S24 Predicting poor outcome at six months following exacerbations of COPD requiring assisted ventilation

Introduction Non-invasive ventilation (NIV) is life-saving in exacerbations of COPD complicated by acute hypercapnic respiratory failure (AHRF). These episodes indicate advanced disease and patients who survive the acute episode show significant morbidity and high one-year mortality.1 In this group, clinicians can be prognostically pessimistic about outcome and quality of life (QoL); understanding predictors of poor outcome six months following an episode of AHRF treated with NIV could help clinical decision making. Methods Unique, consecutive patients were prospectively recruited to the NIV Outcomes study (ISRCTN22921168) in 10 NHS trusts. Consenting patients surviving to discharge were followed up monthly with QoL assessed using validated QoL questionnaires including the COPD Assessment Test (CAT). Poor outcome was defined as a pre-discharge CAT score in the worst 50% of responses (CAT=24–40) with subsequent clinically significant decline (≥2 points), or death, within six months. Multivariate logistic regression identified independent predictors. Continuous variables were subsequently dichotomised, and regression rerun. Results 553 patients survived to discharge; 253 consented and 239 provided >1 QoL assessment. Median(IQR) follow up time was 359(171–367) days, with 8(3–11) months of questionnaires completed, and NIVO score of 3(1–3.75). Mean(SD) age 68.9(9.1) years and FEV136.8(14.7)% predicted. 34.2% had previously received NIV, 24.6% were prescribed LTOT, and 47.1% were unable to leave the house unassisted. 56.6% had persistent hypercapnia post-ventilation. 67/239 suffered poor outcome. Independent predictors of poor outcome were LTOT, Left Ventricular Systolic Dysfunction, Diaphragm height ≤2.3 cm, confusion pre-ventilation, admission haemoglobin ≤14 g/dL, Pre-discharge HADS-depression Score ≥8, Pre-discharge Nottingham Extended Activities of Daily Living score ≤30 (Table 1). A simple tool (1 point assigned to each variable) demonstrated an area under the receiver operating characteristic curve of 0.809 (95% CI’s: 0.747–0.871). Discussion These routinely available clinical indices, which include a measure of hyperinflation, and two simple patient completed questionnaires demonstrate independent prediction of poor outcome in this population. A clinical tool from these indices shows promise but would require external validation before clinical use. If this occurred, it may help challenge prognostic pessimism and with predicting patients who require specialist palliative input. Reference BTS/ICS AHRF Guideline. Thorax(2016);71:ii1–ii35.