br Material and methods br Results
Material and methods
Authorship contribution statement
Disclosure of potential conflicts of interest
Financial support This research did not receive any specific grant from funding agencies in the public, commercial, or not-for-profit sectors. The work of the authors was supported by core funds of their respective institutions, namely the International Agency for Research on Cancer, Danish Cancer Society Research Centre, Georgetown University and Universidad Nacional, Costa Rica respectively.
Acknowledgements The authors are grateful to O. M. Araya, M. Bolaños, N. Gómez, M. G. Mora-Mora, and A. Corrales for their help with the calculations of mid-year population estimates specifically for children younger than 1 year of age. The authors express their special appreciation to L. Córdoba for the preparation of the map of Costa Rica with the MIDEPLAN regions (Fig. 1).
Introduction Prostate cancer is the second most common cancer in men worldwide and the leading type of cancer among men in more developed countries . In 2011, nearly 54,000 cases were diagnosed in France corresponding to a world standardized incidence rate of 97.7 per 100,000 person-year . An increase of incidence rates was observed until the 1990′s in North America  and until the 2000′s in European countries  in relation to the Thymoquinone of Prostate Specific Antigen (PSA) testing. In France, the world standardized incidence rates per 100,000 persons increased from 24.8 in 1980 to 127.1 in 2005, followed by a decrease to 99.4 in 2009 . Prostate cancer belongs to the good prognosis group of cancers. Data from French cancer registries for cases diagnosed during the 1989–1998 period showed 10-year net survival rates of 73% (95%CI: 71%–75%) and 71% (95%CI: 69%–73%) for men aged 55–64 and 65–74, respectively [6,12,13,16]. Survival rates increased with calendar period in most countries, from 79.4% at 5 years for cases diagnosed in France during the 1995–1999 period to 90.5% during the 2005–2009 period [7,8]. Incidence trends and survival rates could differ by age groups considering potential differences in PSA testing, types of cancers [9,10] and medical management . For example, prostate cancers among elderly patients included more high-grade forms categorized as Gleason 8–10 and 48% of cases had metastases at diagnosis , involving many deaths even with aggressive therapy [12,13]. The description of the epidemiology of prostate cancer among men aged 75 and over is of particular interest considering the importance of prostate cancer among this age group. Indeed, incidence rates of prostate cancer increase with age  and the proportion of individuals aged over 75 years among men diagnosed with prostate cancer has rapidly increased due to the increase of the life expectancy. Twenty nine percent of cases were diagnosed after 75 years in 2009 in France whereas 78% of prostate cancer deaths occurred among men aged over 75 years . Although older-people do not lose as many years of life than younger patients due to their shorter remaining life expectancy, patients aged over 75 years still lose three quarters of their life expectancy . The increase of the number of elderly patients contributed to the development of geriatric oncology  whose purpose is to improve the management of patients aged over 75 years  to minimize toxicity and maximize the effectiveness of different treatment options taking into account their specificity [, , , , , ]. This approach could result in changing survival rates .
Material and methods
Theory/calculation Annual standardized incidence and mortality rates were computed for each calendar year from 1991 to 2013, for the 2 age groups considered using a truncated world standard population aged 60–74 and aged over 75. Incidence rates were also computed for the 3 different grades (low, intermediate and high). We then applied the Joinpoint regression model  to identify breakpoints in the trend of age-standardized rates and to estimate average rates of change. Basically, the Joinpoint model finds the best-fit line through several years of data using algorithm that tests whether a multi-segmented line is a significantly better fit then a straight or less-segmented line. The program starts with the minimum number of joinpoint (0 joinpoint, which is a straight line) and tests whether more joinpoints are statistically significant and must be added to the model.