The growing incidence of cancer associated to an aging population represents an epidemiologic reality that requires questioning access to care and prognosis in elderly with cancer, for which disparities have been highlighted. However, generally speaking, studies are limited in that they overlook geriatric-specific factors. The aim of this work was to study sociodemographic, socioeconomic and clinical determinants of access to care (cancer stage, cancer treatment) and prognosis (functional decline, survival) in elderly cancer patients.

The French department of Gironde (1.5 million inhabitants) is covered by cancer registries. In addition, three cohorts were initiated in Gironde on cerebral and functional aging that enrolled subjects aged 65 years and over: the PAQUID study (1988), the 3-City study (1999), and the AMI study (2007). In these cohorts, a large amount of individual data on the participants has been collected longitudinally at different follow-up visits, conducted every two or three years.

Using cancer registries, older subjects with cancer were identified in the cohort studies. Subjects were included if they presented the following: i) aged 65 years and over from one of the three cohorts, ii) alive on January 1st 2005 (common start date for tumor recording) and resident in Gironde, and iii) with a validated cancer diagnosis recorded in one of the cancer registries from January 1st 2005 to December 31st 2014. All tumors for which data is annually sent by cancer registries to the International Agency for Research of Cancer were included: invasive malignant tumors and benign tumors of the central nervous system. As the incidence of skin tumors in older cancer patients is high, they were also included. For patients with multiple tumors, only one was considered: either the first one diagnosed if there were several tumors with a minimum of a 6 months interval between their diagnosis, or the one with the worse prognosis if there were several tumors diagnosed within 6 months.

In order to not consider information concerning patients' characteristics that was too old, subjects with a delay between the last completed cohort follow-up visit before the cancer diagnosis and the diagnosis of cancer equal or superior to 6 years were excluded.

Variables were extracted from cohorts and registries at different times: cohort inclusion, pre-diagnosis visit, cancer diagnosis or registries follow-up.

• Age ≥ 65 years
• Alive on January 1st 2005 ( PAQUID or 3-City cohorts) or included in the AMI cohort study
• Cancer diagnosis recorded in one of the cancer registries in Gironde, French department

Subjects aged 65 and over with a diagnosis of cancer recorded in one of the cancer registries in Gironde, French department.

One year Serious and Other (Not Including Serious) Adverse Events were not monitored/assessed

Serious and Other (Not Including Serious) Adverse Events were not monitored/assessed