Background Despite the often poor outcome of patients hospitalised with acute exacerbations of COPD (AECOPD), it is difficult accurately to identify those at high risk of mortality. To aid prognostication in AECOPD, we have developed a simple, easily memorable and effective tool, based on clinical data available shortly after admission.
Methods Consecutive patients hospitalised with AECOPD were recruited, with clinical and demographic data collected at admission. In-hospital mortality data were collected from hospital records. Variables were dichotomised and the strongest independent predictors of mortality were identified by logistic regression analysis. Tool performance was assessed using ROC curve analysis.
Results 920 patients were recruited: mean (SD) age was 73.1 (10.0) years, with 53.9% female; most had severe airflow obstruction (FEV1 43.6 (17.2) % predicted) and were of normal weight (BMI 24.6 (6.3) kg/m2). 32.5% of patients had coexistent consolidation and 199 (21.6%) received assisted ventilation during their hospital stay. 96 (10.4%) patients died in-hospital. In descending order of strength, the factors independently predicting mortality were: the extended MRC Dyspnoea Scale (eMRCD)1; coexistent radiographic consolidation; eosinopenia (<0.05×109/l); pH <7.3; atrial fibrillation; cough effectiveness; albumin <36 g/dl; age=80 years; cerebrovascular disease; and BMI <18.5 kg/m2. The strongest five variables were selected to form the DECAF (Dyspnoea, Eosinopenia, Consolidation, Acidaemia, atrial Fibrillation) score (Abstract P212 table 1). Each predictor was assigned a score of 1 (present) or 0 (absent), except for eMRCD score which could be 0, 1 or 2, giving a maximum DECAF score of 6. The DECAF score showed good performance for the prediction of in-hospital mortality (area under ROC curve=0.858, 95% CI 0.82 to 0.89), and was a stronger predictor (p<0.0001) than either the APACHE (AUROC=0.727) or CAPS (AUROC=0.710) prognostic scores. In patients with coexistent consolidation (n=299), DECAF was a stronger predictor of mortality than CURB-65 (AUROC=0.77 vs 0.66, p=0.0064).
Conclusion The DECAF score is a strong predictor of in-hospital mortality and may improve the prognostication of patients hospitalised with AECOPD. External validation is required before recommending widespread application.