Birth cohort, time, and age effects in Italian cancer mortality

A. Decarli, C. La Vecchia, G. Mezzanotte, C. Cislaghi

Research output: Contribution to journalArticlepeer-review


Italian death certification data from 1955 to 1979 for total cancer mortality and 30 cancer sites in the population aged 25 to 74, were analyzed using a log-linear Poisson model to isolate the effects of birth cohort, calendar period of death, and age. The most frequent cohort pattern was characterized by increases up to the generations born between 1920 and 1930, followed by stabilization or a slight decrease. This pattern was evident for total cancer mortality in men, and for several common sites, including larynx, lung, esophagus, bladder, female breast, and ovary. Only four sites (pancreas, pleura, intestines in both sexes, and kidney in men) showed cohort values still rising in more recent generations. Stable cohort and period of death curves were observed for cancers of the prostate and testis, whereas trends were steadily going down for neoplasms of the stomach, and (cervix) uteri. Finally, there were a few discontinuous trends (e.g., in the case of brain neoplasms, leukemias, and lymphomas), which probably reflect different effects of improvements in diagnosis and/or treatment. Period of death values increased for lung and other tobacco related sites (chiefly in males) and, up to the early 1970s, for a few other common sites, including intestines, and the female breast. Downward trends over the calendar period were evident for cancers of the stomach and of the (cervix) uteri. Therefore, total cancer mortality trends over the calendar period of death were moderately increasing for men, and slightly decreasing for women.

Original languageEnglish
Pages (from-to)1221-1232
Number of pages12
Issue number6
Publication statusPublished - 1987

ASJC Scopus subject areas

  • Cancer Research
  • Oncology


Dive into the research topics of 'Birth cohort, time, and age effects in Italian cancer mortality'. Together they form a unique fingerprint.

Cite this