The analysis of asbestos count data with "nondetects": the example of asbestos fiber concentrations in the lungs of brake workers

Our plain-English summary of The analysis of asbestos count data with "nondetects": the example of asbestos fiber concentrations in the lungs of brake workers, by M.M. Finkelstein (American Journal of Industrial Medicine, 2013).

The study

• Author: M.M. Finkelstein
• Journal: American Journal of Industrial Medicine
• Published: 2013;56(12):1482-9
• Cited by: 8 articles, as counted by NIH iCite
• Read it: PubMed · Publisher

Abstract

Objectives: In the analysis of tissue for asbestos fibers, some measurements may be below the analytical detection limit (nondetects). The use of maximum likelihood and survival analysis methods have been recommended to perform comparisons between subjects in the presence of nondetects. When the data consist of "counts" another method is useful. This method is discussed, and illustrated with an analysis of asbestos lung burden data among brake mechanics previously analyzed by other methods.

Methods: Statistical models for count data, namely Poisson and negative binomial regression, were used to compare the asbestos fiber concentrations in the lungs of brake mechanics with those of control subjects. The fit of the models was assessed with an analysis of residuals.

Results: The negative binomial regression models fit the data well. The concentrations of Quebec asbestos fibers in the lungs of the brake mechanics were significantly higher than in the control population.

Conclusions: Helsel recommended the use of maximum likelihood and survival analysis methods to perform comparisons in the presence of nondetects. When analyzing asbestos fiber count data, or other count data arising in occupational or environmental health, the use of models such as the Poisson and negative binomial may be added to the analyst's toolbox. Benefits are that neither of these methods requires the substitution of arbitrary values for the nondetects and that programs for the computation of count data models are contained in popular statistical software packages.

© 2013 Wiley Periodicals, Inc.

In plain English

This commentary proposed a statistical method for counts of asbestos fibers in lung tissue when some samples have "nondetects," meaning no fibers were seen. It illustrated the method by re-analyzing earlier data. The data were fiber counts made by scanning electron microscopy in Dr. Victor Roggli's laboratory. They covered 10 brake mechanics and 19 control subjects without occupational asbestos exposure, though the counting sheet for one control was missing. The author added seven more brake mechanics from two later reports, for 17 mechanics in all. Counts of tremolite, anthophyllite and actinolite were combined as "Quebec amphibole," used as a marker of Quebec asbestos. Negative binomial models fit the data better than Poisson models. With all 17 mechanics, the concentration in mechanics compared with controls was 10.4 times higher for chrysotile (95% confidence interval 1.1 to 97) and 4.3 times higher for Quebec amphibole (1.7 to 11). Tissue weights were not reported for the seven added cases, so an average weight was substituted. The author concluded that count models can be used for such data without substituting arbitrary values for nondetects.

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: Commentary presenting a statistical method (Poisson and negative binomial regression for count data), illustrated by a re-analysis of previously published lung fiber burden data
• Who was studied: 10 brake mechanics from Butnor et al. (2003) and 19 control subjects without occupational asbestos exposure from Srebro et al. (1995), with fibers counted in Dr. Victor Roggli's laboratory by scanning electron microscopy. Seven more brake mechanics were added: five with mesothelioma from Marsh et al. (2011) and two with nonmalignant lung disease from Egilman (2012). This gave 17 mechanics in total.
• Exposure: Work as a brake mechanic; friction products in North America were made with Quebec chrysotile asbestos. Lung counts of amosite, crocidolite, chrysotile, tremolite, anthophyllite and actinolite were examined. Tremolite, anthophyllite and actinolite were summed as "Quebec amphibole," taken as a marker of Quebec asbestos.
• Main results:
  • In all comparisons, the dispersion parameter was significantly greater than 0. The negative binomial model therefore fit better than the Poisson model, and only negative binomial results were presented.
  • Chrysotile, original 10 mechanics compared with controls: ratio of fiber concentration 10.2, P = 0.04, 95% confidence interval 1.1 to 93.
  • Chrysotile, all 17 mechanics compared with controls: ratio 10.4, P = 0.04, 95% confidence interval 1.1 to 97, a significantly higher concentration among the mechanics. Bootstrapped confidence intervals could not be computed for chrysotile.
  • Quebec amphibole, original 10 mechanics compared with controls: ratio 5.6, P less than 0.001, 95% confidence interval 2.3 to 13 (bootstrapped interval 2.1 to 14).
  • Quebec amphibole, all 17 mechanics compared with controls: ratio 4.3, P = 0.002, 95% confidence interval 1.7 to 11 (bootstrapped interval also 1.7 to 11).
  • Among the 17 mechanics, 5 of 17 samples (29%) had no Quebec asbestos fibers counted. A negative binomial distribution with the mechanics' mean count of 1.94 predicted about 33% zero counts.
  • Coated and uncoated amosite was found in 6 of 10 mechanics in Butnor's series and in 1 of 18 controls.
  • Examination of the residuals showed that the count models fit the data well.
  • The author concluded that the mean concentration of Quebec asbestos fibers in the lungs of the brake mechanics was significantly higher than in the control population.
• The authors' conclusion: The author concluded that Poisson and negative binomial models may be added to the analyst's toolbox for asbestos fiber counts and other count data, and that these methods do not require substituting arbitrary values for nondetects.
• Limitations the authors state:
  • Tissue weights were not reported for the seven added mechanics, so the average weight from the original mechanics series (0.26 g) was substituted.
  • The counting sheet for one control subject (subject 17) was missing.
  • In small samples, maximum likelihood estimates may be biased. Bootstrap resampling was used to check this, but bootstrapped intervals could not be computed for chrysotile.
  • Bootstrapped confidence intervals could not be computed for the chrysotile models because chrysotile was uncommon and the models failed to converge.

Topics

The National Library of Medicine indexes this study under subjects our site covers: Amosite, Auto mechanics and brake work, Chrysotile, Crocidolite.

Related studies

• Asbestos Fiber Concentrations in the Lungs of Brake Repair Workers: An Updated Analysis Using Several Regression Methods to Handle Nondetectable Measurements — Y. Liu, G.M. Marsh, et al. (2018)
• Asbestos fiber concentrations in the lungs of brake repair workers: commercial amphiboles levels are predictive of chrysotile levels — G.M. Marsh, A.O. Youk, et al. (2011)
• Airborne asbestos concentration from brake changing does not exceed permissible exposure limit — C.L. Blake, D.R. Van Orden, et al. (2003)
• Exposure to brake dust and malignant mesothelioma: a study of 10 cases with mineral fiber analyses — K.J. Butnor, T.A. Sporn, et al. (2003)
• Exposure to chrysotile asbestos associated with unpacking and repacking boxes of automobile brake pads and shoes — A.K. Madl, L.L. Scott, et al. (2008)
• A technical comparison of evaluating asbestos concentration by phase-contrast microscopy (PCM), scanning electron microscopy (SEM), and analytical transmission electron microscopy (ATEM) as illustrated from data generated from a case report — R.F. Dodson, S.P. Hammar, et al. (2008)

Browse more research in American Journal of Industrial Medicine, or return to medical research.