Resuscitation 89 (2015) 1–7

Contents lists available at ScienceDirect

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

Clinical Paper

Association between early arterial blood gas tensions and neurological outcome in adult patients following in-hospital cardiac arrest夽 Chih-Hung Wang a,b , Chien-Hua Huang c , Wei-Tien Chang c , Min-Shan Tsai c , Tsung-Chien Lu c , Ping-Hsun Yu d , An-Yi Wang c , Nai-Chuan Chen e , Wen-Jone Chen c,f,∗ a

Department of Emergency Medicine, National Taiwan University Hospital Yunlin Branch, Yunlin, Taiwan Graduate Institute of Clinical Medicine, College of Medicine, National Taiwan University, Taipei, Taiwan c Department of Emergency Medicine, National Taiwan University Hospital and National Taiwan University College of Medicine, Taipei, Taiwan d Department of Emergency Medicine, Taipei Hospital, Ministry of Health and Welfare, New Taipei City, Taiwan e Department of Emergency Medicine, Tao Yuan General Hospital, Ministry of Health and Welfare, Taoyuan, Taiwan f Department of Emergency Medicine, Lotung Poh-Ai Hospital, Yilan County, Taiwan b

a r t i c l e

i n f o

Article history: Received 23 August 2014 Received in revised form 17 December 2014 Accepted 4 January 2015 Keywords: Heart arrest Cardiopulmonary resuscitation Emergency medicine Critical care Oxygen Carbon dioxide

a b s t r a c t Objective: The early partial pressures of arterial O2 (PaO2 ) and CO2 (PaCO2 ) have been found in animal studies to be correlated with neurological outcome after brain injury. However, the relationship of early PaO2 and PaCO2 to the neurological outcomes of resuscitated patients after cardiac arrest was still not clear. Methods: This was a retrospective observational cohort study in a single medical center. Adult patients who had in-hospital cardiac arrest between 2006 and 2012 and achieved sustained return of spontaneous circulation (ROSC) (ROSC > 20 min without resumption of chest compression) were included. Multivariable logistic regression analysis was used to identify factors associated with favorable neurological outcome at hospital discharge. The first PaO2 and PaCO2 values measured after first sustained ROSC were used for analysis. Results: Of the 550 included patients, 154 (28%) survived to hospital discharge and 74 (13.5%) achieved favorable neurological outcome. The mean time from sustained ROSC to the measurement of PaO2 and PaCO2 was 136.8 min. The mean PaO2 and PaCO2 were 167.4 mmHg and 40.3 mmHg, respectively. PaO2 between 70 and 240 mmHg (odds ratio [OR] 1.96, 95% confidence interval [CI] 1.08–3.64) and PaCO2 levels (OR 0.98, 95% CI 0.95–0.99) were positively and inversely associated with favorable neurological outcome, respectively. Conclusions: The early PaO2 and PaCO2 levels obtained after ROSC might be correlated with neurological outcome of patients with in-hospital cardiac arrest. However, because of the inherent limitations of the retrospective design, these results should be further validated in future studies. © 2015 Elsevier Ireland Ltd. All rights reserved.

1. Introduction Cardiac arrest is a common and lethal emergency condition.1,2 Even when return of spontaneous circulation (ROSC) is achieved, most patients do not survive to hospital discharge.1,2 High postROSC mortality may be attributed to post-cardiac arrest syndrome,

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.2014.12.013. ∗ Corresponding author at: No. 7, Zhongshan S. Rd., Zhongzheng Dist., Taipei City 100, Taiwan. E-mail address: [email protected] (W.-J. Chen). http://dx.doi.org/10.1016/j.resuscitation.2015.01.003 0300-9572/© 2015 Elsevier Ireland Ltd. All rights reserved.

which includes anoxic neurological injury, myocardial dysfunction, and a systemic ischemic/reperfusion response.3 The search for potentially modifiable post-ROSC factors that can improve outcomes has recently focused on the harmful effect that supplemental oxygen has on the pathogenesis of neurological impairment. Results of laboratory studies indicate that reactive oxygen species formed in response to supplemental oxygen lead to elevated brain lipid peroxidation, impaired cerebral oxidative energy metabolism, and increased neurological degeneration.4–7 Besides oxygen, the regulatory function of carbon dioxide8,9 in cerebral vasodilation and vasoconstriction has also gained increasing interests because this function might influence neurological outcome via modulation of cerebral perfusion and tissue oxygen tension.

2

C.-H. Wang et al. / Resuscitation 89 (2015) 1–7

In the only human pilot trial, increased neuronal injuries were observed in post-ROSC patients treated with 100% oxygen.10 Observational studies on the blood levels of oxygen or carbon dioxide and the neurological outcome or survival of post-ROSC patients have had conflicting results,11–17 which might be explained by several reasons. First, there were marked differences on the timing of arterial blood gas levels evaluated by these studies, including using the first,11,14 highest,13 and worst12,15 values during the post-ROSC period. Furthermore, most studies treated oxygen or carbon dioxide tensions as categorical variables with different cutoff points.11−17 Finally, few studies analyzed both oxygen and carbon dioxide tensions in relation to outcome. The guidelines of the American Heart Association (AHA) for the partial pressures of arterial oxygen (PaO2 ) and carbon dioxide (PaCO2 ) in post-ROSC patients were mainly based on animal studies.18 Our retrospective observational study was therefore aimed to address the limitations of previous observational studies and focused on patients of in-hospital cardiac arrest for whom therapeutic hypothermia was less often administered19 to improve the dismal neurological outcome.

2. Methods 2.1. Setting National Taiwan University Hospital is a 2600-bed tertiary medical center with 220 beds in the intensive care units (ICUs). When patients in the general wards suffer from cardiac arrest, a code team is mobilized. A code team consists of a senior resident, several junior residents, a respiratory therapist, a head nurse, and several registered nurses from the ICUs. Each code team member has been certified for advanced cardiac life support and is capable of providing basic and advanced life support according to up-to-date AHA guidelines. A sufficient number of experienced staff is always present in the ICUs; therefore, hospital policy does not mandate mobilization of a code team for ICU patients suffering from cardiac arrest. Resuscitation is performed by the staff of the ICU where the patient is located and by staff from the neighboring ICUs. The Institutional Review Board of National Taiwan University Hospital approved this study and provided a waiver for written informed consent before collection of the data. 2.2. Participants We included patients who had an in-hospital cardiac arrest at the National Taiwan University Hospital between 2006 and 2012. The inclusion criteria were as follows: (1) age ≥ 18 years; (2) documented absence of pulse and performance of chest compression for ≥2 min; and (3) sustained ROSC (ROSC ≥ 20 min without resumption of chest compression). If multiple cardiac arrest events occurred to a patient, only the first event would be recorded. We excluded patients with cardiac arrest related to trauma. We also excluded patients without data on PaO2 or PaCO2 values. 2.3. Data collection We abstracted the following variables: demographics, comorbidities, variables derived from the Utstein template,20 values and time of measurement for the first PaO2 and PaCO2 levels obtained after the first sustained ROSC, post-ROSC interventions (percutaneous coronary interventions, extracorporeal membrane oxygenation, and therapeutic hypothermia), vital signs during the first 24 h following sustained ROSC, and neurological status at

hospital discharge, assessed using the Cerebral Performance Category (CPC) score.21 2.4. Outcome measures The primary outcome was favorable neurological status at hospital discharge, defined as a CPC score of 1 or 2; the secondary outcome was survival to hospital discharge. The CPC score21 is a validated 5-point scale of neurological disability (1, good cerebral performance; 2, moderate cerebral disability; 3, severe cerebral disability; 4, coma/vegetative state; 5, death). The CPC score was developed as a measure of central nervous system function after cardiac arrest and has become the most commonly used outcome assessment tool for this purpose. In this study, the CPC score was retrospectively determined by reviewing medical records for each patient. Patients with a CPC score of 1 or 2 had sufficient cerebral function to live independently. 2.5. Statistical analysis R 2.15.3 software (R Foundation for Statistical Computing, Vienna, Austria) was used for data analysis. Categorical data were expressed as counts and proportions; continuous data were expressed as mean values and standard deviations. Categorical variables were compared using the Fisher’s exact test, and continuous variables were examined using the Wilcoxon rank-sum test. A 2-tailed p-value of ≤ 0.05 was considered statistically significant. Odds ratios (ORs) were obtained using multivariable logistic regression analysis to determine whether independent variables were associated with primary or secondary outcome. Candidate variables without missing values were entered into the regression model regardless of whether they were significant by univariate analysis. The stepwise variable selection procedure (with iterations between the forward and backward steps) was applied to obtain the final regression model. Conservative significance levels for entry and for stay were set at 0.15 to avoid exclusion of potential candidate variables. The final regression model was identified by excluding variables with a p-value > 0.05, one by one, until all regression coefficients were statistically significant. We used generalized additive models (GAM) to identify the nonlinear effects of continuous variables and determine the appropriate cutoff point(s) for discretizing a continuous variable, if necessary, during the stepwise selection procedure of variables. We assessed the goodness-of-fit of the fitted regression model using C-statistics, adjusted generalized R2 , and the Hosmer–Lemeshow test. 3. Results A total of 662 patients met inclusion criteria. Of these, 5 patients were excluded because of trauma-related cardiac arrest, and 107 patients were excluded because of lack of PaO2 or PaCO2 data. The remaining 550 patients were enrolled for further analysis. Tables 1–3 display the pre-, peri-, and postresuscitation characteristics, respectively, of cardiac arrest for all patients in the cohort. The mean age of patients was 65.5 years. Twenty-four hours before cardiac arrest, 298 patients (54.2%) were classified as having favorable neurological status. In the ICUs there were 246 events of cardiac arrest (44.7%), and in the general wards there were 270 events (49.1%). The majority (74.5%) of initial rhythms was non-shockable rhythm. The mean duration of resuscitation was 18.7 min. The mean duration from sustained ROSC to arterial blood gas analysis was 136.8 min. The mean values of PaO2 and PaCO2 were 167.4 mmHg and 40.3 mmHg, respectively. The number of patients with values corresponding to the cutoff points used in previous studies11,14 were as follows: 82 patients (14.9%) with

C.-H. Wang et al. / Resuscitation 89 (2015) 1–7

3

Table 1 Baseline characteristics of the study patients. Variables

Patients with favorable neurological outcome n = 74

Patients without favorable neurological outcome n = 476

Age, y (SDa ) Male, n (%) Comorbidities, n (%) Heart failure, this admission Heart failure, past history Myocardial infarction, this admission Myocardial infarction, past history Arrhythmia Hypotension Respiratory insufficiency Renal insufficiency Hepatic insufficiency Metabolic or electrolyte abnormality Diabetes mellitus Baseline evidence of motor, cognitive, or functional deficits Acute stroke Pneumonia Bacteremia Metastatic Cancer or any blood borne malignancy Favorable neurological status 24 h before cardiac arrest

63.4 (15.7) 54 (73.0)

65.8 (16.6) 284 (59.7)

0.15 0.03

19 (25.7) 15 (20.3) 20 (27.0) 6 (8.1) 15 (20.3) 12 (16.2) 45 (60.8) 25 (33.8) 11 (14.9) 10 (13.5) 25 (33.8) 12 (16.2) 0 (0) 18 (24.3) 4 (5.4) 2 (2.7) 53 (71.6)

83 (17.4) 67 (14.1) 57 (12.0) 14 (2.9) 88 (18.5) 91 (19.1) 343 (72.1) 220 (46.2) 76 (16.0) 97 (20.4) 176 (37.0) 102 (21.4) 31 (6.5) 176 (37.0) 45 (9.5) 96 (20.2) 245 (51.5)

0.11 0.16 0.002 0.04 0.75 0.63 0.06 0.06 1 0.21 0.70 0.36 0.01 0.04 0.38 300 mmHg, 390 patients (70.9%) with PaO2 between 60 and 300 mmHg, 78 patients (14.2%) with PaO2 < 60 mmHg, 106 patients (19.3%) with PaCO2 > 50 mmHg, 291 patients (52.9%) with PaCO2 between 30 and 50 mmHg, and 153 patients (27.8%) with PaCO2 < 30 mmHg. During the initial 24 h following sustained ROSC, 169 patients (30.7%) had fever (body temperature ≥ 38 ◦ C), and 359 patients (65.3%) had hypotension (systolic blood pressure ≤ 100 mmHg). Intubation was performed for all patients with sustained ROSC; 34 patients (6.2%) underwent percutaneous coronary intervention; 62 patients underwent extracorporeal membrane oxygenation (11.3%); and 3 patients (0.5%) underwent therapeutic hypothermia. There were 154 patients (28%) who survived to hospital discharge; of these, 74 patients (13.5%) achieved favorable neurological status.

Except the variables of outcomes, all variables without missing values in Tables 1–3 were put in the regression analysis for variable selection. Table 4 displays the results of the multivariable logistic regression model with favorable neurological status at hospital discharge as the dependent variable. After adjustment for confounding factors, both PaO2 and PaCO2 were found to be significantly associated with neurological outcome. The PaCO2 level was inversely associated with favorable neurological outcome (OR 0.98, 95% confidence interval [CI] 0.95–0.99) (Supplemental Fig. 1). In contrast, GAM revealed that PaO2 had a cap-shaped nonlinear effect on the probability for favorable neurological outcome, indicating that when PaO2 was between 70 and 240 mmHg, the odds for achieving favorable neurological outcome would be greater than 1 [i.e. logit (p) > 0] (Fig. 1). Therefore, PaO2 was dichotomized using these 2 values as cutoff points, and PaO2 between 70 and 240 mmHg

Table 2 Features of cardiac arrest. Variables

Patients with favorable neurological outcome n = 74

Patients without favorable neurological outcome n = 476

p value

Arrest at night, n (%) Arrest on weekend, n (%) Arrest location, n (%) Intensive care unit General ward Monitored status, n (%) Witnessed arrest, n (%) Shockable rhythm, n (%) CPRa duration, min (SDb ) Sodium bicarbonate used in CPR, mEq (SD) Time from sustained ROSCc to ABGd analysis, min (SD) ABG analysis FiO2 e , percent (SD) pH, mean (SD) PaO2 f , mmHg (SD) PaCO2 g , mmHg (SD) Base excess, mmol/L (SD) Lactate, mmol/L (SD)

41 (55.4) 21 (28.4)

288 (60.5) 131 (27.5)

0.45 0.89

40 (54.1) 27 (36.5) 52 (70.3) 56 (75.7) 33 (44.6) 10.3 (9.4) 36.4 (70.1) 147.5 (146.4)

206 (43.3) 243 (51.1) 274 (57.6) 340 (71.4) 107 (22.5) 20.1 (22.0) 74.2 (112.5) 135.1 (144.7)

0.04 0.49