Arterial blood gases during and their dynamic changes after cardiopulmonary resuscitation: A prospective clinical study

Resuscitation 106 (2016) 24–29

Contents lists available at ScienceDirect

Resuscitation journal homepage: www.elsevier.com/locate/resuscitation

Clinical paper

Arterial blood gases during and their dynamic changes after cardiopulmonary resuscitation: A prospective clinical study夽 Walter Spindelboeck a,1 , Geza Gemes b,∗,1 , Christa Strasser c , Kathrin Toescher c , Barbara Kores c , Philipp Metnitz b , Josef Haas d , Gerhard Prause b a

Division of Gastroenterology and Hepatology, Department of Internal Medicine, Medical University of Graz, Austria Clinical Department of General Anaesthesiology, Emergency and Intensive Care Medicine, Department of Anaesthesiology, Medical University of Graz, Austria c Medizinercorps, Austrian Red Cross, Division of Graz, Austria d Department of Obstetrics and Gynaecology, Medical University of Graz, Austria b

a r t i c l e

i n f o

Article history: Received 28 January 2016 Received in revised form 9 June 2016 Accepted 14 June 2016 Keywords: Advanced life support Blood gas analysis CPR

a b s t r a c t Purpose: An arterial blood gas analysis (ABG) yields important diagnostic information in the management of cardiac arrest. This study evaluated ABG samples obtained during out-of-hospital cardiopulmonary resuscitation (OHCPR) in the setting of a prospective multicenter trial. We aimed to clarify prospectively the ABG characteristics during OHCPR, potential prognostic parameters and the ABG dynamics after return of spontaneous circulation (ROSC). Methods: ABG samples were collected and instantly processed either under ongoing OHCPR performed according to current advanced life support guidelines or immediately after ROSC and data ware entered into a case report form along with standard CPR parameters. Results: During a 22-month observation period, 115 patients had an ABG analysis during OHCPR. In samples obtained under ongoing CPR, an acidosis was present in 98% of all cases, but was mostly of mixed hypercapnic and metabolic origin. Hypocapnia was present in only 6% of cases. There was a trend towards higher paO2 values in patients who reached sustained ROSC, and a multivariate regression analysis revealed age, initial rhythm, time from collapse to CPR initiation and the arterio-alveolar CO2 difference (AaDCO2 ) to be associated with sustained ROSC. ABG samples drawn immediately after ROSC demonstrated higher paO2 and unaltered pH and base excess levels compared with samples collected during ongoing CPR. Conclusions: Our findings suggest that adequate ventilation and oxygenation deserve more research and clinical attention in the management of cardiac arrest and that oxygen uptake improves within minutes after ROSC. Hyperventilation resulting in arterial hypocapnia is not a major problem during OHCPR. © 2016 Elsevier Ireland Ltd. All rights reserved.

Introduction In the management of cardiac arrest, arterial blood gas (ABG) analysis can yield important diagnostic information and guide therapeutic management during in-hospital treatment

夽 A Spanish translated version of the summary of this article appears as Appendix in the final online version at http://dx.doi.org/10.1016/j.resuscitation.2016.06.013. ∗ Corresponding author at: Division of General Anaesthesiology, Emergency and Intensive Care Medicine, Department of Anaesthesiology, Medical University of Graz, Auenbruggerplatz 29, 8036 Graz, Austria. E-mail address: [email protected] (G. Gemes). 1 These authors contributed equally to this work. http://dx.doi.org/10.1016/j.resuscitation.2016.06.013 0300-9572/© 2016 Elsevier Ireland Ltd. All rights reserved.

of life-threatening conditions but also during out-of-hospital cardiopulmonary resuscitation (OHCPR), thus serving as part of advanced cardiac life support.1–3 However, clinical data on blood gas analyses during OHCPR are sparse, owing to its technical challenges and personnel requirements on-site.4,5 Recent retrospective studies from our group suggested prognostic importance for the arterial base excess (BE) and oxygen partial pressure (paO2 ), obtained at a singular measurement, especially for the prediction of the return of spontaneous circulation (ROSC),5,6 however, prospective data investigating that issue are lacking at the moment for the prehospital setting. Investigations evaluating in-hospital ABG analyses obtained during the first 24 h after admission from OHCPR reported carbon dioxide partial pressure (paCO2 ) derangements after ROSC,7–9 however these data do

W. Spindelboeck et al. / Resuscitation 106 (2016) 24–29

not necessarily reflect paCO2 during OHCPR or briefly after ROSC, as most intensive care units aggressively treat cardiorespiratory derangements. Despite the lack of systematic evidence for paCO2 values under CPR conditions, current CPR guidelines advise that hyperventilation may contribute to a poor outcome. As ventilation rates are often excessive and outcome in animal models deteriorates with hyperventilation, respiratory rates of 10–12 breaths per minute are recommended.10–12 Moreover, dynamic changes of the acid-base metabolism as well as arterial blood gases from the low-flow state during resuscitation to a reperfusion state immediately after ROSC are mainly known for animal models13 , and few studies on blood gas analysis during CPR exist.14 As this is a current matter of interest, especially in light of the extensive debate of hyperoxia-induced oxygen toxicity, evidence is needed here.15 This study comprises a data set obtained during the “Blood Gas Analysis and Bicarbonate Buffering in Cardiac Arrest” trial (BABICA) which was a prospective randomized controlled multicenter trial investigating bicarbonate buffering targeted by a measured BE value during OHCPR. We herein describe for the first time prospective human data on blood gases and the acid-base metabolism obtained during OHCPR and their dynamics briefly after ROSC. In addition, we investigated prospectively, whether blood gas data may yield an important prognostic value for patient survival to hospital admission. Methods Data were obtained during a prospective randomized controlled multicenter trial (BABICA, Blood gas Analysis and Buffering In Cardiac Arrest, clinicaltrials.gov: NCT01362556, EudraCT: 2010020162-19) in out-of-hospital cardiac arrest patients. Following approval from the ethical committee of the Medical University of Graz (number: 21-357 ex 09/10) five emergency physician bases in Austria participated in the study. The emergency physicians involved were senior doctors with additional training in intensive care medicine or anaesthesia and were experienced in the interpretation of arterial blood gas analysis. All vehicles were equipped with a portable blood gas analyser (OptiCCA, OPTI Medical, Atlanta, USA) and responsible staff were instructed in its operation. Only one single arterial blood sample was gathered and analyzed in one patient, therefore each ABG sample represents a different patient. Samples were drawn on site during OHCPR or briefly after ROSC by arterial cannulation or single arterial sampling as described earlier.4,5,16 All obtained values were collected using a separate case report form, which was based on the Utstein style. The timepoint of measurement was documented exactly. Samples with paO2 < 35 mmHg were considered to be venous and thus excluded. Patients were resuscitated per 2010 guidelines,17 all patients were intubated and ventilated by an ambu bag or a portable emergency respirator. Patient transport under ongoing CPR was not performed; in case of unsuccessful resuscitation attempts patients were declared dead on scene by the attending emergency physician. No ventilation protocols (predefined tidal volume or frequency) were applied, rather, ventilation was left to the discretion of the physician. Endtidal carbon dioxide (etCO2 ) at the time of ABG recording was measured by infrared spectroscopy; etCO2 values were only documented when a stable reading could be obtained. Only etCO2 values documented synchronously with ABG measurements were included. None of the patients included in this dataset received any randomized treatment or buffer therapy before the arterial blood sample was obtained. Data were input into a specific database and statistics were calculated with SPSS 22 (IBM® , USA). Categorical data are reported as proportions and were assessed for significance by Chi-square or Fishers-exact test. Numerical data were tested for normal distribution by the Kolmogorov–Smirnov-test and Shapiro–Wilk-test. Data were compared with Student’s t-test

25

or the Mann–Whitney-U test as appropriate. For better readability all numerical data are presented as median (interquartile range). A p < 0.05 was assumed to indicate statistical significance. Hospital admission (HA), i.e. sustained ROSC upon hospital admission, was used as the outcome endpoint for all calculations. A receiver operating characteristic curve was calculated for each blood gas parameter and Youden’s index was determined. For univariate binary logistic regression coefficients for all variables were calculated separately. Thereafter a multivariate regression (backward conditional, exclusion at >0.05) model was applied. Variables generally accepted to be of possible prognostic interest (age, sex, time from collapse to CPR initiation, bystander CPR and initial rhythm) were included into the analysis in addition to the blood gas parameters paO2 , AaDCO2 (arterio-alveolar carbon dioxide difference, calculated from paCO2 and etCO2 ) and BE. Only 31 of 83 patients received randomized treatment, and this had no effect on the rate of hospital admission (9/16 vs 5/15, p = 0.38) such that we chose not to control for randomized treatment in the analysis. Nagelkerke’s Pseudo-R2 is given for each model as a measure of goodness-of-fit. A multivariate logistic regression with bootstrapping (1000 reiterations) was also applied to estimate the prediction error. Spearman’s rank test was used to determine correlations between blood gas parameters and other variables of interest. Due to the limited number of patients achieving HA, we chose not to report short-term or long-term neurologic outcome because of an anticipated lack of precision of the results. Results During the observation time of 22 months 1295 patients in cardiac arrest were recorded (Fig. 1), of which 676 were declared dead by the attending emergency physician upon the first examination. 91 patients had to be excluded from the study on the grounds of the predefined exclusion criteria and 38 patients achieved ROSC before they could be included in the study. In the remaining 490 patients, a blood gas sample could be obtained during OHCPR in 124 cases

Lifeless person (n=1295)

Advanced life support (n=619)

No ALS attempt (n=676)

Arterial blood sample obtained (n=124)

No blood sample, early ROSC, exclusion criterion (n=495)

Useable arterial blood sample (n=115)

paO2 < 35 mm Hg (n=9)

During chest compressions (n=83)

After ROSC (n=32) ABG dynamics

Died on-scene (n=65)

Hospital admission with sustained ROSC (n=18)

Logistic regression Fig. 1. Patient flow chart. Only one single arterial blood gas (ABG) analysis was obtained per patient, thus one sample represents one patient. ALS = advanced life support, paO2 = arterial oxygen partial pressure, ROSC = return of spontaneous circulation.

W. Spindelboeck et al. / Resuscitation 106 (2016) 24–29

(25%). Nine samples were considered to be venous and excluded such that 115 patients with ABGs were available for further analysis. Of the 115 samples, 83 samples were obtained during ongoing chest compressions and 32 were obtained after ROSC. The median time intervals between the start of chest compressions to the ABG analysis and ROSC to the ABG analysis were 25 min (IQR 15) and 15 min (IQR 19), respectively. Time from start of chest compressions to the ABG analysis showed a significant, but moderate correlation with BE (rho = −0.334, p = 0.004, n = 73) and pH (rho = −0.39, p = 0.001, n = 73).

40

% Hospital Admission

26

30

20

10

Table 1 Baseline characteristics and blood gas parameters of patients with arterial blood gas analysis obtained during out-of-hospital CPR with regard to survival to hospital admission (HA). Numerical data are presented as median (interquartile range, IQR). HA (n = 18) Baseline characteristics Age (median, IQR) Male sex (%) Bystander CPR (%) VF/VT as initial rhythm (%) Presumed cardiac cause of arrest (%) Time from collapse to CPR initiation (mins) Blood gas parameters paO2 (mmHg) paCO2 (mmHg) etCO2 a (mmHg) AaDCO2 a (mmHg) pH BE (mmol l−1 )

No HA (n = 65)

p-value

66 (20) 72 78 56

73 (18) 71 66 31

0.035 0.904 0.347 0.053

39

34

0.210

5 (10)

85 (73) 67 (34) 31.0 (25.0) 19.5 (19.0) 7.07 (0.27) −15.0 (9.0)

9 (11)

66 (39) 62(18) 30.0 (18.0) 29.2 (26.0) 7.02 (0.21) −15.6 (5.0)

0.081

0.050 0.596 0.991 0.071 0.325 0.397

a etCO2 measurement was not available for all patients, thus n = 51. Student’s t-test or Mann–Whitney-U-Test. BE = base excess.

.0 5 >7 .0 5

<7

<7

0 In 83 patients where the ABG sample had been drawn under ongoing OHCPR we found that acidosis (pH < 7.35) was present in 81 (98%), which was purely metabolic in 11/81 patients (14%). In the other cases, acidosis was of mixed metabolic and respiratory origin. No arterial alkalosis was seen in any patient and arterial hypocapnia (paCO2 < 35 mmHg) was only seen in 6% (5/83) of patients. The etCO2 showed a moderate correlation with the paCO2 (rho = 0.395, p = 0.004, n = 51) and a minor correlation with the AaDCO2 (rho = −0.286, p = 0.04, n = 51). Next, we analyzed whether ABG values were associated with the ability or failure to reach the hospital alive (endpoint: HA). Baseline characteristics and ABG parameters of those patients are shown in Table 1. Of note, survivors were significantly younger and showed a lower AaDCO2 and a trend to a higher paO2 . Blood gas parameters were further split by Youden’s index obtained from a receiver operating characteristic curve for HA. With regard to this distribution, paO2 and the AaDCO2 reached statistical significance and, importantly, no patient with an AaDCO2 greater than 33.5 mmHg survived to HA (Fig. 2). A univariate regression did not reveal a major influence for any single independent variable (Table 2A). In a multivariate logistic regression analysis model age, time from collapse to CPR initiation, the initial cardiac rhythm, and the AaDCO2 remained in the final model of hospital admission (Table 2B). Goodness-of-fit for prediction (Nagelkerke’s Pseudo-R2 ) for this model was 48% for all variables combined. In a multivariate logistic regression model with bootstrapping, time from collapse to CPR initiation, the initial cardiac rhythm, and the AaDCO2 were significant predictors of hospital admission (Table 2C).

5 m >7 m H 5 m g m H g >6 9 m <6 m H 9 m g m H g <3 4 m >3 m H 4 m g m H g <14 m >- m 14 ol/ m l m ol /l

ABG analyses during ongoing CPR

paO2

paCO2

AaDCO2

BE

pH

Fig. 2. Arterial blood gas parameters and rates of hospital admission with sustained return of spontaneous circulation. Parameters are split according to Youden’s index obtained by plotting a receiver operating characteristic curve. paO2 = arterial oxygen partial pressure, paCO2 = arterial carbon dioxide partial pressure, AaDCO2 = arterioalveolar carbon dioxide difference, BE = base excess. Chi-squared test or Fisher’s exact test, brackets indicate statistical significance (p for AaDCO2 = 0.02, p for paO2 = 0.04). Note that no patient with an AaDCO2 greater than 34 mmHg survived to hospital admission. End-tidal CO2 was omitted because no meaningful ROC could be generated.

ABG dynamics intra- and post-CPR Thirty-two ABG analyses were available from patients who had already achieved ROSC briefly before sampling and were compared to samples which were obtained under ongoing chest compressions. In order to rule out a selection bias caused by the patients never achieving ROSC, only patients who reached to hospital with sustained ROSC (HA) were entered into the analysis. Baseline data were comparable with respect to age, sex, the incidence of bystander CPR, the initial rhythm detected by the advanced life support team and the presumed aetiology of cardiac arrest. After ROSC, significantly higher paO2 values were observed and there was trend towards higher BE levels as well (Table 3). Discussion Data about blood gases during cardiac arrest and CPR have been gathered in the early nineteen-sixties18–21 in animal studies and retrospective evaluations,22 but to the best of our knowledge, this is the first dataset of prospectively obtained arterial blood gas analyses in humans under ongoing out-of-hospital CPR and very early after ROSC. We believe that these data may have potential implications in formulating future guidelines for the management of cardiac arrest. Our study confirms that an acidosis was present in nearly every patient during OHCPR. Experimental studies showed only minor deviations with pH values in the range of 7.2–7.3, and other human data come from patients after ROSC,23 after hospital admission24–26 or from the setting of intrahospital cardiac arrest.27 Our retrospective data from prehospital CPR conditions5 revealed lower pH levels in the range of 7.0–7.1, and recent data from patients arriving in the emergency department under ongoing CPR also showed blood gas variables which correspond well to our findings.28 Surprisingly, although a metabolic component was present in many of the cases, hypercarbia contributed to a large part of the acidosis. The accumulation of acid metabolites does not seem to be a conditio sine qua non in the state of CPR, as it does not occur in all of the cases. On the other hand, even though all patients

W. Spindelboeck et al. / Resuscitation 106 (2016) 24–29

27

Table 2 Prognostic factors in 83 patients undergoing out-of-hospital CPR with an arterial blood gas analysis obtained under ongoing chest compressions. (A) Univariate logistic regression analysis of variables with assumed prognostic relevance and blood gas parameters. (B) Multivariate logistic regression analysis (backward conditional) of parameters displayed in (A). Endpoint for all calculations: hospital with sustained return of spontaneous circulation. (C) Multivariate logistic regression analysis with bootstrap procedure. Endpoint: as in B. e(ˇ)

p

95% CI

R2 0.07 0.00 0.16 0.01 0.00 0.15 0.02 0.05 0.12

A. Univariate Age (years) Sex (male/female) Time from collapse to CPR initiation (min) Bystander CPR Presumed aetiology (cardiac/non-cardiac) Initial rhythm (VF-VT/other) paO2 (mm Hg) BE (mmol l−1 ) AaDCO2 (mmHg)

0.959 0.889 0.862 0.604 1.295 5.026 1.002 1.076 0.954

0.153 0.893 0.047 0.559 0.718 0.035 0.401 0.214 0.073

0.906 0.160 0.745 0.112 0.317 1.117 0.997 0.959 0.906

1.015 4.949 0.998 3.272 5.286 22.613 1.008 1.207 1.004

B. Multivariate Age (years) Initial rhythm (VF-VT/other) Time from collapse to CPR initiation (mins) AaDCO2 (mmHg) (Constant

0.924 8.409 0.762 0.947 331.4)

0.059 0.043 0.034 0.073

0.851 1.066 0.593 0.892

1.003 66.337 0.979 1.005

0.064 0.003 0.027 0.039

−3.577 0.689 −13.679 −0.888

0.013 70.548 0.979 0.008

0.48

C. Multivariate (bootstrap) Age (years) Initial rhythm (VF-VT/other) Time from collapse to CPR initiation (mins) AaDCO2 (mmHg) Table 3 Baseline characteristics and blood gas parameters during CPR and immediately after return of spontaneous circulation.

Baseline characteristics Age (median, IQR) Male sex (%) Bystander CPR (%) VF/VT as initial rhythm (%) Presumed cardiac cause of arrest (%) Blood gas parameters paO2 (mmHg) paCO2 (mmHg) etCO2 a (mm Hg) AaDCO2 a (mmHg) pH BE (mmol l−1 )

Intra-CPR (n = 18)

Post-ROSC (n = 31)

66 (19) 72 78 56

73 (24) 71 87 55

0.229 1.000 0.443 1.000

39

48

0.565

85 (73) 67 (34) 31 (25) 20 (19) 7.07 (0.27) −15.0 (9.0)

128 (117) 58 (21) 37.5 (17.0) 17 (25) 7.17 (0.24) −10.5 (10.0)

p-value

0.046 0.217 0.271 0.895 0.168 0.093

Baseline characteristics and arterial blood gas values of patients during ongoing chest compressions (Intra-CPR) and immediately after return of spontaneous circulation (Post-ROSC). Data are presented as median (interquartile range). Only patients who reached the hospital with sustained ROSC were entered into the analysis. a etCO2 measurement was not available for all patients, thus n = 10 for intra-CPR versus n = 20 for post-ROSC. BE = base excess. Mann–Whitney-U-Test.

were intubated and mechanically ventilated, hypercarbia was still present in many cases. This is most likely due to dead space ventilation caused by reduced lung perfusion during chest compressions, but may also have been the result of hypoventilation, as we have no information about the actual minute volume delivered to the patients. Acidosis of any kind is most likely detrimental to the circulation as it causes peripheral vasodilatation, negative inotropy and impaired oxygen uptake in the lungs, therefore ventilation and carbon dioxide excretion should possibly be paid more attention in OHCPR and be balanced against the adverse circulatory effects of positive pressure ventilation in extreme low-flow states. Also, every effort to monitor and optimize cardiorespiratory function, including volumetric capnography and arterial blood gas analysis should be considered. Recent resuscitation guidelines emphasize that hyperventilation must be avoided,29,30 however, we cannot

confirm hyperventilation reflected by arterial hypocapnia to be highly prevalent during and shortly after ROSC in an OHCPR population. In a multivariate regression analysis none of the ABG parameters alone showed a predictive value for hospital admission, e.g. reaching the hospital alive, as the endpoint. This contrasts our previous observations, where both BE and paO2 were demonstrated to be predictive of sustained ROSC,5,6 but due to the retrospective nature of these studies, the number of data available for analysis was larger. However, the AaDCO2 , a parameter that was not investigated previously in this context, was associated with hospital admission in a multivariate regression model with bootstrapping. The value of the AaDCO2 as a predictor of mortality after hospital admission following CPR has already been demonstrated.31 In a physiologic context, the AaDCO2 may be regarded as a global surrogate marker representative of both cardiorespiratory function and organ perfusion during OHCPR,32,33 whereas probably neither paCO2 nor etCO2 alone adequately reflect the two principal components of CO2 excretion, which are lung perfusion and ventilation. Although etCO2 monitoring is of indisputable value in the prehospital setting with regard to endotracheal tube placement confirmation, it only weakly correlated with the arterial paCO2 and the AaDCO2 in our dataset of CPR patients. It must be noted that although a significance for paO2 in the prediction of HA was not seen as it was in previous retrospective studies, increased paO2 tensions were still associated with higher rates of HA (Fig. 2). In view of the ongoing debate about oxygen toxicity there is still no evidence to support reductions of FiO2 during CPR in our opinion.6,10,15 The comparison of arterial blood gas analyses which were obtained shortly after ROSC to those that were obtained under ongoing OHCPR also revealed several interesting insights. As only data from patients who achieved HA were used for that calculation few patients remained for statistical analysis. Still, paO2 values were significantly higher, which supports experimental data indicating that as soon as ROSC occurs, oxygen uptake into the blood improves dramatically.13 This may be due to increased lung capillary perfusion as the right ventricle starts to actively eject blood into the pulmonary circulation or decreased pulmonary shunting in the absence of impairment of ventilation by chest

28

W. Spindelboeck et al. / Resuscitation 106 (2016) 24–29

compressions. Interestingly, there was a trend towards higher BE as well. It therefore appears that ROSC is not associated with a rapid overload of acidic metabolites from the tissues into the circulation, rather blood pH remains the same or may even improve rapidly. Although this is the first prospective study exploring arterial blood gases during OHCPR, it has important limitations. Our previous investigations demonstrated that in patients where an arterial blood sample could be obtained during ongoing OHCPR, the ROSC rate was as high as 50%.6 In the absence of ultrasound imaging, gaining arterial access requires a palpable pulsatile flow in a peripheral artery and is only possible when a sufficient circulation is generated by chest compressions. Therefore patients with an ABG under ongoing chest compressions inevitably represent a certain positive selection. Similarly, despite the prospective character of the study, there were relatively few patients were an ABG under ongoing OHCPR was actually obtained. Despite the fact that all emergency physicians were senior doctors with experience in intensive care, insertion of an arterial line during chest compressions can be technically difficult and did not always succeed. The decision to commence OHCPR usually has to be made without any information about the patient’s history, so CPR attempts may be more frequent and also prolonged compared to the intrahospital setting34 in patients with long delays and poor prognosis, where no circulation adequate to puncture an artery can ever be reached. On the other hand, the relatively “better” patients may have regained ROSC with basic measures only such that CPR did not last long enough to gain an arterial blood sample. Thus, only eightythree patients had a usable ABG analysis obtained under ongoing OHCPR, and out of these, not enough patients survived to conduct an adequately powered regression analysis regarding favourable neurological outcome. Also, it was not possible to obtain serial ABGs from the same patients, especially before and after ROSC. This limits us to comparing patient groups that survived to HA with an ABG before and after ROSC. Moreover, despite the universal use of etCO2 monitoring, an etCO2 with an exact documented timepoint corresponding to the ABG sample was not available in all cases, however, a regression analysis still revealed the AaDCO2 to be predictive of hospital admission. In summary, our prospectively obtained ABG data demonstrate that severe acidosis occurs during OHCPR, but there is a large and unpredictable respiratory component to it. Hypocapnia due to hyperventilation is not highly prevalent during or shortly after OHCPR and increasing paO2 levels during OHCPR showed a trend towards better survival to hospital admission. ABG analyses obtained immediately after ROSC had higher pO2 values possibly indicating increased oxygen uptake during spontaneous circulation. The AaDCO2 may serve as a global parameter of cardiorespiratory function during OHCPR and was associated with increased rate of sustained ROSC to HA. Conflict of interest statement On behalf of all authors, the corresponding author states that there is no conflict of interest. Acknowledgements The study was supported by the Österreichische Gesellschaft für Notfall- und Katastrophenmedizin (ÖNK, Austrian Society of Emergency and Disaster Medicine), the Österreichische Gesellschaft Anästhesie, Reanimation und Intensivmedizin (ÖGARI, Austrian Society of Anaesthesia, Resuscitation and Intensive Care Medicine) and the Land Steiermark (Regional Government of Southern Austria). None of the above had any role in the study design, data

collection and analysis and in the writing process of the manuscript. The authors thank Prof. Peter Bauer for critical review of the manuscript.

Appendix A. Supplementary data Supplementary data associated with this article can be found, in the online version, at http://dx.doi.org/10.1016/j.resuscitation. 2016.06.013.

References 1. Vukmir RB, Katz L. Sodium bicarbonate improves outcome in prolonged prehospital cardiac arrest. Am J Emerg Med 2006;24:156–61. 2. Vukmir RB. Survival from prehospital cardiac arrest is critically dependent upon response time. Resuscitation 2006;69:229–34. 3. Bar-Joseph G, Abramson NS, Kelsey SF, Mashiaci T, Craig MT, Safar P. Improved resuscitation outcome in emergency medical systems with increased usage of sodium bicarbonate during cardiopulmonary resuscitation. Acta Anaesthesiol Scand 2005:6–15. 4. Prause G, Hetz H, Doppler R. The value of blood gas analysis in out-ofhospital settings. Part 1: (Die Präklinische Blutgasanalyse, Teil 1). Anaesthesist 1998;47:400–5. 5. Farzi S, Hausler F, Wallner S, Spindelböck W, Prause G, Gemes G. Prehospital measurement of arterial base excess and its role as a possible predictor of outcome after out-of hospital cardiac arrest. Emergencias 2013;25: 47–50. 6. Spindelboeck W, Schindler O, Moser A, et al. Increasing arterial oxygen partial pressure during cardiopulmonary resuscitation is associated with improved rates of hospital admission. Resuscitation 2013:8–13. 7. Roberts BW, Kilgannon JH, Chansky ME, Mittal N, Wooden J, Trzeciak S. Association between postresuscitation partial pressure of arterial carbon dioxide and neurological outcome in patients with post-cardiac arrest syndrome. Circulation 2013;127:2107–13. 8. Schneider AG, Eastwood GM, Bellomo R, et al. Arterial carbon dioxide tension and outcome in patients admitted to the intensive care unit after cardiac arrest. Resuscitation 2013;84:927–34. 9. Vaahersalo J, Bendel S, Reinikainen M, et al. Arterial blood gas tensions after resuscitation from out-of-hospital cardiac arrest. Crit Care Med 2014;42: 1463–70. 10. Neumar RW, Otto CW, Link MS, et al. Part 8: adult advanced cardiovascular life support: 2010 American Heart Association Guidelines for Cardiopulmonary Resuscitation and Emergency Cardiovascular Care. Circulation 2010;122: S729–67. 11. Aufderheide TP, Lurie KG. Death by hyperventilation: a common and lifethreatening problem during cardiopulmonary resuscitation. Crit Care Med 2004;32:S345–51. 12. Stiell IG, Brown SP, Nichol G, et al. What is the optimal chest compression depth during out-of-hospital cardiac arrest resuscitation of adult patients? Circulation 2014;130:1962–70. 13. Cavus E, Bein B, Dörges V, et al. Brain tissue oxygen pressure and cerebral metabolism in an animal model of cardiac arrest and cardiopulmonary resuscitation. Resuscitation 2006;71:97–106. 14. Rivers E, Paradis N, Martin G, Goettnig M, Rosenberg J, et al. Cerebral lactate uptake during cardiopulmonary resuscitation in humans. J Cereb Blood Flow Metabol 1991;11:479–84. 15. Schindler O, Gemes G, Spindelboeck W. Oxygen and cardiac arrest: the timepoint matters. Intensive Care Med 2015;41:952. 16. Wildner G, Pauker N, Archan S, et al. Arterial line in prehospital emergency settings – a feasibility study in four physician-staffed emergency medical systems. Resuscitation 2011;82:1198–201. 17. Morrison LJ, Deakin CD, Morley PT, et al. Part 8: advanced life support: 2010 international consensus on cardiopulmonary resuscitation and emergency cardiovascular care science with treatment recommendations. Circulation 2010;122:S345–421. 18. Stewart JS. Management of cardiac arrest with special reference to metabolic acidosis. Br Med J 1964;1:476–9. 19. Wiklund L, Jorfeldt L, Stjernström H, Rubertsson S. Gas exchange as monitored in mixed venous and arterial blood during experimental cardiopulmonary resuscitation. Acta Anaesthesiol Scand 1992;36:427–35. 20. von Planta M, Weil MH, Gazmuri RJ, Bisera J, Rackow EC. Myocardial acidosis associated with CO2 production during cardiac arrest and resuscitation. Circulation 1989;80:684–92. 21. Redding JS, Pearson JW. Resuscitation from ventricular fibrillation. JAMA 1968;203:255–60. 22. Weng Y-M, Wu S-H, Wen-Cheng Li, Kuo C-W, Chen S-Y, Chen J-C. The effects of sodium bicarbonate during prolonged cardiopulmonary resuscitation. Am J Emerg Med 2013;31:562–5. 23. Takasu A, Sakamoto T, Okada Y. Arterial base excess after CPR: the relationship to CPR duration and the characteristics related to outcome. Resuscitation 2007;73:394–9.

W. Spindelboeck et al. / Resuscitation 106 (2016) 24–29 24. Dybvik T, Strand T, Steen P. Buffer therapy during out-of-hospital cardiopulmonary resuscitation. Resuscitation 1995;29:89–95. 25. Bishop RL, Weisfeldt ML. Sodium bicarbonate administration during cardiac arrest. Effect on arterial pH PCO2 , and osmolality. JAMA 1976;235: 506–9. 26. Aufderheide T. Prehospital bicarbonate use in cardiac arrest: a 3 years experience. Am J Emerg Med 1992;10:4–7. 27. Fillmore S, Shapiro M, Killip T. Serial blood gas studies during cardiopulmonary resuscitation. Ann Intern Med 1970;72:465–9. 28. SOS-Kanto Study Group. Comparison of arterial blood gases of laryngeal mask airway and bag-valve-mask ventilation in out-of-hospital cardiac arrests. Circ J 2009;73:490–6. 29. Link MS, Berkow LC, Kudenchuk PJ, et al. Part 7: adult advanced cardiovascular life support. Circulation 2015;132:S444–64.

29

30. Nolan JP, Soar J, Zideman D, et al. European resuscitation council guidelines for resuscitation 2010 section 1, executive summary. Resuscitation 2010;81:1219–76. 31. Moon S-W, Lee S-W, Choi S-H, Hong Y-S, Kim S-J, Kim N-H. Arterial minus endtidal CO2 as a prognostic factor of hospital survival in patients resuscitated from cardiac arrest. Resuscitation 2007;72:219–25. 32. Prause G, Hetz H, Lauda P, Pojer H, Smolle-Juettner F, Smolle J. A comparison of the end-tidal-CO2 documented by capnometry and the arterial pCO2 in emergency patients. Resuscitation 1997;35:145–8. 33. Belpomme V, Ricard-Hibon A, Devoir C, et al. Correlation of arterial PCO2 and PETCO2 in prehospital controlled ventilation. Am J Emerg Med 2005;23:852–9. 34. Buanes E, Heltne JK. Comparison of in-hospital and out-of-hospital cardiac arrest outcomes in a Scandinavian community. Acta Anaesthesiol Scand 2014;58:316–22.