About the research
The University of Birmingham, supported by the old National Health Service Central Register (NHSCR) at Southport, carried out a standard occupational cohort study which investigates the incidence of myelodysplastic syndrome (MDS) experienced by cohorts of 16,467 petroleum distribution workers and 28,554 oil refinery workers. MDS are a group of cancers in which immature blood cells in the bone marrow do not mature and therefore do not become healthy blood cells. This report was also sponsored by the UK Energy Institute.
This type of study used to be a mainstay of research and is a good example of how follow-up of a cohort needs to be maintained to get reliable results. It also demonstrates how findings from a study in one part of the world leads to related studies being carried out in other parts. In 2012 Schnatter et al. found a statistically significant, monotonic association between the risks of MDS and estimates of cumulative benzene exposure in analysis of occupational cohorts from Australia, Canada and the United Kingdom.
The UK element of the nested case-control study was based on a cohort of UK petroleum distribution workers, which has recently been updated along with a cohort of UK oil refinery workers. Follow-up data are now available for deaths (1951-2011) and cancer registrations (1971-2011). This report presents findings for MDS to see whether simple national comparisons support the recent hypothesis that low-level benzene exposures in the petroleum industry may have a discernible influence on the risks of MDS.
The study subjects were all those male employees first employed at one of 476 UK petroleum distribution centres or eight UK oil refineries in the period of 1946-1974; all subjects had a minimum of twelve months employment with some employment after 1st January 1951. The overall Standardised Registration Ratio (SRR) findings did not provide clear evidence for the presence of an occupational cancer hazard, and provided no support for the hypothesis that low-level benzene exposure has an important effect on the risks of MDS.
The study appears to challenge the results from the 2012 Schatter et al earlier research, although the new study is not comprehensive and does not include MDS health data prior to 1995.
Methodology
The cohort studies used six oil refineries and 403 petroleum distribution centres in England and Wales. The remaining two oil refineries and 73 distribution centres were located in Scotland. The revised cohorts contained dates of birth, work history information, and follow-up information for 28,554 oil refinery and 16,467 petroleum distribution workers first employed in the period 1946-1974. The study subjects had to have at least twelve months employment with some employment after 1 January 1951.
Registration data collected by the General Register Office is provided to the Office for National Statistics (ONS) where the data is coded, additional data added and made secure for research purposes. Registration data including date, cause of death and cancer site and type were made available to the University of Birmingham, based on the pre-defined employment selection described in the previous paragraph.
Expected numbers of registrations were calculated from male incidence rates, and these were compared against observed numbers. Standardised Registration Ratios (SRRs) were calculated as a ratio of observed (Obs) to expected (Exp) numbers, and the significance of differences between them was assessed by a Poisson distribution. As MDS is deemed a disease of old age, the above analysis was carried out for the age range 15-84 and also 15-99y. This was also split between petroleum distribution workers and oil refinery workers.
Sub-group analyses were carried out to investigate any evidence for trends in the pattern of the tendency for SRRs to increase or decrease based on time since first employment. Therefore, four sets of SRRs were investigated; year of registration, period from hire (y), period of commencing employment and duration of employment.
Research findings
The detailed trend analyses are presented into sub-group tables and split into the four defined SRRs. Trends for each SRR are analysed, with 95% confidence intervals (CI) and p-value for trend highlighted.
The analyses carried out on the SRRs for petroleum distribution workers showed no suggestion of an overall increased incidence of MDS. The results shown for oil refinery workers present some more significant numbers. The results presented in the oil refinery tables are outlined below, where the SRR is presented as a percentage:
Age range 15-84:
- no suggestion of an overall increased incidence of MDS
- obs 29, SRR 81, 95% CI 54 to 116
- negative trend for period form hire approached statistical significance (lower SIRs with longer periods form hire)
- negative trend for duration of employment was statistically significant where p=0.03
Age range 15-99:
- no suggestion of an overall increased incidence of MDS
- obs 36, SRR 83, 95% CI 58 to 115
- no significant trends in any of the SRRs
- negative trend for duration of employment was of borderline statistical significance where p=0.05
The University of Birmingham states that the overall SRR findings were not influenced by a major healthy worker effect. The SRR for neoplasm excluding non-melanoma skin cancer in the period 1971 to 2011 and for ages 15-84 was 105 in distribution workers (95% CI 102 to 108) based on 4,344 registrations, and 97 in refinery workers (95% CI 94 to 99) based on 6,394 registrations.
The study has found no overall excess of MDS incidence in historical cohorts of UK oil refinery and petroleum distribution workers, in comparison with rates in the general population and does not support the hypothesis that low-level benzene exposure has a discernible influence on the risks of MDS.
Impact
This piece of research challenges the results of the earlier analysis carried out by Schnatter et al who suggested that MDS may be a more important health risk than Acute Myeloid Leukaemia (AML) for lower benzene exposures. However the overall findings in this report suggest that this is not the case for these cohorts of UK petroleum distribution workers and oil refinery workers. There is no reason to believe that benzene exposures in the UK petroleum industries were lower than those in Canada or Australia, but further evidence on this topic would be beneficial.
The UK study argued that if the nested case-control study has correctly identified that a large proportion of a cohort is at a highly elevated risk of some disease, then an excess for this disease should be apparent in simple national comparisons for the overall cohort. It will be important to know whether there is any overall excess of MDS based on national comparisons in those Canadian and Australian cohorts. If national comparisons of these cohorts find no excess, then the findings of Schnatter et al. cannot be regarded as reflecting causality.
It would also be helpful to have information on MDS risks in those members of the Chinese benzene cohort with lower levels (say < 10 parts per million [ppm].year) of benzene exposure. Data on MDS mortality are already available from a US cohort of 2266 workers exposed to benzene in a chemical manufacturing plant. The single death from MDS occurred in the highest (>25 ppm.year) of the three selected exposure groups; the expected number of MDS deaths in the entire cohort was 0.15, however it would be dangerous to draw any conclusion from a single death. An expansion of this US study to include cancer incidence data would be helpful and informative for this study.
It is important to note the limitations of the study; most notably, that the absence of MDS health outcome data prior to 1995, leading to no data on observed or expected numbers in the first twenty years from hire. University of Birmingham notes the possibility that an MDS effect from benzene exposure is only found in this early period of follow-up and consequently may have been missed. Further clarity and investigation in this area would be a good direction for future research.
A Working Group from the International Agency for Research on Cancer (IARC) met recently to review the entire worldwide literature on cancer risks and benzene exposure, and a monograph prepared by the Working Group (Volume 120) will be published in 2018. It will be interesting to see how the UK findings are interpreted by IARC.
Read the Report
Publications and outputs
Lymphatic and hematopoietic cancers among benzene-exposed workers
