Abstract

Objective: To study the correlation and agreement between end-tidal carbon dioxide (EtCO₂) and arterial carbon dioxide (PaCO₂) in ventilated extremely low birth weight (ELBW) infants in the first week of life. Methods: Retrospective chart review of all ELBW (<1000 g) infants admitted to a level III NICU from January 2003 to December 2003. Data collected included demographic details and simultaneous EtCO₂ (mainstream capnography) and arterial blood gas values (pH, PaCO2, PaO2). Outcome: The correlation coefficient, degree of bias with 95% confidence interval between the EtCO₂ and PaCO₂. Results: There were 754 end-tidal and arterial CO₂ pairs from 31 ELBW infants (21 male and 10 female). The overall EtCO2 values were significantly lower than PaCO₂ value (EtCO₂=37.3 ± 7.8 mmHg Vs PaCO₂=43.5 ± 9.6 mmHg, P<0.0001). In only 89/754(11.8%) pairs, the EtCO2 was higher than the PaCO₂. The overall bias was 5.6 ± 6.9 mmHg (95% C.I. 5.11-6.09). The intraclass correlation coefficient was 0.81. Using EtCO₂ ranges of 30 to 50 mmHg, the capnographic method was able to identify 84% of instances where PaCO₂ was between 35 (<35 = hypocarbia) and 55 mmHg (>55= hypercapnia). Conclusions: There is good correlation and agreement between end-tidal CO₂ and arterial CO₂ in ELBW infants in the EtCO₂ range 30-50 mmHg. End-tidal CO₂ monitoring can be helpful in trending or for screening abnormal PaCO₂ values in ELBW infants in first week of life.

Key words: Capnography, End-tidal carbon dioxide, ELBW, PaCO₂.

Arterial blood gas analysis is essential for monitoring the adequacy of ventilation and oxygenation in ventilated babies. It is often difficult to obtain arterial blood gases frequently especially where resources are limited. Arterial blood sampling leads to blood loss, pain, stress and arteriospasm in neonates and hence need for non invasive monitoring. Pulse oximetry provides the non-invasive method of assessing the oxygenation in neonatal intensive care unit (NICU) patients(1) It is necessary to have a reliable, non-invasive method for detecting hypocarbia and hypercarbia. Both hypocarbia and hypercarbia are detrimental to the ELBW infants and have been implicated as a causative factor in periventricular leukomalacia (PVL), intra-ventricular bleed (IVH) and chronic lung disease (CLD)(2-7). Maintaining the PaCO₂ within the desired range by frequent arterial sampling can increase the need for multiple transfusions in these infants(8). Transcutaneous CO₂ monitoring is a non-invasive technique for measuring the carbon dioxide(9). However, transcutaneous CO₂ monitoring is not tolerated in VLBW infants because of their fragile skin and is affected by acidosis and hypoxia(10).

End-tidal carbon dioxide (EtCO₂) measurement by capnography is a continuous and non-invasive indirect measurement of blood carbon dioxide tensions. Capnography is done by either side-stream or mainstream gas sampling. The technique, which measures the concentration of CO₂ in exhaled gas, has been used extensively in adults and children(11,12) The reliability of mainstream capnography has also been reported in term and preterm neonates(13-15). The EtCO₂ monitoring has some clear advantages over the transcutaneous CO₂ monitoring, such as a much faster response time to changes in blood CO₂ levels, internal calibrating ability and no thermal injury to the fragile skin of the newborn(16). This noninvasive method can reduce the blood loss and pain associated with arterial blood gas sampling. This study was done to find out the correlation and agreement between EtCO₂ and arterial CO₂ in ventilated ELBW infants in first week of life.

This study is based on a retrospective chart review of all ELBW babies admitted in a level III NICU in Southern Alberta over one year period (January 2003 to December 2003). The data are derived from a quality improvement study to audit the usefulness of EtCO₂ monitoring in the ELBW babies. ELBW babies (<8 days) receiving mechanical ventilation who had indwelling arterial catheter and end- tidal CO₂ monitor by mainstream capnography were included in the study. Babies with structural cardiopulmonary malformation and those ventilated on high frequency ventilation were excluded from the study. Demographic details, gestational age, birth weight, details of surfactant therapy, simultaneous EtCO₂ and arterial blood gas values (pH, PaCO₂, PaO₂) along with ventilatory settings were retrieved from the case record.

Exhaled CO₂ was continuously monitored by using a commercially available mainstream capnography (Capnogard ETCO₂ monitor, Model 1265, Novametrix Medical Systems Inc, Wallingford, Connecticut, USA). This device measures changes in the CO2 concentration of the infants exhaled tidal volume, which is passing through an airway adapter containing capnostat sensor. Infrared light is generated in one leg of the U shaped sensor and then beamed through the windows of the airway adapter to a detector in the other leg of the sensor. The adapter (< 0.5 mL dead space) is placed just proximal to the endotracheal tube. The capnograph was calibrated everyday according to the manu-facturer’s recommendations. The arterial blood gas analysis was performed on blood obtained from the indwelling arterial catheters via umbilical or radial arteries, using an automatic blood gas analyzer (Radiometer ABL 725 Operation) that was calibrated daily.

To determine whether EtCO₂ were representative of PaCO₂, the relationship between the end-tidal and arterial CO₂ was analyzed by simple linear regression. The precision of the EtCO₂ and the agreement between EtCO₂ and PaCO₂ (gold standard) were assessed by bias, standard deviation and calculating the 95% confidence interval (CI) for the bias (bias = PaCO₂-EtCO₂)(17). Data were correlated using intraclass correlation coefficient. Differences between EtCO₂ and PaCO₂ were compared by using paired "t" test. Values were express as mean ± SD of the mean (and range when appropriate). A statistically significant difference was defined as p <0.05.

Thirty-one ELBW infants (21 boys and 10 girls) were included in the study. The mean birth weight was 740 ± 130 g (range 500- 970 g) and the mean gestational age was 25 ± 1 weeks (range 23-27 wk). All infants had hyaline membrane disease and treated with surfactant. All infants were ventilated using pressure-limited, time-cycled ventilators in a variety of modes, including intermittent mandatory ventilation and synchronized intermittent mandatory ventilation (Infant star and Drager Babylog 8000 ventilators).

A total of 754 EtCO₂-PaCO₂ pairs were available for analysis from 31 infants. The overall EtCO₂ was significantly lower than the corresponding PaCO₂ values (EtCO₂ = 37.3 ± 7.8 mmHg vs PaCO₂= 43.5 ± 9.6 mmHg respectively, P <0.001). In 89/754 (11.8%) pairs, the EtCO₂ was higher than the PaCO₂. Figure 1 shows the scattergram of the end-tidal CO₂ and PaCO₂. Pearson correlation coefficient was 0.71.The overall EtCO₂ bias was 5.6 ± 6.8 mmHg with a 95% confidence interval for mean of 5.11-6.09. The bias in the relationship between the EtCO₂ and PaCO₂ is illustrated in Fig. 2 where the difference is plotted against the average of the two values (Bland-Altman Plot). There was a good agreement between EtCO₂ and PaCO₂ with the intraclass correlation coefficient of 0.81 (P <0.0001).

Fig. 1. Scatter gram of end-tidal CO2 and PaCO2 relationship. End-tidal CO2 and PaCO2 are presented in mmHg.

Fig. 2. Bland-Altman plot of the difference between the end-tidal and arterial CO2 versus the average of the two simultaneous readings.

Using end-tidal CO₂ ranges of 30 to 50 mmHg, the capnographic method was able to identify 84% of instances where PaCO₂ was between 35 (<35 = hypocarbia) and 55 mmHg (>55 = hypercapnia). When the end-tidal CO₂ is beyond the limit i.e., EtCO₂ <30 mmHg (n = 87) or >50 mmHg (n = 49), capnographic method was able to identify 78/136 (57.3%) of instances where PaCO₂ was <35 mmHg (51/87) and PaCO₂ >55 mmHg (27/49).

This study showed that there is a good correlation and agreement between EtCO₂ and PaCO₂ in surfactant treated, ventilated ELBW babies during the first week of life. Although arterial blood gas analysis remains the most accurate method of ascertaining blood CO₂ tensions, alternative non-invasive methods are crucial and desirable because these infants have limited blood volumes and frequent blood gas sampling contributes to the high transfusion rates seen in ELBW infants. The need to quickly assess and respond to changes in CO₂ pressures has become more critical with the increased recognition that both low and high PaCO₂ for even brief periods of time are associated with long term morbidity in these patients.

Rozycki, et al.(13) and Wu, et al.(14) reported a good correlation between the EtCO₂ and PaCO₂ in neonates including preterm babies. These authors suggested that EtCO₂ monitoring by mainstream capnography was accurate method for trending and screening infants with abnormal PaCO₂. In the pre surfactant era, EtCO₂ monitoring was reported to be unreliable for estimating PaCO₂ in preterm babies with respiratory distress(15). Nangia, et al.(15) also reported similar findings in preterm babies with hyaline membrane disease. The poor correlation may be due to small sample size, different devices for exhaled CO₂ monitor and use of side stream capnography(15,16). Other published studies have shown that factors such as poor cardiac output, hypothermia, size of tidal volume and lung disease (especially if an increased in dead space is present) adversely affect the EtCO₂-PaCO₂ relation-ship(16,18,19)

The EtCO₂ is usually lower compared to corresponding PaCO₂. The consistent bias of the EtCO₂ value in this study is within the range reported in previous studies of main stream capnometry in neonates(13,14). The lower value for the end-tidal measurement may be attributable to gas mixing proximal to the endotracheal tube. End-tidal carbon dioxide measurement from gas sampled distally to the endotracheal tube-ventilator connection more closely matches the arterial value(20,21). The severity of lung disease, the ventilation index and the oxygenation index had negligible influences on the degree of bias(13).

The normal range of carbondioxide to be maintained during assisted ventilation is 35 mmHg to 55 mmHg. Hypocapnia and hypercapnia are harmful to the brains of preterm and term infants. There is a strong association between PaCO₂ values less than 25 to 30 mmHg and an increase incidence of cystic periventricular leukomalacia, periven-tricular echo densities or chronic lung disease in preterm infants(2-7). The upper limit of benign PaCO₂ levels is less clear. In a randomized trial of permissive hypercapnia, mild hypercapnia (PaCO₂ 45-55 mmHg) did not increase the incidence of IVH in VLBW infants(22). However, severe hypercapnia (PaCO₂ >60 mmHg) was associated with intraventricular hemorrhage in a large retrospective study(6).

The goal of mechanical ventilation should be avoiding hypocapnia or hypercapnia, rather than achieving a specific level or range of PaCO₂. Hence, the continuous noninvasive monitoring of CO₂ by capnometry is helpful in these vulnerable preterm infants. Based on this approach, we have demonstrated that the mainstream end-tidal CO₂ monitor will indicate that the PaCO₂ is within prescribed parameters (PaCO₂ between 35 and 55 mm Hg) 84% of the time.

Since this study consisted of retrospective data collection, data regarding the waveform, its characteristics or changes in the waveform were not recorded in this study. However, it is not known in the literature that the addition of wave form analysis to the EtCO₂ value would improve the ability of the end-tidal CO₂ monitor to act as a warning device that the PaCO₂ is approaching undesirable levels. Other limitations is that when the end-tidal monitor was outside the prescribed limits (EtCO₂ <30 or >50 mmHg), the PaCO₂ was truly in the hypocapnia (PaCO₂ <35 mm Hg, 51/87) or hypercapnia (PaCO₂ >55 mm Hg, 27/49) range in 78/136 instances (57.3%). On the other hand, out of 640 instances where PaCO₂ was >35 mmHg, there were only 31 instances where EtCO₂ value was <30 mmHg. [i.e. specificity 609/640 (90%)].

In summary, we found good correlation and agreement between EtCO₂ and PaCO₂ in ventilated ELBW infants during the first week of life. We suggest using the EtCO₂ measurement to adjust ventilator settings when the monitor is within the range of 30 to 50 mmHg but recommend periodic check of blood gases to make sure it is within normal range of PaCO₂, and to obtain arterial blood gases when the end-tidal monitor shows readings beyond these limits. This protocol may reduce the amount of barotrauma by allowing for more prompt adjustment of the ventilatory settings in response to changes in PaCO₂. It may also reduce the exposure time to hypocapnia and hypercapnia and therefore decrease the brain injury; and finally it may reduce the number and total volume of blood samples required for gas measurement.

Contributors: SAS plan the study, collected the data and drafted the manuscript. NS supervised the drafting of the manuscript. SAS would act as the guarantor of the paper .

Funding: None.

Competing interests: None.

1. Dziedzec K, Vidyasagar D. Pulse oximetry in neonatal intensive care. Clin Perinatol 1989; 16: 177-197

2. Giannakopoulou C, Korakaki E, Manoura A, Bikouvarakis S, Papageorgiou M, Gourgiotis D, et al. Significance of hypocarbia in the development of periventricular leukomalacia in preterm infants. Pediatr Int 2004; 46: 268-273.

3. Erickson SJ, Grauaug A, Gurrin L, Swaminathan M. Hypocarbia in the ventilated preterm infant and its effect on intraventricular hemorrhage and bronchopulmonary dysplasia J Paediatr Child Health 2002; 38: 560-562.

4. Ambalavanan N, Carlo WA. Hypocapnia and hypercapnia in respiratory management of newborn infants. Clin Perinatol 2001; 28: 517-531.

5. Okumura A, Hayakawa F, Kato T, Itomi K, Maruyama K, Ishihara N, et al. Hypocarbia in preterm infants with periventricular leuko-malacia: the relation between hypocarbia and mechanical ventilation. Pediatrics 2001; 107:469-475.

6. Wallin LA, Rosenfeld CR, Laptook AR, Maravilla AM, Strand C, Campbell N, et al. Neonatal intracranial hemorrhage: II. Risk factor analysis in an inborn population. Early Hum Dev 1990; 23:,129-137.

7. Wiswell TE, Graziani LJ, Kornhauser MS, Stanley C, Merton DA, McKee L, et al. Effects of hypocarbia on the development of cystic periventricular leukomalacia in premature infants treated with high-frequency jet ventilation. Pediatrics 1996; 98: 918-924.

8. Strauss RG. Transfusion therapy in neonates. Am J Dis Child 1991; 145: 904-911.

9. Rennie JM. Transcutaneous carbon dioxide monitoring. Arch Dis Child 1990; 65: 345-346

10. Bhat R, Kim WD, Shukla A, Vidyasagar D. Simultaneous tissue pH and transcutaneous carbon dioxide monitoring in critically ill neonates. Crit Care Med 1981; 9: 744-749.

11. Campbell FA, McLeod ME, Bissonnette B, Swartz JS. End-tidal carbon dioxide measurement in infants and children during and after general anaesthesia. Can J Anaesth 1994; 41: 107-110.

12. McDonald MJ, Montgomery VL, Cerrito PB, Parrish CJ, Boland KA, Sullivan JE. Comparison of end-tidal CO2 and PaCO2 in children receiving mechanical ventilation. Pediatr Crit Care Med 2002; 3: 244-249.

13. Rozycki HJ, Sysyn GD, Marshall MK, Malloy R, Wiswell TE. Mainstream end-tidal carbon dioxide monitoring in the neonatal intensive care unit. Pediatrics 1998; 101: 648-653.

14. Wu CH, Chou HC, Hsieh WS, Chen WK, Huang PY, Tsao PN. Good estimation of arterial carbon dioxide by end-tidal carbon dioxide monitoring in the neonatal intensive care unit. Pediatr Pulmonol 2003; 35: 292-295.

15. Nangia S, Saili A, Dutta AK. End tidal carbon dioxide monitoring– its reliability in neonates. Indian J Pediatr 1997; 64: 389-394.

16. Watkins AM, Weindling AM. Monitoring of end tidal CO2 in neonatal intensive care. Arch Dis Child 1987; 62: 837-839.

17. Bland JM, Altman DG. Statistical methods for assessing agreement between two methods of clinical measurement. Lancet 1986; 1: 307-310.

18. Weinger MB, Brimm JE. End tidal carbon dioxide as a measure of arterial carbon dioxide during intermittent mandatory ventilation. J Clin Monit 1987; 3: 73-79.

19. Sharma SK, McGuire GP, Cruise CJE. Stability of the arterial to end tidal carbon dioxide difference during anesthesia for prolonged neurosurgical procedures. Can J Anaesth 1995; 42: 498-503.

20. Hillier SC, Badgwell JM, McLeod ME, Creighton RE, Lerman J. Accuracy of end-tidal PCO2 measurements using a side stream capnometer in infants and children ventilated with the Sechrist infant ventilator. Can J Anaesth 1990; 37: 318-321.

21. Rich GF, Sconzo JM. Continuous end-tidal sampling within the proximal endotracheal tube estimates arterial CO2 tension in infants. Can J Anaesth 1991; 38: 201-203.

22. Mariani G, Cifuentes J, Carlo WA. Randomized trial of permissive hypercapnia in preterm infants. Pediatrics 1999; 104: 1082-1088.