The separate effect of alcohol and tobacco on laryngeal cancer was analysed in two case -control studies from Italy and Switzerland, comprising 40 non-smoking and 68 non-drinking cases, and 160 non-smoking and 161 non-drinking controls. The multivariate odds ratio was 2.46 for heavy drinkers non-smokers, and 9.38 for current smokers non-drinkers. Alcohol drinking and tobacco smoking increase the risk of laryngeal cancer, and show a multiplicative effect on risk (IARC, 1986(IARC, , 1988Tuyns et al, 1988;Franceschi et al, 1990;Sankaranarayanan et al, 1990;Dosemeci et al, 1997). In developed countries, laryngeal cancer is extremely rare in individuals who do not smoke and do not drink alcohol and information on risk in such people is therefore limited.A hospital-based US study found an association with tobacco in non/moderate alcohol drinkers, but had no information on alcohol in non-smokers (Wynder et al, 1976). A case -control study from Canada, reported an increasing risk of laryngeal cancer in relation to alcohol drinking in non-smokers, based, however, on three case -control pairs only (Burch et al, 1981). A case -control study conducted in Italy, Spain, Switzerland and France reported odds ratios (OR) of 1.7 for 580 g day 71 of alcohol among nine nonsmoking cancer cases of the hypopharynx and epilarynx, and of 6.7 for 540 g day 71 of alcohol among 22 non-smoking cancer cases of the endolarynx (Tuyns et al, 1988). A Polish study found a strong association with number of cigarettes among non-vodka drinkers, but the reference category was based on a single case (Zatonski et al, 1990).
PATIENTS AND METHODSThe separate effect of alcohol and tobacco on laryngeal cancer risk was analysed in the combined data of two case -control studies, including a uniquely large number of non-smoking and non-drinking subjects. The first study, conducted between 1986 and 1992 in the province of Pordenone and the greater Milan area, northern Italy, included 162 subjects with incident, histologically confirmed cancer of the larynx La Vecchia et al, 1990). The second one was conducted between 1992 and 2000 in the province of Pordenone and the greater Milan area, Italy, and in the Swiss Canton of Vaud, on 527 incident, histologically confirmed laryngeal cancer cases. Forty non-smoking cases (median age 60 years, range 30 -72) and 68 non-drinking ones (median age 61 years, range 42 -74) were identified in the two studies, and included in the present analysis. Non-smoking cases included 20 glottis, two supraglottis and 18 other or unspecified laryngeal cancers; non-drinking ones included 21 glottis, 12 supraglottis and 35 other or unspecified laryngeal cancers. They were matched on study, sex, age and study center to 160 non-smoking controls (median age 60 years, range 31 -79) and 161 non-drinking controls (median age 59 years, range 40 -77), selected from a database of 4781 patients, admitted to the same network of hospitals as cases, for a wide spectrum of acute, non-neoplastic conditions, unrelated to alcohol consumption and tobacc...