Introduction Predicting which patients are likely to benefit from assisted ventilation in exacerbations of COPD (ECOPD) is difficult. Existing prognostic scores did not assess the prognostic value of timing of acidaemia onset relative to admission during development. The 2011 National Chronic Obstructive Pulmonary Disease Resources and Outcomes Project (NCROP) report identified this as a potentially important marker of non-invasive ventilation (NIV) success but further study has been limited. We investigated the relation of timing of respiratory acidaemia to NIV outcomes in ECOPD. Of importance, in both participating hospitals few patients are denied NIV on grounds of assumed futility.

Methods A retrospective cohort of consecutive, unique patients, hospitalised with a primary diagnosis of ECOPD were identified from known cohorts, hospital coding records coding and ventilation service records. Other selection criteria included: age 35+years, smoking history 10+pack years, airflow obstruction on spirometry; received assisted ventilation (either NIV or invasive) for acidaemic respiratory failure; and absence of comorbidity expected to limit survival to <12 months (principally metastatic cancer). Population descriptors, ventilation details and outcome data were collected from notes and electronic records.

Results 489 consecutive patients were identified between 30/11/08 and 19/5/13; 124 (25.4%) died in-hospital. Median time to ABG prompting ventilation was 2 hour 42 m (IQR 1 hour 2 m – 15 hour 28 m). Most (94.5%) received NIV alone, 5.5% received invasive ventilation (+/-NIV). In patients requiring assisted ventilation within 12 hours of admission, mortality was 18.3% (65/356), compared to 44.3% (59/133) in all those treated after 12 hours (p<0.01).

Discussion Our study has several strengths including objective confirmation of COPD, capture of consecutive patients and liberal NIV use. Compared to patients with respiratory acideamia on or shortly after admission, later development was associated with progressively higher mortality. 12 and 48 hours were identified as clinically useful thresholds. Of note, lower pH, FEV1 and prior LTOT prescription do not account for worse outcome. Older age, greater comorbidity, frailty (eMRCD5b: requiring help washing and dressing when recently stable), and a strong trend towards increasing pneumonia are associated with later development of acidaemia. Timing of acidaemia should be considered when deciding whether to initiate NIV.