The Combined Effects of Alcohol and Tobacco on Esophageal Cancer Risk: An Epidemiological and Statistical Modeling Approach

Abstract

This study investigates the synergistic effects of alcohol and tobacco consumption on esophageal cancer risk through comprehensive statistical modeling of case-control data. Using logistic regression with interaction terms, we compared additive (AIC=221.39) versus interactive (AIC=233.94) risk models, finding no significant improvement in fit from interaction terms (χ²=5.45, p=0.79). Age-adjusted odds ratios revealed strong independent effects: highest alcohol consumption (120+ g/day: OR=65.1, 95% CI[20.9-229.7]) and heaviest tobacco use (30+ g/day: OR=8.6, 95% CI[2.3-30.1]). Contingency analyses showed non-significant alcohol-cancer associations (χ²=4.21, p=0.24) but suggested dose-response trends. Alternative modeling approaches including Poisson (deviance=78.40) and multinomial regression (AIC=77.74) confirmed robustness of findings. Propensity score matching (nearest-neighbor, n=29 pairs) and bootstrap validation (500 replicates) supported model stability. Visual analytics through correspondence analysis (χ²=7.39, p=0.60) and effect plots elucidated complex exposure-risk relationships. The results demonstrate significant independent effects of alcohol and tobacco, while suggesting their combined impact may be additive rather than multiplicative in this population. These findings underscore the importance of dual abstinence strategies in esophageal cancer prevention while highlighting methodological considerations for analyzing interacting risk factors in epidemiological studies.