Article Text


0385 Bias in Exposure Assessment from Worst-Case Selection of Workplaces in OSHA’s Integrated Management Information System Databank IMIS
  1. Philippe Sarazin1,
  2. Laurel Kincl2,
  3. Igor Burstyn3,
  4. Jérôme Lavoué1,4
  1. 1University of Montreal, Montreal, Quebec, Canada
  2. 2Oregon State University, Corvallis, Oregon, USA
  3. 3Drexel University, Philadelphia, Pennsylvania, USA
  4. 4University of Montreal Hospital Research Centre, Montreal, Quebec, Canada


Objectives The Integrated Management Information System (IMIS), with over 1M measurements taken by inspectors of the Occupational Safety and Health Administration, is an important source of information for occupational epidemiology. We assessed the association of the reason for conducting inspection with the reported levels of chemical exposure in IMIS.

Method Time weighted averaged measurements made during each of the un-programmed inspection types (employee complaint, referral by safety officer, follow-up, monitoring) were compared to those made during programmed inspections for 50 chemicals. Ratios of the median of detected results (dM), and the differences in the proportion of non-detects (dPr) for each category compared to the programmed inspections were calculated for each chemical.

Results The analysis included 218 916 measurement records. 32% were collected during programmed inspections, 48% -- complaints, 13% -- referral, 5% -- follow-up, and 2% -- monitoring. The detected concentrations were similar for complaint (dM=0.98, interquartile range across chemicals, IQR=[0.83;1.11]) and referral (dM=0.91, IQR=[0.76;1.08]) inspections and greater for follow-up (dM=2.18, IQR=[1.38;3.13]) and monitoring (dM=1.59, IQR=[1.24;2.44]) inspections relative to presumed representative inspections. Similarly, the proportion of non-detects were similar to programmed inspections during complaint-driven (dPr=3%, IQR=[-1;8]) and referrals (dPr=0%, IQR=[-6;5]) and lower during follow-up (dPr=-11%, IQR=[-19;-2]) and monitoring (dPr=-8%, IQR=[-12;3]) inspections.

Conclusions Despite the absence of consistent differences across chemicals for the most frequent categories, exposure levels during non-programmed surveys can be significantly higher than those obtained during presumably representative measurement campaigns. Great care has to be taken in determining typical exposure distributions from OSHA’s IMIS data.

Statistics from

Request permissions

If you wish to reuse any or all of this article please use the link below which will take you to the Copyright Clearance Center’s RightsLink service. You will be able to get a quick price and instant permission to reuse the content in many different ways.