Comparison of standard methods with G-estimation of accelerated failure-time models to address the healthy worker effect.

Details

• Personal Author:
  Chevrier J ; Eisen EA
• Description:
  Background and Objective: The health consequences of occupational exposures may be underestimated because workers who are less healthy are more likely to leave work or transfer to jobs with lower exposure. We compared results obtained using standard methods to control for the healthy worker effect (HWE) with g-estimation of accelerated failure-time models in a mortality study of autoworkers exposed to metalworking fluids (MWF). Methods: Annual exposure to oil-based MWF was estimated for about 45 000 workers in three Michigan automobile manufacturing plants. To adjust for HWE, we used standard Cox proportional hazards models controlling for time since hire, cumulative time off work, employment status, or restricted to individuals with >10 years of follow up, and compared results with those obtained using g-estimation. We investigated associations with all-causes of death combined, all cancer mortality, chronic obstructive pulmonary disease (COPD), heart disease, and specific cancers previously associated with oil-based MWF exposure. Results: While standard models suggest that exposure to oil-based MWF has a null or protective effect for all causes combined, all cancers, COPD, and heart disease mortality, the g-estimation results suggest that exposure may be causally related to these outcomes. The largest increases in hazard ratios (HRs) were observed for mortality due to diseases that cause more disability (ie, COPD and heart disease) based on Global Burden of Disease disability weights. In addition, the increase in HR was largest for rectal cancer, which has higher disability weights than bladder and laryngeal cancers. Conclusion: Bias may arise because leaving work is associated with mortality, determines future exposure and is predicted by past exposure and employment history. G-estimation accounts for health status being both a confounder and an intermediate variable between exposure and disease. The results support the benefits of g-estimation relative to traditional models to control for the bias due to HWE. [Description provided by NIOSH]
• Subjects:
  Automobiles Cardiovascular Diseases Cardiovascular System Lung Metals Morbidity Mortality Neoplasms Occupational Health Respiratory System Respiratory Tract Diseases Risk Assessment Risk Factors Safety Workplace

  More +
• Keywords:
  Workers; Exposure-levels; Risk-factors; Work-environment; Models; Automotive-industry; Metalworking-fluids; Mortality-rates; Morbidity-rates; Cancer; Pulmonary-system-disorders; Pulmonary-system; Cardiovascular-system-disease; Cardiovascular-system; Cardiopulmonary-system; Cardiopulmonary-system-disorders
• ISSN:
  1351-0711
• Document Type:
  Text
• Funding:
  Grant
• Genre:
  Abstract or Summary
• Place as Subject:
  California ; OSHA Region 9
• CIO:
  National Institute for Occupational Safety and Health (NIOSH)
• Topic:
  Workplace Safety & Health
• Location:
  North America
• NIOSHTIC Number:
  nn:20045664
• Citation:
  Occup Environ Med 2010 Sep; (Suppl 1):A21
• Federal Fiscal Year:
  2010
• Performing Organization:
  University of California, Berkeley
• Peer Reviewed:
  False
• Start Date:
  20070801
• Source Full Name:
  Occupational and Environmental Medicine
• Supplement:
  1
• End Date:
  20110731
• Collection(s):
  National Institute for Occupational Safety and Health
• Main Document Checksum:
  urn:sha-512:7dc847c09aedf140bc731323e8446b4703641ae07d6ad9e7b7238120dab30b8f7dcaa9b938d34d035f87873a0fe7fec262d5d13dcaf290925445e82b6de3b57d
• Download URL:
  https://stacks.cdc.gov/view/cdc/248273/cdc_248273_DS1.pdf
• File Type:
  [PDF - 168.02 KB ]