Annals of Work Exposures and Health, 2017, 1–10 doi: 10.1093/annweh/wxx080 Original Article Original Article Occupational Exposure to Inhalable Manganese at German Workplaces Benjamin Kendzia1, Rainer Van Gelder2, Tobias Schwank2, Cornelia Hagemann2, Wolfgang Zschiesche1, Thomas Behrens1, Tobias Weiss1, Thomas Brüning1†, and Beate Pesch1*† Institute for Prevention and Occupational Medicine of the German Social Accident Insurance, Institute of the Ruhr University Bochum (IPA), Bochum, Germany; 2Institute for Occupational Safety and Health of the German Social Accident Insurance (IFA), Sankt Augustin, Germany 1 *Author to whom correspondence should be addressed. Tel: +49-(0)234-302-4536; fax: +49-(0)234-302-4505; e-mail firstname.lastname@example.org Equally contributed. † Submitted 14 March 2017; revised 19 July 2017; editorial decision 22 August 2017; revised version accepted 4 September 2017. Abstract Due to mounting evidence of neurotoxic effects of manganese (Mn) already at low concentrations, occupational exposure limits (OELs) have been adopted. We analyzed 5771 personal measurements of inhalable manganese (Mn) together with information on sampling conditions and job tasks from the German exposure database Messdaten zur Exposition gegenüber Gefahrstoffen am Arbeitsplatz (MEGA) to assess exposure levels in welders and other occupations between 1989 and 2015. Geometric means (GMs) of exposure to Mn were estimated for various occupational settings adjusted for 2-h sampling duration and analytical method, centered at 2009. Measurements below the limit of quantification (LOQ) were multiply imputed. The median concentration was 74 µg m−3 (inter-quartile range 14–260 µg m−3) in welders and 8 µg m−3 (inter-quartile range <LOQ–31 µg m−3) in other occupations. Every third measurement was higher than 100 µg m−3, 20% exceeded 200 µg m−3, and 5% of welders inhaled concentrations ≥1000 µg m−3. GMs >100 µg m−3 were observed in gas metal and flux-cored arc welders and in shielded metal arc welders using consumables of high Mn content (>5%). Tungsten inert gas welding, laser welding and working in other occupations such as foundry worker, electroplater, or grinder were associated with GMs <10 µg m−3. A shorter sampling duration was associated with higher Mn concentrations. High-emission welding techniques require protective measures to cope with adopted OELs. Results of this study are useful to assess cumulative Mn exposure in community-based studies on neurotoxic effects. Keywords: manganese; modelling; occupational exposure; welding © The Author 2017. Published by Oxford University Press on behalf of the British Occupational Hygiene Society. 2 Annals of Work Exposures and Health, 2017, Vol. XX, No. XX Introduction Mounting evidence of neurotoxic effects prompted a review of exposure to manganese (Mn) that was carried out for a comprehensive Criteria Document on adopting occupational exposure limits (OELs; Levy, 2005). The largest use of metallic Mn is in steel production to improve firmness, formability, and corrosion resistance (Postle et al., 2015). Mn is also added to nonferrous alloys. Most welding consumables contain up to 6% Mn. Furthermore, Mn compounds are used in paints and zinc-carbon and alkaline batteries, and permanganates are powerful oxidizers. Due to reports on neurotoxic effects of Mn concentrations at low doses, standard-setting committees and governmental bodies reduced OELs for airborne Mn (Bevan et al., 2017). In 2013, the American Conference of Governmental Industrial Hygienists (ACGIH) lowered the threshold limit value (TLV) for inhalable Mn to 100 µg m−3 and for respirable Mn to 20 µg m−3 (ACGIH, 2011, 2013). In 2015, German OELs were reduced to 20 µg m−3 for respirable Mn and to 200 µg m−3 for inhalable Mn (http://www.baua.de/de/Themen-von-A-Z/ Gefahrstoffe/TRGS/TRGS-900.html). However, the relation between Mn concentrations in respirable and inhalable particles of metal fumes may vary by setting and has been studied in parallel in few studies only. In welding fumes, most of Mn was respirable (Pesch et al., 2012; Jeong et al., 2016). Personal exposure to Mn varies widely between and within different industrial settings (Levy, 2005). The technological process, Mn content of materials, workplace conditions, and protective measures may influence the concentration of Mn in the breathing zone of workers as could be quantified for welders (Pesch et al., 2012). Several job tasks can hardly comply with these OELs, for example gas metal arc welding (GMAW) (Wallace et al., 2001; Flynn and Susi, 2010; Pesch et al., 2012; Taube, 2013). In high-exposure circumstances such as welding with high-emission techniques in confined space, powered air-purifying respirators may be needed reduce Mn to levels below OELs (Lehnert et al., 2014). So far, estimates of exposure to Mn were usually based on few measurements in selected industrial settings (Levy, 2005). Large exposure databases have been established in many countries, which could be mined to assess personal exposure to Mn (Flynn and Susi, 2010; Peters et al., 2012). These databases compile personal measurements since the 1980s, where measurements in the respirable particle fraction are limited. Here, we took advantage of a large number of personal measurements of inhalable Mn compiled in the German database ‘Messdaten zur Exposition gegenüber Gefahrstoffen am Arbeitsplatz’ (MEGA) along with ancillary information on the sampling conditions, analytical methods, and technological processes (Stamm, 2001; Gabriel et al., 2010; Pesch et al., 2015; Kendzia et al., 2017). The analysis of these measurements serves for several aims, for example to publish reports on exposure levels at German workplaces and in connection with the chemical safety assessment of REACH (Gabriel et al., 2010). Our objective was to estimate exposure to Mn in a variety of job tasks and industrial settings in order to evaluate compliance with OELs and to provide quantitative data for the exposure assessment within the framework of research on neurotoxic effects in community-based studies. Our results were used to calculate cumulative exposure to Mn of metal workers enrolled in the Heinz Nixdorf Recall Study (HNRS, https://www. uni-due.de/recall-studie/) to estimate neurotoxic effects (Casjens et al., 2017). Materials and Methods Measurements of inhalable Mn The concentrations of inhalable Mn were obtained from the exposure monitoring and measurement system (MGU) of the Institute for Occupational Safety and Health (IFA) of the German Social Accident Insurance (DGUV) compiled in the exposure database MEGA as previously reported for an analysis of hexavalent chromium (Cr(VI)) and nickel (Ni) (Pesch et al., 2015; Kendzia et al., 2017). The measurements were carried out by technicians of the Employer’s Liability Insurance Associations (Berufsgenossenschaften) based on standard operating procedures and a computer-assisted data documentation of the measurements and ancillary information. For example, the Mn content of welding consumables is shown in the material safety data sheet. Here, this analysis was based on 5771 personal measurements collected between 1989 and 2015 together with data on the sampling duration, analytical method, and occupational setting, including information on the welding technique and materials. The measurements were performed with a GSP sampler operating at a flow rate of 3.5 l min−1 according to the standard EN 481 to capture inhalable particles in the breathing zone of workers (EN 481, 1993). Workplace samples were analysed at the central laboratory at IFA by means of atomic absorption spectrometry (AAS), reflection X-ray fluorescence (TXRF), inductively coupled plasma mass spectrometry (ICP-MS) and inductively coupled plasma optical emission spectrometry Annals of Work Exposures and Health, 2017, Vol. XX, No. XX3 (ICP-OES). Concentrations below the limit of quantification (LOQ) were documented by their individual LOQs, which mainly depend upon the analytical method, pump flow rate, and duration of sampling. Assessment of occupational exposure to manganese Job titles were coded in MEGA according to the German classification of occupations (Bundesanstalt für Arbeit, 1988). We assigned measurements according to the information on job title, more detailed job tasks and occupational settings to predefined at-risk occupations of a supplemental questionnaire applied in the second follow-up of HNRS to investigate neurotoxic effects. The modelling of exposure levels in these settings based on the measurements corresponds to an approach that has been applied to analyse Cr(VI) and Ni (Pesch et al., 2015; Kendzia et al., 2017). We categorized welders by major process and consumable material as major source of airborne Mn (Pesch et al., 2012). We assigned a high Mn content if welding consumables contained >5%. We considered the contents of aluminium (Al), Cr, or Ni in settings with Mn ≤5% as follows. High Al means a content usually higher than 95%. Next, we classified consumable electrodes as stainless steel if the Cr or/and Ni content exceeded 10%. Otherwise, electrodes were classified as mild steel. In case of techniques not using consumables, this classification was applied to the base material. Metal workers comprised torch cutters, electroplaters, foundry workers, steel workers, grinders, metal mixer, blast-furnace workers, solderers, rolling-mill workers, scrap-metal workers, and sinters. We included metal workers and all other occupations with measurements of Mn in the category ‘non-welders’. Statistical analysis All calculations were performed with the statistical software SAS, version 9.4 (SAS Institute Inc., Cary, NC, USA). We modelled exposure to Mn according to the methods applied to measurements of Cr(VI) and Ni (Pesch et al., 2015; Kendzia et al., 2017). In brief, the fraction of measurements below the LOQ and selected percentiles were used to describe the distribution of inhalable Mn, in particular the median, inter-quartile range (IQR), 90th, and 95th percentile. Measurements below the LOQ were multiply imputed with 10 runs (Lotz et al., 2013). We defined an interval where the lower bound was set to zero and the upper bound was set to the specified LOQ and postulated that the probability distribution of concentrations within this interval (0 to LOQ) depends on all measurements above LOQ. For 10 imputation runs, we applied mixed regression models and adjusted the log-transformed Mn concentrations for occupational setting, calender year (mediancentered at 2009), analytical method, and duration of sampling: ln (Mn) = Intercept + β × Occupational setting + β × Year 1 2 + β × Analytical method + β × ln (Duration ) + ε . 3 4 For welders, we additionally implemented the welding technique and consumable material with regard to the occupational setting. Following, point estimates of all runs were combined with PROC MIANALYZE. We estimated geometric means (GMs) of exposure to Mn for the median duration of sampling (2 h) for the various occupational settings. Adjusted R2 was estimated for goodness-of-fit of the respective regression models (Harel, 2009). Results Table 1 presents the characteristics of 5771 personal measurements of inhalable Mn. The median of all concentrations was 35 µg m−3 (IQR 6–180 µg m−3). Major analytical method was ICP-MS, especially in more recent years. We observed a halving of the concentrations from 1989 to 2015 based on all measurements. Low-emission welding technologies such as tungsten inert gas welding (TIG) or laser welding were more frequently used in more recent years (data not shown). Sampling duration was mostly 2 h. Individual LOQs ranged from 0.0015 to 67 µg m−3 (data not shown). Table 2 and Fig. 1A and B shows the distribution of inhalable Mn in welders (n = 3985) and other occupations. Median Mn concentrations were much higher in welders (74 µg m−3; IQR 14–260 µg m−3) than in other occupations (8 µg m−3; IQR <LOQ-31 µg m−3). About 5% of welders exceeded concentrations of 1000 µg m−3. Flux-cored metal arc welding (FCAW) ranked highest (median 240 µg m−3; IQR 27–860 µg m−3), followed by GMAW (median 130 µg m−3; IQR 35–340 µg m−3), in particular when welding mild steel. Also, a high Mn content of the consumables in shielded metal arc welding (SMAW) can result in concentrations >100 µg m−3 (median 110 µg m −3; IQR 46–835 µg m −3). Mn concentrations < 20 µg m−3 were mostly measured in TIG welding (median 8 µg m−3; IQR 4–21 µg m−3), plasma, and laser welding. Also, all other metal workers had median Mn concentrations <20 µg m−3 except torch cutters (median 28 µg m−3; IQR 8–74 µg m−3). For example, foundry workers had a median value of 7 µg m−3 and grinders presented on average with 11 µg m−3. Annals of Work Exposures and Health, 2017, Vol. XX, No. XX 4 Table 1. Distribution of personal measurements of inhalable manganese compiled in the German MEGA database between 1989 and 2015. Characteristics N Total 5771 Analytical method AAS 1694 ICP-OES 665 ICP-MS 2529 X-ray fluorescence 883 Time of measurement [years] 1989 to <2005 1482 2005 to <2009 1297 2009 to <2012 1470 2012 to ≤2015 1522 Sampling time [hours] <2 903 2 to <3 4054 3 to <4 537 ≥4 277 N < LOQ (%) P25 (µg m−3) Median (µg m−3) P75 (µg m−3) P90 (µg m−3) P95 (µg m−3) 14 6 35 180 500 880 16 13 9 22 10 14 4 7 58 67 23 29 250 260 140 130 650 700 400 430 1100 1200 680 810 17 14 10 13 10 8 6 4 55 46 28 21 270 210 140 120 740 560 410 350 1300 1000 700 610 11 14 17 12 20 6 4 4 110 30 20 21 400 160 130 110 1300 430 392 320 2600 710 800 500 AAS = atomic absorption spectroscopy; ICP-OES = inductively coupled plasma emission spectroscopy; ICP-MS = inductively coupled plasma mass spectroscopy; LOQ = limit of quantification; P25 = 25th percentile; P75 = 75th percentile; P90 = 90th percentile; P95 = 95th percentile. Table 3 depicts the adjusted effect estimates from the regression model as factors modifying the average exposure level. GMAW and ICP-MS served as reference groups. Only welders applying FCAW had higher Mn concentrations (exp(β) 1.51; 95% confidence interval [CI]: 1.03–2.19) than GMAW. Welding with electrodes of >5% Mn was associated with nearly 2-fold higher Mn concentrations as compared to mild steel. AAS and X-ray fluorescence did not differ from ICP-MS but ICPOES resulted in 1.42-fold higher concentrations. We could not estimate a significant time trend between 1989 to 2015 when stratifying by setting and adjusting for covariates. Mn concentrations were significantly higher in exposure circumstances with a shorter sampling time. Adjusted R2 was 0.42 (95% CI 0.40–0.44). Table 4 presents GMs in these occupational settings centered for the median year 2009 with adjustment for the average sampling duration of 2 h. GMs >100 µg m−3 were estimated for GMAW using consumables of high Mn content (201 µg m−3; 95% CI 138–291 µg m−3) and when applying FCAW (115 µg m−3; 95% CI 79–168 µg m−3). We estimated GMs >50 µg m −3 for all welders using GMAW (78 µg m−3; 95% CI 72–84 µg m−3) and SMAW with electrodes of high Mn content (84 µg m−3; 95% CI 56–127 µg m −3). GMs <20 µg m −3 were estimated for welding with Al, submerged arc welding with the exception of electrodes of high Mn content. GMs <5µg m −3 were observed in laser, plasma, and resistance welding. Among the non-welding settings, GMs between 10–20 µg m −3 were estimated in torch cutters, blastfurnace workers, grinders, and metal mixers. All other settings had lower GMs, for example foundry workers (5 µg m−3). Discussion Mn is an essential additive for improving the hardness of steel and is a major constituent of welding fumes, besides a variety of other applications and exposure circumstances (Postle et al., 2015). High exposure to airborne Mn has been associated with neurotoxic symptoms such as movement disorders (Olanow, 2004; Lucchini et al., 2009). Here, we estimated occupational exposure to inhalable Mn using 5771 personal measurements and ancillary data compiled in the German exposure database MEGA between 1989 and 2015. The data distribution was skewed with a median of 35 µg m−3 and 880 µg m−3 as 95th percentile but welders had nearly 10-fold higher concentrations than other metal workers. GMAW, FCAW, and SMAW with consumables of high Mn content ranked highest with P95s >1000 µg m−3. Following mounting evidence of neurobehavioral dysfunctions already at lower doses, regulatory bodies adopted OELs for Mn (Bevan et al., 2017). For inhalable Mn, ACGIH recommends 100 µg m −3 as TLV (ACGIH, 2013), and Germany established an OEL at Annals of Work Exposures and Health, 2017, Vol. XX, No. XX5 Table 2. Distribution of personal measurements of inhalable manganese in occupations with anticipated exposure (MEGA database, 1989–2015). Occupation Welder Gas metal arc welding Tungsten inert gas welding Shielded metal arc welding Flux-cored metal arc welding Laser welding Submerged arc welding Plasma welding Resistance welding Others or not specified Non-welders Torch cutter Electroplater Foundry worker Steel worker Grinder Metal mixer Blast-furnace worker Solderer Rolling-mill worker Scrap-metal worker Sinter Other occupations Welding material N N < LOQ (%) P25 (µg m−3) Median (µg m−3) P75 (µg m−3) P90 (µg m−3) P95 (µg m−3) Total Mild steel Stainless steel Al content >5% Mn content >5% Other content Total 3985 2721 1418 342 68 47 846 529 8 3 2 5 29 2 2 18 14 35 53 20 <LOQ 40 34 4 74 130 180 50 10 113 130 8 260 340 410 135 30 400 320 21 610 690 890 380 73 1800 590 47 1100 1100 1400 620 150 2900 940 71 Mild steel Stainless steel Al content >5% Mn content >5% Other content Total 60 332 34 10 93 283 25 18 48 0 15 6 3 4 <LOQ 30 5 15 15 7 3 30 8 57 31 15 8 74 22 210 43 37 22 100 50 930 92 51 27 120 70 2200 Mild steel Stainless steel Al content >5% Mn content >5% Other content 85 84 22 92 95 3 7 8 6 2 28 10 46 15 27 95 33 110 57 240 270 79 835 210 860 1200 180 3000 830 1678 2200 210 5300 1600 2293 47 36 27 103 144 1786 251 60 212 10 461 144 14 26 32 46 20 511 36 5 56 50 17 27 8 50 23 10 22 32 7 50 28 52 50 33 <LOQ 8 <LOQ <LOQ 6 <LOQ 8 <LOQ 2 4 4 <LOQ 6 <LOQ <LOQ <LOQ <LOQ <LOQ 3 21 <LOQ 3 52 8 28 5 7 6 11 5 15 4 4 <LOQ 8 5 8 88 6 6 235 31 74 11 24 21 34 38 45 5 14 5 13 23 37 150 24 20 730 130 290 24 86 48 100 550 60 8 34 13 140 130 48 370 77 30 1100 370 630 45 190 66 280 830 240 18 69 20 320 350 LOQ = limit of quantification; P25 = 25th percentile; P75 = 75th percentile; P90 = 90th percentile; P95 = 95th percentile. 200 µg m−3 (http://www.baua.de/de/Themen-von-A-Z/ Gefahrstoffe/TRGS/TRGS-900.html). In our analysis of measurements of inhalable Mn at German workplaces, the 75th percentile of all measurements was 180 µg m−3, 34% exceeded 100 µg m−3, and 20% were >200 µg m−3, mostly among welders. We observed a 50% reduction of 6 Annals of Work Exposures and Health, 2017, Vol. XX, No. XX Figure 1. (A and B) Density function of the concentrations of inhalable manganese among welders (left) and non-welders (right) (MEGA database, 1989–2015). exposure levels between 1989 and 2015, which may be due to an increasing use of low-emission welding technologies such as TIG and laser welding as well as an improvement of ventilation and local fume extraction at the workplaces. The distribution of Mn concentrations in an exposure database is influenced by the purpose of the measurements and, hence, the selection of settings for monitoring workplace exposure. The MEGA database documents a large data set of measurements from multiple purposes, including inspection, compliance, research, and surveys. The majority of measurements was available for welders, where GMAW comprised 47% of all concentrations of inhalable Mn. The median concentration was 74 µg m−3 in all welders, which corresponds to 73 µg m−3 as median among 230 welders of the German study WELDOX using the same sampling device in the breathing zone of the welders (Pesch et al., 2012). FCAW was associated with the highest exposure levels among all at-risk settings and frequently exceeded 200 µg m−3 (median MEGA: 240 µg m−3, WELDOX: 585 µg m−3). An intervention study in a high-exposure setting with FCAW demonstrated that Mn concentrations <20 µg m−3 could be achieved inside powered air-purifying respirators (Lehnert et al., 2014). GMAW of mild steel ranked also high with a median concentration of 180 µg m−3 in MEGA as well as in WELDOX. Inhalable Mn >100 µg m−3 was measured in SMAW when using consumable electrodes with Mn content >5%. TIG welding was mostly associated with inhalable Mn <20 µg m−3, which confirms observations in WELDOX and other studies (Flynn and Susi, 2010; Pesch et al., 2012). New techniques such as laser welding resulted in even lower concentrations of inhalable Mn with GMs ≤2 µg m−3. Exposure to inhalable Mn was on average lower than 20 µg m−3 in most of the non-welding settings, like in steel production. We found only minor differences between average concentrations for certain occupations estimated with data compiled in MEGA and in the US Chemical Exposure Health Database (CEHD, https://www.osha.gov/opengov/healthsamples.html). For example, median exposure of blast-furnace workers was 14 µg m−3 which is similar to the median of 13 µg m−3 from measurements compiled CEHD. Even lower averages were observed for several other settings, like scrapmetal workers (MEGA: 2 µg m−3, CEHD: 4 µg m−3). The respirable mass concentration of particulate matter is a more suitable metric for neurotoxic effects of Mn, which can enter the brain via blood (Baker et al., 2014). However, only few measurements of respirable Mn were available in MEGA or other databases. Whereas most measurements of Mn compiled in MEGA were derived from inhalable particulate matter, CEHD compiled measurements of Mn in total dust. Progress has been made in defining size-specific sampling criteria, and the inhalable particle fraction has replaced ‘total dust’ (Levy, 2005). The relation between Mn determined in inhalable versus respirable particulate matter may vary by setting and other factors and has been estimated with a factor of 5 (ACGIH) or 10 (Germany) for particle-size-specific OELs. When preparing the comprehensive report on Mn (Levy, 2005), only one study in the alloy production was identified, which reported a ratio of 10 from side-byside measurements of Mn in the inhalable and respirable fraction (Ellingsen et al., 2000). Newer studies found lower ratios of around 2 for measurements in welding fumes where Mn is mostly respirable (Pesch et al., 2012; Jeong et al., 2016). One strength of this modelling approach is the large number of Mn concentrations in a wide range of job tasks together with ancillary data about the measurements, workplaces, and welding techniques. In published studies, the Mn content of welding consumables as major source of airborne Mn was rarely considered Annals of Work Exposures and Health, 2017, Vol. XX, No. XX7 Table 3. Influence of occupation, year of measurement, welding material, and sampling duration on the concentration of inhalable manganese (MEGA database, 1989–2015). Intercept Welder Gas metal arc welding Tungsten inert gas welding Shielded metal arc welding Flux-cored metal arc welding Laser welding Submerged arc welding Plasma welding Resistance welding Others or not specified Welding material Mild steel Stainless steel Al content >5% Mn content>5% Other or mixed content Non-welders Torch cutter Electroplater Foundry worker Steel worker Grinder Metal mixer Blast-furnace worker Solderer Rolling-mill worker Scrap-metal worker Sinter Other occupations Analytical method AAS ICP-OES ICP-MS X-ray fluorescence Year of measurement [Ref = 2009] Sampling time [log(h)] 95% CI N Exp(β) 5771 194 169–222 2721 529 283 95 47 36 27 103 144 1.00 0.10 0.42 1.51 0.03 0.24 0.02 0.02 0.28 0.09–0.13 0.33–0.53 1.03–2.19 0.01–0.04 0.13–0.44 0.01–0.06 0.01–0.03 0.20–0.38 1995 1096 137 96 2447 1.00 0.57 0.10 2.10 0.77 0.49–0.66 0.07–0.14 1.44–3.07 0.68–0.87 251 60 212 10 461 144 14 26 32 46 20 511 0.20 0.02 0.08 0.07 0.11 0.07 0.19 0.01 0.05 0.01 0.03 0.05 0.16–0.26 0.01–0.03 0.06–0.11 0.02–0.22 0.09–0.14 0.05–0.09 0.07–0.49 0.00–0.03 0.03–0.09 0.00–0.01 0.01–0.08 0.04–0.06 1694 665 2529 883 5771 5771 0.93 1.42 1.00 1.10 0.97 0.44 0.79–1.10 1.20–1.67 0.94–1.28 0.96–1.00 0.39–0.50 CI = confidence interval of Exp(β). R2 0.42 (95% CI = 0.40–0.44). (Hobson et al., 2011). Facing the lognormal distribution of measurements at workplaces, GMs should be preferred to present average exposure levels instead of arithmetic means. A limitation of our analysis is the duration of measurements which lasted usually 2 h. In high-exposure settings, the sampler could be loaded with particles even within a shorter time. As discussed in the modelling of Cr(VI) concentrations with MEGA data, we refrained from calculating 8-h shift estimates (Pesch et al., 2015). For example, the measurements are commonly performed during the activity under evaluation, for example when welding. However, the welding torch does not burn during the whole shift. The modelling of average exposure concentrations in at-risk occupations is a pivotal step in exposure assessment. The information compiled in MEGA allowed us to describe exposure patterns not only by welding technique but also for various types of electrodes. Annals of Work Exposures and Health, 2017, Vol. XX, No. XX 8 Table 4. Model-based estimates of the geometric means of occupational exposure to inhalable manganese predicted for the year 2009 with adjustment for sampling time. (MEGA database, 1989–2015). Occupations Welding material Welder Gas metal arc welding Mild steel Stainless steel Al content >5% Mn content >5% Tungsten inert gas welding Mild steel Stainless steel Al content >5% Mn content >5% Shielded metal arc welding Mild steel Stainless steel Al content >5% Mn content >5% Flux-cored metal arc welding Laser welding Submerged arc welding Plasma welding Resistance welding Others or not specified Non-welders Torch cutter Electroplater Foundry worker Steel worker Grinder Metal mixer Blast-furnace worker Solderer Rolling-mill worker Scrap-metal worker Sinter Other occupations 95% CI (µg m−3) N Geometric mean (µg m−3) 2721 1418 342 68 47 529 60 332 34 10 283 85 84 – 22 95 47 36 27 103 144 78 95 54 9 201 6 10 6 1 21 33 40 23 4 84 115 2 18 1 1 21 72–84 87–105 47–62 7–13 138–291 5–7 8–12 5–7 1–1 14–31 26–41 32–51 18–29 3–6 56–127 79–168 1–3 10–34 1–4 1–2 15–29 251 60 212 10 461 144 14 26 32 46 20 511 16 1 5 6 8 5 13 1 4 1 2 3 13–20 1–2 4–7 219 6–9 3–6 5–35 0–2 2–7 0–1 1–6 3–4 CI = confidence interval of geometric mean. However, some information was not documented in former MEGA data, like the position of the sampler head (inside or outside of a helmet) or the type of wire (solid or flux-cored). In WELDOX, also the type of wire, efficiency of the local exhaust ventilation and welding in confined space was assessed and could explain >70% of the variance of respirable Mn together with the major welding technique (Pesch et al., 2012). When assessing exposure in large cohort or community-based studies, the retrospective assessment of such detailed information is less feasible. We already used the estimates of this analysis to calculate cumulative Mn exposure in welders and other occupations for the investigation of neurotoxicity in a community-based study (Casjens et al., 2017). A welding process exposure matrix was developed for a cohort study among European welders where the job axis was stratified by major technique and type of steel (Gerin et al., 1993). Our former analysis of Cr(VI) and Annals of Work Exposures and Health, 2017, Vol. XX, No. XX9 Ni using MEGA data may improve this JEM (Pesch et al., 2015; Kendzia et al., 2017). Mn can be an additional agent of such a matrix. Conclusions This analysis of a large data set of inhalable Mn aimed at estimating average exposure levels in occupational settings in Germany between 1989 and 2015. Welders had a median value of 74 µg m−3 compared to 8 µg m−3 in other occupations. GMAW, FCAW, and SMAW frequently exceeded 100 µg m−3 and require protective measures to comply with OELs. We recommend collecting information about the welding technique in addition to the job title in community-based studies when estimating neurotoxic effects of Mn in welders. Acknowledgements We are grateful for access to concentrations of manganese compiled in the database MEGA of the Institute for Occupational Safety and Health of the German Social Accident Insurance (IFA), Sankt Augustin, Germany. Conflict of interest Authors from the Institute for Prevention and Occupational Medicine (IPA) and Institute for Occupational Safety and Health (IFA) work for the German Social Accident Insurance (DGUV). The authors are independent from the DGUV in study design, access to the collected data, responsibility for data analysis and interpretation, and the right to publish. References ACGIH (Ed.). (2011) TLVs and BEIs, Threshold Limit Values for Chemical Substances and Physical Agents and Biological Exposure Indices. Cincinnati, OH: Signature Publications. ACGIH (Ed.). (2013) Manganese, Elemental and Inorganic Compounds: TLV(R) Chemical Substances. 7th edn Documentation. Cincinnati, OH: Signature Publications. Baker MG, Simpson CD, Stover B et al. (2014) Blood manganese as an exposure biomarker: state of the evidence. J Occup Environ Hyg; 11: 210–7. Bevan R, Ashdown L, McGough D et al. (2017) Setting evidence-based occupational exposure limits for manganese. Neurotoxicology; 58: 238–48. Bundesanstalt für Arbeit. (1988) Klassifizierung der Berufe: Systematisches und alphabetisches Verzeichnis der Berufsbenennungen. Nürnberg, Germany: Bundesanstalt für Arbeit. Casjens S, Pesch B, Robens S et al. (2017) Associations between former exposure to manganese and olfaction in an elderly population: Results from the Heinz Nixdorf Recall Study. Neurotoxicology; 58: 58–65. Ellingsen D, Bast-Pettersen R, Hetland S et al. (2000) Health Survey of Individuals Exposed to Manganese in Smelting Plants: A Cross Section Survey. Norway: Statens Arbeidsmiloinstitutt (STAMI). EN 481. (1993) Workplace atmospheres; size fraction definitions for measurement of airborne particels: European Standard EN 481. Brussels, Belgium: European Standardization Committee. Flynn MR, Susi P. (2010) Manganese, iron, and total particulate exposures to welders. J Occup Environ Hyg; 7: 115–26. Gabriel S, Koppisch D, Range D. (2010) The MGU - a monitoring system for the collection and documentation of valid workplace exposure data. Gefahrstoffe - Reinhalt Luft; 70: 43–9. Gerin M, Fletcher AC, Gray C et al. (1993) Development and use of a welding process exposure matrix in a historical prospective study of lung cancer risk in European welders. Int J Epidemiol; 22 (Suppl 2): 22–8. Harel O. (2009) The estimation of R2 and adjusted R2 in incomplete data sets using multiple imputation. J App Stat; 36: 1109–18. Hobson A, Seixas N, Sterling D et al. (2011) Estimation of particulate mass and manganese exposure levels among welders. Ann Occup Hyg; 55: 113–25. Jeong JY, Park JS, Kim PG. (2016) Characterization of total and size-fractionated manganese exposure by work area in a shipbuilding yard. Saf Health Work; 7: 150–5. Kendzia B, Pesch B, Koppisch D et al. (2017) Modelling of occupational exposure to inhalable nickel compounds. J Expo Sci Environ Epidemiol; 27: 427–33. Lehnert M, Weiss T, Pesch B et al.; WELDOX Study Group. (2014) Reduction in welding fume and metal exposure of stainless steel welders: an example from the WELDOX study. Int Arch Occup Environ Health; 87: 483–92. Levy LS. (2005) Occupational exposure limits: Criteria document for manganese and inorganic manganese compounds. Leicester, UK: MRC Institute for Environment and Health, pp. 285. Lotz A, Kendzia B, Gawrych K et al. (2013) Statistical methods for the analysis of left-censored variables. GMS Med Inform Biom Epidemiol; 9: 1–9. Lucchini RG, Martin CJ, Doney BC. (2009) From manganism to manganese-induced parkinsonism: a conceptual model based on the evolution of exposure. Neuromolecular Med; 11: 311–21. Olanow CW. (2004) Manganese-induced parkinsonism and Parkinson’s disease. Ann N Y Acad Sci; 1012: 209–23. Pesch B, Kendzia B, Hauptmann K et al. (2015) Airborne exposure to inhalable hexavalent chromium in welders and other occupations: Estimates from the German MEGA database. Int J Hyg Environ Health; 218: 500–6. Pesch B, Weiss T, Kendzia B et al. (2012) Levels and predictors of airborne and internal exposure to manganese and iron among welders. J Expo Sci Environ Epidemiol; 22: 291–8. Peters S, Vermeulen R, Olsson A et al. (2012) Development of an exposure measurement database on five lung carcinogens (ExpoSYN) for quantitative retrospective occupational exposure assessment. Ann Occup Hyg; 56: 70–9. 10 Annals of Work Exposures and Health, 2017, Vol. XX, No. XX Postle M, Nwaogu T, Upson S, et al. (2015) Manganese, The Global Picture: A Socio Economic Assessment. Report for the International Manganese Institute; Loddon, Norfolk, UK. http://www.manganese.org/images/uploads/publications/ Manganese_SEA_-_Complete.pdf. Stamm R. (2001) MEGA-database: one million data since 1972. Appl Occup Environ Hyg; 16: 159–63. Taube F. (2013) Manganese in occupational arc welding fumes– aspects on physiochemical properties, with focus on solubility. Ann Occup Hyg; 57: 6–25. Wallace M, Shulman S, Sheehy J. (2001) Comparing exposure levels by type of welding operation and evaluating the effectiveness of fume extraction guns. Appl Occup Environ Hyg; 16: 771–9.