Interaction of Alcohol and Tobacco in Laryngeal Cancer

Our plain-English summary of Interaction of Alcohol and Tobacco in Laryngeal Cancer, by W.D. Flanders, K.J. Rothman (1982).

The study

• Authors: W.D. Flanders, K.J. Rothman
• Published: American Journal of Epidemiology, Volume 115 No3. October 26th 1981 (1982)
• Type of article: Journal article
• Our copy: from Johns-Manville's own medical library. The article is still under copyright, so we describe it here rather than post it.

In plain English

This study re-analyzed existing case-control data to measure how alcohol and tobacco together relate to laryngeal cancer in men. It did not study asbestos, which the authors mentioned only as a known risk factor. The authors used two sources. The first was interview data from the Third National Cancer Survey (seven cities and two states, 1969 to 1971), covering 87 male laryngeal cancer cases and 956 male controls with other cancers. The second was published data from a study by Wynder and colleagues, covering 224 male cases and 414 male controls. They estimated an index comparing the observed effect of combined exposure with the effect expected if the two effects simply added together; a value of 1 means no interaction. The estimate was 1.5 (90 per cent confidence limits 0.6 to 2.3) for the survey data and 2.5 (1.1 to 3.9) for the Wynder data. Adjusting for age in the survey data gave 1.7. The authors concluded there was moderate synergy. They cautioned that sparse data and uncontrolled confounding could affect the estimates.

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: Analysis of case-control data from two earlier sources: interview data from the Third National Cancer Survey, analyzed with a case-control approach, and published data from a matched case-control study by Wynder et al.
• Who was studied: Men only. Third National Cancer Survey (seven cities and two states, 1969 to 1971): 87 male laryngeal cancer cases and 956 male controls with cancers at other sites, excluding sites linked to alcohol, tobacco or occupational exposures. Wynder et al. data: 224 male cases and 414 male hospital controls.
• Exposure: Alcohol and tobacco use (cigarettes, cigars, pipes). Total lifetime consumption in the survey data; daily use in the Wynder data. Asbestos was mentioned only as a background risk factor and was not studied.
• Main results:
  • Third National Cancer Survey data: the maximum-likelihood estimate of the interaction index was 1.5, with 90 per cent confidence limits from 0.6 to 2.3 (1 means no interaction).
  • Third National Cancer Survey data: a weighted average of the nine category-specific estimates was 1.4.
  • Third National Cancer Survey data: after controlling for age, the estimated interaction index was 1.7.
  • Wynder et al. data: the maximum-likelihood estimate of the interaction index was 2.5, with 90 per cent confidence limits from 1.1 to 3.9.
  • Third National Cancer Survey data: the smoothed rate ratio estimate for the heaviest combined alcohol and tobacco use was 18.5, compared with the lowest category of both.
  • Using only categories that were not open-ended, the estimate was 2.5 for the survey data and 1.4 for the Wynder data.
  • The authors explained that an index of 1.5 implies the combined effect of alcohol and tobacco is 50 per cent greater than if their effects simply added together.
• The authors' conclusion: The authors concluded that there was moderate synergy between alcohol and tobacco in increasing the risk of laryngeal cancer.
• Limitations the authors state:
  • Data were sparse for heavy alcohol and light tobacco use, so estimates for those categories were unstable.
  • Uncontrolled confounding between alcohol and tobacco within categories may have raised the interaction estimate.
  • The Wynder data had no cases among nonsmokers who were moderate or heavy drinkers. This underestimated the effect of alcohol among nonsmokers, which inflates the interaction estimate.
  • Adding 0.5 to each cell frequency to avoid division by zero may introduce bias.
  • The published Wynder data were not detailed enough to control for age.

Topics

Medical terms in this study, explained in our Lexicon: Laryngeal Cancer, Confounding.

Browse return to medical research.