Statistical modeling to determine sources of variability in exposures to welding fumes

Abstract

Background/aims: Exposures to total particulate matter (TP) and manganese (Mn) received by workers during welding and allied hot processes were analyzed to assess the sources and magnitudes of variability.

Methods: Compilation of data from several countries identified 2065 TP and 697 Mn measurements for analysis. Linear mixed models were used to determine fixed effects due to different countries, industries and trades, process characteristics, and the sampling regimen, and to estimate components of variance within workers (both intraday and interday), between workers (within worksites), and across worksites.

Results: The fixed effects explained 55 and 49% of variation in TP and Mn exposures, respectively. The country, industry/trade, type of ventilation, and type of work/welding process were the major factors affecting exposures to both agents. Measurements in the USA were generally higher than those in other countries. Exposure to TP was 67% higher in enclosed spaces and 43% lower with local exhaust ventilation (LEV), was higher among boilermakers and was higher when either a mild-steel base metal or a flux cored consumable was used. Exposure to Mn was 750% higher in enclosed spaces and 67% lower when LEV was present. Air concentrations of Mn were significantly affected by the welding consumables but not by the base metal. Resistance welding produced significantly lower TP and Mn exposures compared to other welding processes. Interestingly, exposures to TP had not changed over the 40 years of observation, while those of Mn showed (non-significant) reductions of 3.6% year(-1). After controlling for fixed effects, variance components between worksites and between-individual workers within a worksite were reduced by 89 and 57% for TP and 75 and 63% for Mn, respectively. The within-worker variation (sum of intraday and interday variance components) of Mn exposure was three times higher than that of TP exposure. The estimated probabilities of exceeding occupational exposure limits were very high (generally much >10%) for both agents.

Conclusions: Welding exposures to TP and Mn vary considerably across the world and across occupational groups. Exposures to both contaminants have been and continue to be unacceptably high in most sectors of industry. Because exposures to the two agents have different sources and characteristics, separate control strategies should be considered to reduce welders' exposures to TP and Mn.