected, especially from studies that followindividuals from entry into the welding profession acrossa significant fraction of their working lifetimes.

2.2 Metal Fume Fever. Ulvik (1983) reported a case ofa healthy 34-year-old welder suffering from zinc fever.The welder was occupationally exposed for approxi-mately five hours to welding fumes that penetrated hismask because of a defective filter. These fumes wereanalyzed and found to contain 3.5% zinc. Further analy-sis of the fumes did not reveal the presence of leadand cadmium, which often contaminate zinc-containingmaterial and may also cause similar symptoms.

Symptoms of zinc fever began to appear 2 hours afterexposure and reached a plateau after 5 to 6 hours beforegradually subsiding within 24 hours after the diseasestarted. Blood samples were taken at 6 hours, 23 hours,and 6 days (normal reference value) after onset of symp-toms. The leucocyte count and lactate dehydrogenaseconcentration were expectedly high, but the erythrocytesedimentation rate, hemoglobin concentration, serumprotein electrophoreses, bilirubin, and electrolytes werenormal. Relative to the patient's own reference value(day 6), serum zinc concentrations were decreased byapproximately 15% and 29% at 6 hours and 23 hours,respectively, after the symptoms appeared.

Serum zinc levels have been reported only rarely in thezinc fever literature. One 1926 study reported elevatedzinc levels, but Ulvik questioned the accuracy of methodsin use then. Two studies have reported normal levels. Inone, however, the serum zinc levels were not measureduntil two weeks after the initial exposure; in the other,neither the time after exposure at which the blood sam-ple was analyzed nor the method and its limitations aredescribed.

Ulvik's study does not support the notion that increasedserum zinc concentration is a basic part of the pathogenicmechanism in zinc fever. The investigator is uncertainwhy the serum zinc concentration was lower in the acutephase of zinc fever but cites a previous hypothesis that itis a nonspecific response to local inflammation in thelung tissue. (Low serum zinc is a common characteristicof inflammatory conditions and tissue damage.) The

author suggests that more detailed studies are needed toclarify the pathogenic mechanism of zinc fever.

Acute exposure to aluminum can also cause metalfume fever. In a paper discussed in detail elsewhere,Boylen et al. (1984) reported that 65% of the nonsmok-ing aluminum welders and 33% of the stainless steel andmild steel welders experienced metal fume fever symp-toms. It is apparent from the report that metal fume feveris more prevalent among aluminum welders than amongstainless steel or mild steel welders.

2.3 Effects on the Cardiovascular System. Changes inthe bronchopulmonary and cardiovascular systems ofabout 500 welders exposed to low- and high-alloyed steelwelding relative to 200 control workers were described ata recent meeting by Krasnyuk et al. (1985), but only anabstract was available for review. Chronic bronchitiswas detected in low-alloyed steel welders, and neuro-circulatory dystony and metabolic changes in the cardio-vascular system were found in the high-alloyed steelwelders. Details on the nature of these metabolic changeswere not provided in the abstract. The differences foundin the pathology of respiratory and cardiovascular sys-tems were attributed to the different chemical composi-tions of aerosols generated by high- and low-alloyedsteels. This conclusion was based on the higher contentof chromium in blood and manganese in hair of thehigh-alloyed steel welders in comparison with the low-alloyed steel welders. No other details of the experimentwere provided.

2.4 Effects on Kidney Function. The etiology of kidneydiseases induced by metal welding fumes is unclear.Epidemiological evidence of the occurrence of kidneydisease among welders is generally lacking except for afew recent studies attempting to demonstrate a correla-tion between welding and renal disease.

Because of an apparent cluster of kidney disease inwelders, Lindquist (1983) reported on a pilot study inUmea, Sweden. From 1978 through 1981, 18 welderswere admitted to the Umea Medical Department withkidney disease (Lindquist, 1983). One woman and seven-teen men with an average age of 34 years (range, 19 to65 years) were diagnosed by kidney biopsy immuno-fluorescence. The kidney biopsy material was not ana-lyzed for the presence of metals. Ten patients had weldedfor more than ten years, and eight for less. Actual expo-sure levels were not reported, but a subjective descriptionof either "poor" or "good" ventilation was provided.

Six welders were diagnosed as having interstitialnephritis, three with IgA glomerular nephritis, two withmembranous glomerular nephritis, one with glomerularnephritis type minimalis, and two with arteritis. Fourpatients had only proteinuria/hematuria. An obvious

21

correlation between the quality of ventilation and se-verity of these pathological changes could not bedemonstrated by the investigators. Lindquist did notreport on the number of welders in the area served byUmea Hospital. If this is an area of heavy welding, 18welders hospitalized for kidney disease may not be inexcess. The authors specify that it is unclear whether therisk is eliminated. The epidemiologic studies that areplanned should answer that question.

Changes in low molecular weight (LMW) proteinuria,an indicator of defective reabsorption in the proximaltubuli, and /3-hexosaminidase excretion in urine, whichis a sign of damaged tubular cells, are sensitive indicatorsof renal disease. In a study by Littorin et al. (1984) usingthese methods, no signs of kidney damage were detectedin 17 male, manual metal-arc stainless steel welderswhose mean age was 45 years old (range, 34 to 64) andwhose exposure to welding was considered "intense" fora mean period of 20 years (range, 17 to 41). For eachwelder, an individual matched on age (within 5 years),sex, smoking, approximate alcohol intake, social class,and drug consumption was selected. The authors did notprovide information on the sampling frame for the con-trols, on the criteria for control, or on factors besides ageor sex other than that the controls were individualswithout occupational exposure to stainless steel orchromium. Interviews and clinical examinations revealedthat only one welder was under treatment for hyperten-sion; all other subjects showed no evidence of significantdisease. Analysis of blood and urine samples revealed nosignificant differences between welders and controls forerythrocyte sedimentation rate, hemoglobin level, ery-throcyte count, white cell count, and uric acid, albumin,creatinine, and glucose levels, although the authors didnot present the data in detail. They noted that chromiumhas previously been associated with proximal tubularkidney damage in experimental animals and also inhumans after heavy oral doses of chromium. Therefore,the chromium content in the urine of welders wasassayed by Littorin et al. (1984) using the direct flame-less atomic absorption spectroscopy method. Urinarychromium levels, both in the morning and afternoon,were found to be far higher in welders than in controls(median 23 vs 1.5 jumole creatinine,/? < 0.001), indicat-ing an accumulation of chromium in the kidney. Unfor-tunately, the authors ignored the matching in the analy-sis of chromium, and thus the true association is strongerby an unknown amount. However, the amount of chro-mium in the kidneys of these welders was considered bythe authors to be probably much less than in animalexperiments or in previous reports of accidentallypoisoned humans. The authors concluded that there wasnot evidence of damage to either the glomeruli or renaltubules in these stainless steel welders.

Chronic renal tubular dysfunction was observed in astudy of 22 welders, 22 to 55 years of age, who had beenexposed to cadmium oxide fumes for a period of 7months to 23 years (Toda et al., 1984). Welders weredivided into three groups (high-, middle-, and low-excretion) based on determination of urinary cadmiumlevels, which varied from 5.7 to 184.9 /ug per day. Cad-mium excretion had a correlation coefficient of 0.39 withnumber of years of cadmium exposure. Hematological,liver function, respiratory function, and radiologicalexamination confirmed the exposure to cadmium. Thus,it would appear that cadmium excretion can be taken asa surrogate of total cadmium exposure in view of theprobable variation in intensity of cadmium exposureover the years, which would cause some misclassificationin using duration of exposure as an indicator of totaldose. Workers in the high-excretion group complainedof coughing, sputum, dryness of nose, and rhinitis andwere characterized by increased urinary calcium and j82-microglobulin; however, decreased levels of creatinineclearance and percent of tubular reabsorption of phos-phate, calcium, and /32-microglobulin were observed.

In all welders, the following relationships were signifi-cant: urinary cadmium and phosphate excretions [corre-lation coefficient (r) r = 0.4689, p < 0.05]; urinarycadmium and calcium excretions (r = 0.6844,/? < 0.01);and urinary cadmium and /32-microglobulin (r = 0.5083,p < 0.05).

Therefore, this study indicates that exposure of weldersto cadmium oxide fumes can be associated with kidneydamage because chronic renal tubular damage occurredin all five of the welders placed in the high-excretiongroup, with one showing considerable damage.

2.5 Biological Monitoring. There has been a continuingsearch for methods to establish more exact dose-responserelationships between welding fume exposure and healtheffects. A number of professional years in welding hastypically been used as a rough indicator of exposure inthe absence of other data. This section reviews the recentefforts of various investigators to identify reliable bio-logical parameters of exposure to chromium, nickel,cadmium, fluoride, aluminum, and manganese.

The measurement of chromium levels in the urine ofwelders has been extensively investigated as a possibilityfor monitoring occupational exposure to chromium(Sjogren et al., 1983a). However, lack of informationon the kinetics of chromium excretion from the bodyhas hampered efforts to make estimates of exposure.Welinder et al. (1983) measured chromium and personalair chromium levels in three groups of stainless steel arcwelders. By measuring urinary levels in 9 welders whohad been retired for an average of four years, 14 weldersboth before and after a 30-day vacation, and 20 active

22

welders, a two-compartment model for elimination ofchromium was demonstrated. The fast-eliminationcompartment consisted of a chromium half-time of sev-eral hours (range 4 to 35 hours), significantly correlatingchromium in total personal air to urinary chromiumlevels (p < 0.01). The slow compartment half-time wasmore approximate, ranging from 14 days to infinity.Retired workers excreted significantly higher levels ofchromium than nonexposed controls (7 and 1.5 yumol/mol creatinine respectively,/? < 0.001) but did not differfrom the welders who had been on vacation for 30 days(9 /xmol/mol creatinine). Active welders and weldersbefore vacation had urine levels of 34 and 45 /umol/ molcreatinine, respectively. Chromium levels in activewelders decreased from 34 to 19 /xmol/mol creatinineafter a two-day weekend (p < 0.005, one-tailed), andfrom 45 to 33 jumol/mol creatinine from postshiftMonday to preshift Tuesday (p < 0.01, one-tailed). Thispattern indicates a rapid initial decrease in the urinarychromium level after the end of exposure, then a slowerchange much later. There was good correlation betweenair and urine concentrations at the higher air concen-trations, but the variation was too extreme at low airconcentrations to make urinary chromium a good pre-dictor of occupational exposure in these welders.

Sjogren et al. (1983a) measured personal air andpostshift urine levels of chromium in 53 stainless steel arcwelders on the same working day to determine to whatextent postshift urine chromium measurements could beused to predict current chromium concentrations in theair. A linear relationship between concentration of totalchromium (including hexavalent chromium) and meanpostshift urinary level among welders was observed(r = 0.72, p < 0.0001). The time-weighted air chromiumconcentration was 124 Mg/m3> and the mean postshiftchromium concentration in urine was 718 rjmol/1. Therewas no relationship between exposure time, expressed asworking years, and urine concentration, and there wasno difference between smokers and nonsmokers. Sincenot only current exposure but also previous exposurecontributes to urinary chromium concentrations, it isnot sufficient to make a single urinary measurement toestimate chromium exposure.

Mutti et al. (1984) attempted to relate time-weightedworkroom air exposure to various forms of chromium(such as hexavalent, trivalent, and zinc and lead chro-mium compounds) with urine levels in 36 MMA/SSwelders and 12 MIG/SS welders. An increase in thechromium concentration in urine over the working daywas related only to hexavalent, water-soluble chromiumin air, consistent with the hypothesis that only water-soluble hexavalent chromium compounds are readilyabsorbed by inhalation. Neither trivalent chromium norlead or zinc chromates were observed to increase during

the work day, as determined from urine analysis. There-fore, urine analysis was recommended only for the esti-mation of hexavalent chromium.

Rahkonen et al. (1983) examined chromium andnickel levels in 10 healthy MMA/SS welders with anaverage of 13 working years with stainless steel welding.During one workweek, personal air was measuredthroughout the workday; blood, plasma, and urinesamples were taken pre- and post-shift; and spot urinesamples were taken during a follow-up period. Magne-topneumography was conducted to measure retentionrate of magnetic dust in the lungs. As in the study ofMutti et al. (1984), a linear relationship (p < 0.05)between hexavalent chromium in the air and chromiumin urine was observed. Correlations were not observedbetween chromium concentrations in whole blood andplasma and exposure to hexavalent chromium; the bestcorrelations were found between the daily mean increasein the chromium in whole blood and the total chromiumor hexavalent chromium in air (p < 0.001). Using asingle exponential functional fit, biological half-time ofchromium was estimated at 2.0 ± 0.5 days in plasma.Correlations were observed between chromium andhexavalent chromium in personal air samples and themean daily increase of chromium in blood (p = 0.001)and the retention rate of magnetic dust in lungs {p <0.001). Large variations were observed in the preshifturine and blood samples (0.1-2.7 /umol/1 and 0.05 to1.43 /zmol/1, respectively), possibly indicating differentbody burdens of chromium in the welders.

In summary, further investigation is necessary todetermine whether urinary chromium levels are a relia-ble index of exposure, because of the variability ofindividual factors, such as pulmonary ventilation andchromium body burden, environmental factors, andphysical properties, including the portion of chromiumsoluble in water.

In the same study, the concentration of nickel in theurine increased 11% during the work day, which corre-lated with the concentrations of nickel in the air (p <0.01); the mean nickel concentrations, 11.5 jug/g creati-nine (range, 7.8 to 26.5 Mg/g creatinine), also correlatedwith the nickel concentrations in air (p < 0.01). Thenickel concentration in whole blood did not differbetween welders and controls.

Sjogren et al. (1983b) measured aluminum in both theblood and the urine of various industrial workers,including aluminum welders, to determine whetheraluminum was absorbed from the occupational envi-ronment. A linear relationship was observed betweenyears of welding exposure and urinary aluminum con-centration when adjusted for creatinine (r = 0.64); thisrelationship was not explained by concentrations ofaluminum in the drinking water or ingestion of drugs

23

containing aluminum. Blood aluminum levels showedno such relationship. There was a linear relationshipbetween blood and urine aluminum concentrations,which was due largely to some individuals with extremelyhigh values of both.

Jarvisalo et al. (1983) measured manganese levels inthe urine and personal air of five nonsmoking, healthyshipyard MMA/ MS welders who had five to nine yearsof work experience. The core rod they currently used forwelding contained less than 1% manganese. Total dustconcentrations were measured gravimetrically, and filterswere placed within the face masks of welders to measuretwo 4-hour samples — one of dust and one of air. Spoturine samples and venous blood specimens were col-lected six times during the week. The study revealed widevariations in airborne concentrations of manganese inthe personal air of the workers, perhaps reflecting differ-ent working habits. These values ranged from 8.9 to56.3 mg/m3 of dust for the worker with the greatestexposure; manganese varied from 0.30 to 2.30 mg/m3.The urinary concentration varied from 11.7 to 35.6rjmol/1; the highest concentrations were observed innighttime samples. Good correlations (r = 0.89 and 0.63)between the time-weighted average of manganese levelsin air and urinary levels were observed in only twoworkers; otherwise, there was no systematic correlation.Urinary concentrations of manganese in most of thewelders were significantly elevated over levels in nonex-posed controls (9.7 ± 11.7 r}mol/\,p < 0.005). The rangeof concentrations of manganese in the blood did not varysignificantly; however, blood concentrations in welderswere elevated significantly over blood concentrations inthe nonexposed control group (25 to 32 »7mol/l com-pared with 14 + 2 T/mol/1, p < 0.05). Because of thesmall numbers of samples taken during the study, it wasnot possible to determine the suitability of blood sam-pling for estimating manganese exposure.

Bergert et al. (1982) measured manganese concentra-tions in urine, feces, and hair of welders and the nonex-posed controls. Manganese concentrations in urine were271.75 + 84.98 r)mo\/1 in welders, compared with 163.07± 54.42 r/mol/l in controls. In the feces, levels of 594.23± 239.15 /umol/g were found in welders, compared with506.23 ± 177.63 ^mol/g in controls. In hair samples,341.90 + 207.16 /umol/g in welders were found, com-pared with 58.42 ± 39.49 /rniol/g in controls.

The possibility of monitoring individual manganeseexposures from concentrations in biological fluids iscomplicated by the wide interpersonal variations inobserved concentration due to differences in workinghabits, and the fact that sources of manganese other thanoccupational sources may be important.

Baddely et al. (1983) used a neutron activation tech-nique to measure cadmium levels in 130 potentially

exposed workers, including smelters, electroplaters, andwelders. Other industrial workers served as controls.This method was used because blood and urine concen-trations generally have not reflected total body burdenfrom chronic low-level exposure because of the kineticsof cadmium excretion; in this investigation, blood andurine concentrations also correlated poorly with neutronactivation analysis. In the exposed worker population,36 out of 130 had liver tissue levels above 20 ppm and 11had liver tissue levels above 30 ppm. The investigatorsdid not consider these levels to indicate a serious healthhazard in those industries since only 2 workers out of 130had levels greater than 50 ppm, the level associated withchronic cadmium nephropathy. This study demonstratedthe utility and convenience of neutron activation analysisfor industrial screening and confirmed the lack of utilityof blood and urine sampling for monitoring cadmiumexposure.

To monitor exposure of welding workers to fluorides,Sjogren et al. (1984) measured fluorides in the personalair, and the postshift urine of two groups of weldersduring four successive workdays. One group consisted of11 stainless steel arc welders with 8 to 42 years workexperience, and the other group consisted of 64 railwaytrack welders who had worked with the OK 74.78 elec-trode, the fume of which consists of 18% to 20% fluo-rides, for 5 to 41 years. Welders had higher postshifturinary concentrations of fluoride than the controls (58/umol/1, range 10 to221 ,umol/l, compared with 26nmolj1, range < 10 to 137 /umol/l,p < 0.001, chi-square test).

In both groups of welders, a linear relationship wasobserved between postshift urinary fluoride concentra-tion and total welding fume concentration (calculated asa time-weighted average) and airborne concentration ofparticulate fluoride (p < 0.001). Postshift urinary fluo-ride concentrations remained similar during the courseof the workweek, and there was no relationship betweenage and urinary fluoride excretion, suggesting that therewas no tendency for bioaccumulation. From these data,it was considered possible to use urinary analysis to esti-mate exposure to particulate fluoride in welding fumes; avalidity test for this prediction could not be performedbecause the number of test subjects was too small. How-ever, the regression line indicated, with a 95% confidencelimit, that a mean particle air concentration of 5 mg/m3,the occupational exposure limit for nonspecific respir-able particles in Sweden and the United States, corre-sponds to a urinary fluorides concentration of 80/umol/1. It is of interest to note that the levels of urinaryfluoride observed in this study were lower than the levelsreported to cause adverse health effects (370 /xmol/1).

2.6 General Morbidity. Jinadu (1982) reported the re-sults of a survey, in four towns in Nigeria, of three

24

occupational groups, one of which was welders, con-cerned with the repairs of motor vehicles. Because theconditions of this work in Nigeria appear to bear noresemblance — in terms of worker conditions and safe-guards — to conditions in the developed world, thispaper will not be reviewed in detail. In addition, the onlycomparison data reported by the author was for twoother groups of workers in the industry — mechanicsand painters — and it is not clear how comparable andappropriate these control groups are.

The results of a cohort morbidity study of 46 malewelders by Kalic-Filipovic (1985) was presented to asymposium, but only the abstract of the presentation isavailable for review in this report. The sampling framefor the controls is chemical industry workers, but themethod of selecting of the controls is not stated in theabstract. One potential cause of concern with selection ofcontrols from one industry is that the results could reflectunusual risks in the control industry rather than in theindustry of interest. For example, if welders were shownto have an increased risk of gastrointestinal diseases incomparison with the chemical industry workers, theseresults might reflect a very low risk of gastrointestinaldisease among chemical industry workers and a risksimilar to — or even lower than — that of the generalpopulation among the welders, or it could represent anincreased risk of gastrointestinal disease among thewelders. In the absence of any other data, there would beno way to distinguish between these possibilities.

The abstract reports brief results for broad categoriesof disease. A few results are presented in the analysis.The wording in the abstract is somewhat ambiguous,but, we assume that the percents given in the analysis arepercents of welders affected rather than percents of totaldiseases. On that assumption, welders were over 15 timesas likely as the controls to have skin diseases, nearly3 times as likely to have diseases of the bones and joints,and about 2/3 more likely to have gastrointestinaldiseases, particularly peptic ulcer. However, they were30% less likely to develop chronic diseases of the respira-tory system. Because of the need for brevity in theabstract, the author does not discuss the basis for thediagnoses. If the diagnoses were made, as seems likely,by different physicians in the two industries, these differ-ences could easily reflect differences in diagnostic criteriarather than real differences in risk. However, if one ormore physicians made the diagnoses, both for thewelders and for the chemical workers, and used objectivediagnostic criteria, these results would be much morevaluable in determining whether welders are at increasedrisk.

Clearly, the value of this paper can be determined onlyupon publication of the full presentation. However, theinformation in the abstract suggests that this study is

unlikely to be helpful in determining whether there arehealth hazards from welding.

2.7 Cancer of the Nasopharynx. Lam and Tan (1984)studied this condition in Hong Kong, where it is rela-tively common. They carried out a proportional mortal-ity analysis based on validated known deaths from 1976to 1981 classified by occupation as given on the deathcertificate. The authors recognized the limits of this typeof study, including limitations in proportional mortalityanalyses (discussed elsewhere in this report) and inaccur-acies in occupation recorded on the death certificates. Inthe results that they reported, welders and plumbers wereanalyzed together. Since the exposures of plumbers arequite different from those of welders, and since it appearssafe to assume that there are considerably more plumbersthan welders in Hong Kong, the reported results havelittle application to the risk of welders. The authorsrightly point out that the study can only be used togenerate hypotheses. However, in view of the rarity ofthis cancer in Western countries, it appears unlikely thatwelding exposure is implicated in the etiology of cancerof the site.

2.8 Nasal Cancer. A case control study of nasal cancerand occupational exposure, including welding, wascarried out in Finland, Sweden, and Denmark, wherethere are nationwide cancer registries (Hernberg etal., 1983a, 1983b). (However, in Denmark alone, theNational Cancer Registry was not used as a source ofcases and controls; instead, for practical reasons, casesand controls were ascertained from four of the five exist-ing oncology centers in the country.) The cases studiedwere all new patients with primary malignant tumors ofthe nasal cavity and paranasal sinus diagnosed andreported to the sources in these three countries betweenJuly 1, 1977, and December 31, 1980. Each case wasmatched to an individual control for nationality, sex,and age at diagnosis (within 3 years) the controls beingliving patients with malignant tumors of the colon orrectum ascertained through the same National CancerRegistries during the same period of time. The firstpatient who met the matching criteria was alwaysselected as the control, but at the same time a secondcontrol was selected according to the same criteria to beused in case the first control was dead or otherwiseunavailable. The authors do not report the proportion ofthe time that a second control had to be used, but domention that 6 cases had to be excluded from the analysisbecause controls could not be selected through thismethod.

Of the 287 cases, 167 were included in the analysis.In addition to the 6 who could not be matched witha control within the limits of the methodology, 45were dead and an additional 69 were nonrespondents,

25

according to the authors in most cases either because ofpoor health or lack of a telephone. The authors provideinformation on cases included and excluded from theanalysis, which is helpful in determining the generaliza-bility of the results. The greatest difference in inclusionand exclusion was by country; the proportion of identi-fied cases included in the analysis was approximatelytwice as high for Sweden as for Finland. This may reflectthe proportion of households having telephones or thequality of medical care in the two countries. The othermajor difference is the primary site of the tumor; indi-viduals having cancer of the maxillary sinus were theleast likely to be interviewed, perhaps because this cancerhas a poor prognosis. The proportions of cases includedin the analysis according to cell type of cancer were quitesimilar in the three countries. The mean age of cases notincluded was higher than that age of those who wereincluded, especially in Finland and least in Denmark,and more so for females than for males. This last findingmay represent true differential noncooperation.

For each subject, a detailed occupational history wasobtained for each separate occupation of more than1 year's duration except for the 10 years preceding diag-nosis in order to account for latency. An industrialhygienist blindly evaluated the exposure by category,one of which was "welding, flame cutting, and soldering."

This investigation has a number of strengths. At firstglance, a case control study rather than a cohort studymay be surprising. For most diseases, studying a cohortof welders is considerably more efficient in determiningwhether they are at increased risk of the disease ratherthan taking people with and without the disease, becausesuch a small proportion of the population are welders.However, nasal cancer is so rare that it is considerablymore efficient to study the proportion of welders in nasalcancer cases and controls than to determine the numberof observed and expected nasal cancers in a cohort ofwelders. In SRI's two largest cohort studies (ClientPrivate) combined, which represent a total study popula-tion of approximately 45,000 individuals and over650,000 person-years of observation, there were no nasalcancer deaths. Consequently, it is not surprising that inmuch smaller cohort studies of welders (for example,that of Polednak reviewed in Effects of Welding andHealth IV), no deaths from nasal cancer were noted.

In addition, a panel of pathologists reviewed all thecases to make certain that they were nasal cancers, afeature that is often absent from other such studies (fre-quently on the grounds of lack of funding or availabilityof interested pathologists). Because the cases were ascer-tained through population-based cancer registries, theuse of such registries to act as a sampling frame forcontrols is most appropriate. Finally, the thoroughreview of occupational histories by an industrial hygien-

ist, including the allowance for the elapse of a latentperiod, is likely to minimize the potential misclassifica-tion by exposure.

Therefore, it is indeed unfortunate that the authors,having designed the study so commendably, introducedsome features that made the results difficult to interpret.With a control group of all nonrespiratory cancersmatched with these cancers from the same data source,there would be no question that the finding of an oddsratio of 2.8 for "welding, flame cutting, and soldering"would mean an increased risk of nasal cancer in welders.Unfortunately, the investigators chose as their controlgroup cancers of the colon and rectum, so that thisfinding could reflect either a low risk of colorectal canceramong welders or a high risk of nasal cancer amongwelders, flame cutters, and solderers (or both). In fact,there may be some suggestion that welders are at low riskof cancer of the colon and rectum, as Polednak (1981), ina study reviewed in Effects of Welding on Health IV,found that the number of observed deaths from all diges-tive cancers (many of which are cancers of the colon andrectum) was less than half the number predicted from theexperience of the comparable general population. Inaddition, Hernberg and coworkers reported data thattook the matching into account only for welders, flame-cutters, and solderers combined, and in trying to inter-pret the risk for welders, solderers should be excluded.The authors do present a table giving some exposureinformation for subjects exposed to welding, flame cut-ting, soldering, chromium, or nickel. However, even thisincomplete information does not indicate whether any ofthese cases are matched to any of these controls; in otherwords, the matching is broken and, as is well known,efficiency is lost. From the fragmentary informationavailable, it is clear that there are 7 cases and 3 controlswith welding exposure. If the matching is ignored (as ithas to be in the absence of information on matching), anodds ratio of 2.4 is calculated. This, as already noted,could represent a lower risk than reality for colorectalcancer.

In summary, a well-conducted (in many respects) casecontrol study shows that patients with nasal cancer aremore than twice as likely to have an occupational historyof at least 1 year of welding at least 10 years before theonset of the disease than are patients with cancer of thecolon or rectum. However, it is not clear whether thisrepresents a decreased risk among welders of cancer ofthe colon or the rectum or an increased risk of cancer ofthe nose and nasal sinuses among them.

2.9 Cancer of the Larynx. Sidenius et al. (1983) pro-vided a case report of five cancers of the larynx withina five-year period among workers in a Copenhagen

26

machine factory with an average work force of 90workers. Based on the incidence rate for cancer of thelarynx in Copenhagen, the number of cancers of thelarynx was 114 times the number expected. Although it isnot clear whether the author used crude or adjusted ratesto calculate the expected number of cancers of thelarynx, age differences cannot explain such an extremefinding. As the authors note, chance is an extremelyunlikely explanation. Two of the five laryngeal cancerswere in welders. However, one had been employed foronly four years in the factory and the other for only twoyears before the cancers were diagnosed. For these twoindividuals it would appear that the interval betweenonset of employment and occurrence of the cancer is tooshort for the cancers to be caused by these occupationalexposures. In fact, of the five cases, only two had morethan a five-year interval between onset of employmentand diagnosis of cancer (one interval of 11 years and oneof 20 years).

However, the working conditions, particularly withregard to industrial hygiene in this factory, are mostunusual. The welding itself apparently has been stainlesssteel using argon gas shielding. Before this method,plated electrodes containing manganese, molybdenum,and titanium were used. There were continual improve-ments in the welding technology, but the methods usedearlier than the last named are unknown. The factory didnot install exhaust hoods in the welding area until 1974.At the time of the report, there was no fresh air intake inthe factory except what came through cracks in the doorsand windows. Heating in the factory was by the use ofseveral heat turbines, which circulated the air in theroom by sucking in the air, heating it to 70° C, and thenexpelling it. As the authors note through this process, thepolluted air is mixed, heated, and spread over the onelarge room in the factory where the welding work andgrinding (which is screened)-vice work, assembling, anddrilling operations are located. In December 1977, per-sonal and area sampling was carried out within this largeroom for chromium, copper, lead, manganese, zincoxide, and iron oxide. Although all these substanceswere found in the air everywhere in this large room, inthe welding and grinding areas the hygienic limits wereexceeded. Depending on the sample, between 43 and80% of the total dust was in substances other than whatwere measured. In contrast, oil mist measurements wereonly 30 to 40% of the allowed maximum values. In viewof the work conditions, it would appear that not only thewelders, but other workers as well, were exposed tochromium, nickel, and other substances from the weld-ing processes.

In terms of other causes of cancer of the larynx, all fivecases were cigarette smokers. Three smoked one pack aday and the other two smoked less but also smoked

cigars and took snuff. None of the five had a history ofexcessive alcohol ingestion.

The authors originally intended to carry out a casecontrol study to determine the reasons for the excess ofcancer of the larynx. They were unable to do so becauseof incompleteness of the records, both in terms of lackof availability of records on former employees and interms of obtaining a complete work history on presentemployees. In the absence of this investigation, all thatcan be said is that there is a huge excess of cancer of thelarynx in this factory, which may be related to occupa-tional exposures originating in the welding area, butother occupational and nonoccupational causes cannotbe ruled out. Three of the five cases have too short aninterval between onset of an employment and diagnosesof the cancer for an occupational cause in the factory tobe suspected, but, as the authors point out, something inthe factory could be acting as a promoter.

Because of this report, a case control study of cancer ofthe larynx in welding in the entire country of Denmarkwas carried out (Olsen et al., 1984). The cases consistedof all newly diagnosed laryngeal cancer patients inDenmark between March 1980 and March 1982 whowere under 75 years of age at the time of diagnosis. Thesecases were ascertained from hospital records of the fivehospitals involved in treatment of laryngeal cancer inDenmark. For each case, four controls were identifiedusing the municipal person registry in which the case waslisted; these controls were matched to cases by residence,sex, and closest possible birthdate. Each subject receiveda postal questionnaire about their occupational expo-sure to a number of specific chemical and physicalagents, including welding dust from stainless steel weld-ing and welding fumes in general, as well as tobacco andalcohol use and other items.

Because very few females were exposed to weldingfumes, the analysis reported by Olsen and colleagues wasconfined to males. Three mailings were sent out andmissing information was handled either by returning thequestionnaires to the subject or by interviewing the sub-ject over the telephone. However, information on weld-ing exposure was missing for 4% of the cases and 5% ofthe controls.

Of the 271 cases included in the analysis, 42 hadwelding fume exposure. Of the 971 controls included inthe analysis, 115 had welding fume exposure. The riskvaried consideraly by anatomical location within thelarynx. (Patients with cancer of the hypo- and oro-pharynx were excluded from the analysis.) There were176 cases with glottic laryngeal cancer (i.e., cancer of thetrue vocal cord); of these, 23 were exposed to weldingfumes. There were 79 supraglottic laryngeal cancers (i.e.,cancers located above the vocal cords); of these patients,13 were exposed. Five of the eleven subglottic laryngeal

27

cancers (i.e., cancers, located below the true vocal cord)were exposed. For all laryngeal cancers there was a weakassociation with welding fumes, which was of borderlinestatistical significance. (Although the authors reportedseveral different statistical analyses, the results of eachanalysis were somewhat similar except that the analysisthat retained the matching characteristics of the designshowed stronger associations in each case.) When thecases were broken down by location within the larynx,there was essentially no association with glottic cancer,an association with supraglottic cancer (which in thematched analysis was of borderline statistical signifi-cance) that was a little stronger than for all cancers of thelarynx, and a 6- to 8-fold increased risk for subglotticcancer, which was statistically significant in spite of thesmall numbers.

Among all laryngeal cancer patients, 50% of thoseexposed to welding fumes welded more than 15 hours aweek, compared to only 21% of the exposed controls.Welders among the cases also had been welding longeron the average (22 years vs an average of 18 years for thewelders in the control group). These data were notreported by anatomic location for laryngeal cancer butare compatible with a causal association.

The authors found similar findings for exposure towelding dust from stainless steel. However, the numberexposed was much smaller (30 controls and 12 cases — 8glottic, 2 supraglottic, and 2 subglottic). Therefore, sta-tistical significance was not attained although the rela-tive risk estimates were similar.

The authors were concerned about the so-called recallor memory bias (i.e., that the cases would be more likelyto remember about exposures because they were stimu-lated to think about possible causes of their disease). Inthe introduction to the questionnaire, the participantswere told that this was a study of the occupationalcausation of cancer of the larynx, but welding was notspecifically mentioned. According to the authors, mostother occupational exposures occurred with equal fre-quency among cases in controls (although they did notpresent the data) which would indicate that the positivefindings for welding were not a result of this recall ormemory bias.

The authors also examined the possibility of recallbias in another way. They believe that if that biasexisted, cases would be especially likely to remembershort-term welding exposures better than controls.Therefore, they reanalyzed the data by reclassifying allthose exposed to welding for less than five years in thenonexposure category. This somewhat weakened theassociation with supraglottic laryngeal cancer and some-what strengthened the one with glottic laryngeal cancer,and left the subglottic cancer association essentiallyunchanged.

For all cancer of the larynx patients, the increased riskwith exposure to welding dust was seen only in smokers.However, among the cases, there were only 13 non-smokers, so these data do not permit a determinationof whether exposure to welding dust in nonsmokersincreases the risk.

The authors were rightly concerned about the differ-ential nonresponse between cases and controls. Ninety-six percent of the cases participated but only seventy-eight percent of the controls did so. The refusal rateamong the controls was higher in the rural districts thanin Copenhagen and was at its highest among the elderly.As the authors noted, a high proportion of weldersamong the nonrespondents in the control group couldexplain the present finding in the absence of an increasedrisk from welding. However, their impression was thatpeople who had not been occupationally exposed to anyof the potential hazards mentioned in the questionnairewere less likely to respond; if so, the association with weld-ing exposures is underestimated. The authors noted thatmost of the nonrespondents came from the agriculturalareas of the country which reinforced this impression.

The authors specifically noted that the cases reportedin the factory by Sidenius and her colleagues were notincluded among the cases (because they had been diag-nosed earlier).

This is, in general, a very well-conducted study. Itshows a weak association between cancer of the larynx asa whole and either welding fume or stainless steel weld-ing dust (which contains the carcinogens, hexavalentchromium and nickel), between welding fume andsupraglottic cancer, and between stainless steel weldingdust and glottic cancer. These relatively weak associa-tions could be easily explained by a common underlyingfactor. For cancer of the larynx as a whole, the riskincreases with the number of years as a welder andparticularly with the number of hours per week spentwelding, which weakens the possibility of a commonunderlying factor explaining the results. In addition, thestudy provides strong evidence that welding exposurescause the relatively rare (4% of this Danish series, 1% ofall laryngeal cancers in the U.S.) subglottic laryngealcancer. Because subglottic cancers make up such a smallproportion of all laryngeal cancers, a causal associationwith subglottic cancers but no association with otherlaryngeal cancers would only result in a weak overallassociation for cancer of the larynx.

Just such a weak association was found by Sjogrenand Carstensen (1985); they found 22 cancers of thelarynx compared to 17.3 expected. Their abstract is dis-cussed in more detail elsewhere in this report.

2.10 Lung Cancer. Blotetal. (1980) carried out a casecontrol study of lung cancer in the Tidewater area of

28

Virginia. This part of the country was selected becausethe county mortality rates in the U.S. between 1950 and1969 for lung cancer were elevated in this part of Vir-ginia, an area where shipbuilding had been a majorsource of employment. The emphasis of the study wasprimarily on shipbuilding. Blind interviews were com-pleted with 89% of the lung cancer patients and 82% ofthe controls. In each case, the interview was with the nextof kin or other relatives and friends, and the proportionof interviews in cases and controls for each of these typesof surrogate was comparable, after excluding 2% of theindividuals when information was incomplete, 3% whohad lived in Tidewater less than five years, and another2% who "did not qualify as cases or controls." The finalanalysis was based on 336 patients who died from lungcancer during the year 1976 and 361 controls who weresystematically selected from the mortality listings, afterexcluding all respiratory cancer and chronic lung dis-ease, to resemble the cases with respect to race, age, yearof death, and county or city of residence.

Most of the results reported and the discussion dealtwith whether the subject had worked in a shipyard.However, the authors did report that 11 lung cancercases and 9 controls had worked as a welder or burner inshipyard trades prior to 1950. These data give a crudeodds ratio of 1.32 which is slightly lower than the oddsratio reported by the authors for shipbuilding in general.(The authors noted that comparisons may have beenaffected not only by the slightly lower response rateamong the controls, but also by publicity surroundingthe issue of asbestos and cancer and by limitation ofoccupational information obtained from next of kin, asopposed to the subjects who, had they been able to beinterviewed — would presumably be able to provide amore complete occupational history.)

This is a reasonably well-conducted study, and theauthors are well aware of the limitations of their data.However, a cohort study of the shipbuilding industrywould have been a better way to test the hypothesisbecause lung cancer is a relatively common cause ofdeath. The slightly increased risk among welders, as willbe noted, is similar to risk estimates found in cohortstudies of welders and probably does represent a mildlyincreased risk. Because the increased risk of lung canceramong all shipyard employees is secondary to this riskfrom asbestos exposure, which has been shown numer-ous times, in the absence of further information it seemsreasonable to conclude that if there is an increased risk ofdeath from lung cancer among these welders, it is proba-bly a result of the asbestos exposure rather than weldingfume exposure in the shipyard.

In a 1981 review, Stern stated that he combined theresults of four cohort studies and one case control studyof lung cancer in welders, but he did not identify the

studies that were combined. According to Stern, thesefive studies have an aggregate study population of150,000 welders with approximately 420,000 person-years of observation. There were 415 observed deathsfrom lung cancer, and 308 expected to give an SMR of135. Without knowing which studies Stern combined,the appropriateness of this calculation cannot be judged.As a minimum, it seems dubious to combine a casecontrol study with cohort studies in this manner becauseof the way the data are ascertained.

Several studies that have been discussed extensivelyelsewhere in this report with regard to their strengths andweaknesses have reported results for lung cancer as well.Sjogren and Carstensen (1985) found 193 lung cancersreported to the Cancer Registry for a Standardized Inci-dence Ratio (SMR) of 1.42, and Sjogren et al. (1985)reported an SMR of 134 for pulmonary tumors when thesame sample of welders was followed until death.Gallagher and Threlfall (1983) reported a Proportional-ized Mortality Ratio (PMR) for lung cancer of 145 basedon 74 deaths. Coggon et al. (1984) found a relative risk oflung cancer of 1.3 for individuals with an occupationalhistory of solder exposure and 1.0 for individuals with"high exposure" to solder. (As noted elsewhere, it is notclear how much these last results relate to welding expo-sure.) In their abstract, Santi et al. (1985) state that theyfound a slight, nonsignificant increase in lung cancer riskamong welders, but they gave no further data. McMillanand Pethybridge (1983) found that the number ofobserved deaths from lung cancer among welders wasexactly equal to the expected number. In an abstract,McMillan (1985) mentioned that in the dockyardmortality study there were five deaths from bronchialcarcinoma, but he did not give the expected number.

Newhouse et al. (1985) presented a paper on mild steelwelders in a shipyard. Only the abstract of their presen-tation is available for this review. The records were ofmen employed between 1940 to 1968 and included 1027welders, analyzed separately. Over 97% of the individu-als were successfully traced for vital status, an excellentresult. Newhouse and her colleagues reported an SMRfor lung cancer of 114 among the welders. No furtherdata are available, but the authors state that theyintended this investigation as a "a useful pilot study" andare likely to carry out more comprehensive studies ofstainless steel workers.

Beaumont and Weiss (1985) also presented at thissymposium; only the abstract is available for review atpresent. Virtually all the results reported by the authorsappear to be reported in their earlier papers reviewed inEffects of Welding on Health IV. However, they didreport that if welding exposure can be considered causalfor lung cancer, then among 100,000 welders there were23.1 lung cancer deaths occurring as a result of welding

29

each year. In their abstract, they stated that the increasedrisk of death from lung cancer increases with increasinglength of employment and interval since initial employ-ment, which was the strongest predictor of excess risk.They also mentioned that the risk increased with increas-ing age, which is not surprising and not particularlyhelpful, because such a finding would be expected tooccur whether or not there was an occupationally relatedrisk of lung cancer. The increase with increasing dura-tion of employment is compatible with an occupation-ally related cause. This paper cannot be completely eval-uated until the proceedings of the entire synmposium areavailable.

Sjogren et al. (1982) wrote a letter to the editor inresponse to the earlier paper of Beaumont and Weiss onlung cancer among welders that was reviewed in Effectsof Welding on Health IV. In their letter, they presentedsome data from their survey of cancer among weldersand gas cutters from the Swedish census data, with 96lung cancers observed and 66.64 expected, a findingwhich is highly statistically significant. They suggested,therefore, that the findings of Beaumont and Weiss forlung cancer in welders could be related to asbestos expo-sure. In their letter, they reported data from a 1980 studyby Sjogren also reviewed in Effects of Welding on HealthIV. In that study, Sjogren found 3 deaths from lungcancer with 0.68 expected, which, in spite of the lownumber of deaths, was statistically significant. Sjogren etal., therefore suggested that hexavalent chromium fromstainless steel welding might also play a role in the resultsreported by Beaumont and Weiss.

Beaumont and Weiss (1982) responded to the letter bysaying that they did not think asbestos and chromiumwere major factors in their study. For example, theyfound no mesothelioma deaths. [As noted in our sectionon mesothelioma, Sjogren et al. (1982) reported fourmesothelioma deaths, which was more than three timesthe expected number.] Beaumont and Weiss also pointedout that "stainless steel welding constituted only a smallpercentage of the total welding performed (mild steelwelding was predominant)." They also pointed out thatStern had proposed (in yet another paper reviewed inEffects of Welding on Health IV) that a very high rela-tive risk for stainless steel welding might explain theoverall excess lung cancer risk of about 40% in thewelding literature. (As can be seen from the data that wehave reported above, risks this high or somewhat lowerhave been found recently.) Beaumont and Weiss give astheir opinion, however, that further study of welders notexposed to chromium or asbestos is needed before it canbe inferred that workers not exposed to chromium orasbestos are free of excess lung cancer risk. They recom-mended that populations of mild steel welders notexposed to asbestos be identified and studied.

Waddington et al. (1983) described a number ofinvestigations launched by the WHO Regional Officein Europe. These included a study of health effects ofoccupational exposure of welders to chromium andnickel. One component of this study is a mortality studyof welders exposed to chromium-containing fumes,welders not so exposed (who will be used apparently asan internal control) and external general populationcontrols. If these plans achieve fruition, papers and otherreports of these investigations will be very importantportions of future editions of the Effects of Welding onHealth.

In summary, a number of studies have reported eitherno increased risk of lung cancer or a slight to moderateincrease of less than 50% above expectation in welders.The data are consistent with an increased risk second-ary to asbestos exposure; however, until some of themore specific studies mentioned above are actuallycarried out and published, it will not be clear how seriousthe problem is and whether, to the extent it exists, theincreased risk can be entirely explained by asbestosexposure.

As is well known, cigarette smoking is a cause of anumber of diseases, especially respiratory tract diseases.In cohort studies, where welders are compared with thegeneral population for risk of diseases, information oncigarette smoking is nearly always not available to theinvestigators. When an increased risk of these diseasesis found in such studies, this increased risk sometimeshas been facilely attributed to cigarette smoking in asuperficial analogy to the attribution of excess risk ofmesothelioma in shipyard welders who were exposed toasbestos.

This analogy is not appropriate, because shipyardwelders are all exposed to asbestos as a consequenceof working in a shipyard, whereas not all welders smoke.Because some welders are nonsmokers, indirect exam-ination of whether cigarette smoking can explain anincreased risk of these diseases is possible, even inthe absence of smoking information on individualwelders.

As the magnitude of the SMR observed increases, it isless likely that cigarette smoking will be the underlyingexplanation for the increase. In evaluating the effects ofwelding on health, an even more important considera-tion is whether the SMR increased with increasing inten-sity or duration of welding exposure. If such a dose-response relationship is seen, the possibility that cigarettesmoking may explain the association becomes quiteremote because in order to explain the finding, cigarettesmoking would have to increase with increasing intensityor duration of welding exposure. Because of their impli-cations we have emphasized the magnitude of the SMRand particularly its relationship to the intensity and

30

duration of welding exposure in our presentation of theresults of cohort studies of welders.

2.11 Mesothelioma. Barnes (1983) discussed all com-pensable, asbestos-related diseases occurring in NewSouth Wales, Australia, between February 1968, whenasbestosis became recognized by the Board in NewSouth Wales as compensable, and December 31, 1983.There had been 197 compensable deaths among workersduring this period — five in welders, all from mesotheli-oma. It is difficult to tell the number of deaths expectedfrom mesothelioma in the absence of excess risk becauseBarnes does not give the total number of welders in NewSouth Wales during this period and because the scientificliterature contains little information about mortalityrates for mesothelioma. However, it would appear thatfive deaths in welders is in excess of expectation, and thatthese welders therefore were at increased risk.

It seems reasonable to implicate asbestos exposure asthe most likely cause of this probable increased risk ofmesothelioma among welders in New South Wales.According to Barnes, during World War II asbestos waswrapped around welding rods to make them burn moreslowly. Barnes examined several such rods at dock yardsand found they were wrapped in crocidolite threads.Experimental use of one of these rods found air levels"many times greater" than 2 milligrams per cubic meterof air for chrysotile asbestos and over 1 milligram percubic meter of air for crocidolite asbestos. Asbestos is theonly known cause of mesothelioma, and the only otherhypothesized cause — a mineral named zeolite — hasonly been reported to be associated with mesotheliomain Turkey.

McMillan (1983) also has described case reports ofmesothelioma among welders. According to McMillan,Dr. Geoffrey Sheers has maintained a registry of all casesof mesothelioma occurring in Plymouth, England, whichis the location of a very large British dockyard in Devon-port. Also according to McMillan, the completeness ofcancer registration has been "better than average,"although it has been estimated that 1 death in 13 is notdetected because of men moving from the area. Between1964 and 1978, 108 diagnoses of mesothelioma weremade and confirmed histologically. Not surprisingly, 96were among men who worked in the dockyard, and 3 ofthese decedents were welders. According to McMillan,between 1978 and 1981, four other welders in Plymouthare known to have died of mesothelioma. Although forreasons given above, expected numbers are not given,these seven deaths from mesothelioma among weldersduring this time period appear to be considerably inexcess. Because shipyards are known to be places wherethere are high levels of occupational exposure to asbes-tos, once again it appears reasonable to ascribe this

probable excess of mesothelioma among welders toasbestos exposures in the shipyard.

McMillan and Pethybridge (1983) reported a propor-tional mortality analysis among welders and otherworkers in the same dockyard. Included in the analysiswere all individuals who had been employed for at leastsix months and had retired or died while still employedat the dockyard in Devonport between January 1, 1955and December 31, 1974. All these men were traced todetermine which had died by December 31, 1975. Of the2,568 men, 656 were deceased.

The authors stated that it was not possible to calculatestandardized mortality ratios because the total popula-tion at risk was not assessed, presumably because thefunding did not permit this to be done. Therefore, theyhad to settle for calculating proportional mortality ratios(i.e., the observed number of deaths divided by anexpected number derived from the proportion of deathsfrom that cause in the comparison population). Thereare limitations in interpreting proportional mortalityratios when the expected numbers are derived from thegeneral population proportional mortality, so perhapsfor this reason, the authors used two other groups ofdeceased workers as control groups. These were boiler-makers and shipwrights who had moderate, intermittentexposure to welding fumes and gases, and electricalfitters, painters, and joiners who had little or no exposureto welding fumes and gases. This, unfortunately, onlymakes it even more difficult to interpret the results. Eventhe finding of a gradient of an increasing proportionalmortality ratio for a cause by degree of welding exposuredoes not necessarily mean that welders are at increasedrisk of death from this cause. Such a finding could beexplained through several other mechanisms in theabsence of an increased risk of death from this cause inwelders. Among such mechanisms are a decreased risk ofdeath from this cause among boilermakers and ship-wrights and an even lower risk among the electricalfitters, painters, and joiners and, as a second possibility,an increased risk from all other causes among boiler-makers and shipwrights, and an even more increased riskfrom all other causes among electrical fitters, painters,and joiners.

For mesothelioma, the PMR was 259 for welders, 116for the boilermakers, and 24 for the electrical fitters andpainters. This certainly describes a gradient of increasingPMR with increasing exposure to welding fume. In fact,as noted above, these data in themselves do not necessar-ily indicate that welders are at increased risk of mesothe-lioma; however, because it is well known that shipyardworkers are at increased risk of mesothelioma, it is clearthat even in comparison with other shipyard workers,welders are at an increased risk of mesothelioma. Itappears unlikely that this gradient is caused by welding

31

fume exposure. In fact, the authors correctly ascribe thisincreased risk of pleural mesothelioma to asbestos expo-sure. McMillan and Pethybridge (1983) do note thatthere were no deaths from sinonasal tumors, whose riskis increased from certain exposures to nickel.

In summary, all of these papers show an increased riskof mesothelioma among welders in shipyards, and —justas clearly — in each case, asbestos exposure is respon-sible for this increased risk.

In the linkage of the 1960 Swedish Census with theSwedish Cancer Registry, Sjogren et al. (1982), in aletter to the editor, noted that there were 4 cases ofmesothelioma in the welders and gas cutters identified inthe 1960 National Census. According to these authors,the expected values, which are sex-, age-, and region ofSweden-specific, are 1.17 for mesothelioma, giving astandardized incidence ratio of 3.43, which was statis-tically significant. Sjogren and Carstensen (1985) pre-sented linked data from the same sources at a 1985meeting. According to the abstract of that presentation(all that is available for this update), there were still 4plural mesothelioma cases, but the expected number hadrisen to 2.7, lowering the "risk ratio" (actually, as notedabove, the standardized incidence ratio) to 1.49, a valuethat was no longer statistically significant. Presumably,the higher expected number represents later data, al-though until the presentation is actually published wewill not be able to determine this. Evaluation of thesedata will depend on publication of the full presentationin order to determine the basis for the expected numberof mesotheliomas. Until then, it is not clear what the riskof mesothelioma is among Swedes who identified them-selves as welders during the 1960 census.

Baba (1983) reported the first case of mesothelioma ina welder in Japan. Because this man also had a diagnosisof asbestosis, his death in 1979 was presumably the resultof asbestos exposure.

2.12 Bladder Cancer. There have been two recentreports of increased risk of bladder cancer among indi-viduals occupationally exposed to solder in England andWales (Coggon et al., 1984) and among solderers andwelders combined in the United States (Miller et al.,1984). Neither of these papers will be discussed in detailbecause the exposed groups included both welders andnonwelders and no separate analysis was done onwelders alone.

Preliminary results of a retrospective cohort mortalitystudy of 2,190 longshoremen, including 214 weldersin the Genoa dockyards between January 1, 1960 andDecember 31, 1979, have been reported (Santi et al.,1985). According to the abstract of the presentation,which is all that is available in time for this review, thesepreliminary results indicated an increased, statistically

significant risk of bladder cancer. In the publishedabstract, the authors did not report results separately forthe welders but stated that in their presentation, theyprovided an analysis of risk of lung cancer amongwelders by length of exposure. Evaluation of this studyand the results must await publication of the full presen-tation, probably a few months after the completion ofthis update.

2.13 Kidney Cancer. Sjogren and Carstensen (1985)have recently presented results of a cohort study at ameeting. For this review, only the abstract of their pre-sentation is available. These authors took advantage ofthe fact that during the 1960 Swedish census, individualswere asked their occupation. This occupational informa-tion was linked with newly diagnosed cancers in thenationwide Cancer Registry. In addition, the authorsstate that they took deaths into account, although theabstract does not state how this was done. One correctway to do this would be to accumulate person-years forcalculating expected numbers of cancers for individualswho died from non-neoplastic causes up until the date ofdeath and then to cease accumulating these person-yearsfor these individuals. The expected number of cancersare adjusted for sex, age, and region of the country.

The authors' presentation is confined to the 23,464males 20 to 69 years old at the time of the 1960 census(October 2-8) and gave their occupation during thatweek as welders or gas cutters. Females were excludedbecause of small numbers. There were 70 newly diag-nosed kidney cancers and 54.0 expected among thesemen. This 30% excess is statistically significant {p = 0.04)because of the large numbers. In the abstracts, theauthors briefly discuss the problems with misclassifica-tion, which are discussed in the review of this paperunder Brain Cancer. They also pointed out that theremay be subsets of the welders with much higher risk, buttypes of exposure within welding were not characterizedin the census data.

Because the association is weak, it could easily beexplained by a common underlying factor such ascigarette smoking or by biased misclassification if it wereof sufficient magnitude. Consequently, all we can say isthat the Swedish data, at least as reported in the abstract,suggest that welders are at an increased risk of kidneycancer. This finding needs to be confirmed or refuted byother independent studies.

2.14 Brain Cancer. Englundetal. (1982) have reportedon brain cancer by occupation in Sweden. During theSwedish national census, which took place October 2-8,1960, individuals stated their occupation. One of thecategories was welders and gas cutters combined.Because each Swede has a unique nine-digit personal

32

identification number used on all records, investigatorshave been able to link occupational information recordedin the 1960 census with the occurrence of cancer in theNational Cancer Registry between 1961 and 1973. Afterelimination from the data of children born after 1960 andimmigrants to Sweden after 1960, 98.8% of the cancersreported from 1961 to 1973 have been linked to the 1960census reports. According to the authors, unmatchedrecords resulted almost entirely because of incompleteidentification numbers.

The authors calculated age- and sex-adjusted expectednumbers of brain cancers for different occupationalgroups, including welders. This enabled them to calcu-late "risk ratios" (in actuality, standardized incidenceratios) by dividing the observed numbers of braincancers by expected numbers thus calculated. Therewere 50 brain cancers reported in the 25,935 welders,which gave an excess of observed-to-expected of approx-imately 35%. The authors reported 99% confidenceintervals, which shows that the difference betweenobserved and expected was not significant at the 1%level. Our calculations at the more conventional 5% levelof significance show borderline statistical evidence.

In a recent abstract of a presentation at a meeting,Sjogren and Carstensen (1985) reported observed andexpected brain cancers among the 23,464 males aged 20to 69 during the 1960 Swedish census. The abstract didnot give specific follow-up dates, but it appears that thisfollow-up is more recent because there were 68 observedbrain cancers among this subset of all welders with 56.9expected for a "risk ratio" of 1.20 which is no longerstatistically significant at the 5% level. (Although wehave conventionally referred to these cancers as braincancers, they actually are malignant neoplasms of allparts of the central nervous system, virtually all of whichare malignant neoplasms of the brain.)

As the authors point out, "certain misclassifications ofboth diagnoses and occupation are inevitable and haveto be borne in mind when one interprets the results."Some of the welders on that date may have spent most oftheir working lives at different occupations, and some ofthe other individuals may have worked as welders butwere not doing so during that census week. If this mis-classification was completely random, it would notcreate a spuriously positive result but would tend toweaken any association found. If the misclassificationwere nonrandom, anything could happen, even falsepositive results. However, if the misclassification wasminimal, even such biased nonrandom misclassificationwould not affect the results to a large extent. It seemslikely that most welders were classed appropriately, andthat the number of former welders among the rest of theSwedish population included in the analysis was toosmall a fraction to affect the results.

More serious is the probelm regarding the accuracy ofthe diagnosis. The authors did not attempt to confirmany of the diagnoses. Brain cancers are sometimesundiagnosed or misdiagnosed. In the United States,there is some evidence (Greenwald et al , 1981) thatemployed populations in certain industries with excel-lent benefits in health insurance and fine employee medi-cal services may be more likely to have their braincancers detected. Greenwald and his colleagues call thiseffect "diagnostic sensitivity bias." Because of the univer-sal availability of social insurance in Sweden, the Swed-ish authors point out there is no reason to believe thatdiagnostic sensitivity bias would be stronger per se insome industries than in others. However, they do showthat in certain areas of Sweden where major cities arelocated, and particularly in Malmo and vicinity the "riskratio" for brain cancer is extremely high — perhapsbecause of diagnostic sensitivity bias. If welders are morelikely to be found in the Malmo area than in the rest ofSweden, the geographic differences could expain thefindings.

Englund et al. (1982) performed an additional analysison those individuals born between 1896 and 1940 andhence, would have been between 20 and 65 at the time ofthe 1960 census. They calculated incidence rates (the typewas not specified, but it appears that these are annualincidence rates per 100,000, which are age- and sex-adjusted) for the entire Swedish population and forspecific occupational groups. For the entire Swedishpopulation, the incidence rate for all brain tumors was12, and for glioblastomas, which are relatively commonand the most serious prognostically, the rate was 5. Forwelders, the corresponding rates were 15 and 9. Thesefigures indicate that if there is an increased risk of braincancer in welders, the risk of glioblastomas is evenhigher.

Because of the uncertainties in the study describedabove and because such a relatively weak association caneasily be explained by common underlying factors, atpresent, one can only say that there is a suggestion fromsome Swedish data that welders may be at increased riskof brain cancer, particularly gliolastoma. However, thisfinding needs to be confirmed before the association canbe accepted as causal.

A case control study for brain cancer in Maryland(Lin et al., 1985) reported an odds ratio of 1.95 (in otherwords, nearly a twofold risk) of the specific brain cancersglioblastomas, other gliomas, and astrocytomas com-bined for a category that included welders. This cate-gory, which included electricians, dispatchers, highwaypatrolmen, electric, electronic, aerospace, and telecom-munication engineers, television and radio repairmen,electronic service repairmen, and telecommunicationindustries repairmen was set up by the authors because

33

they felt that individuals in these occupations were prob-ably exposed to electromagnetic fields. Results forwelders were not provided separately, so it is not clearhow much the results reported by Lin et al. (1985) applyto welders.

2.15 Neuroblastoma. This is a childhood tumorfound primarily in the adrenal gland, although it canoccur elsewhere. Recently, there has been interest inpossible parental occupational causes of this cancer inthe offspring. A case control study for this disease hasrecently been reported in an abstract by Kramer et al.(1984). They compared 104 cases diagnosed between1970 and 1979 in a well-defined region known as theGreater Delaware Valley with 104 controls individuallymatched on year of birth (plus or minus three years),race, and first eight digits of the telephone exchange ofthe case of diagnosis. They reported statistically signifi-cant results, which indicate how likely the findings are tobe caused by chance but do not present the strength ofthe association, which is more valuable as an indicator ofhow likely a confounding variable can explain the find-ings. Maternal occupational exposure to metal fumesand dust any time during the five years prior to the birthof the child was significantly (j> = 0.01) associated withrisk of neuroblastoma. Paternal exposure to metal fumesand dust during the five years prior to birth was alsosignificantly associated with the risk of neuroblastoma(p = 0.05) but less so than the previously reported mater-nal exposure finding. Spitz and Johnson (1985) havenoted that the exposure to metal fumes suggests theoccupation of welders. However, they feel that thewelders are exposed to electromagnetic radiation, whichthey feel is the important risk factor for neuroblastoma.In any event, the findings of Kramer and her colleaguesimply that a large number of tests of significance weremade, which means that even if there were no relation-ship, 1 in 20 would be conventionally statistically sig-nificant by chance alone. Therefore, the statisticalsignificance of the exposure is less impressive than itmight otherwise seem. Because not enough details of thestudy by Kramer et al. (1984) are given in the abstract, acomplete evaluation of the relevance of that study to theeffects of welding on health must necesarily await publi-cation or provision of the full paper.

2.16 Hodgkin's Disease. Gallagher and Threlfall( 1983)reported on a proportional mortality analysis on deathsin the province of British Columbia between 1950 and1978. During that period there were 457,083 deaths in theprovince. Approximately 8% were eliminated from thedata either because the individuals were under the age of20 or because the records were part of "a small number"having invalid information on sex, occupation, or cause

of death. The authors examined the proportional mor-tality of 10,036 male metal workers. Because only sevendeceased women were recorded as metal workers, theywere excluded from the analysis. The authors apparentlycompared the proportion of deaths by cause in malemetal workers and subsets, including 1002 deceasedwelders, with the proportion of deaths by cause amongthe 254,901 men over 20 who died in the province duringthose years and had valid information on their deathcertificates. (The word "apparently" is used because thewording in the paper regarding the source of theexpected numbers is somewhat ambiguous.) Becausethese authors examined 216 occupations and 158 causesof death, the statistical significance of any of their find-ings has to be interpreted with caution; by chance alone,5% of their findings would be statistically significant, andover 34,000 such comparisons were made.

The authors presented only limited results. They spe-cifically stated that for other types of cancer within eachoccupational group for which the data were not pre-sented, the proportional mortality ratios were "asexpected." For welders, three causes were presented. Theproportional mortality ratio for the 207 deaths from allcancers combined was 114, a finding which was ofborderline significance. This is not an unusual value forsuch a PMR, and the statistical significance of the find-ing is not impressive in view of the number of compari-sons that were made. This ratio is likely to be explainedby chance and, regardless of the reason, is unlikely to beof clinical significance. More important, the nine deathsfrom Hodgkin's desease were nearly 2.5 times theexpected number. Although this finding was statisticallysignificant (p - 0.03), approximately 1000 findings withthat degree of statistical significance would be expectedby chance alone. The other reported result for lungcancer is discussed under that disease.

The authors discussed a number of limitations in theirdata. First, they raised the possibility of inaccurate causeof death coding. As they note, this is reasonably reliablefor most causes and even more reliable for specificcancers. This problem is not extensive enough to affectthe results markedly. A potentially more extensive prob-lem is inaccuracy of occupational information on thedeath certificates. However, Petersenand Milham(1974)found approximately 80% agreement of occupationgiven on the death certificate with that provided oninterview with next of kin or other relatives. The biggestlimitation of the data is the forced use of the propor-tional mortality ratio, the limitations of which arediscussed extensively elsewhere in this report. However,according to Gallagher and Threlfall, the PMR for arteri-osclerotic heart disease was near 100 (the null value),which makes less likely, a compensating excess in theproportional mortality ratio for specific cancers to coun-

34

terbalance a deficit in the most common cause of deathnevertheless. However, an excess in a particular cause inthe proportional mortality ratio could still occur in theabsence of an increased risk, simply because the authorsare forced to measure the proportion of deaths ratherthan the force of mortality.

The authors conclude that "the higher risk ofHodgkin's disease and multiple myeloma in welders inBritish Columbia is a new finding and will require con-firmation in more detailed studies." Their conclusion isjustified even more about Hodgkin's disease by the lim-itations of their methodology, particularly the forced useof proportional mortality ratio. Therefore, the findingson Hodgkin's disease require confirmation throughbetter designed studies, although by themselves they dogive some ground for concern.

The reference in passing to an increased risk of multi-ple myeloma in welders in this study is puzzling, becausethe authors do not present any data for multiplemyeloma in welders. As noted earlier, they specificallystated that any data not presented was omitted becausethe observed and expected deaths were in very goodagreement. They did find a proportional mortality ratioof over 200 for multiple myeloma in machinists anddiscussed this finding elsewhere in the paper. Presuma-bly, the sentence was in error and the authors werereferring to the elevated Proportional Mortality Ratiosfor Hodgkin's disease among welders and for multiplemyeloma in machinists (not welders).

2.17 References

Baba, K. Indications of an increase of occupational pleu-ral mesothelioma in Japan./ . UOEH. 5(1):3-15; 1983.

Baddeley, B., Thomas, B. J., Thomas, B. W., andSummers, V. Liver cadmium concentrations in metalindustry workers. Brit. J. Radiol. 56:449-451; 1983.

Barnes, R. Compensable asbestos-related disease in NewSouth Wales. Med. J. Aust. 2:221-224; 1983.

Beaumont, J. J. and Weiss, N. S. Lung cancer in welders:the experience of 3247 members of a union local.Abstract of presentation at International Conference ofHealth Hazards and Biological Effects of WeldingFumes and Gases. Copenhagen, 18-21 February 1985.

— Authors' response (letter to the editor). / . Occup.Med. 24:81-75; 1982.

Bergert, V. K. D., Voigt, H., and Holler, U. Zum Expos-tionsnachweis bei Schweibern durch die Manganbes-timmung im biologischen Material. Z. gesamte. inn.Med. 37:504-507; 1982.

Blot, W. J., Morris, L. E., Strube, R., Tagnon, I., andFraumeni, Jr., J. F. Lung and laryngial cancers in rela-

tion to shipyard employment in coastal Virginia. J. Natl.Cancer Inst. 65:571-575; 1980.

Boylen, C , Kilburn, K., Balmes, J., Johnson, S. J., andWarshaw, R. Effects of acute exposure to aluminumstainless steel and mild steel welding fumes on symptomsand airflow. Am. Rve. Respir. Dis. 129(4):A174; 1984.

Coggon, D., Pannett, B., and Acheson, E. D. Use of jobexposure matrix in an occupational analysis of lung andbladder cancers on the basis of death certificates. / . Natl.Cancer Inst. 72:61-65; 1984.

Di Maria, G. U., Palermo, F., Crimi, N., Fogliano, S.,and Mistretta, A. Respiratory morbidity in workersexposed to welding fumes. A cross-sectional study(Abstract). Am. Rev. Respir. Dis. 129(4):A177; 1984.

El-Gamal, F., Chinn, D., Bridges, N., and Cotes, J.Relative sensitivities of single breath nitrogen and flow-volume indices as guides to impaired lung function inyoung shipyard welders and caulker/burners. Abstractof presentation at International Conference on HealthHazards and Biological Effects of Welding Fumes andGases. Copenhagen, 18-21 February 1985.

Englund, A., Ekman, G., and Zabrielski, L. Occupa-tional categories among brain tumor cases recorded inthe Cancer Registry in Sweden. Ann. N. Y. Acad. Sci.381, 188-196; 1982.

Fawer, R. F., Gardner, A. W., and Oakes, D. Absencesattributed to respiratory diseases in welders. Brit. J. Ind.Med. 39:149-152; 1982.

Freedman, A. P., Robinson, S. E., and Green, F. H. Y.Magnetopneumography as a tool for the study of dustretention in the lungs. Ann. Occup. Hyg. 26(1-4):319-335; 1982.

Freedman, A. P., Robinson, S. E., and Huber, H. S.Magnetopneumographic investigation of determinantsof occupational lung dust retention. In OccupationalLung Disease (J. Bernard, L. Gee, W. Keith, C. Morgan,and S. M. Brooks, eds.). New York: Raven Press; 1984.

Gallagher, R. P. and Threlfall, W. J. Cancer mortality inmetal workers. Can. Med. Assoc. J. 129:1191-1194;1983.

Greenwald, P., Friedlander, B. R., Lawrence, C. E.,Hearn, T., and Earle, K. Diagnostic sensitivity bias — anepidemiologic explanation for an apparent brain tumorexcess. J. Occup. Med. 23:690-694; 1981.

Hernberg, S. Y., Collin, Degerth, R., Englund, A.,Engzell, U., Kuosma, E., Mutanen, P., Nordlinder,H., Hansen, H. S., Schultz-Larsen, K., Sogaard, H.,and Westerholm, P. Nasal cancer and occupational

35

exposures. Stand. J. Work Environ. Health. 9:208-213;1983a.

Hernberg, S., Westerholm, P., Schultz-Larsen, K.,Degreth, R., Kuosma, E., Englund, A., Engzell, U.,Hansen, H. S., and Mutanen, P. Nasal and sinonasalcancer. Scand. J. Work Environ. Health. 9:315-326;1983b.

Hogstedt, P., Kadefors, R., and Naslund, P.-E. Magne-topneumography in welders: observations on relaxationand clearance. Nuovo Cimeto. 2D(2):608-616; 1983.

Jarvisalo, J., Olkinuora, M., Tossavainen, A., Virtamo,M., Ristola, P., and Aitio, A. Urinary and blood man-ganese as indicators of manganese exposure in manualmetal arc welding of mild steel. Chem. Toxicol. Clin.Chem. Proc. Clnt. Conf. March 19, 1983.

Jinadu, M. K. Occupational health problems of motorvehicle mechanics, welders and painters in Nigeria. Soc.Health J. 102:130-132; 1982.

Kalic-Filipovic, D. Some aspects of welders' chronicmorbidity. Results of follow-up study (3-year). Abstractof presentation at International Conference of HealthHazards and Biological Effects of Welding Fumes andGases. Copenhagen, 18-21 February 1985.

Kalliomaki, P.-L., Kalliomaki, K., and Korhonen, O.Lung retention of welding fumes and ventilation lungfunctions — A follow-up study of shipyard welders.Abstract of presentation at International Conference onHealth Hazards and Biological Effects of WeldingFumes and Gases. Copenhagen, 18-21 February 1985.

Kalliomaki, P.-L., Kalliomaki, K., Korhonen, O.,Nordman, H., Rakhonen, E., and Vaaranen, V. Respira-tory status of stainless steel and mild steel welders.Scand. J. Work Environ. Health. 8:117-121; 1982.

Keimig, D. G., Pomrehn, P. R., and Burmeister, L. F.Respiratory symptoms and pulmonary function inwelders of mild steel: across-sectional study. Am. J. Ind.Med. 4:489-499; 1983.

Kramer, S., Ward, E., and Meadows, A. T. Case controlstudy of risk factors for neuroblastoma. Proc. Am.Assoc. Cancer Res. 25:222; 1984.

Krasnyuk, E. P., Lubyanova, I. P., Timofeyeva, N. T.Changes of respiratory and cardiovascular systems inalloyed steel welders. Abstract of presentation at Inter-national Conference of Health Hazards and BiologicalEffects of Welding Fumes and Gases, Copenhagen18-21, February 1985.

Lam, Y. M., and Tan, T. C. Mortality from nasopharyn-geal carcinoma and occupation in men in Hong Kong

from 1976-1981. Ann. Acad. Med. Singapore. 13(2):361-365; 1984.

Levy, B., Hoffman, L., and Gottsegen, S. Boilermakerbronchitis. J. Occup. Med. 26:567-570; 1984.

Lin, R. S., Dischinger, P. C , Conde, J., and Farrell, K.P. Occupational exposure to electromagnetic fields andthe occurrence of brain tumors. J. Occup. Med. 21:413-419; 1985.

Lindquist, B. Kidney disease in welders. IRCS Med. Sci.11:99; 1983.

Lippman, M. Magnetopneumography as a tool formeasuring lung burden of industrial aerosols. Abstractof presentation at International Conference of HealthHazards and Biological Effects of Welding Fumes andGases, Copenhagen, 18-21 February 1985.

Littorin, M., Welinder, H., and Hultberg, B. Kidneyfunction in stainless steel welders. Int. Arch. Occup.Environ. Health. 53:279-282; 1984.

Lyngenbo, O., Groth, M., Groth, S., Dirksen, H., andOlsen, O. Respiratory symptoms and pulmonary func-tion among Danish welders. Abstract of presentation ofInternational Conference on Health Hazards and Bio-logical Effects of Welding Fumes and Gases. Copen-hagen, 18-21 February 1985.

MacMahon, B., and Pugh, T. Epidemiology:principlesand methods. Boston: Little, Brown, and Co.; 1970: p.267.

McMillan, G. H. G. The health of workers in navaldockyards: final summary report. /. Roy. Nav. Med.Serv. 69:25-131; 1983.

— The health of welders in her majesty's dockyardsmortality study. Abstract of presentation at InternationalConference of Health Hazards and Biological Effects ofWelding Fumes and Gases. Copenhagen, 18-21 Febru-ary 1985.

McMillan, G. H. G. and Pethybridge, R. J. The health ofwelders in naval dockyards: proportional mortalitystudy of welders and two control groups. J. Soc. Occup.Med. 33:75-84; 1983.

— A clinical radiological and pulmonary functioncase control (sic) study of 135 dockyard welders aged45 years and over. J. Soc. Occup. Med. 34:3-23;1984.

Miller, E. R., Smith, E. M., Silverman, D. T., Woolson,F., and Brown, C. K. Risk of bladder cancer amongsolderers and welders. Presented at the Tenth Scien-tific Meeting of the International Epidemiological

36

Association, University of British Columbia, Vancouver,Canada, August 19-25 1984.

Mur, J., Pham, Q., Meyer-Bisch, C , Diebold, F., Salsi,S., Massin, N., Moulin, J., Teculescu, D., and Cavelier,C. Effects on respiratory function of long term exposureto arc welding fumes. Overall results of 3 epidemiologi-cal surveys. Abstract of presentation at InternationalConference on Health Hazards and Biological Effects ofWelding Fumes and Gases. Copenhagen, 18-21 Febru-ary 1985.

Mutti, A., Pedroni, C , Arfini, G., and Franchini, I.Biological monitoring of occupational exposure to dif-ferent chromium compounds at various valency states.Intern. J. Environ. Anal. Chem. 17:35-41; 1984.

Naslund, P.-E. and Hogstedt, P. Welding and bronchi-tis. Eur. J. Respir. Dis. 63:69-72; 1982.

Newhouse, M. L., Oakes, D., and Woolley, A. J. Mortal-ity of welders and other craftsmen at a shipyard innortheast England. Abstract of presentation at Inter-national Conference on Health Hazards and BiologicalEffects of Welding Fumes and Gases. Copenhagen, 18-21 February 1985.

Olsen, J., Sabroe, S., and Lajer, M. Welding and cancerof the larynx: a case control study. Eur. J. Cancer Clin.Oncol. 20:639-643; 1984.

Petersen, G. R. and Milham, Jr., S. Hodgkin's diseasemortality and occupational exposure to wood. / . Natl.Cancer Inst. 53: 957-958; 1974.

Polednak, A. P. Mortality among welders including agroup exposed to nickel oxides. Arch. Environ. Health.36:235-242; 1981.

Rahkonen, E., Junttila, M. L., Kalliomaki, P. L.,Olkinourora, M., Koponen, M., and Kalliomaki, K.Evaluation of biological monitoring among stainlesssteel welders. Int. Arch. Occup. Environ. Health. 52:243-255; 1983.

Santi, L., Vercelli, M., Bonassi, S., Ceppi, M., Stagnaro,E., Valerio, F., and Puntoni, R. Bladder and (lung)cancer among autogenous and electrical welders in theport of Genoa in relation to lenght (sic) of exposure.Abstract of presentation at International Conference ofHealth Hazards and Biological Effects of WeldingFumes and Gases. Copenhagen, 18-21 February 1985.

Schneider, W. D., Grund, W., Schmidt, G., Wiesner, W.Respiratory diseases in welders on stainless steel and/ oraluminum — A cross-sectional study. Eur. J. Respir.Dis. 62: Suppl, 179-180; 1980.

Sidenius, K., Sorensen, H. H., and Svane, O. Strubek-raeft blandt metalarbejdere. Ugeskr. Laeg. 145:1623-1628; 1983.

Sjogren, B. and Carstensen, J. Cancer among Swedishwelders and gas cutters. Abstract of presentation atInternational Conference on Health Hazards and Bio-logical Effects of Welding Fumes and Gases. Copen-hagen, 18-21 February 1985.

Sjogren, B., Hedstrom, L., and Lindstedt, G. Urinaryfluoride concentration as an estimator of welding fumeexposure from basic electrodes. Brit. J. Ind. Med.41:192-196; 1984.

Sjogren, B., Hedstrom, L., and Ulfvarson, U. Urinechromium as an estimator of air exposure to stainlesssteel welding fumes. Int. Arch. Occup. Environ. Health.51:347-354; 1983a.

Sjogren, B., Hogsteddt, C , and Malker, H. Chromiumand asbestos, two probable risk factors in lung canceramong welders (letter to the editor). J. Occup. Med.24:874-875; 1982.

Sjogren, B., Lundberg, I., and Lidums, V. Aluminum inthe blood and urine of industrially exposed workers.Brit. J. Ind. Med. 40: 301-304; 1983b.

Sjogren, B., Weiner, J., Horte, L. G., and Carstensen, J.Mortality among Swedish welders and gas cutters.Abstract of presentation at International Conference ofHealth Hazards and Biological Effects of WeldingFumes and Gases. Copenhagen, 18-21 February 1985.

Spitz, M. R. and Johnson, C. C. Neuroblastoma andpaternal occupation: a case control analysis. Am. J.Epidemiol. 121:924-292; 1985.

Stem, R. M. A protocol for the study of health effects ofwelding. Report from the National Technical Informa-tion Service, Springfield, Virginia, 1981.

Stern, R. M. and Drenck, K. Susceptibility bridgemagnetopneumography of individuals with and withoutoccupational dust exposures. Svejuscentralen. TheDanish Welding Institute Report: 1982.

Stern, R. M., Drenck, K., Lyngenbo, O., Dirksen, H.,and Groth, S. Health, occupational exposure, and thor-acic magnetic moment of highly exposed lifetime-nonsmoking shipyard welders. Abstract of presentationat International Conference of Health Hazards and Bio-logical Effects of Welding Fumes and Gases, Copen-hagen 18-21, February 1985.

Toda, K., Mori, Y., Koike, S., Ohmori, K., and Yamaga,S. Chronic renal dysfunction in workers exposed to

37

cadmium oxide fume. Jpn. J. Ind. Health. 268:212-223;1984.

Ulvik, R. J. Subnormal serum zinc concentration in apatient with zinc fever. J. Soc. Occup. Med. 33:187-189;1984.

Waddington, J., Tarkowski, S., and Farkas, I. WHO/EURO monitoring an Epidemiology Programmes Re-lated to Heavy Metals. In Fourth Heavy Metals in theEnvironment International Conference, Heidelberg.Edinburgh, Scotland: CEP Consultants, Ltd.; 1983:Vol. I, pp. 33-40.

Welinder, H., Littorin, M., Gullberge, B., and Skerfving,S. Elimination of chromium in urine after stainless steelwelding. Scand. J. Work Environ. Health. 9: 397-403;1983.

3. Experimental Investigations3.1 Animal Studies

3.1.1 Carcinogenesis. The possible bronchocarcino-genic effects of fumes generated by two stainless steelprocesses, manual metal arc (MMA) welding and metalinert gas (MIG) welding, were examined in 70 hamstersby intratracheal instillation of either 0.5 or 2 mg of MIGand MMA fumes (Reuzel et al., 1985). Following once-weekly administrations for 340 days, the hamsterstreated with high doses of MMA and MIG fumesshowed increased lung weight, interstitial pneumonia,alveolar broncholization, and emphysema. Hamsterstreated with 0.5 mg of MMA fumes or calcium chromate(VI) (CaCrO4) had similar changes that were much lesssevere. Histological examination revealed 2 malignantlung tumors in the MMA-exposed group; none weredetected in any other group, including 791 historicalcontrols.

In an attempt to study the possible bronchocarcino-genic effects of welding fumes containing chromium,Berg et al. (1985) collected and implanted fume particlesas pellets in the bronchi of groups of 100 rats. Theparticles were shown to contain both tri- and hexavalentchromium in soluble and low-soluble forms. After 34months, no significant differences were noted in growthrates, survival times, terminal organ weights, macro-scopic and microscopic findings, and precancerouschanges or tumors at the implantation site between thetest and negative control groups. By contrast, a positivecontrol group receiving pellets containing benzo(a)py-rene developed epithelial cell tumors in all rats. Thesignificance of apparent negative findings, however, maybe questionable because the implantation technique maynot adequately mimic true human exposure by inhala-

tion in terms of absorption and bioavailablility of thewelding fume particles.

3.1.2 Pneumoconiosis The biological effects of var-ious welding processes and their respective generatedfumes were studied in rats by three investigators. Hickset al. (1984) focused on the effects of MMA weldingusing flux-coated mild steel electrodes and MIG weldingusing stainless steel. Al-Shamma et al. (1983) examinedan array of flux-coated electrode welding fume parti-clates, in more detailed analysis of fume effects. Finally,a study by Kennebeck (1985) was available for reviewonly as an abstract, which did not provide details on thespecific type of welding materials.

Hicks et al. (1984) compared the biological effects inrats of fume materials from MMA welding with flux-coated mild steel electrodes to MIG stainless steelwelding by either inhalational exposure or intratrachealinjection. The MMA fumes were obtained by therepeated consumption of flux-coated (FC) rutile-ironmild steel (MS) rod electrodes generating iron (metaland oxide forms) and silica (25% W/W as SiO2) parti-cles, oxides of nitrogen (NO2) plus a mixture of otherparticles. A group of 40 rats were exposed to an averageMMA/ MS fume-particle concentration of 1178 mg/m3

for 46 minutes. Metal inert gas fumes were producedfrom stainless steel wire, Bostrand 61, consisting mainlyof iron, chromium, and nickel. Twenty-four rats wereexposed for one hundred eighty-four minutes to MIG/SS welding fumes having an average concentration of400 mg/m3. In a second experiment, 30 rats wereexposed to an average concentration of 580 mg/m3 for173 minutes. Animals were exposed to fumes drawnfrom the region of the arc and passed into a "head only"chamber. The animals were exposed for different lengthsof time in these experiments because the concentrationsof particulate material differed.

Samples of particulate material were collected fromthe welding processes and were suspended in saline forintratracheal administration of 50 mg of MMA/MSfume particles to each of 30 rats. In serial experiments,groups of 20 rats were given intratracheal doses of either10 or 50 mg of MIG/MS or MMA/SS fume particles.The groups of animals were sacrificed at various timesfor histopathological examination.

Histopathologic changes in the rat lung tissue werecompared at various times after inhalational exposuresto MMA/ MS and MIG/ SS welding fumes. After two tofive hours, widespread pneumonitis was observed in thelungs of rats exposed to either the MMA/MS orMIG/SS welding fumes. Two days after exposure toMMA/MS welding fumes, the pneumonitic exudatewas reduced with some aveolar septal thickening. Incomparison, pneumonitis persisted with more extensive

38

alveolar septal thickening two days after MIG/ SS expo-sure. Peripheral alveoli became clear and free of visibleparticles between 7 and 21 days after MMA/MS expo-sure, whereas the alveolar epithelial thickening becamemore extensive, with particle-laden macrophages widelydistributed in MIG/SS-exposed. Even 30 days afterexposure, alveolar septal thickening was much moreextensive in the MIG/ SS-exposed rats. By 300 days, fewparticle-containing macrophages remained in the alveoli,thickening had mostly abated, peribronchial aggregateswere well demarcated, and numerous well-formed,particle-engorged giant cells had appeared in theMMA/MS-exposed rats. At this time in the MIG/SS-exposed rats, the peribronchial clusters were also welldeveloped, but more irregular, particle-laden cells andepithelial thickening were still extensive, and numerousgiant-cells appeared that were misshapen and devoid ofparticles.

The histopathological changes in rat lungs were alsocompared after a single intratracheal injection of 50 mgof either MMA/MS or MIG/SS welding fumes. At 90to 100 days after exposure, extensive and massive depos-its of particles, some collagen formation, and aggrega-tion of some multinucleated giant cells were observed inMMA/MS-exposed rats. At that time in the MIG/SS-exposed rats, deposits of particles were irregular andmassive; evidence of macrophage activity was extensive,fibrous formation was slight; thickening of alveolarepithelium was extensive; and giant cells were present.By 450 days, distinct collagen development and occur-rence of numerous, regularly shaped giant cells wereobserved in MMA/MS-exposed rats. The MIG/SS-exposed rats were characterized at this time by massiveconsolidation and epithelial-cell proliferation of particle-laden cells, little or no collagen, and numerous, irregu-larly shaped giant cells.

In summary, the study reported by Hicks et al. (1984)revealed some differences between the effects of the twoforms of welding. The lung damage observed in ratsincluded an initial effect consisting of acute inflamma-tory responses and a later effect associated with cellularchanges and the persistence of dispersed or aggregateddeposits of particles. The proliferative changes asso-ciated with pneumonitis included a thickening of alveo-lar epithelial septa, which occured in response to eithertype of welding fume. Using alveolar thickening as anindicator, MIG/SS fumes appear to have caused greaterand more prolonged pneumocyte damage than MMA/MS fumes.

Al-Shamma et al. (1983) conducted a similar rat studytesting the separate components of flux-coated electrode(OK 46.00) welding fumes. Analysis showed the fumes tobe composed of silicon dioxide particles, metals, metaloxides, nitrogen oxides, and hydrogen fluoride.

A group of 36 rats were exposed by inhalation toparticulate matter for four hours then sacrificed, ingroups of six, immediately and at 1,7,14,28, and 60 daysafter exposure. In a long-term experiment, groups of 24rats were injected intratracheally with 50 mg of a mixtureof nitrous fumes, hydrogen fluoride, carbon monoxide,manganese, silicon dioxide, and iron and sacrificed at 3,6, 8, and 10 months.

In both experiments, one day after exposure, rats hadedema, acute pneumonitis, and hyperplasia or alveolarwall thickening, which were not apparent in rats sacri-ficed at later times. The authors suggest that thepneumonitis and edematous reaction were probablyattributable to the high concentration of nitrous fumesand hydrogen fluoride, which are both known lung irri-tants. A statistically significant increase in the ratio oflung weight to body weight ratio of exposed rats wasobserved. The animals sacrificed at the later stages hadfibrosis and nodules surrounded by bundles of collagen.The high levels of collagen were indicated by a significantincrease in hydroxyproline levels. Most of the weldingparticles and macrophages were localized at the peri-vascular and peribronchial areas. The fibrotic changesobserved suggest that silica was probably responsible.

In conclusion, each of the fume constituents had dis-tinct harmful effects when administered either by intra-tracheal or inhalational exposure in rats.

The toxicity in rats of six welding materials that areused in shielded metal, gas metal, and flux core arcwelding is described in a meeting abstract by Kennebeck(1985). After six hours of exposure, urine samples weretaken and rat lungs were necropsied for microscopicexamination. The lung tissues of rats that had beensubjected to lung lavage were bloody or had reduced cellpopulations at 20 hours postexposure. Signs of pig-mented macrophages and occasional focal acute pneu-monitis, multifocal perivascular edema, and changes inlung weight were observed 14 days after exposure to eachgroup of fumes tested. Out of the six fumes tested, thestainless steel electrode was determined to be the mosttoxic to rats.

Uemitsu et al. (1984) conducted an inhalational toxic-ity study on fumes generated from MMA/MS with alime electrode (L fumes) and from MMA/SS with alime-titania electrode (S fumes), a process that producesparticles comparatively rich in nickel and water-solublehexavalent chromium. Rats were exposed to 1000mg/m3 of either MMA/MS L fumes or MMA/SS Sfumes for 1 hour, or to 400 mg/m3 for 30 min/day, 6days/week, for 2 weeks.

In the single-exposure study, microscopic abnormali-ties were found only in the lungs of rats exposed tofumes. The damaged lungs gradually returned to normalin rats exposed to L fumes, but the rats exposed to S

39

fumes after 14 days continued to show slight lungdamage.

From repeated exposure to fumes, the rat lungs hadhistopathological signs of discoloration, weight change,and morphological changes only in rats treated toS-fumes. Significant (p < 0.01) increases in the wetweight and relative lung weights approximately 1.2 or1.7 times over controls were observed in L fumes — or Sfumes—exposed groups, respectively. Respiratory irri-tation, such as mucous granules in the alveoli and hyper-plasia of mucous cells in the bronchial epithelium, andbronchiectasia were also observed in the group exposedto S fumes. The authors suggest that the morphologicalabnormalities and respiratory irritation are attributableto allergic and irritant reaction to chromium, since thisresponse cannot be explained by acidity or alkalinity ofthe weakly acidic fumes (pH = 6.6).

In conclusion, the stainless steel welding particulatesinduced greater and longer-lasting effects in the lung ofrats than did mild steel welding particulates.

3.1.3 Whole-Body Effects. The acute and subchronictoxicity of the following trivalent and hexavalent chro-mium compounds were studied in mice by Bryson andGoodall (1983) to determine the active forms of thesecompounds: chromium nitrate [Cr(NO3)3-9H2O],chromium trichloride (CrCL3 • 6H2O), chromic sulfate[Cr(SO4)3-15H2O], and chromium trioxide (CrO3).The acute toxicity of chromium nitrite, chloride, andsulfate salts; chromium trioxide; and potassium di-chromate was tested by intraperitoneal administration inNZC male mice and in CXO male and female mice. TheLD50 estimates on day 3 and day 10 were determinedusing 2 or 4 animals per group, at 5 to 10 different doselevels. Hexavalent chromium compounds were moreacutely toxic than the trivalent chromium salts; the meanLD5Os (day 3) were 17.8 /ig/g and 39.1 Mg/g, respec-tively. However, the mean distal LD5Os were similar,regardless of the oxidation states of these compounds.Slightly more toxicity was seen in female than in malemice.

Subchronic toxicity and whole body retention werestudied in mice by intraperitoneal injection of eithertrivalent chromium (III) (3.25 Mg/g) o r chromium (3.23/xg/g), administered one time or weekly for 8 weeks(Bryson and Goodall, 1983). Because mice receivingmultiple treatments of chromium (VI) showed signs ofbody weight loss (range 15% to 20% at the end of theeighth week, the intervals of injections were changed toevery two weeks. (Dosages were one-sixth of the distalLD50 per injection.) The whole-body retention of trival-ent chromium was 6.5 times higher than that of hexaval-ent chromium at 21 days after a single injection. Follow-ing multiple doses, approximately nine times more

trivalent chromium was retained than hexavalentchromium (VI). Both were measured by whole-bodyclearance rates and cumulative total excretion of chro-mium in urine and feces. Analyses of fecal and urinaryexcretion confirmed that most of the urinary chromiumclearance that occurred soon after injection, and chro-mium excretion from animals treated with hexavalentchromium was much faster in both urine and feces thanfrom those treated with trivalent chromium. However,the rate of chromium excretion in the urine was greaterthan that in the feces for both forms of chromium. Afteradministration of a single dose of trivalent chromium,87% was retained 3 days after treatment, 73% after 7days, and 45% after 21 days. The high retention rate oftrivalent chromium was attributed to its binding to formcoordination compounds or polynuclear complexes. Bycomparison, potassium dichromate showed a much fas-ter removal rate. At 3 days, 31% was retained, 16% at 72days, and only 7.5% at 21 days. The low initial retentionof hexavalent chromium ion is attributed to its lowability to form co-ordination compounds at physiologi-cal pH. Most of the hexavalent chromium is excreted inthe urine, but the chromium remaining in the peritonealcavity would be expected to be reduced from hexavalentto trivalent in a few hours and then to be released by thetrivalent chromium mechanism. Similar retention rateswere observed when chromium was administered re-peatedly to mice.

The toxicokinetics of hexavalent chromium was stud-ied in the rat after intratracheal administration ofchromates of different solubilities (Bragt and van Dura,1983). After a single intratracheal dose of radiolabeledsodium zinc, or lead chromate, sodium chromate andless soluble zinc chromate were absorbed into the blood,resulting in increased excretion of chromium into theurine. The maximum blood chromium level from sodiumchromate was 0.35 /xg Cr/ml at 30 minutes after intra-tracheal administration. Zinc and lead chromates showedmaximum levels of 0.60 and 0.007 jug Cr/ ml respectively,at 24 hours. The cumulative excretion of radiolabeledchromium in urine up to 10 days after dosing was about25% and 20% of the administered dose for the zinc andsodium chromates, respectively. Only about 2% ofadministered lead chromate was excreted in the urinewithin 10 days. The more insoluble a chromate was inwater, the higher its elimination in the feces. For exam-ple, the cumultive excretion in feces was almost 80% ofthe administered dose of lead chromate for up to 10 daysafter dosing. Chromium was retained in the spleen andbone marrow at higher levels than in blood for all threechromates. For sodium chromate, the levels in the kid-neys and liver were also significantly higher than inblood. The authors suggested that biological monitoringof chromium in urine and blood of workers dealing with

40

lead chromate is not indicative of previous exposurebecause of the possibility of minor resorption of leadchromate.

A study (reported as an abstract) of the levels ofwelding fume components in various bodily tissues andfluids of rats was done to demonstrate the physical-chemical differences between fumes generated in MMAand MIG welding (Aitio, 1985). After exposure to MIGwelding fumes, chromium and iron were cleared veryslowly from the lungs, and even nickel and manganesedisappeared slowly, with half-times of 85 and 125 days,respectively. In contrast, after exposure to MMA weld-ing fumes, some nickel and chromium disappeared fromrat lungs immediately after exposure, followed a second,slower phase of disappearance with half-times of 40 and20 days for nickel and chromium, respectively. Urinaryexcretion rates for chromium after exposure to MMAwelding fumes corroborated the biphasic rate of dis-appearance and suggested that there may be accumula-tion of chromium in the body during the working week.

For humans, Aitio (1985) reported that data on kine-tics of nickel and manganese after exposure to differentwelding fumes are lacking. There appeared to be increasedurinary excretion of both after MMA welding, but forMIG welders, there was no increased urinary excretionof nickel, and there were no data for manganese.

All metal components of welding fumes were knownto accumulate in the lungs, according to Aitio (1985).However, little is known about the metal contents ofother tissues in welders. The concentration of chromiumincreased five-fold in the liver and kidney of rats exposedto MM A welding fumes, but the concentration of nickeldid not increase. No increases were detected in eitherchromium or nickel levels in rat brain.

3.1.4 Retention and Clearances. Several studies havebeen reported recently by Kalliomaki and coworkers onthe retention and clearance of different forms of weldingfumes in rats (Lakomaa et al., 1982; Kalliomaki et al.,1982; Kalliomaki et al., 1983a; Kalliomaki et al., 1983b;Kalliomaki et al., 1983c; Kalliomaki et al., 1983d; andAnttila et al., 1985). In general alveolar retention offume particles from stainless steel MMA welding isgreater than from mild steel MMA welding.

For example, Kalliomaki et al. (1983a) compared theretention and clearance of MMA/SS and MMA/MSwelding fumes in rats. In this study, rats were exposed to43 mg/m3 of either of the two types of welding fumes byinhalation using a "nose-only" exposure chamber. ForMMA/SS welding fumes, the concentrations for Cr, Ni,and Fe (exogenous) in the lungs correlated well (p <0.001) with cumulative duration of exposure, showing alinear retention of these elements up to 20 days. Theretention of Mn became saturated after 19 hours of

exposure and was low in comparison with the otherelements. The biological half-times for lung clearancewere 30 to 50 days for these elements, indicating a quiteslow clearance mechanism after exposure to MMA/SSwelding fumes.

In constrast, the retention rate of Fe (exogenous)found for exposure to MMA/MS welding fumes wasrapid (28 /xg/h) during the first 10 days of exposure, butthereafter, the concentration saturated. The concentra-tion of Mn also saturated during the first five days ofexposure. The lung clearance patterns of Fe (exogenous)and Mn were also different from those observed forMMA/SS welding fume exposure. Bilogical half-timesof clearance were 6 and 0.5 days for Fe (exogenous) andMn, respectively, indicating rapid clearance.

In summary, after four weeks of exposure, the maxi-mum alveolar retention of MMA/ SS welding fume par-ticles was about four times higher than retention ofMMA/MS particles in rats. This finding apparentlyagrees with earlier observations made in welders. Totaliron concentration in the lungs of rats exposed toMMA/MS welding fumes was two times higher thanthat in unexposed rats. In contrast, MMA/SS weldingfumes did not significantly affect total iron concentra-tion. Clearance of MMA/MS welding fume particleswas much more rapid than that of MMA/ SS particles.

3.2 In Vitro Studies

3.2.1 Mutagenesis Assays. Previous studies testingMIG/SS or MMA/SS welding fumes in the Ames Sal-monella typhimurium assay have identified the hexaval-ent form of chromium as the active mutagen. It has beenreported that hexavalent chromium is not stable inMIG/SS fumes but decays rapidly, defying standardcollection methods so that measures of the mutagenicityof fumes experienced by welders in an occupationalsetting are likely to be inaccurate (Stern, 1983).

For this reason, Stern (1983) investigated the possibil-ity of using a modification of the Ames S. typhimuriumassay to evaluate fumes from MIG/SS welding. MIG/ SSand MMA/SS fumes were generated by a referencewelding robot and either collect for analysis or passedthrough an impinger system designed to collect freshfumes that would theoretically mimic the actual fumes towhich welders are exposed during work. Fumes wereroutinely analyzed for water-soluble and -insoluble frac-tions of iron and chromium. The hexavalent form ofchromium was analyzed using atomic absorption spec-trometry. Strain TA100 cells growing in suspensionwithout a metabolic activation system were placed in asingle impinger system producing MIG/ SS fumes. Theywere exposed for durations varying from 200 seconds to3600 seconds, in 200-second intervals, during which time

41

a maximum of 4.8 mg of fume was collected; the cellswere then plated for mutagenicity and survival. Celltoxicity increased with exposure, but a concentration-related significant increase in the number of histidinerevertants per plate was observed. The investigator sug-gested that, for the purposes of assessing health risksfrom welding fumes, the potency of hexavalent chro-mium component found in welding fumes is identical onan equimolar basis to that of hexavalent chromium fromother sources. Although the methods used in this assayare a departure from the standard assay protocol, theresults are generally in agreement with the results ofother researchers who have used the standard plate-incorporation method (Stern et al., 1982; Reuzel et al.,1985).

3.2.2 Cell Transformation Assays. Hansen and Stern(1983, 1985) studied the ability of nickel components ofMIG welding fumes to transform baby hamster kidneycells (BHK-21) and Syrian hamster embryo (SHE) cellsin culture. These assay systems were modified fromstandard assay protocol, although the modificationswere not specified. This protocol was chosen so thattransformation rates due to the presence of nickelcompounds in the MIG/SS welding fumes could becompared with the transformation rates of nickel subsul-fide (a known animal carcinogen), two different nickeloxides, and corundum on an equally toxic basis. Thismodification was based on the assumptions that onlynickel taken up by the cells would be toxic, that thetoxicity of these materials in transformation assays is adirect measure of available nickel, and that nickel is theultimate carcinogen independent of the uptake mecha-nism. Hansen and Stern observed that the transforma-tion frequencies at the 50% survival level were identical,suggesting to them that the nickel component of all thecompounds was responsible for both their toxicity andtheir transforming ability.

In a study using the same modified protocol, MMA/SS welding fumes were evaluated (Hansen and Stern,1985). Apparently, the water-soluable component, hex-avalent chromium, was the active compound in trans-forming BHK-21 and SHE cells. Trivalent chromiumcompounds were not active, nor were mild steel weldingfume particles. In a further modification of this assaysystem, growing BHK-21 and SHE cells were exposed tofresh welding fumes via a water impinger system used forcollection of fresh fumes. By this method, Hansen andStern were able to demonstrate the presence of a short-lived, biologically reactive species that is not typicallycollected and cannot be assayed in fumes obtained bystandard monitoring techniques. In this system, solublechromium compounds were toxic at 1 to 2 jigj ml, where-as soluble nickel compounds were toxic as 75

It should be noted that these results were obtainedthrough a modification of the standard in vitro BHK-21SHE cell transformation assays that has not been vali-dated by other researchers.

3.2.3 Cytogenetic Assays. MMA/SS welding fumeswere found to induce sister chromatid exchanges (SCEs)in vitro in Chinese hamster ovary (CHO) cells (Reuzel etal., 1985). Potassium chromate, when tested in the samesystem, was similar in potency to MMA/SS weldingfumes, but MIG/ SS welding fumes were negative in thissystem. No further details using this assay were reportedin the meeting abstract that was available for this review.

No significant increase in the frequency of choromo-some aberrations (CAs) and SCEs was observed in theperipheral lympocytes of 24 male MMA/SS workersfrom six industries in different parts of Sweden, whencompared with controls (Littorin et al., 1983). Controlswere matched for sex, age, smoking and drug consump-tion history, socioeconomic status, and living area. Thetest subjects were chosen for their long experience(median 19 years, range 7 to 41 years) with stainless steelwelding. Exposure to welding fumes during the samplingperiod was confirmed by (1) counting the number ofelectrodes used during the work day, (2) measuring thechromium in workplace air, and (3) measuring chro-mium levels in urine collected during the work day.

In contrast, there was a statistically significant (p <0.007) increase in frequency of the number of certaintypes of CAs (interchanges, dicentrics, ring chromo-somes, and marker chromosomes) of smokers comparedwith nonsmokers plus exsmokers. The frequency ofother types of CAs did not differ, not did the totalnumber CAs and micronuclei or the frequency of SCEs.However, negative results in this assay system may notbe relevant to the question of the genotoxicity of weldingfumes in man. Husgafvel-Pursiainen et al. (1982, 1983)conducted a similar study of CAs and SCAs in theperipheral lymphocytes of 23 MMA/SS welders fromthree different companies in Finland. These welders wereselected because they had been exposed almost exclu-sively to MMA/SS welding fumes in the preceding 10years (median length of employment in welding was 21 ±10 years). Exposure to welding fumes at the time of theblood sampling was confirmed by urine samples thatwere taken at the end of the work day and analyzed forchromium and nickel; a personal and medical historywas recorded for each welder and control. The urinelevels of chromium for some of the welders exceeded thestandard (0.6 /umole/1) for exposure recommended inFinland. Nevertheless, no significant difference wasobserved, in either CAs or SCEs in the blood of weldersand controls. These negative findings could not be attrib-uted to lack of exposure to welding fumes; however,

42

peripheral lymphocytes may not be the appropriatetarget site for measuring genotoxic effects of exposure.As has been observed elsewhere, there was a significant(p < 0.01) positive association between smoking andSCE frequency within both the welder group and thecontrol group.

3.2.4 In Vitro Toxicity Assays. The possibility ofusing bovine alveolar macrophages in culture to measurethe potential inhalational toxicity of welding fumes hasbeen investigated in two preliminary studies (Hooftmanet al., 1985, and White et a l , 1983), since alveolar macro-phages have been shown to be important in the lung'sability to detoxify inhaled particles.

In the study of White et al. (1983), welding particleswere obtained from three sources: stainless steel weldingwith a rutile-basic-coated electrode, mild steel weldingwith a basic-coated electrode, and mild steel weldingwith a rutile-coated electrode. The chemical composi-tions of the particles were analyzed so that chromiumsalt solutions could be made up to concentrations equiva-lent to the hexavalent chromium in the welding fumes.Titanium dioxide and chrysotile asbestos were negativeand positive controls, respectively. Alveolar macro-phages were harvested and welding fume particles, tri-valent and hexavalent chromium salts, titanium dioxideparticles, or chrysotile asbestos fibers were cultured withthe cells for 17 hours in serum-free medium. Toxicity wasmeasured by cell detachment and trypan blue exclusion.

In cultures exposed to welding fume particles, detach-ment increased with dose; at the highest tested dose(40 mg/ ml), about 95% of the cells were detached. A highpercentage of cells were nonviable by trypan blue exclu-sion, in ratios roughly proportional to dose. Weldingfumes from mild steel rutile-coated electrodes were theleast toxic, and stainless steel welding fumes obtainedfrom rutile-basic-coated electrodes were the most toxic,by both tests. Hexavalent chromium, in equimolar con-centrations to those in the stainless steel, was less toxicthan the whole particulate, but trivalent chromium wasnontoxic at equimolar concentrations. Therefore, it ispossible that compounds other than hexavalent chro-mium could be contributing to the toxicity of thesewelding fumes.

In additional studies, decreased toxicity was observedwhen the culture medium was supplemented with 10%ne"7V>orn-calf serum, but not when bovine serum albuminwas added to the medium. When the main constituent ofpulmonary surfactant, dipalmitoyl phosphatidycholine(DPPC), was added to samples in serum-free medium,welding fume particles tended to aggregate, making tox-icity difficult to assess. However, it was observed thatDPPC did not affect the toxicity of either hexavalentchromium or the stainless steel rutile-basic-coated elec-

trode welding particles, but did reduce the toxicity ofboth chrysotile asbestos fibers and mild steel rutile-coated electrode welding particles.

The study of Hooftman et al. (1985) confirmed thathexavalent chromium at concentrations greater than0.03 jUg/ml, but not trivalent chromium, was toxic toalveolar macrophages in culture. Although experimentaldetails were not reported in this abstract, the authorsstated that particles from MM A/ SS welding fumes werethe most toxic, and that hexavalent chromium toxicitywas comparable on an equimolar basis.

3.3 References

Aitio, A. Levels of welding fume components in tissuesand body fluids. Abstract of presentation at Inter-national Conference of Health Hazards and BiologicalEffects of Welding Fumes and Gases. Copenhagen, 18-21 February 1985.

Al-Shamma, K., Karim, K., and Al-Genno, G. Biologi-cal effects of inhaled welding fumes from flux-coatedelectrodes (OK 46.00). In: Developments in the scienceandpractice of toxicology. Hayes, A., Schnell, R., Miya,T. (eds.). Elsevier Science Publishers; 1983: pp. 487-490.

Anttila, S., Sutinen, S., Kalliomaki, P., Sivonen, S.,Grekula, A., and Nickels, J. Distribution, structure andcomposition of inhaled manual metal arc and shieldgasstainless and mild steel welding fumes in rat lung.Abstract of presentation at International Conference ofHealth Hazards and Biological Effects of WeldingFumes and Gases, Copenhagen, 18-21 February.

Berg, N., Berlin, M., Ruddel, B., Schutz, A., andWarfvingi, K. Possible branchocarcinogenic effects ofwelding and thermal spray fumes containing chromium.Abstract of presentation at International Conference ofHealth Hazards and Biological Effects of WeldingFumes and Gases. Copenhagen, 18-21 February 1985.

Bragt, P., and E. Van Dura. Toxicokinetics of hexava-lent chromium in the rat after intratracheal administra-tion of chromates of different solubilities. Ann. Occup.Hyg. 27:315-322; 1983.

Bryson, W., and C. Goodall. Differential toxicity andclearance kinetics of chromium (III) or (VI) in mice.Carcinogenesis. 4:1535-1539; 1983.

Hansen, K., and Stern, R. In vitro toxicity and trans-formation potency of nickel compounts. Environ. HealthPerspect. 51:223-226; 1983.

— Chromium and nickel compounds and stainless steelwelding fumes in cell transformation assays in vitro.

43

Abstract of presentation at International conference onHealth Hazards and Biological Effects of WeldingFumes and Gases. Copenhagen, 18-21 February 1985.

Hicks, R., Lam, H., Al-Shamma, K., and Hewitt, P.Pneumoconiotic effects of welding-fume particles frommild and stainless steel deposited in the lung of the rat.Archieves of Toxicology. 55:1-10; 1984.

Hooftman, R., Roza, P., and Arkesteyn, C. In vitrocytotoxicity testing of welding fume particles withbovine alveolar macrophages. Abstract of presentationat International Conference on Health Hazards and Bio-logical Effects of Welding Fumes and Gases. Copen-hagen, 18-21 February 1985.

Husgafvel-Pursiainen, K., Kalliomaki, P.-L., and Sorsa,M. A chromosome study among stainless steel welders.Occup. Med. 24:762-765; 1982.

— Chromosome aberrations and sister chromatid ex-changes among MMA-weldings of stainless steel. In:Prevention of Occupational Cancer—InternationalSymposium, Geneva: International Labour Office; 1983:p. 488-490.

Kalliomaki, P.-L., Junttila, M.-L., Kalliomaki, K.,Lakomaa, E.-L., and Kivela, R. Comparison of theretention and clearance of different welding fumes in ratlungs. Am. Ind. Hyg. Assoc. J. 44:733-738; 1983 a.

— Comparison of the behavior of stainless and mild steelmanual metal arc welding fumes in rat lung. Scand. J.Work Environ. Health. 9:176-180; 1983b.

Kalliomaki, P.-L., Kiilunen, M., Vaaranen, V.,Lakomaa, E.-L., Kalliomaki, K., and Kivela, R. Reten-tion of stainless steel manual metal arc welding fumesin rats. J. Toxicol. Environ. Health. 10:223-232; 1982.

Kalliomaki, P.-L., Lakomaa, E., Kalliomaki, K.,Kiilumen, M., Kivela, R., and Vaaranen, V. Stainlesssteel manual metal arc welding fumes in rates. Br. J. Ind.Med. 40:229-234; 1983.

Kalliomaki, P.-L., Tuomisaari, M., Lakomaa, E.-L.,Kalliomaki, K., and Kivela, R. Retention of and clear-ance of stainless steel sheildgas welding fumes in ratlungs. Am. Ind. Hyg. Assoc. J. 44:649-654; 1983d.