Prediction of in-hospital mortality in exacerbations of chronic obstructive pulmonary disease requiring assisted ventilation, and patient-centred outcomes over the subsequent year

Abstract

Background: COPD exacerbations account for over 141,000 hospitalisations annually in the UK, and 20% are complicated by respiratory acidaemia. In such cases non-invasive ventilation (NIV) substantially reduces mortality, but only 50% of patients receive NIV and delivery is often suboptimal. Prognostic pessimism may partially explain underuse. Little is known about the quality of life (QoL) of patients following ventilation. Aims and Methods: The NIV Outcomes (NIVO) study aimed to validate a simple prognostic tool for inpatient mortality in this population and assess patient-centred outcomes over one year among those surviving to discharge. Prospective recruitment of patients occurred in 10 trusts. Model performance was assessed. QoL and mortality were tracked for one year following discharge. Results: 733 patients were included. The NIVO score comprised eMRCD, acidaemia occurring >12 hours after admission, atrial fibrillation, chest x-ray consolidation, Glasgow coma scale ≤14 and pH <7.25. Stepwise increase in inpatient mortality was observed, with a c-statistic of 0.79. The NIVO score significantly outperformed preidentified comparator scores. 586 patients survived to discharge; QoL was assessed in 272 patients over one year. Symptom burden was substantial. QoL peaked 2-months after discharge and overall was well maintained subsequently. Most patients would accept NIV again. Poor outcome, defined as death within six months or poor QoL with a clinically significant deterioration, could be accurately predicted (c-statistic 0.81) using: LTOT, admission haemoglobin, left ventricular systolic dysfunction, diaphragm height, confusion, plus measures of depression and independence. Discussion: Using six simple indices strong prediction of inpatient mortality was achieved. In most survivors, QoL remained stable over one year post-discharge. Predictors of patients likely to have poor outcome were identified. Clinical applications include facilitating informed discussions with patients and shared decisions about escalation to ventilation, challenging pessimism, comparing risk-adjusted outcomes between units (national audit), advance care planning and referral to specialist services.