Background Non-invasive ventilation (NIV) is an established treatment for patients with acute ventilatory failure. It can be successfully provided on a specialist ward, rather than intensive care (ICU) when certain criteria are met. It is frequently delivered outside ICU when a patient is deemed not suitable for invasive ventilation.
Methods Deaths in 2012 on our dedicated ventilation unit were analysed as part of ongoing clinical governance. Information on demographics, admission diagnosis, respiratory and metabolic acidosis, consolidation or pulmonary oedema on chest radiograph reports, Glasgow Coma Score (GCS), serum creatinine and hospital length of stay prior to NIV were recorded. Escalation of care and resuscitation decisions were noted.
Results There were 228 admissions for acute NIV, with 31 recorded deaths (13.6%), 22 case notes were available for review. Mean age was 79 years, 77.3% had known COPD, admission median MRC score of 4, and 18.2% had been in hospital for >7 days before NIV. All had acute hypercapnic respiratory failure. Not for resuscitation decisions had been made for 95.5% prior to NIV, and 100% had NIV as a ‘ceiling of care’. Mean pH was 7.25 (SD 0.06), similar to previous reports of admissions to our unit1, 22% had mixed acidosis (BE <-2.0 mmol/l). GCS was <8 in 9% and 36.4% had serum creatinine >100 µmol/l, all triggering alerts for acute kidney injury. Admission diagnoses are shown in figure 1. Radiographic consolidation was reported in 59.1% and pulmonary oedema in 18.2%.
Conclusion The mortality of patients receiving acute NIV is low2. Most deaths had an underlying diagnosis of COPD, they were an elderly frail group, deemed inappropriate for escalation to critical care. There were multiple risk factors for NIV failure on initiation of therapy. Whilst a trial of NIV may have been appropriate based purely on blood gases, it was at high risk of failure and discussion about end of life care may have offered an alternative approach.
Chakrabarti et al. Thorax 2009;857–62.
Roberts et al. Thorax 2010;43–8.