Sir-We have read with interest the paper of Lassise et al. recently published by the IJE on the relation between use of intrauterine devices (IUD) and risk of invasive cervical cancer.' That study, conducted in five metropolitan areas in the US on 481 cases and 801 general population controls, showed a nonsignificant reduced risk of invasive cervical cancer associated with copper IUD use (relative risk [RR] 0.6, 95% confidence interval [CI] : 0.3-1.2), but not with inert IUD use (RR 1.1, 95% CI : 0.9-1.7). To offer further data on this issue we would like to add the following.

This study confirms the existence of and quantifies a number of strong socioeconomic correlates of cancer risk and indicates a few points open to additional investigation, such as the different pattern of risk for rectal and colon cancer, the strong negative gradient for endometrial cancer, and the absence of any clear association with education for cancers of the ovary, prostate, urinary tract, lymphomas, and myeloma.

Between June 1978 and January 1979 300 abortions were performed at the Bassini hospital in Milano, Italy; they were the first abortions performed after the promulgation of law 194 which made abortion legal in Italy. This number represented 57% of the number of requests for abortion during that period. Only 3% of abortions were performed before the 8th week of gestation; 19% of women were below 20, and 10% were minors. 39% of patients had not even completed elementary school education; 29% were housewives. Socioeconomic reasons represented 51% of all motivation for abortion; the partner was favorable to the decision in 90% of cases. Contraceptive practice prior to abortion was ineffective in 95.7% of patients; 40% of patients were young nulliparae, and 60% women with high parity. The Italian law does not contemplate, as yet, a follow-up visit after abortion.

Differences in the frequency of consumption of 30 selected foods and in the estimated intake of total calories and selected nutrients in relation to alcohol drinking, tobacco smoking, and education were described using information obtained from 1,774 controls of a case-control study of digestive tract cancers conducted in northern Italy. Heavy alcohol consumption, tobacco smoking, and lower level of education were associated with a diet poorer in several aspects, including lower consumption of fresh fruit and green vegetables and higher intake of specific indicator foods, such as sausages and canned meat. For instance, the mean number of portions of fresh fruit per week was 10.5 among male nondrinkers vs. 9.0 among heavy drinkers, 10.4 among male nonsmokers vs. 8.1 among heavy smokers, and 8.8 in less educated individuals vs. 10.7 among those more educated. Consequently, intake of beta-carotene, ascorbic acid, and calcium tended to be inversely related to alcohol and tobacco and directly related to education. Most associations were stronger in males, for whom alcohol consumption was also more common in less educated individuals. Calorie intake was directly related to alcohol consumption, largely reflecting calories provided by alcohol itself. However, alcohol drinking was also directly related to fat consumption. In both sexes, there was a strong positive correlation between cigarette smoking and coffee drinking. These results provide quantitative documentation that alcohol drinking, tobacco smoking, and education, three of the major determinants of cancer risks, were also correlates of dietary patterns and, hence, may exert an important confounding or modifying effect on the diet and cancer relationship.

Changing to cigarettes with a lower tar yield is not an effective means of reducing tobacco related morbidity from myocardial infarction.

The relationship between intake of various indicator foods and beverages and risk of prostatic cancer was assessed in 271 cases of prostatic cancer and 685 hospital controls recruited in two areas of northern Italy, the province of Pordenone and the greater Milan area. Increased risks were found for more frequent intake of meat [odds ratio (OR) in the highest vs. lowest consumption tertile = 1.4, 95% confidence interval (CI) 1.0-2.0], milk (OR = 1.6, 95% CI 1.1-2.4), fresh fruit (OR = 1.4, 95% CI 1.0-2.1), and vegetables (OR = 1.4, 95% CI 0.9-2.2). After allowance for the reciprocal confounding effect of various dietary habits, only frequent intake of milk seemed to be a significant independent indicator of prostatic cancer risk. There was also a clue that the unfavorable influence of frequent intake of a few food items (i.e. meat, fish, liver, ham and salami, milk and butter, and retinol) may be greater or restricted to older individuals (i.e., > or = 70 yrs of age). In conclusion, the present study confirms the presence of a moderate adverse effect of high intake of foods of animal origin, chiefly milk, while it suggests that a diet rich in fresh fruit and vegetables does not convey a protection.

To provide further insight on the relationship between macronutrients and colorectal cancer, overall and by specific subsite(s), we carried out between 1992 and 1996 in 6 Italian areas a case-control study on 1,953 individuals of both sexes with incident colorectal cancer (age range 19-74) and 4,154 controls (age range 19-74) in hospital with acute, non-neoplastic diseases. A validated food-frequency questionnaire was used, including questions on 78 foods or recipes and on individual fat-intake pattern. The risk of cancer of the colon and rectum increased with total energy intake (odds ratio in highest vs. lowest quintile 1.43 and 1.50, respectively). The risk also rose significantly with an increase of starch intake, whereas it moderately decreased with an increase of protein intake. Monounsaturated fat intake appeared uninfluential, while saturated fat intake showed a modest direct association with rectal cancer. Polyunsaturated fat intake was inversely associated with colon cancer risk, particularly with the right colon.

To evaluate risk factors for cervical intraepithelial neoplasia (CIN), data were collected in a case-control study based on 366 patients (58 with CIN class 1, 70 with CIN class 2, and 238 with CIN class 3) and 323 control subjects with normal cervical smears interviewed on selected days at the same screening clinics where cases had been identified. No relationship emerged between indicators of socioeconomic status (education and social class) and risk of mild/moderate (considered together) and severe dysplasia. A total of 55 (43%) patients with CIN class 1 or 2, 107 (45%) patients with CIN class 3, and 94 (29%) controls were current smokers. The corresponding relative risk (RR) estimates for current versus never smokers were 1.9 (95% confidence interval [CI] 1.2 to 3.0) for patients with CIN class 1 or 2 and 2.5 (95% CI 1.7 to 3.6) for patients with CIN class 3, and the risk increased with the number of cigarettes smoked per day. No relationship was observed between oral contraceptive use, parity, spontaneous or induced abortions and the risk of CIN, but patients tended to report earlier age at first birth than control subjects. Compared with women reporting their first birth before the age of 20 years, the risk estimates were 0.5 and 0.4, respectively, for patients with CIN 1 or 2 and patients with CIN 3 in women reporting first birth at 20 to 24 years of age. The risk estimates were 0.5 and 0.6 for those reporting their first birth at age 25 or later, but the trends in risk were not statistically significant. The number of sexual partners was directly associated with the risk for both histopathologic subgroups. Compared with women reporting no intercourse or their first intercourse after 22 years of age, women with first intercourse before the age 18 had a RR estimate of CIN class 1 or 2 of 2.3 and of CIN class 3 of 2.4, with the trends in risk being statistically significant. This study confirms considerable similarities in the epidemiology of mild/moderate and severe cervical dysplasia. In addition, it suggests consistency between the epidemiology of intraepithelial and invasive cervical neoplasia for risk factors that are likely to act on one of the first stages of the process of carcinogenesis (i.e., indications of sexual habits) but differences for hormone-mediated factors (i.e., reproductive variables or oral contraceptives).

To evaluate whether some form of mild immunosuppression may influence the geographical distribution of non-AIDS Kaposi's sarcoma (KS), we correlated incidence rates of KS and non-Hodgkin's lymphoma in individuals aged 60 or more in 18 European countries and Israel. Significant positive correlations emerged but, within highest risk countries (i.e. Italy and Israel), internal correlations were inconsistent.