Mortality prediction models in AECOPD
| Study | Predictors | AUROC | Calibration | EV |
|---|---|---|---|---|
| Roche et al,29 103 French EDs, n=794 (14.5% pneumonia). Outcome: in-hospital mortality (7.4%). | Age >70, MRC score at baseline, cyanosis, use of accessory muscles, paradoxical abdominal movement, asterixis, neurological impairment, oedema. | 0.79; IV 0.83 | – | Roche et al38 0.78 |
| Ruiz-González et al,30 Spanish single centre, n=160. Excluded if infiltrates on CXR. Outcome: composite death in hospital or <15 days of discharge; ICU transfer or developed acute heart failure (24%). | CRP ≥50 mg/L, current smoker, ≥2 comorbidities and confusion. | 0.80, (0.72 to 0.88) | – | – |
| Tabak et al (BAP-65),31 US multicentre study, n=88 074. Outcome: in-hospital mortality (2%). | Urea >8 mmol/L, acute mental status change, pulse ≥110/min, age >65. | 0.72 (0.70 to 0.74) IV bootstrap; EV 0.71 (0.70 to 0.73) | Plotted derivation vs validation risk groups | Shorr et al,39 0.79 (0.78 to 0.80); Hu et al,40 0.67 (0.59 to 0.74); Steer et al,32 0.68 |
| Asiimwe et al,33 UK single-centre study, derivation n=4986. Outcome: in-hospital mortality (15.5%). | Albumin, urea and arterial pCO2. | 0.73; IV 0.70 | – | – |
| Steer et al,32 (DECAF) 2 UK centres; n=920. Outcome: in hospital mortality (10.4%). | Extended MRC DECAF. | 0.86 (0.82 to 0.89) IV bootstrap 0.86 (0.82 to 0.89) | H-L test p=0.39 | Echevarria et al,41 0.82 (0.77 to 0.87) |
| Tabak et al,34 derivation 172 US hospitals n=69 299 (2005–2006), validation (2007) n=33 327. Outcome: mortality (3.2% derivation vs 2.9% validation). | Age, urea, pH, albumin, WBC, pCO2, Troponin I or CK-MB, CPK, pO2, Na, Hb, PT time or INR, bands, platelets, pulse, temperature, RR, BP, altered mental state, CCF, pulmonary circulation disease, weight loss, metastatic cancer, malignancy. | 0.83 (0.82 to 0.84) IV 0.84 (0.83 to 0.85) | Calibration plotted; H-L test derivation p<0.0001, validation p<0.01 | – |
| Stiell et al,35 6 Canadian EDs, n=844. Outcome: serious adverse event* (7.8%). | PVD, CABG, prior intubation, ischaemic ECG, pulmonary congestion on CXR, unable to walk in ED, HR ≥110/min, Hb <100 g/L, urea ≥12 mmol/L, serum CO2 ≥35 mmol/L. | 0.80 (0.74 to 0.85); IV bootstrap 0.79 (0.79 to 0.80) | H-L test p=0.7 | – |
| Quintana et al,36 Spain, n=2487. Outcome: mortality in-hospital or <7 days of discharge (2.4%). | Age, baseline dyspnoea, previous LTOT or NIV, altered mental status, use of accessory muscles or paradoxical breathing on arrival. | 0.85 (0.77 to 0.93) | H-L test p>0.62 | – |
*Death <30 or <14 days and admission to monitored unit, MV or NIV, MI, CABG, PCI, new haemodialysis or, if discharged, readmission <14 days.
AECOPD, acute exacerbation of COPD; AUROC, area under the receiver operating characteristic curve (95% CI, where reported); Bands, immature bands on full blood count; BP, blood pressure; CABG, coronary artery bypass grafting; CCF, congestive cardiac failure; CK-MB, creatine kinase MB isoenzyme; CPK, creatine kinase; CRP, C reactive protein; CXR, chest X-ray; DECAF, Dyspnoea, Eosinopenia, Consolidation, Acidaemia and Atrial Fibrillation; ED, emergency department; EV, external validation; Hb, haemoglobin; H-L test, Hosmer-Lemeshow test (of calibration); ICU, intensive care unit; INR, international normalised ratio; IV, internal validation; LTOT, long-term oxygen therapy; MI, myocardial infarction; MRC, Medical Research Council; MV, mechanical ventilation; Na, Sodium; NIV, non-invasive ventilation; PCI, percutaneous coronary intervention; PT time, prothrombin time; RR, respiratory rate; PVD, peripheral vascular disease; WBC, white blood cells.