Abstract
BACKGROUND: High electrode temperature during transcutaneous monitoring is associated with skin burns in extremely premature infants. We evaluated the accuracy and precision of CO2 and O2 measurements using lower transcutaneous electrode temperatures below 42°C.
METHODS: We enrolled 20 neonates. Two transcutaneous monitors were placed simultaneously on each neonate, with one electrode maintained at 42°C and the other randomized to temperatures of 38, 39, 40, 41, and 42°C. Arterial blood was collected twice at each temperature.
RESULTS: At the time of arterial blood sampling, values for transcutaneously measured partial pressure of CO2 (PtcCO2) were not significantly different among test temperatures. There was no evidence of skin burning at any temperature. For PtcCO2, Bland-Altman analyses of all test temperatures versus 42°C showed good precision and low bias. Transcutaneously measured partial pressure of O2 (PtcO2) values trended arterial values but had large negative bias.
CONCLUSION: Transcutaneous electrode temperatures as low as 38°C allow an assessment of PtcCO2 as accurate as that with electrodes at 42°C.
- transcutaneous monitoring
- carbon dioxide monitoring
- oxygen monitoring
- electrode temperature
- blood gas monitoring
Introduction
Extremes and fluctuations of arterial CO2 and O2 tension pose well-documented risks to the neonate, including periventricular leukomalacia,1–3 intraventricular hemorrhage,3–6 retinopathy of prematurity,7 bronchopulmonary dysplasia, and mortality.8–10
Therefore, these levels must be closely monitored. Arterial blood sampling is the accepted standard, but this provides only a single measurement of what is often a constantly changing clinical picture. It is also invasive by nature, requiring either an indwelling catheter or multiple needle sticks and can lead to anemia, which is an already prevalent issue in premature infants.
Transcutaneous monitors can provide continuous CO2 and O2 monitoring. One limitation to the widespread use of transcutaneous monitors in neonatal ICUs is the previously reported association of skin burning in extremely low birthweight infants due to the elevated electrode temperature required to achieve arterialization of the capillary bed, particularly in the measurement of PtcO2.11–13 High electrode temperature may not be needed for PtcCO2 assessment.
We hypothesized that a transcutaneous electrode temperature as low as 38°C would be accurate and precise in the measurement of CO2 levels. We sought (1) to assess the accuracy and precision of 4 lower electrode temperatures (38, 39, 40, 41°C) compared to the recommended temperature of 42°C in the measurement of CO2 levels in preterm infants between 1–2 kg; (2) to assess the accuracy and precision of all measured electrode temperatures (38–42°C) in the measurement of CO2 compared to the accepted standard, arterial blood gas PCO2; (3) to reassess the accuracy and precision of PtcO2 compared to PaO2 and assess PtcO2 at the 4 lower electrode temperatures (38, 39, 40, 41°C).
QUICK LOOK
Current knowledge
Transcutaneous monitoring provides continuous CO2 and O2 measurements without the need for blood sampling. Premature infants are at higher risk for complications secondary to abnormal CO2 and O2 levels, therefore continuous monitoring is vital. However, transcutaneous monitoring requires warming of the skin, which has been associated with skin injury in extremely low birthweight infants.
What this paper contributes to our knowledge
Transcutaneous temperatures as low as 38°C provide an accurate assessment of CO2 compared to the previously recommended temperature of 42°C. Use of lower transcutaneous temperatures to track CO2 trends should allow this technology to be applied in the smallest preterm infants. Transcutaneous monitoring may be useful for tracking O2 levels.
Methods
Subjects
This study was approved by the Institutional Review Board of the University of Arkansas for Medical Sciences. A total of 20 neonates were enrolled after written parental consent was obtained. Inclusion criteria included birthweight 1–2 kg, umbilical arterial access, and hematocrit levels > 35%. Exclusion criteria included pressor support, hematocrit levels < 35%, congenital anomalies, nitric oxide use, persistent pulmonary hypertension, and cyanotic heart disease.
Methods
The precision and accuracy of PtcCO2 and PtcO2 were examined at 5 different electrode temperatures (38, 39, 40, 41, and 42°C) in each infant. All newborns were monitored using 2 transcutaneous monitors applied simultaneously (TCM CombiM, Radiometer, Copenhagen, Denmark). This allowed for one electrode to remain unchanged at 42°C (ie, control) while the temperature of the other electrode was randomized to 38, 39, 40, 41, or 42°C (to assure accuracy of the control monitor). The transcutaneous electrodes were applied to the skin of the trunk as recommended by the manufacturer. The accepted standard to which the transcutaneous values were compared was the PaCO2 and PaO2. After stabilization of both monitors, 2 arterial blood gases were collected at each temperature, approximately 15 min apart from an indwelling umbilical arterial catheter (for a total of 10 blood gases). For the purposes of this project, stabilization was defined as a change of ≤ 2 mm Hg PtcCO2over 5 min. Approximately 0.2 mL of arterial blood was collected with each sample, which totaled 2 mL of blood per subject. Total blood collected per subject was recorded during the study. Arterial blood gas sampling and readings of the PtcCO2 and PtcO2 levels were performed simultaneously and recorded. After changing the electrode temperature, an appropriate amount of time was allowed for calibration (usually 5–10 min) and stabilization of the new electrode temperature (usually 10–15 min). Transcutaneous data were continuously recorded and downloaded from the monitor after the study. The collection of data, per subject, took approximately 4 hours.
Statistical Analysis
For a sample size estimate, with the design of 5 repeated measurements at each pair of temperatures, a compound symmetry structure was used, and we assumed every pair had the same correlation for within-subject covariance. A sample size of 20 neonates achieved 80% power to detect a minimum difference of 0.67, 0.97, and 1.2 per standard deviation, respectively, when the correlation between observations on the same subject was 0.1, 0.5, and 0.9. We used a repeated-measures design for this study, thereby requiring substantially fewer subjects and increasing trial feasibility compared to a 2-group or n-group comparison study.
Summary statistics such as mean and standard deviations were reported. Several comparisons were carried out for the main analysis. Comparison of average PtcCO2 or PtcO2 between each of the 5 test temperatures and the control temperature were analyzed using repeated-measures analysis of variance followed by Dunnett's method adjusting for multiple comparisons. Comparison of average PtcCO2 or PtcO2 levels between test temperatures and blood gas measurements, and average warm-up times for stabilization between test temperatures and the control temperature were analyzed by testing whether the difference was something other than zero, using the Wilcoxon signed-rank test due to the small sample size. For continuous data, the differences between test and control temperatures were assessed using generalized linear mixed-effect models.
For each neonate, measurements of PtcCO2 and PtcO2 between the 2 blood gas draws were used for analysis of agreement. Agreement between test and control temperatures, as well as test temperatures versus arterial measurements of PtcCO2 and PtcO2 was assessed using Bland-Altman plots. Limits of agreements and mean difference were reported. P values < .05 indicate statistical significance. All statistical analysis was performed using STATA 14.2 (StataCorp, College Station, Texas) or R v.3.3.2 (R Foundation for Statistical Computing, Vienna, Austria).
Results
Twenty infants were enrolled. Subject characteristics were (mean ± SD) gestational age 30 ± 2 wk, birthweight 1,449 ± 304 g, weight at study 1,378 ± 323 g, and age at study 2 ± 1 d. PtcCO2 levels of the control and test sites at 42°C were not different (45.5 ± 1.0 vs 45.7 ± 1.2 mm Hg, P = .87). However, PtcO2 levels of the control and test sites were different (57.9 ± 3.2 vs 47.6 ± 4.0, P = .02). At the time of blood gas analyses, the mean PtcCO2 among all test temperatures versus control (42°C) were not different, while for PtcO2 there were significant differences between test and control sites for all temperatures (Table 1). The differences between PtcCO2 and PaCO2 values were similar at all temperatures, while PtcO2 values varied markedly with large mean differences compared to blood gas values (Table 2).
Figure 1 shows continuous data for both PtcCO2 and PtcO2 at the lowest test temperature of 38°C versus control (42°C). For PtcCO2, while differences were statistically significant (longitudinal analysis, test of fixed effects, P = .005), they are not clinically important. For PtcO2, the difference between 42°C versus 38°C is quite large. For both PtcCO2 and PtcO2, results at 38°C aligned well with results at 42°C. Bland-Altman analysis of 38°C versus the control of 42°C showed an average bias of 2.2 for PtcCO2 at 38°C. Bias is defined as the tendency of a statistic to overestimate or underestimate a parameter. The 95% limit of agreement is between −7.04 and 11.44, which indicates poor precision (Fig. 2A). Precision refers to how close measurements from different samples are to each other. Results of other test temperatures versus control for PtcCO2 were similar (not shown). Analysis of PtcCO2 versus PCO2 at 38°C showed a positive bias of 4.85 and poor precision with a 95% limit of agreement between −2.69 and 13.39 (Fig. 2B). For PtcO2, 38°C versus 42°C (Fig. 3A), Bland-Altman analysis showed poor precision and a strong negative bias of −26.1 (95% limit of agreement between −48.51 and −3.61). Analysis of PO2 versus PtcO2 (Fig. 3B) showed a strongly negative bias of −31.85 and poor precision (95% limit of agreement between −64.05 and 0.35). Bland-Altman analysis was done for all test temperatures versus control as well as all temperatures versus PCO2 and PO2. Electrode temperature had minimal effect on bias and precision of PtcO2 measurement. However, results of PtcO2 measurements were similar with strong negative biases and poor precision at all test temperatures except for 42°C. At 42°C, there was a significantly lower bias and poor precision (Table 3). Time to warm-up for each temperature was not different, ranging from 10.5 to 12.3 min.
Discussion
Extremes and fluctuations in levels of CO2 and O2 are known to be dangerous and can result in morbidity and mortality for premature infants. The effects of hypercapnia (both permissive and incidental) have been associated with worse neurodevelopmental outcome.4,8–10 Hypercapnia is also associated with impaired cerebral autoregulation, leading to an increased incidence of intraventricular hemorrhage.5 In fact, extremes and fluctuations in both hypercapnia and hypocapnia are associated with severe intraventricular hemorrhage, especially in the first few days of life.6 Not only can over-ventilation damage the lung, but hypocapnia has also been found to cause or worsen cerebral ischemia1 and is known to be associated with periventricular leukomalacia.2,3 Extremes in O2 levels also have negative effects on the neonate. Hyperoxia is associated with periventricular leukomalacia and retinopathy of prematurity, while hypoxia has been shown to cause ischemia and is related to a higher chance of mortality.7
Transcutaneous monitoring provides a method to continuously monitor CO2 and O2, and in a recent Cochrane review, Bruschettini et al14 called for additional studies demonstrating safety and efficacy in transcutaneous monitoring. More studies are needed to evaluate long-term outcomes of the use of transcutaneous monitoring. There is evidence of earlier diagnosis of pneumothorax prior to decompensation as well as assessment for clinical disease severity in infants with bronchiolitis using continuous transcutaneous monitoring.15,16 Transcutaneous monitoring is not the only option for continuous monitoring of CO2 and O2, but at this time it is the most useful. End-tidal CO2 monitoring has been used, but there are documented disadvantages to end-tidal monitoring that make it less feasible in the neonate. In infants with lung disease, who comprise a large proportion of neonatal subjects, ventilation-perfusion mismatch can lead to inaccurate results. Most neonates have a high breathing frequency requiring a faster response time than end-tidal sensors can deliver. This population also has a lower tidal volume resulting in proportionally increased dead space of the sensors. In addition, end-tidal monitoring can underestimate CO2 levels.17–19 Monitoring end-tidal CO2 at the distal end of a double-lumen endotracheal tube may be useful, but these tubes are not always used for neonates.20 Given these issues, transcutaneous monitoring is the best option at this time for continuous CO2 monitoring.
To measure O2 saturation, the modality used most frequently is pulse oximetry (SpO2). SpO2 is widely available and easy to use, but it has weaknesses as well. Due to the O2-dissociation curve, SpO2 is unable to provide accurate estimates of PO2 levels at higher O2 saturation levels, which may be associated with very high PO2, which is harmful to the infant.21,22 Continuous accurate monitoring of PO2 could be useful to avoid hyperoxia and hypoxia. Infants with pulmonary hypertension and infants of extremely low birthweight are examples of subjects in whom very close monitoring of PaO2 levels could be especially useful. It may be of more limited use in extremely premature infants, given their lower oxygenation targets. In our study, wide differences in PtcO2 and PaO2 occurred, even at 42°C. As noted above, there were also electrode-site differences for PtcO2. Because arterialization of the capillary bed is so important for PtcO2 monitoring, perhaps a higher electrode temperature is required to accurately assess PtcO2 trends. Transcutaneous monitoring may have an important role in continuous PO2 assessment in larger infants who can tolerate higher electrode temperatures, and its use should perhaps be re-evaluated, particularly in the setting of pulmonary hypertension.
In our study, we assessed both PtcCO2 and PtcO2 at the recommended electrode temperature of 42°C and at several lower temperatures. We found that electrode temperatures as low as 38°C provide an accurate assessment of PCO2 and similar trends compared to the recommended electrode temperature of 42°C. In comparison to the accepted standard of arterial blood gas measurement, transcutaneous electrode temperatures were accurate at every test temperature (38, 39, 40, 41, and 42°C), with mean differences from arterial values ranging from 2 to 8 mm Hg.
Using an older transcutaneous monitor, Sorensen et al13 found that electrode temperatures of 40 and 41°C could be used to track PtcCO2; they recommend, however, using a correction factor of 12–15%. More recently, Hirata et al23 found results similar to ours with electrode temperatures of 38, 39, and 40°C in comparison to a control of 42°C, but they recommended a correction factor of 6 mm Hg. Hirata et al,23 however, did not use 2 monitors simultaneously, which is a major strength of our study. Given our average difference of 2.2 mm Hg between 38°C and 42°C, we do not feel a correction factor is needed, although occasional correlation with an arterial gas must be done. It is important to note that the limit of agreement in the neonate between arterial and transcutaneous values may be more dependent on skin perfusion due to blood flow changes in the neonatal skin. The trend in any one infant is more important than the absolute difference between the transcutaneous reading and the arterial value, therefore it may be more important to be aware of trends than to use a correction factor.
For PtcO2, transcutaneous monitoring has been associated with skin burns as well as measurement inaccuracies at lower electrode temperatures. For accurate measurement of PO2 trends, transcutaneous monitoring requires an electrode temperature sufficient for arterialization of the capillary bed.13,24 In our study we reassessed PtcO2 using newer electrode technology recently developed by Radiometer (Copenhagen, Denmark). We tested this technology to assess any improvements in PtcO2 measurements. The new sensors reportedly have a very small surface and a site time-heat function monitor that ensures that the heating of the sensor is automatically switched off after a monitoring session, theoretically decreasing the risk of skin damage. We found that while PtcO2 values were much lower than arterial values, especially at low electrode temperatures, PtcO2 appeared to track the higher electrode temperature results relatively well. Thus, following both PtcO2 and PtcCO2 may be clinically useful, even when low electrode temperatures are employed.
Limitations
This was a small study with an enrollment of 20 subjects. We did not include babies with birthweight of < 1 kg due to the potential of skin burns in this population at the higher electrode temperatures. The majority of our infants required only minimal respiratory support, and the study was short in duration, lasting approximately 4–5 h. It is also worth noting that, in general, the monitor performed with limited precision, likely due to changes in perfusion. For this reason, we recommend occasional correlation with an arterial blood gas.
Conclusions
Continuous CO2 and O2 monitoring are of utmost importance in extremely premature infants. Our results support and expand current literature. Use of transcutaneous monitor temperatures as low as 38°C allow accurate monitoring and tracking of PCO2 and therefore can be used instead of the recommended 42°C. For PO2, transcutaneous monitoring may be useful for trending, although caution must be used in the interpretation of PtcO2 values. Use of lower transcutaneous temperatures to track PCO2 values should allow this technology to be applied in the smallest preterm infants.
Footnotes
- Correspondence: Jessica F Jakubowicz MD, University of Arkansas for Medical Sciences, 4301 W Markham St, Slot 512–5B, Little Rock, AR 72205. E-mail: jmjakubowicz{at}uams.edu.
The authors report partial funding for this project by an intramural grant from the UAMS College of Medicine Children's University Medical Group Grant Program, grant #2822-3. Transcutaneous monitors were provided for this study by Radiometer, which did not have any participation in study design, data collection or statistical analysis.
Dr Jakubowicz presented a version of this article at the Pediatric Academic Societies Meeting held May 6, 2017, in San Francisco, California.
- Copyright © 2018 by Daedalus Enterprises
References
- 1.
- 2.
- 3.
- 4.
- 5.
- 6.
- 7.
- 8.
- 9.
- 10.
- 11.
- 12.
- 13.
- 14.
- 15.
- 16.
- 17.
- 18.
- 19.
- 20.
- 21.
- 22.
- 23.
- 24.