Assessing the power and quality of epidemiologic studies of asbestos-exposed populations

Our plain-English summary of Assessing the power and quality of epidemiologic studies of asbestos-exposed populations, by D.L. Davis, B. Mandula, J.V. Van Ryzin (Toxicology and Industrial Health, 1985).

The study

• Authors: D.L. Davis, B. Mandula, J.V. Van Ryzin
• Journal: Toxicology and Industrial Health
• Published: 1985;1(4):93-110
• Cited by: 2 articles, as counted by NIH iCite
• Read it: PubMed · Publisher

Abstract

This paper briefly discusses criteria for evaluating epidemiologic studies for risk assessment purposes, using asbestos as an example. Asbestos is one of the few carcinogens for which substantial data exist on exposures to humans. However, there are major difficulties in using these data for conducting risk assessments. In particular, exposure data are often incomplete, and risk assessments usually involve extrapolating from the higher exposures of the occupational environments to the lower levels typically encountered in the nonoccupational environment. The term "asbestos" refers to the fibrous form of several minerals, and levels of exposures to these fibers are not easily assessed. Criteria for evaluating epidemiologic studies used in an Ontario Royal Commission report on asbestos are discussed. The importance of considering the statistical power of studies to detect an excess risk is examined using as examples major cohort studies of asbestos-exposed workers, as summarized in a report by the U.S. National Research Council.

In plain English

This paper was a methodological discussion rather than a new study of workers. The authors used asbestos to illustrate how epidemiologic studies are judged for risk assessment. They described problems in measuring asbestos exposure and the criteria an Ontario Royal Commission used to score six asbestos mortality studies. They then explained statistical power, the chance a study will detect a true excess risk. They applied it to 23 major cohort studies of asbestos-exposed workers summarized by the U.S. National Research Council. Most studies had fairly low power to detect a relative risk of 1.5 for lung cancer, yet virtually all detected an excess. For gastrointestinal (GI) cancers, only five showed a statistically significant increase. Fewer than half of the 16 studies without a significant GI excess had power above 0.8 to detect a relative risk of 2.5. Using a 5% one-sided significance test, the authors calculated less than a 1% probability that chance alone would lead 5 of 21 studies to show a significantly increased relative risk. They concluded that power matters both in interpreting studies that show no excess risk and in designing new ones.

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 review and discussion paper, with power calculations applied to previously published cohort studies (no new data collected)
• Who was studied: 23 major published epidemiologic (cohort) studies of asbestos-exposed workers, as summarized by the U.S. National Research Council (1984). Per the table notes, the studies came from the United States, Britain, Canada, Finland, Italy, Denmark and Australia. The paper also discussed six studies scored by the Ontario Royal Commission.
• Exposure: Occupational asbestos exposure. The paper names chrysotile, crocidolite and amosite as the major industrially used forms. Power was calculated for detecting a relative risk of 1.5 for GI and respiratory cancer at a 0.05 significance level.
• Main results:
  • Lung cancer, across the cohort studies: most studies had fairly low power to detect a relative risk of 1.5, but the observed relative risk was high enough that virtually all studies detected it.
  • GI cancers, across the cohort studies: only five studies showed a statistically significant increase.
  • Nicholson et al. (1979) study, respiratory cancer: with a true relative risk of 2.5 and about 11 expected deaths, the study had more than a 98% probability (power 0.986) of detecting the excess.
  • Nicholson et al. (1979) study, GI cancer: with a true relative risk of 1.1 and 9.5 expected deaths, power was 0.090, only a 9% probability of detecting it.
  • Sixteen of the occupational studies did not show a statistically significant excess of GI cancers. Fewer than half of these negative studies had power greater than 0.8 to detect a relative risk of 2.5 for GI cancers.
  • In the discussion of GI cancers, the authors stated there was less than a 1% probability that chance alone would lead 5 of 21 studies to show a significant increased relative risk, using a 5% one-sided significance test.
  • Citing NRC (1984) and CPSC (1983), the authors reported that exposure-response relationships for lung cancer were evident in almost half of the cohort studies.
• The authors' conclusion: The authors concluded that power is an important factor in epidemiologic studies, both for interpreting studies that show no excess risk and for designing further studies.
• Limitations the authors state:
  • The power formula used is a simplification that assumes the expected number of deaths is without error.

Related studies

• Assessing quality of life when planning and evaluating worker notification programs: two case examples — B.E. Meyerowitz (1993)
• Ordinal regression models for epidemiologic data — B.G. Armstrong, M. Sloan (1989)
• Asbestos risk among full-time workers in an electricity-generating power station — A. Hirsch, L. Di Menza, et al. (1979)
• Critical review of epidemiologic studies related to ingested asbestos — G.M. Marsh (1983)
• Comparing epidemiologic studies of ingested asbestos for use in risk assessment — L.S. Erdreich (1983)
• Additional thoughts on the review of epidemiologic studies related to ingested asbestos — G.M. Marsh (1983)

Browse more research in Toxicology and Industrial Health, or return to medical research.