The PaCO 2 -ETCO 2 gradient in pre-hospital intubations of all aetiologies from a single UK helicopter emergency medicine service 2015

Abstract

Background

Control of the arterial partial pressure of carbon dioxide (PaCO₂) is important in the ventilated patient. End-tidal carbon dioxide (ETCO₂) levels are often used as a proxy, but are clinically limited. The difference between the PaCO₂ and ETCO₂ has been suggested to be 0.5–1.0 kPa. However, this has not been consistently reflected in the physiologically unstable pre-hospital patient. This study aims to elucidate the PaCO₂-ETCO₂ gradient for pre-hospital intubated patients.

Methods

This was a retrospective, cohort study using data identified from the HEMSbase 2 database (Feb 2015–Nov 2018). Patients were included if they had documented ETCO₂ and arterial PaCO₂ measurements. Arterial PaCO₂ data that could not be linked to within 5 minutes of ETCO₂ were excluded. Bland-Altman plots were calculated to describe agreement.

Results

A total of 73 patients were identified. Aetiology was arranged into three categories: 13 (17.8%) medical, 22 (30.1%) traumatic and 38 (52.1%) out-of-hospital cardiac arrest (OHCA). The median PaCO₂-ETCO₂ gradient was 2.0 [1.3–3.1] kPa. A PaCO₂-ETCO₂ gradient of 0–1 kPa was seen for only 11 (15.1%) of total patients. The Bland-Altman agreement for all aetiologies was more than the accepted gradient of 0-1 kPa with the largest bias and widest limits of agreement seen for OHCA (–3.2 [0.3 – –6.8]).

Conclusion

The magnitude of the differences between the ETCO₂ and PaCO₂, levels of variation and inability to predict this suggest that ETCO₂ is not a suitable surrogate upon which to base ventilatory settings in conditions where pH or PaCO₂ require precise control.

Keywords

Intubation capnography mechanical ventilation out-of-hospital cardiac arrest blood gas analysis

Introduction

Capnography provides a non-invasive and continuous assurance of tracheal tube placement.^1–3 End tidal carbon dioxide (ETCO₂) also contributes to circulatory assessment,^1–3 particularly in cardiac arrest⁴,⁵ and has a role in both metabolic^6–9 and respiratory monitoring.¹⁰,¹¹ Mechanical ventilation enables clinicians to control the arterial partial pressure of carbon dioxide (PaCO₂) and consequently arterial pH with potential physiological benefits across a range of illness. Hypercapnia (PaCO₂ >6.0 kilopascals, kPa) causes acidosis and cerebral vasodilation, so increasing intracranial pressure.¹²,¹³ Conversely, hypocapnia (PaCO₂ <4.0 kPa) causes alkalosis and cerebral vasoconstriction, which may lead to cerebral ischaemia.¹²,¹⁴,¹⁵ Unintentional hypercapnia and hypocapnia frequently occur in clinical practice and may worsen outcomes for patients, particularly those with a traumatic brain injury (TBI).^14–16 In patients successfully resuscitated from cardiac arrest, acidosis has a negative inotropic and pro-arrhythmogenic effect, and is associated with an increased risk of multi-organ failure and mortality.^17–19 However, recent data observed that mild to moderate hypercapnia may be associated with a high probability of good neurological outcome, suggesting that acidosis may be of benefit in out-of-hospital cardiac arrest (OHCA).²⁰

ETCO₂ is frequently used by clinicians as a surrogate for PaCO₂¹,²,³ but does not necessarily correspond to the PaCO₂ levels obtained on an arterial blood gas (ABG).¹,² Understanding the relationship between the PaCO₂ and ETCO₂ is important in guiding ventilation decisions for physiologically unstable pre-hospital patients.^21–24 The Association of Anaesthesists of Great Britain and Ireland (AAGBI) recommend maintaining ETCO₂ 4.0–4.5 kPa to achieve normocapnia post-intubation with pre-hospital emergency anaesthesia.²⁵ In healthy adults the acceptable difference between PaCO₂ and ETCO₂ is widely quoted as 0.5–1 kPa.²⁶ Most studies exploring this relationship have been undertaken in spontaneously breathing patients^11,27–29 or in the operating theatre.³⁰,³¹ In ventilated unstable pre-hospital patients the difference may vary as a consequence of ventilation-perfusion mismatch, acid-base disturbance, haemodynamic instability and ventilator set-up.³²,³³ Studies exploring the relationship between PaCO₂ and ETCO₂ in unstable patients have shown significant heterogeneity.

The literature on the relationship between ETCO₂ and PaCO₂ is conflicting and may differ across disease processes, with an increasing gradient associated with increasingly severe metabolic and physiologic disturbance.¹⁴,¹⁶,²² These studies have largely been undertaken in an Emergency Department (ED)¹¹,¹⁴,¹⁶,³⁴,³⁵ or intensive care unit (ICU)¹⁰,²⁶,³⁶ setting. Four studies have examined the PaCO₂-ETCO₂ gradient in pre-hospital intubated patients, all of which concluded that the PaCO₂ cannot be estimated by the ETCO₂^21–24 Methodological limitations of these studies include a variable time lag between ETCO₂ and PaCO₂ sampling,²²,²⁴ focus on a single aetiology alone,²²,²⁴ and small sample size.²¹,²³,²⁴

PaCO₂ is rarely available to pre-hospital clinicians and consequently ETCO₂ values are often used to adjust ventilation.^1–3 There is a paucity of data on patients with OHCA and TBI, areas where acid base and PaCO₂ levels are known to affect outcomes.¹⁴,¹⁵,¹⁷,¹⁹ This study aims to describe the relationship between PaCO₂ and ETCO₂ for pre-hospital intubated patients with a priori subgroup analysis for patients with medical, OHCA and major trauma as pathological groups.

Methods

This was a retrospective, cohort study using data drawn from the electronic medical records (HEMSbase 2, V1.0 2015, Medic One Systems, London, England) of a doctor-paramedic pre-hospital critical care team covering a mixed urban – rural population of around 6.3 million.³⁷,³⁸ A consecutive sample of all eligible patients on HEMSbase 2 (February 2015 to November 2018) was collected. Patients were included if they were invasively ventilated with a documented PaCO₂ drawn and analysed within 5 minutes of the documented end-tidal value. Venous, capillary and unlabelled samples were excluded. Samples analysed but with device failure and any sample that could not be definitively identified as drawn within 5 minutes of the ETCO₂ were also excluded. The study was registered locally as a service evaluation, approved by the Caldicott Guardian, and was conducted on retrospective, non-identifiable patient data. The study was run through the NHS Health Research Authority Ethical Approval Decision Tool and followed the principles of the Declaration of Helsinki, consequently ethical approval was not required.

ETCO₂ values were measured by sidestream sampling (Microstream, FilterLine Set, Oridion Medical, Jerusalem, Israel) using a Zoll X monitor (Zoll Medical Corporation, Cheshire, England) with the values and timing directly exported from the monitor into the HEMSbase 2 database. PaCO₂ samples were processed using the i-STAT point of care device (i-STAT, Abbott Point of Care Inc., Princeton, USA), which was calibrated as per the manufacturers guidelines³⁹ with a factory programmed internal quality check system. It was serviced every 6 months with ad-hoc servicing following any failure or repair. The values from the i-STAT device were manually entered by clinicians into HEMSbase 2. The Zoll X allowed recording of heart rate and blood pressure whilst the i-STAT device also recorded pH. Timings of the PaCO₂ values were identified from the recording of the i-STAT analysis time on HEMSbase2 and/or the hand written Patient Report Form (PRF), or from the timing of first invasive blood pressure measurement on the patients monitor (when it was clearly documented that the i-STAT sample was taken on insertion of an arterial line). Aetiology was classified as either medical, trauma or OHCA based upon HEMSbase 2 coding. Haemodynamic status was calculated as shock index from the measurements closest to (and within 5 minutes of) PaCO₂ and ETCO₂ sampling. Failure of i-STAT (device use which did not result in PaCO₂ measurement) was identified by manually reviewing all free text on the database. All data points were derived from different individual patients.

Classification error was identified and defined as when the ETCO₂ classification (hypocapnia (<4kPa)/ normocapnia (4–6 kPa)/hypercapnia (>6 kPa))²⁵,⁴⁰ did not match the PaCO₂ classification.

Analysis

Data were analysed in Excel for Windows (Microsoft Office Professional 2013, V 15.0, Microsoft Corporation, Redmond, Washington, USA). Data are reported as n (%), and median [inter-quartile range]. Statistical analysis was conducted using R (R Core Team (2018), R; A Language and environment for statistical computing. R Foundation for Statistical Computing, Vienna, Austria. https://www.R-project.org). Bland-Altman plots were calculated for each aetiology to describe the agreement between the PaCO₂-ETCO_2. A regression model incorporating age, gender, pH and shock index was undertaken. For all analyses, significance was accepted at α = 0.05 with 95% confidence intervals.

Results

During the study period a total of 73 patients had complete data (Figure 1). The aetiology was arranged into three categories: medical (13, 17.8%), traumatic (22, 30.1%), and OHCA (38, 52.1%). Demographics are described in Tables 1 and 2. The ETCO₂, PaCO₂ and PaCO₂-ETCO₂ gradient for medical, OHCA and trauma aetiologies is presented in Table 1.

Figure 1.

A consort diagram describing the number of intubated patients with usable PaCO₂-ETCO₂ data (n = 73).

Table 1.

Table demonstrating demographics and ETCO₂, PaCO₂ and PaCO₂-ETCO₂ gradient for medical, OHCA and trauma aetiologies (n = 73).

Aetiology (N)	Median Age [IQR]	Number male (%)	Median shock index [IQR]	Median pH [IQR]	Median base excess [IQR]	Median ETCO₂ [IQR]	Median PaCO₂ [IQR]	Median PaCO₂-ETCO₂ [IQR]
Total (73)	62 [45–72]	48 (65.7)	0.84 [0.66–1.01]	7.21 [7.01–7.35]	–7 [−13–−3]	4.2 [3.8–4.7]	6.3 [5.5–7.8]	2.01 [1.29–3.1]
Medical [13]	72 [47–77]	7 (53.8)	0.85 [0.49–0.88]	7.27 [7.18–7.38]	–3.4 [−7.3–1.3]	4.6 [4.1–4.7]	6.6 [5.3–6.9]	1.9 [1.2–2.4]
OHCA [38]	61 [49.5–70.5]	26 (68.4)	0.93 [0.71–1.24]	7.14 [6.91–7.22]	–9 [−17–−6.4]	4.2 [3.9–4.9]	7.3 [5.9–8.8]	2.7 [1.8–4.8]
Trauma [22]	58.5 [32.75–66.75]	15 (68.2)	0.75 [0.67–0.88]	7.35 [7.22–7.41]	–3 [−10–1]	4.2 [3.8–4.7]	5.5 [5.1–5.8]	1.3 [0.9–1.9]

*Missing data 1 OHCA pH, 2 trauma shock index, 5 OHCA shock index, 1 Medical BE, 3 OHCA BE, 1 Trauma BE.

Table 2.

Table demonstrating further demographics for medical and trauma aetiologies.

Medical		Trauma
Mechanism	Number of cases	Number of injured areas per patient	Number of cases	Injury location	Number of injuries
Asthma	2	1	13	Head	16
Collapse	3	2	4	Thorax	5
CVA	4	3	1	Abdomen	2
Fitting	3	Not recorded	4	Limb	1
Other	1			Not recorded	4

The levels of agreement between ETCO₂ and PaCO₂ are presented using Bland-Altman plots (Figure 2). The Bland-Altman bias for all aetiologies (−2.5 (CI −3.0 to −2.1) was considerably greater than the putatively acceptable gradient of 0 −1 kPa and limits of agreement were wide (upper 1.1 (CI 0.3 to 1.8) and lower −6.2 (CI −6.9 to −5.5). The largest bias (−3.2 (CI −3.9 to −2.7)) and widest limits of agreement (upper 0.3 (CI −0.7 to 1.4) and lower −6.8 (CI −7.9 to – 5.8)) seen for OHCA.

Figure 2.

Bland-Altman Plots demonstrating PaCO₂-ETCO₂ gradient in pre-hospital intubations of all aetiologies from a single UK Helicopter Emergency Medicine Service (n = 73).

The commonly accepted PaCO₂-ETCO₂ gradient of 0–1 kPa (median 0.7 [0.51 – 0.93]) was seen for 11 (15.1%) of total patients whilst a gradient between 1–5 kPa (median 2.1 [1.6–2.85]) was seen for 60 (82.2%) of patients). The PaCO₂-ETCO₂ gradient was observed to be ≥5 kPa (median 5.7 [5.5–6.9]) for 11 patients (15.1%). Classification error is demonstrated in Table 3. As recommended by AAGBI an ETCO₂ of 4.0–4.5 kPa was seen in 21/73 (28.7%) and 11/21 (52.3%) of these were misclassified as normocapnic (4–6 kPa) when the PaCO₂ was hypercapnic (>6.0 kPa).

Table 3.

Table demonstrating classification error in pre-hospital intubations of all aetiologies.

Cases	Number (%) with classification error	Classification error subtype	Number (%) classification error subtype	Number per mechanism and (%) of the classification error subtype
Cases	Number (%) with classification error	Classification error subtype	Number (%) classification error subtype	Medical	OHCA	Trauma
Total (73)	53 (72.6)	ETCO₂ hypocapnic (<4 kPa) & PaCO₂ normocapnic (4–6 kPa)	15 (28.3)	3 (20.0)	3 (20.0)	9 (60.0)
		ETCO₂ hypocapnic (<4 kPa) & PaCO₂ hypercapnic (>6 kPa)	9 (16.9)	0 (0.0)	8 (88.9)	1 (11.1)
		ETCO₂ normocapnic (4–6 kPa) & PaCO₂ hypercapnic (>6 kPa)	28 (52.8)	6 (21.4)	18 (64.3)	4 (14.3)
		ETCO₂ hypercapnic (>6 kPa) & PaCO₂ normocapnic (4–6 kPa)	1 (1.8)	0 (0.0)	0 (0.0)	1 (100.0)

Note: Classification error defined as when the ETCO₂ classification (hypocapnia/normocapnia/hypercapnia) does not match the PaCO₂ classification (hypocapnia/normocapnia/hypercapnia).

In 71 (97.3%) of patients PaCO₂ was higher than the ETCO₂. In two (2.7%) patients with trauma the converse was observed. One patient was a 45 yr pedestrian involved in a collision with a heavy transport vehicle sustaining injuries to the head and abdomen. She had a shock index of 0.66, respiratory rate of 19, pH 7.4, PaCO₂ 5.4 kPa, ETCO₂ 6.1 kPa and PaCO₂-ETCO₂ –0.7 kPa. The second patient was a 19 yr who fell from a building sustaining an isolated head injury. He had a shock index of 0.64, respiratory rate of 25, pH 7.5, PaCO₂ 3.6 kPa, ETCO₂ 3.9 kPa and PaCO₂-ETCO₂ –0.3 kPa.

A regression model (Table 4) demonstrated that a significant effect was seen for age (p < 0.05) and pH (p < 0.001).

Table 4.

Table demonstrating P values from linear regression model for PaCO₂-ETCO₂ gradient in pre-hospital intubations of all aetiologies (n = 73).

Variable	T Value	P Value
Age	2.44	<0.05
Gender	–0.45	0.65
Shock Index	–1.71	0.09
pH	–9.41	<0.001

Discussion

The Bland-Altman bias for all aetiologies was more than the commonly accepted gradient of 0-1 kPa and limits of agreement were wide. The variation in the PaCO₂-ETCO₂ gradient was observed to differ based on aetiology with the largest bias and widest limits of agreement seen for OHCA, perhaps in relation to the low cardiac output in this patient group and subsequent high ventilation/perfusion (V/Q) mismatch. Classification error was common and a high percentage of patients who were hypercapnic (>6.0 kPa) were misclassified as normocapnic (4.0–6.0 kPa) by the ETCO₂. The magnitude of the differences between the ETCO₂ and PaCO₂, levels of variation and inability to predict this suggest that ETCO₂ is not a suitable surrogate upon which to base ventilatory settings in clinical conditions where pH or PaCO₂ require precise control. However there is not sufficient data to quantify potential harms.

In this study all but two pre-hospital intubated patients had an ETCO₂ < PaCO₂ and the majority displayed a PaCO₂-ETCO₂ gradient >1 kPa. In two cases a negative PaCO₂-ETCO₂ gradient was observed. The reasons for this unexpected findings are not readily apparent. This has previously been described most commonly in pregnancy,⁴¹ obesity and infants.⁴²,⁴³ It may reflect the use of large tidal volumes with low respiratory rates, an increased cardiac output, low functional residual capacity or lung compliance.^41–43 The patients in our study did not meet these criteria. It may be measurement or calibration error or reflect that readings were taken prior to a steady state being reached following a period of hypoventilation – for example following adjustments to ventilatory parameters or pause in ventilation.

In healthy patients^27–29 and in selected studies in the ED^11,35 and ICU setting^10,26,34,36 a close relationship and acceptable difference between the ETCO₂ and PaCO₂ has been observed. However, the literature is conflicting on this and a predictable relationship between the ETCO₂ and PaCO₂ has not been demonstrated in the pre-hospital environment.^21–24 The authors were able to identify only four studies performed in the prehospital arena.^21–24 In a study of 100 pre-hospital intubated patients of mixed aetiologies Belpomme et al.²¹ recorded ABG results after intubation (with a stabilising ventilation period of approximately 15 minutes) and on arrival at hospital, comparing these to the ETCO₂ measurements. Similarly to the study reported here PaCO₂ was measured by the i-STAT analyser.²¹ Bland-Altman plots were not performed and analysis of variance was completed by the Fisher test.²¹ Belpomme et al. observed a mean pre-hospital of PaCO₂-ETCO₂ gradient of 9 mmHg ± 13.6 with a range of -19.7 to 75 mmHg (1.2 kPa ± 1.8 (range –2.6 to 9.9 kPa) and concluded that there was wide variation in the PaCO₂-ETCO₂ gradient.²¹ Similarly in the study reported in this journal the overall median PaCO₂-ETCO₂ gradient was 2.01 kPa [1.29 – 3.1] with a range of -0.7 to 7.15 kPa. Belpomme et al. did not identify aetiology specific differences in the PaCO₂-ETCO₂ gradient, however subgroup analysis was limited by small sample sizes with each aetiology including only 3–30 patients.²¹ 20% of all patients were described as cardiac aetiology but it is not clear if any patients were post OHCA.²¹

Prause et al. compared ETCO₂ to PaCO₂ for 61 ABG samples (OPTI 1 blood Gas Analyzer (AVL, Graz, Austria)) drawn from 27 patients intubated in the pre-hospital environment.²³ The authors concluded that there was no correlation (using linear regression analysis) between the ETCO₂ and PaCO₂ in patients who underwent cardiopulmonary resuscitation (CPR) (R₂ = 0.04) and in those with respiratory insufficiency of cardiopulmonary origin (R₂ = 0.02). A weak (R₂ = 0.34) but statistically insignificant (p = 0.053) correlation was reported for patients with extrapulmonary respiratory disturbances (largely cerebral).²³ The data included samples from patients receiving CPR in whom a higher Vd/Vt is expected.²³ The accuracy of ETCO₂ measurement is a major limitation to Prause et al’s study.²³ The capnometry device (Capnometry module of the Defigard 2000 (Schiller, Linz, Austria)) was not equipped with CO₂ curve display and the authors report that CO₂ was likely measure in mid-expiration rather than end, unlike in the study reported here.²³ The small sample size and lack of analysis for pH and haemodynamic status (as conducted in this study) are further limitations.²³

Cooper et al.²² explored the PaCO₂-ETCO₂ gradient in a prospective study of 162 pre-hospital invasively ventilated trauma patients. Pre-hospital ETCO₂ values maintained enroute were recorded and an arterial PaCO₂ was obtained within 15 minutes of hospital arrival using bench-top laboratory analyser (rather the portable analysers as in this study).²² Haemodynamic status was not explicitly stated but inferred from the injury severity score.²² The authors concluded that overall the ETCO₂ did not correlate (linear regression analysis) with the PaCO₂ (R₂ = 0.08).²² Cooper et al. observed that patients with more severe injuries had a greater difference between the ETCO₂ and PaCO₂ (injury severity score ≥25 (R₂ = 0.0003)) than those with less severe injures (injury severity score ≤ 15 (R₂ = 0.46)).²² Analysis by injury severity score is an advantage of the combined pre-hospital/hospital data set in Cooper’s study, which we were unable to obtain.²² The variable time between the pre-hospital ETCO₂ and hospital PaCO₂ is a limitation to Cooper’s study, which may not accurately reflect the PaCO₂-ETCO₂ gradient.²² In the study reported here all patients where there was >5 minutes delay between end tidal and PaCO₂ measurement were excluded from analysis but analysis included OHCA. Cooper et al excluded patients with OHCA.²²

Recently Price et al. conducted a retrospective cohort study of 40 adult patients with TBI who were intubated pre-hospital.²⁴ The study was conducted at the same pre-hospital organisation as the study reported here.²⁴ Price et al. compared the ETCO₂ to PaCO₂ measurements taken within 30 minutes of hospital arrival.²⁴ The authors concluded that there was only a moderate correlation (R₂ = 0.23, p = 0.002) between PaCO₂ and ETCO₂.²⁴ The Bland Altman bias was 1.7 (CI 1.4 to 2.0) kPa with upper and lower limits of agreement of 3.6 (CI 3.0 to 4.1) kPa and −0.2 (CI −0.8 to 0.3) kPa.²⁴ There was no evidence of a larger gradient associated with more severe TBI (p = 0.29).²⁴ For patients with coexisting serious thoracic injury a larger gradient was observed (2.0 (±1.1) kPa) but no significant gradient correlation was reported (R₂ =0.13, p = 0.10).²⁴ The pre-hospital gradient was not analysed, however, a subgroup of seven (17.5%) of patients had a pre-hospital arterial blood sample via the I-STAT and on arrival had a lower and more appropriate PaCO₂ compared with the no ABG group (4.7 (±0.2) kPa).²⁴ This may suggest that the pre-hospital ABG allowed early and appropriate adjustment of ventilation. Although undertaken at the same organisation as this study, there were only a small number of patients with a pre-hospital ABG and these were not closely matched to a pre-hospital ETCO₂ reading, making it likely that few to none patients crossed over between study datasets. Similarly to Cooper et al. the wide and variable time between ETCO₂ and PaCO₂ measurements is a limitation as it is possible that changes in the gradient may reflect changes in the ventilation strategy during this period.²²,²⁴

The above four studies and the data presented in the study published here highlight the poor levels of correlation and agreement between the PaCO₂ and ETCO₂. All offer a similar message regarding the limitation of using ETCO₂ to estimate the PaCO₂ for the pre-hospital patent.^21–24 The differences between ETCO₂ and PaCO₂ levels, the lack of a predictable relationship and the variation with differing levels of resuscitation and cardiac output suggest that in conditions such as TBI where outcomes may be altered by PaCO₂/pH then measured arterial levels are required. However in many clinical scenarios, such as ventilating a critically unwell patient with bronchospasm the clinical focus is on the ventilatory pressures and PaCO₂ a secondary concern, so ETCO₂ levels may be acceptable to the treating team.

The gradient in trauma is further explored in studies by Warner et al. and Lee et al.¹⁴,¹⁶ Warner et al collected PaCO₂ and ETCO₂ data on 180 intubated ED trauma patients.¹⁴ There was a significant but weak correlation between the ETCO₂ and PaCO₂ (R₂ = 0.277, p = <0.001) but with wide confidence intervals. Warner et al. reported that the strongest correlation was for patients with isolated TBI (R₂ = 0.52) whilst the weakest were seen in patients with severe abdominal injuries (R₂ = 0.19) or with a shock index ≥0.9 (R₂ = 0.17).¹⁴ Lee et al.¹⁶ reported a moderate correlation (r = 0.67) between the PaCO₂-ETCO₂ for 66 patients with TBI cared for in the ED. Patients with severe chest trauma (p = 0.049), hypotension (p = 0.017) and metabolic acidosis (p = 0.018) had a greater difference in the PaCO₂-ETCO₂ gradient.¹⁶ In a subgroup of 28 patients with no hypotension, acidosis or trauma a strong correlation was observed (r = 0.893, p = <0.001).¹⁶ The study reported in this journal identified that with decreasing pH the PaCO₂-ETCO₂ gradient increases. The data reported in this study was insufficient to allow adjustment for base excess. The overall effect of acidosis on the PaCO₂-ETCO₂ gradient is likely in part a consequence of the acidosis reflecting tissue hypoperfusion and increased V/Q mismatch.³³ In the studies by Warner et al.¹⁴ and Lee et al.¹⁶ it is possible that the most physiologically deranged patients with expectedly greater differences between the PaCO₂ and ETCO₂ were not included as they died prehospital or were transferred directly to operating theatre. Overall, the previous literature suggests that in trauma patients with higher injury severity, chest wall injuries and haemodynamic instability the difference and variability between the ETCO₂ and PaCO₂ increases.¹⁶,²² The study reported here involved 13/22 (59%) single system trauma from a small number of patients with a median SI 0.75 [0.67–0.88] and pH of 7.35 [7.22–7.41] suggesting a less injured cohort.

For OHCA previous literature describing the PaCO₂-ETCO₂ gradient is limited to the small sample of 12 patients analysed by Prause et al.²³ Previous studies have reported an increased mortality associated with hypercapnea.^17–19 However, in a recent prospective multi-centre study of 280 post-ROSC patients undergoing targeted temperature management Kilgannon et al observed that PaCO₂ had an inverted “U” shaped associated with mild to moderate hypercapnia having the highest probability of good neurological outcome with further increases reducing the likelihood of good neurological outcome.²⁰ Whilst in a retrospective, observational study of 58 post-ROSC patients Kim et al.¹⁹ evaluated the PaCO₂ and ETCO₂ values at 6, 12 and 24 hours. PaCO₂-ETCO₂ gradients were significantly lower in survivors than non-survivors at 12 hours (p = 0.040) and 24 hours (p = <0.001).¹⁹ This may reflect the increased gradient in the more critically unwell patients.¹⁹ In the study reported in this journal the greatest variance between the ETCO₂ and PaCO₂ was observed for patents with OHCA with most patients being acidotic (pH 7.14 [6.91 - 7.22]). Clinical consideration of the greater difference and variance in the PaCO₂-ETCO₂ gradient for OHCA cases may prompt close arterial monitoring and more aggressive ventilation for this aetiology.

In a physiology based retrospective cross-sectional study of 493 data points from 56 mechanically ventilated paediatric ICU patients McSwain et al.³⁶ calculated the PaCO₂-ETCO₂ gradient across a range of deadspace volume/tidal volume (V_d/V_t) values using the Bohr equation.³⁶ The authors reported that the strength of the correlation between ETCO₂ and PaCO₂ decreased as V_d/V_t increased (V_d/V_t ≤ 0.40 (R = 0,95); V_d/V_t >0.70 (R2 = 0.78)³⁶ and observed that the gradient between the PaCO₂ and ETCO₂ increased with increasing V_d/V_t (from an average of 0.04 kPa (0.3 millimetres of mercury (mmHg)) at low V_d/V_t to 2.4 kPa (18 mmHg) at high V_d/V_t).³⁶ The physiological principals can be expected similar in adult populations.³⁶ McSwain et al concluded that the ETCO₂ is a useful indicator of PaCO₂ provided that the physiologically expected increase in the PaCO₂-ETCO₂ gradient is considered.³⁶ However, increased variability and an inconsistent difference in the PaCO₂-ETCO₂ gradient with increasing dead space makes clinical use challenging as PaCO₂ cannot be reliably predicted.³⁶ Pre-hospital patients with critical illness have high V_d/V_t and the variability in the gradient, particularly for trauma and OHCA patients, suggests that the ETCO₂ cannot consistently predict the PaCO₂ and therefore arterial sampling of the PaCO₂ may be useful in the pre-hospital environment.

Limitations

This study involved the use of point of care i-STAT (i-STAT, Abbott Point of Care Inc., Princeton, USA). In a prospective study of intubated adult patients in an ICU setting by Thomas et al.⁴⁴ the i-STAT samples were concluded as being acceptable surrogates for bench-top laboratory measurements. However, the ICU setting differs from the pre-hospital setting where differing external and operator factors may affect machine performance. Indeed we identified 46/298 (15.4%) attempted blood gas analyses failed (Figure 1). Reasons included error codes, temperature, clotted blood, underfilled/overfilled samples, the machine failing to turn on and operator failure. The number of attempted blood gas samples was less than the number of intubations (298/1800 (16.5%)), suggesting clinicians selected patients for analysis. The median shock index of 0.84 [0.66–1.01] and high percentage of single system trauma infers that sicker patients may not have been chosen for pre-hospital arterial sampling or did not have this recorded due to the concurrent pressures on clinicians for these patients. The small sample size also risks analysis being effected by extreme outliers, for example five cases involved a PaCO₂-ETCO₂ gradient of >6 kPa.

Furthermore, as data collection was retrospective the authors were unable to ensure that data was consistently taken in a steady state in relation to intubation or changes in ventilator settings.

Overall the number of missing data risks high levels of bias and reflects the difficulties with retrospective data collection in the pre-hospital environment and limited further sub-group analysis. However our findings are consistent with previous work.

Conclusion

The Bland-Altman bias for all aetiologies was greater than the widely accepted gradient of 0–1 kPa and limits of agreement were wide. A high percentage of cases involved hypercapnic patients being misclassified as normocapnic by the ETCO₂. The data in this study suggests that in a high proportion of patients, particularly those with major trauma or OHCA, ETCO₂ is not a sufficiently accurate predictor of PaCO₂ to guide ventilator settings. Arterial blood gas monitoring may therefore provide useful data for prehospital teams where control of acidaemia and carbon dioxide levels are of clinical relevance. This data presented here is insufficient to draw firm conclusions and larger, prospective studies are required.

Disclaimers

Views expressed in the submitted article are our own and not an official position of East Anglian Air Ambulance.

Footnotes

Acknowledgements

The authors would like to acknowledge and thank Richard Hindson, Dr Ed Barnard and Dr Jeremy Mauger for their valued support on this project.

Declaration of conflicting interests

The author(s) declared no potential conflicts of interest with respect to the research, authorship, and/or publication of this article.

Funding

The author(s) received no financial support for the research, authorship, and/or publication of this article.

ORCID iD

Owen Hibberd

References

Masimo International. Emerging uses of capnography in Emergency Medicine. Masimo White Paper, www.masimo.co.uk/siteassets/uk/documents/pdf/lab7850a_whitepaper_emerging_uses_of_capnometry_in_emergency_medicine_british.pdf (2013, accessed 23 October 2019)

Rieves

Bleess

Be all end-tidal: the expanding role of capnography in prehospital care. EMS Med Website, www.naemsp-blog.com/emsmed/2017/3/22/be-all-end-tidal-the-expanding-role-of-capnography-in-prehospital-care (2017, accessed 23 October 2019).

Aminiahidashti

Shafiee

Kiasari

, et al. Applications of end-tidal carbon dioxide monitoring in emergency department: a narrative review. Emerg (Tehran) 2018; 6: e5.

Hartmann

Farris

Di Gennaro

, et al. Systematic review and meta-analysis of end-tidal carbon dioxide values associated with return of spontaneous circulation during cardiopulmonary resuscitation. J Intensive Care Med 2015; 30: 426–435.

Sutton

French

Meaney

, et al. Physiologic monitoring of CPR quality during adult cardiac arrest: a propensity-matched cohort study. Resuscitation 2016; 106: 76–82.

Nagler

Wright

Krauss

End-tidal carbon dioxide as a measure of acidosis among children with gastroenteritis. Pediatrics 2006; 118: 260–267.

Kartal

Eray

Rinnert

, et al. ETCO₂: a predictive tool for excluding metabolic disturbances in non-intubated patients. Am J Emerg Med 2011; 29: 65–69.

Hunter

Silvestri

Ralls

, et al. A prehospital screening tool utilizing end-tidal carbon dioxide predicts sepsis and severe sepsis. Am J Emerg Med 2016; 34: 813–819.

Kim

GB.

The effectiveness of end tidal carbon dioxide for improvement of the metabolic acidosis patients. Ann Emerg Med 2013; 62: S87.

10.

Bhanot

Simpson

Stites

, et al. End-tidal carbon dioxide (ETCO₂) as a surrogate for PaCO₂ during spontaneous breathing trial. Chest 2015; 148: 302A.

11.

Cinar

Acar

Arziman

, et al. Can mainstream end-tidal carbon dioxide measurement accurately predict the arterial carbon dioxide level of patients with acute dyspnoea in ED. Am J Emerg Med 2012; 30: 358–361.

12.

Rentola

Hastbacka

Heinonen

, et al. Estimation of arterial carbon dioxide base on end-tidal gas pressure and oxygen saturation. J Clin Med 2018; 7: 290–301.

13.

Marhong

Fan

Carbon dioxide in the critically ill: too much or too little of a good thing?

Respir Care 2014; 4: 1597–1605.

14.

Warner

Cuschieri

Garland

, et al. The utility of early end-tidal capnography in monitoring ventilation status after severe injury. J Trauma 2009; 66: 26–31.

15.

Davis

DP.

Early ventilation in traumatic brain injury. Resuscitation 2008; 77: 418.

16.

Lee

Hong

Han

, et al. Concordance of end-tidal carbon dioxide and arterial carbon dioxide in severe traumatic brain injury. J Trauma 2009; 67: 526–530.

17.

Wang

Prince

Drennan

, et al. Post-resuscitation arterial oxygen and carbon dioxide and outcomes after out-of-hospital cardiac arrest. Resuscitation 2017; 120: 113–118.

18.

Jamme

Ben Hadj Salem

Guillemet

, et al. Severe metabolic acidosis after out-of-hospital cardiac arrest: risk factors and association with outcome. Ann Intensive Care 8: 62–70.

19.

Kim

Hwang

Kang

, et al. The gradient between arterial and end-tidal carbon dioxide predicts in hospital mortality in post-cardiac arrest patient. Am J Emerg Med 2019; 37: 1–4.

20.

Kilgannon

Hunter

Puskarich

, et al. Partial pressure of arterial carbon dioxide after resuscitation from cardiac arrest and neurological outcome: a prospective multi-center protocol-directed cohort study. Resuscitation 2019; 135: 212–220.

21.

Belpomme

Ricard-Hibon

Devoir

, et al. Correlation of PCO₂ and PETCO₂ in prehospital controlled ventilation. Am J Emerg Med 2005; 23: 852–859.

22.

Cooper

Kraatz

Kubiak

, et al. Utility of prehospital quantitative end tidal CO₂? Prehosp Disaster Med 2013; 28: 1–6.

23.

Prause

Hetz

Lauda

, et al. A comparison of the end-tidal CO₂ documented by capnometry and the arterial PCO₂ in emergency patients. Resuscitation 1997; 35: 145–148.

24.

Price

Sandbach

Ercole

, et al. End-tidal and arterial carbon dioxide gradient in serious traumatic brain injury after prehospital emergency anaesthesia: a retrospective observational study. Emerg Med J 2020; 1–6, doi: 10.1136/emermed-2019-209077.

25.

Lockey

Crewdson

Davies

, et al. AAGBI: safer pre-hospital anaesthesia 2017. Anaesthesia 2017; 72: 379–390.

26.

Razi

Moosavi

Omidi

, et al. Correlation of end-tidal carbon dioxide with arterial carbon dioxide in mechanically ventilated patients. Arch Trauma Res 2012; 1: 58–62.

27.

Takano

Sakamoto

Kiyofuji

, et al. A comparison of the end-tidal CO₂ measured by portable capnometer and the arterial PCO₂ in spontaneously breathing patients. Respir Med 2003; 97: 476–481.

28.

Fuke

Miyamoto

Ohira

, et al. Evaluation of transcutaneous CO₂ responses following acute changes in PaCO₂ in healthy subjects. Respirology 2009; 14: 436–442.

29.

Barten

Wang

ESJ.

Correlation of end-tidal CO₂ measurements to arterial PaCO₂ in non-intubated patients. Ann Emerg Med 1994; 23: 560–563.

30.

Ali

Dubikaitis

Raheem Al Qattan

The relationship between end tidal carbon dioxide and arterial carbon dioxide during controlled hypotensive anaesthesia. Med Principles Pract 2002; 11: 35–37.

31.

Tremper

Shoemaker

Shippy

, et al. Transcutaneous PCO₂ monitoring on adult patients in the ICU and the operating room. Crit Care Med 1981; 9: 752–755.

32.

West

JB.

Respiratory physiology: the essentials. 9th ed. Baltimore, MD, USA: Lippincott Williams & Wilkins, 2012.

33.

West

JB.

Pulmonary pathophysiology: the essentials. 8th ed. Baltimore, MD, USA: Lippincott Williams & Wilkins, 2013.

34.

McDonald

Montgomery

Cerrito

PB.

Comparison of end-tidal CO₂ and PaCO₂ in children receiving mechanical ventilation. Pediatr Crit Care Med 2002; 3: 244–249.

35.

Yosefy

Hay

Nasri

, et al. End tidal carbon dioxide as a predictor of the arterial PCO₂ in the emergency department setting. Emerg Med J 2004; 21: 557–559.

36.

McSwain

Hamel

Smith

, et al. End-tidal and arterial carbon dioxide measurements correlate across all levels of physiologic dead space. Respir Care 2010; 55: 288–293.

37.

Barnard

EBG

Sandbach

Nicholls

, et al. Prehospital determinants of successful resuscitation after traumatic and non-traumatic out-of-hospital cardiac arrest. Emerg Med J 2019; 36: 333–339.

38.

Office for National Statistics. Revised population estimates for England and Wales: mid-2012 to mid-2016. UK: Office for National Statistics.

39.

Abbott Point of Care Inc. Calibration verification and the i-STAT system. I-STAT Technical Bulletin, www.pointofcare.abbott (accessed 9 April 2020).

40.

Resuscitation Council UK. Advanced life support. 7th ed. London, England: Resuscitation Council UK, 2018.

41.

Shankar

Moseley

Kumar

, et al. Arterial to end-tidal carbon dioxide tension during caesarean section anaesthesia. Anaesthesia 1986; 41: 698–702.

42.

Fletcher

Jonson

Deadspace and the single breath test for carbon dioxide during anaesthesia and artificial ventilation. Br J Anaesth 1984; 56: 109–119.

43.

Shankar

Moseley

Kumar

Negative arterial to end-tidal gradients. Can J Anaesth 1991; 38: 260–261.

44.

Thomas

Hoffman

Handrahan

, et al. The measure of treatment agreement between portable and laboratory blood gas measurements in guiding protocol-driven ventilator management. J Trauma 2009; 67: 303–314.

The PaCO 2 -ETCO 2 gradient in pre-hospital intubations of all aetiologies from a single UK helicopter emergency medicine service 2015–2018

Abstract

Background

Methods

Results

Conclusion

Keywords

Introduction

Methods

Analysis

Results

Discussion

Limitations

Conclusion

Disclaimers

Footnotes

Acknowledgements

Declaration of conflicting interests

Funding

ORCID iD

References