From the authors

We appreciate your interest in our paper 1 and are grateful for the opportunity to clarify some issues.

We agree that several types of nail polish can affect pulse oximeter readings. Actually, none of the patients in our study were wearing any nail polish during pulse oximetry; hence, we believe that the results obtained are reliable.

As for the question of hypercapnia, most authors establish hypercapnic status on the basis of carbon dioxide arterial tension (P_a,CO2) levels >45 mmHg (>5.99 kPa). J. Hinkebein and H.V. Genzwerker consider that a mean of 45.3±8.0 mmHg (6.04±1.07 kPa) does not exactly reflect hypercapnia. In this respect, it is important to remember that the mean is a descriptive value that addresses only the central trend, not the variability. As we show in table 2 1, at least one-third of our patients (upper tertile in table 2; n = 274) had P_a,CO2values >48 mmHg (>5.99 kPa). In fact, we believe that the wide spectrum of P_a,CO2values included is one of the strong points of the study.

With regard to the comments on our use of Bland and Altman analysis, we consider that this is not an issue since the difference between two values is completely symmetrical and the interpretation of the results is not altered. It is also important to note that the mean bias as described by Bland and Altman assumes that there is no trend along the mean of the two measurement axis. We included the complementary regression analysis with the sole purpose of addressing this issue.

The mechanism by which P_a,CO2 variation can affect agreement between arterial oxygen saturation (S_a,O2) and arterial oxygen saturation measured by pulse oximery (S_p,O2) is unknown. J. Hinkebein and H.V. Genzwerker suggest the possibility that hypercapnia-induced vasoconstriction might be responsible for the discordance between these determinations. However, the effect that hypercapnia plus hypoxaemia (the situation of most of the patients in our study) might have on vascular tone is uncertain. Although this hypothesis cannot be excluded, the fact that a decrease in P_a,CO2 also alters agreement between S_a,O2 and S_p,O2undermines this interpretation. It could, however, support our idea that the carbohaemoglobin level is actually the factor affecting the pulse oximeter readings.

Lastly, we must mention that our study provides the first observation that carbon dioxide arterial tension can have an influence on the arterial oxygen saturation measured by pulse oximery reading. Future studies should confirm this finding and investigate the mechanisms responsible for this effect.

Statement of interest

None declared.