Derivation and validation of a novel scoring tool to predict inpatient mortality in exacerbations of chronic obstructive pulmonary disease requiring assisted ventilation

Abstract

Background: Exacerbations of Chronic Obstructive Pulmonary Disease (COPD) are common and account for approximately 12% of UK hospital admissions. A significant proportion will be complicated by respiratory acidaemia which has a high mortality. Non-Invasive ventilation (NIV) confers a 2-3 fold reduction in mortality in this setting. Despite this, practice is suboptimal; the intervention is underused, infrastructure is lacking, and complex decisions are made by a wide range of clinicians. It is feasible that pessimism contributes. Aims: To derive and separately validate a simple, bedside, clinical tool to predict in-hospital mortality in exacerbations of COPD complicated by respiratory acidaemia requiring assisted ventilation. Methods: The study was split into two parts with similar methods. The derivation study was a single trust (2 sites: one urban and one rural) retrospective study. In patients meeting selection criteria, data were collected, and multivariable regression analysis identified independent predictors of in-hospital death. A simple predictive model was created. The validation study captured a more limited dataset in prospectively recruited patients across 10 trusts. The predictive model’s performance was assessed. Results: 489 patients were identified in the derivation study and 733 in the validation. Independent predictors of outcome were confirmed, and a final, simple bedside model entitled the NIVO score produced. Using atrial fibrillation, chest X-ray consolidation, eMRCD score, Glasgow coma scale, timing of acidaemia relative to admission time and pH in a simple scoring system stratified risk was obtained with an area under the receiver operated curve of 0.79 in the validation cohort. Discussion: Using only simple, readily available indices good prediction of in-hospital mortality is feasible. The NIVO score outperformed pre-identified comparator scores in both its derivation and validation studies. Potential practical applications include but are not limited to guiding level of care, setting treatment limitations and objectifying discussion with patients or family members