Analytical implications of epidemiological concepts of interaction

Our plain-English summary of Analytical implications of epidemiological concepts of interaction, by N. Pearce (International Journal of Epidemiology, 1989).

The study

• Author: N. Pearce
• Journal: International Journal of Epidemiology
• Published: 1989;18(4):976-80
• Cited by: 43 articles, as counted by NIH iCite
• Read it: PubMed · Publisher
• In J-M's library: Johns-Manville's company medical library held a copy (J-M no. 002542).

Abstract

In contrast to definitions based on statistical or biological concepts, Rothman has adopted an unambiguous epidemiological definition of interaction in which two factors are not 'independent' if they are component causes in the same sufficient cause. This leads to the adoption of additivity of incidence rates as the state of 'no interaction'. However, there are other considerations which generally favour the use of multiplicative models. This implies an apparent dilemma as to how an analysis can be conducted which combines the advantages of ratio measures of effect with the assessment of independence in terms of a departure from additivity. These apparently contradictory goals can be reconciled through the analysis of separate and joint effects. This approach is discussed with reference to studies of asbestos exposure, cigarette smoking and lung cancer.

In plain English

This methods paper discussed how to analyse 'interaction': whether two exposures together produce more or less disease than expected. No new data were collected. To illustrate, the author used lung cancer death rates from a study by Selikoff and colleagues of asbestos and cigarette smoke exposure. Asbestos exposure added 471.9 lung cancer deaths per 100,000 person-years in non-smokers and 736.3 in smokers. Compared with people with neither exposure, and adjusted for age and calendar period, the relative risks were 7.0 for smoking alone, 17.5 for asbestos alone and 32.7 for both. The author calculated that if the two effects added together, the combined figure would have been 23.5. The joint effect was therefore more than additive. The author stated this was consistent with asbestos and smoking acting together in at least one cause of lung cancer. The author added that using this to refute the idea that the two never biologically interact would depend on the untestable assumption of no unknown confounding or bias. The author concluded that these competing goals can be reconciled by analysing separate and joint effects.

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 commentary on how to analyse interaction between risk factors, illustrated with published data from another study; no new data were collected
• Who was studied: No study population of its own. Illustrative data came from a study by Selikoff and colleagues of lung cancer death rates. The reference list titles that study as one of amosite asbestos factory workers.
• Exposure: Asbestos exposure and cigarette smoking, alone and together, in relation to lung cancer death rates (illustrative data)
• Main results:
  • In the illustrative data, the lung cancer death rate difference due to asbestos exposure was 471.9 per 100,000 person-years in non-smokers and 736.3 per 100,000 person-years in smokers
  • Measured as a ratio, the effect of asbestos was lower in smokers (rate ratio 4.7) than in non-smokers (rate ratio 17.5)
  • Relative to people with neither exposure, with figures adjusted for age and calendar period, the relative risks were 7.0 for smoking only, 17.5 for asbestos only and 32.7 for both. The author computed 23.5 as the value expected if the effects were additive, so the joint effect was more than additive.
  • The author stated that the more-than-additive joint effect is consistent with asbestos and smoking being joint components of at least one sufficient cause of lung cancer
• The authors' conclusion: The author concluded that using ratio measures and assessing interaction as a departure from additivity can be reconciled by analysing separate and joint effects. The author also concluded that confounding should be evaluated before interaction.
• Limitations the authors state:
  • Estimating separate and joint effects may be difficult when the factors of interest are closely correlated
  • Using non-additivity to refute the hypothesis that asbestos and smoking never biologically interact depends on the untestable assumption of no unknown confounding or bias
  • Even in ideal circumstances, additivity would not rule out that asbestos and smoking interact

Topics

Medical terms in this study, explained in our Lexicon: Incidence, Asbestos Exposure, Lung Cancer.

Related studies

• Toxicology and epidemiology: improving the science with a framework for combining toxicological and epidemiological evidence to establish causal inference — H.O. Adami, S.C. Berry, et al. (2011)
• Estimation with vanishing baseline risk — R.M. Park (2012)
• Analytical characterization of cell-asbestos fiber interactions in lung pathogenesis — S. Yao, G. DellaVentura, et al. (2010)
• Multistage modelling of lung cancer mortality in asbestos textile workers — N. Pearce (1988)
• Asbestos and lung cancer: an analysis of the epidemiological evidence on the asbestos-smoking interaction — R. Saracci (1977)
• Inhibition of mesothelin-CA-125 interaction in patients with mesothelioma by the anti-mesothelin monoclonal antibody MORAb-009: Implications for cancer therapy — R. Hassan, C. Schweizer, et al. (2010)

Browse more research in International Journal of Epidemiology, or return to medical research.