To compare the separate and combined effects of alcohol drinking and smokin
g between the 2 sites, we evaluated 274 men with oral cancer, 364 with phar
yngeal cancer and 1,254 controls, frequency-matched for age and area of res
idence, from Italy and Switzerland. Extremely elevated risk increases for o
ral cancer (odds ratio, OR = 228) and pharyngeal cancer (OR = 100) were fou
nd for the highest joint level of drinking (greater than or equal to 77 dri
nks/week) and smoking (greater than or equal to 25 cigarettes/day), Ratios
of ORs between oral cancer and pharyngeal cancer vs. controls, obtained by
polytomous logistic regression, suggested that the risk increase for oral c
ancer was about 2-fold greater than that for pharyngeal cancer at each comb
ined level of smoking and drinking, except at low levels of drinking in smo
kers. A clear departure from risk difference additivity was present for bot
h oral and pharyngeal cancer in individuals heavily exposed to both factors
versus non-smoking abstainers/light drinkers. Our findings thus help expla
in observations from descriptive epidemiology that, if smoking level in a p
opulation does not change substantially, but alcohol consumption increases,
increase in oral cancer would be greater than at any other site in the upp
er aero-digestive tract, including cancer of the pharynx. (C) 1999 Wiley-Li
ss, Inc.