To the Editor:
We thank the editor for the opportunity to respond to the letter by Dr. Möhner1 regarding our article.2 In this response, we address concerns raised and offer additional context regarding the validity of our estimates.
In his letter, Dr. Möhner1 largely focuses on the lack of adjustment for smoking and history of high-risk jobs. This information was not available for the entire Diesel Exhaust in Miners Study (DEMS) cohort, so we were unable to adjust for it.2 However, partial information on both covariates was available in the DEMS nested case–control sample.3 A comparison of the adjusted case–control estimates with the unadjusted cohort study estimates4 does not support the notion that these variables caused bias away from the null; in fact, the adjusted estimates were stronger.
Dr. Möhner1 correctly points out that causal inference based on our application of the parametric G-formula in DEMS (or any other analysis) is subject to assumptions such as conditional exchangeability and counterfactual consistency. Conditional exchangeability stipulates that the observed probability of exposure is independent of counterfactual outcomes, given measured covariates. If our data included unexposed gardeners who smoked more, then a comparison with the exposed underground miners would be problematic, as conditional exchangeability would fail without control for smoking. This was not the case, however. The counterfactual scenario of a miner working as a gardener is also irrelevant to the regulation of diesel exposures in the mining industry. Our study focused on interventions we judged to be the most relevant for questions regarding regulatory limits for diesel exposures in the mining industry. We compared counterfactual outcomes under interventions on exposure to diesel exhaust without intervening on other covariates such as employment status, job type, or location.
Similarly, we did not restrict the population of underground miners because we were interested in the target population that includes all those at risk of exposure in this industry. Including workers with history of other high-risk jobs causes problems for the identifiability of our target parameters only if these histories are also independent predictors of the exposure in our study. Evidence from the case–control study,3 however, does not suggest that this factor was an important confounder.
Last, we disagree with the claim that evidence of an exposure–response between respirable elemental carbon and lung cancer is “very limited.” The International Agency for Research on Cancer reclassified diesel exhaust as a group 1 human carcinogen in 2012 after reviewing the body of scientific evidence, including studies such as DEMS with quantitative exposure metrics.5 An independent panel further concluded that these studies addressed major issues in the diesel and lung cancer literature and highlighted the importance of epidemiologic studies as evidence for exposure–response, despite analytical challenges.6
In summary, we acknowledge that our study is subject to limitations, as are all observational studies, and agree that assumptions required for parameter estimation should always be explicitly considered. Our study built on strong prior evidence that diesel exhaust is a risk factor for lung cancer and leveraged observational data to provide estimates of risk that are particularly relevant for regulatory purposes.
Andreas M. Neophytou
Ellen A. Eisen
Division of Environmental Health Sciences
UC Berkeley School of Public Health
1. Möhner M. Re: Occupational diesel exposure, duration of employment, and lung cancer: an application of the parametric G-formula. Epidemiology. 2017;28:e63–e64.
2. Neophytou AM, Picciotto S, Costello S, Eisen EA. Occupational diesel exposure, duration of employment, and lung cancer: an application of the parametric G-formula. Epidemiology. 2016;27:21–28.
3. Silverman DT, Samanic CM, Lubin JH, et al. The Diesel Exhaust in Miners Study: a nested case-control study of lung cancer and diesel exhaust. J Natl Cancer Inst. 2012;104:855–868.
4. Attfield MD, Schleiff PL, Lubin JH, et al. The Diesel Exhaust in Miners Study: a cohort mortality study with emphasis on lung cancer. J Natl Cancer Inst. 2012;104:869–883.
5. IARC. IARC Monographs on the Evaluation of the Carcinogenic Risks to Humans. Diesel and Gasoline Engine Exhausts and Some Nitroarenes. 2014.Vol 105. Lyon, France; International Agency for Research on Cancer.
6. HEI Diesel Epidemiology Panel. Diesel Emissions and Lung Cancer: An Evaluation of Recent Epidemiological Evidence for Quantitative Risk Assessment. 2015.Boston, MA; Health Effects Institute.