Advertisement

Cancer mortality in cohorts of workers in the European rubber manufacturing industry first employed since 1975

      ABSTRACT

      Background

      Increased cancer risk has been reported among workers in the rubber manufacturing industry employed before the 1960s. It is unclear whether risk remains increased among workers hired subsequently. The present study focused on risk of cancer mortality for rubber workers first employed since 1975 in 64 factories.

      Patients and methods

      Anonymized data from cohorts of rubber workers employed for at least 1 year from Germany, Italy, Poland, Sweden, and the UK were pooled. Standardized mortality ratios (SMRs), based on country-specific death rates, were reported for bladder and lung cancer (primary outcomes of interest), for other selected cancer sites, and for cancer sites with a minimum of 10 deaths in men or women. Analyses stratified by type of industry, period, and duration of employment were carried out.

      Results

      A total of 38 457 individuals (29 768 men; 8689 women) contributed to 949 370 person-years. No increased risk of bladder cancer was observed [SMR = 0.80, 95% confidence interval (CI) 0.46; 1.38]. The risk of lung cancer death was reduced (SMR = 0.81, 95% CI 0.70; 0.94). No statistically significant increased risk was observed for any other cause of death. A reduced risk was evident for total cancer mortality (SMR = 0.81, 95% CI 0.76; 0.87). Risks were lower for workers in the tyre industry compared with workers in the general rubber goods sector. Analysis by employment duration showed a negative trend with SMRs decreasing with increasing duration of employment. In an analysis of secondary end points, when stratified by type of industry and period of first employment, excess risks of myeloma and gastric cancer were observed each due, essentially, to results from one centre.

      Conclusion

      No consistent increased risk of cancer death was observed among rubber workers first employed since 1975, no overall analysis of the pooled cohort produced significantly increased risk. Continued surveillance of the present cohorts is required to confirm the absence of long-term risk.

      Key words

      introduction

      Epidemiological studies have reported increased risk of bladder cancer and leukaemia among workers in the rubber manufacturing industry, mainly among those employed before the 1960s: these excess risks have been associated with exposure to aromatic amines, and solvents, respectively [
      IARC. The rubber industry. Monographs on the Evaluation of the Carcinogenic Risks to Humans, Suppl 7. Overall Evaluations of Carcinogenicity: An Updating of IARC Monographs Volumes 1 to 42.
      ]. A narrative (qualitative) review of epidemiological studies published up to 1997 concluded that bladder, laryngeal, lung cancer, and leukaemia were the cancers with an increased risk among rubber workers, with evidence of substantial heterogeneity of results [
      • Kogevinas M.
      • Sala M.
      • Boffetta P.
      • et al.
      Cancer risk in the rubber industry: a review of the recent epidemiological evidence.
      ]. However, a meta-analysis based on cohort studies published through 2003 showed no excess risk for any of the cancer sites, although there was not differentiation made between natural rubber and synthetic rubber (styrene butadiene) [
      • Alder N.
      • Fenty J.
      • Warren F.
      • et al.
      Meta-analysis of mortality and cancer incidence among workers in the synthetic rubber-producing industry.
      ]. Recently, the IARC has re-evaluated the rubber industry, and has classified rubber production as a group 1 carcinogen for cancers of the stomach, the lung, and lymphoma in addition to bladder cancer and leukaemia [
      • IARC
      A review of human carcinogens. Volume 100 Part F: Chemical agents and related occupations/IARC Working Group on the Evaluation of Carcinogenic Risks to Humans.
      ], employing all data available irrespective of the period of exposure.
      The rubber manufacturing industry, which employed ∼380 000 workers in EU-15 in 1990–1993 [
      • Kauppinen T.
      • Toikkanen J.
      • Pedersen D.
      • et al.
      Occupational exposure to carcinogens in the European Union.
      ], has undergone radical technological changes since the 1950s, entailing major reductions in rubber dust and fume exposure or the removal of known carcinogenic agents, although the changes in technology were gradual and took place at variable pace in different countries. Data collected by the EU-ExAsRub consortium in the rubber manufacturing industry showed a general reduction in the exposure of workers to inhalable dust from the 1970s to 2003, however with marked differences between countries [
      • de Vocht F.
      • Vermeulen R.
      • Burstyn I.
      • et al.
      Exposure to inhalable dust and its cyclohexane soluble fraction since the 1970s in the rubber manufacturing industry in the European Union.
      ]. These results are of interest but are of an ecological nature and such data are not available on a personal level.
      Two studies of workers first employed after 1980 in the rubber manufacturing industry in Germany [
      • Taeger D.
      • Weiland S.K.
      • Sun Y.
      • et al.
      Cancer and non-cancer mortality in a cohort of recent entrants (1981–2000) to the German rubber industry.
      ] and UK [
      • Dost A.
      • Straughan J.
      • Sorahan T.
      A cohort mortality and cancer incidence survey of recent entrants (1982–91) to the UK rubber industry: findings for 1983–2004.
      ] showed no excess risk of overall cancer mortality, as well as bladder, stomach, and lung cancers. The present study reports data on risk of death from cancer and major causes of death from a large-scale epidemiologic study of workers first employed since 1975 in the rubber manufacturing industry from five European countries. This study provides therefore a direct evaluation of risk of cancer in the modern rubber manufacturing industry.

      materials and methods

      A protocol specifying inclusion criteria and a detailed statistical analysis plan was prepared between local principal investigators of the present study before data analysis (available upon request from the corresponding author). The cohort consisted of rubber workers employed in Germany, Italy, Poland, Sweden, and the UK. Workers were included if they (i) were employed for at least 1 year in one of the cohorts, and (ii) had a first employment in the rubber manufacturing industry at 1 January 1975 or later. This first employment period varied from country to country, being 1981 for the German cohort, 1982 for the UK, and 1975 for all other countries. Factories were distributed throughout all regions in Poland, Sweden, and the UK. For Italy, workers were employed in factories of the north of Italy: in the region of Piemonte (Italy). For Germany, workers were employed in factories located in western Germany in the Federal States of Lower Saxony and North Rhine-Westphalia.
      For all countries but Germany, follow-up data were complete until a right censoring date (Table 1). For Germany, the follow-up data and vital status were complete until 2000, and from 2001 to 2003, cause of death and migration status was missing for the majority of workers. The period 2001–2003 was therefore excluded from the analysis for Germany. From 2004 to 2012, vital status became partially available and enabled evaluating vital status and causes of death. In order to avoid underestimation of mortality rates, correction factors, depending on migration and risk of deaths from all causes, were applied to person-years from 2001 onward. These correction factors were estimated from the 2042 workers in Hannover city for which complete information on vital status and migration was available. Sensitivity analyses were conducted to evaluate the influence of these correction factors.
      Table 1Characteristics of the five European cohorts of workers first employed in the rubber manufacturing industry since 1975
      GermanyItalyPolandSwedenUK
      Number of workers76165115970074248602
      Number of factories8221141
      Follow-up (years)18.719.129.421.724.1
      Number of deaths1362161451376546
      Gender (% men)85.991.163.266.387.4
      Type of industry (%tyre/%GRG/%other
      Other includes mixed factories (both tyre and GRG).
      )
      0/34.1/65.9100/0/077.1/22.9/05.5/32.2/62.451.0/36.2/12.8
      Age of hire (median, IQR)24 (21; 29)25 (22; 31)21 (19; 26)25 (20; 35)25 (21; 33)
      Duration of Employment (median, IQR)5.3 (2.5; 10.5
      Not estimable as more than 25% of workers were still employed at the last job history update.
      )
      10.7 (4.9; 16.3
      Not estimable as more than 25% of workers were still employed at the last job history update.
      )
      7.4 (3.0; 17.2)4.6 (2.2; 10.3
      Not estimable as more than 25% of workers were still employed at the last job history update.
      )
      4.9 (2.6; 12.7)
      Date of first recruitment5 January 19811 January 19751 January 19751 January 19751 January 1982
      Date of last follow-up31 December 2012 (LS) or 31 December 2013 (NRW)31 March 201330 June 201231 December 201131 December 2011
      Date of last job history update31 December 200031 March 201331 October 20111 July 2002
      • – 31 December 2011 (4005 subjects/47%)
      • – Variable between 1988 and 1995 (4597 subjects/53%)
      % Still employed at last job history update51.2%46.4%0.5%44.8%
      • – Last job history update 2011: 14.5% among 4005 subjects
      LS, Lower Saxony; NRW, North Rhine-Westphalia; GRG, general rubber goods.
      a Other includes mixed factories (both tyre and GRG).
      b Not estimable as more than 25% of workers were still employed at the last job history update.
      The primary outcomes of interest were mortality from bladder cancer and lung cancer. Secondary outcomes were mortality from all cancers combined, stomach cancer, leukaemia, multiple myeloma. and non-Hodgkin's lymphoma (NHL). Exploratory outcomes included all other neoplasms with more than 10 observed deaths in men or women in the combined analysis and mortality from cardiovascular diseases, respiratory diseases (without pneumonia), and all-causes. If an exploratory outcome was considered for a particular site, the SMR for the other gender group was reported only if the number of deaths was >5. The list of ICD-9 codes used in the present article is reported in supplementary Table S1, available at Annals of Oncology online.
      The observed numbers of deaths for each cause of death considered were compared with the expected number calculated on the basis of national or regional (Italy) gender, age-, and period-specific mortality rates. Five-year age groups were used for age and time period. National reference rates were obtained from the WHO Mortality Database (Revision of November 2014). For Italy, reference rates could be obtained from the region of Piedmont. Data from the Central Statistical Office of Poland were used when data from the WHO mortality database were not available (earlier years for a few cancer sites). For each country, standardized mortality ratios (SMRs, i.e. the ratio of observed to expected deaths) were calculated together with their confidence intervals (CIs) based on the Poisson distribution of observed deaths [
      • Breslow N.E.
      • Day N.E.
      Statistical Methods in Cancer Research, Vol. 2. The Design and Analysis of Cohort Studies.
      ,
      • Owen D.B.
      Handbook of Statistical Tables.
      ]. Country-specific SMRs were combined using random-effects models [
      • van Houwelingen H.C.
      • Arends L.R.
      • Stijnen T.
      Advanced methods in meta-analysis: multivariate approach and meta-regression.
      ], which take into account potential heterogeneity among cohorts. Forest plots were reported for bladder and lung cancer mortality (primary outcomes) to evaluate the relative contribution of each country in the overall risk estimate.
      Measures of heterogeneity was reported using I2 statistics [
      • Higgins J.P.
      • Thompson S.G.
      Quantifying heterogeneity in a meta-analysis.
      ] as well as test for heterogeneity based on Cochran's Q statistic, although this test is known for having poor statistical power [
      • Gavaghan D.J.
      • Moore R.A.
      • McQuay H.J.
      An evaluation of homogeneity tests in meta-analyses in pain using simulations of individual patient data.
      ].
      As a sensitivity analysis, several stratified analyses were conducted. To evaluate whether the risk differed by type of industry, workers were separated between tyre production, general rubber goods, and a mixed category for factories that manufactured both tyres and general rubber goods. The role of the period of first employment on the risk of death was also investigated separating workers employed before 1985 to those employed from 1985 onwards.
      The role of the duration of employment on the risk of death was investigated in a Poisson model with gender and a gender-specific smooth function of duration of employment as explanatory variables, and the logarithm of the expected number of deaths as ‘offset’. Duration of employment was modelled with cubic natural splines and three degrees of freedom. Cut-points (knots) of duration of employment were built from duration of employment of deceased subjects, such as at least eight deaths occurred between two points. The knots for the spline were built in a way that the knots divide duration of employment in intervals that contain the duration of employment of eight deceased workers each. This enables a stable estimation of SMR while keeping enough points for modelling the splines for the parameter of duration.
      All data were anonymized before statistical analysis. P-values below 5% were considered as statistically significant.

      results

      In the five European countries, a total of 38 457 workers (29 768 men and 8689 women) were included in the present study. Each country contributed rather homogeneously to the number of workers with Poland and the UK providing the greatest number of workers (Table 1). The average follow-up was 26 years and ranged from 18.7 years in Germany to 29.4 years in Poland, contributing to a total of 949 370 person-years of observation. The majority of workers (77.4%) were men. More women were recruited in Poland and Sweden. No major difference was observed between the cohorts assembled by each country. Overall, 2725 deaths were observed during the follow-up.
      Table 2 shows observed deaths and SMRs for primary, secondary, and exploratory outcomes for men, women, and both genders combined. In men, there were significantly reduced SMRs for lung cancer, colorectal cancer, all cancers combined, all causes combined, and cardiovascular diseases. In women, there were significantly reduced SMRs for all cancers combined, all causes combined, and cardiovascular diseases. In the total study population (both genders combined), there were significantly reduced SMRs for lung cancer, colorectal cancer, all cancers combined, all causes combined, and cardiovascular diseases.
      Table 2Observed deaths and standardized mortality ratios (SMRs) among 38 457 workers first employed in the European rubber manufacturing industry since 1975, by gender
      Cause of deathsMen (n = 29 768; PY = 714 612)

      Women (n = 8689; PY = 234 758)

      Combined (n = 38 457; PY = 949 370)

      ObservedSMR (95% CI)ObservedSMR (95% CI)ObservedSMR (95% CI)
      Primary outcomes
       Bladder cancer160.87 (0.49; 1.52)10.36 (0.06; 2.29)170.80 (0.46; 1.38)
       Lung cancer1570.80 (0.68; 0.94)290.91 (0.61; 1.35)1860.81 (0.70; 0.94)
      Secondary outcomes
       All cancers5820.81 (0.74; 0.88)1870.83 (0.72; 0.97)7690.81 (0.76; 0.87)
       Stomach cancer471.15 (0.84; 1.56)101.23 (0.61; 2.46)571.16 (0.88; 1.53)
       Leukaemia180.97 (0.58; 1.63)40.76 (0.14; 4.02)220.84 (0.53; 1.35)
       Non-Hodgkin's lymphoma140.74 (0.40; 1.37)40.85 (0.28; 2.57)180.79 (0.47; 1.34)
       Multiple myeloma121.55 (0.81; 2.98)41.68 (0.56; 5.11)161.54 (0.89; 2.68)
      Exploratory outcomes
      Results of exploratory analysis with <5 deaths observed were not reported.
       All causes of death22590.80 (0.71; 0.89)
      Significant heterogeneity between the five countries (P < 0.05)
      4660.86 (0.78; 0.94)27250.81 (0.73; 0.89)
      Significant heterogeneity between the five countries (P < 0.05)
       Cardiovascular diseases6550.80 (0.71; 0.90)1100.75 (0.61; 0.90)7650.79 (0.70; 0.89)
       Respiratory diseases excluding pneumonia680.93 (0.72; 1.19)120.74 (0.38; 1.42)800.90 (0.71; 1.13)
       Brain and central nervous system cancer330.95 (0.66; 1.39)90.99 (0.46; 2.14)420.96 (0.69; 1.34)
       Colorectal cancer390.61 (0.44; 0.86)201.16 (0.71; 1.91)590.72 (0.55; 0.94)
       Kidney cancer180.83 (0.49; 1.38)3210.82 (0.51; 1.32)
       Larynx cancer171.18 (0.69; 2.04)0171.12 (0.65; 1.93)
       Liver cancer150.73 (0.40; 1.31)2170.69 (0.40; 1.20)
       Melanoma110.88 (0.45; 1.71)1120.68 (0.36; 1.28)
       Oesophagus230.90 (0.57; 1.41)1240.84 (0.54; 1.30)
       Oral cavity and pharynx230.97 (0.62; 1.53)1240.92 (0.59; 1.43)
       Pancreas cancer330.93 (0.64; 1.36)131.31 (0.70; 2.44)460.98 (0.72; 1.35)
       Prostate cancer260.79 (0.52; 1.21)
       Breast cancer1360.86 (0.60; 1.23)
       Cervical cancer171.25 (0.74; 2.10)
       Ovarian cancer120.73 (0.38; 1.40)
      Statistically significant associations are in bold.
      a Results of exploratory analysis with <5 deaths observed were not reported.
      b Significant heterogeneity between the five countries (P < 0.05)
      Table 3 shows observed deaths and SMRs for primary and secondary outcomes for men, women, and both genders combined, by industry sector. In the tyre sector, there were significantly reduced SMRs for lung cancer in men and in both genders combined, and for all cancers combined in men, women, and both genders combined. There were no significant excesses. In the general rubber goods sector, there were significantly elevated SMRs for stomach cancer in men and in both genders combined, and for multiple myeloma in men and both genders combined. There were no significant deficits. In the mixed sector, there were significantly reduced SMRs for all cancers combined in men and both genders combined. There were no significant excesses.
      Table 3Observed deaths and standardized mortality ratios by type of industry among 38 457 workers first employed in the European rubber manufacturing industry since 1975, by industry sector
      Cause of deathsTyre

      General rubber goods

      Mixed

      ObservedSMR (95% CI)ObservedSMR (95% CI)ObservedSMR (95% CI)
      Menn = 13 962; PY = 359 318n = 7684; PY = 185 206n = 8122; PY = 170 021
       Bladder cancer91.03 (0.48; 2.20)60.84 (0.32; 2.18)10.23 (0.03; 1.90)
       Lung cancer
      Primary outcomes.
      720.72 (0.57; 0.92)670.95 (0.74; 1.23)180.78 (0.47; 1.31)
       All cancers2350.67 (0.59; 0.77)2631.05 (0.92; 1.19)840.70 (0.56; 0.88)
       Stomach cancer200.97 (0.60; 1.57)241.75 (1.12; 2.71)30.58 (0.12; 2.75)
       Leukaemia111.15 (0.60; 2.19)60.84 (0.34; 2.08)10.14 (0.02; 1.15)
       Non-Hodgkin's lymphoma61.12 (0.21; 5.93)60.95 (0.35; 2.64)20.44 (0.08; 2.57)
       Multiple myeloma30.94 (0.25; 3.55)83.02 (1.40; 6.51)10.32 (0.04; 2.68)
      Womenn = 3437; PY = 107 540n = 2628; PY = 68 544n = 2624; PY = 58 672
       Bladder cancer00.00 (0.01; 2.98)00.00 (0.01; 1.62)11.30 (0.14; 11.92)
       Lung cancer
      Primary outcomes.
      80.61 (0.29; 1.26)161.68 (0.96; 2.92)50.63 (0.23; 1.73)
       All cancers660.68 (0.53; 0.88)801.01 (0.74; 1.38)410.85 (0.61; 1.17)
       Stomach cancer30.64 (0.19; 2.19)52.07 (0.77; 5.59)21.37 (0.23; 8.14)
       Leukaemia10.29 (0.04; 2.29)10.36 (0.05; 2.51)20.84 (0.04; 16.90)
       Non-Hodgkin's lymphoma10.45 (0.06; 3.56)21.08 (0.20; 5.81)10.54 (0.06; 4.73)
       Multiple myeloma21.69 (0.37; 7.73)21.08 (0.12; 9.56)00.00 (0.01; 2.56)
      Men and women combinedn = 17 399; PY = 466 858n = 10 312; PY = 253 750n = 10 746; PY = 228 693
       Bladder cancer90.97 (0.46; 2.08)60.76 (0.29; 1.98)20.50 (0.11; 2.30)
       Lung cancer
      Primary outcomes.
      800.71 (0.56; 0.90)831.02 (0.81; 1.29)230.73 (0.47; 1.14)
       All cancers3010.68 (0.60; 0.76)3431.03 (0.91; 1.18)1250.74 (0.61; 0.88)
       Stomach cancer230.92 (0.59; 1.44)291.83 (1.23; 2.72)50.71 (0.24; 2.13)
       Leukaemia120.96 (0.52; 1.78)70.80 (0.35; 1.84)30.60 (0.15; 2.37)
       Non-Hodgkin's lymphoma71.08 (0.24; 4.90)81.08 (0.46; 2.50)30.50 (0.13; 1.99)
       Multiple myeloma51.12 (0.41; 3.04)103.18 (1.61; 6.27)10.25 (0.03; 2.06)
      Statistically significant associations are in bold.
      a Primary outcomes.
      Based on 17 deaths observed during the follow-up, the risk of death from bladder cancer was SMR = 0.80 (95% CI 0.46–1.38) with no evidence of heterogeneity between countries (Figure 1). This absence of increased risk for bladder cancer remained when stratified by gender (Table 2), employment period (supplementary Table S2, available at Annals of Oncology online), and type of industry (Table 3). The risk of death from lung cancer was significantly decreased in rubber workers when compared with the general population with an SMR of 0.81 (95% CI 0.70–0.94) with no heterogeneity between countries (Figure 2).
      Figure 1
      Figure 1Forest plot of the risk of bladder cancer mortality (men and women combined) from five cohorts of European rubber workers. Numbers of deaths are reported in brackets. The pink line corresponds to a standardized mortality ratio (SMR) of 1, i.e. no difference in mortality between the cohort and the general population. The blue line corresponds to the SMR for all-cause mortality.
      Figure 2
      Figure 2Forest plot of the risk of lung cancer mortality (men and women combined) from five cohorts of European rubber workers. Numbers of deaths are reported in brackets. The pink line corresponds to a standardized mortality ratio (SMR) of 1, i.e. no difference in mortality between the cohort and the general population. The blue line corresponds to the SMR for all-cause mortality.
      For all other cancer sites as well as all-cause mortality, cardiovascular and respiratory mortality, the risk of death was not significantly increased for neither men nor women (Table 2). On the contrary, a statistically significant decrease was observed for mortality from all causes of cancer (SMR = 0.81, 95% CI 0.76–0.87), all causes of death (SMR = 0.81, 95% CI 0.73–0.89), cardiovascular diseases (SMR = 0.79, 95% CI 0.70–0.89), and for risk of death from colorectal cancer (SMR = 0.72, 95% CI 0.55–0.94). Even if only eight deaths in men were observed for mesothelioma, the SMR was estimated for this cancer site and was 1.32 (95% CI 0.72–2.43). The overall results were not affected by the correction factors applied to person-years in Germany to account for migration and risk of death in the 2001–2003 period with no follow-up information.
      Workers in the tyre industry were hired at around the same age as workers in the general rubber goods industry (median age of hire being 23 and 25 years, respectively), and they were followed for about the same duration (25 and 23.5 years, respectively). However, workers in the tyre industry were more frequently men (80% men) than in the general rubber goods industry (74% men). The greatest difference was the duration of employment which was a median of 10.5 years in the tyre industry, while it was 4.2 years in the general rubber good industry. When stratifying workers by type of industry, a clear difference could be observed between these two populations with lower SMR in the tyre industry when compared with general rubber goods (Table 3). For total cancer mortality, the risk of death was SMR = 0.68 (95% CI 0.60–0.76) among workers in the tyre industry while it was close to 1 in workers in the general rubber goods industry with an SMR of 1.03 (95% CI 0.91–1.18) (Figure 3). In this stratified analysis, workers in the general rubber goods industries had a significant increased risk of stomach cancer (SMR = 1.83, 95% CI 1.23–2.72) and multiple myeloma (SMR = 3.18, 95% CI 1.61–6.27). Elevated SMRs were observed only among workers employed before 1985 (supplementary Table S2, available at Annals of Oncology online). These results were not consistent between countries and primarily driven by a majority of deaths in Poland for stomach cancer and a majority of deaths in the UK for multiple myeloma.
      Figure 3
      Figure 3Forest plot of the risk of all cancers mortality (men and women combined) from five cohorts of European rubber workers in the tyre industry (pink), in the general rubber goods (GRG) industry (green), and in mixed industry (blue). Numbers of deaths are reported in brackets. The black line corresponds to a standardized mortality ratio (SMR) of 1, i.e. no difference in mortality between the cohort and the general population.
      Workers employed from 1985 onwards had an apparently, though not significantly lower risk of total cancer mortality with an SMR of 0.74 (95% CI 0.65–0.84) when compared with workers employed before 1985 for which the SMR was 0.86 (95% CI 0.79–0.94). Such a decreased risk in more recently employed workers was also observed for other primary and secondary outcomes (supplementary Table S2, available at Annals of Oncology online).
      The risk of death from any cause was further investigated in a Poisson regression with a spline function applied to the duration of employment. With a risk initially above the general population, especially among men and women in general rubber goods production, a marked decline of the risk of death was observed with increasing years of employment (Figure 4, supplementary Figure S2, available at Annals of Oncology online). This trend seemed slightly stronger in men but went in the same direction for women. A similar trend was observed when limiting the analysis to total cancer mortality (supplementary Figure S1, available at Annals of Oncology online). Although difficult to interpret because of small numbers, a similar trend was observed for lung cancer mortality.
      Figure 4
      Figure 4Analysis of risk of death from any cause from five cohorts of European rubber workers by duration of employment in the rubber manufacturing industry. The horizontal pink line corresponds to the global SMR for all causes of death. The plain blue lines represent the spline curve of SMR by duration of employment (bold line) with its 95% confidence interval (black lines).

      discussion

      In this cohort of workers in the rubber manufacturing industry first employed since 1975 in five European countries, there was no consistent indication of an increased risk of cancer mortality among the cancer sites pre-identified as primary or secondary outcomes of interest, i.e. total cancer mortality, and site-specific mortality for bladder, lung, stomach, leukaemia, myeloma, and NHL. In addition, none of the exploratory outcomes had their risk significantly increased.
      There are a number of noticeable issues among a number of stratified analyses showing inconsistencies in the outcomes. In an analysis of secondary end points, when stratified by type of industry and period of first employment, excess risks of myeloma and gastric cancer were observed due, essentially, to results from the UK and Poland, respectively. Increases in risk of multiple myeloma and stomach cancer were observed mainly for the workers first employed before 1985, but not workers employed more recently. It is possible that these increased risks, which are further decreasing in more recently exposed workers, are the result of decrease in exposure to carcinogens. However, these results should be interpreted with caution as they were not consistent across countries, gender, and type of industry, and mainly driven by a few excess deaths observed in the UK for multiple myeloma and a few excess deaths observed in Poland for stomach cancer.
      Lung cancer, one of the primary hypotheses of the study, showed decreased risks in most countries. Secondary outcomes such as colorectal cancer, total cancer mortality, all-cause mortality, and cardiovascular diseases also showed decreased risks. The reduction in risks for causes of death was lower for workers in the tyre industries than for the general rubber goods industries. Several factors could be responsible for such a reduced risk. Such a pattern of decreased risk could be the sign of a ‘healthy worker effect’ [
      • Checkoway H.
      • Pearce N.
      • Kriebel D.
      Cohort studies.
      ], in particular because the observed numbers of deaths from any cause and for total cancer mortality were systematically lower than expected. The healthy worker effect is a classical bias in epidemiology of occupational mortality and results in biased estimation of SMR when mortality of workers is compared with mortality of the general population. Its impact could play at the inception of cohort of workers with a ‘healthy population selection effect’ and during follow-up with a ‘survivor population effect’ [
      • Fox A.J.
      • Collier P.F.
      Low mortality rates in industrial cohort studies due to selection for work and survival in the industry.
      ].
      In the cohorts of European rubber workers, a decreased risk of all-cause mortality is consistent with a healthy population selection effect, in particular because the follow-up started at recruitment in each cohort [
      • Goldblatt P.
      • Fox J.
      • Leon D.
      Mortality of employed men and women.
      ]. It has been suggested that the initial health advantage conferred by selection is expected to progressively dissipate with prolonged follow-up [
      • Goldblatt P.
      • Fox J.
      • Leon D.
      Mortality of employed men and women.
      ]. This would be in contradiction with the decreasing trend in SMR for longer duration of employment in rubber workers. However, staying active in the industry could be a secondary effect of selection with less healthy workers being more likely to be removed from hazardous exposure than healthier workers [
      • Applebaum K.M.
      • Malloy E.J.
      • Eisen E.A.
      Reducing healthy worker survivor bias by restricting date of hire in a cohort study of Vermont granite workers.
      ]. Longer duration in employment could be interpreted as a marker of improved economic status, better access to medical care, and changes in life-style [
      • Wen C.P.
      • Tsai S.P.
      Anatomy of the health worker effect—a critique of summary statistics employed in occupational epidemiology.
      ]. The healthy worker effect has been reported in previous studies on SMRs for cancer mortality [
      • Fox A.J.
      • Collier P.F.
      Low mortality rates in industrial cohort studies due to selection for work and survival in the industry.
      ,
      • Applebaum K.M.
      • Malloy E.J.
      • Eisen E.A.
      Reducing healthy worker survivor bias by restricting date of hire in a cohort study of Vermont granite workers.
      ], but this bias usually affected less the SMRs for cancer than the SMRs for total mortality [
      • Fox A.J.
      • Collier P.F.
      Low mortality rates in industrial cohort studies due to selection for work and survival in the industry.
      ,
      • Goldblatt P.
      • Fox J.
      • Leon D.
      Mortality of employed men and women.
      ]. In the present study, SMRs were significantly decreased for several cancer sites with values even lower than the total mortality.
      The difference between tyre industry and general rubber goods could be an artefact with workers in the general rubber goods industry were employed for a shorter duration. However, the effect of type of industry and duration of employment seemed independent when conducting the spline regression separately by type on industry (supplementary Figure S2, available at Annals of Oncology online). This indicates that both working in the tyre industry and for a long duration were actually associated with a decreased risk of overall mortality. In contrast, in the general rubber goods industry, a similar healthy worker effect was not evident for total mortality: elevated SMRs were evident up to 15 years duration of employment.
      Rubber production is an occupation classified as group 1 carcinogen by IARC, without clear indication of what the responsible agent might be [
      • IARC
      A review of human carcinogens. Volume 100 Part F: Chemical agents and related occupations/IARC Working Group on the Evaluation of Carcinogenic Risks to Humans.
      ]. A particular focus has been made on tyre production, in particular the tyre curing process during which high temperatures are reached to stimulate reaction between different chemical compounds of the tyre. Measurements conducted with the European project ExAsRub [
      • De Vocht F.
      • Straif K.
      • Szeszenia-Dabrowska N.
      • et al.
      A database of exposures in the rubber manufacturing industry: design and quality control.
      ] where the results can be consulted online (http://exasrub.iras.uu.nl/) revealed higher exposure to several agents in workers working in the general rubber goods when compared with workers in the tyre industry. For example, in the period 1989–1993 in the UK, personal representative measurement of rubber dust was 2.82 mg/m3 (geometric mean) in the general rubber goods and 1.18 mg/m3 in the tyre industry. The differences in the overall SMRs between tyre and general rubber goods sectors might also be due to differences in smoking habits; however, no clear data were identified to confirm this hypothesis.
      Major industries in Europe have experienced major changes in occupational hygiene through increased surveillance and actions to protect workers. In the rubber manufacturing industry, these changes seemed to have started earlier as the analysis of the data collected within the EU-ExAsRub consortium showed a continuous decreasing time trends of inhalable dust from 1975 to 2005, although there was national variation [
      • de Vocht F.
      • Vermeulen R.
      • Burstyn I.
      • et al.
      Exposure to inhalable dust and its cyclohexane soluble fraction since the 1970s in the rubber manufacturing industry in the European Union.
      ]. The results stratified by period of employment showed that decreased risks of lung cancer and total cancer mortality were even stronger after 1985 than before 1985 which is in line with data showing a progressive decline in exposure to inhalable dust measured in rubber industries in Europe. Caution must be expressed in interpreting these findings. Comparison with mortality data in the prospective cohort, whether overall or by subgroup (such as type of manufacturing), and the dust levels are on an ecological basis, a weak type of study design compared with the prospective cohort study reported here.
      The findings from this study are reassuring. However, it is recommended that these cohorts continue to be monitored regularly to ensure that the absence of increased risk is maintained for longer follow-up, and the long-term positive effects of the continual implementation of technology innovations to reduce exposures to workers can be maintained. Longer follow-up will also allow examination of cohorts employed in more recent years (after 1985). In addition, it will allow more detailed evaluation of the healthy worker effect.

      funding

      The study was partially funded by an unrestricted research grant from the European Tyre and Rubber Manufacturers’ Association1 (ETRMA) and was conducted and reported in full independence from the sponsor. The German study was partially funded and supported by the BG RCI2 (formerly Berufsgenossenschaft der Chemischen Industrie, Heidelberg) and the Wirtschaftsverband der deutschen Kautschukindustrie e V, Frankfurt/M. No grant numbers apply.
      European Tyre and Rubber Manufacturers’ Association
      BG RCI2

      disclosure

      The authors have declared no conflicts of interest.

      acknowledgements

      The study Working Group comprised iPRI staff (Mathieu Boniol and Peter Boyle) and each national Principal Investigator [Beata Świątkowska (Poland), Tom Sorahan (UK), Jürgen Wellmann (Germany), Kristina Jakobsson (Sweden), Enrico Pira (Italy)]. AK and CP were iPRI staff working on the project. DT was the co-principal investigator with Jürgen Wellmann working on the German cohort. CLV and PB were co-principal investigators with Enrico Pira working on the Italian cohorts. The authors would like to thank Faustine Valentini and Kim Coppens (iPRI) for the editorial support during the preparation of this manuscript, and Maria Bota (iPRI) for help in literature scoping. Prof. Neonila Szeszenia-Dąbrowska and Urszula Wilczyńska, PhD, provided background information on the rubber industry workers and allowed the follow-up of the already existing rubber manufacturing cohorts in Poland. Zoli Mikoczy, and Ulf Bergendorf were responsible for the assembly of the Swedish database.

      Supplementary data

      references

      1. IARC. The rubber industry. Monographs on the Evaluation of the Carcinogenic Risks to Humans, Suppl 7. Overall Evaluations of Carcinogenicity: An Updating of IARC Monographs Volumes 1 to 42.
        IARC, Lyon, France1987: 332-334
        • Kogevinas M.
        • Sala M.
        • Boffetta P.
        • et al.
        Cancer risk in the rubber industry: a review of the recent epidemiological evidence.
        Occup Environ Med. 1998; 55: 1-12
        • Alder N.
        • Fenty J.
        • Warren F.
        • et al.
        Meta-analysis of mortality and cancer incidence among workers in the synthetic rubber-producing industry.
        Am J Epidemiol. 2006; 164: 405-420
        • IARC
        A review of human carcinogens. Volume 100 Part F: Chemical agents and related occupations/IARC Working Group on the Evaluation of Carcinogenic Risks to Humans.
        IARC, Lyon, France2012
        • Kauppinen T.
        • Toikkanen J.
        • Pedersen D.
        • et al.
        Occupational exposure to carcinogens in the European Union.
        Occup Environ Med. 2000; 57: 10-18
        • de Vocht F.
        • Vermeulen R.
        • Burstyn I.
        • et al.
        Exposure to inhalable dust and its cyclohexane soluble fraction since the 1970s in the rubber manufacturing industry in the European Union.
        Occup Environ Med. 2008; 65: 384-391
        • Taeger D.
        • Weiland S.K.
        • Sun Y.
        • et al.
        Cancer and non-cancer mortality in a cohort of recent entrants (1981–2000) to the German rubber industry.
        Occup Environ Med. 2007; 64: 560-561
        • Dost A.
        • Straughan J.
        • Sorahan T.
        A cohort mortality and cancer incidence survey of recent entrants (1982–91) to the UK rubber industry: findings for 1983–2004.
        Occup Med. 2007; 57 (Oxford, England): 186-190
        • Breslow N.E.
        • Day N.E.
        Statistical Methods in Cancer Research, Vol. 2. The Design and Analysis of Cohort Studies.
        IARC, Lyon, France1987
        • Owen D.B.
        Handbook of Statistical Tables.
        Addison-Wesley Publishing Co., MA1962
        • van Houwelingen H.C.
        • Arends L.R.
        • Stijnen T.
        Advanced methods in meta-analysis: multivariate approach and meta-regression.
        Stat Med. 2002; 21: 589-624
        • Higgins J.P.
        • Thompson S.G.
        Quantifying heterogeneity in a meta-analysis.
        Stat Med. 2002; 21: 1539-1558
        • Gavaghan D.J.
        • Moore R.A.
        • McQuay H.J.
        An evaluation of homogeneity tests in meta-analyses in pain using simulations of individual patient data.
        Pain. 2000; 85: 415-424
        • Checkoway H.
        • Pearce N.
        • Kriebel D.
        Cohort studies.
        in: Research Methods in Occupational Epidemiology. 2nd edition. Oxford University Press, New York, USA2004: 123-178
        • Fox A.J.
        • Collier P.F.
        Low mortality rates in industrial cohort studies due to selection for work and survival in the industry.
        Br J Prev Soc Med. 1976; 30: 225-230
        • Goldblatt P.
        • Fox J.
        • Leon D.
        Mortality of employed men and women.
        Am J Ind Med. 1991; 20: 285-306
        • Applebaum K.M.
        • Malloy E.J.
        • Eisen E.A.
        Reducing healthy worker survivor bias by restricting date of hire in a cohort study of Vermont granite workers.
        Occup Environ Med. 2007; 64: 681-687
        • Wen C.P.
        • Tsai S.P.
        Anatomy of the health worker effect—a critique of summary statistics employed in occupational epidemiology.
        Scand J Work Environ Health. 1982; 8: 48-52
        • De Vocht F.
        • Straif K.
        • Szeszenia-Dabrowska N.
        • et al.
        A database of exposures in the rubber manufacturing industry: design and quality control.
        Ann Occup Hyg. 2005; 49: 691-701