P_aO2/F_IO2 Is Not a Guesstimation

To the Editor:

I read with great interest and with some concern the paper by Cao et al¹ that compared volume-controlled and pressure-limited noninvasive ventilation (NIV) in subjects with acute hypercapnic respiratory failure. The authors reported on P_aO2/F_IO2 rather than P_aO2 as part of their baseline demographics in Table 1 and in their results in Figure 5. The methods for NIV described that supplemental oxygen was delivered through a port in the mask with flow adjusted to maintain oxygen saturation using a specific machine (FLEXO ST30, Curative Medical Technology, Suzhou, China). They used a formula to determine F_IO2 delivered to each subject and reported results as P_aO2/F_IO2, which causes me some concern. The authors did state that the conversion factor provided an approximation of F_IO2, which was influenced by minute ventilation, breathing pattern, and gas leakage. I feel that an approximation of F_IO2 cannot accurately predict P_aO2/F_IO2 where a completely closed system and accurate F_IO2 is not provided. I found one abstract² that used the formula to help rapidly predict correlation of S_pO2/F_IO2 in non-intubated patients as a surrogate to arterial blood gas for determining P_aO2/F_IO2 in emergency room patients, which may be a novel noninvasive way to assess respiratory failure.^3–5

I could not find any reference to the formula to validate and verify an F_IO2 via a port for NIV. I have used REMstar NIV machines (Philips Respironics, Murrysville, Pennsylvania) for several decades, and the general procedure limits oxygen to no more than 10 L for F_IO2 approximately 0.4–0.5. I would need a machine that has a blender that would guarantee a delivered F_IO2 for higher concentrations to obtain an accurate P_aO2/F_IO2. Examples of calculated flows based on the formula would be as follows: 5 L = 36%, 10 L = 51%, 15 L = 66%, or 20 L = 81%.

I recently cared for a patient with acute hypoxemic respiratory failure and an S_pO2 of 89% on 6 L nasal cannula. The patient was placed on a CPAP device without an oxygen blender as I waited for a different machine with a blender. No matter what flow, based on the limitations of that machine, the saturation did not change. The patient required an F_IO2 of >0.6 before the S_pO2 began to increase. At 8, 10, 15, and 20 L bled into the mask port, S_pO2 remained doggedly fixed at 89–90%. Indeed, the limitations on this device did not agree with the formula for F_IO2 described in the paper.

The use of this equation may represent some false acceptance that the F_IO2 is truly delivered to the patient and may lead to false assumptions regarding the seriousness of the subject's condition. Although the paper has merit with regard to volume or pressure targeting, the authors should have reported flow and P_aO2 so that readers could make their own assumptions about oxygenation.

• Copyright © 2017 by Daedalus Enterprises