From personal interviews obtained for 7,518 incident cases of invasive cancer from the population-based Third National Cancer Survey, the quantitative lifetime use of cigarettes, cigars, pipes, unsmoked tobacco, wine, beer, hard liquor, and combined alcohol were recorded, as well as education and family income level. In an initial screening analysis of these data, Mantel-Haenszel 2 X 2 contingency tabulations and multiple regression analyses were used to compare each specific cancer site with controls from other sites to test for associations with the "exposure variables." Significant positive associations with cigarette smoking were found for cancers of the lung, larynx, oral cavity, esophagus, stomach, pancreas, bladder, kidney, and uterine cervix. Other forms of tobacco were associated with cancers of the oral cavity, larynx, lung, and cervix. Consumption of wine, beer, hard liquor, and all combined showed positive associations with neoplasms of the oral cavity larynx, esophagus, colon, rectum, breast, and thyroid gland. College educaton and high income both showed positive associations with cancers of the breast, thyroid gland, uterine corpus, and melanomas in males. These same indicators of high socioeconomic status showed inverse associations with invasive neoplasms of the uterine cervix, lung, lip-tongue, and colon in females. College attendance (but not income) showed an inverse association with stomach cancer and positive association with pancreatic cancer in males. Still other tumor sties showed "suggestive" associations with each of these exposure variables. In the analyses producing these results, age, race, sex, smoking, drinking, education, income, parity, foreign birth, marital status, and geographic location were used as stratification variables separately or in combination when appropriate to assess and control for their potentially confounding affects and to examine results in different strata to assess interaction.
