Assessing the component associations of the healthy worker survivor bias: occupational asbestos exposure and lung cancer mortality

Our plain-English summary of Assessing the component associations of the healthy worker survivor bias: occupational asbestos exposure and lung cancer mortality, by A.I. Naimi, S.R. Cole, M.G. Hudgens, et al (Annals of Epidemiology, 2013).

The study

• Authors: A.I. Naimi, S.R. Cole, M.G. Hudgens, M.A. Brookhart, D.B. Richardson
• Journal: Annals of Epidemiology
• Published: 2013;23(6):334-41
• Cited by: 27 articles, as counted by NIH iCite
• Read it: PubMed · Publisher · Free full text

Abstract

Background: The healthy worker survivor bias is well-recognized in occupational epidemiology. Three component associations are necessary for this bias to occur: i) prior exposure and employment status; ii) employment status and subsequent exposure; and iii) employment status and mortality. Together, these associations result in time-varying confounding affected by prior exposure. We illustrate how these associations can be assessed using standard regression methods.

Methods: We use data from 2975 asbestos textile factory workers hired between January 1940 and December 1965 and followed for lung cancer mortality through December 2001.

Results: At entry, median age was 24 years, with 42% female and 19% non-Caucasian. Over follow-up, 21% and 17% of person-years were classified as at work and exposed to any asbestos, respectively. For a 100 fiber-year/mL increase in cumulative asbestos, the covariate-adjusted hazard of leaving work decreased by 52% (95% confidence interval [CI], 46-58). The association between employment status and subsequent asbestos exposure was strong due to nonpositivity: 88.3% of person-years at work (95% CI, 87.0-89.5) were classified as exposed to any asbestos; no person-years were classified as exposed to asbestos after leaving work. Finally, leaving active employment was associated with a 48% (95% CI, 9-71) decrease in the covariate-adjusted hazard of lung cancer mortality.

Conclusions: We found strong associations for the components of the healthy worker survivor bias in these data. Standard methods, which fail to properly account for time-varying confounding affected by prior exposure, may provide biased estimates of the effect of asbestos on lung cancer mortality under these conditions.

Copyright © 2013 Elsevier Inc. All rights reserved.

In plain English

This methods study used records from the South Carolina Chrysotile Asbestos study, covering 2,975 asbestos textile factory workers followed for lung cancer deaths from 1940 through 2001. The authors did not estimate how much asbestos raises lung cancer risk. Instead, they checked for three links that together create the 'healthy worker survivor bias'. This distortion can arise because staying at work, and so exposure, is not random. All three links were present. Person-years at or above the median cumulative chrysotile exposure (5.4 fiber-years/mL, counted up to the prior year) had about half the adjusted hazard of leaving work of those below it (hazard ratio 0.50, 95% CI 0.46 to 0.55). While workers were employed, 88.3% of person-years (95% CI 87.0 to 89.5) were classed as exposed to any asbestos. After they left work, none were. After leaving work, the adjusted hazard of lung cancer death was lower than at work (hazard ratio 0.52, 95% CI 0.29 to 0.94). The authors concluded that, for these data, exposure lagging or employment status adjustment will not reduce this bias. They had no smoking histories.

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: Methodological analysis of an existing occupational cohort, using de-identified records. It tested for the three component associations of the healthy worker survivor bias using Kaplan-Meier curves, Cox regression and logistic and linear regression. It did not estimate the effect of asbestos on lung cancer.
• Who was studied: 2,975 workers at an asbestos textile factory in the southern United States (the South Carolina Chrysotile Asbestos study cohort), followed for lung cancer mortality between 1940 and 2001, for 115,643 person-years in total. The original cohort was 3,072 people employed 6 months or more with at least 1 month of employment between 1940 and 1965. The authors excluded 97 who left work before age 18. At entry, median age was 24, 42% were female and 19% were non-Caucasian.
• Exposure: Occupational chrysotile asbestos exposure in textile work, estimated from a department, task and calendar-period job-exposure matrix. Exposure was expressed as annual and cumulative fiber-years/mL. Median cumulative exposure over follow-up was 5.4 fiber-years/mL. The other variable of interest was employment status (at work versus having left work).
• Main results:
  • Component 1 (prior exposure and leaving work), comparing person-years at or above the median cumulative exposure up to the prior year (5.4 fiber-years/mL) with those below it: adjusted hazard ratio for leaving active employment 0.50 (95% CI 0.46 to 0.55).
  • For each 100 fiber-year/mL increase in cumulative exposure up to the prior year, the adjusted hazard ratio for leaving active employment was 0.48 (95% CI 0.42 to 0.54).
  • With a 10-year exposure lag, the hazard ratio for leaving active employment per 100 fiber-year/mL was 0.71 (95% CI 0.61 to 0.83).
  • Component 2 (employment status and later exposure): 88.3% of actively employed person-years (95% robust CI 87.0 to 89.5) were classified as exposed to any asbestos.
  • No person-years after leaving employment were classified as exposed.
  • With a 10-year lag on the indicator of any exposure, the adjusted odds ratio linking employment status to lagged exposure was 0.13 (95% CI 0.12 to 0.15).
  • Component 3 (employment status and lung cancer death): the adjusted hazard ratio for lung cancer mortality, comparing time after leaving active employment with time actively employed, was 0.52 (95% CI 0.29 to 0.94).
  • Restricted to person-years with no asbestos exposure in the prior year, this hazard ratio was 0.42 (95% CI 0.06 to 3.11), which the authors describe as less precise.
• The authors' conclusion: The authors concluded that all three component associations of the healthy worker survivor bias were present. For these data, they said, exposure lagging or adjusting for employment status will not reduce the bias, and standard methods may give biased estimates of the effect of asbestos on lung cancer mortality.
• Limitations the authors state:
  • No information on individual smoking history or on intermittent time off work.
  • Exposure values were estimated by a regression model from a job/task exposure matrix, not measured for each person.
  • The survival analyses assumed that censoring was independent, given the measured covariates.
  • The authors did not test whether the associations differed by relevant characteristics.

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: Asbestos Exposure, Lung Cancer, Epidemiology, Confounding, confidence interval.

Related studies

• Estimating the effect of cumulative occupational asbestos exposure on time to lung cancer mortality: using structural nested failure-time models to account for healthy-worker survivor bias — A.I. Naimi, S.R. Cole, et al. (2014)
• Quality of life of survivors of malignant pleural mesothelioma in Japan: a cross sectional study — Y. Nagamatsu, I. Oze, et al. (2018)
• Mortality by cause of death and risk behaviors in farmers versus non-farmers: the importance of avoiding the healthy worker effect — G. Zhao, E. Ronda, et al. (2019)
• Analysis of occupational asbestos exposure and lung cancer mortality using the g formula — S.R. Cole, D.B. Richardson, et al. (2013)
• Mortality from occupational exposure to relatively pure chrysotile: a 39-year study — L. Sichletidis, D. Chloros, et al. (2009)
• Occupational exposure to carcinogens in Australian road transport workers — S. Si, R. Carey, et al. (2016)

Browse more research in Annals of Epidemiology, or return to medical research.