Computer simulation of industrial hazards

Our plain-English summary of Computer simulation of industrial hazards, by E.G. Knox (British Journal of Industrial Medicine, 1973).

The study

• Author: E.G. Knox
• Journal: British Journal of Industrial Medicine
• Published: 1973;30(1):54-63
• Cited by: 10 articles, as counted by NIH iCite
• Read it: PubMed · Publisher · Free full text

Abstract

Knox, E. G. (1973).Brit. J. industr. Med.,30, 54-63. Computer simulation of industrial hazards. A computer simulation system for a range of industrial hazards provided for model experiments which manipulated (a) the sub-structure of an exposed population in terms of age-distributions and levels of exposure, (b) the nature of the dose/response relationship, (c) the latent interval and its variability, (d) normal life-table expectations, and (e) employment turnover rates. The development of the system led to clarification of terms and concepts with ambiguous current usages, notably in relation to latency. Distinction is made between the notions of `biological' and `observable' latent intervals. Hypothetical exercises with the model tested its technical validity and at the same time demonstrated in quantitative terms the relationships between `biological' and `observable' latent intervals, employment turnover rates, total mortalities, and the distribution of illnesses and death between those currently employed in the exposing industry, those employed elsewhere, and those retired. Prospects of success for personnel engineering techniques, which manipulate age-distributions of exposed work people in relation to diseases with long latent intervals, were examined. Published asbestos cancer data were used as a basis for specific model fitting and resulted in a numerical formulation of the exposure/response relationships. Severe exposure results in an increment of risk of death of about 0·02 unit per person per annum for those exposed for around six years, but with higher rates for shorter exposures and lower rates for longer ones. The mean biological latent interval was about 25 years with a coefficient of variation of about 25%. These suppositions explained a range of published data comprehensively and at the same time predicted that (a) persons exposed at severe levels for a working lifetime of 50 years have a 40% risk of dying from asbestos cancer, and (b) industrial populations with moderate to high turnover rates effect a form of extended dose sharing, and non-linearity of the exposure/response relationship results in substantially more deaths than would be the case if the turnover rate were lower.

In plain English

This single-author paper described a FORTRAN simulation model that predicted deaths from industrial hazards in a steady-state workforce, using no new data on workers. In hypothetical runs, longer latent intervals (time from exposure to disease) shifted deaths from current employees to ex-employees and retired people. With a five-year latent interval and 10% annual turnover, 54% of cases occurred in employees; at 10 years, 29%. The author distinguished a 'biological' latent interval from observed intervals, which the author regarded as biased estimates that tend to be low. The model was fitted mainly to Newhouse (1969) asbestos cancer data, with an assumed 25-year mean biological latent interval. It predicted around 40% eventual asbestos cancer mortality in those most severely exposed for 50 years. A simulated workforce with 10% turnover gave 10.37 asbestos cancer deaths per year, in employees and ex-employees, per thousand current employees in moderate to severe exposure grades. That exceeded the 9.24 for a zero-turnover population that was all severely exposed. The author concluded that labour turnover harms the population as a whole under the non-linear relationship assumed for asbestos, in which shorter and lower exposures carried proportionally more risk. Several inputs were called guesses or arbitrary.

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: Computer simulation (deterministic, steady-state model) tested on hypothetical data, then fitted to published asbestos cancer data from other studies
• Who was studied: Simulated industrial populations. These were hypothetical homogeneous populations in the test runs. For the asbestos application, the simulated workforce had four sub-groups: younger high-turnover and older low-turnover groups, each split into severe and moderate exposure. Inputs were drawn from published studies such as Newhouse (1969).
• Exposure: In the test runs, an abstract exposure measured as fractions of a lethal dose. In the asbestos application, asbestos dust at an annual exposure rate of 0.02 lethal unit for a severely exposed group and one-third of that for a moderately exposed group. Slight exposures were ignored. No fiber types were specified.
• Main results:
  • In hypothetical runs with 10% annual turnover, 54% of all cases occurred in current employees with a five-year latent interval. This fell to 29% with a 10-year interval and to less than 10% with a 20-year interval.
  • In practice, ascertained cases are likely to be biased towards shorter intervals, so estimated mean latent intervals tend to be low. The author also noted that measuring from first exposure would tend to overestimate the interval for prolonged exposures.
  • For a model (biological) latent interval of 35 years, the observable mean interval was only 24 years with complete follow-up. It was 15.1 years if only cases still in employment were counted.
  • Employing older rather than younger workers reduced total specific mortality only modestly for short latent intervals: at 5 years, the older population had 92% of the younger population's mortality. The reduction was larger at longer intervals: 68% at 15 years, 42% at 25 years and 33% at 30 years.
  • Model fitting to the Newhouse data implied an eventual observable asbestos cancer mortality of around 40% in those most severely exposed for 50 years.
  • The simulated asbestos workforce with 10% turnover gave 10.37 asbestos cancer deaths per year among employees and ex-employees, per thousand current employees working in moderate to severe grades of exposure (slight exposures were ignored). This exceeded the 9.24 per thousand from a zero-turnover population all severely exposed for 50 years. Only 14.1% of these deaths occurred in current employees.
  • In the asbestos simulation, the modal age at death was 50 years and the mean age was 53.1 years.
• The authors' conclusion: The authors concluded that the biological latent interval to death from asbestos cancer is about 25 years, and that exposure for a working lifetime at severe levels results in a 40% risk of death from asbestos cancer. They also concluded that, under the probable non-linear relationship assumed for asbestos, in which yields fall with increasing intensity and duration of exposure, the dose-sharing caused by labour turnover is harmful to the population as a whole. Elsewhere the author noted that under a different non-linear relationship, a power function, increased turnover would be beneficial.
• Limitations the authors state:
  • The employment turnover rate used in the asbestos simulation was a guess, as no recorded data were found.
  • The exposure-intensity relationship (equation 4) was arbitrary, because exposure intensity was not expressed in physical terms.
  • The 25-year latent interval was chosen arbitrarily, although the author said it was compatible with published data.
  • The initial risk estimates taken from the Newhouse data were regarded as minimum values because follow-up was incomplete.
  • The author stated that the practical usefulness of computer modelling in this field was unproven.

Topics

Medical terms in this study, explained in our Lexicon: Latency.

Related studies

• Is there a single spot size and grid for intensity modulated proton therapy? Simulation of head and neck, prostate and mesothelioma cases — L. Widesott, A.J. Lomax, et al. (2012)
• Latency periods in neoplastic diseases — A.P. Polednak (1974)
• The Production of Corporate Research to Manufacture Doubt About the Health Hazards of Products: An Overview of the Exponent Bakelite® Simulation Study — D. Egilman (2018)
• Retrospective cohort study of a microelectronics and business machine facility — S.R. Silver, L.E. Pinkerton, et al. (2014)
• Mesothelioma mortality in asbestos workers: implications for models of carcinogenesis and risk assessment — J. Peto, H. Seidman, et al. (1982)
• Asbestos as an industrial health hazard — (1973)

Browse more research in British Journal of Industrial Medicine, or return to medical research.