Predictors of 1-year mortality in patients discharged from hospital following acute exacerbation of chronic obstructive pulmonary disease

Introduction: acute exacerbation of COPD (AECOPD) is a major cause of hospital admission, and predicts subsequent medium-term mortality. We aimed to examine mortality predictors in patients discharged from hospital after AECOPD. Methods: we obtained baseline demographic and clinical data from 100 patients (mean age (range) = 73 (60-98) years; 48 males) admitted with AECOPD. All completed the following validated questionnaires: a quality of life questionnaire (Breathing Problems Questionnaire; BPQ); a screening questionnaire for depression (Brief Assessment Schedule Depression Cards; BASDEC); a disability questionnaire (Manchester Respiratory Activities of Daily Living questionnaire; MRADL). Following discharge all were prospectively followed and survival/mortahty at 12 months confirmed from hospital notes and by contacting general practitioners. Results: the prevalence of depression at recruitment was 56%. One-year mortality in the whole group was 36%. Odds ratios (95% confidence intervals) for mortality predictors (univariate logistic regression analysis) were: use of long-term oxygen therapy = 2.72 (1.06-6.97); subsequent readmission = 2.57 (1.08-6.12); MRADL score = 0.87 (0.80-0.94) (disability predicting death); BASDEC score = 1.13 (1.02-1.26) (depression predicting death); BPQ score = 1.08 (1.04-1.12) (low quality of fife predicting death); length of original hospital stay = 1.03 (1.00-1.07). On multivariate logistic regression analysis the only mortality predictor was BPQ with an odds ratio (95% confidence limits) of 1.13 (1.04-1.22). In terms of mortality prediction for individuals, a threshold MRADL score of < 12 gave a sensitivity of 86%, specificity of 55%, positive predictive value of 88% and negative predictive value of 52%, with similar predictive values using BPQ as an independent variable. Conclusions: 1-year mortality after AECOPD admission is high. The presence of depressive illness (which is extremely common), and levels of both disability and impairment of quality of fife are univariate predictors of 1-year mortality in this patient group. This model may be useful in predicting prognosis for individuals and thus in guiding treatment decisions.