HEALTH EFFECTS OF WELDING AND CUTTING FUME - AN UPDATE

Gary M. Liss, MD, MS, FRCPC
Ontario Ministry of Labour
Final Report December 1996

SUMMARY

In 1985, this author prepared a document on the Health Effects of Welding and Cutting Fume, and in 1995, was asked by the Occupational Disease Panel to update the document. As in the initial review, this report is limited to effects associated with the gaseous and particulate components of the fume. For each outcome, recent findings are summarized and consistency was sought with the evidence noted in the earlier report.

Welding is an important occupational activity, in part because from 0.2 to 2.0% of the working population in industrialized countries has been reported to be engaged in welding. Although there are a vast number of processes, it has been estimated that shielded metal arc welding (SMAW, also known as manual metal arc welding, MMAW) and gas metal arc welding (GMAW or metal inert gas welding) applied to mild steel (MS), stainless steel (SS) and aluminum account for combinations practised by 70% of welders. The welding environment is very complex, with the presence of numerous gaseous and particulate components. The consumable is the major source of fume, although exposures extraneous to the process itself can be generated, such as decomposition products from chlorinated hydrocarbons or metal coatings. As a result of the multiple exposures, in general it is not possible to attribute health effects to specific fume contaminants or welding processes.

Welders continue to suffer both acute and chronic health problems that appear to be associated with work. With respect to acute effects, as found prior to 1985, acute intoxications continue to be observed, consisting mostly of case reports of lead intoxication, and upper and lower respiratory tract inhalation (bronchitis, pneumonitis) which may be process- or metal-dependent (cadmium, decomposed chlorinated hydrocarbons, coatings on metal). Severe cases of pneumonitis from inhalation of welding fume, such as that due to cadmium, may cause permanent sequelae or fatalities. The reports of increased mortality among welders due to pneumonia documented in the early time period have also been confirmed since 1985, and prompted one group of investigators (Coggon et al, 1994) to conclude that "there are strong grounds for the classification of lobar pneumonia as an occupational disease in welders". Metal fume fever (MFF) is an acute febrile illness of short duration due to inhalation of freshly generated fume that affects important proportions of welders (up to 30% in some estimates) and results from the inhalation of freshly formed oxides, most commonly that of zinc. The cause is not known but the syndrome resolves leaving no apparent chronic disorder. A new finding which was reported by Blanc et al (1993) suggested that various cytokines released from pulmonary cells may be involved in the pathogenesis of MFF.

Compared to the knowledge base in 1985, there is now evidence for acute (short-term) changes in pulmonary function among welders, that appear to be related to exposure (that is, present in groups with higher exposure or the absence of ventilation systems). A case of asthma was recently described by Vandenplas et al (1995) associated with GMAW on MS, which was confirmed on inhalation challenge with both late and dual reactions. Hendrick (1996) reported (in abstract form) that there was increased nonspecific methacholine reactivity among welders. Occupational asthma was also reported more frequently among welders than among the working population in general, in the Surveillance of Work-Related and Occupational Respiratory Diseases (SWORD) scheme in the United Kingdom (Meredith et al, 1991). The incidence of asthma among welders in one prospective study (Wang et al, 1994) was similar among MS vs SS welders. However, because the incidence of asthma observed in this study was not much different than that estimated in the general population, the evidence for an association with asthma cannot be considered conclusive.

For the studies of non-malignant respiratory diseases (symptoms, pulmonary function) and lung cancer, the quality of the studies was rated independently by two reviewers. With respect to chronic health effects, given the respirable nature of welding fume, chronic non-malignant respiratory disease continues to be an important focus of attention. Fourteen new studies were identified. Longitudinal studies have now been reported (as recommended in 1985); thus, the available studies are not all cross-sectional with the inherent weaknesses of this study design although the presence of confounders (especially smoking) still plagues the interpretation of these studies. Some but not all studies continue to demonstrate an increased prevalence of symptoms (chronic bronchitis as well as other symptoms) among welders compared to referents. While non-smokers were affected in some studies, there was evidence for an interaction with smoking in those reports where examined, with greater effects seen among smokers. Welders developed some respiratory symptoms (although not chronic bronchitis) more frequently than controls in 2 of 3 longitudinal studies. One study (Groth, 1989) found dose-response relationships for pulmonary symptoms among both smokers and non-smokers. Similarly, there is evidence from some but not all of the recent studies for enhanced deterioration of pulmonary function among welders compared to referents, involving the small airways somewhat more frequently than the large airways, and again with greater effects observed among smokers in some studies. These changes were not seen consistently across studies, however. In one study, obstructive changes were seen more frequently among older smoking welders than controls but restrictive changes were more prevalent among non-smoking welders. Enhanced deterioration of some pulmonary function measures was seen in all 3 longitudinal studies. Summarizing this issue, Wanders et al (1992) examined "medical wastage" among shipyard welders and controls (shipwrights and engine fitters at the same shipyard) as part of a 40-year cohort study. Medical wastage was defined as leaving for medical reasons under the Disability Act, invalidity pensions, death, and voluntary discharge for medical reasons. The only diagnostic category of medical wastage that was significantly different between the two groups was for non-malignant respiratory disease (greater among welders), with an adjusted incidence density ratio of 4.2 (95% confidence interval [CI] 2.4-7.4). The authors concluded that it underscores the "need to reduce the large excess of respiratory diseases among shipyard welders". Overall, the evidence is somewhat better than that existing a decade ago.

As was found prior to 1985, there continue to be case reports documenting pulmonary fibrosis among welders, but these have now been shown to occur in the absence of silicosis. However, there are still no studies of the incidence of fibrosis among welders. Siderosis, a pneumoconiosis due to pulmonary deposition of iron particles, appears as nodular radiodensities on chest x-ray. Siderosis itself is usually considered to be benign, without fibrosis, and if fibrosis is present, is considered to be due to exposure to other dusts, such as crystalline silica. However, one epidemiological study by Funahashi et al (1988) involved histological examination of lung tissue obtained from 10 symptomatic welders with abnormal radiographs. They found evidence of interstitial pulmonary fibrosis that was moderate to pronounced in five and this did not appear to be related to co-existing silicosis.

With respect to nonrespiratory chronic effects, an increased prevalence of neuropsychiatric symptoms has been reported among welders, but confirmation with objective neuropsychological tests is required. Most of the new studies of renal function continue to provide little evidence for renal function abnormalities among welders. However, a recent report by Nuyts et al (1995) from Belgium found an increased risk of chronic renal failure associated with exposure to welding fumes (odds ratio (OR) 2.1, 95% CI 1.1-4.0), but a dose-response relationship was not demonstrated. Some but not most studies of semen quality found a deterioration among welders. However, the studies suffer from weaknesses in methodology including low response rate and possible recall bias. There is limited evidence that the risk of Wilms' tumour is increased among the offspring of welders.

The recent in vitro studies continue to show that SS fume (particularly MMA/SS) is mutagenic and the activity correlated with the hexavalent chromium content, while MS fume demonstrated much less activity. New findings that were not found in 1985 are provided by the cytogenicity studies examining lymphocytes of welders. These studies continue to show no increases in sister chromatid exchanges among welders; however, several reports demonstrated genotoxic effects on lymphocytes of MMA/SS welders (who are most intensely exposed to chromium), consisting of chromosomal aberrations and DNA-protein cross-linking. The implications of these markers of genotoxicity for disease is uncertain at present.

With respect to cancer, associations with several types of non-respiratory cancer have been reported but there is little evidence that these findings are related to time since first exposure or to duration of employment, and will need to be confirmed.

Given the respirable nature of the fume, and the presence in welding fume of metals known or suspected to be carcinogenic in other settings, attention has naturally been drawn to assessing the possible risk of lung cancer in welders. There is no experimental evidence documenting the carcinogenicity of welding fume. Three possible explanations for the excess of lung cancer among welders discussed in the past included that it may be due to long-term MS welding of any type; that MS welding may involve little or no risk but certain welding processes, in particular SS welding, constitute a "hot spot" with a higher risk of lung cancer; or finally, that the increased risk may be due to the confounding effects of smoking, and possibly other carcinogens such as asbestos. Most of the epidemiological studies reported during 1985 or after, consistent with those reported previously, continue to document a small to moderate (about 30-40%) increase in lung cancer risk among welders. However, the findings are not consistent. The large (European) study conducted by the International Agency for Research on Cancer (IARC; Simonato et al, 1991) showed an overall increase in lung cancer mortality of 34%. However, a moderately large, well-conducted study of U.S. nonshipyard MS welders by Steenland et al (1991) found no excess (SMR 107 among welders, with SMR 117 among the comparison group of nonwelders), no trend with duration of employment, and rate ratios less than 1 on internal comparisons (0.90, 95% CI 0.60-1.36 for welders overall; 0.66 (0.33-1.30) for welding 10-20 years; and 0.65 (0.29-1.42) for welding greater than 20 years). This study was ranked as the best in quality, and was probably the one with the least potential for exposure of subjects to asbestos.

The studies of SS welders have been limited by small population size, and other than one very small study from Sweden by Sjogren et al (1987), have not shown increases in lung cancer mortality that are greater than those among MS welders in the same cohorts. Lung cancer risk among SS welders in the IARC study was not related to cumulative exposure. Thus, although it is known that there is exposure to hexavalent chromium in this group, that welders absorb hexavalent chromium, that SS welding fume is mutagenic, and that this exposure is documented to cause increased risks for lung cancer in other industries, the evidence for SS welding to date as a cause of lung cancer is unclear. Future reports may clarify this.

With respect to confounding exposures, there is evidence from a number of studies that welders are exposed to asbestos based on finding mesotheliomas in these cohorts (especially at shipyards). Of the studies providing data on smoking prevalence, there is evidence that welders smoked more than the general population in the U.S., Sweden, France, and Norway, and smoked more than internal controls in a German study. However, these differences in smoking prevalence are unlikely to explain more than a 20% increase in risk.

IARC (1990) concluded that there is inadequate evidence for the carcinogenicity of welding fumes and gases in animals and limited evidence in humans, with a classification of Group 2B (possibly carcinogenic to humans), which means "a positive association was found for which a causal interpretation is considered to be credible, but chance, bias or confounding could not be ruled out with reasonable confidence". Given the evidence for confounding exposures, this would appear to be the case. In a recent case-control study by Jöckel et al (1994), the OR for lung cancer among welders was reduced (from 1.5 to 1.2) when adjusted for asbestos. In order to clarify this issue, case-control studies within the existing cohort studies should be undertaken, in order to examine the association of lung cancer with welding, including detailed exposure regarding MS vs SS, while controlling for smoking and asbestos.

Taken as a whole, there continues to be evidence that exposure to welding fumes and gases that existed under past and perhaps current conditions is associated with certain acute and chronic health problems among welders including relatively common conditions such as pneumonia, metal fume fever, non-malignant respiratory disease, and possibly lung cancer and other conditions. A number of new outcomes have been observed which had not been described a decade ago.

DEFINITIONS OF SOME TERMS IN THE REPORT

Welding is a process for joining metals rendered liquid by heating, by pressure, or both. Fusion welding, which includes gas and electric arc welding, involves heating the metals to a suitable temperature. In arc welding, the metal temperature is around 4000 when workpieces fuse together, although the arc temperature is much higher. The various welding processes will not be described here. However, Stern (1983a) has pointed out that two major technologies (shielded metal arc welding using stick electrodes and gas metal arc welding using continuous wire) applied to mild steel, stainless steel and aluminum account for combinations practised by 70% of welders.

Cutting or scarfing may involve cutting slab steel, removing gates and risers on castings, or gouging out defective metal in castings. Gates are the end of the runner in a mold where molten metal enters the casting or mold cavity. Risers are the reservoirs of molten metal provided to compensate for the contraction of the casting as it solidifies. In gouging, a strip of surface metal is removed but the metal is not cut through. Carbon arc cutting is an arc cutting process in which metals are severed by melting them with the heat of an arc between a carbon electrode and the base metal. In oxygas cutting, the metal is heated by a flame and a jet of pure oxygen is directed to the point of cutting, and moved along the line to be cut. Cutting differs from welding in that only one piece of metal may be involved and temperatures are somewhat lower than in welding.

Burning is the removal of surface materials, especially paint and is also improperly used to describe the process of cutting. Lead welding, often called "lead burning", is the fusion welding of lead. The heat required for lead welding is much lower than that for welding cast iron and steel.

Brazing can be defined as a technique for joining metals using a filler metal with melting point greater than 427 C (800 F) and is thus differentiated from soldering which uses filler metal or solder with a melting point less than 427 C. The processes differ from arc welding in that the filler metal is flame or contact melted rather than melted by an electric arc.

FEF_25-75 is the forced expiratory flow between the 25% and 75% of the forced vital capacity (FVC), also known as the maximum mid-expiratory flow (MMF). This is a small airway test, largely determined by the later, effort-independent part of the forced expiratory manoeuvre.

FEF₇₅ (also known as MEF_{25% FVC}) is the flow when 75% of FVC has been expired (that is, 25% of FVC left). Similarly, FEF₅₀or MEF_50%FVC is the flow when 50% of the FVC remains to be expired. These flow rates are also reduced by narrowing of the smaller lung airways.

Abbreviations for Processes

ABBREVIATIONS USED IN THE DOCUMENT

BACKGROUND

In 1985, this writer prepared a document "The Health Effects of Welding and Cutting Fume" (Liss, 1985). The Executive Summary of the document is reproduced here as Appendix 1. In March 1995, this author received a request from the Occupational Disease Panel (ODP) to update the Health Effects Document. Since it had been 10 years, it seemed timely to examine what had been published during the past decade. Preparation of this report commenced in May of this year and was completed in November 1995. This final version of December 1996 incorporates minor revisions to incorporate comments of the peer reviewers.

Terms of Reference

This review of the health effects of welding, as in the first document, is limited to the gaseous and particulate components of the fume. It should be noted that welders may also suffer an important burden of illness associated with exposure to physical agents including noise, burns, eye injuries including "arc eye"/ ultraviolet radiation, electric shock, as well as ergonomic and other musculoskeletal problems. These aspects have been reviewed elsewhere (Villaume et al. 1979; Tinkler and Graville, 1983; Zakhari et al 1983, and others) and are not considered here.

The overall approach taken is to examine the associations between health effects and welding and cutting fume exposure reported in the past 10 years. For each outcome, a brief summary is provided of what was known 10 years ago but the individual references from the earlier document are not reviewed, and then the recent findings are presented. Consistency is sought with the evidence noted in the earlier report. In essence, the recent time period is taken to be a separate or independent "experiment" to look for consistency with what was known previously.

ACKNOWLEDGEMENTS: Staff of the Occupational Disease Panel and Ms. Irene Rule obtained the references. Ms. Marlene Vaz helped in the preparation of the tables. Dr. Ron House, St. Michael's Hospital, University of Toronto, kindly provided an independent rating of the methodology of certain articles.

1.0 INTRODUCTION

Welding is an important activity worthy of being reviewed, in part because from 0.2 to 2.0% of the working population in industrialized countries has been reported to be engaged in welding (Stern, 1981). In the 1980s, it was estimated that up to 36,000 residents were employed as welders in Ontario (Hatch Associates, 1985). In the United States, it was estimated that more than 185,000 workers were employed as welders, brazers or thermal cutters in the US, and up to 700,000 workers carried out some welding during their work (NIOSH, 1988).

Although there are a large number of processes, a small group of these make up a large part of welding activity: the five processes, shielded metal arc welding, SMAW (also known as manual metal arc welding, MMA) and gas metal arc welding, GMAW (or metal inert gas, MIG) on mild steel (MS) and stainless steel (SS), and gas tungsten arc welding, GTAW, on aluminum make up 60-70% of all welding activity (Australian National Occupational Health and Safety Commission, 1990). Similarly, Stern (1983a) noted that SMAW using stick electrodes, and GMAW (using continuous wire) applied to MS, SS, and aluminum account for combinations practised by 70% of welders.

The welding environment is very complex, with the presence of numerous gaseous and particulate components (Figure 1). These constituents of welding fumes and gases can be classified into those intrinsic, and those outside or extraneous to the process (e.g. bystander exposure to asbestos) (Liss, 1985; IARC, 1990). Details of the processes, and brief reviews of the individual health effects of some of these specific contaminants (constituents of welding fume and gases) were included in the earlier review (Liss, 1985). No attempt is made to review these individual constituents here.

The consumable is the major source of fume. Fume, which makes up most of the particulate component generated during welding, is generated by volatilization of melted substances with subsequent condensation of solid particles from the gaseous state. Examples of the range of likely exposures in different processes were summarized in Liss (1985), and more recently by IARC (1990) and van der Wal (1988); the reader is referred to those sources. Morgan (1989) also commented on the multiplicity of processes, as well as how emissions and types of fume vary. In general as a result of the multiplicity of exposures, one cannot attribute health effects to specific components of the fume or to individual welding processes.

Examples of processes and process-dependent exposures:

Exposures to various other compounds can occasionally occur with the use of specific consumables or from surface coatings extraneous to the welding process.

Approach taken: For the most part the focus is on reports in humans; reports in animals are included where it adds or human data are lacking. We review acute effects first and then chronic effects. For the discussion of respiratory diseases, it should be emphasized that there is a high likelihood of welding fume reaching the alveoli (Morgan, 1989) due to its respirable size range.

2.0 METHODS

The following sources were used to identify as many relevant studies and reviews as possible:

1. In March 1995, a literature search of the Medline and NIOSHTIC data bases was conducted for this writer by the ODP, with medical subject headings "welding" and "health effects". All abstracts were reviewed and those papers which appeared relevant were retrieved, if not already in personal files.
2. Personal files
3. Recent review articles including Morgan (1989), IARC (1990), and Sferlazza and Beckett (1991).
4. References from these articles were obtained if considered relevant.

Rating Quality of Studies

The quality of the cohort studies of lung cancer and the morbidity studies of chronic non-malignant respiratory disease (symptoms, lung function) was rated independently by this writer and a second reviewer. For the studies of chronic respiratory disease (which were predominantly cross-sectional in design), we used a rating scheme (shown in Appendix 2) adapted from one previously used in reviewing carpal tunnel syndrome (Stock, 1991) and Dupuytren's Contracture (Liss, 1993). Similarly, for the cohort studies of lung cancer and welding published since 1985, a rating instrument was developed (Appendix 3) adapted from the Ontario Advisory Council (1983). Differences in rating were resolved by consensus.

3.0 ACUTE EFFECTS

3.1 Acute Bronchitis, Pneumonitis and Intoxications

1985 Review: Prior to 1985, numerous cases of fatal and non-fatal pneumonitis in welders had been described, associated with exposure to ozone, oxides of nitrogen, cadmium, lead, polytetrafluoroethylene, and phosgene (due to decomposition products generated from chlorinated hydrocarbons). Many of these incidents were due to welding in confined spaces, or welding and cutting on metal with unknown coatings or paints.

Recent findings: Unfortunately, case reports continued to be reported since 1985. The reports included acute pulmonary reactions associated with welding-induced decomposition of trichloroethylene (Sjogren et al, 1991); welding copper pipe using silver brazing solder containing cadmium (Inoue et al, 1994); a plate welder working with oxyacetylene torch on cadmium alloy (Yates and Goldman, 1990); and fever and symptoms after welding on steel painted with a polymer lacquer (Sjogren et al, 1991a). In this latter report, the paints contained epoxydized vegetable oils hardened by hexachloro-endomethylenetetrahydrophthalic acid anhydride (HETacid anhydride). Some of the decomposition products are chlorinated, and possibly hexachlorocyclopentadiene may be generated from HETacid. This highlights once more the need for welders to be aware of the materials used to clean or coat the base metal as well as the composition of consumable and base metal themselves.

Other acute intoxications: Lead can be considered to have both acute and chronic effects. Reports of acute lead intoxication continued to be published (Holness and Nethercott, 1988; Marino et al, 1989). In addition, lead exposure associated with cutting lead-painted plates with oxyacetylene torches during the refitting of ships was described by Landrigan and Straub (1985).

3.2 Pneumonia

1985 Review: There were reports of increased deaths from pneumonia among welders, that were not observed in their wives or compared to a referent group. Increased morbidity from pneumonia among welders, however, was not observed in two studies where it was examined.

Recent findings: In previous analyses of occupational mortality from England and Wales, the Registrar General found that welders had an excess of pneumonia (see Liss, 1985) with standardized mortality ratios (SMRs) of 184 (95% Confidence Interval [CI] 150-224) in 1959-63, and 157 (95% CI 121-200) in 1970-72. Coggon et al (1994) extended these findings by analyzing data for 1979-80 and 1982-90; they found that the mortality risk from pneumonia was attributable mainly to an excess of pneumococcal and unspecified lobar pneumonia (proportional mortality ratio 255, 95% CI 192-332). No excess was seen in men above age 65.

Because increases were also seen among men in other jobs that entail exposure to metal fume or heated metal (e.g. moulders and coremakers), the authors suggested that metal fume might be the cause. Alternately, the gases such as oxides of nitrogen and ozone, which increase susceptibility to bacterial infections in animals could be involved (e.g. Copestake, 1987). Newhouse et al (1985) found increased deaths from pneumonia among welders at a shipyard in England (10 vs 5.4 expected), as did Beaumont and Weiss (1980) (SMR 1.67; observed 19). In a cross-sectional morbidity survey of British shipyard employees, a history of pneumonia was not associated with exposure to welding fume but a history of pleurisy was (Cotes et al, 1989). Coggon et al (1994) concluded that "there are strong grounds for the classification of lobar pneumonia as an occupational disease in welders". Although the consistent findings over time are certainly striking, one peer reviewer suggested that what is needed "to test this hypothesis further, is a well-designed case-control study nested in an industrial population in which pneumonia mortality can be easily ascertained".

3.3 Metal Fume Fever

As summarized in my 1985 document, metal fume fever (MFF) (also known as brass founders' ague or Monday morning fever), is a relatively common, acute febrile illness of short duration that may occur during and after welding operations. The condition is thought to be caused by freshly formed fume. The symptoms, which may simulate a flu-like illness, come on 4 to 8 hours after exposure, and include thirst, dry cough and throat, metallic taste in the mouth, chills, dyspnea, malaise, muscle aches, headaches, nausea, and fever and chills. The symptoms and signs resolve by the next morning. It is associated with many types of fume, most commonly zinc, copper and magnesium.

The mechanism is not known. In my 1985 report, I noted "specific features of MFF, namely, that it can be reproduced in experimental animals, that it can be experienced on the first day by a new employee (no latent period) and that large proportions of the welding workforce (about 30%) are affected, are not consistent with features suggestive of an allergic basis for disease. ...while MFF has manifestations resembling an immunologic disorder... Conceivably, the mediators leading to the manifestations of chills, fever and leukocytosis, could be released by a direct (toxic) pharmacologic interaction of human cells with some component in the welding fume and not necessarily by an allergic reaction. It is possible that the more recently recognized elements of the immune system, such as interleukins are involved, but evidence is lacking at this time" (Liss, 1985).

Recent findings: In fact, a recent investigation by Blanc et al (1993) suggested that various cytokines released from pulmonary cells, tumour necrosis factor, interleukin-6 (IL-6), and interleukin-8 (IL-8), may be involved in the pathogenesis of MFF.

German investigators (Vogelmeier et al, 1987) examined exposure during welding by a locksmith who had had MFF symptoms. During challenge, the blood zinc level was elevated but the cadmium level was not; the temperature and peripheral blood leukocyte count rose. The total cell count in bronchoalveolar lavage fluid was 10 times normal, while the differential count indicated a marked increased in polymorphonuclear leukocytes (PMNs). Pulmonary function tests showed a slight decline in airway resistance but a significant decrease in inspiratory vital capacity, transfer factor, and arterial oxygen partial pressure. The authors concluded that MFF induces a high-grade alveolitis. No pulmonary fibrosis has been observed following MFF but the authors considered that the question remains open regarding whether the marked increase in PMNs is really harmless to pulmonary tissue.

Noel and Ruthman (1988) demonstrated elevated serum zinc levels in two cases of MFF, while Langley (1991) described a welder who developed MFF and reactive airways dysfunction syndrome (RADS) after welding.

Recent animal report: Lam et al (1988) reported that although previous work had shown 1-3 hr exposures to zinc oxide (ZnO) at or below 5 mg/m³ had only minor transient changes on lung function parameters, guinea pigs exposed for 3 hr per day for 5 days to 7 mg/m³ of freshly generated respirable ZnO particles demonstrated decreases in flow resistance, lung compliance, and total lung capacity (which recovered in 48-72 hr). Vital capacity and functional residual capacity and DL_co were also depressed. This suggested that the recommended TLV for ZnO of 5 mg/m³ may be inadequate to protect workers.

3.4 Acute (Short-Term) Changes in Pulmonary Function

1985 Review: The main previous study of this subject (McMillan and Heath, 1979) found no significant differences between lung function changes over the day between welders and controls; of note, the observers appeared to be blinded to exposure status in that study.

Recent findings: Four new studies addressed this issue. Akbar-Khanzadeh (1993) obtained pulmonary function tests (FEV₁, FVC, and transfer factor for lung for carbon monoxide (TL)) before and after the shift in 209 welders and 109 nonwelder controls in England. The controls were selected from office workers, jointers, fitters, drivers, staging makers and electricians. It was not stated whether the observers were blinded to exposure status. The two groups had identical age, height and proportion of cigarette smokers and daily number of cigarettes. The welders did not wear respiratory protection. In 24.5% of welding samples, the concentration of welding fume exceeded 5 mg/m³ , and in 6.7% of samples the concentration of iron oxide exceeded 5 mg/m³.

There were significant decreases from morning to afternoon in the mean change in all three pulmonary function indices among both welders and controls but the reduction was approximately four times greater among welders. There were no statistically significant correlations between the diurnal variations in lung function and daily tobacco, fumes or gases sampled except for acute reduction of FEV₁ which was positively correlated with levels of iron oxide produced while welding (p<0.05). Welders who did not use any ventilation system showed maximum reductions in TL value while those who used combined ventilation systems had increased TL values in the afternoon. The overall difference in mean diurnal variations in TL between groups working under different conditions was significant (p<0.01). Those with reduced pulmonary function did not differ from those without reductions in terms of age, height, baseline lung function or smoking habits.

Marquat et al (1989) reported contradictory findings. They studied lung function measurements before and after the shift for 5 consecutive shifts among 11 welders of zinc-coated steel, 10 nonwelders who were indirectly exposed, and 17 controls. Exposures were lower than in the study by Akbar-Khanzadeh (1993) with geometric mean concentrations in the three groups of 0.91, 0.44 and 0.40 mg/m³ for dust, and 34.0, 19.1, and 4.1 g/m³ for zinc, respectively. No statement regarding blinding of observers was given. On average, lung function parameters were higher at the end of the work shift in keeping with normal diurnal variation. Differences in mean changes between exposure groups were small compared to the intragroup variation; there were no significant differences (p>0.1) between the three groups in mean changes in lung function over the week. There was no consistent direction or value of correlation coefficient between changes in lung function over a work shift and concentration of dust and zinc measured during the shift. Of 14 comparisons (7 lung function tests and 2 exposures measures), only the change in peak expiratory flow was significantly correlated with dust (r=-0.18, p<0.05).

As part of a larger study, Kilburn et al (1990) examined pulmonary function across a Monday workshift in 31 subjects (7 welded SS, 14 black steel, and 10 did not weld but were fitters and helpers). No indication of blinding to exposure group was stated. No exposure data were given but among the men who welded SS, levels of serum chromium (Cr) increased 66% and urinary Cr increased 22% over the shift. Pulmonary function changes were less than 2%, did not differ between groups, and were not statistically significant. However, the group sizes were limited.

Most recently, Donoghue et al (1994) examined 20 welders and 20 nonwelders (representing 95% of those invited to participate). All had essentially never smoked and had similar baseline age, height and spirometry. They were studied on Monday, with no welding since noon on the previous Saturday; there was no indication regarding whether the observers were blinded to exposure status. The peak expiratory flow (PEF) was measured before the start of work and 15 minutes, 30 minutes and 1, 2, 4, 7, and 12 hours after the start of welding. The mean change in PEF among welders at 15 minutes (a decline of 1.67%) was significantly different from that among nonwelders (an increase of 0.79%; p=0.028), and the mean for maximum decline recorded among the welders (-4.85%) was significantly greater than that in the nonwelders (-2.14%; p=0.011). None of the welders had PEF declines of 20% considered diagnostic of asthma. However, 10 (50%) of the 20 welders compared to only 1 (5%) of the 20 nonwelders experienced a 5% decline in PEF (p<0.005). The authors considered the results suggestive of an immediate type reaction in welders similar to that seen in occupational asthma (OA). Unfortunately, exposure was not measured. There was no local exhaust ventilation and low use of respiratory protection.

Summary: Two of the four recent reports did not find significant acute changes in lung function over the shift, in agreement with the early report of McMillan and Heath (1979); however one should note that these were small studies with low power and therefore inconclusive. One of the two studies that did find such changes (Akbar-Khanzadeh, 1993) documented what appeared to be the highest exposures, and effects were seen among those not using any ventilation system. This is consistent with the occurrence of acute pulmonary effects during welding, but that the effects may be dose- (or indirectly ventilation-) related. This would not be surprising given the known irritant effects of gases such as ozone and NO₂, as well as other constituents of welding fume.

3.5 Asthma

Vandenplas et al (1995) recently described a welder who developed asthma after 3 months of GMA welding on MS. He demonstrated both late and dual reactions after occupational-type challenge with welding on uncoated MS, suggesting a hypersensitivity rather than an irritant mechanism. Recently, Hendrick et al (1996) demonstrated (in abstract form), an increase in methacholine airway reactivity among shipyard welders and caulker burners in the U.K, compared with male school leaver controls, associated with exposure to welding fume, presumably through a nonspecific "irritant" mechanism.

Wang et al (1994) conducted a follow-up study of the welding work force at four factories in Sweden. They used company records to locate those who had been employed after January 1980, and who had at any time in the previous 10 years worked as welders; they also attempted to locate those who had left the company. The subjects were interviewed by questionnaire and the investigators established two cohorts based on data from the questionnaire: SS welders, defined as those who had been welding less than 10 years before the start of the study and who had welded for at least 6 months (welding time in SS > 50% of total time welding, exceeded 4 day/week and 4 hr/day for at least 6 months), and MS welders, selected with similar criteria. Both cohorts were studied for evidence of asthma, including ex-welders and active welders. There were 209 active welders (67 SS, 142 MS), and of 227 who had left welding, 209 (92%) of the ex-welders were located, and 187 responded (57 former SS, 130 former MS). A reference group of vehicle assembly workers who had never welded was also asked to complete the questionnaire. Bronchial responsiveness was determined among active welders by methacholine challenge. Subjects were considered to have asthma if they had two or more symptoms of excessive cough, wheeze, chest tightness or episodic dyspnea and the methacholine test showed bronchial hyper-responsiveness. Subjects with symptoms of asthma who did not perform this test were denoted as possible asthma. There was no difference in the incidence of welding-associated asthma (5% for SS, 7% for MS; 5 and 7 per 1000 welding-years, respectively). Bronchial responsiveness and lung function in active welders was normal, and did not differ between MS and SS welders. These estimated incidence rates of asthma can be compared with estimates of 2.1 per 1000 per year in a U.S. population (McWhorter et al, 1989), and rates of physician-diagnosed asthma of 4 per 1000 per year (Lebowitz, 1989).

Summary: In a review article, Sferlazza and Beckett (1991) commented that given the prevalence of asthma and the large number of regular welders, this suggests an infrequent occurrence of OA associated with welding. On the other hand, Meredith et al (1991) found OA was reported more commonly among welders than in the average working population. The recent prospective study by Wang et al (1994) provides little evidence of an increased incidence of asthma among welders compared to the general population, and gave no support for an increase, if any, being due to nickel and chromium in SS welding as suggested by the cases reported previously by Keskinen et al (1980).

4.0 CHRONIC RESPIRATORY DISEASES

4.1 In Vitro Cytotoxicity and Fibrogenicity of Welding Fume

1985 Review: Several studies, for example those by White and colleagues (1983) and Stern and Pigott (1983) showed that welding fumes from SS welding processes were particularly toxic. Cytotoxicity was noted to increase in the following order: fumes from MS rutile electrodes < MS basic < SS. Cr(VI) particles were toxic to macrophages; the toxicity of SS fume was attributed to Cr(VI) content, and MMA/SS fume was more toxic than GMA/SS fume.

Recent findings: Pasanen et al (1986) also examined the cytotoxic effects of fume on rat alveolar macrophages (AMs), and examined cell viability with trypan-blue dye exclusion. They found MMA/SS and MMA/MS to be cytotoxic, while MIG fumes of SS and MS had markedly smaller effects. Only the effects of MMA/SS were diminished when water-soluble components were extracted from the particles. Diluted solutions of potassium chromate similarly showed that hexavalent chromium may be responsible for the cytotoxicity of MMA/SS. They considered that the low cytotoxicity of MIG welding fumes was related to their chemical composition, which differed from MMA/SS (as the former contains mostly poorly water-soluble components).

Hooftman et al (1988) investigated whether toxicity was restricted to SS welding fume and estimated the contribution of chromium to the effects observed. They examined fume from MMA/SS, MIG/SS, MMA/CI (cast iron), MMA/MS, and MIG/MS using test criteria of viability (using trypan blue exclusion), phagocytic capacity, and morphology of the macrophages on bovine AMs. Consistent with the earlier reports, they found, for cytotoxicity, that the toxicity of the fume increased in the order: MIG/MS, MMA/MS, MMA/CI and MIG/SS, MMA/SS. They considered that MS fume (and MIG/MS particles in particular) are hardly toxic at all, having the same effects as glass beads. Soluble Cr(VI) appeared to be far more toxic than Cr(III). In terms of soluble chromium content, the results for the welding fume particles were comparable with those from chromium from K₂CrO_4. In other words, it seemed that the toxicity of most welding fume particles paralleled their content of soluble chromium. Fume from MIG/SS and MMA/CI were more toxic than one would expect based on this parallel, suggesting other factors play a part.

Stern et al (1988) also examined the in vitro cytotoxicity of welding fume and found somewhat similar results. They concluded that MMA/SS fumes are toxic (and genotoxic) apparently because of their soluble Cr(VI) content, and MIG/SS fumes are toxic at sufficiently high doses. The toxicities of MIG/Ni and MAG/SS which have insoluble fractions are within a factor of 2 of MS fumes; none of these fumes contain appreciable concentrations of Cr(VI).

Li and Yun (1988) compared the effects on collagen silica (95% quartz) and welding fume dust following intratracheal installation into lungs of rats. They examined collagen using ELISA staining and histological findings. Silica produced marked (significant) increased in both Types I and III collagen (about 8-9 fold) and increased the ratio of I/III from about 1.4 to 2. Welding fume produced a much smaller increase (less than doubling) and the I/III ratio stayed at about 1.4. They concluded that welding fume induced a slower and milder lung fibrosis compared to silica.

Finally, Ben Otmane et al (1992) examined ultrastructural changes on electron microscopy in guinea pig AMs and enzyme leaking (ß-glucuronidase activity) in guinea pig AM homogenate following cell culture with welding fume. They found that MMA/SS welding fume had relatively high cytotoxicity confirmed by an increase in culture medium -glucuronidase activity which they felt might be due to a soluble Cr(VI) compound. MIG/Ni and MIG/SS showed similar ultrastructural changes but the cellular damage was much more prominent for MMA/SS.

Summary: Welding of stainless steel generates the most toxic fumes, and MMA/SS particles were far more toxic than MIG/SS particles. The observed cytotoxic effects are largely due to the soluble chromium content of the fume particles, particularly those from MMA/SS welding. Insoluble compounds may play a part. The authors concluded that since the fibrogenic power of particles is known to be related to their cytotoxicity to AMs, particles from MMA/SS welding should be regarded as potentially fibrogenic. This was a similar conclusion to that reached earlier by Stern and Pigott (1983). Since phagocytosis is affected, this could make the lungs more vulnerable to inhaled particles, including microorganisms (perhaps this could explain the findings of increased pneumonia).

4.2 Chronic Respiratory Morbidity

1985 Report: Although not all studies were of very good quality, the investigations available at that time were consistent in suggesting an increased prevalence of one or more chronic respiratory symptoms (dyspnea, cough, sputum, wheezing, chronic bronchitis) in welders compared to referents. Of 13 studies, 9 revealed such evidence for one or more symptoms that was "statistically significant" and in 7 of these, seven reported data separately for non-smokers, and 4 showed significant excesses in non-smokers for one or more symptoms, with prevalence rate ratios of 2 to 4. Few studies examined the evidence for an effect due to duration of welding, and most did not use multivariate analyses. For pulmonary function, keeping the limitations of cross-sectional studies in mind, the differences in lung function seemed fairly consistent, with decreases in FEV₁ and measures of small airway function (MEFR, FEF₅₀, FEF₇₅), in a number of studies. Fewer studies documented such decreases in non-smokers. The evidence was limited for an association with duration of exposure. Kilburn (1986) suggested an interaction between smoking and welding fume but this needed to be confirmed.

Recent findings:

The results are presented separately for (I) symptoms (focusing mostly on chronic bronchitis, as a possible marker of chronic lung disease), and (ii) pulmonary function.

Assessment of validity of studies: All of the studies identified from the literature search were assessed independently by two raters using a validity assessment questionnaire (Appendix 2), adapted from Stock (1991). It included the seven criteria considered by Stock (1991): absence of selection bias, absence of nonrespondent bias, comparability of study and control groups, accounting for confounders, validity of exposure measures, validity of outcome measures, and blinding of assessors. Particular note was made as to whether the findings were reported separately for non-smokers. The interobserver agreement of the ranking was calculated using the kappa statistic (Sackett et al, 1985). Differences were resolved by consensus. A separate criterion, for study design, was added subsequently (scored 1 for cross-sectional only, 3 if a longitudinal component was included).

Studies Identified

Fourteen studies were identified; Table 1 shows the main features of these studies. Two studies (Kilburn et al, 1989; Hjortsberg et al, 1992) did not report symptoms. Three studies had longitudinal components: Mur et al (1989) a follow-up of the 1985 study, Chinn et al (1990), and Chinn et al (1995). The findings of the validity assessment of the methodology of the studies are shown in Table 2. Table 3 presents the study results for chronic bronchitis, and Table 4 the study results for pulmonary function.

Validity Assessment (Table 2): The kappa for interobserver rating was 0.40, indicating fair to moderate agreement. The quality (ranking) did not vary widely among the studies, ranging from 13 to 17 out of 21, with median rank about 14. The greatest weakness was the lack of blinding of examiners. All studies except one (Chinn, 1990) had scores of 2 or more on the three criteria of comparability of groups, valid exposure measures, and valid outcome measures considered by one investigator (Stock, 1991) as the major threats to validity in such studies, and so none were excluded from further assessment. However two studies (Kilburn et al, 1990; Nielsen et al, 1993) did not present the findings stratified by smoking status.

Study Findings

4.2.1 Symptoms (Chronic Bronchitis; Table 3)

Studies without stratification for smoking status: Two studies (Kilburn et al, 1990; Nielsen et al, 1993) did not present their findings stratified by smoking status. In these studies, the prevalence of chronic bronchitis among welders was 3 to 4 times that in the controls in these studies. A third report (Marquat et al, 1989) stated there was no difference in symptoms but no data were provided.

Nonsmokers:

Of the other 8 studies, one found a slightly (but not significantly) lower prevalence of chronic bronchitis among welders (Mur et al, 1985) (Table 3). Two studies found little or no difference (Sjogren et al, 1985; Mur et al, 1989), and in a third, Chinn et al (1995), in their cross-sectional analysis using multivariate techniques, found that among smokers and non-smokers combined, trade as a welder or caulker/burner was not a significant predictor for chronic bronchitis but that smoking and trade interacted significantly, with increased symptoms among smokers.

On the other hand, the prevalence of chronic bronchitis was reported to be higher among welders in the other four studies: two-fold higher by Zober (1985), 16% vs 0% (p<0.05; OR not calculable) by Lyngenbo, three-fold higher (including ex-smokers) by Cotes (1989), and four to five-fold higher by Groth (1989; p<0.01). Groth demonstrated evidence of a dose-response relationship with level of welding exposure (Table 3).

Other symptoms: At the initial assessment, Chinn et al (1995) found that trade as welder or caulker/burner (W/CB) among nonsmokers was related to the prevalence of wheeze, current cough, cough > 3 months, and a history of producing phlegm for > 3 months. Cotes (1989) found wheeze significantly increased among non-smokers (OR 2.7).

Smokers:

Of the 8 studies available, the prevalence of chronic bronchitis was less frequent (but not significantly so) among welders in the follow-up study by Mur (1989) and among heavier smokers in the initial report by Mur (1985). There was no difference found in chronic bronchitis prevalence among smoking welders and controls by Zober (1985) and Chinn (1995, at cross-sectional analysis), found that trade was "not significant" in the logistic regression for cough and phlegm for > 3 months among smokers and non-smokers combined.

On the other hand, the prevalence of chronic bronchitis was increased but not significantly, by Sjogren (1985; OR 1.7) and among light smokers by Mur (1985; OR 1.9). The prevalence among smoking welders was significantly increased in three studies: Cotes (OR 2.3), and Groth (about 2-fold in each smoking group). Among smokers alone, Chinn (1995) found that the prevalence of cough and phlegm for > 3 months was related to trade (OR 2.8). There was some evidence of dose-response given by Groth (1989) in each of the smoking groups.

Other symptoms: Chinn et al (1995) found trade to be a significant predictor among smokers alone for cough > 3 months (OR 3.5), phlegm > 3 months (OR 2.2), and wheeze (1.8), and among smokers and nonsmokers combined for phlegm (OR 2.9), and wheeze (OR 2.3), while controlling for smoking.

Longitudinal analyses: Mur (1989) found that the evolution over 5 years of respiratory symptoms (chronic bronchitis, breathlessness, and asthma) was not significantly different between welders and controls. Chinn (1990) found that during follow-up, an increase in the grade of breathlessness (OR 2.8) but not chronic bronchitis or wheeze to be significantly related to trade as a welder but this analysis was limited to those subjects seen at follow-up, only a small proportion of whom were still active. These analyses were adjusted for smoking, but smoking status-specific results were not presented. After controlling for smoking, Chinn (1995) found that trade as a W/CB was significantly associated with the development of chronic phlegm during follow-up (OR 2.8) but was not a significant predictor of the development of chronic cough, chronic bronchitis or wheeze.

Evidence for interaction: Cotes et al (1989) suggested that there was evidence for an interaction between smoking and exposure to fumes as a W/CB in causing chronic bronchitis and other symptoms. Chinn et al (1995) (cross-sectional analysis) found a significant interaction between smoking and trade for cough and for chronic bronchitis.

4.2.2 Pulmonary function (Table 4)

4.2.2.1 Large airways

No differences were found in six studies: Sjogren (1985; in FEV₁ and FVC); Mur (1985; and in 1989 follow-up), Kilburn (1989), Hjortsberg (1992) (only non-smoking welders studied), and Nielsen (1993) (not stratified for smoking). In a seventh study, Marquat (1989) found that the regression coefficients for welders vs controls for tests of large airways were negative but not significant.

Kilburn (1990) found only small non-significant decreases among nonsmokers but significant declines among smokers of about 5%. Lyngengbo (1988) studied nonsmokers only and found significant declines in FVC and FEV₁. In the cross-sectional analysis (initial assessment), Chinn (1995) found that trade as a W/CB was associated with an increase in expiratory reserve volume, and functional residual capacity. At follow-up, analyzed cross-sectionally, these investigators found that the lung function of referents (other trades) was superior to those of welders and caulker/burners who remained. However, at the initial assessment for the same workers, the lung function (e.g. FEV₁) of welders who were to remain (working) did not differ from other subjects, suggesting that the decline was associated with work but the role of selection bias is not clear, given the small proportion followed up.

Duration of exposure or exposure index: Marquat (1989) found that the regression coefficient for years of welding, after adjusting for smoking and other factors, was negative and of borderline significance for FEV₁ and FEV₁/FVC. Cotes (1989) found that among ever smoking welders, the fume exposure index contributed to FEV₁ and FEV% (but not FVC).

Longitudinal analyses (Table 4): Mur (1989) did not find that the "evolution" of pulmonary function differed between groups. Chinn (1990) found that the annual decline in FEV₁ was related to trade as a welder, with that among a smoking welder about 3 times that of a non-smoking nonwelder. In welders but not others, atopic status contributed to the annual decline in lung function but not to the risk of developing respiratory symptoms. These investigators also found that for PEF and PEF, the annual decline was greater in men whose trade was a W/CB. Chinn et al (1995) presented data showing that trade as a W/CB, independently of whether or not a man had left the shipyard, was associated with adverse changes in some lung function tests including inspiratory vital capacity and residual volume. Smoking appeared to interact with shipyard trade to impair development of expiratory vital capacity. Continued shipyard work as a W/CB was associated with enhanced deterioration in FEV₁and in PEF (the latter confined to smokers - that is, there was a significant interaction). Chinn (1995) also found that the decline in FEV₁ was related to whether exhaust ventilation was used 100% of time.

Presence or Absence of Restrictive or Obstructive Patterns:

Restrictive Patterns (Table 4): Among nonsmokers, the prevalence of restriction was found to be increased by Zober, 5-fold (but not significant) (1985) and by Lyngenbo (1988) 8% vs 0% (also not significant). Among smokers, Zober (1985) found no significant difference in restrictive changes.

Obstructive Patterns: Among nonsmokers, obstruction was reported to be about 2 times more prevalent by Zober and Lyngengbo (both not significant). Among smokers, Zober found obstruction was less frequently present among smoking welders than smoking nonwelders at young ages (2% vs 14%), while at older ages, it was more frequent among welders (8% vs 0%, p<0.05).

4.2.2.2 Small airways

Results not stratified for smoking status: One study (Nielsen et al, 1993) did not present results stratified by smoking status. They found no difference in spirometry or in one small airway test (closing volume, CV), but another test, slope of the alveolar plateau (phase III) by single breath N₂ washout was increased significantly after methacholine (p=0.01) in welders compared to controls. In addition, long-term welders had significantly higher phase III readings than short-term welders and controls, thought mainly to apply to aluminum welders. Mur (1985) reported that MEF_50%, not stratified for smoking, was significantly lower for welders using manual vs semi-automatic processes (p<0.05).

Nonsmokers: No differences in small airways function were reported by Mur (1985) at the initial assessment but in the cross-sectional analysis of the follow-up study (Mur, 1989), there was evidence for decreased MEF_25%. Significant decreases were observed by Kilburn (1989) and Kilburn (1990) for FEF_75-85, and Lyngenbo for an increase in the slope of the alveolar plateau. Hjortsberg et al (1992) reported larger volume of trapped gas (VTG), of borderline significance, among welders compared to referents and significant increases in VTG after methacholine among welders but not among referents.

Smokers: Mur (1985) found no differences at the initial assessment (1985) or at follow-up (1989) analyzed cross-sectionally. Kilburn (1989) and (1990) found significant declines among welders in FEF_25-75 and FEF_75-85 of 7-15% and 9-25%, respectively.

Longitudinal analyses: Mur (1989) found the decline in MEF_25% over time was significant only among welders. Chinn (1990) found that for various spirometric indices (MEF_50%, MEF_50%) after allowing for age and smoking, the annual decline was greater in men whose trade had ever been W/CB compared with trades that did not involve these tasks. Chinn (1995) found that over follow-up, after adjusting for smoking, continued trade as a W/CB was associated with enhanced deterioration in FEF_75-85.

Diffusing capacity: Among nonsmokers, diffusing capacity was normal among welders in most studies when examined, although Lyngengbo (1988) found it to be decreased. In smokers, it was reported to be decreased by Mur (1985) and Mur (1989), but normal by Kilburn (1990). Chinn (1990) found that changes in diffusion over follow-up were not related to trade. At the cross-sectional analysis, Chinn (1995) reported that trade as a W/CB was associated with a decrease in transfer factor.

Evidence for interaction between welding and smoking: Mur (1985) found evidence of an interaction for lung transfer (greater in smokers) and found the effects of welding in nonsmokers and light smokers to be negligible. Kilburn (1989) suggested an interaction of welding with smoking in producing lung function decrements. Cotes (1989) also found that the effects of fume exposure were mainly confined to smokers and ex-smokers. Kilburn (1990) reported that nonsmokers showed smaller decreases in lung function than did smokers, consistent with a synergistic effect on airways of welding fume and tobacco. Chinn (1995) reported that smoking appeared to interact with shipyard trade to impair development of expiratory vital capacity, and that the decline in PEF (longitudinally) was limited to welders who smoked (significant T.Sm term). On other hand, as noted, Zober (1985) found the prevalence of restrictive lung function abnormalities was increased in nonsmoking but not smoking welders compared to their respective controls.

Summary:

Symptoms: There was improvement in study quality since 1985; some studies examined nonsmokers only. The data suggest an increase in the prevalence of chronic airways symptoms (including chronic bronchitis as a marker of severe disease as well as other symptoms), that was seen in some but not all studies. In the 11 cross-sectional studies, increases were seen in 2 studies not stratified for smoking (3-4 fold higher prevalence), and in both nonsmokers (4 of 8 studies) and smokers (significant increase in 3 of 8 studies and non-significant increase in 2 others). This is consistent with the earlier review (Liss, 1985). There was evidence of an interaction in some studies with the effect greater among smokers. The strength of the association is moderate (about 2-fold, in those studies demonstrating an effect), but effects were not seen consistently. Evidence for a dose-response relationship was provided in only one study, with increasing "degree of welding exposure" (a measure which included duration of welding) (Groth, 1989), and this held across smoking strata.

The publication since 1985 of studies with longitudinal design (as was recommended) also represents an improvement over the quality of studies available a decade ago, as one hoped this would clarify some of the limitations of cross-sectional studies. In these longitudinal studies, some symptoms such as phlegm and increase in breathlessness (but not chronic bronchitis or wheeze) were related to trade. However, as noted, these studies were limited by the small proportion available for examination or who were currently active at follow-up. For example, Chinn (1990) studied 488 (88%) of 609, but 425 of these had retired or been made redundant, and Chinn (1995) restudied only 346 (55%) of 625; hence, selection bias was possible.

Pulmonary function: As was found previously (Liss, 1985), the assessment of pulmonary function is limited in that most studies were cross-sectional in design. Impairment in large airways was observed in only a small minority of studies, and only two studies provided evidence supporting an association of pulmonary function with duration or index of exposure. Small airway function was affected somewhat more frequently than large airways (up to 7 studies in smokers, nonsmokers or both); nonsmokers were also affected but again, there was some evidence for an interaction with smoking. Zober (1985) found a higher prevalence of obstructive abnormalities in older smoking welders.

As with symptoms, longitudinal studies have now been conducted, and tend to show changes of enhanced deterioration in both large airways (in two studies) and small airways (in all three). However, caution is required in interpreting the data, in part because the participant drop-out rate has been too high to make definitive statements. The clinical importance of the pulmonary function changes reviewed above remains to be determined.

Are these findings causal?

Magnitude of effects: Chinn (1990) found that the proportion of variance explained by trade was on average half that due to smoking except when atopy was included in the analysis. For welders who smoked, the decline in FEV₁ was three times that of non-smoking nonwelders. Chinn (1995) found that the effects on pulmonary function were independent of and at least as large as those due to smoking. As noted, Kilburn (1989; 1990) found declines in small airways of 6-17% in nonsmokers and 15-25% in smokers. Cotes (1989) considered that the effect of fume, on average, was similar to that of being a smoker.

Consistency and temporality: Adverse effects were not seen consistently, that is, in some but not all studies (generally fewer than half). Temporality is not really addressed in the cross-sectional studies but is taken into account in the longitudinal studies. Welders had increased development of respiratory symptoms (but not chronic bronchitis) in two of the three longitudinal studies, and some markers of excess deterioration in pulmonary function in all three.

Dose-response: Groth (1989) presented evidence of dose-response relationships across smoking strata for symptoms but there is scant evidence available for an association with dose or duration of exposure for pulmonary function. On the other hand, several other findings add to the consistency of the adverse effects of welding fume and gases on pulmonary function and constitute indirect evidence for an exposure-response relationship. First, Mur (1985) reported that small airway function (MEF_25%) was poorer among welders working in confined compared to open spaces. Second, this study also found small and large airways tests were poorer among those performing manual vs semi-automatic welding processes (Table 4). Third, Chinn (1995) found that the excess decline in FEV₁ was less among those who reported using exhaust ventilation 100% of the time. An association between welding exposure and chronic obstructive pulmonary disease makes biologic sense, given that a number of fume and gas components are respiratory irritants. Occupational exposure to fumes or gases estimated by job exposure matrices was related to a decrease in FEV₁ in some analyses in the French Cooperative PAARC (Pollution Atmosphérique et Affections Respiratoires Chroniques) survey (Le Moual et al, 1995). Thus, only some of the criteria for causality are met.

Overall, the studies are of better quality than a decade ago, and the findings are somewhat more convincing for an association with non-malignant respiratory effects. However, many of the studies involved shipyard welders, and the applicability to Ontario welders is uncertain.

Consistent in general with these findings, Wanders et al (1992) recently examined "medical wastage" among 548 Dutch shipyard welders and 1409 controls (shipwrights and engine fitters) employed at the same shipyard as part of a 40-year cohort study. They contributed 5227 and 17,005 person-years of observation, respectively. Total medical wastage was defined as leaving for medical reasons including under the Work Disability Act, invalidity pension, death, selected employment and voluntary discharge for medical reasons. The only diagnostic category of medical wastage that was significantly different between the two groups was for non-malignant respiratory disease (greater among welders), with an incidence density ratio of 4.2 (90% CI 2.4-7.4), adjusted for age, calendar period and duration of employment as welder or control. The authors concluded that it underscores the "need to reduce the large excess of respiratory diseases among shipyard welders". A contrary view was recently provided by Morgan (1995) who concluded that McMillan and colleagues "studied all aspects of lung function and diffusing capacity and shown that welding and welder's siderosis was not associated with significant respiratory impairment, with the possible exception of a greater prevalence of airways obstruction among smoking welders".

Final note: For completeness, one could note criticisms of the reports by Kilburn (1989, 1990), raised recently by Jones et al (1995). The reference group used by Kilburn had been studied in the early 1980s, and involved a different group of subjects (from Michigan), collected by different investigators (Miller et al, 1986) for a different purpose. The study group was volunteer, which may have resulted in bias towards a worse health status, while the reference group was a random sample of a geographically defined population. The reference group was also screened to exclude those with common respiratory symptoms or diagnoses, angina or diagnosis of coronary artery disease. In response, Kilburn et al (1995) replied that "much rhetoric has been expended by other critics as well as Jones" with respect to selection bias, and emphasized that both groups were volunteers.

4.3 Siderosis and Fibrosis

1985 Report: Siderosis, due to prolonged pulmonary deposition of iron (oxide) particles appearing as nodular radiodensities on chest x-rays, is the most commonly occurring form of pneumoconiosis in welders (Palmer, 1983). Siderosis is usually classified among the benign pneumoconiosis (Parkes, 1982) as iron oxide is considered to be inert. Attfield and Ross (1978) reported the most detailed study of the relationship between the prevalence of small round opacities of class 0/1 or higher (larger than 1 mm "siderosis") and welding exposure based on chest x-rays from 661 British shipyard welders (see Liss, 1985). No cases of small round opacities 0/1 or greater appeared before 15 years exposure and the prevalence then increased with age or with length of exposure reaching over 30% among those with greater than 45 years' exposure.

Siderosis has generally been assumed to be benign and unassociated with respiratory symptoms (Morgan and Seaton, 1984) based on the absence of fibrosis both in animals administered iron oxide and in lungs of welders, and on the absence of pulmonary function abnormalities in welders with marked radiographic abnormalities (Morgan and Kerr, 1963). On the other hand, there are case reports of welders with dyspnea and respiratory symptoms, impairment of lung function, x-ray opacities and extensive fibrosis when biopsied (Liss, 1985). However, some of the cases worked in environments with silica as well as iron (suggesting mixed dust fibrosis). Stern et al (1983) reviewed the fibrogenic potential of welding fume, including the 90 or so published case reports which referred to irreversible pulmonary fibrosis in welders supported by tissue pathology at biopsy or autopsy. They suggested the possible role of NO₂ in fibrosis. In 1985, I concluded that "as there was no estimate for the incidence of pulmonary fibrosis in welders (or for that matter, in the general population), it is not possible to demonstrate if, in fact, an excess of pulmonary fibrosis in welders exists". Moreover, if welders face a possible fibrogenic hazard (Stern and Pigott, 1983), it has not been possible to establish a (the) fibrogenic agent by epidemiological methods.

Recent findings: Morgan (1989) noted, as have previous writers, that welders' siderosis is characterized by the deposition of iron particles in the lung parenchyma. Most of the iron particles have been ingested by AMs while some are seen lying free in the alveoli. He noted, referring to his early report (Morgan and Kerr, 1963), that the alveolar septa are not thickened and there is no alveolitis. He commented on the occasional reports of fibrosing alveolitis or chronic interstitial fibrosis and one case in particular in which he felt the subject had idiopathic pulmonary fibrosis and that the little iron in the lungs was not ferric oxide but hemosiderin (Morgan, 1989). More recently, Morgan (1995) commented that there was "little doubt that in its pure form, siderosis does not lead to pulmonary impairment".

Despite this comment, the epidemiologic studies (see above) continue to suggest increased symptom prevalence and some abnormalities of pulmonary function among groups of welders. What evidence is provided by the recent original reports regarding fibrosis?

Case Reports: Lasfargues et al (1991) described a 56 year old nonsmoking welder (arc welder for 30 years) who was noted to have siderosis (discovered from radiological abnormalities) and who developed fibrosis in the absence of any associated silicosis. Lung biopsy revealed patchy interstitial fibrosis. A case of chronic beryllium pneumonitis was reported from Scotland associated with exposure due to "spot welding" in an electrical engineering firm (Monie and Roberts, 1991). Glass et al (1994) described a 32-year old welder from New Zealand who had welded since age 17 repairing farm machinery, and manufacturing animal feeders and cages, and then had not welded for 4 years. He smoked five cigarettes per day. In August 1989, after he began (MIG) welding again, on galvanized steel frames, he suffered two relatively brief intense exposures to welding fumes each under relatively confined space conditions, indoors with poor ventilation. Three months after starting work he developed shortness of breath, could no longer participate in recreational running, and by February 1991, was no longer able to continue welding. Investigations revealed restrictive lung disease with severe reduction in diffusing capacity without airway obstruction or bronchial hyperresponsiveness. The results were indicative of chronic interstitial lung disease (CILD). The authors noted that to date, CILD had not been reported in association with galvanized welding or other zinc oxide exposures. Figueroa et al (1992) described two cases of hard metal interstitial pulmonary disease (one fatal) in workers employed in the same area of a metal coating plant using the detonation gun process for applying a durable metal surface to metal parts. In this welding, a mixture of powdered metals, including tungsten carbide and cobalt, was heated by ignition of a flammable gas and propelled from the end of the "gun" at high temperature and velocity to form a welded metal coating.

Epidemiological investigations: Kennedy et al (1985) attempted to determine if mineral dust or fume exposure in smokers resulted in additional changes in small airways structure and function. They studied 25 subjects exposed to dust or fume for more than 10 years selected from an ongoing study of pulmonary structure and function based on patients admitted to hospital for surgical resection of a lobe or lung because of lung cancer (13 miners exposed to dusts and 12 others exposed to other mineral dust or fumes, including 5 welders), and 25 control subjects admitted to the same study who reported no occupational exposure to dust or fume of any kind, matched for lobe resected, age ±3 years and smoking history. Membranous and respiratory bronchioles were graded for emphysema and the presence and degree of mural inflammation, fibrosis, smooth muscle hypertrophy, pigment, squamous metaplasia and goblet cell hypertrophy. There was no statement regarding whether observers were blinded to exposure. There was significantly greater fibrosis in the walls of membranous bronchioles and increased goblet cell metaplasia in the epithelium of the airways of the exposed group (p<0.01). There were no significant differences between the groups with respect to the respiratory bronchioles. When the exposed group was divided according to occupation into miners and nonmining exposed group and each subgroup compared with the corresponding control group, the results revealed that the excess fibrosis and goblet cell metaplasia was significant in both groups. Separate results for the welders were not presented. There were no differences in pulmonary function between the groups. The authors concluded that this supported the concept of airway fibrosis as a nonspecific reaction to mineral dust and fume; and that occupational exposure to mineral dust and fume results in a chronic inflammatory response (structural changes) in peripheral airways that are greater than those seen with smoking alone but that the changes were not associated with greater abnormalities in lung function.

Funahashi et al (1988) conducted histological examinations on lung tissue obtained from 10 symptomatic welders, ages 31 to 63 (mean 49) with duration of welding 8 to 40 years (mean 23), who had symptoms of cough and dyspnea, and abnormal radiographs (diffuse nodular or reticulonodular patterns). Pulmonary function tests showed restrictive impairment in seven, mild to moderate airways obstruction in two, and reduced diffusing capacity in three of nine (who also had restrictive impairment). Nine had open lung biopsies for diagnosis and in one case the tissue was obtained at autopsy. The tissue specimens were processed for tissue elemental analysis by energy dispersive x-ray analysis (EDXA); silicon/sulphur [Si/S] and iron/sulphur [Fe/S] ratios were compared with 10 age-matched controls and 10 cases of silicosis. The following table shows the main pathological findings, as interpreted by two pathologists.

The pathologists agreed within one grade; differences were resolved to the lower grade. All subjects had some degree of fibrosis and this was moderate to pronounced in five; alveolar wall thickening was moderate to pronounced in five. The elemental content in tissue showed that the mean Si/S ratio was no different between controls and welders whereas patients with silicosis had a significantly higher ratio than controls and welders (p<0.005). The mean Fe/S ratio was slightly but not significantly higher in patients with silicosis than in controls, while the Fe/S ratio was significantly higher in welders than in controls and silicotics (p<0.0025). No other elements were found to be raised in welders. Interstitial fibrosis is a reaction of lung tissue to various insults. They noted that the fact that many iron containing particles were seen in the fibrosed alveolar septa suggested that the observed fibrosis is a reaction to those particles (although this did not constitute proof). They felt that it was important for clinicians to recognize that "welders who have no exposure to silica or asbestos may develop interstitial fibrosis leading to respiratory symptoms and physiological abnormalities although it is uncommon". They concluded that (1) interstitial pulmonary fibrosis is seen in some welders; (2) the association of welders' pneumoconiosis and interstitial fibrosis appears to be more than coincidental; and (3) the cause of the fibrosis did not appear to be coexisting silicosis, as silica concentrations in the lung were within normal limits.

Summary: These recent investigations show that interstitial fibrosis occurs in welders (in one study, among those with radiological changes); that, unlike early reports from the 1960s, this does not seem to depend on the presence of silicosis; and that there are small airway (possibly early emphysematous) changes. It is not clear if the changes are related to iron or other constituents of welding fume and gases; the possible role of cytotoxicity of welding fume, especially SS fume, has been discussed above. As was the case in 1985, there are still no estimates of the incidence of interstitial fibrosis among welders to compare to that in the general population. It is possible, as noted by Billings and Howard (1993), that radiological siderosis could be a marker of serious welding fume exposure.

5.0 NONRESPIRATORY CHRONIC EFFECTS

5.1 Neuropsychiatric symptoms

Sjogren et al (1990) assessed neuropsychiatric symptoms in 65 welders exposed to aluminum (Al) and 217 railroad track welders in Sweden using a previously validated questionnaire (Q16). The questionnaire was initially sent to 308 welders, with a 92% response rate. Logistic regression was employed to examined the relation between exposure, age and occurrence of neuropsychiatric symptoms. Welders exposed to aluminum, lead or manganese for long periods had significantly more neuropsychiatric symptoms than welders not exposed to these metals; the authors suggested that detailed psychometric studies should be performed. It should be emphasized that the outcome here was subjective, and may reflect metal exposure in general, not necessarily features of the welding environment in particular. Sjogren and Elinder (1992) subsequently noted that long term Al welders have been shown to have increased urinary Al, and suggested, based on this study by Sjogren et al (1990) and the known effects of Al seen with dialysis encephalopathy at very high Al concentrations, that a dose-response relationship existed between Al welding fume exposure and effects on the central nervous system.

More recently, Hänninen et al (1994) examined 17 male Al welders from a ship building company in Finland engaged in welding for 2 to 27 (mean 15) years but engaged in MIG welding on Al for only the last four years. The authors conducted a series of neuropsychological tests, and assessed subjective symptoms and moods (symptom questionnaire and Profile of Mood States, POMS), serum and urine Al, and quantitative EEG (QEEG). The mean scores for the tests for psychomotor, visual and spatial abilities in the group fell within the "good-average range" and the means of memory tests and tests for verbal ability were within the average range of unpublished standardization data of the Finnish Institute of Occupational Health. However, there was a negative association between all four memory tests and urinary Al, and a positive association between the variability (standard deviation) of visual reaction times and exposure (based on serum Al). Low scores for the combined symptoms and mood scores were associated with a high serum Al (S-Al) (fatigue, memory, depression or tension or irritability). In the QEEG, the amount of delta and theta activity in the frontal region correlated positively and the amount of alpha activity in the frontal region correlated negatively with S-Al.

The authors noted that because of the small sample size, the statistical significance of the exposure-effect associations for the neuropsychological tests was modest and the "results are only suggestive". However the type of neuropsychological impairment is consistent with previous documentation seen in workers exposed to Al in other (nonwelding) settings (Hosovoski et al, 1990; White et al, 1992) which will not be reviewed here. The negative association between mood disturbance and S-Al was contrary to expectation and difficult to explain. EEG abnormalities have been found among patients with dialysis encephalopathy. The authors noted that as far as they were aware, no studies had been published concerning EEG findings from occupationally exposed subjects with normal renal function. In the present study, welders with higher levels of S-Al had more slow (delta and theta) activity and less alpha activity in the frontal areas compared with welders with lower levels of S-Al. They felt that these results were "principally consistent" with the findings from patients with Al encephalopathy and from animal experiments, and suggested that Al uptake from welding may have an effect on cerebral function. They noted that their study "does not allow any definitive conclusions" but that Al welding may involve hazards for the CNS, with disturbances of memory and attention possibly the most likely early effects.

A recent U.S. case-control study involving cases of amyotrophic lateral sclerosis and controls with other neuromuscular disorders and from the community (Strickland et al, 1996), found the strongest association with welding or soldering, OR 5.3 (95% CI 1.4-20.1). The authors wondered whether lead might be responsible.

Summary: The findings to date remain limited, in large part, to associations with neuropsychological symptoms. Further confirmation with objective neuropsychological tests is required. It is possible that any effects, if they exist, may be due to metals including Al or lead which are seen in other occupational (nonwelding) settings.

5.2 Renal Effects

1985 Review: The limited studies available at that time did not permit one to conclude that renal impairment occurred as a result of welding.

Recent findings: A case-control study of severe renal disease and glomerulonephritis cases and population-based referents matched by sex, age and community in Sweden (Hagberg et al, 1986) assessed exposure to welding fume by means of a questionnaire. There were no differences in exposure to welding between cases and referents, with rate ratios for all welding fume of 0.94; for welding fumes more than 100 days per year of 1.53 (95% CI 0.4-1.9), for SS welding of 0.83, and for welder given as occupation of 0.82.

Verschoor et al (1988) examined the influence of occupational exposure to water-soluble chromium (VI) on renal function in chrome platers, SS welders, boilermakers and a control group of workers in the Netherlands. The serum creatinine was significantly higher among the welders than the other groups, and the combined Cr(VI) group (chrome platers and welders) had some differences in glomerular function parameters compared with the Cr-dust or non-exposed group (boilermakers and controls) but the renal function parameters were not related to chromium concentration in the urine or to chromium clearance. The authors concluded that some evidence for alterations in glomerular function was observed in chrome-plating workers and welders which "may be caused by exposure to Cr(VI)" but that it is also possible that other characteristics such as a static workload may be causal factors. It was difficult to find relationships between exposure and renal function parameters.

Subsequently, Vyskocil et al (1992) examined biochemical markers of kidney damage in 52 male SS (MMA) welders exposed to chromium and nickel and matched referents. The prevalence of abnormal values in the welders did not differ from that in the controls and no consistent or clinically significant renal impairment was observed among the SS welders exposed to chromium air concentrations slightly above the current ACGIH TLV for water-soluble Cr(VI) of 50 µg/m³.

Very recently, however, a Belgian study (Nuyts et al, 1995) examined occupational risk factors for chronic renal failure (CRF). Cases of CRF were identified from five renal units interviewed within six months of the start of their dialysis therapy, and controls were people with normal renal function randomly selected from lists of voters in the same communities. A structured questionnaire was used to obtain information on lifestyle characteristics (smoking, drinking and medication intake), medical history, and occupational history. A significant increased risk of CRF was found for welding fumes (OR 2.06, 95% CI 1.05-4.04). However, there was no exposure trend suggestive of a dose-response relation.

Studies of welders and brazers (Elinder et al, 1985a,b) who have exposure to cadmium in solders have shown tubular damage with various markers such as -microglobulinuria, similar to what has been demonstrated with cadmium exposure in other settings.

Summary: Most of the new studies available provide little evidence for renal function abnormalities among welders as a group or SS welders exposed to water-soluble Cr. However, one recent study from Belgium found an increased risk of CRF for welding fume, although no trend with exposure was observed. Welders with cadmium exposure may have changes that had been observed for many years among groups with exposure to cadmium in other settings.

5.3 Skin Effects

Kanerva et al (1991) reported from Finland a repeated urticarial reaction associated with fever occurring in a welder while welding steel profiles filled with polyurethane. Provocation tests with two pyrolysis products (4,4'-diphenylmethane diisocyanate and 4,4'-diaminodiphenylmethane) were negative.

5.4 Reproductive Effects

5.4.1 Effects on Semen and Subfertility

1985 Review: Rachootin and Olsen (1983) compared infertile couples (cases) with fertile couples (referents) and obtained information on occupational exposures by postal questionnaires completed by the women (response rate 87%). They focused on exposures preceding hospital care for either infertility or childbirth. Welding on SS but not on other types of steel was associated with reduced sperm quality (deterioration in concentration, motility or morphology). However, in an internal analysis among the case group, there was no difference in welding exposure among cases of idiopathic origin and cases with urogenital disorders. Moreover, in an internal analysis among the referents, welding of other metals but not SS welding was associated with delayed conception. Thus, the results were not entirely consistent as an association was found between SS welding and reduced semen quality, but not with delay of conception, and conversely, between non-SS welding and delay of conception but not with reduced semen quality. This 1983 report was described by Bonde (1993) as an hypothesis-generating study. In animal studies, high doses of hexavalent chromium cause testicular atrophy and decreased sperm count.

Recent findings: Most of the recent reports have been published by one investigator (J.P. Bonde) from Denmark. Mortensen (1988) reported the findings of a case-referent study involving 3,119 men who had delivered semen sample(s) to one of three hospitals in Denmark in connection with fertility problems. A mail questionnaire was sent to the men with an 81% response rate. The sperm samples of the men were categorized according to sperm concentration, motility and appearance; if any of the criteria were abnormal, the sperm quality was considered to be poor and the man was classified as a "case". The questionnaire information was used to categorize the men as (i) welders; (ii) metal workers not exposed to welding; (iii) other industrial workers; and (iv) unexposed workers. After adjusting for confounders, there was an increased risk for reduced sperm quality among welders compared to unexposed workers (OR 2.0 (95% CI 1.2-3.4)) but not in the other groups. This was concentrated among those working on SS (OR 2.3).

On the other hand, Jelnes and Knudsen (1988) examined sperm quality among 145 men, representing 64% of the eligible workers. The subjects included 77 SS welders and 68 age- and smoking-matched referents, who were mostly but not exclusively from the same plants as the welders. There were no differences in semen quality parameters between the welders and nonwelders. This study was limited by selection bias (the low participation) and because hexavalent Cr exposures were lower than in other studies. Additional comparisons of a sample of 20 welders to external controls also did not show differences between the groups. The authors concluded that the data suggested that SS welding did not present a major risk of reduced semen quality.

In a cross-sectional study, Bonde (1990) examined semen quality in 35 SS welders using TIG, 46 MS welders using MMA and MAG, and 54 nonwelding metalworkers and electricians (participation rates of 37% in welders and nonwelders). They were asked to deliver three semen samples at monthly intervals, and these were provided by 27, 46 and 35 of the subjects, respectively, with mean values used for those providing multiple samples. The referents were selected from the same geographical, occupational and social setting as the welders, and were similar in age and smoking but differed in alcohol and bathing. There were no differences between MS or SS welders in sperm concentration, proportion with low concentrations (<20 million), or proportion with <50% normal morphology. The sperm count per ejaculate, the proportion with normal morphology, the degree of sperm motility, and the linear penetration rate of the sperm were significantly decreased in the MS welders. A dose-response relation across high and low MS exposure and referents was observed for these parameters (sperm count excepted), with semen quality decreasing and FSH concentrations increasing with increasing exposure. For SS welders, total sperm count per ejaculate, proportion of motile sperm, motility, and concentration of testosterone were also decreased. Selection bias was possible due to the low participation rate. Based on an initial questionnaire (with 80% response rate), previous knowledge of reduced semen quality and history of urogenital disorder was more prevalent among participating referents than non-participating referents but not among welders, suggesting that bias towards the null hypothesis was possible. The results concerning the MS welders with dose-response and consistency of sperm and FSH findings were felt to add to the reliability of the findings but those for SS welding were considered "more dubious".

Bonde and Ernst (1992) did not find that semen parameters deteriorated with increasing level of internal exposure to chromium among 30 TIG/SS, 30 MS welders and 47 non-welding workers in Denmark, and concluded that low-level exposure to hexavalent chromium in these welding groups did not appear to be a major hazard for human spermatogenesis.

A longitudinal design was adopted (Bonde, 1990a) to deal with the methodological drawbacks of cross-sectional studies. The analysis was designed to detect changes in semen parameters in the postvacation relative to the prevacation period among subjects in the previous study (of whom 19 MS welders, 18 SS welders and 16 referents participated). The participants were asked to deliver subsequent semen samples 3, 5 and 8 weeks after the last day of summer vacation. No significant improvement relative to the prevacation level was observed for any semen parameter among the welders during the follow-up. Contrary to expectation, among the non-exposed referents but not among welders, there was a significant time-dependent increase in sperm concentration and the proportion of normal sperm forms during follow-up, which could not be explained by sexual abstinence, urogenital infection, or fever in the preceding three months. The author hypothesized that this difference might be interpreted to be the result of a "vacation effect" improving the semen quality among non-exposed subjects but not among welders because of a lasting suppression of spermatogenesis by welding exposure (i.e. that the effect of welding is non-reversible within the rather short period of non-exposure), or alternatively, that the reported associations between exposure to welding and poor semen quality are of a non-causal nature.

A case-referent study within a cross-sectional sample was conducted among workers at six workplaces in Denmark (Bonde, 1990b). Delayed conception was defined by questionnaire as "have you and your wife (former or current) ever had difficulties in conceiving? - Only answer YES if you have tried at least two years without success (no pregnancy)". An association with welding exposure at the time of delayed conception was observed (OR 2.2 (95% CI 1.1-4.6)); when adjusted for confounders the risk dropped to 1.9 (0.8-4.1). When exposure was classified according to exposure at the time of case/ birth, welding was associated with delayed conception (OR 1.9 (0.99-3.59)) and this remained unchanged after adjustment for confounders (OR 2.0 (1.0-4.0)). Such studies are susceptible to various biases including recall bias.

Bonde et al (1990), as part of a large study of 10,059 production workers described below (Bonde et al, 1992), examined the relationship between fertility and welding exposure in a cohort of Danish male metalworkers. The occurrence of a birth in a given year was analyzed by logistic regression conditional on exposure (welding or not, MS or SS) during the preceding year as well as on several potentially confounding factors, including age, birth cohort, paternal parity, birth of a child in the preceding five years, smoking and alcohol consumption. The probability of a birth in years at risk from welding exposure was slightly but significantly reduced (OR 0.89, 95% CI 0.83-0.97). When examined according to MS or SS welding, MS welders experienced a significantly reduced likelihood of having a child during years at risk from welding (OR 0.86, 95% CI 0.76-0.99) but SS welders did not (OR 0.98, 95% CI 0.84-1.31). Among SS welders (after excluding years of MS welding), there was no difference in fertility between the not-at-risk years before and after welding (OR 0.96) nor was fertility decreased during years at risk from exposure (OR 0.95). Among never welders, the fertility of SS grinders who also have chromium and nickel exposure, was similar during years with and without exposure risk (OR 1.0). The authors noted that the effect associated with MS welding, although a small decrement, was similar to the effects in other studies of conception delay (Rachootin et al, 1983; Bonde, 1990b) and of reduced semen quality (Mortensen, 1988). It should be noted, however, that the small effects could be due to uncontrolled confounding, and that there was no trend in fertility with duration of welding exposure or with time elapsed since last exposure. The authors suggested that additional investigations are needed before a clear picture will emerge.

Radiant heat: Bonde (1992) investigated whether semen quality was affected before, during and after exposure to radiant heat in 17 MMA welders working on alloyed steel compared with two reference groups (54 nonwelding metal workers, with participation rate 37% and 19 flexoprinters, with participation rate 91%). The semen samples were examined blindly. For sperm samples delivered during exposure, the mean total sperm counts were lower in the welders (of borderline statistical significance) but no statistical differences were observed for motile sperm count, sperm concentration, proportion with normal morphology, and proportion motile. Prospective examination among the welders showed that sperm count, motile sperm count, sperm concentration and proportion of normal shaped sperm decreased within 4 to 6 weeks of the onset of exposure to heat in a subgroup of 10 welders, and then increased 4 weeks after cessation of heat exposure in a subgroup of 8 welders. There was a significant difference (p<0.01) in the proportions of sperm with normal shape found between the start of exposure (within 2 weeks) and during exposure (4 to 8 weeks), and between exposure (12 to 32 weeks) and after vacation. During exposure to heat, the skin temperature in the groin increased on average by 1.4°C.

5.4.2 Other adverse reproductive outcomes: Bonde et al (1992) conducted a large pregnancy outcome study among children of a cohort of fathers (drawn from 10,059 male metal workers employed at Danish SS or MS manufacturing companies for a minimum one year during 1964 to 1984), created from records in the Danish Pension Fund. Data on job title and department of potential cohort members were obtained at each of the 79 companies by standard interviews with managers, foremen and long-term workers. Only subjects with verified employment as MS welders, SS welders, SS grinders, or nonwelding and nongrinding production workers were admitted into the cohort. During 1986, cohort members were mailed a questionnaire on lifetime occupational exposures and smoking and drinking habits, with an 85% response rate. Liveborn offspring of all cohort members were identified from records in the Danish Central Population Register. There was a total of 3569 children fathered during the time period. The occurrence of low birthweight, preterm delivery, infant mortality and congenital malformations was not increased among children at risk from parental welding exposure compared with children not at risk. The risk of spontaneous abortion (SA) in previous pregnancies was moderately increased for pregnancies at risk from paternal exposure to SS welding (adjusted OR 2.0, 95% CI 1.1-3.5) but not from MS welding (OR 1.1, 95% CI 0.5-2.4). The risk was somewhat higher for pregnancies at risk from MMA/SS welding where Cr exposures were higher (OR 2.0) than from TIG/SS welding (OR 1.7) but the confidence intervals overlapped. The number of years with paternal SS welding experience did not influence the rate of SA.

However, there were several design limitations. First, the occurrence of SAs was monitored conditional on a subsequent birth. If SS welding has a strong effect on abortion rate, a relatively smaller fraction of pregnancies proceed to birth, which might lead to an underestimation of risk. On the other hand, women experiencing a SA might be expected to become pregnant again sooner. Second, the frequency of SAs was distorted by a high frequency of induced abortions. Third, the risk estimates were blurred by misclassification of exposure (with no data on the actual date of the SA preceding the index birth). An analysis undertaken including only births with either exposure or non-exposure through all three years preceding the birth resulted in a risk estimate for SA among SS welders that was unchanged (OR 1.7, 95% CI 0.9-3.1). The authors felt that recall bias could be excluded because the data on occupational exposure had been collected prior to the planning of the study and information on outcome variables was obtained independently. Differences between wives of the various groups of metal workers were possible; although only limited data were available, there were not expected to be major differences. The authors concluded that the hypothesis of SA increases among SS welders should be tested in further studies.

In contrast to this Danish study, Lindbohm et al (1991) investigated parental exposure and SAs in Finland by linking a nationwide data base of medically diagnosed SAs with national census data. The risk of SA was not increased in a number of parental exposure categories, including metals (OR 1.0, 95% CI 0.8-1.1).

5.4.3 Cancer in Offspring: Bunin et al (1989) explored multiple parental occupations among cases of Wilms' tumour and controls selected through random digit dialing in the U.S. using a matched case-control design. The only significant findings were linked to aromatic hydrocarbons, aliphatic hydrocarbons, metals and inorganic compounds. Welding among fathers was reported among 5 cases but only 1 control. One previous study (Kantor et al, 1979) also demonstrated this association. A similarly designed study of retinoblastoma by these investigators (Bunin et al, 1990), found an association with parental employment in the metal industry for the sporadic heritable form (OR (95% CI 1.4- ; p=0.02)) but this was not significant among welders and machinists (OR 2.3 (0.5-14)). For the nonheritable form of retinoblastoma, a significant association was observed for the job cluster consisting mostly of welders and machinists (OR 4.0 (1.1-22.1; p=0.04)).

In the follow-up study of reproductive end points among metal workers noted above, Bonde et al (1992) found that the overall incidence of childhood malignancies among 23,264 children born to the 10,059 metal workers was equal to national rates (relative risk 0.97) and was even lower among children of MS welders (0.93) or SS welders (0.77).

Summary: Bonde (1993) presented an overview of these various reproductive outcomes attributable to welding of metals, noting that hexavalent chromium and perhaps also manganese might be toxicants of interest. He noted that the internal load of these metals following short-term welding is probably "rather low in comparison with the lowest observed spermatotoxic effect levels found in animal experiments". Overall, some (Mortensen, 1988; Bonde, 1990b) but not most studies found deterioration in semen quality and fertility in welders. However, the inconsistencies between studies have not been explained and problems in methodology, low response rate, possible recall bias and so on, have to be considered, as well as the possibility that the observed associations are "of a non-causal nature" (Bonde, 1993). The evidence for a relation of welding exposure to SA (and congenital anomalies) is inadequate to reach conclusions. There is limited evidence that the risk of Wilms' tumour (and possibly retinoblastoma) is increased among the offspring of welders.

5.5 Mutagenesis

5.5.1 In vitro studies

The basis for these investigations includes that welders are exposed to and absorb chromium and nickel compounds, and that chromium and nickel in other settings are recognized to be human carcinogens by inhalation (IARC, 1990).

1985 Review: Studies reviewed 10 years ago indicated that SS welding fume appeared to be mutagenic (genotoxic) in bacteria and in cultured mammalian cells, that hexavalent chromium compounds might constitute the active component, and that gas shielded welding (MIG) produces much less hexavalent chromium than MMA.

Recent findings: Studies published during the recent period are consistent with this. Hansen and Stern (1985) showed that in the baby hamster kidney (BHK) transformation test and in primary Syrian Hamster Embryo cells, fumes from MMA/SS had transforming effects attributable to the Cr(VI) content. MIG/SS and MMA/MS had only small effects even at relatively high doses but neither contain appreciable amounts of Cr(VI). The fumes from MIG/SS had a toxic effect but this was 2-3 times greater than expected from their soluble Cr(VI) content. Relatively insoluble Cr(VI) compounds in this fume showed a higher toxic and transforming effect in the BHK assay than could be ascribed to the soluble Cr(VI) content of the medium, indicating the importance of phagocytosis as a pathway for uptake of Cr(VI) and other toxic substances from particulate.

Reuzel et al (1986) found positive results in the Ames assay and for sister chromatid exchanges (SCEs) for MMA/SS welding fume but negative results in the HGPRT test and in a DNA repair test in human cells (unscheduled DNA synthesis). The findings were similar to those of K₂CrO₄, if compared on the basis of soluble chromium quantities. Baker et al (1986) examined the effects of soluble and insoluble fractions of welding fume generated from MMA in inducing mitotic delay and SCEs in cultured Chinese hamster lung cells. They found that the water-soluble and water-insoluble fractions induced SCEs in proportion to their Cr(VI) content, while the contribution from other elements was minimal. However, mitotic delay could not be explained by Cr(VI) concentration alone, suggesting that other components of the fractions were capable of inducing mitotic delay. The activity of the insoluble fume fraction in both tests was markedly less than that of the soluble fraction, attributed to the lower bioavailability of "insoluble" components of the fume particles.

Biggart et al (1987) noted that in past studies using the modified Salmonella/ microsome assay, particles from welding SS containing 15-25% chromium were mutagenic and MS particles containing 0.1% chromium were not mutagenic or toxic. However, these investigators showed that MS welding fume induced a dose-dependent increase in reversions with or without S9 mix in the assay; thus MS particles were thought to contain direct-acting and promutagenic compounds that induced frameshift mutations, but not base-pair substitutions. The chemical nature of the mutagenic compounds was investigated. Because the mutagenicity and toxicity of the particles could not be recovered in the filtrate from particle suspension, the agents were not water soluble, and did not contain chromium compounds. Moreover, other studies indicated that an ionic metal species was not released from the particles. They felt that this demonstration of mutagenic activity in welding fume particles was an important factor to consider in assessing the potential toxicity and carcinogenicity of welding fumes.

De Raat and Bakker (1988) examined the induction of SCEs in Chinese hamster ovary cells from different welding fume processes. MMA/SS fume induced the greatest amount of SCEs, while doses about 100 times higher were required to produce effects with particles from MMA cast-iron, MIG/SS and MMA/MS, while no significant induction of SCE was demonstrated with MIG/MS. The effects could be explained by soluble chromium only for MMA/SS, while the effects associated with MMA/Cast Iron and MIG/SS particles may be caused by insoluble chromium and/ or nickel, as previously suggested. While MMA/MS particles contain some soluble chromium, the amount was not great enough to explain the effects, and no explanation for the induction of SCEs by these particles was available.

5.5.2 Cytogenicity Studies in Welders

1985 Review: Two studies performed before 1985 (Husgafvel-Pursiainen et al, 1982; Littorin et al, 1983) did not show chromosomal changes in lymphocytes of welders.

Recent findings: Elias et al (1989) studied chromosomal aberrations in France among 55 controls and 55 welders (22 MMA on coated electrodes - usually MS but SS occasionally used; 18 welders using semi-automatic MAG on nickel wire with MS metal, exposed to iron, manganese and nickel; and 15 welders using semi-automatic TIG for SS wires). The frequencies of cells with aberrations and of total structural chromosomal aberrations in welders was significantly greater (P<0.001) than in the non-exposed controls, mainly attributable to a higher frequency of chromosomal gaps in welders. In the second group of welders (MAG/Ni), the proportion of total aberrations as well as cells with aberrations was significantly higher than matched controls (P<0.01, gaps included and P<0.05, gaps excluded). In this group, there was a significant correlation between length of welding employment and frequencies of breaks, both chromatid and chromosomal (although multivariate control for smoking was not conducted). Smoking habits had no effect on the incidence of gaps but induced increased chromatid and chromosome-type breaks, acentric fragments, and in welders only, the exchange type aberrations. The authors concluded that certain welding processes may generate fumes that seem to have clastogenic activity; it was detected in a group of welders not usually welding SS who were exposed to nickel, from welding wire.

Popp et al (1991) from Germany investigated the frequencies of chromosomal breakage, cross-linking and SCEs in lymphocytes of 39 welders engaged in MAG and MMA welding using Cr/ Ni/ Mn electrodes on low- to medium-alloy steel and a control group of 18 nonexposed men whose smoking habits and age distribution (35, 36-45, 46) were standardized to the exposed group. The average SCE frequency for welders was significantly lower (P<0.05, Wilcoxon test) than that for the control group, consistent with the findings of Littorin et al (1983) and Husgafvel-Pursiainen et al (1982). The SCE frequency was raised among smokers in both groups, as expected, but the differences were not significant. However, among the welders, the SCE frequency and frequency of DNA strand breakage was significantly (P<0.05) correlated with the chromium concentration in the urine. These investigators also found a reduced frequency of strand breakage and increased frequency of protein cross-linking in the welders (as measured by reduced elution rate through filters). The authors explained this as follows: if one assumes a high proportion of DNA-protein cross-links in welders, then the reduced elution rate through polycarbonate (PC) filters can be explained in terms of a relative reduction in the numbers of shorter DNA fragments, which are partially cross-linked to protein and therefore cannot pass through PC filters. It is known that chromium can cause the formation of DNA-protein cross-links that remain stable, whereas chromium-induced DNA single-strand breaks are rapidly repaired. Thus, the main carcinogenic effect of chromates is now attributed to DNA-protein cross-link formation (Popp et al, 1991). The authors therefore concluded that the present results were in good agreement with in vitro and in vivo investigations confirming the importance of DNA-protein cross-links for the carcinogenic effect of chromium.

Knudsen et al (1992) examined various measures of genotoxic exposures in peripheral lymphocytes among 127 SS welders from 15 work places in the metal industry in Denmark (63 MMA and TIG, 52 predominantly TIG, and 12 predominantly MIG) compared with referents involved in cutting, repair work and administrative work at the same workplaces as the welders. Results were tabulated separately according to smoking status. The welders had an increased frequency of chromosomal aberrations, especially those involved with MMA welding which generates a higher chromate content in fume and greater urinary chromium. There was a decrease in unscheduled DNA synthesis and in SCE exchanges among welders. This is consistent with the previous studies which did not show significant differences in this latter outcome (Husgafvel-Pursiainen et al, 1982; Litorrin et al, 1983).

Costa et al (1993) in the U.S. used a new method for detecting DNA-protein cross-links in peripheral white blood cells among 21 railroad arc welders exposed to MS fume (none were exposed to SS welding recently) and 26 unexposed controls who were office workers, field railroad workers in supervisory positions, union representatives, janitors and laboratory technicians. The distribution of levels of DNA-protein cross-links in the two groups overlapped; however the levels were higher among the welders (mean ± SD, 1.85±1.14%) than among the controls (1.17±0.46%) (p=0.01). Of the 21 welders, 5 (24%) showed cross-link values above 2.33%, which was the upper range of the controls. Cross-link values did not vary according to age, smoking status, body weight, or race. The Ni levels in blood were not elevated in welders and there was only a slight elevation of Cr levels. Despite the observed increased in cross-linking, the authors were unsure of the agent involved. They concluded that welders were burdened with an excess of DNA-protein cross-links in cells indicating not only a biomarker of possible exposure to cross-linking agents (i.e. nickel and chromium) but also the presence of a lesion that may be an early indicator of other potential genotoxic consequences (cancer).

Jelmert et al (1994) in Norway examined chromosome damage in lymphocytes from 42 MMA/SS welders (defined as using MMA/SS for more than two-thirds of active welding time), and a subgroup of 20 SS welders studied before starting MMA/SS welding and retested after 1-4 months of welding. For each welder, a referent subject matched for gender, age and smoking habit was selected among office workers or other subjects not exposed to welding fume (reference group I), of whom 2 had unsuccessful cultures yielding 40 referents. These referents were also included within another reference group (reference group II), comprising 91 office workers, one gardener and two scaffolders. There was a significant increase among welders in chromatid breaks (1.4 vs 0.9 and 0.8 for groups I and II, respectively) and for cells with aberrations (2.2 vs 1.6 in group II). In the SCE assay, the MMA/SS welders showed no increase compared to referents. In smoking-specific analyses, in conventional cultures, a statistically significant increase in the number of chromatid breaks was observed only among non-smoking MMA/SS welders vs referents in Group I (1.5 vs 0.7, p=0.007) and Group II (1.5 vs 0.7, p=0.0009). There were no differences in smoking-specific strata for SCEs. No synergistic effect between smoking and welding fumes was observed for any type of aberration. The results indicated that the increase in chromatid breaks was associated with cumulated welding fume exposure for more than a year, and with not using any type of respirator, but not with current welding fume exposure during the week before sampling. For the subgroup of 20 welders analyzed for cytogenetic endpoints before and 1-4 months after MMA/SS welding, there was a significant increase found for chromatid breaks. However, compared to their matched referents after welding, there was no increase observed. The authors offered possible explanations for the inconsistency (and confusing findings) with the matched referents: that there were difficulties in selecting reference groups; alternatively, individual variation in results of repeated cytogenetic analyses could not be completely excluded. The authors noted that in two of the three previous non-positive studies, there were non-significant increases. The authors concluded that the small but statistically significant increase in chromatid breaks, but not in other types of chromosomal aberrations, observed for MMA/SS welders compared to matched referents, or with the results for the same welders prior to exposure, suggests that MMA/SS welders are exposed to cytotoxic agents capable of inducing chromatid breaks.

In a subsequent report from this group (Jelmert et al, 1995), the authors did not find evidence of cytogenetic damage among TIG/SS or MIG/MAG/SS welders, exposed to fumes containing a relatively lower content of Cr and Ni than the MMA/SS fume studied in the previous report. The fact that these welders are exposed to several of the same agents as the MMA/SS welders (ozone, NO_x, electromagnetic fields and ultraviolet light) but did not have aberrations, suggests that these common exposures were unlikely to be responsible for the increase in aberrations in the MMA/SS welders.

Summary : During the recent time period, although lymphocytes of welders again have been found not to demonstrate increases in SCEs, several studies have demonstrated genotoxic effects on lymphocytes of MMA/SS welders, mainly with respect to chromosomal aberrations (and DNA-protein cross-linking