A Bayesian approach to assess interaction between known risk factors: the risk of lung cancer from exposure to asbestos and smoking

Our plain-English summary of A Bayesian approach to assess interaction between known risk factors: the risk of lung cancer from exposure to asbestos and smoking, by D. Wraith, K. Mengersen (Statistical Methods in Medical Research, 2008).

The study

• Authors: D. Wraith, K. Mengersen
• Journal: Statistical Methods in Medical Research
• Published: 2008;17(2):171-89
• Cited by: 13 articles, as counted by NIH iCite
• Read it: PubMed · Publisher

Abstract

We review the literature on the combined effect of asbestos exposure and smoking on lung cancer, and explore a Bayesian approach to assess evidence of interaction. Previous approaches have focussed on separate tests for an additive or multiplicative relation. We extend these approaches by exploring the strength of evidence for either relation using approaches which allow the data to choose between both models. We then compare the different approaches.

In plain English

This review re-analysed, with Bayesian models, 18 case-control and cohort data sets from earlier reviews by Lee and Liddell, asking how asbestos exposure and smoking combine in lung cancer risk. They included crocidolite miners and millers (Wittenoom, Australia), Quebec chrysotile miners and millers, asbestos insulation workers and shipbuilding-area hospital patients. Each factor was classed only as exposed or not. Versus no exposure, overall lung cancer relative risks were 5.51 (95% credible interval 3.78 to 7.89) for smoking alone, 3.13 (1.80 to 5.41) for asbestos alone and 13.69 (8.20 to 22.76) for both. A synergy index of 1.94 (1.29 to 2.84) and multiplicativity index of 0.83 (0.46 to 1.40, including 1) indicated, per the authors, more than additive, less than multiplicative. Given a choice between only additive and multiplicative relations, a mixture model gave an additive probability of 0.02 (0.00 to 0.10) for case-control and 0.52 (0.42 to 0.62) for cohort studies. The authors concluded the relation is more than additive and less than multiplicative. They noted varied studies, unadjusted summary data, and few lung cancers in non-smokers.

We wrote this plain-English summary from the full article with the help of AI, and a second AI model checked each sentence against the article. It is not the authors' wording; the abstract and the article are their own account.

What the study did and found

• Design: Literature review with a Bayesian meta-analysis (statistical re-analysis) of published case-control and cohort studies
• Who was studied: 18 study data sets listed in Table 1, taken from the earlier reviews by Lee and Liddell. They were case-control and cohort studies from Australia, England, Italy, Norway, the USA, Canada, China and Finland, with periods ranging from 1891 to 1991. The logistic and Poisson regression analysis excluded five cohort studies that relied on external rates.
• Exposure: Asbestos exposure and smoking, each classed only as exposed or not exposed. The included studies covered crocidolite miners and millers, chrysotile miners, millers and products workers, anthophyllite miners, amosite factory workers, asbestos insulation workers, asbestos factory workers, and shipbuilding and industrial areas.
• Main results:
  • Overall estimated relative risk of lung cancer for smoking alone, compared with no exposure: 5.51 (95% credible interval 3.78 to 7.89)
  • Overall estimated relative risk of lung cancer for asbestos exposure alone, compared with no exposure: 3.13 (95% credible interval 1.80 to 5.41)
  • Overall estimated relative risk of lung cancer for combined asbestos exposure and smoking, compared with no exposure: 13.69 (95% credible interval 8.20 to 22.76)
  • Overall synergy index S was 1.94 (95% credible interval 1.29 to 2.84) and multiplicativity index V was 0.83 (0.46 to 1.40, an interval that includes 1). The authors described this as indicating a more than additive and less than multiplicative relationship.
  • In the mixture model, the overall posterior probability of an additive model was 0.06 (0.00 to 0.12). Study-specific estimates ranged from 0 to 0.98.
  • When the mixture model was given a choice between only an additive and a multiplicative model, the probability of an additive model was 0.02 (0.00 to 0.10) for case-control studies and 0.52 (0.42 to 0.62) for cohort studies
  • The three relative risk models did not clearly support either a multiplicative or an additive model
• The authors' conclusion: The authors concluded that the combined effect of asbestos and smoking on lung cancer is more than additive and less than multiplicative, consistent with recent reviews
• Limitations the authors state:
  • The combined studies varied in study type, country, size, definitions of non-smokers, asbestos exposure times, and the types and sizes of asbestos particles
  • The small number of lung cancer cases among non-smokers greatly increased uncertainty
  • The analysis relied on summary statistics and unadjusted relative risks, without control for factors such as extent and duration of smoking or other lung carcinogens
  • The form of interaction seen may depend on the length of follow-up in the studies
  • Using a binary exposed/not-exposed classification is likely weaker than an analysis based on continuous data

Topics

The National Library of Medicine indexes this study under subjects our site covers: Lung cancer.

Medical terms in this study, explained in our Lexicon: Lung Cancer, Asbestos Exposure.

Related studies

• A new trajectory approach for investigating the association between an environmental or occupational exposure over lifetime and the risk of chronic disease: Application to smoking, asbestos, and lung cancer — E. Lévêque, A. Lacourt, et al. (2020)
• Asbestos and smoking as risk factors for idiopathic retroperitoneal fibrosis: a case-control study — M. Goldoni, S. Bonini, et al. (2014)
• Bayesian estimation of the probability of asbestos exposure from lung fiber counts — S. Weichenthal, L. Joseph, et al. (2010)
• Bayesian analysis of the rate of spontaneous malignant mesothelioma among BAP1 mutant mice in the absence of asbestos exposure — D.M. Nielsen, M. Hsu, et al. (2025)
• Risk factors for malignant mesothelioma in people with no known exposure to asbestos — B. Musk, L. Gordon, et al. (2017)
• A Bayesian approach to strengthen inference for case-control studies with multiple error-prone exposure assessments — J. Zhang, S.R. Cole, et al. (2013)

Browse return to medical research.