In Kenyan and Ugandan children with severe pneumonia and low blood oxygen, higher-tech oxygen delivery increased costs without better outcomes.

In the COAST trial, high-flow nasal therapy cost more than low-flow oxygen for children with severe hypoxaemia. For children with less severe hypoxaemia, both high-flow nasal therapy and low-flow oxygen cost more than permissive hypoxaemia. Across groups, health outcomes were similar, and the extra cost of high-flow nasal therapy was mainly driven by equipment and consumables.

Quick summary

• What the study found: High-flow nasal therapy had higher mean total costs than low-flow oxygen in severe hypoxaemia, and higher costs than permissive hypoxaemia in hypoxaemia; outcomes were similar across groups.
• Why it matters: When budgets are tight, choosing the oxygen strategy that delivers similar outcomes at lower cost can free resources for staffing, medications, and essential equipment.
• What to be careful about: The trial stopped early, and the authors note ongoing uncertainty about clinical benefits; these results are about costs and resource use over 28 days, not long-term effects.

What was found

The journal article [A cost-consequence analysis of the children’s administration oxygenation strategies trial (COAST) in severe pneumonia] examined the relative costs of different oxygen delivery strategies for children with severe pneumonia and low blood oxygen in Kenya and Uganda. The researchers used data from the Children’s Oxygen Administration Strategies Trial (COAST), which recruited children with severe pneumonia and hypoxaemia (low blood oxygen).

Children were grouped (“stratified”) by baseline oxygen saturation measured by peripheral capillary oxygen saturation (SpO2, an estimate of blood oxygen from a fingertip or toe sensor). In the severe hypoxaemia group (SpO2 less than 80%), children were randomised to high-flow nasal therapy (HFNT) or low-flow oxygen (LFO). In the hypoxaemia group (SpO2 80–91%), children were randomised to HFNT, LFO, or permissive hypoxaemia (a strategy that accepts lower oxygen levels instead of giving supplemental oxygen right away).

The analysis tracked resource use for 28 days after randomisation in 1,842 children. Resources included oxygen delivery, medications, blood and fluid products, diagnostic tests, point of care tests (tests performed near the patient rather than in a central lab), hospital admission, and length of stay.

Costs differed clearly by strategy. In the severe hypoxaemia group, the mean total cost was $393.04 for HFNT and $218.73 for LFO. In the hypoxaemia group, mean total costs were $391.95 for HFNT, $198.26 for LFO, and $167.80 for permissive hypoxaemia.

The paper also reported adjusted cost differences (differences in mean costs between groups after adjusting for baseline differences). The adjusted cost difference between HFNT versus LFO and liberal versus permissive was $184.43 (95% confidence interval: $127.90 to $240.95) and $124.01 (95% confidence interval: $99.53 to $148.49), respectively. HFNT and LFO versus permissive hypoxaemia differed by $216.22 (95% confidence interval: $160.77 to $271.68) and $31.80 (95% confidence interval: $11.49 to $52.11), respectively.

Importantly, other costs were similar across treatment groups in both strata, and health outcomes were similar across groups. The authors identify the main driver of HFNT costs as the high cost of equipment and consumables.

What it means

This is a straightforward economic message: in this dataset, HFNT increased costs compared with simpler oxygen approaches, without differences in health outcomes. When outcomes are similar, decision-makers often focus on cost and feasibility, because every dollar spent on a higher-cost option is a dollar not spent elsewhere.

The biggest practical implication is about what drives the bill. If HFNT cost is mainly due to equipment and consumables, then scaling HFNT across hospitals is not just a clinical decision; it is a supply-chain and maintenance decision. Equipment-dependent care can fail not only because of clinical complexity, but because of breakdowns, stock-outs, and inconsistent access to compatible consumables.

For hospitals and health systems, the similarity of “other costs” across groups matters too. It suggests the extra spending is not being offset by reductions in medications, tests, transfusions, fluids, or hospital bed-days within the 28-day window measured here. In other words, HFNT did not “pay for itself” through reductions in other categories of spending in this analysis.

For families and frontline clinicians, the take-home is not that HFNT is “bad,” but that it is “costly,” and cost needs to be justified when outcomes are comparable. In settings where severe pneumonia and hypoxaemia are common and budgets are constrained, cost-effective choices can translate into broader access to care.

Where it fits

This work addresses a gap the authors explicitly note: limited evidence on the relative costs of alternative oxygen delivery for critically ill children in low- and middle-income countries. Oxygen is recommended for children with severe pneumonia or hypoxaemia, but “how” oxygen is delivered can vary widely in complexity and cost.

The study uses a cost-consequence analysis, which means it presents costs alongside a range of outcomes rather than compressing everything into a single metric. For busy decision-makers, this can be more actionable: it shows what you spend and what you get, without requiring agreement on one “master” outcome value.

The clinical backdrop also matters: the COAST trial stopped early, and the paper notes ongoing uncertainty about the clinical benefits of the alternative strategies. That uncertainty makes cost evidence more—not less—important, because when benefits are unclear, high-cost approaches need especially strong justification before they become routine.

Finally, it is notable that the study spans two countries and includes real-world hospital resources: oxygen delivery, tests, medications, blood products, fluids, admissions, and length of stay. That breadth reduces the chance that results are driven by a single narrow cost category—except in the case of HFNT, where equipment and consumables clearly dominate the difference.

How to use it

If you are a hospital leader deciding what to buy, this paper supports a cautious approach to scaling HFNT when budgets are fixed and outcomes are similar. A practical interpretation is “prioritise reliable access to lower-cost oxygen delivery first,” especially if HFNT scale-up would reduce the budget available for other essentials.

If you are designing protocols, separate the clinical decision from the procurement reality. Protocols that rely on HFNT should be paired with explicit plans for device maintenance, staff training, and consistent consumable supply. Otherwise, HFNT can become an intermittent resource that disrupts care rather than improving it.

If you are commissioning services at a district or national level, use these findings to model opportunity costs. Opportunity cost is the benefit you give up by choosing one option over another; here, the extra dollars spent on HFNT could alternatively fund other high-impact care components. Even without claiming what those alternatives “would” achieve, the logic of trade-offs is unavoidable in constrained systems.

If you are a clinician advocating for better oxygen access, this study can support a “coverage-first” argument: build dependable oxygen systems that reach more children before investing heavily in higher-cost delivery modes. In many health systems, expanding basic capacity is the most immediate path to reducing avoidable deaths, even when advanced technology is appealing.

Limits & what we still don’t know

The biggest limitation is built into the study’s context: the COAST trial stopped early, and there is ongoing uncertainty about the clinical benefits of the alternative strategies. That means you should not treat cost differences as the final word on which strategy is “best” clinically.

The time horizon here is 28 days post-randomisation. Costs and outcomes beyond that window were not reported in the excerpt, so this analysis cannot answer questions about longer-term recovery, later complications, or downstream health service use.

The results also do not specify where, within each health system, the price of equipment and consumables comes from, how stable those prices are, or how costs might change with bulk purchasing, reuse policies, or local manufacturing. Those real-world factors can change the magnitude of cost differences, even if the direction remains similar.

Finally, “health outcomes were similar across treatment groups” is a key statement, but the excerpt does not list which outcomes were measured or how. Without those details, readers should avoid assuming the results apply equally to every possible patient subgroup or to every clinically meaningful outcome.

Closing takeaway

In this cost-consequence analysis using COAST data, HFNT was more expensive than LFO for children with severe hypoxaemia, and both HFNT and LFO were more expensive than permissive hypoxaemia for children with hypoxaemia. The extra spending on HFNT was largely driven by equipment and consumables, while other costs and health outcomes were similar across groups. For resource-limited settings, the clearest message is to demand strong, demonstrated clinical benefit before paying the premium for high-cost oxygen delivery technologies.

Data in this article is provided by PLOS.

Related Articles

• Nasal temperature drops during stress, especially social speech stress
• Self-centered reflection increased sense of agency; selfless reflection decreased it
• Autistic adults report lasting mental health benefits from psychedelics
• Working memory links broadly to preadolescent psychopathology in network analysis
• Parents of autistic children reported heavy strain and relied on coping rituals
• Teachers on the Front Line of Bullying What Drives Action and What Gets in the Way
• Turning Heartbreak Into a Story: How Writing About a Breakup Changes What You Remember and Expect Next
• Why Some Mental Health Apps Stick While Others Don’t: Lessons from People Using PolarUs for Bipolar Self‑Management
• When ADHD Care Works, It’s Usually Because the System Finally Does
• When the Brain Stops Staying in Its Lane: What LSD Reveals About Flexibility and Synchrony
• When Campus Noise Becomes More Than a Nuisance: What a 2,080-Student Study Says About Sound Sensitivity, Autistic Traits, Social Skills, and Gender