N415 Module 2 Critique of a Randomized Oxygen Trial Example

Not Inferior, Not Yet Proven: A Critique of a Randomized Trial of Dry Versus Bubble-Humidified Oxygen

Student Name

Pre-licensure BSN Program, Aspen University

N415: Research / Evidence-Based Practices

Instructor Name

Month Day, Year

Not Inferior, Not Yet Proven: A Critique of a Randomized Trial of Dry Versus Bubble-Humidified Oxygen

A study's conclusion is only as good as the methods under it. This paper critiques a quantitative study that bears directly on a question from my medical unit, whether oxygen delivered by nasal cannula needs a bubble humidifier. The Oxyrea trial randomized intensive care patients to dry or bubble-humidified oxygen and measured their discomfort (Poiroux et al., 2018). The critique follows the usual order, from problem and purpose through design, sample, measurement, ethics, results and applicability (Polit & Beck, 2021), and judges each part against the claims the authors make.

Problem, Purpose and Hypothesis

The authors framed a clear problem: bubble humidifiers are used widely even though earlier studies and guidelines found little evidence that they help at low flows, and discomfort from dry oxygen is the main reason given for using them. Their hypothesis was that dry oxygen would be no worse than humidified oxygen for comfort, which makes this a non-inferiority trial. That choice fits the practical question. A unit does not need proof that dry oxygen is better; it needs reasonable confidence that removing the humidifier does not make patients less comfortable. The trial was registered and was funded through a French national program for nursing research, which adds to its relevance for nursing practice.

Design and Non-Inferiority Margin

Oxyrea was a multicenter randomized trial run in nine intensive care units from 2011 to 2014, with randomization stratified by oxygen flow at 4 liters a minute or less and above 4. Non-inferiority depends on a margin, the largest difference that would still count as no worse. The authors set it at 20% of the humidified group's mean discomfort score, which at the first measurement came to 5.4 points on a 150-point scale. Reporting guidance for non-inferiority trials asks authors to justify the margin clinically (Piaggio et al., 2012). The paper explains the 20% rule but not why a difference of about five points would be unimportant to a patient, so the reader must accept the margin on trust.

Sample, Attrition and Analysis Sets

Adults receiving dry oxygen at 1 to 15 liters a minute by cannula or mask for less than two hours were eligible, and patients who could not answer questions, had a tracheostomy or were receiving palliative care were excluded. Of 354 patients randomized, 172 went to humidified oxygen and 182 to dry oxygen. At six to eight hours, the per-protocol analysis included 136 and 141 patients, and at 24 hours only 72 and 86, a loss the reader has to trace through the study flowchart. Losing more than half the sample by 24 hours weakens every later comparison. The authors analyzed both per-protocol and modified intention-to-treat sets and reported the same conclusions, which is appropriate, since in a non-inferiority trial a lenient analysis can make two treatments look alike.

Allocation, Blinding and Measurement

The trial was open-label. Patients and the health practitioner who read each of the 15 discomfort items aloud could both tell whether a humidifier was attached. For a subjective outcome, that is the trial's most serious weakness, because expectations can shift scores in either direction. The outcome measure was new. The 15 items, from dry mouth and nasal dryness to headache and noise, came from the literature and from a survey of 50 intensive care patients, and each was scored from 0 to 10. The authors acknowledge that the scale was not validated before the trial and that its validation used the same dataset, so its reliability and validity cannot be judged independently.

Ethical Considerations

An ethics committee approved the trial. Because both strategies were considered standard care under French law, written consent was not required; patients were informed orally and in writing and could decline at any point. The reasoning is defensible for a low-risk comparison of two accepted practices, although a United States institutional review board might have asked for written consent or a formal waiver.

Results

Discomfort was low in both groups. In the per-protocol analysis at six to eight hours, mean scores were 26.9 with humidification and 29.7 without, a difference of 2.8 points with a 90% confidence interval from -0.4 to 6.0. Because the interval crossed the 5.4-point margin, non-inferiority could not be shown, and the result was inconclusive rather than negative. At 24 hours, dry oxygen was non-inferior overall and in patients at 4 liters a minute or less, whose mean score was 24.8 on dry oxygen and 29.6 on humidified oxygen. For patients above 4 liters, a post hoc analysis suggested more discomfort without humidification (36.4 against 23.8). Fewer than one patient in five scored above 45 in either group, and intubation, ear, nose and throat infection, length of stay and mortality did not differ (Poiroux et al., 2018).

Threats to Validity

Four problems limit confidence. First, the sample size was not calculated formally, because the new scale had no prior data; the authors expected a mean near 45 with little spread, observed a mean near 27 with wide spread, and later estimated that about 1,478 patients would have been needed. The trial was underpowered for its own question. Second, the flow-rate findings come from subgroups, one of them post hoc, so they suggest rather than show. Third, the open-label design and new scale add measurement bias. Fourth, a careful reader will notice that the abstract reports slightly different figures from the main text, because it gives the modified intention-to-treat values (a difference of 3.2 and a margin of 5.3); the conclusions match, but the difference is a reminder to read beyond the abstract.

Applicability to a Medical Unit

Intensive care patients in France are not the same as adults on a medical unit in the United States receiving 2 liters a minute for pneumonia, and applying these findings to my unit requires caution. Still, the evidence points the same way: the subgroup most like our patients, those at 4 liters or less, did no worse on dry oxygen by 24 hours, and an older American study of patients at higher flows also found that complaints of dry nose and throat were common with or without humidification (Campbell et al., 1988). As a single randomized trial, Oxyrea supports removing routine humidifiers at low flows but does not settle the question for higher flows.

Conclusion

Oxyrea asked a practical nursing question with an appropriate design, randomized a reasonable number of patients and reported its uncertainty honestly. Its weaknesses, an unjustified margin, an unvalidated scale, open-label measurement, heavy attrition and too few patients, mean that its main result was inconclusive. Read with those limits, it adds modest support to the case against routine humidification at low flows and a caution for patients on higher flows.

References

Campbell, E. J., Baker, M. D., & Crites-Silver, P. (1988). Subjective effects of humidification of oxygen for delivery by nasal cannula: A prospective study. Chest, 93(2), 289-293. https://doi.org/10.1378/chest.93.2.289

Piaggio, G., Elbourne, D. R., Pocock, S. J., Evans, S. J. W., Altman, D. G., & CONSORT Group. (2012). Reporting of noninferiority and equivalence randomized trials: Extension of the CONSORT 2010 statement. JAMA, 308(24), 2594-2604. https://doi.org/10.1001/jama.2012.87802

Poiroux, L., Piquilloud, L., Seegers, V., Le Roy, C., Colonval, K., Agasse, C., Zinzoni, V., Hodebert, V., Cambonie, A., Saletes, J., Bourgeon, I., Beloncle, F., & Mercat, A. (2018). Effect on comfort of administering bubble-humidified or dry oxygen: The Oxyrea non-inferiority randomized study. Annals of Intensive Care, 8, Article 126. https://doi.org/10.1186/s13613-018-0472-9

Polit, D. F., & Beck, C. T. (2021). Nursing research: Generating and assessing evidence for nursing practice (11th ed.). Wolters Kluwer.