You are here

  • Home
  • Archive
  • Volume 76, Issue 4
  • Lifetime exposure to rubber dusts, fumes and N-nitrosamines and cancer mortality in a cohort of British rubber workers with 49 years follow-up

Article Text

Article menu

This article has a correction. Please see:

Download PDFPDF

XML
Workplace
Original article
Lifetime exposure to rubber dusts, fumes and N-nitrosamines and cancer mortality in a cohort of British rubber workers with 49 years follow-up
  1. Mira Hidajat1,
  2. Damien Martin McElvenny2,
  3. Peter Ritchie2,
  4. Andrew Darnton3,
  5. William Mueller2,
  6. Martie van Tongeren4,
  7. Raymond M Agius4,
  8. John W Cherrie2,5,
  9. Frank de Vocht1
  1. 1 Population Health Sciences, Bristol Medical School, University of Bristol, Bristol, UK
  2. 2 Research Division, Institute of Occupational Medicine, Edinburgh, UK
  3. 3 Statistics and Epidemiology Unit, Health and Safety Executive, Bootle, UK
  4. 4 Centre for Occupational and Environmental Health, Centre for Epidemiology, School of Health Sciences, The University of Manchester, Manchester, UK
  5. 5 Institute of Biological Chemistry, Biophysics and Bioengineering, Heriot Watt University, Edinburgh, UK
  1. Correspondence to Dr Frank de Vocht, Population Health Sciences, Bristol Medical School, University of Bristol, Bristol, BS8 2PS, UK; frank.devocht{at}bristol.ac.uk

Abstract

Objectives To quantitatively evaluate exposure-response associations between occupational exposures to rubber dust, fumes and N-nitrosamines and cancer mortality in the UK rubber industry.

Methods Competing risk survival analyses were used to examine cancer mortality risk in a cohort of 36 441 males aged 35+ years employed in the British rubber industry in 1967, followed up to 2015 (94% mortality). Exposure measurements are based on a population-specific quantitative job-exposure matrix for rubber dust, rubber fumes and N-nitrosamines from the EU-EXASRUB project.

Results Exposure (lifetime cumulative (LCE))-response associations were found for N-nitrosomorphiline and all cancers (subdistribution HR (SHR) 1.48, 95% CI 1.39 to 1.57) and cancers of the bladder, stomach, multiple myeloma, oesophagus, prostate and pancreas, as well as for N-nitrosodimethylamine and all cancers (SHR 2.08, 95% CI 1.96 to 2.21) and cancers of the bladder, stomach, leukaemia, multiple myeloma, prostate and liver. LCE to the N-nitrosamines sum were associated with increased risks from all cancers (SHR 1.89, 95% CI 1.78 to 2.01) and cancers of the lung, non-Hodgkin’s lymphoma and brain. LCE to rubber dust and fumes are associated with increased mortality from all cancers (rubber dust SHR 1.67, 95% CI 1.58 to 1.78; rubber fumes SHR 1.91, 95% CI 1.80 to 2.03) and cancers of the bladder, lung, stomach, leukaemia, multiple myeloma, non-Hodgkin’s lymphoma, oesophagus, prostate, pancreas and liver.

Conclusions Consistent with previous studies, N-nitrosamines exposures are associated with mortality from cancers of the bladder, lung, stomach, leukaemia, multiple myeloma, oesophagus, prostate, pancreas and liver. The long follow-up with nearly complete mortality enabled estimations of lifetime cancer mortality risk from occupational exposures in the rubber industry.

  • rubber
  • occupational exposures
  • cohort
  • exposure-response
  • cancer

This is an open access article distributed in accordance with the Creative Commons Attribution 4.0 Unported (CC BY 4.0) license, which permits others to copy, redistribute, remix, transform and build upon this work for any purpose, provided the original work is properly cited, a link to the licence is given, and indication of whether changes were made. See: https://creativecommons.org/licenses/by/4.0/.

https://doi.org/10.1136/oemed-2018-105181

Statistics from Altmetric.com

    Request Permissions

    If you wish to reuse any or all of this article please use the link below which will take you to the Copyright Clearance Center’s RightsLink service. You will be able to get a quick price and instant permission to reuse the content in many different ways.

    Key messages

    What is already known about this subject?

    • Employment in the rubber industry has been concluded to cause cancer by the International Agency for Research in Cancer.

    What are the new findings?

    • This paper updates exposure-response associations between cancer mortality and specific occupational exposures in the rubber industry for 49 years of follow-up and nearly complete mortality.

    • Exposure-response associations were found for N-nitrosamines exposure and mortality from cancers of the bladder, stomach, oesophagus, leukaemia, multiple myeloma, prostate, pancreas and liver), and for rubber dust exposure and the lung.

    • Elevated risks for cancer mortality without exposure-response patterns were also found for occupational exposures to rubber dust, rubber fumes and nitrosamines.

    • Occupational exposure in the rubber industry was not found to be associated with laryngeal cancer.

    How might this impact on policy or clinical practice in the foreseeable future?

    • Results from this study contributed to the evidence of elevated cancer mortality risks from occupational exposures in the rubber industry by further clarifying the relationship between each carcinogen and cancer.

    Introduction

    Employment in the rubber industry has been concluded to cause cancer by the International Agency for Research in Cancer (IARC). In particular, cancers of the bladder, lung, stomach, leukaemia and malignant lymphoma are designated as having sufficient evidence for excess incidence and/or mortality among workers in the rubber industry.1 Important carcinogenic exposures encountered in this industry are N-nitrosamines, rubber (process) dust, rubber fumes, polycyclic aromatic hydrocarbons including phthalates, aromatic amines including β-naphthylamine and solvents including benzene, among others.2

    Exposures vary throughout the rubber manufacturing process. Rubber dust tends to have highest exposure in the beginning of the production process, particularly in handling raw materials. Rubber fumes (measured as the cyclohexane soluble fraction of rubber dust) and N-nitrosamines are generated during the heating and curing processes. Due to the complexity of exposure patterns and the numerous chemicals used in the rubber production process, disentangling exposure-response associations between specific suspected carcinogens and cancer risk in this industry remains difficult. Excess mortality from bladder cancer among rubber workers, particularly those employed before 1950, has been documented in multiple studies.2–6 Nevertheless, others found this association only for higher exposures to aromatic amines and not rubber dust.7 An exposure-response relationship between rubber dust and asbestos with lung cancer mortality was observed in Germany,8 but not in Poland.7 While some studies suggested possible associations between lung cancer mortality and N-nitrosamines exposure,9 others did not.8 10 11 Excess risk of stomach cancer has been found among workers early in the production line and in curing where rubber fumes exposure is highest, as well as those exposed to talc.1 Nevertheless, stomach cancer mortality was not associated with higher levels of inhalable aerosols7 or N-nitrosamines.8 10 11 Increased risk for leukaemia was associated with benzene and butadiene exposures,1 12 but mortality from leukaemia and other malignant lymphomas were not associated with inhalable aerosols among male rubber factory workers.7

    IARC further reported limited, insufficient or inconsistent evidence for associations between rubber dust, rubber fumes and nitrosamines and several cancers.1 Oesophageal cancer mortality was found to have an exposure-response relationship with N-nitrosamines exposure.8 Non-statistically significant increases in brain and prostate cancer mortality were observed with higher exposures to N-nitrosamines.8 No evidence of associations between prostate and brain cancer mortality with rubber dust were found in the Polish industry.7 The relationship between pancreatic cancer and occupational exposure is unclear; small increased risks of incidence and mortality have been documented13–15 but not consistently.7 16 Laryngeal cancer mortality was found in excess in the highest exposure categories of rubber dust in Germany,11 but not in Poland.7 Evidence of an exposure-response association between liver cancer mortality and rubber dust was not statistically significant, possibly due to a low number of deaths.7 Possible explanations for these discrepancies may lie in differences in the industry and chemicals used across countries and time periods. Further difficulties include measurement error in the assessment of personal exposures over long periods of time.1

    Previous studies in the British rubber industry have found similar trends of excess mortality or incidence of cancers of the bladder, particularly before 1949 due to exposures to 1-naphthylamine and 2-naphthylamine,3 lung due to exposures to rubber fumes5 and stomach as associated with rubber dust.5 Nevertheless, measurements of exposures to carcinogenic agents were only indirectly assessed through employment characteristics such as job title and work hours.

    The current study aims to assess the specific association between cancer mortality and cumulative occupational exposures to rubber dust, rubber fumes and N-nitrosamines with follow-up of 49 years and a 94% mortality rate.

    Materials and methods

    Population

    We used data from a cohort of male UK rubber factory workers aged 35 years or older as of 1 February 1967 (n=36 443 from 381 factories) followed up for mortality to December 2015. The median employment start year was 1954 (mean=1951.8, IQR 1946–1961). The cohort was set up by the predecessor to UK Health and Safety Executive.4 5 17

    Exposure assessment

    Exposure assessment was based on estimates from the EU-EXASRUB database of measurements of compounds in rubber factories in Europe.18 Linear mixed effects models with random factory intercepts were used to estimate average exposure in each year between 1915 and 2000 and build a job-exposure matrix for rubber dusts, rubber fumes and N-nitrosamines.19 N-nitrosamines included in this study are N-nitrosodimethylamine (NDMA) and N-nitrosomorpholine (NMor) and N-nitrosamines sum score (NSS), a sum of NDMA, NMor, N-nitrosodibutylamine (NDBA), N-nitrosodiethylamine and N-nitrosopiperidine. Because only job information in 1967 was available, the primary analyses assumed all subjects remained in the same factory department (ie, not necessarily in the same job) throughout their careers and were employed until retirement at age 70, death or emigration. Lifetime cumulative exposures (LCE) to rubber dust, rubber fumes and N-nitrosamines were calculated for each worker based on the assumed number of years worked and department. Sensitivity analyses using different backcasting assumptions of constant average exposure level from the first year of available measurement were shown to have only minimal impact on individual cumulative exposure estimates (data not shown).

    Statistical methods

    To examine the probability of dying from specific causes in a cohort with nearly complete mortality (94.1%), competing risk survival analysis was used to model time to death either from specific cancers, a competing event (death by another cause) or censored due to attrition (such as through emigration). Following the method by Fine and Gray,20 the current model is specified as:

    Embedded Image

    where Embedded Image  is the subdistribution hazard20 of cause k at time t, T is time of the first observed event and D is a random variable denoting type of event occurring. Subjects who experienced a competing event before the event of interest remain in the risk set and are weighted using the inverse probability of censoring weighting approach.21 22 This is in contrast to a standard Cox proportional hazard approach which would consider deaths from competing risks to be censored and would be removed from the risk set. Censoring competing events violates the assumption that censoring occurred at random and is independent from the risk of dying from the cause of death of interest, leading to a biased Kaplan-Meier estimator.23 Furthermore, within the context of competing risks, the interpretation of HRs from a standard Cox proportional hazard approach changes to the hazards of dying if no other deaths occurred,21 which is untenable in a cohort with 94.1% mortality rate. Subdistribution HRs (SHRs) are estimated using stcrreg in Stata V.1524 and comparable in interpretation to proportional HRs in Cox models.25

    Cumulative exposure was divided into four quartiles and as a continuous metric to assess linearity of the exposure-response association. Outcomes are mortality from cancers previously associated with the rubber industry1: all cancers, cancers of the bladder, lung, stomach, multiple myeloma, leukaemia, larynx, oesophagus, prostate, non-Hodgkin’s lymphoma, pancreas, brain, liver and in situ, benign or unknown behaviour neoplasms.

    Analyses were adjusted for birth year and LCE (in quartiles) to rubber dust, rubber fumes or N-nitrosamines. Because the baseline mean age of the sample is 50.1 years and median ages at death from cancer among males are between ages 60 and 75 years26, or 10 and 25 years after the baseline, and a lag or latency period between exposure and cancer development has been reported elsewhere,27–31 the analyses used 15-year lags, approximately around the midpoint of median cancer deaths.26 Analyses without lags for multiple myeloma, non-Hodgkin’s lymphoma and leukaemia deaths were tested and found lower effect sizes but similar directions and p values to lagged analyses.

    Sensitivity analyses were conducted with alternative simulated employment durations based on information on employment durations from another, partly overlapping, cohort of British rubber factory workers.4 5 32 Ten simulated employment durations for each worker were generated from log-normal distributions (mean=2.355, SD=0.470) that for the whole sample had similar characteristics to this cohort; the 47th percentile of the distribution have 10 years of employment and 88th percentile have 18 or more years. Log-normal distributions were chosen over other functional forms because they provided the best fit to these parameters. Lifetime cumulative exposures were subsequently recalculated and competing risk survival analyses were performed for each of the 10 simulated employment durations to allow for assessment of the variability in the SHRs for the LCEs. Sensitivity analyses were conducted for all cancers and cancers of the bladder, lung, stomach, myeloma and leukaemia. Results of the primary analyses were mostly supported by the sensitivity analyses, including for all malignant neoplasms for all agents.

    Results

    The cohort included 36 441 male workers in the UK rubber industry followed from 1967 to 2015 with 880 794 person years of time at risk and under observation. International Statistical Classification of Diseases and Related Health Problems codes and number of deaths are provided in OSM. We describe risk patterns as absence of exposure-response, linear (in log-space) exposure-response (increasing risk in quartiles (Q) 1–4), plateauing exposure-response (increasing risk in Q1–3, but plateau or reduction in Q4), and increased risk without an exposure-response pattern. Bladder cancer mortality represented 4.7% of all cancer deaths (n=417). The primary analyses (tables 1–2) showed a linear exposure-response relationship for NDMA (SHRs up to 2.82 (Q4) and NMor (SHRs up to 2.59 (Q4), plateauing exposure-response for NSS (SHRs up to 2.19 (Q3)) and increased risks without exposure-response pattern for rubber dust (SHRs up to 2.56 (Q4)). Sensitivity analyses (online suplementary materials) supported the primary analyses for all exposures. However, results for Q2 were not fully supported by the sensitivity analyses for NDMA where SHRs ranged from 1.14 (95% CI 0.85 to 1.52) to 1.56 (95% CI 1.18 to 2.06), and for NSS, where SHRs ranged from 1.26 (95% CI 0.95 to 1.68) to 1.51 (95% CI 1.15 to 1.99). Lung cancer mortality comprised the largest proportion of cancer deaths in the cohort at 37.1% (n=3377). A linear exposure-response relationship was found for exposures to rubber dust (SHRs up to 1.44 (Q4)). Sensitivity analyses supported this exposure-response pattern where Q4 SHRs ranged from 1.21 (95% CI 1.10 to 1.33) to 1.27 (95% CI 1.15 to 1.39), which is within the 95% CI of the primary analysis (1.31 to 1.58). Increased risks were found for NDMA exposure (SHRs up to 1.70 (Q4)). Plateauing exposure-response was observed for rubber fumes (SHRs up to 1.55 (Q3)), NSS (SHRs up to 1.60 (Q3)) and NMor (SHRs 1.19 (Q2 and Q3)). Sensitivity analyses support the exposure-response patterns observed for the primary analyses (OSM). Other differences between SHRs in the sensitivity and primary analyses were minimal with the largest difference found was for NDMA Q4 where primary analysis SHR=1.70 (95% CI: 1.41 to 2.10) and sensitivity analyses SHRs range from 1.40 (95% CI 1.27 to 1.54) to 1.46 (95% CI 1.32 to 1.60).

    Supplemental material

    View this table:
    Table 1

    Risk of death from selected cancers by 15-year lagged cumulative exposure to rubber dust and rubber fumes in a cohort of rubber factory workers in the UK

    View this table:
    Table 2

    Risk of death from selected cancers by 15-year lagged cumulative exposure to nitrosamines sum score, NDMA and NMor in a cohort of rubber factory workers in the UK

    For stomach cancer mortality (n=768, 8.4% of all cancer deaths), a linear exposure-response relationship was found for exposures to NDMA (SHRs up to 1.72 (Q4)) and NMor (SHRs up to 1.49 (Q4)). However, this exposure-response relationship was not supported by the sensitivity analyses. Rather, only elevated SHRs for Q3 (NDMA, NMor) and Q4 (NMor) were consistently found across the simulations. Increased risks were observed for rubber dust (SHR up to 1.78 Q3)), rubber fumes (SHRs up to 1.75 (Q3)) and NSS (SHRs up to 1.78 (Q3)). This was supported by the sensitivity analyses.

    A linear exposure-response association was found for leukaemia mortality (n=195, 2.1% of all cancer deaths) and NDMA exposures (p for trend <0.001) (SHRs up to 3.47 (Q4)). Plateauing exposure-response were found for rubber dust (SHRs up to 2.42 (Q3)) and NSS (SHRs up to 3.08 (Q3)). Increased risks were found for rubber fumes (SHRs up to 2.10 (Q3 and Q4)) and NMor (SHRs up to 1.96 (Q4)). Sensitivity analyses results show consistently elevated and statistically significant SHRs for Q3 for NSS and Q4 for rubber dust, rubber fumes and NMor.

    A linear exposure-response relationship was found for multiple myeloma mortality (n=462, 5.1% of all cancer mortality) with exposures to NDMA (SHRs up to 2.81 (Q4), p for linear trend <0.01) and NMor (SHRs up to 1.82 (Q4), p for linear trend <0.01). Sensitivity analyses support this pattern. Plateauing exposure-response was found for rubber fumes (SHRs up to 2.12 (Q3), p=0.04 for trend) and NSS (SHRs up to 2.35 (Q3), p=0.02 for trend). Increased risks were found for rubber dust (SHRs up to 1.99 (Q4)). Sensitivity analyses support the patterns for NSS and rubber fumes exposures and partially support NMor (Q4) and rubber dust (Q3 and Q4).

    Increased risks for non-Hodgkin’s lymphoma mortality (n=141, 1.5% of all cancer mortality) were observed for rubber dust (SHR up to 1.68 (Q3)), rubber fumes (SHRs up to 2.27 (Q3)), NSS (SHRs up to 2.24 (Q3)), NDMA (SHRs up to 2.26 (Q4)) and NMor (SHRs up to 1.58 (Q3)).

    A linear exposure-response relationship for oesophageal cancer mortality (n=333, 3.7% of all cancer deaths) as found for NMor (SHRs up to 2.25 (Q4), p<0.001 for trend) and plateauing exposure-response relationship was found for rubber fumes (SHRs up to 2.55 (Q3)). Increased risks were found for rubber dust (SHRs up to 2.26 (Q3)), NSS (SHRs up to 2.42 (Q3)) and NDMA (SHRs up to 3.04 (Q4)).

    Linear exposure-response relationships between prostate cancer mortality (n=885, 9.7% of all cancer deaths) and exposures to NDMA (SHRs up to 5.36 (Q4)) and NMor (SHRs up to 2.71 (Q4)) were found. Plateauing exposure-response were found for rubber fumes (SHRs up to 4.03 (Q3)) and NSS (SHRs up to 3.75 (Q3)). Increased risks were found for rubber dust (SHRs up to 3.37 (Q3)).

    Results for laryngeal cancer mortality (n=62, 0.7% of all cancer deaths) show elevated SHRs for all exposures, but did not achieve statistical significance, except for NSS (Q3).

    Results for brain cancer mortality (n=106, 1.2% of all cancer deaths) show plateauing exposure-response relationships for NMor (SHRs up to 3.16 (Q3)), elevated SHRs in Q3 for rubber dust (SHR 2.22) and Q4s for NSS (SHR 1.75) and NDMA (SHR 2.50).

    A linear exposure-response relationship was found for pancreatic cancer mortality (n=328, 3.6% of all cancer deaths) and NMor (SHRs up to 1.96 (Q4)). Plateauing exposure-response relationship was found for rubber fumes (SHRs up to 2.48 (Q3)) and increased risks without a trend were found for rubber dust (SHRs up to 2.00 (Q3)), NDMA (SHRs up to 2.60 (Q4)) and NSS (SHRs up to 2.20 (Q2)).

    A linear exposure-response relationship between liver cancer mortality (n=122, 1.3% of all cancer deaths) and exposures to NDMA (SHRs up to 2.86 (Q4)) were observed. Plateauing exposure-response was found for NMor exposure (SHRs up to 1.91 (Q3)). Increased risks were found for rubber dust (SHRs up to 3.40 (Q4)), rubber fumes (SHRs up to 3.35 (Q3)) and NSS (SHRs up to 3.20 (Q3)).

    Discussion

    This study examined cancer mortality risks associated with occupational exposures to rubber dust, rubber fumes and nitrosamines in a large occupational cohort of UK rubber manufacturing workers with 49 years of follow-up (880 794 person years). We found increased risks associated with cumulative exposures to rubber dust, rubber fumes and N-nitrosamines, which can be attributed to the unique combination of very long follow-up of the cohort, 94% mortality and selection of the exposures based on prior knowledge on carcinogenicity.1

    Linear exposure-response relationships were found for cumulative exposure to NDMA (bladder, stomach, leukaemia, multiple myeloma, prostate, liver), NMor (bladder, stomach, multiple myeloma, oesophagus, prostate, pancreas) and rubber dust (lung). Plateauing exposure-response relationships were found for rubber dust (leukaemia), rubber fumes (lung, leukaemia, multiple myeloma, oesophagus, prostate, pancreas), NSS (bladder, lung, leukaemia, multiple myeloma, prostate) and NMor (lung, brain, liver). Increased risks were found for rubber dust (bladder, stomach, multiple myeloma, non-Hodgkin’s lymphoma, oesophagus, prostate, pancreas, brain, liver), rubber fumes (bladder, stomach, leukaemia, non-Hodgkin’s lymphoma, liver), NSS (stomach, non-Hodgkin’s lymphoma, brain, liver), NDMA (lung, non-Hodgkin’s’ lymphoma, oesophagus, brain, pancreas) and NMor (leukaemia, non-Hodgkin’s lymphoma). Occupational exposures were not associated with laryngeal cancer mortality.

    Exposure-response associations between occupational exposures to nitrosamines and mortality from oesophageal and prostate cancers have been reported previously.8 11 We additionally observed increased stomach cancer risks with NDMA and NMor, in contrast to several studies8 11 and consistent with others.10 33

    Although this study found exposure-response associations between lung cancer mortality and occupational exposure to rubber dust and elevated risks associated with rubber fumes, NSS, NDMA and NMor, several previous studies did not observe associations between lung cancer mortality and N-nitrosamine exposure8 10 11 and others found inconsistent evidence on associations with rubber dust.7 8 It is possible that these results are capturing effects of smoking on lung cancer mortality rather than occupational exposures because individual smoking histories were unavailable. To obtain some indication of the possible confounding effect of smoking in this cohort, we used a statistical external adjustment method.34 External Monte Carlo analyses based on information on smoking prevalence, ex-smokers and never smokers from a cohort of rubber industry entrants after 198218 indicated that mean bias is only 1.6% compared with the general population. To achieve as much as 10% (attenuating) bias internally, 12%–14% more smokers and 12%–14% fewer never smokers would have to be present in Q4 compared with Q1 (data not shown), which seems unlikely; suggesting that confounding by smoking in this cohort was likely not a significant confounding factor.

    There are several strengths to this study. First, the 49-year follow-up period constitutes the longest in cohort studies of rubber workers in the UK. With a nearly entirely deceased cohort, these analyses provide the most precise and complete lifetime risks from exposures encountered in the rubber manufacturing industry. Second, instead of qualitative information on jobs, exposure assessments from a quantitative job-exposure matrix19 based on historic exposure measurements previously collated in the EXASRUB database18 were used.

    Aside from these strengths, several limitations exist in this study. Information regarding individual employment histories were unavailable before 1967 and during the follow-up period. As such, the main analysis assumed continuous employment until retirement at age 70, emigration or death. To address this limitation, we conducted sensitivity analyses with varying employment durations, which supported the main results of this study in all but a few instances. Differences observed between the results of the main analyses and the sensitivity analyses imply that results for low incidence cancers such as leukaemia (n=195) are more sensitive than high-incidence cancers such as lung cancer (n=3377) to the length of occupational exposure to each agent, which is further dependent on employment duration. Second, we assumed a 15-year lag between exposure and clinical manifestation of the cancer. However, these lags are not uniform across all cancers and are generally somewhat shorter for bloodborne cancers.35 Nonetheless, as an overall approximation 15 years seems appropriate. Third, although some cancers are less fatal than others and this study used underlying cause of death from the death certificate without any cancer incidence data, some cancers may have been undercounted and comparison to cancer mortality in the general population would yield lower rates. However, results for selected cancers from SMR analyses of the same data33 showed either higher cancer mortality or no difference from the general population. Fourth, information on important confounders such as smoking and other lifestyle factors were unavailable, although additional simulations indicated smoking was unlikely to be a significant confounding factor. Fifth, the cohort was subject to selection effects at recruitment because workers had to survive until age 35 to be included in the cohort. Sixth, the JEM provides estimates of average exposure for workers which introduces measurement error in individual assessments. However, group-based exposure assessment generally leads to Berkson-type rather than classical measurement error, which results in attenuation of exposure-response associations rather than biased results.36 Finally, cross-contamination between departments could not rule out the need for multipollutant models, but given the high correlations between exposures this requires different and complex statistical modelling, with currently unknown validity in this context.

    Conclusions

    In summary, we examined the exposure-response association between occupational exposures to rubber dust, rubber fumes and nitrosamines with cancer mortality in a cohort of 36 441 UK rubber factory workers with a follow-up from 1967 to 2015. Consistent with previous studies, N-nitrosamines exposures in the rubber industry, were associated with mortality from cancers of the bladder, lung, stomach, leukaemia, multiple myeloma, oesophagus, prostate, pancreas and liver. We also found linear exposure-response associations where the highest exposures to NMor more than double the risks for mortality from cancers of the bladder, oesophagus and prostate, similarly to the highest exposures of NDMA and cancers of the bladder, leukaemia, multiple myeloma, prostate, liver. Linear exposure-response relationships resulting in modest increased risks were also found for exposures to NMor and cancers of the stomach, multiple myeloma and pancreas and NDMA with cancers of the stomach. Furthermore, exposures to rubber dust and rubber fumes were found to be associated with higher risks for these cancers as well. Results from this study contribute to the evidence of elevated cancer mortality risks from occupational exposures in the rubber industry by further clarifying the relationship between each carcinogen and cancer with implications for the industry today where occupational exposures to N-nitrosamines continues to persist, although at greatly reduced levels compared with several decades ago.

    Acknowledgments

    The authors would like to thank Professors Richard Morris and Margaret May for statistical advice. The authors would also like to thank staff at the Health and Safety Executive, The University of Manchester and the Institute of Occupational Medicine for computerising thousands of study cards.

    References

    1. ↵
      IARC Working Group. Chemical Agents and Related Occupations IARC Monographs of the Evaluation of Carcinogenic Risks to Humans. Lyon, France: IARC, 2012.
    2. ↵
      2. Kogevinas M ,
      3. Sala M ,
      4. Boffetta P , et al
      . Cancer risk in the rubber industry: a review of the recent epidemiological evidence. Occup Environ Med 1998;55:1–12.doi:10.1136/oem.55.1.1
      OpenUrlAbstract/FREE Full Text
    3. ↵
      2. CASE RA ,
      3. Hosker ME ,
      4. McDONALD DB , et al
      . Tumours of the urinary bladder in workmen engaged in the manufacture and use of certain dyestuff intermediates in the British chemical industry. I. The role of aniline, benzidine, alpha-naphthylamine, and beta-naphthylamine. Br J Ind Med 1954;11:75–104.doi:10.1136/oem.11.2.75
      OpenUrlPubMedWeb of Science
    4. ↵
      2. Sorahan T ,
      3. Parkes HG ,
      4. Veys CA , et al
      . Cancer mortality in the British rubber industry: 1946-80. Br J Ind Med 1986;43:363–73.doi:10.1136/oem.43.6.363
      OpenUrlPubMedWeb of Science
    5. ↵
      2. Sorahan T ,
      3. Parkes HG ,
      4. Veys CA , et al
      . Mortality in the British rubber industry 1946-85. Br J Ind Med 1989;46:1–10.doi:10.1136/oem.46.1.1
      OpenUrlPubMedWeb of Science
    6. ↵
      2. Delzell E ,
      3. Monson RR
      . Mortality among rubber workers: VIII. Industrial products workers. Am J Ind Med 1984;6:273–9.doi:10.1002/ajim.4700060405
      OpenUrlPubMed
    7. ↵
      2. de Vocht F ,
      3. Sobala W ,
      4. Wilczynska U , et al
      . Cancer mortality and occupational exposure to aromatic amines and inhalable aerosols in rubber tire manufacturing in Poland. Cancer Epidemiol 2009;33:94–102.doi:10.1016/j.canep.2009.06.013
      OpenUrlPubMed
    8. ↵
      2. Straif K ,
      3. Keil U ,
      4. Taeger D , et al
      . Exposure to nitrosamines, carbon black, asbestos, and talc and mortality from stomach, lung, and laryngeal cancer in a cohort of rubber workers. Am J Epidemiol 2000;152:297–306.doi:10.1093/aje/152.4.297
      OpenUrlCrossRefPubMedWeb of Science
    9. ↵
      2. Li K ,
      3. Yu S
      . Economic status, smoking, occupational exposure to rubber, and lung cancer: a case-cohort study. J Environ Sci Health C Environ Carcinog Ecotoxicol Rev 2002;20:21–8.doi:10.1081/GNC-120003926
      OpenUrlCrossRefPubMedWeb of Science
    10. ↵
      2. Straif K ,
      3. Chambless L ,
      4. Weiland SK , et al
      . Occupational risk factors for mortality from stomach and lung cancer among rubber workers: an analysis using internal controls and refined exposure assessment. Int J Epidemiol 1999;28:1037–43.doi:10.1093/ije/28.6.1037
      OpenUrlCrossRefPubMedWeb of Science
    11. ↵
      2. Straif K ,
      3. Weiland SK ,
      4. Bungers M , et al
      . Exposure to high concentrations of nitrosamines and cancer mortality among a cohort of rubber workers. Occup Environ Med 2000;57:180–7.doi:10.1136/oem.57.3.180
      OpenUrlAbstract/FREE Full Text
    12. ↵
      2. Cheng H ,
      3. Sathiakumar N ,
      4. Graff J , et al
      . 1,3-Butadiene and leukemia among synthetic rubber industry workers: exposure-response relationships. Chem Biol Interact 2007;166:15–24.doi:10.1016/j.cbi.2006.10.004
      OpenUrlCrossRefPubMed
    13. ↵
      2. Andreotti G ,
      3. Silverman DT
      . Occupational risk factors and pancreatic cancer: a review of recent findings. Mol Carcinog 2012;51:98–108.doi:10.1002/mc.20779
      OpenUrlCrossRefPubMed
    14. ↵
      2. Delzell E ,
      3. Monson RR
      . Mortality among rubber workers: IX. Curing workers. Am J Ind Med 1985;8:537–44.doi:10.1002/ajim.4700080605
      OpenUrlPubMed
    15. ↵
      2. Li K ,
      3. Yu S
      . A nested case-control study on risk of pancreatic cancer among workers in the rubber industry. Pancreas 2002;24:417–8.doi:10.1097/00006676-200205000-00016
      OpenUrlPubMed
    16. ↵
      2. Fritschi L ,
      3. Benke G ,
      4. Risch HA , et al
      . Occupational exposure to N-nitrosamines and pesticides and risk of pancreatic cancer. Occup Environ Med 2015;72:678–83.doi:10.1136/oemed-2014-102522
      OpenUrlAbstract/FREE Full Text
    17. ↵
      2. Dost A ,
      3. Straughan J ,
      4. 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;57:186–90.doi:10.1093/occmed/kql171
      OpenUrlCrossRefPubMed
    18. ↵
      2. De Vocht F ,
      3. Straif K ,
      4. 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.doi:10.1093/annhyg/mei035
      OpenUrlCrossRefPubMedWeb of Science
    19. ↵
      2. Hidajat M ,
      3. McElvenny DM ,
      4. Mueller W , et al
      . Job-exposure matrix for historical exposures to rubber dust, rubber fumes and n-Nitrosamines in the British rubber industry. Occup Environ Med 2019;76:259–67.doi:10.1136/oemed-2018-105182
      OpenUrlAbstract/FREE Full Text
    20. ↵
      2. Fine JP ,
      3. Gray RJ
      . A proportional hazards model for the subdistribution of a competing risk. J Am Stat Assoc 1999;94:496–509.doi:10.1080/01621459.1999.10474144
      OpenUrlCrossRefWeb of Science
    21. ↵
      2. Haller B ,
      3. Schmidt G ,
      4. Ulm K
      . Applying competing risks regression models: an overview. Lifetime Data Anal 2013;19:33–58.doi:10.1007/s10985-012-9230-8
      OpenUrlCrossRefPubMedWeb of Science
    22. ↵
      2. Robins JM ,
      3. Rotnitzky A
      . Recovery of information and adjustment for dependent censoring using surrogate markers AIDS epidemiology: Springer, 1992:297–331.
    23. ↵
      2. Putter H ,
      3. Fiocco M ,
      4. Geskus RB
      . Tutorial in biostatistics: competing risks and multi-state models. Stat Med 2007;26:2389–430.doi:10.1002/sim.2712
      OpenUrlCrossRefPubMedWeb of Science
    24. ↵
      StataCorp. Stata 15 Base Reference Manual. College Station, TX: Stata Press, 2017.
    25. ↵
      2. Ingle S ,
      3. Margaret M ,
      4. Uebel K , et al
      . Outcomes in patients waiting for antiretroviral treatment in the Free State Province, South Africa: prospective linkage study. 24. London, England: AIDS, 2010:2717.
      OpenUrl
    26. ↵
      Cancer Research UK. Cancer mortality by age. http://www.cancerresearchuk.org/health-professional/cancer-statistics/mortality/age (Accessed Mar 2018).
    27. ↵
      2. Checkoway H ,
      3. Pearce N ,
      4. Hickey JLS , et al
      . Latency analysis in occupational epidemiology. Arch Environ Health 1990;45:95–100.doi:10.1080/00039896.1990.9935932
      OpenUrlCrossRefPubMedWeb of Science
    28. ↵
      2. Hodgson JT ,
      3. McElvenny DM ,
      4. Darnton AJ , et al
      . The expected burden of mesothelioma mortality in Great Britain from 2002 to 2050. Br J Cancer 2005;92:587–93.doi:10.1038/sj.bjc.6602307
      OpenUrlPubMedWeb of Science
    29. ↵
      2. Langholz B ,
      3. Thomas D ,
      4. Xiang A , et al
      . Latency analysis in epidemiologic studies of occupational exposures: application to the Colorado Plateau uranium miners cohort. Am J Ind Med 1999;35:246–56.doi:10.1002/(SICI)1097-0274(199903)35:3<246::AID-AJIM4>3.0.CO;2-6
      OpenUrlCrossRefPubMedWeb of Science
    30. ↵
      2. Richardson DB ,
      3. Cole SR ,
      4. Chu H , et al
      . Lagging exposure information in cumulative exposure-response analyses. Am J Epidemiol 2011;174:1416–22.doi:10.1093/aje/kwr260
      OpenUrlCrossRefPubMedWeb of Science
    31. ↵
      2. Salvan A ,
      3. Stayner L ,
      4. Steenland K , et al
      . Selecting an exposure lag period. Epidemiology 1995;6:387–90.doi:10.1097/00001648-199507000-00010
      OpenUrlPubMedWeb of Science
    32. ↵
      2. Sorahan T ,
      3. Hamilton L ,
      4. Jackson JR
      . A further cohort study of workers employed at a factory manufacturing chemicals for the rubber industry, with special reference to the chemicals 2-mercaptobenzothiazole (MBT), aniline, phenyl-beta-naphthylamine and o-toluidine. Occup Environ Med 2000;57:106–15.doi:10.1136/oem.57.2.106
      OpenUrlAbstract/FREE Full Text
    33. ↵
      2. McElvenny D ,
      3. Mueller W ,
      4. Ritchie P , et al
      . Cohort study of workers in the British rubber industry: mortality follow-up extended to 49 years. Occup Environ Med 2018;75:848–55.
      OpenUrlAbstract/FREE Full Text
    34. ↵
      2. Steenland K ,
      3. Greenland S
      . Monte Carlo sensitivity analysis and Bayesian analysis of smoking as an unmeasured confounder in a study of silica and lung cancer. Am J Epidemiol 2004;160:384–92.doi:10.1093/aje/kwh211
      OpenUrlCrossRefPubMedWeb of Science
    35. ↵
      2. Nadler DL ,
      3. Zurbenko IG
      . Estimating cancer latency times using a weibull model. Adv Epidemiol 2014;2014:1–8.doi:10.1155/2014/746769
      OpenUrl
    36. ↵
      2. Burstyn I ,
      3. de Vocht F ,
      4. Kim H-M , et al
      . Bayesian correction for measurement error following group-based exposure assessment in a case-referent study. Occup Environ Med 2011;68:A44.doi:10.1136/oemed-2011-100382.145
      OpenUrlAbstract/FREE Full Text

    Footnotes

    • Contributors FdV and DMMcE conceived of the study. FdV, DMMcE, AD, RMA and JWC obtained funding. FdV, DMMcE, PR, AD and WM worked on coding of data not in digital format. MH conducted the statistical analyses and wrote the first draft version of the manuscript. All authors contributed to interpretation of the results and commented on the various iterations of the manuscript. All authors approved the final version.

    • Funding This study was funded by Cancer Research UK (C29425/A16521). Additional funding for tracing of the cohort was provided by the UK Health and Safety Executive (PRJ787).

    • Disclaimer The views expressed are those of the author(s) and not necessarily those of the Cancer Research UK or UK Health and Safety Executive.

    • Competing interests None declared.

    • Ethics approval The study obtained clearances from an NHS ethics committee (Ref: 13/NW/0543), the Health Research Authority’s Confidentiality Advisory Group (Ref: CAG 5-08(d)/2013), the Office for National Statistics and NHS Digital’s Data Access Advisory Group (now Independent Group Advising on the Release of Data, Ref: NIC-323309-L2G9T).

    • Provenance and peer review Not commissioned; externally peer reviewed.

    • Data sharing statement Use of these data was granted by the NHS ethics committee, the Health Research Authority’s Confidentiality Advisory Group, the Office for National Statistics and NHS Digital’s Data Access Advisory Group (now Independent Group Advising on the Release of Data) for the specific purpose of this study only.

    • Patient consent for publication Not required.

    Linked Articles