Retrospective Exposure Assessment for Occupational Disease of an Individual Worker Using an Exposure Database and Trend Analysis

Our plain-English summary of Retrospective Exposure Assessment for Occupational Disease of an Individual Worker Using an Exposure Database and Trend Analysis, by D.A. Chung, R.R. Yang, D.K. Verma, et al (Journal of Occupational and Environmental Hygiene, 2015).

The study

• Authors: D.A. Chung, R.R. Yang, D.K. Verma, J. Luo
• Journal: Journal of Occupational and Environmental Hygiene
• Published: 2015;12(12):855-65
• Type of article: Historical article
• Cited by: 3 articles, as counted by NIH iCite
• Read it: PubMed · Publisher

Abstract

This article outlines a hierarchy of data required for retrospective exposure assessment for occupational disease of an individual worker. It then outlines in a step-wise manner how trend analysis using a relatively large exposure database can be used to estimate such exposure. The process of how a large database containing exposure measurements can be prepared for estimating historic occupational exposures of individual workers in relation to their illnesses is described. The asbestos subset from a large government collected air monitoring database called Medical Surveillance (MESU) was selected to illustrate the cleaning and analysis processes. After unidentifiable values were removed, the cleaned dataset was examined for possible sources of variability such as changes to sampling protocol. Limit of detection (LOD) values were substituted for all non-detectable values prior to the calculation of descriptive statistic using left censored analysis methods (i.e., maximum likelihood estimation (MLE), Kaplan Meier (KM), and simple substitution). The JoinPoint Regression Program was used to perform trend analysis and calculate an annual percentage change (APC) value for the available sampling period. An asbestos case study is presented to illustrate how the APC can then be combined with more recent job and/or process specific exposure data to estimate historic levels. The MESU asbestos dataset contained 1,610 samples from 1984-1995. An average of 17% of this data was left censored. The asbestos air sampling methods in Ontario changed around 1990. LOD values of 0.06 f/cc and 0.02 f/cc were substituted for LOD values pre- and post-1990, respectively. The annual mean fiber levels for the MLE method were an average of 44% lower than KM and substitution methods. The corresponding APC for MLE method was -6.5% and -7.7% for KM and simple substitution. The findings of this paper illustrate how the temporal trend of an exposure databases can be used to efficiently estimate historic contaminant levels in the presence of limited historical information.

In plain English

This was a methods paper from Ontario, Canada. The authors showed how a government air-monitoring database could be prepared and used to estimate one worker's past asbestos exposure. They used the asbestos portion of the Ontario Ministry of Labour's Medical Surveillance (MESU) database. After removing entries whose meaning was unknown, 1,610 air samples from 1984 to 1995 remained, and 17% were below the detection limit. Using two statistical methods, they estimated an average change in asbestos levels of -6.5% per year (95% confidence interval -22.0% to 12.1%) and -7.7% per year (-20.3% to 6.8%). No point where the trend changed direction or speed was found. Projecting backward, they estimated Ontario levels in 1972 at about 2.3 fibers per cubic centimetre (95% confidence interval 0.4 to 16.2). For an example carpenter/assistant who worked from 1978 to 1980, they estimated exposure during asbestos-handling work at about 0.3 fibers per cubic centimetre (95% confidence interval 0.05 to 1.67). The authors concluded the approach can estimate changes in exposure levels. They cautioned that projections should extend only a few years beyond the measured period.

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: Methods paper: preparation and trend analysis (JoinPoint regression) of a government air-monitoring database, with an illustrative individual case study in an appendix
• Who was studied: Asbestos air samples from the Ontario Ministry of Labour MESU database, collected across industrial sectors in Ontario, Canada (Mining and Construction excluded); 1,610 samples from 1984 to 1995 after cleaning. The appendix case is a single worker employed as a carpenter/assistant from 1978 to 1980.
• Exposure: Airborne asbestos fiber levels (fibers per cc) from area and personal samples. Fiber types are not specified. No job titles were recorded.
• Main results:
  • About 28% of the asbestos data points were alphanumeric values of unknown meaning and were excluded.
  • 17% of the cleaned asbestos dataset was below the detection limit (left censored).
  • Limit of detection values of 0.06 f/cc (pre-1990) and 0.02 f/cc (post-1990) were substituted for non-detectable values.
  • Annual mean fiber levels calculated by the MLE method were on average 44% lower than those from the KM and substitution methods.
  • Average annual percent change in asbestos levels, 1984 to 1995, by the MLE method: -6.5 (95% confidence interval -22.0%, 12.1%).
  • Average annual percent change in asbestos levels, 1984 to 1995, by the KM method: -7.7 (95% confidence interval -20.3%, 6.8%).
  • No change points were identified in either the MLE or the KM trend line.
  • Using the KM rate, Ontario asbestos levels in 1972 were estimated at about 2.3 f/cc (95% confidence interval 0.4, 16.2 f/cc).
  • Comparison of this 1972 estimate with an earlier Ministry of Labour report by Rajhans et al. (average about 3.5 f/cc) found no statistical evidence that the two means differed (z = -0.14, p = 0.89).
  • In the appendix case (carpenter/assistant, 1978 to 1980), exposure during asbestos-handling work was estimated at about 0.3 f/cc (95% confidence interval 0.05, 1.67).
• The authors' conclusion: The authors concluded that simple descriptive statistics from a large database, combined with JoinPoint regression, can generate a reliable estimate of changes in exposure levels. They also concluded that the resulting annual percent change may be an adequate method for extrapolating historical exposure levels.
• Limitations the authors state:
  • Projections should extend only a few years outside the period covered by the measurements.
  • The time trend approach is limited where confidence intervals around the trend values are large.
  • The asbestos data had no job title identifiers, so job-specific trends could not be built.
  • Alphanumeric values of unknown meaning were excluded from the dataset.
  • The dataset lacked details on calibration, analytical methods and sampling strategies, so systematic errors could not be further reduced.
  • The difference between MLE and the other methods may reflect that the data did not follow the assumed lognormal distribution.
  • Because exposure trends may not follow a consistent pattern, simple extrapolation beyond the trend line may not be appropriate.

Topics

Medical terms in this study, explained in our Lexicon: Occupational Disease, Occupational Exposure, Air Sampling.

Related studies

• Development of an exposure measurement database on five lung carcinogens (ExpoSYN) for quantitative retrospective occupational exposure assessment — S. Peters, R. Vermeulen, et al. (2012)
• Incidence of Cancer and Asbestos-Related Diseases among Residents Living near Abandoned Asbestos Mines in South Korea: A Retrospective Cohort Study Using National Health Insurance Database — K. Kwak, K.E. Zoh, et al. (2021)
• Global trends in mortality from malignant mesothelioma: Analysis of WHO mortality database (1994-2013) — O. Abdel-Rahman (2018)
• Publication trend, resource utilization, and impact of the US National Cancer Database: A systematic review — C. Su, C. Peng, et al. (2018)
• Occupational exposure to mineral fibres: analysis of results stored on colchic database — E. Kauffer, R. Vincent (2007)
• EV@LUTIL: An open access database on occupational exposures to asbestos and man-made mineral fibres — E. Orlowski, S. Audignon-Durand, et al. (2015)

Browse more research in Journal of Occupational and Environmental Hygiene, or return to medical research.