1405PA PROGNOSTIC MODEL OF SURVIVAL AFTER FIRST HOSPITAL DISCHARGE OF CANCER PATIENTS

ABSTRACT Aim: Unplanned hospitalization of cancer patients has been associated to low survival. Although prognostic tools exist for palliative patients, no tools are available to prognostically stratify patients after discharge from a Medical Oncology unit, probably resulting in missed opportunities for palliative care. Methods: We collected data from admissions and survival of patients admitted to a Medical Oncology department between March-2011 and March-2013. Cases were defined as patients discharged after first unplanned admission. Sample was randomly split in a derivation and a validation set (2:1). A multivariate Cox model for overall survival was built and a clinical score derived from the model was tested on the derivation and validation sets Results: We collected 1303 admissions corresponding to 745 single patients, of which 723 admissions were included. A multivariate Cox model for survival after discharge from first admission included the following covariates: Stage IV (HR: 2.8; 95%CI: 2-3.9), type of tumor (no breast or gynecological tumor; HR: 1.7; 95%CI: 1.2-2.3), no active treatment (HR: 1.9; 95%CI: 1.4-2.6) and admission for a reason other than toxicity (HR:1.9; 95%CI: 1.4-2.6). We integrated these factors in a prognostic score (from 0 to 4, with 1 point assigned to each factor), which was able to stratify patients according to survival (HR: 2.1, 95%CI: 1.8-2.3; p Survival outcomes by score in the derivation and validation sets Derivation set (n = 506) Validation set (n = 217) Score Median survival, weeks (n) Median survival, weeks (n) 0 Not reached (50) Not reached (20) 1 103,1 (71) 89,9 (20) 2 48,1 (139) 68,9 (62) 3 23,7 (191) 27,4 (99) 4 8 (55) 9 (16) Global 40 weeks 48 weeks Conclusions: A simple clinical score is useful for prognostic classification of patients discharged from a Medical Oncology unit. Patients with a high score show a low survival after discharge; consideration of prompt referral to palliative care services might lead to a better integration of care in this group Disclosure: All authors have declared no conflicts of interest.