WHICH RISK FACTORS COULD HAVE AFFECTED BREAST CANCER INCIDENCE IN SLOVENIA IN THE PAST, AND WHAT ARE THE PREDICTIONS FOR THIS DECADE

• Main Authors: Vera Pompe Kirin, Barbara Japelj, Mojca Primic Žakelj, Milivoja Šircelj
• Format: Article
• Language: English
• Published: Slovenian Medical Association 2001-06-01
• Series: Zdravniški Vestnik
• Subjects: female breast cancer; trends in incidence; prediction; APC models; Sloveni
• Online Access: http://vestnik.szd.si/index.php/ZdravVest/article/view/2569
• ISSN: 1318-0347; 1581-0224

<p><strong>Background</strong>. Since 1950, the incidence in female breast cancer has been increasing in Slovenia. The curve of annual incidence rates shows an uneven increase. So, the calculation of linear trend and any prediction depend only on the chosen period of observation. In this paper, the analysis of female breast cancer incidence in Slovenia for the period 1964–1998, and the predictions till the years 2003 and 2008 are presented.<br /><strong>Methods</strong>. First, the basic APC (age-period-cohort) model was applied. Because of an unexpectedly high number of new cases predicted by this model for the next decade, we applied three modified models taking into account some known risk factors: the published fertility rates, the average number of children and the age at first birth in individual birth cohorts of the 20th century. We used the GLIM program.<br /><strong>Results</strong>. As risk factors have a different impact on patients diagnosed with breast cancer before and after menopause, we divided the patients by the age at the diagnosis into the following three groups: 25–44 years, 45–54 years and 55 years and more. According to this model, the number of new breast cancer cases will not increase in the first half of this decade, while in the second half, it will be close to the number registered in the years 1990–1994.<br /><strong>Conclusions</strong>. According to our estimates, the third modified model gave us the most appropriate picture of the impact of known risk factors on different generations in the past. We therefore assume that the prediction of incidence for the future five years according to this model is the most reliable, though at a first glance, too optimistic.</p>