Transcutaneous carbon dioxide monitoring during prolonged apnoea with high‐flow nasal oxygen

The duration of apnoeic oxygenation with high‐flow nasal oxygen is limited by hypercapnia and acidosis and monitoring of arterial carbon dioxide level is therefore essential. We have performed a study in patients undergoing prolonged apnoeic oxygenation where we monitored the progressive hypercapnia with transcutaneous carbon dioxide. In this paper, we compared the transcutaneous carbon dioxide level with arterial carbon dioxide tension.


| INTRODUCTION
Apnoeic oxygenation with high-flow nasal oxygen has been introduced as sole mean of oxygenation for short laryngeal procedures where the technique allows for better visualisation and working space for the surgical intervention. 1 However, the limiting factors for the duration of apnoeic conditions by any known apnoeic technique are hypercapnia and acidosis. Of particular importance, the respiratory acidosis can develop unpredictably quickly on an individual level during apnoea, 2 and it is therefore paramount to monitor the patient's carbon dioxide level reliably when using apnoeic techniques.
Arterial blood gas (ABG) analysis is considered as the gold standard for measuring the carbon dioxide partial pressure (PCO 2 ) but each ABG sample only represents a snapshot in time and patients need to have either an arterial line or numerous arterial blood samples taken with risk of complications. 3 Transcutaneous carbon dioxide (TcCO 2 ) allows for real-time non-invasive continuous monitoring of PCO 2 . It is commonly used in neonatal intensive care units 4 but has also been proved useful in critically ill patients and intraoperatively. 5 TcCO 2 has previously been found to be a reliable surrogate measure of PaCO 2 that remains accurate during hemodynamic instability and use of vasopressors; however, comparisons have mainly been performed at near normocapnia. 6 We have conducted a clinical trial investigating apnoeic oxygenation with high-flow nasal oxygen where we assessed ABGs over time and simultaneously measured continuous TcCO 2 . During the trial, all patients reached considerable hypercapnia with a mean arterial carbon dioxide (PaCO 2 ) of 11.2 kPa. 2 The aim of this secondary analysis was to compare TcCO 2 measurements with PaCO 2 in patients exposed to prolonged apnoea with high-flow nasal oxygen to determine the agreement between common methods of monitoring PCO 2 during increasing hypercapnia. Patients were healthy, non-obese, 18 years or older, and scheduled for elective surgery under general anaesthesia. Detailed inclusion and exclusion criteria are described elsewhere. 2 The aim of assessing TcCO 2 was specifically described in the study protocol. After loss of consciousness a patent airway was ensured with jaw thrust and minimal mouth opening for passive egress of air. High-flow nasal oxygen flow rate was increased to 70 L/min and the inclination of head and truncus was reduced to 20 . Immediately afterwards, to confirm easy ventilation, the patient was ventilated once via a facemask and subsequently high-flow nasal oxygen was maintained throughout apnoea until a stop criterion, PaCO 2 reaching 12 kPa or pH reaching 7.15, was met.

| Anaesthesia
At termination of apnoeic oxygenation, bag-mask ventilation by facemask was performed, followed by airway management by insertion of either a laryngeal mask or a tracheal tube. Positive pressure ventilation was continued for the remainder of the procedure.

| Monitoring
Preoperatively, an arterial line was placed in the radial artery and connected to a closed blood sample system for repetitive ABG sampling.
ABGs were collected at baseline before preoxygenation, immediately after induction of anaesthesia and repeated every 5 min during the apnoea period. All samples were analysed immediately with either the ABL90 FLEX or the ABL800 (Radiometer) depending on availability.
Continuous TcCO 2 monitoring was performed with the TCM5 FLEX with a Stow-Severinghaus-type sensor, the TC Sensor 54, according to manufacturer guidelines (Radiometer). The sensor temperature was set to 43.5 C. After calibrating the sensor for 10 min in the calibration chamber, the skin of the measuring site superficial to the anterior aspect of the trapezius muscle was cleansed with an alcohol swap before the Tosca fixation ring was applied. Two drops of contact gel were applied to the skin area in the centre of the fixation ring before the sensor was attached. The sensor was calibrated before every new patient was monitored and the membrane was replaced once a week according to the guidelines of the manufacturer.
End-tidal CO 2 (ETCO 2 ) was measured at the termination of apnoeic oxygenation by giving the patient one full tidal volume breath via a tightfitting face mask, followed by a full passive expiration.

| Statistical analysis
Baseline data were reported with mean and standard deviation (SD) or median with interquartile range (IQR) for continuous variables and with frequency and percent for categorical variables.
The agreement between PaCO 2 and TcCO 2 was assessed using Bland-Altman analyses. We calculated the bias (mean difference) and limits of agreement (LOA) which include 95% of the data points.
The statistical analysis and graphs were made in R Studio version 2022.02.2.

| RESULTS
Data were available from a total of 36 patients. Of these, one patient developed hiccups immediately after induction of anaesthesia and thus did not receive high-flow nasal oxygen and was omitted from the analysis. In one patient, the TcCO 2 monitor needed recalibration after induction of anaesthesia, and data were only available from 15 min of apnoea and onwards. In two patients, the ABG at the end of apnoea was erroneously not drawn before ventilation had been started.  Figure S1).  A Bland-Altman analysis comparing PaCO 2 at the end of apnoea with ETCO 2 at the first ventilation revealed that ETCO 2 was lower than PaCO 2 with a bias of À2.7 kPa and LOA from À4.9 to À0.5 kPa.

| DISCUSSION
We found that the association between TcCO 2 and PaCO 2 changed during apnoea as the TcCO 2 initially at normocapnia slightly underestimated PCO 2 whereas at hypercapnia above 10 kPa, TcCO 2 overestimated PCO 2 .
The primary strength of our study is the unique opportunity to study the agreement of monitoring techniques during prolonged apnoea with very high CO 2 levels, providing essential knowledge for the use of TcCO 2 monitoring during procedures with prolonged apnoea or hypopnea. In addition, another strength is the frequent ABG sampling that enabled multiple comparisons between TcCO 2 and PaCO 2 and the highly standardised set up.
One limitation of this study is that there are missing data at baseline due to unsuccessful calibration of the TCM5 on primary attempt.
This resulted in a large SD and the agreement analysis was based on fewer patients at this timepoint. Another limitation is that we only had data from non-obese patients without comorbidity. This limits the generalisability, and we cannot extrapolate the findings to patients with cardiorespiratory comorbidity or higher BMI.  We found that the difference between TcCO 2 and PaCO 2 changed over time when the PCO 2 increased. This PCO 2 dependency was also described by Soerensen et al. 9 in a study investigating the effects of different electrode temperatures in 40 infants. In that study, the difference between PaCO 2 and TcCO 2 increased from 5% at 44 C to 17% at 39 C with overestimation using transcutaneous monitoring with bias increasing approximately 2% per kPa rise in PCO 2 . The correlation between TcCO 2 and PaCO 2 depends on the temperature in the capillary bed, the perfusion of the skin and the local production of CO 2 . As a result, the temperature of the skin is a possible bias. The longer the electrode is applied to the skin, the higher skin surface temperature. Soerensen et al. allowed the sensor to equilibrate and stabilise for 30 min between each reading, which may increase the precision of the TcCO 2 , but preheating the skin before measuring at lower temperatures did not make any difference. Furthermore, a study by Janssens et al. 10 investigated the drift of the TcCO 2 signal over 8 h of monitoring in 10 adult patients and found no significant drift comparing TcCO 2 and PaCO 2 . A study by Nishiyama et al. 11 investigating the best electrode placement also found no drift in TcCO 2 at an electrode temperature of 43 after 2 h of monitoring. This speaks against the assumption that skin perfusion is clinically important for the accuracy of these measurements.
Placement of the transcutaneous sensor may also affect the accuracy of the monitoring. It is recommended that the sensor is placed in a highly vascularised area 12 and potential sites are the forearm, earlobe, cheeks or the forehead amongst others. We chose to place the sensor superficial to the anterior aspect of the trapezius muscle with a sensor temperature of 43.5 C according to manufacturers' recom- Hinkelbein et al. 20 found a good agreement between TcCO 2 and PaCO 2 during interhospital ground transport of intubated critically ill patients with a PCO 2 ranging between 3.2 and 9.7 kPa (bias À0.08 kPa; LOA À2.1 to 1.9 kPa.).
In the normal range of PCO 2 the observed difference between TcCO 2 and PaCO 2 does not seem to be clinically important whereas at hypercapnia, reaching PaCO 2 above 10 kPa the TcCO 2 overestimated PCO 2 and must be interpreted with caution and optimally verified with an ABG prior to clinical action such as terminating a procedure with apnoeic oxygenation and converting to tracheal intubation when it is not yet necessary.
In conclusion, TcCO 2 provided an acceptable substitute for PaCO 2 at PCO 2 levels below 10 kPa, whereas TcCO 2 overestimated PCO 2 at higher levels in patients undergoing apnoeic oxygenation with high-flow nasal oxygen.

AUTHOR CONTRIBUTIONS
All authors contributed to the study conception and design. Material