RT Journal Article SR Electronic T1 Risk factors of poor outcomes after admission for a COPD exacerbations. Multivariate logistic predictive models JF European Respiratory Journal JO Eur Respir J FD European Respiratory Society SP PA403 DO 10.1183/13993003.congress-2015.PA403 VO 46 IS suppl 59 A1 Juan Luis Garcia-Rivero A1 Cristina Esquinas A1 Miriam Barrecheguren A1 Patricia Garcia-Sidro A1 Elsa Naval A1 Carlos Martinez A1 Rosa Malo de Molina A1 P.J. Marcos A1 J.M. Diez A1 Alberto Herrejón A1 J.A. Ros A1 Sagrario Mayoralas A1 Manuel Valle A1 Marc Miravitlles YR 2015 UL http://erj.ersjournals.com/content/46/suppl_59/PA403.abstract AB Banckground: The aim of the study was to identify an age-adjusted multivariate model to predict failure after admission for a COPD exacerbation.Methods: Multicenter, observational and prospective study. COPD admitted patients were followed during 3 months. Relevant clinical variables at admission were selected. For each variable, the best cut-off for a new exacerbation were indentified using receiver operating characteristic (ROC) curves, Finally, a stepwise logistic regression model were performed.Results: A total of 106 patients were included. Mean age 71.1 (9.8), mean FEV1(%): 45.2%. Mean CAT at admission 24.8 (7.1). At 3 months 39 (36.8%) patients presented failure: death (2.8%), readmission (20.8%) or new ambulatory exacerbation (13.2%). Variables included in logistic model were: previous hospital admission, FEV1<45%, Charlson≥3, hemoglobin (Hb) <13 mg/dL, PCO2≥46 mmHg, Fibrinogen≥554 mg/dL, CRP≥45g/L, leukocytes<9810 x109/L, presence of sputum purulence and CAT at admission≥31. Final model showed that Hg<13 mg/dL(OR=2,86, CI95% 1,09-7,49) and CRP≥45g/L (OR=2,80, CI95%=1,05-7,46) increased the probability for failure up to 62%. After adding the presence of CAT ≥31 at admission, the probability increased to 84% (AUC=0.785(p=0.001))Conclusions: Up to 38.8% of patients have a poor outcomes at 3 months after admission. Low Hb and high CRP are risk factors for failure. High CAT score at admission increase the predictive value of the model.