Automated Oxygen Titration: Advancing But Not Yet Free

A longtime gap in the care of patients with chronic lung disease is the lack of a highly portable, convenient, and effective way of ameliorating dyspnea and improving exercise endurance during ambulation. Over the past several decades, multiple approaches have been tried, with varied success. Oxygen supplementation has been one of those approaches, especially in patients with COPD.¹ Unfortunately, evidence to support its value in augmenting exercise capacity and reducing dyspnea has been weak and inconsistent.^2,3 Problems with prior studies include lack of standardization for increasing O₂ supplementation with exercise and failure to ascertain that O₂ supplementation is sufficient to avoid severe desaturation.⁴

Prior approaches to improve exercise endurance include mask noninvasive ventilation and high-flow nasal cannula (HFNC).^5,6 These modalities improve exercise endurance and oxygenation, with reduced dyspnea, but are not portable. The bulkiness of the devices themselves and the need for a source of electric power via a wall socket or heavy battery have limited their use to stationary exercise. Another approach is exemplified by noninvasive open ventilation (NIOV) (Life2000, Hillrom, Chicago, Illinois). It consists of a 1 pound device that can be strapped to a belt and provides adjustable volumes (up to 250 ccs) of pulsed oxygen via nasal prongs during inspiration to accommodate rest, moderate and higher levels of activity. It improves cycle ergometer exercise endurance and O₂ saturation compared to continuous low-dose nasal O₂.⁷ However, the requirement for a 50-psi source of O₂, such as an E-tank, to power the device limits its usefulness during ambulation.

In this issue of Respiratory Care, Vézina et al report their evaluation of automated O₂ titration (auto-O₂) using the FreeO₂ device (OxyNov, Québec, Canada) that is a closed-loop system that monitors S_pO2 and automatically adjusts supplemental O₂ flow, ranging from 1–20 L/min, to maintain a target O₂ saturation.⁸ The authors are transparent in acknowledging that some are shareholders and one is a cofounder of OxyNov, which is important to consider in a study that is difficult to blind. They have performed a number of prior studies on their device dating back over 10 years including on healthy subjects⁹ and subjects with severe COPD.¹⁰ They have shown that auto-O₂ compared to fixed-flow O₂ (fixed O₂) from the same device maintains a target S_pO2 for a greater percentage of time, reducing duration of both hypoxia and hyperoxia, and prolongs exercise duration in subjects with COPD.^9,10

Their current study extends these findings by examining effects on dyspnea, the study's main outcome variable, using the Borg score after a 3-min constant-speed shuttle test (CSST), a validated test for exertional dyspnea. Contrary to the main hypothesis, Borg score was not significantly reduced by auto-O₂ compared to fixed O₂, attributed by the authors to the slow response time of the auto-O₂ setting in the context of a 3-min test. However, an interesting and very important outcome (albeit exploratory) was that the time for the endurance shuttle walk test (ESWT) increased 56% during auto-O₂ versus fixed O₂ for all enrolled subjects. Among the participants with COPD (10) and interstitial lung disease (ILD) (10), the increases in endurance time with auto-O₂ versus fixed O₂ were 58% and 74%, respectively. Similar increases were seen in subjects with pulmonary hypertension (PH) and cystic fibrosis (CF), but participant numbers were only 5 and 3, respectively, too few for meaningful statistical analysis. In the whole cohort, Borg dyspnea scores during the ESWT registered statistically and clinically important improvements at isotime with auto-O₂ versus fixed O₂. The time at target S_pO2 (94 ± 2%) was also significantly greater during the ESWT with auto-O₂ than with fixed O₂, but not during the CSST. The duration of the ESWT was, on average, more than double that of the CSST, allowing more time for auto-titration; and, notably, adjunct use of HFNC with the auto-O₂ did not show further improvements in exercise endurance or dyspnea scores.

Strengths of the study include the crossover design, the enrollment of subjects with diverse etiologies, and the selection of only subjects with significant exertional O₂ desaturation (drop in S_pO2 of ≥ 5% to < 88%), making them more likely to respond. Weaknesses include the single-center and small sample size, especially in the PH and CF subgroups, as well as the difficulty with blinding. Both auto-O₂ and fixed O₂ were delivered via the same FreeO₂ system in an attempt to blind participants and assessors, but it is likely that at least some were able to sense the higher oxygen flows during auto-O₂. Furthermore, some of the findings are confirmatory rather than novel, including the enhanced ability of auto-O₂ to maintain target S_pO2 and prolong exercise endurance in subjects with COPD compared to fixed O₂. New findings include the reduction in exertional dyspnea at isotime during the ESWT and significant benefits in the ILD subgroup paralleling those seen in patients with COPD. In ILD participants, perhaps related to the greater exertional O₂ desaturations seen, the response to auto-O₂ appeared to be even more robust than with COPD. These findings are notable because of the dearth of evidence on response of patients with ILD with exertional O₂ desaturation to O₂ supplementation.

The lack of enhancement with HFNC was contrary to the initial hypothesis and sheds light on possible mechanisms of auto-O₂ benefit. Although only one HFNC gas flow was tested and other flows could have been more beneficial, the lack of benefit with HFNC suggests that high gas flows and hence flushing of dead space that improves ventilatory efficiency is not as important as supplementing O₂ and maintaining a target O₂ saturation. Supporting this is the lack of differences in S_pO2 and dyspnea scores between auto-O₂ and fixed O₂ during the CSST in contrast with the sizable improvements in S_pO2 associated with better dyspnea scores and exercise endurance with auto-O₂ versus fixed O₂ during the ESWT. Also, this is compatible with prior findings in the NIOV study⁶ where use of NIOV with compressed air yielded no improvements in exercise endurance (or S_pO2) compared to room air alone; only when NIOV was powered with compressed O₂ were improvements in S_pO2 and exercise endurance seen.

Evidence from this study and previous ones by these authors convincingly shows that auto-O₂ using the FreeO₂ device maintains target S_pO2 better than fixed O₂ as they applied it. It must be acknowledged though that providing fixed nasal O₂ at 2 L/min or the usual flow +1 L/min “as [is] commonly done in clinical practice” stacked the odds in favor of auto-O₂. Had the authors used a range of higher continuous O₂ flows, especially at 6 L/min or more, the results might have looked quite different. On the other hand, the automaticity of the FreeO₂ device is likely an advantage over relying on busy caregivers to make prompt and appropriate adjustments in O₂ flows, reducing the risks not only of prolonged hypoxia but also hyperoxia, although this hasn't been demonstrated in a clinical setting.

Future research should focus on mechanisms of benefit and allowing greater device portability. Additionally, the impact of variable oxygen flows on patient important clinical outcomes require study. However, as long as aides are necessary to push a cart behind the patient to permit ambulation and O₂ flows up to 20 L/min are required, use will be limited to higher acuity settings such as with early mobilization of ICU patients or in rehabilitation settings with sufficient resources. Until these constraints can be addressed, home applications and use with unassisted ambulation as the name FreeO₂ implies will remain out of reach.