COAST data analysis

Oxygen supplementation is a recommended treatment for children with severe pneumonia or hypoxaemia. The open, fractional-factorial Children’s Oxygen Administration Strategies Trial (COAST) recruited Kenyan and Ugandan children with severe pneumonia and hypoxaemia. Participants in the severe hypoxaemia stratum (SpO 2 < 80%) were randomised to high-flow nasal therapy (HFNT) or low-flow oxygen (LFO), and in the hypoxaemia stratum (SpO 2 80–91%) to HFNT, LFO or permissive hypoxaemia (ratio 1:1:2). The trial stopped early and there is ongoing uncertainty about the clinical benefits of the alternative strategies. There is a lack of evidence about the relative costs, of alternative oxygen delivery for critically-ill children in low- and middle- income countries. We used data from COAST to conduct a cost-consequence analysis of the treatment strategies. We measured resource use for 28 days post-randomisation (n = 1,842). Resources included oxygen delivery, medications, blood and fluid products, diagnostic tests, point of care tests, hospital admission and length of stay. We calculated the total costs and reported the incremental costs as the difference in the mean total costs between groups, adjusting for baseline differences. In the severe hypoxaemia stratum, the mean total cost was $393.04 for HFNT and $218.73 for LFO. In the hypoxemia stratum, the mean total costs were $391.95 (HFNT), $198.26 (LFO) and $167.80 (permissive). The adjusted cost difference between HFNT versus LFO and liberal versus permissive was $184.43 (95% CI l: $127.90, $240.95), and $124.01 (95% CI: $99.53, $148.49), respectively. The differences of HFNT and LFO versus permissive were $216.22 (95% CI: $160.77, $271.68) and $31.80 (95% CI: $11.49, $52.11), respectively. For children with severe hypoxaemia, HFNT is more costly than LFO. For children with hypoxaemia, either of HFNT or LFO were more costly than permissive hypoxaemia. The main driver of costs for HFNT is the high cost of equipment and consumables; other costs were similar across treatment groups in both strata, as were health outcomes.

Keywords