Quality of evidence must guide risk assessment of asbestos

Our plain-English summary of Quality of evidence must guide risk assessment of asbestos, by V. Lenters, A. Burdorf, R. Vermeulen, et al (The Annals of Occupational Hygiene, 2012).

The study

• Authors: V. Lenters, A. Burdorf, R. Vermeulen, L. Stayner, D. Heederik
• Journal: The Annals of Occupational Hygiene
• Published: 2012;56(8):879-87
• Cited by: 10 articles, as counted by NIH iCite
• Read it: PubMed · Publisher

Abstract

In 2011, we reported on the sensitivity of lung cancer potency estimates for asbestos to the quality of the exposure assessment component of underlying evidence. Both this meta-analysis and a separate reassessment of standards published by the Health Council of the Netherlands (Gezondheidsraad) have been commented on by Berman and Case. A criticism is that we used a truncated data set. We incrementally excluded poorer-quality studies to evaluate trends in meta-analyzed lung cancer potency estimates (meta-K (L) values). This was one of three analysis approaches we presented. The other two used the full set of studies: a meta-analysis stratified by covariates and dichotomized by poorer and better exposure assessment aspects; and a meta-regression modeling both asbestos fiber type and these covariates. They also state that our results are not robust to removal of one study. We disagree with this claim and present additional sensitivity analyses underpinning our earlier conclusion that inclusion of studies with higher-quality asbestos-exposure assessment yield higher meta-estimates of the lung cancer risk per unit of exposure. We reiterate that potency differences for predominantly chrysotile- versus amphibole-asbestos-exposed cohorts are difficult to ascertain when meta-analyses are restricted to studies with fewer exposure assessment limitations. We strongly argue that the existence of any uncertainty related to potency issues should not hamper the development of appropriate evidence-based guidelines and stringent policies in order to protect the public from hazardous environmental and occupational exposures.

In plain English

This article was the authors' reply to a commentary by Berman and Case that criticized the authors' 2011 meta-analysis of asbestos and lung cancer. That meta-analysis combined 19 risk estimates. It asked whether studies with better exposure assessment gave higher estimates of lung cancer risk per unit of exposure, a measure called the potency factor KL. In this reply the authors presented additional sensitivity analyses. When they left out one study at a time, the trend of higher estimates in better-quality studies was generally stable, but two studies were influential. Removing the South Carolina chrysotile textile plant weakened the trend: the ratio of best to all studies was 2.3 instead of 4.2. Removing the Libby, Montana vermiculite mines made it stronger, at 12.6. Changing the order in which studies were excluded did not change the overall trend. The authors noted that several quality criteria were well above the significance threshold of 0.10 and could have arisen by chance, and that only 19 data points were available. The authors concluded that exposure assessment quality influences the asbestos-lung cancer exposure-response slope. They argued that remaining uncertainty should not hamper protective policies.

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: Reply to a commentary (Berman and Case, 2012), with additional sensitivity analyses of the authors' previously published 2011 meta-analysis. These included leave-one-out (jackknife) analyses, subgroup meta-analysis, meta-regression and varied exclusion order, using a random effects model (REML).
• Who was studied: 19 risk estimates from the 2011 meta-analysis: 18 industry-based cohorts (mines and mills, textile, friction products, insulation, cement and vermiculite operations in North America, Europe and Australia) and one population-based case-control study from Stockholm County, Sweden.
• Exposure: Occupational asbestos exposure, with studies grouped by fiber type: chrysotile (studies 1 to 5), amphiboles (studies 6 to 9) and mixed (studies 10 to 19). Studies were also classified by five exposure assessment quality criteria: documentation, conversion factor for particle counts, completeness of job histories, contrast between the highest and lowest cumulative exposure categories, and coverage of the exposure history by measurement data.
• Main results:
  • In leave-one-out analyses, the trend of increasing meta-KL values with better exposure assessment quality was generally stable, but two studies were influential.
  • The ratio of the meta-KL from the best studies to the meta-KL from all studies was 4.2 in the original results. It fell to 2.3 when Study 4 (South Carolina, predominantly chrysotile-exposed textile workers) was excluded.
  • When Study 9 (Libby, Montana, predominantly amphibole-exposed vermiculite workers) was excluded, the ratio of best to all studies rose to 12.6.
  • Excluding studies in different orders did not change the overall trend. Meta-KL times 100 rose from 0.13 for all 19 studies to 0.55 or 0.56 when 2 studies remained.
  • Higher meta-KL values were most evident for two quality criteria. For studies with more than 30% coverage of the exposure history by measurements, meta-KL times 100 was 0.27 (95% CI 0.08, 0.46), versus 0.08 (95% CI minus 0.01, 0.18) for 30% or less (P 0.08). For studies with sufficient job histories it was 0.19 (95% CI 0.08, 0.30), versus 0.03 (95% CI minus 0.10, 0.17) for insufficient job histories (P 0.08).
  • With South Carolina excluded, the all-studies meta-KL times 100 was 0.10 (95% CI 0.02, 0.18), compared with 0.13 (95% CI 0.04, 0.22) originally. The authors reported that the pattern of larger meta-KL values in higher-quality studies still held.
  • By fiber type, meta-KL times 100 was 0.04 (95% CI minus 0.05, 0.12) for chrysotile studies, 0.33 (95% CI 0.09, 0.56) for amphibole studies and 0.13 (95% CI 0.03, 0.23) for mixed-exposure studies (P 0.06 for the difference). The P value was 0.02 with South Carolina excluded.
  • When the authors repeated the analyses with their own KL values but the modeling approach of Berman and Case, they observed similar trends.
• The authors' conclusion: The authors concluded that their meta-analysis demonstrated that the quality of exposure assessment influences the magnitude of the asbestos-lung cancer exposure-response slope. They reiterated that differences in lung cancer potency between predominantly chrysotile- and amphibole-exposed cohorts are difficult to ascertain when analyses are restricted to studies with fewer exposure assessment limitations. They argued that remaining uncertainty should not hamper evidence-based guidelines and stringent policies.
• Limitations the authors state:
  • With only 19 data points, adding more variables to the meta-regression would make the models too unstable.
  • Several of the quality criteria were well above the significance threshold of 0.10, and the differences could have arisen by chance.
  • Including the quality covariates did not substantially reduce heterogeneity between studies or improve model fit.
  • Differences in potency between studies may be explained by factors other than quality, such as misclassification of the endpoint, age at first exposure, residual confounding by smoking, fiber type and fiber dimensions.
  • There is not yet a scientific consensus on what the quality threshold for including observational studies should be.
  • Mesothelioma was not addressed in the 2011 meta-analysis, and the authors did not comment on the Health Council's mesothelioma estimates.

Topics

The National Library of Medicine indexes this study under subjects our site covers: Chrysotile, Mesothelioma.

Medical terms in this study, explained in our Lexicon: Lung Cancer, Meta-Analysis, Chrysotile, Amphibole Group, Occupational Exposure.

Related studies

• Quality of evidence must guide risk assessment of asbestos, by Lenters, V; Burdorf, A; Vermeulen, R; Stayner, L; Heederik, D — J.T. Hodgson (2013)
• Quality of evidence must guide risk assessment of asbestos, by Lenters, V; Burdorf, A; Vermeulen, R; Stayner, L; Heederik, D — D.W. Berman, B.W. Case (2013)
• Overreliance on a single study: there is no real evidence that applying quality criteria to exposure in asbestos epidemiology affects the estimated risk — D.W. Berman, B.W. Case (2012)
• A meta-analysis of asbestos and lung cancer: is better quality exposure assessment associated with steeper slopes of the exposure-response relationships? — V. Lenters, R. Vermeulen, et al. (2011)
• A ban on asbestos must be based on a comparative risk assessment — M. Camus (2001)
• Reply: response to the letter by Drs Berman and Case — D. Heederik, V. Lenters, et al. (2013)

Browse more research in The Annals of Occupational Hygiene, or return to medical research.