Tumori, 77: 1-6, 1991 CANCER MORTALITY IN ITALY, 1985-1987 Adriano Decarli 1 and Carlo La Vecchia• (' Istituto di Biometria e Statistica Medica, Universita di Milano, lstituto Nazionale Tumori, Milano and lstituto di Statistica e Ricerca Operativa, Universita di Trento, Trento, 2 lstituto di Ricerche Farmacologiche «Mario Negri», Milano, and Institute of Social and Preventive Medicine, University of Lausanne, Lausanne, Switzerland) Data and statistics on cancer mortality in Italy from 1955 to 1984 per each sex, age group and single calendar year or quinquennium have been published (6). We now present an updtate of this work, based on aggregate death certifica- tion data of 3-year period 1985-87 1 • The material and methods are similar to those previously used (1, 6). Briefly, death certification numbers by cause and estimates of the resident population, stratified by sex and quinquennia of age, were derived from official publications of the Central Institute of Statistics (ISTAT) (2, 3). All cancers, originally classified according to the standard International Classifica- tion of Diseases (ICD), Ninth Revision, were grouped ac- cording to the same 30 categories (plus a broad category of « others and non specified ») adopted for previous reports. In particular, all intestinal sites, all skin cancers, cancers of the cervix and corpus uteri, and all histological types of brain tumors (malignant, benign and undefined) were grouped together, in view of the difficulties in effecting reliable distinctions on death certification data alone; for liver cancer only deaths certified as primary liver cancer were included. Eight standard tables were produced, providing the following information: 1) average annual crude, age-standardized death certification rates and stand- ardized percentages of all cancer deaths for the population at all ages and truncated to 35-64 years (Table 1 for males and 2 for females). Two different standards were used: i) the 1971 Italian Census population, corrected for Census undercount and subdivided into the usual 16 quinquennia of age from 0-4 to 75-79 plus ::::, 80, and ii) the European standard population, for comparative purposes with other countries; 2) average annual age-specific death certification rates for each sex and quinquennium of age from 0-4 to 75-79 plus ::::, 80 (Table 3 for males and 4 for females); 3) total number of registered deaths in the period 1985-87 for each cancer or group of cancers, sex and age group (Table 5 for males and 6 for females); 4) average annual percentages of all cancer deaths for each sex and age group (Table 7 for males and 8 for females). ' Data per single calendar year are available on request from AD. The cancer sites or groups of cancers considered are listed in Tables 1 and 2 together with the corresponding ICD codes (9th Revision), whereas in subsequent tables only the ICDs are reported (and, consequently, Tables 1 and 2 can be used as legend for subsequent ones). This is clearly a statistical technical report. A short com- ment is included only to assist the reading and interpreta- tion of data for major cancer sites, and to recall underlying long-term tendencies. Comparisons were generally made with the previous 5-year calendar period, 1980-84 (6). Nonetheless, this can be easily ignored by the reader interested in the data only, or more specifically interested in considering details in mortality statistics for any particular cancer site, since any inference should be based on careful examination of the whole matrix of age-specific data. Discussion In the triennium 1985-87, the average number of cancer deaths per year was over 138,000, and reached 141,500 in 1987. Thus, the annual rise in total number of cancer deaths was over 2.5 % . This is partly due to an increased proportion of older subjects in the population, but the overall age-adjusted rate increased in males as well, from 246.6 in 1980-84 to 254.4 in 1985-87 (Italian standard). In females, the rise was much smaller, from 158.1 to 159.8. These upward trends were due to increases in the older age groups. In contrast, some decline was observed in both sexes in the truncated rates (from 281.9 to 277.9 in males, from 159.4 to 154.1 in females). Even more clear-cut was the pattern of age- specific trends, since all rates declined in males below age 55 and in females below 65, but increased at older ages. These patterns were appreciably influenced by lung cancer trends, which were still upward in older males (and, hence, in the overall rate, increased from 72.0 to 86.3) but not in middle ages (the truncated rate declined from 99.9 to 98.5). Inspection of age-specific rates shows that a major component of the apparently favorable trends is interpretable in terms of cohort effect, since lung cancer mortality showed an appreciably downward trend in the age group 45 to 54. This was the cohort born around Acknowledgments: This work was conducted within the framework of the National Research Council (C.N.R.) Applied Projects «Oncology» (Contract No. 87.01544.44) and «Risk Factors for Disease,., and with the contributions of the Italian Association for Cancer Research and the Italian League against Tumors, Milan. The authors wish to thank Mrs. J. Baggott, Mrs. M.P. Bonifacino, and the G.A. Pfeiffer Memorial Library for editorial assistance. To whom correspondence should be addressed: Prof. Adriano Decarli, Istituto di Biometria e Statistica Medica, Universita di Milano, lstituto Nazionale dei Tumori, via Venezian 1, 20133 Milano, Italy. Received June 20, 1990. 1