Because our original publication (1) generated considerable interest over the past two years, we now provide additional data from our nested case-control study that are valuable in evaluating the relation between diesel exhaust exposure and lung cancer. Here we include results based on alternative exposure metrics (Table 1) and alternative approaches for adjusting for cigarette smoking as a confounder (Table 2), for comparison with results that appeared in Table 3 of Silverman et al. (1).
Table 1.
Odds ratios and 95% confidence intervals for average and cumulative REC lagged 15 years based on alternative exposure metrics*, †
| Exposure metric | “.58” | “5-Year Mean” | “Median” | ||||||
|---|---|---|---|---|---|---|---|---|---|
| Case patients | Control subjects | OR (95%CI) | Case subjects | Control subjects | OR (95%CI) | Case subjects | Control subjects | OR (95%CI) | |
| Average REC intensity | |||||||||
| Quartiles, μg/m3 | |||||||||
| 0 to < 1 | 47 | 187 | 1.0 (referent) | 47 | 191 | 1.0 (referent) | 47 | 190 | 1.0 (referent) |
| 1 to < 6 | 49 | 173 | 1.04 (0.55 to 1.96) | 48 | 161 | 1.28 (0.68 to 2.41) | 52 | 189 | 1.10 (0.59 to 2.05) |
| 6 to < 57 | 35 | 109 | 1.46 (0.66 to 3.23) | 53 | 167 | 1.51 (0.72 to 3.17) | 54 | 167 | 1.80 (0.86 to 3.76) |
| ≥ 57 | 67 | 197 | 1.99 (0.96 to 4.11) | 50 | 147 | 2.00 (0.93 to 4.33) | 45 | 120 | 2.65 (1.22 to 5.75) |
| P trend = .0497 | P trend = .127 | P trend = .017 | |||||||
| Cumulative REC | |||||||||
| Quartiles, μg/m3-yr | |||||||||
| 0 to < 3 | 49 | 157 | 1.0 (referent) | 49 | 155 | 1.0 (referent) | 49 | 158 | 1.0 (referent) |
| 3 to < 72 | 48 | 219 | 0.73 (0.39 to 1.37) | 48 | 226 | 0.71 (0.38 to 1.31) | 50 | 233 | 0.71 (0.38 to 1.32) |
| 72 to < 536 | 40 | 131 | 1.48 (0.72 to 3.05) | 57 | 169 | 1.73 (0.84 to 3.59) | 58 | 172 | 1.67 (0.81 to 3.45) |
| ≥ 536 | 61 | 159 | 2.34 (1.08 to 5.10) | 44 | 116 | 2.28 (1.02 to 5.08) | 41 | 103 | 3.03 (1.35 to 6.79) |
| P trend = .004 | P trend = .015 | P trend = .001 | |||||||
* Adjusted for mine location/smoking status/packs/day combination (surface work only/never smoker, surface work only/unknown/occasional smoker, surface work only/former smoker/< 1 pack/day, surface work only/former smoker/1 to < 2 packs/day, surface work only/former smoker/≥ 2 packs/day, surface work only /current smoker/< 1 pack/day, surface work only/current smoker/1 to < 2 packs/day, surface work only/current smoker/≥ 2 packs/day, ever underground [UG] work/never smoker, ever UG work/unknown/occasional smoker, ever UG work/former smoker/< 1 pack/day, ever UG work/former smoker/1 to < 2 packs/day, ever UG work/former smoker/≥ 2 packs/day, ever UG work/current smoker/< 1 pack/day, ever UG work/current smoker/1 to < 2 packs/day, ever UG work/current smoker/≥ 2 packs/day); history of respiratory disease five or more years before date of death/reference date; and history of a high-risk job for lung cancer for at least 10 years. CI = confidence interval; CO = carbon monoxide; DEMS = Diesel Exhaust in Miners Study; OR = odds ratio; REC = respirable elemental carbon; UG = underground.
† Alternative exposure metrics were defined as:
“.58” REC metrics were based on RECx = RECR(COx/COR)b, where R and X refer to estimates for the reference and for other years, respectively, and the constant, b = 0.58, was estimated from the DEMS measurements. This contrasts with b = 1, which assumes changes in REC over time were directly proportional to changes in CO over time, as in Silverman et al. (1).
“5-year mean” REC metrics used five-year averages in CO between 1976 and 1998–2001 and the ratio of horsepower over ventilation before 1976 in time trend models to predict historical REC levels.
“Median” REC metrics were based on the median of the REC measurements instead of the mean of the REC measurements to derive 1998–2001 reference values used in the time trend models, as in Silverman et al. (1).
Table 2.
Odds ratios and 95% confidence intervals for average and cumulative REC lagged 15 years with alternative approaches for smoking adjustment
| Exposure metric | Status: duration*, † | Status: pack-years‡ | Status: packs/day, duration†, § | ||
|---|---|---|---|---|---|
| Case subjects | Control subjects | OR (95%CI) | OR (95%CI) | OR (95%CI) | |
| Average REC intensity | |||||
| Quartiles, μg/m3 | |||||
| 0 to < 1 | 47 | 190 | 1.0 (referent) | 1.0 (referent) | 1.0 (referent) |
| 1 to < 6 | 52 | 187 | 1.25 (0.67 to 2.31) | 1.25 (0.68 to 2.31) | 1.34 (0.70 to 2.57) |
| 6 to < 57 | 49 | 141 | 1.89 (0.90 to 3.98) | 1.88 (0.89 to 3.97) | 2.23 (1.02 to 4.85) |
| ≥ 57 | 50 | 148 | 2.48 (1.16 to 5.29) | 2.25 (1.05 to 4.84) | 2.66 (1.20 to 5.87) |
| P trend = .040 | P trend = .090 | P trend = .051 | |||
| Cumulative REC | |||||
| Quartiles, μg/m3-yr | |||||
| 0 to < 3 | 49 | 158 | 1.0 (referent) | 1.0 (referent) | 1.0 (referent) |
| 3 to < 72 | 50 | 228 | 0.86 (0.47 to 1.59) | 0.79 (0.43 to 1.45) | 0.87 (0.45 to 1.67) |
| 72 to < 536 | 49 | 157 | 1.58 (0.77 to 3.27) | 1.47 (0.71 to 3.03) | 1.73 (0.80 to 3.74) |
| ≥ 536 | 50 | 123 | 3.19 (1.44 to 7.06) | 2.88 (1.31 to 6.36) | 3.72 (1.61 to 8.60) |
| P trend = .0004 | P trend = .001 | P trend = .0002 | |||
* P values based on two-sided Wald test for linear trend; Adjusted for mine location/smoking status/duration smoked combination (surface work only/never smoker, surface work only/unknown/occasional smoker, surface work only/former smoker/< 20 yrs, surface work only/former smoker/20 to < 30 yrs, surface work only/former smoker/30 to < 40 yrs, surface work only/former smoker/40 to < 50 yrs, surface work only/former smoker/≥ 50 yrs, surface work only/current smoker/< 20 yrs, surface work only/current smoker/20 to < 30 yrs, surface work only/current smoker/30 to < 40 yrs, surface work only/current smoker/40 to < 50 yrs, surface work only/current smoker/≥ 50 yrs, ever underground [UG] work/never smoker, ever UG work/unknown/occasional smoker, ever UG work/former smoker/< 20 yrs, ever UG work/former smoker/20 to < 30 yrs, ever UG work/former smoker/30 to < 40 yrs, ever UG work/former smoker/40 to < 50 yrs, ever UG work/former smoker/≥ 50 yrs, ever UG work/current smoker/< 20 yrs, ever UG work/current smoker/20 to < 30 yrs, ever UG work/current smoker/30 to < 40 yrs, ever UG work/current smoker/40 to < 50 yrs, ever UG work/current smoker/≥ 50 yrs); history of respiratory disease five or more years before date of death/reference date; and history of a high-risk job for lung cancer for at least 10 years. CI = confidence interval; OR = odds ratio; pkyr = pack-year; REC = respirable elemental carbon.
† Duration smoked was computed by subtracting age started smoking from current age (current smokers) or age quit smoking (former smokers) minus any intermittent years stopped smoking. For the 72 case subjects and 107 control subjects with missing age started smoking, age started was assumed to be 17 years old, which was the average age started in the controls.
‡ P values based on two-sided Wald test for linear trend; Adjusted for mine location/smoking status/pkyr combination (surface work only/never smoker, surface work only/unknown/occasional smoker, surface work only/former smoker/< 33 pkyrs, surface work only/former smoker/33 to < 49 pkyrs, surface work only/former smoker/49 to < 70 pkyrs, surface work only/former smoker/≥ 70 pkyrs, surface work only/current smoker/< 33 pkyrs, surface work only/current smoker/33 to < 49 pkyrs, surface work only/current smoker/49 to < 70 pkyrs, surface work only/current smoker/≥ 70 pkyrs, ever UG work/never smoker, ever UG work/unknown/occasional smoker, ever UG work/former smoker/< 33 pkyrs, ever UG work/ former smoker/33 to < 49 pkyrs, ever UG/former smoker/49 to < 70 pkyrs, ever UG work/former smoker/≥ 70 pkyrs, ever UG work/current smoker/< 33 pkyrs, ever UG work/current smoker/33 to < 49 pkyrs, ever UG /current smoker/49 to < 70 pkyrs, ever UG work/former smoker/≥ 70 pkyrs); history of respiratory disease five or more years before date of death/reference date; and history of a high-risk job for lung cancer for at least 10 years.
§ P values based on two-sided Wald test for linear trend; Adjusted for mine location/smoking status/packs/day/duration smoked combination (surface work only/never smoker, surface work only/unknown/occasional smoker, surface work only/former smoker/< 1 pack/day/< 20 yrs, surface work only/former smoker/< 1 pack/day/20 to < 30 yrs, surface work only/former smoker/< 1 pack/day/30 to < 40 yrs, surface work only/former smoker/< 1 pack/day/40 to < 50 yrs, surface work only/former smoker/< 1 pack/day/≥ 50 yrs, surface work only/former smoker/1 to < 2 packs/day/< 20 yrs, surface work only/former smoker/1 to < 2 packs/day/20 to < 30 yrs, surface work only/former smoker/1 to < 2 packs/day/30 to < 40 yrs, surface work only/former smoker/1 to < 2 packs/day/40 to < 50 yrs, surface work only/former smoker/1 to < 2 packs/day/≥ 50 yrs, surface work only/former smoker/≥ 2 packs/day/< 20 yrs, surface work only/former smoker/≥ 2 packs/day/20 to < 30 yrs, surface work only/former smoker/≥ 2 packs/day/30 to < 40 yrs, surface work only/former smoker/≥ 2 packs/day/40 to < 50 yrs, surface work only/former smoker/≥ 2 packs/day/≥ 50 yrs, surface work only/current smoker/< 1 pack/day/< 20 yrs, surface work only/current smoker/< 1 pack/day/20 to < 30 yrs, surface work only/current smoker/< 1 pack/day/30 to < 40 yrs, surface work only/current smoker/< 1 pack/day/40 to < 50 yrs, surface work only/current smoker/< 1 pack/day/≥ 50 yrs, surface work only/current smoker/1 to < 2 packs/day/< 20 yrs, surface work only/current smoker/1 to < 2 pack/day/20 to < 30 yrs, surface work only/current smoker/1 to < 2 packs/day/30 to < 40 yrs, surface work only/current smoker/1 to < 2 packs/day/40 to < 50 yrs, surface work only/current smoker/1 to < 2 packs/day/≥ 50yrs, surface work only/current smoker/≥ 2 packs/day/< 20 yrs, surface work only/current smoker/≥ 2 packs/day/20 to < 30 yrs, surface work only/current smoker/≥ 2 packs/day/30 to < 40 yrs, surface work only/current smoker/≥ 2 packs/day/40 to < 50 yrs, surface work only/current smoker/≥ 2 packs/day/≥ 50 yrs, ever UG work/never smoker, UG only/unknown/occasional smoker, ever UG work/former smoker/< 1 pack/day/< 20 yrs, ever UG work/former smoker/< 1 pack/day/20 to < 30 yrs, ever UG work/former smoker/< 1 pack/day/30 to < 40 yrs, ever UG work/former smoker/< 1 pack/day/40 to < 50 yrs, ever UG work/former smoker/< 1 pack/day/≥ 50 yrs, ever UG work/former smoker/1 to < 2 packs/day/< 20 yrs, ever UG work/former smoker/1 to < 2 packs/day/20 to < 30 yrs, ever UG work/former smoker/1 to < 2 packs/day/30 to < 40 yrs, ever UG work/former smoker/1 to < 2 packs/day/40 to < 50 yrs, ever UG work/former smoker/1 to < 2 packs/day/≥ 50 yrs, ever UG work/former smoker/≥ 2 packs/day/< 20 yrs, ever UG work/former smoker/≥ 2 packs/day/20 to < 30 yrs, ever UG work/former smoker/≥ 2 packs/day/30 to < 40 yrs, ever UG work/former smoker/≥ 2 packs/day/40 to < 50 yrs, ever UG work/former smoker/≥ 2 packs/day/≥ 50 yrs, ever UG work/current smoker/< 1 pack/day/< 20 yrs, ever UG work/current smoker/< 1 pack/day/20 to < 30 yrs, ever UG work/current smoker/< 1 pack/day/30 to < 40 yrs, ever UG work/current smoker/< 1 pack/day/40 to < 50 yrs, ever UG work/current smoker/< 1 pack/day/≥ 50 yrs, ever UG work/current smoker/1 to < 2 packs/day/< 20 yrs, ever UG work/current smoker/1 to < 2 packs/day/20 to < 30 yrs, ever UG work/current smoker/1 to < 2 packs/day/30 to < 40 yrs, ever UG work/current smoker/1 to < 2 packs/day/40 to < 50 yrs, ever UG work/ current smoker/1 to < 2 packs/day/≥ 50 yrs, ever UG work/current smoker/≥ 2 packs/day/< 20 yrs, ever UG work/current smoker/≥ 2 packs/day/20 to < 30 yrs, ever UG work/current smoker/≥ 2 packs/day/30 to < 40 yrs, ever UG work/current smoker/≥ 2 packs/day/40 to < 50 yrs, ever UG work/current smoker/≥ 2 packs/day/≥ 50 yrs.
In Table 1, we estimate the lung cancer risk associated with average respirable elemental carbon (REC) intensity and cumulative REC, lagging exposure by 15 years, based on three alternative exposure metrics (2–6). The “.58” REC metrics assume that changes in REC over time are not directly proportional to changes in carbon monoxide (CO) over time, whereas the original REC metrics assumed direct proportionality (1). The “5-year mean” REC metrics use 5-year averages in CO between 1976 and the last year of the exposure assessment at each mine (1998–2001) and the ratio of horsepower over ventilation before 1976 (when CO measurements were unavailable) in the time trend models to predict historical REC levels. The “median” REC metrics were based on the median of the REC measurements instead of the mean of the REC measurements to derive the 1998–2001 reference values used in the original time trend models. Results based on these alternative exposure metrics closely reflect those originally reported in Table 3 (1), each with consistent positive trends in risk with increasing exposure. Trends tended to be highly statistically significant for cumulative REC and statistically significant or borderline significant for average REC intensity.
In Table 2, we estimate the lung cancer risk associated with average REC intensity and cumulative REC, lagging 15 years, based on three approaches for smoking adjustment. We present odds ratios by increasing exposure levels adjusting for: 1) smoking status (never, former, current smoker) and duration smoked; 2) status and pack-years smoked; and 3) status and both duration and packs/day in the same model. Regardless of the approach for smoking adjustment, we observed consistent highly statistically significant trends for cumulative REC and borderline significant trends for average REC intensity. Results are strikingly consistent with the previously published odds ratios in Table 3 (1).
In conclusion, these sensitivity analyses underscore the robustness of our original findings (1) and provide further evidence that diesel exhaust causes lung cancer in humans.
Funding
Intramural Program of the National Institutes of Health, National Cancer Institute, Division of Cancer Epidemiology and Genetics and the National Institute of Occupational Safety and Health, Division of Respiratory Disease Studies .
We thank Nathan Appel of IMS, Inc for computer support and Gabriela Cadena for clerical support.
The findings and conclusions in this letter have not been formally disseminated by the National Institute for Occupational Safety and Health and should not be construed to represent any agency determination or policy. The authors declare no potential conflict of interest.
References
- 1. 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(11):855–868. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 2. Stewart PA, Coble J, Vermeulen R, et al. The Diesel Exhaust in Miners Study: I. Overview of the exposure assessment process. Ann Occup Hyg. 2010;54(7):728–746. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 3. Coble J, Stewart PA, Vermeulen R, et al. The Diesel Exhaust in Miners Study: II. Exposure monitoring surveys and development of exposure groups. Ann Occup Hyg. 2010;54(7):728–746. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 4. Vermeulen R, Coble JB, Yereb D, et al. The Diesel Exhaust in Miners Study: III. Interrelations between respirable elemental carbon and gaseous and particulate components of diesel exhaust derived from area sampling in underground non-metal mining facilities. Ann Occup Hyg. 2010;54(7):762–773. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 5. Vermeulen R, Coble J, Lubin J, et al. The Diesel Exhaust in Miners Study: IV. Estimating historical exposures to diesel exhaust in underground mines. Ann Occup Hyg. 2010;54(7):774–788. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 6. Stewart PA, Vermeulen R, Coble JB, et al. The Diesel Exhaust in Miners Study: V. Evaluation of the exposure assessment methods. Ann Occup Hyg. 2012;56(4):389–400. [DOI] [PMC free article] [PubMed] [Google Scholar]