Current Issues

2021 : Volume 1, Issue 1

Assessing Agreement of Transcutaneous Carbon Dioxide Monitoring and Blood Gas Analysis in a Neonatal Population

Author(s) : Mia Kahvo 1 , Ajit Mahaveer 2 and Ranganath Ranganna 2

1 Regional Newborn Intensive Care Unit , St Michael's Hospital , England

2 Newborn Intensive Care Unit , St Mary's Hospital , England

Open J Pediatr Neonatol

Article Type : Research Article



Objective: To assess agreement between transcutaneous carbon dioxide (TcCO2) monitoring and blood gas analysis in neonates.

Study Design: This was a prospective observational study performed in a tertiary neonatal intensive care unit. 19 infants with a mean postmenstrual age of 35+3 weeks were included. Agreement was assessed by Bland-Altman analysis and concordance correlation coefficient. End-user feedback was collected from staff and infants were assessed for evidence of skin damage.

Results: Overall bias from 698 paired samples was -0.30 (SD 1.21, p<0.0001) with good concordance (CCC 0.80). 69% (95% CI 65%-72%, p=0.0003) of samples fell within the predefined clinically acceptable difference of 1kPa. Agreement was more favorable for non-invasively ventilated infants (bias -0.11, CCC 0.91). Staff feedback was positive, and no infants suffered skin damage.

Conclusion: TcCO2 monitoring is a reliable assessment tool for both invasively and non-invasively ventilated neonates. It can be used as an adjunct to blood gas analysis, reducing the frequency of invasive blood tests.

Keywords: Neonate; Transcutaneous Monitoring; Rspiration; Blood Gas Analysis



Mechanical ventilation (MV) is an important tool when managing unwell infants in the neonatal intensive care unit (NICU)[1], with up to 85% requiring some form of MV[2]. However, there remain a number of risks associated with the use of MV, particularly with fluctuations in partial pressure of carbon dioxide (PCO2), such as increased risks of intraventricular haemorrhage, periventricular leukomalacia and bronchopulmonary dysplasia [3]. The current ‘gold standard’ for measurement of PCO2 is blood gas analysis. While this provides vital information, it is linked with a number of disadvantages, namely those associated with physiological instability from repeated painful procedures [4], risks of infection form repeated central line access [5] and potential for iatrogenic anaemia [6]. In addition, recognition of fluctuations in PCO2 may be delayed and opportunities to wean ventilation missed, thereby prolonging the length of ventilation time. Transcutaneous CO2 (TcCO2) is an alternative technique which allows continuous monitoring of the infant and avoids harms associated with repeated blood sampling. A heated sensor, separated from the skin by a membrane, is attached to the body causing local vasodilatation, increasing permeability of skin to CO2 allowing its diffusion through the membrane to the sensor. A correction factor is applied to calculate PCO2, offsetting the effect high temperatures have on the solubility of CO2 in blood [7].

Many studies have sought to investigate the applicability of transcutaneous monitoring during neonatal surgery [8, 9] or through the course of neonatal transport [10, 11]. Time of monitoring in these studies was relatively short, making the results difficult to apply for use on a neonatal unit. Of recent studies which have looked at the use of these monitors on the neonatal unit, results have either not differentiated between invasively or non-invasively ventilated neonates [12] or have focused on only ventilated [7] or premature neonates [13]. Decision making on the neonatal unit can be difficult. Knowing how to apply transcutaneous monitoring for these subdivisions of infants, and the clinically relevant differences, is important. Therefore, the aim of this study was to evaluate the reliability and accuracy of TcCO2 (SenTec© monitors) in a range of infants in a Tertiary NICU. In particular, the feasibility of using transcutaneous monitoring on a mixed population of infants (term and premature), including those who were invasively and non-invasively ventilated, was examined.


This was a single center prospective observational study performed from August 2019 to November 2019. Infants admitted to a Tertiary NICU were prospectively enrolled. As this was an observational trial, written consent and ethical approval was not indicated. Decisions on which infants to commence TcCO2 monitoring on was made by the Consultant Neonatologist on the unit that day. Routine standard care was continued while TcCO2 was simultaneously monitored. Written, hourly recordings of TcCO2 were made and paired with blood gases, taken only as deemed clinically indicated. Transcutaneous CO2 was measured using the SenTec© Digital Monitoring System (SenTec AG, Therwil, Switzerland) and utilised as per manufacturer guidelines with sensors heated to 41°C. Blood gas analysis was performed by the GEM Premier 5000 analyser (IL, Bedford, MA, United States). Statistical analysis was performed with MedCalc Statistical Software version 16.4.3 (MedCalc Software, Ostend, Belgium).

Patient characteristics are described using qualitative (frequencies) and quantitative variables (mean and range). As there is no statistical rule to help define an acceptable level of agreement between two different methods, and is rather more a matter of clinical judgment [14], a difference of +/-1kPa (7.5mmHg) was felt to be clinically acceptable based on current literature [15]. To assess agreement between TcCO2 and ‘gold standard’ BG PCO2 a combination of statistical tools was used: concordance correlation coefficient (CCC) [16], Bland-Altman analysis [17] and percentage agreement (using 75% agreement to be acceptable). End-user feedback of the SenTec© monitors as well as any evidence of skin damage from the heated sensors was collected, as judged by the nurse caring for the infant.


19 infants were enrolled, giving a total of 698 paired TcCO2/BG PCO2 samples. Data on patient characteristics was unavailable for two infants. Mean corrected gestational age was 35+3 weeks (range 25+0 - 51+5) and mean infant weight was 1985g (range 675g - 4900g). The mean number of paired blood gas/TcCO2 recordings was 36 (range 5 – 97) while each infant was monitored for a mean of 195 hours (range 17 – 521). Bland-Altman analysis showed an overall bias of -0.30kPa (-2.25mmHg). Overall concordance was good (CCC 0.80, 95% CI 0.77-0.82) and 69% of samples (95% CI 65%-72%, p=0.0003) were within 1kPa of each other (Table 1 and Table 2).

Paired samples

Bias (SD)

Lower LOA (95% CI)

Upper LOA (95% CI)

P value

All infants


-0.30 (1.21)

-2.67 (-2.82 to -2.52)

2.08 (1.92 to 2.23)


Without outliers


-0.30 (1.01)

-2.28 (-2.41 to -2.14)

1.67 (1.54 to 1.80)


>30 weeks cGA


-0.24 (1.28)

-2.74 (-2.93 to -2.55)

2.27 (2.08 to 2.45)


<30 weeks cGA


-0.44 (0.92)

-2.24 (-2.48 to -1.99)

1.35 (1.11 to 1.60)


Invasive ventilation


-0.34 (1.27)

-2.83 (-3.01 to -2.65)

2.15 (1.97 to 2.32)


Non-invasive ventilation


-0.11 (0.88)

-1.83 (-2.10 to -1.57)

1.61 (1.34 to 1.87)


Table 1: Comparison of blood gas CO2 and transcutaneous CO2 monitoring (results shown in kPa, 1kPa=7.5mmHg), cGA: corrected gestational age; 95 % CI: 95% confidence interval; LOA: limit of agreement.


Proportion in agreement (%)

95% CI (%)

P value

All infants


484 (69)



>30 weeks cGA


361 (69)



<30 weeks cGA


115 (71)



Invasive ventilation


379 (66)



Non-invasive ventilation


105 (83)



Table 2: Proportion of paired samples in agreement.

Figure 1 (Panel A) is a graphical representation of Bland-Altman analysis. There are a number of paired samples lying outside the upper and lower limits of agreement. On review of these outliers, a number of samples were recorded from a baby undergoing therapeutic hypothermia (25 paired samples). Of the remaining 34 outliers, 17 were recorded at a time when the TcCO2 sensor was due re-calibration. If these outliers are removed, there was no difference in the overall bias and only a moderate improvement in the limits of agreement (Table 1).

Figure 1: Panel A: Bland-Altman plot showing bias (mean difference) for TcCO2 and BG CO2. The mean line represents bias. The dashed lines represent the upper and lower limits of agreement. The full lines represent the clinically acceptable difference (1kPa/7.5mmHg); Panel B: Blood gas CO2 values versus TcCO2 values are plotted. The dashed line represents the identity line. The solid line represents the line of best fit.

Infants were also subdivided based on corrected gestation at the time of monitoring (>30 weeks or <30 weeks) and their mode of ventilation (invasive or non-invasive). Bias remained within the pre-defined clinically acceptable range (Table 1, and Figure 1, Panels C-F). Concordance remained good but was stronger for infants >30 weeks (CCC 0.81, 95% CI 0.78 – 0.84) and those who were non-invasively ventilated (CCC 0.91, 95% CI 0.87 – 0.94). For infants aged <30 weeks gestation (CCC 0.72, 95% CI 0.64 – 0.78) and undergoing invasive ventilation (CCC 0.75, 95% CI 0.71 – 0.78) concordance remained acceptable.

Figure 2:
Bland-Altman plots showing bias (mean difference) for TcCO2 and BG CO2. The mean line represents bias. The dashed lines represent the upper and lower limits of agreement. The full lines represent the clinically acceptable difference(1kPa/7.5mmHg). Panel A: Bland-Altman plot for infants with corrected gestation >30 weeks; Panel B: Bland-Altman plot for infants with corrected gestation <30 weeks; Panel C: Bland-Altman plot for infants undergoing invasive ventilation only; Panel D: Bland-Altman plot for infants with undergoing non-invasive ventilation only.

Feedback from staff was largely positive (Table 3). None of the 19 infants were found to have suffered from skin damage from the heated sensors.


Average Score

Overall impression




Pre-use test/calibration



Clear/intuitive user interface


Monitor display- numerics


Monitor display- graphics


Monitor display- trends





Ease of application to skin


Ease of changing sensor membrane


Ease of sensor/ring removal from skin




Total number of responders 16; Scores: 1= Very poor, 10= Excellent. Number of infants with skin damage noted secondary to the heated sensor= 0.

Table 3: End user feedback.


This was a single center study that analyzed a predominantly older cohort of ventilated neonates. It is reasonable to assume that infants most at risk from fluctuations of PCO2 would be those at the extremes of prematurity and very low birth weight infants. Therefore, results may not be generalizable to this cohort of infants, but still considered to be used with close monitoring. While a large number of paired samples were analyzed, these came from only 19 individual infants and there were limitations in the availability of equipment and staff that were familiar with its use. For the purposes of statistical analysis, there was no differentiation between venous/capillary/arterial samples, though the majority is assumed to be capillary samples as that are standard practice in the unit. However, studies have shown that there is good correlation between all three sampling methods [18].


Although direct comparison with similar studies is difficult due to differences in trial methodologies, results from this study have shown more favorable agreement between transcutaneous monitors and blood gas analysis when compared to previous similar trials. In sub-analysis according to gestation, agreement was better when compared to studies performed in a purely preterm population [19], showing less variation between individual infants. Results were also more favorable for ventilated neonates, irrespective of gestation, when compared to previously published data [7]. In addition to the range of infants studies, another strength of the study was the length of time infants were monitored for, and was significantly longer than comparable trials [20]. In addition to showing favorable agreement with blood gas analysis, this study has also found the heated sensors to be safe to use with no evidence of skin damage as assessed by clinical staff. This was particularly encouraging for the extremely low birth weight babies (<1000g) of which there were 4 (weight range 675g-960g at the time of monitoring) and included 3 babies with corrected gestation <30 weeks. Studies have been performed on extremely low birth weight infants with no evidence of skin damage identified [19] confirming our results, as well as at lower probe temperatures with no difference found in results when compared to the standard recommended probe temperature [21].

One potential disadvantage of transcutaneous monitors is the apparent time-consuming calibration process and need for frequent electrode changes. In particular, this poses challenges when caring for unstable infants where a minimal handling approach is necessary. However, feedback gained from staff in this study was reassuring. With appropriate training, the electrode can be changed with minimal disturbance to the infant. As far as we are aware, this is the first study to seek end-user feedback on transcutaneous monitors in a NICU. In addition, TcCO
2 monitoring has been shown to reduce the frequency of blood gas sampling, thus easing the work for the clinical team and reducing the number of times infants are exposed to these painful stimuli [7, 23]. While it may not be possible to state from these results that TcCO2 is sufficiently reliable and accurate to completely substitute blood gas PCO2, agreement is sufficiently high to allow use as an adjunct to blood gas sampling, allowing continuous monitoring of trends. This can alert the clinical team of the need for more detailed assessment of the infant, which may otherwise be missed if blood gases alone were used. It can guide weaning of ventilation for preterm infants, avoiding extreme values of PCO2 which have been shown to be detrimental to outcomes and lead to less frequent blood gas sampling.


Results from this observational study have shown that agreement between TcCO2 and blood gas results is good, allowing use as an adjunct to blood gas analysis in a neonatal population. Monitors were positively reviewed by staff and no infants were found to have suffered skin damage secondary to the heated sensors. This study can be a basis for larger studies to further assess the suitability of TcCO2 monitoring and long-term outcomes, particularly in an extremely preterm infant cohort.


Funding Source: No external funding was secured for this study.
Conflict of Interest: All transcutaneous monitors and consumables were provided for by SenTec© (SenTec AG, Therwil, Switzerland). However, SenTec© were not involved in data collection, analysis or write-up of the report and did not provide any funding.

Author Contributions

RR and AM were responsible for study concept and design. MK performed data collection, analysis and drafted the manuscript. RR and AM reviewed and edited the manuscript.


1. Brown MK, DiBlasi RM. Mechanical Ventilation of the Premature Neonate. Respir Care. 2011;56:1298-311; discussion 311-3.

2. van Kaam AH, Rimensberger PC, Borensztajn D, et al. Ventilation Practices in the Neonatal Intensive Care Unit: A Cross-Sectional Study. J Pediatr. 2010;157:767-71.e1-3.

3. Bruschettini M, Romantsik O, Zappettini S, et al. Transcutaneous Carbon Dioxide Monitoring for the Prevention of Neonatal Morbidity and Mortality. Cochrane Database Syst Rev. 2016;2:CD011494.

4. Medicine COFANaSOAAP. Prevention and Management of Procedural Pain in the Neonate: An Update. Pediatrics. 2016;137:e20154271.

5. Hermansen MC, Hermansen MG. Intravascular Catheter Complications in the Neonatal Intensive Care Unit. Clin Perinatol. 2005;32:141-56, vii.

7. Mukhopadhyay S, Maurer R, Puopolo KM. Neonatal Transcutaneous Carbon Dioxide Monitoring--Effect on Clinical Management and Outcomes. Respir Care. 2016;61:90-7.

8. Karlsson V, Sporre B, Ågren J. Transcutaneous PCO2 Monitoring in Newborn Infants During General Anesthesia is Technically Feasible. Anesth Analg. 2016;123:1004-7.

9. Chandrakantan A, Jasiewicz R, Reinsel RA, et al. Transcutaneous CO2 Versus End-Tidal CO2 in Neonates and Infants Undergoing Surgery: A Prospective Study. Med Devices (Auckl). 2019;12:165-72.

10. Tingay DG, Stewart MJ, Morley CJ. Monitoring of end Tidal Carbon Dioxide and Transcutaneous Carbon Dioxide During Neonatal Transport. Arch Dis Child Fetal Neonatal Ed. 2005;90:F523-6.

11. O'Connor TA, Grueber R. Transcutaneous Measurement of Carbon Dioxide Tension during Long-Distance Transport of Neonates Receiving Mechanical Ventilation. J Perinatol. 1998;18:189-92.

12. Lambert LL, Baldwin MB, Gonzalez CV, et al. Accuracy of Transcutaneous CO2 Values Compared With Arterial and Capillary Blood Gas. Respir Care. 2018;63:907-12.

13. Janaillac M, Labarinas S, Pfister RE, et al. Accuracy of Transcutaneous Carbon Dioxide Measurement in Premature Infants. Crit Care Res Pract. 2016;2016:8041967.

14. Hollis S. Analysis of Method Comparison Studies. Ann Clin Biochem. 1996;33:1-4.

15. Restrepo RD, Hirst KR, Wittnebel L, et al. AARC Clinical Practice Guideline: Transcutaneous Monitoring of Carbon Dioxide and Oxygen: 2012. Respir Care. 2012;57:1955-62.

20. Sandberg KL, Brynjarsson H, Hjalmarson O. Transcutaneous Blood Gas Monitoring During Neonatal Intensive Care. Acta Paediatr. 2011;100:676-9.

21. Jakubowicz JF, Bai S, Matlock DN, et al. Effect of Transcutaneous Electrode Temperature on Accuracy and Precision of Carbon Dioxide and Oxygen Measurements in the Preterm Infants. Respir Care. 2018;63:900-6.

22. Lambert LL, Baldwin MB, Gonzalez CV, et al. Accuracy of Transcutaneous CO. Respir Care. 2018;63:907-12.

23. Geven WB, Nagler E, de Boo T, et al. Combined Transcutaneous Oxygen, Carbon Dioxide Tensions and end-Expired CO2 Levels in Severely Ill Newborns. Adv Exp Med Biol. 1987;220:115-20.


Corresponding Author: Dr Mia Kahvo (MBChB), Regional Newborn Intensive Care Unit, St Michael’s Hospital, Southwell Street, Bristol, BS2 8EG, UK.

Copyright: © 2021 All copyrights are reserved by Mia Kahvo, published by Coalesce Research Group. This This work is licensed under the terms of the Creative Commons Attribution License, which permits unrestricted use, distribution and reproduction in any medium, provided the original author and source are credited.

Support Links

Track Your Article

Twitter Tweets