ABSTRACT

Objects: The purpose of this study was to observe a correlation between P_ETCO₂ and PaCO₂ in intubated neonates under intermittent mandatory ventilation with spontaneous breathing. Material and methods: A total of 55 paired P_ETCO₂ measured by mainstream capnometry and PaCO₂ values were obtained from 4 intubated neonates in our neonatal intensive care units at Nagano Children’s Hospital, Nagano, Japan. Results: P_ETCO₂ and PaCO₂ were significantly correlated (r² = 0.928, p < 0.0001). For samples in ventilated neonates with spontaneous breathing, maximum P_ETCO₂ and mean PETCO2 correlated strongly with PaCO₂ (maximum P_ETCO2: r² = 0.9401, p< 0.0001; mean P_ETCO₂: r² = 0.8587, p < 0.0001). Although PaCO₂ also correlated with minimum P_ETCO₂ (r² = 0.2884, p < 0.01) in ventilated infants with spontaneous breathing, a significant difference was seen with maximum P_ETCO₂ (p < 0.05) and mean P_ETCO₂ (p < 0.05) in the correlation coefficient r between PaCO₂ and P_ETCO₂. Conclusion: Present study showed that a good correlation exists between P_ETCO₂ and PaCO₂ in intubated neonates under intermittent mandatory ventilation with spontaneous breathing. Lightweight with low amounts of dead space mainstream capnometry can be used as noninvasive monitor in incubated neonates with spontaneous breathing.

1. INTRODUCTION

Capnography, which displays the level and waveform of CO₂ in exhaled air, is a simple technique that appears to accurately indicate arterial PCO₂ (PaCO₂) and provides information on cell metabolism, blood perfusion, and alveolar ventilation [1-3]. The use of end-tidal CO₂ (P_ETCO₂) for monitoring and as a tool for verifying endotracheal tube (ETT) position is another common practice in the operating room and in adult and pediatric intensive care units [3]. This procedure was also introduced to neonatal intensive care units (NICUs), but there are limitations of the technique in smaller babies, especially for extremely low birth weight infants due to issues such as the weight of sensors or water droplets within circuits, dead space, and leakage from tracheal intubation tubes. Recently, a lightweight mainstream capnometer was developed. We have previously reported a strong correlation between P_ETCO₂ and PaCO₂ under controlled ventilation when tidal volume/body weight (TV/BW) was 6 - 15 mL/kg and the leakage rate was <60% in rabbits. Furthermore, under conditions of 6 mL/kg of tidal volume (TV), PaCO₂ was significantly increased by a dead space increase of only 1 mL, representing >7% of TV [4].

Capnometry is expected as one of the non-invasive monitor in NICUs. Although several investigators have demonstrated good relationships between values for P_ETCO₂ and PaCO₂ in infants [5-8], the value of capnometry in estimating PaCO₂ has been questioned during anesthesia with spontaneous ventilation [9,10]. On the other hand, while TV is larger than spontaneous breaths, P_ETCO₂ is close to the PaCO₂ and to that observed with voluntary maximal expiration [11]. The purpose of this study was to observe a correlation between P_ETCO₂ and PaCO₂ in intubated neonates under intermittent mandatory ventilation (IMV) with spontaneous breathing.

2. MATERIAL AND METHODS

A total of 55 paired P_ETCO₂ and PaCO₂ values were obtained from the 4 neonates who had been admitted to the NICU at Nagano Children’s Hospital between April and July 2009. Neonates deemed as non-viable by the attending physician were excluded, as were those with conditions such as acute shock, infection, or hemodynamic instability. Before enrollment into this study, informed consent was obtained from the parents or guardian of each infant. Study subjects comprised 4 neonates. Mean (±standard deviation (SD)) gestational age and birth weight were 36.7 ± 2.1 weeks and 2446 ± 487 g, respectively. Data on demographics, clinical characteristics, and laboratory findings for subjects were collected by referring to the clinical laboratory records.

The infants were ventilated mechanically using a timecycled pressure-limited ventilator (Calliope^®; Metran, Saitama, Japan). Peak inspiratory pressure (PIP), fraction of inhaled oxygen (FiO₂), inspiratory time, positive end expiratory pressure (PEEP) and respiratory rate were settled to provide the optimal arterial PaO₂ and PaCO₂ as defined by the neonatologists and were not manipulated for the purposes of the study.

Mainstream P_ETCO₂ was measured via a capnograph connected to the proximal end of the endotracheal tube (Cap-One^®; TG-970P, Nihon-Kohden, Tokyo, Japan). Data was continuously recorded on a laptop computer using the software programmed by LabVIEW (National Instruments, Texas, USA) through CO₂ monitor (OLG-2800; Nihon-Kohden, Tokyo, Japan) each patients. We distinguished between spontaneous breaths and ventilator breaths on the basis of capnography for 20 s at the time of blood gas analysis. For each 20-s sample period, we determined the maximum, mean and minimum values of P_ETCO₂ (Figure 1). Measurements of P_ETCO₂ that did not show an alveolar plateau due to a large amount of leakage were excluded.

The tidal volume was measured by mainstream capnography (CO2SMO 8100, Fukuda Denshi, Tokyo, Japan). The leakage ratio was calculated using the following equation:

Blood samples were drawn from indwelling arterial lines into a 0.1-mL heparinized syringe to prevent coagulation. Blood sampling was performed by heel puncture when arterial line was not placed. Measurements were then immediately made using a bedside blood gas analyzer (ABL 700; Radiometer, Copenhagen, Denmark) for PaCO₂. Blood gas analysis was performed for the purposes of evaluation of the patient (including PaO₂, PaCO₂, electrolytes or lactate, etc.) only. Calibrations were performed automatically for the blood gas analyzer and the accuracy of the capnography was checked by 5% CO₂ gas cylinder.

All statistical analyses were conducted using SPSS Statistics version 17.0 (SPSS, Chicago, Illinois). To determine whether P_ETCO₂ were representative of PaCO₂, the relationship between P_ETCO₂ and PaCO₂ was analyzed by simple linear regression. The standard technique of Fisher’s Z transformation was performed to determine whether a significant difference existed in the correlation coefficient r between PaCO₂ and each group of P_ETCO₂.

Furthermore, Bland-Altman plots were performed to assess measurements of P_ETCO₂. Bland-Altman plots demonstrate “good agreement” not only when differences between methods are consistent across all measurements but also when the differences are small. In a situation in which the difference between measurements is expected to change based on a third variable, Bland-Altman plots lose importance. Precision of P_ETCO₂ and the agreement between P_ETCO₂ and PaCO₂ were assessed by bias, SD and calculating the 95% confidence interval (CI) for the bias (bias = P_ETCO₂ – PaCO₂). Values of p < 0.05 were determined to be significant.

This study was carried out under the control of the Ethics Committee of Medicine and Medical Care, Nagano Children’s Hospital, Nagano, Japan.

3. RESULTS

Mean TV/BW and leakage ratio were 7.3 ± 1.7 mL/kg and 9.5% ± 12.0%, respectively (Table 1). All patients were treated using sedative drugs.

P_ETCO₂ and PaCO₂ were significantly correlated (r² = 0.928, p < 0.0001). In the Bland-Altman plot test, the mean difference (bias) and SD of the differences for

P_ETCO₂ was –0.88 ± 2.69 mmHg (95% CI for the bias, −1.61 to –0.16 mmHg) (Figure 2). We chose the maximum for P_ETCO₂ on the basis of capnograms for each 20-s period at the time of blood gas analysis.

Due to breath-to-breath variation, we evaluated three measurements of P_ETCO₂ to determine which one most consistently and accurately predicted PaCO₂. From 55 measurements, we have selected 24 paired P_ETCO₂ and PaCO₂ values which were obtained at the time when spontaneous breathing was present.

For samples in ventilated infants with spontaneous breathing, maximum P_ETCO₂ and mean P_ETCO₂ correlated strongly with PaCO₂ (maximum P_ETCO₂: r² = 0.9401, p < 0.0001; mean P_ETCO₂: r² = 0.8587, p < 0.0001). Although PaCO₂ also correlated with minimum P_ETCO₂ (r² = 0.2884, p < 0.01) in ventilated infants with spontaneous breathing, a significant difference was seen with maximum P_ETCO₂ (p < 0.05) and mean P_ETCO₂ (p < 0.05) in the correlation coefficient r between PaCO₂ and P_ETCO₂ (Figures 3(a)-(c)).

Bland-Altman analysis showed that P_ETCO₂ underestimated PaCO₂ by a mean difference (bias) of –0.175 ± 2.31 mmHg (95% CI for the bias, –1.15 to 0.799 mmHg) in the maximum P_ETCO₂, –5.01 ± 3.55 mmHg (95% CI for the bias, –6.50 to –3.51 mmHg) in mean P_ETCO₂, and –14.6 ± 8.82 mmHg (95%CI for the bias, –18.3 to –10.9 mmHg) in minimum P_ETCO₂ (Figures 3(d)-(f)).

4. DISCUSSION

Advances in the treatment of neonatal respiratory failure, including exogenous surfactant [12,13], inhaled nitric oxide (iNO) [14,15], and a growing repertoire of assisted ventilation strategies [16] have decreased morbidity and mortality rates. Patient monitoring has played a critical role in the safe and effective application of these advanced therapies.

Pulse oximetry provides a noninvasive method of assessing oxygenation and continuous surveillance of the partial pressure of arterial oxygen [17]. Maintaining PaCO₂ within the desired range by frequent arterial sampling can increase the need for multiple transfusions in the NICU [18], highlighting the need for methods of continuous non-invasive monitoring of CO₂ levels. Both hypocarbia and hypercarbia are detrimental to extremely low birth weight infants and have been implicated as

causative factors in periventricular leukomalacia, intraventricular hemorrhage and chronic lung disease [19-21]. Critical event analyses have documented that hypoxemia secondary to depressed respiratory activity is a principal risk factor for near misses and death [22,23].

Monitoring of P_ETCO₂ is a simple and noninvasive technique that appears to accurately indicate PaCO₂ in a variety of clinical situations [1,24]. However, levels of P_ETCO₂ and PaCO₂ depend on ventilation, cardiac output, CO₂ output, and pulmonary function; a change in any of these will cause a change in P_ETCO₂ [25]. For instance, a growing degree of difference between P_ETCO₂ and PaCO₂ can indicate the severity of pulmonary embolism [26] or even the effects of thrombolytic therapy [27].

P_ETCO₂ varied appreciably from breath to breath. The reliability of this value under conditions of significant ventilation perfusion inequality or heterogeneous tidal volumes has thus been questioned [28]. In many cases, spontaneous breaths of variable tidal volumes far outnumbered ventilator breaths, but still contributed relatively little to alveolar minute ventilation. We found that maximum P_ETCO₂ during each sampling period showed the best correlation with PaCO₂. A number of investigators have suggested that larger tidal volumes are necessary to measure P_ETCO₂ accurately, as small (e.g., spontaneous) breaths may fail to “wash out” the anatomic dead space [11,29].

Weinger et al. [30] also found a wide range in differences between P_ETCO₂ and PaCO₂ over time. However, they measured mean peak P_ETCO₂ over a period of 8 min. Although averaging maximal P_ETCO₂ values over a longer time span might improve the stability and reliability of PaCO₂ - P_ETCO₂, assessing the respiratory status to measure P_ETCO₂ over a long time would be difficult. The respiratory apparatus might change second to second.

Takano et al. [31] reported that reliable P_ETCO₂ was obtained when a vital capacity maneuver was performed on each nonintubeted patient, indicating that full expiration to the maximal expiratory position is necessary for precise estimation of PaCO₂.

Comparison with spontaneous breathing, end tidal CO₂ measured from ventilated breath is close to the PaCO₂ and to that observed with a voluntary maximal expiration. This PCO₂ gradient between ventilator and spontaneous breathing indicates a large dead-space to tidal volume ratio; much of the expired CO₂ appearing with spontaneous breathing is diluted, with dead space air lowering the concentration at any point during expiration.

Although our prospective observational study revealed a good correlation and agreement between PaCO₂ and maximum P_ETCO₂, the present study included only a very small number of participants, and the conditions of patients were not constant. Furthermore, in children with congenital cyanotic heart disease, right-to-left intracardiac shunting reportedly causes an obligatory difference between PaCO₂ and P_ETCO₂ [32,33]. Our subjects included patients with congenital heart disease because they showed stable cardiorespiratory status in this study.

This study consisted of only a small number of participants, the statistical implications of repeated measurements in the same infants, and the conditions of the cases were not constant. In addition, we did not adjust for cardiac output their measurements, though differences in cardiac output are known to affect P_ETCO₂ measurements [25]. Therefore, we would like to study these issues in the future.

5. CONCLUSION

Our results indicate that a good correlation exists between P_ETCO₂ and PaCO₂ in intubated neonates under intermittent mandatory ventilation with spontaneous breathing. Furthermore, maximum P_ETCO₂ correlated strongly with PaCO₂ compared to minimum P_ETCO₂. Lightweight with low amounts of dead space mainstream capnometry can be used as noninvasive monitor in incubated neonates with spontaneous breathing.

6. ACKNOWLEDGEMENTS

The authors acknowledge the help and support of Mr. Dainobu, Mr. Inoue, and Mr. Yamamori from Nihon Kohden.

REFERENCES

NOTES

^*Competing interests: The authors have no competing interests to declare.