|Home | About | Journals | Submit | Contact Us | Français|
S. J. Hutchings1, L. Rushton1, T. Brown2. 1Department of Epidemiology and Public Health, Imperial College London; 2Health and Safety Laboratory, Buxton, Derbyshire
ObjectivesThe aim of this project is to update the 1981 Doll and Peto estimates of attributable burden of cancer due to occupation for Great Britain. Estimates of the relative contributions of different occupational exposures are also required.
MethodsThe primary measure of burden used in the project is the attributable fraction (AF) which combines the relative risk associated with exposure (RR) with the proportion exposed (Pr(E)). For each cancer/exposure pairing, a “best epidemiological study” is identified for the RR estimate, and depending on the source of the estimate (industry‐ or population‐based studies, meta‐analysis or review), either Levin's equation, which requires an independent estimate of Pr(E), or Miettinen's equation, which uses an internal (study‐based) estimate of the proportion of the cases exposed (Pr(E|D)), is used. Where possible, RRs by specific exposure level are used, but in general exposed numbers are split between high and low exposure levels that match the “best study” exposure scenarios. For Levin's equation, an estimate of the numbers ever exposed over the relevant exposure period (REP), (which depends on the cancer latency) is required. CAREX data are the main source, or where these are not available, estimates are obtained from national surveys (the Labour Force Survey and Census of Employment). An employee turnover factor is used to estimate numbers ever exposed, and national estimates of the population of working age during the REP are used for the proportion denominator.
ResultsIn practice, RRs from meta‐analyses and reviews and UK industry‐based cohorts have nearly always been used, therefore requiring Levin's equation for the AF. Estimates by defined exposure level could rarely be used, as matching estimates of numbers exposed were not generally available. Limited data on cancer latency is also a weakness.
ConclusionA robust and transparent method of estimating occupation attributable fraction has been developed for Great Britain. Future work will include the development of “credibility intervals” for the AFs which take account of bias uncertainty as well as random error, and the prediction of future cancer burden.
Key wordscancer burden; attributable fraction; occupational exposure
L. Rushton1, S. J. Hutchings1, T. B. Brown2. 1Imperial College London; 2Health and Safety Laboratory
ObjectivesThe aim of the project is to update the 1981 Doll and Peto estimates of attributable burden of cancer due to occupation for Great Britain. Doll and Peto estimated about 4% of US cancer deaths were attributable to occupation, with an uncertainty range of 2%–8%. Estimates of current burden, future burden, and the relative contributions of different occupational exposures are required.
MethodsWe are tackling the problem on a cancer‐by‐cancer basis, and are estimating attributable fractions (AFs) for the current (2004) burden first. Our approach has been to estimate AF for each cancer–exposure combination, and then to combine these estimates firstly into an AF for each cancer, and then into an overall AF for the cancers covered (six so far). We have divided exposures identified by IARC and Siemiatycki et al (2004) into “established” and “uncertain” groups based on the strength of evidence for causality, and present separate estimates for these two groups.
ResultsThe cancers for which AF has been estimated so far are lung, bladder, non‐melanoma skin (NMSC) and sino‐nasal cancers, leukaemia and mesothelioma. Overall AFs of 7.9% for men and 1.5% for women, based on deaths, and 6.7% and 1.2% based on cancer registrations, are provisionally estimated for these six cancers for established plus uncertain exposures. Results indicate that asbestos is responsible for the highest numbers of current deaths and cancer registrations (for lung cancer and mesothelioma), followed by mineral oils and solar radiation (mostly NMSC), silica (lung cancer) and diesel engine exhaust (lung and bladder cancers). Regarding occupations, employment in the construction industry accounts for half of the total occupational attributable deaths estimated so far, exposing workers to 12 different occupational carcinogens.
ConclusionResults are similar to those of other authors for the USA and Europe. However only six cancers have been covered so far, for exposures in IARC groups 1 and 2A only. Lung cancer and mesothelioma produce the greatest numbers of deaths due to occupation and NMSC the highest numbers of registrations.
Key wordscancer burden; occupation; attributable fraction
A. M. ‘tMannetje1, E. Dryson2, C. Walls2, D. McLean1, F. McKenzie1, M. Maule3, S. Cheng1, N. Pearce1. 1Centre for Public Health Research, Massey University, Wellington; 2Occupational Medicine Specialists, Auckland; 3Cancer Epidemiology Unit, CeRMS and CPO Piemonte, University of Turin
ObjectivesPrevious studies into occupational risk factors for non‐Hodgkin's lymphoma (NHL) in New Zealand have indicated that farmers and meat workers are at increased risk for these neoplasms. We conducted a new nationwide case‐control study to assess whether previously observed associations persist and to identify other occupations that may contribute to the risk of NHL in the New Zealand population.
MethodsA total of 291 incident cases of NHL (age 25–70 years) notified to the New Zealand Cancer Registry during 2003 and 2004, and 471 population controls, were interviewed face‐to‐face. The questionnaire collected demographic information and the full occupational history. The relative risks for NHL associated with being ever employed in particular occupations and industries were calculated by unconditional logistic regression adjusting for age, sex, smoking, ethnicity and socio‐economic status. Estimates were subsequently semi‐Bayes adjusted to account for the large number of occupations and industries being considered.
ResultsAn elevated NHL risk was observed for field crop and vegetable growers (odds ratio (OR) 2.74, 95% CI 1.04 to 7.25) and nursery growers (OR 3.16, 95% CI 1.03 to 9.69), particularly for women. Sheep, beef and dairy cattle farming were not associated with an increased risk of NHL. Meat processors had an elevated risk (OR 1.97, 95% 0.97 to 3.97), as did heavy truck drivers (OR 1.98, 95% CI 0.92 to 4.24) and metal product manufacturing (OR 1.92, 95% CI 1.12 to 3.28). Although not considered an a priori high risk occupation for NHL, cleaners had an increased risk (OR 2.11, 95% CI 1.21 to 3.65) which remained statistically significant after semi‐Bayes adjustment (ORsemi‐bayes 1.80, 95% CI 1.11 to 2.84).
ConclusionThis study has confirmed that crop farming (particularly horticulture and fruit growing) and meat working remain high risk occupations for NHL in New Zealand, and has identified several other occupations and industries of high NHL risk that merit further study.
Key wordsnon‐Hodgkin lymphoma; occupation; farming
P. Boffetta1, F. de Vocht2. 1IARC; 2Occupational and Environmental Health Research Group, School of Translational Medicine, Faculty of Medical and Human Sciences, The University of Manchester
ObjectivesStrictly speaking, the fraction of non‐Hodgkin lymphoma (NHL) attributable to occupation is zero, since no occupational agents have been classified as established causes of NHL by the IARC Monograph programme. However, since a number of jobs and industries have been suggested as entailing an increased risk of NHL, we conducted a series of meta‐analyses for employment in farming, teaching, meat working, the printing industry and wood processing.
MethodsThe Medline database was searched for publications on NHL in relation to any of these occupations, and meta‐ and sensitivity analyses were conducted. A total of 47 publications on farming, 18 on teachers, 8 for meat workers, 11 for wood workers and 7 for the printing industry were included in the meta‐analyses.
ResultsA summary RR for farming of 1.11 (95% CL 1.05 to 1.17) was found, with a concentration of risk in animal breeding (RR 1.31 (1.08 to 1.60)) rather than in crop farming (RR 0.96 (0.83 to 1.09)). Employment as a teacher (RR 1.47 (1.34 to 1.61)) and in the printing industry (RR 1.86 (1.37 to 2.52)) were also associated with NHL risk. Although woodworking was also associated with NHL risk (RR 1.15 (1.00 to 1.31)), this RR might possibly be inflated due to publication bias. Employment in the meat industry was not associated with NHL (RR 0.99 (0.77 to 1.29)).
ConclusionThe concentration of NHL risk in animal breeding points towards a viral rather than chemical aetiology, although farmers involved in animal breeding are also exposed to organic chemicals. The increased summary risk for teachers also points towards a viral aetiology, although this is contradicted by the absence of an increased risk in the meat processing industry. In addition, an increased risk in the printing industry can be explained by exposure to chemicals, with organic solvents being a likely candidate. For no job or industry is there at present conclusive evidence of a causal association with NHL. This might be due to methodological problems, but despite these, research on jobs and industries possibly associated with NHL risk might provide clues on possible causal agents. Nonetheless, these results suggest that it is unlikely that occupation represents a major risk factor for NHL in most populations.
Key wordsnon‐Hodgkin lymphoma; occupational exposure; meta‐analysis
G. M. H. Swaen1, L. G. P. Amelsvoort2. 1The Dow Chemical Company; 2Maastricht University
ObjectivesTo develop an empirically based weight of evidence approach to causal inference from epidemiology studies.
MethodsThere is substantial debate about how to draw causal inferences from epidemiological data. The conventional approach taken in this debate has been theoretical and not empirical. For many risk factors the available evidence is not sufficient to draw definitive conclusions about causality. However, so far there is no method by which this insufficient evidence can be used to assess the probability that an association is causal. We have used the agents classified by IARC as category 1 or 2A carcinogens to evaluate the contribution to causal inference of each of the nine Bradford Hill criteria. The 159 chemicals/agents classified by IARC as category 1 or 2A carcinogens were scored based on the evidence available at the time of the IARC evaluation following a predetermined set of rules. Discriminant analysis was used to estimate the weight of evidence of each of the nine criteria.
ResultsThe discriminant analysis yielded weights for each of the nine criteria. The weights can be used to combine the evidence of all nine criteria into one overall assessment of the probability of a causal association. The criteria strength, consistency and experimental evidence contributed most to the probability. The model correctly predicted 130 of the 159 agents (81.8%).
ConclusionThe weight of evidence of each of the nine Bradford Hill criteria can be quantified. The proposed approach allows estimation of the probability that a certain association is causal. Examples show that the weight of evidence approach works well.
Key wordscausal inference; weight of evidence; risk assessment
K. Straif, R. Baan, Y. Grosse, B. Secretan, F. ElGhissassi, V. Bouvard, A. Altieri, V. J. Cogliano. IARC
ObjectivesThe Preamble describes the scientific principles and procedures used in developing IARC Monographs. In 2005 the Preamble was amended through a public process including suggestions from recent meeting chairs and subgroup chairs, an Advisory Group meeting to recommend updates for consideration, posting of the Draft Preamble on IARC's website for public comment and an Advisory Group meeting to review the amended Preamble. The amended Preamble was published on IARC's website and came into effect in January 2006.
MethodsSome key issues that were addressed included measures to avoid conflicts of interests, enhancing the Monographs' status as a peer review, clarifying points that had led to inconsistent approaches in the past, updating scientific criteria for molecular epidemiology and carcinogenic mechanisms; and including quantitative analyses in a Monograph or subsequent IARC publication.
ResultsDuring the last 2 years the IARC Monographs programme has evaluated or re‐evaluated the human carcinogenicity of several substances or mixtures with potential exposures in the workplace: volume 92 (some non‐heterocyclic polycyclic aromatic hydrocarbons and some related industrial exposures); and volume 93 (carbon black, titanium dioxide and non‐asbestiform talc). Summary findings and evaluations of these Monographs, particularly, the Working Group's reasoning in making the final evaluations and critical research gaps will be highlighted. Additional (re‐)evaluations which are scheduled for June 2007 (volume 97: 1,3‐butadiene, ethylene oxide and vinyl halides: vinyl fluoride, vinyl chloride and vinyl bromide) and October 2007 (volume 98: fire‐fighting, painting and shift‐work) will be presented.
ConclusionThe IARC Monographs have been published continuously since 1971. The 100th volume of IARC Monographs is an historic occasion for the series, and a fitting topic for this volume is a review of the human carcinogens that have been identified to date. For each agent that is classified as carcinogenic to humans (group 1), volume 100 will contain a concise Monograph. After the Monographs in volume 100 are developed, two additional Working Groups will be convened to develop related scientific publications that build on the data that will be summarised in volume 100: (1) tumour‐site concordance between humans and experimental animals; and (2) mechanisms involved in human carcinogenesis.
Key wordsoccupational cancer; review; carcinogen identification