Background: Workers on coke oven plants may be exposed to potentially carcinogenic polycyclic aromatic hydrocarbons (PAHs), particularly during work on the ovens tops. Two cohorts, employees of National Smokeless Fuels (NSF) and the British Steel Corporation (BSC) totalling more than 6,600 British coke plant workers employed in 1967, had been followed up to mid-1987 for mortality. Previous analyses suggested an excess in lung cancer risk of around 25%, or less when compared with Social Class IV ('partly skilled').Analyses based on internal comparisons within the cohorts identified statistical associations with estimates of individual exposures, up to the start of follow-up, to benzene-soluble materials (BSM), widely used as a metric for mixtures of PAHs. Some associations were also found with times spent in certain coke ovens jobs with specific exposure scenarios, but results were not consistent across the two cohorts and limitations in the exposure estimates were noted. The present study was designed to reanalyse the existing data on lung cancer mortality, incorporating revised and improved exposure estimates to BSM and to benzo[a]pyrene (B[a]P), including increments during the follow-up and a lag for latency.
Methods: Mean annual average concentrations of both BSM and B[a]P were estimated by analysis of variance (ANOVA) from concentration measurements at all NSF and six BSC plants, and summarised by job and plant, with a temporal trend (for the BSM only). These were combined with subjects' work histories, to produce exposure estimates in each year of follow-up, with a 10-year lag to allow for latency. Exposures to BSM and to B[a]P were sufficiently uncorrelated to permit analysis in relation to each variable separately.Lung cancer death risks during the follow-up were analysed in relation to the estimated time-dependent exposures, both continuous and grouped, using Cox regression models, with adjustment for age.
Results: Changing the exposure estimates changed the estimated relative risks compared with earlier results, but the new analyses showed no significant trends with continuous measures of exposure to either BSM or B[a]P, nor with time spent on ovens tops. Analyses with grouped exposures showed mixed results. Across all BSC plants, the relative risk coefficient for working 5 or more years on ovens tops, where the exposures were highest, was 1.81, which was statistically significant. However, results for those with 0-5 years on ovens tops did not suggest a trend; the evidence for an underlying relationship was thus suggestive but not strong.
Conclusions: The new results are in line with previous findings; they show some signs consistent with an effect of coke ovens work on lung cancer risk, especially on ovens tops, but the preponderant absence of significant results, and the inconsistencies between results for NSF and BSC, highlight how little evidence there is in these data of any effect.