Bubble continuous positive airway pressure for children with severe pneumonia and hypoxaemia in Bangladesh: an open, randomised controlled trial

Background: In developing countries, mortality in children with very severe pneumonia is high, even with the provision of appropriate antibiotics, standard oxygen therapy, and other supportive care. We assessed whether oxygen therapy delivered by bubble continuous positive airway pressure (CPAP) improved outcomes compared with standard low-flow and high-flow oxygen therapies.

Methods: This open, randomised, controlled trial took place in Dhaka Hospital of the International Centre for Diarrhoeal Disease Research, Bangladesh. We randomly assigned children younger than 5 years with severe pneumonia and hypoxaemia to receive oxygen therapy by either bubble CPAP (5 L/min starting at a CPAP level of 5 cm H2O), standard low-flow nasal cannula (2 L/min), or high-flow nasal cannula (2 L/kg per min up to the maximum of 12 L/min). Randomisation was done with use of the permuted block methods (block size of 15 patients) and Fisher and Yates tables of random permutations. The primary outcome was treatment failure (ie, clinical failure, intubation and mechanical ventilation, death, or termination of hospital stay against medical advice) after more than 1 h of treatment. Primary and safety analyses were by intention to treat. We did two interim analyses and stopped the trial after the second interim analysis on Aug 3, 2013, as directed by the data safety and monitoring board. This trial is registered at ClinicalTrials.gov, number NCT01396759.

Findings: Between Aug 4, 2011, and July 17, 2013, 225 eligible children were recruited. We randomly allocated 79 (35%) children to receive oxygen therapy by bubble CPAP, 67 (30%) to low-flow oxygen therapy, and 79 (35%) to high-flow oxygen therapy. Treatment failed for 31 (14%) children, of whom five (6%) had received bubble CPAP, 16 (24%) had received low-flow oxygen therapy, and ten (13%) had received high-flow oxygen therapy. Significantly fewer children in the bubble CPAP group had treatment failure than in the low-flow oxygen therapy group (relative risk [RR] 0·27, 99·7% CI 0·07-0·99; p=0·0026). No difference in treatment failure was noted between patients in the bubble CPAP and those in the high-flow oxygen therapy group (RR 0·50, 99·7% 0·11-2·29; p=0·175). 23 (10%) children died. Three (4%) children died in the bubble CPAP group, ten (15%) children died in the low-flow oxygen therapy group, and ten (13%) children died in the high-flow oxygen therapy group. Children who received oxygen by bubble CPAP had significantly lower rates of death than the children who received oxygen by low-flow oxygen therapy (RR 0·25, 95% CI 0·07-0·89; p=0·022).

Interpretation: Oxygen therapy delivered by bubble CPAP improved outcomes in Bangladeshi children with very severe pneumonia and hypoxaemia compared with standard low-flow oxygen therapy. Use of bubble CPAP oxygen therapy could have a large effect in hospitals in developing countries where the only respiratory support for severe childhood pneumonia and hypoxaemia is low-flow oxygen therapy. The trial was stopped early because of higher mortality in the low-flow oxygen group than in the bubble CPAP group, and we acknowledge that the early cessation of the trial reduces the certainty of the findings. Further research is needed to test the feasibility of scaling up bubble CPAP in district hospitals and to improve bubble CPAP delivery technology.

Funding: International Centre for Diarrhoeal Disease Research, Bangladesh, and Centre for International Child Health, University of Melbourne.