Vasopressin, Epinephrine, and Corticosteroids for In-Hospital Cardiac Arrest

Background  Animal data on cardiac arrest showed improved long-term survival with combined vasopressin-epinephrine. In cardiac arrest, cortisol levels are relatively low during and after cardiopulmonary resuscitation. We hypothesized that combined vasopressin-epinephrine and corticosteroid supplementation during and after resuscitation may improve survival in refractory in-hospital cardiac arrest.

Methods  We conducted a single-center, prospective, randomized, double-blind, placebo-controlled, parallel-group trial. We enrolled 100 consecutive patients with cardiac arrest requiring epinephrine according to current resuscitation guidelines. Patients received either vasopressin (20 IU per cardiopulmonary resuscitation cycle) plus epinephrine (1 mg per resuscitation cycle) (study group; n = 48) or isotonic sodium chloride solution placebo plus epinephrine (1 mg per resuscitation cycle) (control group; n = 52) for the first 5 resuscitation cycles after randomization, followed by additional epinephrine if needed. On the first resuscitation cycle, study group patients received methylprednisolone sodium succinate (40 mg) and controls received saline placebo. Postresuscitation shock was treated with stress-dose hydrocortisone sodium succinate (300 mg daily for 7 days maximum, with gradual taper) (27 patients in the study group) or saline placebo (15 patients in the control group). Primary end points were return of spontaneous circulation for 15 minutes or longer and survival to hospital discharge.

Results  Study group patients vs controls had more frequent return of spontaneous circulation (39 of 48 patients [81%] vs 27 of 52 [52%]; P = .003) and improved survival to hospital discharge (9 [19%] vs 2 [4%]; P = .02). Study group patients with postresuscitation shock vs corresponding controls had improved survival to hospital discharge (8 of 27 patients [30%] vs 0 of 15 [0%]; P = .02), improved hemodynamics and central venous oxygen saturation, and more organ failure–free days. Adverse events were similar in the 2 groups.

Conclusion  In this single-center trial, combined vasopressin-epinephrine and methylprednisolone during resuscitation and stress-dose hydrocortisone in postresuscitation shock improved survival in refractory in-hospital cardiac arrest.

Trial Registration  clinicaltrials.gov Identifier: NCT00411879

The incidence of in-hospital cardiac arrest is 1 to 5 per 1000 patient admissions.¹ Survival to hospital discharge is approximately 20%.¹ Survival after refractory cardiac arrest, ie, refractory ventricular fibrillation/pulseless ventricular tachycardia or asystole/pulseless electrical activity, ranges from 5% to 15%.²

In nonsurvivors of cardiopulmonary resuscitation (CPR), the plasma vasopressin level is lower than in CPR survivors.³ Vasopressin acts directly via V₁ receptors on vascular contractile elements. In cardiac arrest, vasopressin is released as adjunct vasopressor to epinephrine.⁴ Recent animal data showed improved survival and postresuscitation neurologic status after treatment with vasopressin-epinephrine compared with epinephrine alone.⁵ Combination treatment was associated with fewer postresuscitation cardiovascular complications and similar neurologic status relative to vasopressin alone.⁵

Relative to other stress states, cardiac arrest is associated with lower cortisol levels during and after CPR.^4,6,7 Return of spontaneous circulation is associated with plasma cytokine elevation,^4,6 endotoxemia,⁴ coagulopathy,⁴ and adrenal insufficiency contributing to postresuscitation shock.^4,7 Corticosteroid supplementation during and after CPR might confer benefits with respect to hemodynamics, intensity of postresuscitation systemic inflammatory response, and organ dysfunction.^4,7

We hypothesized that, in refractory in-hospital cardiac arrest, treatment with combined vasopressin-epinephrine during CPR and corticosteroid supplementation during and after CPR, compared with epinephrine alone during CPR and no corticosteroid supplementation, may (1) facilitate return of spontaneous circulation, (2) attenuate postresuscitation systemic inflammatory response and cardiac arrest–associated organ injuries, and (3) improve survival to hospital discharge.

We conducted our study in the intensive/coronary care units (ICUs/CCUs), emergency department, general wards, and operating rooms of Evaggelismos Hospital, a tertiary care teaching hospital. The patient eligibility criterion was refractory cardiac arrest, defined as epinephrine requirement for ventricular fibrillation/ventricular tachycardia or asystole/pulseless electrical activity according to the European Resuscitation Council Guidelines for Resuscitation 2005.⁸ Exclusion criteria were age younger than 18 years, terminal illness² or do-not-resuscitate status, cardiac arrest due to exsanguination, cardiac arrest before hospital admission, treatment with intravenous corticosteroids before the cardiac arrest, and previous enrollment in or exclusion from the current study. Consent was not obtained for the CPR-drug combination.² The patients and their families were informed about the trial.⁹ Informed, written next-of-kin consent and nonwritten patient consent (whenever feasible) were obtained for stress-dose hydrocortisone sodium succinate in postresuscitation shock and for blood sampling to determine plasma cytokine levels. The Scientific Council of Evaggelismos Hospital approved the study.

This was a single-center, prospective, randomized, double-blind, placebo-controlled, parallel-group clinical trial. Group allocation was conducted by the director of the hospital's pharmacy with the Research Randomizer (http://www.randomizer.org). Random numbers from 1 to 100 were generated in sets of 4. Each number of each set was unique and was assigned to 1 of the 100 consecutively enrolled patients as his or her code. Vasopressin and methylprednisolone were prepared by the hospital's pharmacy in identical, preloaded 5-mL syringes and placed along with epinephrine ampules in boxes bearing patient codes (for details, see the supplemental material available at http://www.mentzelopoulos-et-al.com). After patient randomization, a box was opened and study drugs were injected intravenously according to protocol. Drug injection was followed by 10 mL of isotonic sodium chloride solution.

Adult inpatients with cardiac arrest induced by ventricular fibrillation/ventricular tachycardia not responsive to 2 defibrillations separated by 2 to 3 minutes of CPR⁸ or patients with asystole/pulseless electrical activity were randomized to receive either combined vasopressin (20 IU per CPR cycle; Monarch Pharmaceuticals, Bristol, Tennessee) and epinephrine (1 mg per CPR cycle; Demo, Athens, Greece) (study group), or isotonic sodium chloride solution placebo and epinephrine (1 mg per CPR cycle) (control group), for the first 5 CPR cycles after randomization. Forty milligrams of methylprednisolone sodium succinate (Pfizer, Athens, Greece) and isotonic sodium chloride solution placebo were administered during the first CPR cycle after randomization to study group and control group patients, respectively. If return of spontaneous circulation was not achieved on completion of the experimental treatment, CPR was continued according to current guidelines.⁸ Our protocol is schematically presented in Figure 1. Experimental drug stability in the syringes was confirmed by high-performance liquid chromatography (see the online supplemental material). Advanced life support was conducted according to current standards⁸ (also described in the online supplemental material).

At 4 hours after resuscitation, surviving study group patients with postresuscitation shock received stress-dose hydrocortisone sodium succinate (300 mg daily for 7 days maximum, and gradual taper; Pfizer).¹⁰ Hydrocortisone was available in vials containing 100 mg of hydrocortisone sodium succinate powder. Each daily dose was diluted in 100 mL of isotonic sodium chloride solution at the hospital's pharmacy and administered to study group patients as a continuous infusion. On vasopressor cessation or on day 8 after cardiac arrest, the daily hydrocortisone sodium succinate dose was consecutively reduced to 200 mg and 100 mg and then discontinued (see the online supplemental material). Control group patients with postresuscitation shock received daily infusions of 100 mL of isotonic sodium chloride solution placebo. Isotonic sodium chloride solution infusion bags bore the patient codes.

Circulatory failure was defined as inability to maintain mean arterial pressure greater than 70 mm Hg without using vasopressors after volume loading.¹⁰Respiratory failure was defined as a ratio of arterial oxygen partial pressure to inspired oxygen fraction of 200 mm Hg or less. Coagulation failure was defined as a platelet count of 50 × 10³/μL or less. Hepatic failure was defined as serum bilirubin concentration of 6 mg/dL (to convert to micromoles per liter, multiply by 17.104) or less. Renal failure was defined as serum creatinine level of 3.5 mg/dL (to convert to micromoles per liter, multiply by 88.4) or greater and/or requirement of renal replacement therapy. Neurologic failure was defined as a Glasgow Coma Score of 9 or less.

After resuscitation, cardiac arrest–induced cardiac and microcirculatory dysfunction lasts approximately 24 hours.⁶ Postresuscitation shock was defined as sustained (>4 hours), new postarrest circulatory failure or postarrest need for at least a 50% increase in any prearrest vasopressor/inotropic support targeted to maintain mean arterial pressure above 70 mm Hg.

Attempts at CPR were documented according to the Utstein style.⁹ Additional data comprised periarrest arterial pressure, gas exchange, electrolyte and lactate levels, vasopressor/inotropic support, and intravenous fluids given. Daily follow-up was conducted by 4 blinded investigators (N.K., S.G., A.P., and A.S.). Follow-up to day 60 after cardiac arrest included medication, organ or system failures, and ventilator-free days. Morbidity and complications throughout ICU/CCU and hospital stay and times to ICU/CCU and hospital discharge were also recorded. Encoded patient data were entered into a database by 2 investigators (N.K. and S.G.) and independently cross-checked by another 2 investigators (A.P. and A.S.). Data were independently scrutinized by a steering committee.

Venipuncture blood samples were obtained on day 0 (at 6 hours after randomization) from the last 35 surviving patients with postresuscitation shock; additional blood samples were obtained on days 1, 3, and 7 after randomization. Serum concentrations of tumor necrosis factor, interleukin (IL)-1β, IL-6, IL-8, and IL-10 were measured by an enzyme-linked immunosorbent assay (Quantikine; R&D Systems Europe Ltd, Abingdon, England) according to manufacturer instructions.

Primary end points were return of spontaneous circulation for 15 minutes or longer and survival to hospital discharge, defined as presence of an attending physician discharge order to home or to a rehabilitation facility. Secondary end points were arterial pressure during and 15 to 20 minutes after CPR, intensity of postarrest systemic inflammatory response, number of organ failure–free days until completion of follow-up, and cerebral performance according to the Glasgow-Pittsburgh scale at hospital discharge (see online supplemental material for details on determination of end points).

Initial rhythm is asystole in 75% to 80% of the refractory cardiac arrests occurring in our hospital. Sample-size calculation (G*Power version 3.0.8; Heinrich Heine University, Düsseldorf, Germany) was based on a possible, drug-related, overall 3.1-fold improvement in survival to hospital discharge of the study group vs the control group. Survival improvement was expected mainly for patients with asystole.⁹ Thus, our overall prediction was equivalent to an experimental treatment–induced 3.8-fold rise in the survival of patients with asystole. This corresponds to an improvement of 22.6% relative to a recently reported vasopressin-induced 3.1-fold rise in survival after asystolic cardiac arrest.⁹ Predicted overall survival of the control group was 5%.² Calculated χ² effect size was 0.34. For an α value of .05 and a power of 0.80, the estimated sample size was 68 (ie, 34 patients per group). The inclusion of 100 patients resulted in a safety margin of 32 of 68 (47%).

An intention-to-treat analysis was conducted with SPSS version 12.0 statistical software (SPSS Inc, Chicago, Illinois). Data are reported as mean (SD), median (interquartile range [IQR]), or number (percentage), unless otherwise specified. Distribution normality was tested by the Kolmogorov-Smirnov test. Dichotomous and categorical variables were compared by the χ² or Fisher exact test. Continuous variables were compared by a 2-tailed, independent-samples t test or the Mann-Whitney exact test.

In postresuscitation shock, we used linear mixed-model analysis to determine the overall effects of group, time, and their interaction (group × time) on log-transformed plasma cytokine concentrations throughout the first 7 days after randomization. The effects of group, time, and group × time on (1) average daily central venous oxygen saturation and arterial blood lactate (measured every 12 hours), mean arterial pressure (recorded every 3 hours), and infusion rates of vasopressors; (2) daily fluid balance; and (3) hemoglobin concentration (measured every 24 hours) were also analyzed for the first 10 days after randomization. Fixed-effects significance was determined by the F test. Model selection was based on the minimum values of −2 restricted log-likelihood and Akaike information criteria. Between-group comparisons at individual, consecutive time points were conducted with the independent-samples t test; P values were not corrected for multiple comparisons.

Survival was analyzed by the Kaplan-Meier method, and survival data were compared by (1) the Fisher exact test to determine any nonrandom association between group and survival to hospital discharge and (2) the log-rank test to test the null hypothesis that the probability of death did not differ between the study and control groups throughout patient follow-up. Univariate and multivariate backward-stepwise Cox regression analysis was used to identify independent predictors of death and to determine the respective proportional hazards and their 95% confidence intervals. Variable entry and removal criteria were 0.05 and 0.10, respectively. Reported P values are 2 sided. Statistical significance was set at P < .05.

From June 8, 2006, to March 16, 2007, there were 139 potentially eligible patients with cardiac arrest. Overall survival to hospital discharge was 37 of 139 patients (26.6%). Thirty-nine patients were excluded and 100 patients (52 in the control group and 48 in the study group) were enrolled (Figure 2). Patient encoding was disclosed to one of us (S.D.M.) on April 9, 2007 (hospital discharge date for the last surviving patient). Data from the first 50 patients enrolled were independently analyzed by the steering committee on December 13, 2006. This interim analysis established study safety and proper working of randomization.

Table 1 displays baseline patient characteristics and causes of cardiac arrest. Study group patients vs control group patients had significantly higher rates of return of spontaneous circulation for 15 minutes or longer (39 of 48 patients [81%] vs 27 of 52 [52%]; P = .003) (Table 2). In the study group, average mean arterial pressure during CPR (determined only in the ICU/CCU among patients with an arterial line in place) and 15 to 20 minutes after CPR (determined in all CPR survivors) was higher by 32.1% (P = .009) and 25.9% (P = .02), respectively (Table 3).

At 4 hours after resuscitation, 27 of 29 surviving study group patients and 15 of 20 surviving controls had postresuscitation shock and were assigned to stress-dose hydrocortisone and isotonic sodium chloride solution placebo, respectively (Figure 2). Within 12 hours after cardiac arrest, all surviving patients were in the ICU or CCU. Survival to hospital discharge was significantly higher in the study group than in the control group (9 of 48 [19%] vs 2 of 52 [4%]; P = .02 by Fisher exact test) (Figure 3A; P = .003 by log-rank test). Multivariate Cox regression analysis showed that independent risk factors for death were assignment to study group and completion of a full postarrest course of hydrocortisone according to protocol (relative risk, 0.15; 95% confidence interval, 0.06-0.38; P < .001), periarrest lactate level (see also Table 3, footnote e) (1.07; 1.02-1.11; P = .003), and successful resuscitation after 3 or fewer CPR cycles (0.49; 0.29-0.83; P = .008). Post hoc analysis showed that rapid (ie, requiring ≤3 CPR cycles) successful resuscitation was more frequent in the study group than in the control group (22 of 48 [46%] vs 11 of 52 [21%], P = .01).

Full follow-up data were obtained for all CPR survivors. In those who survived for 4 hours or longer, prescribed medication was similar, except that vasopressor use throughout follow-up was significantly lower in the study group (P = .002) (see the online supplemental materials). Median (IQR) days free of all organ failure were 0.0 (0.0-36.0) and 0.0 (0.0-0.0) in the study group and control group, respectively (P = .27). Postarrest morbidity, complications, and causes of death were similar in both groups (Table 4). For the 11 long-term survivors, ICU/CCU and hospital discharge occurred at a mean (SD) of 37.6 (27.4) and 58.8 (31.2) days after arrest, respectively.

Study group patients with postresuscitation shock vs corresponding controls had significantly improved survival to hospital discharge (8 of 27 [30%] vs 0 of 15 [0%]; P = .02 by Fisher exact test) (Figure 3B; P = .01 by log-rank test), a trend toward significantly more all–organ failure–free days (median [IQR], 0.0 [0.0-32.0] vs 0.0 [0.0-0.0]; P = .06), and significantly more renal failure–free days (3.0 [1.0-59.0] vs 0.0 [0.0-5.0]; P = .03). Study group patients who completed a full course of hydrocortisone according to protocol (n = 12) vs corresponding controls (n = 6) had significantly more days free of all organ failure and circulatory, neurologic, hepatic, renal, coagulation, and respiratory failure (P = .001 to P = .04) (Figure 3C).

Linear mixed-model analysis showed significant effects of group on log-transformed plasma IL-6 levels (P < .001), central venous oxygen saturation (P < .001), and mean arterial pressure (P < .001). There was a significant effect of group × time on central venous oxygen saturation (P = .007). There was a time-dependent decrease in arterial blood lactate levels (P < .001), daily norepinephrine infusion rate (P = .004), and positivity of daily fluid balance (P = .001) (see the online supplemental material).

Plasma IL-6 level was significantly lower in the study group than in the control group throughout the first week after randomization (P = .002 to P = .02). Six hours after randomization, plasma tumor necrosis factor level was significantly lower (P = .04), and plasma IL-1β level exhibited a trend toward significantly lower values (P = .06) in the study group (Figure 3D). Central venous oxygen saturation and mean arterial pressure were significantly higher in the study group than in the control group throughout the first 10 days (P < .001 to P = .04) and at days 2, 4, and 10 (P = .006 to P = .03) after randomization, respectively (Figures 3E and 3F). Arterial oxygen saturation, hemoglobin concentration, and dobutamine and epinephrine daily infusion rates were similar in the 2 groups (data not shown).

Additional analyses are presented in the online supplementary material at http://www.mentzelopoulos-et-al.com. Prearrest physiologic disturbances and medication had similar distributions in the 2 groups. Four study group patients (8%) and 4 controls (8%) with acute coronary syndromes received periarrest revascularization therapy⁸ (P > .99).

Post hoc analyses were conducted according to use or no use of additional epinephrine during resuscitation (Figures 1 and 2). In the additional-epinephrine subgroup, return of spontaneous circulation for 15 minutes or longer was significantly more frequent in study group patients than in controls (9 of 17 [53%] vs 6 of 29 [21%]; P = .048). Two (12%) of the study group patients survived to hospital discharge, 1 with moderate and 1 with severe cerebral disability; all 29 controls died before hospital discharge. After exclusion of 1 study group patient and 2 controls, the subgroup without additional epinephrine included 51 successfully resuscitated patients. During resuscitation, the 30 study group patients had a significantly greater total number of “potentially reversible” major disorders (eg, hypoxemia, hyperkalemia, hypovolemia)⁸ per patient than the 21 controls (median [IQR], 1.0 [0.8-2.0] vs 0.0 [0.0-1.0]; P = .01) (Table S3 of the online material). Seven (23%) of the 30 study group patients (6 with good cerebral performance and 1 with moderate cerebral disability) and 2 (10%) of the 21 controls (both with good cerebral performance) survived to hospital discharge.

Within the first 10 days after randomization, blood glucose level was 201 mg/dL or more (to convert to millimoles per liter, multiply by 0.0555) in 325 of 1098 (29.6%) and 181 of 678 (26.7%) ICU/CCU chart recordings in the study and control groups, respectively (P = .19). Relative to controls, study group patients had more hyperglycemic episodes on days 2 and 3 (P < .001 and P = .01, respectively). Those who survived for more than 48 hours in the study group (n = 19) and control group (n = 12) developed a median (IQR) of 0.0 (0.0-2.0) and 0.0 (0.0-1.0) ICU-associated infectious complications, respectively (P = .64); ventilator-free days were 0.0 (0.0-42.0) and 0.0 (0.0-0.0), respectively (P = .21). Six study group patients (32%) and 3 controls (25%) underwent tracheostomy after weaning and/or extubation failure (P > .99). Among those who survived for 10 days or longer, paresis¹¹ was noted in 4 of 13 study group patients (31%) and 2 of 6 controls (33%) (P > .99).

The conduct of this study could constitute a change in the working conditions of resuscitation teams and ICU/CCU physicians and staff. This might result in enhanced productivity and improved patient outcomes (Hawthorne effect).¹² To investigate this possibility, after study completion, we retrospectively analyzed CPR and postarrest data from 93 consecutively identified patients who (1) received advanced life support⁸ for refractory in-hospital cardiac arrest within the period from December 1, 2005, to May 31, 2006; (2) fulfilled the present study's enrollment criteria; and (3) were not assigned to the experimental arm of any ongoing trial. Data were collected from CPR records of the Department of Anesthesiology and from patient records and ICU/CCU charts retrieved from the hospital's archive. Data collection was conducted by 2 independent reviewers blinded to the objectives of the analysis.

Historical controls and actual control and study group patients had similar characteristics and causes of cardiac arrest (data not shown). Within 12 hours after arrest, all successfully resuscitated and surviving historical controls were admitted to the ICU or CCU. Regarding the primary end points, historical controls vs the study group had a significantly lower rate of return of spontaneous circulation for 15 minutes or longer (47 of 93 patients [51%] vs 39 of 48 [81%]; P < .001); this rate was similar to the rate of the actual control group (P > .99). Survival to hospital discharge was also similar in historical controls and the actual control group (Figure 4A) and significantly lower in historical controls than in the study group (Figure 4B). Thus, there was no Hawthorne effect on the primary outcomes of this trial.

The findings of this single-center study constitute the first evidence, to our knowledge, of increased efficacy of adding vasopressin and methylprednisolone to epinephrine during CPR and treating postresuscitation shock with stress-dose hydrocortisone.

Methylprednisolone was chosen for initial treatment because it enhances the contractile function of both the heart during and after myocardial ischemia¹³ and the peripheral arteries during endotoxemia.¹⁴ Myocardial dysfunction¹⁵ and sepsislike vasoplegia⁶ are key components of early postresuscitation shock.^6,15 The early cardiovascular effects of the methylprednisolone dose used may be partly nongenomic^16,17 and are expected within 30 to 60 minutes after administration.^16,17 Thus, the results on arterial pressure during CPR (Table 3) are explained mainly by the combined and simultaneous vasopressin-epinephrine action. Increased mean arterial pressure suggests improved coronary perfusion,¹⁸ facilitating restoration of spontaneous cardiac rhythm. This explains the more frequent return of spontaneous circulation.^5,19

Hydrocortisone was chosen for postresuscitation shock for its vascular^17,20 and immune^21,22 modulatory effects. In postresuscitation shock, study group results on cytokine levels indicate attenuation of the systemic inflammatory response. Furthermore, mean arterial pressure was higher during the early and late postresuscitation periods (Table 3 and Figure 3F). Central venous oxygen saturation was also higher for more than 72 hours after resuscitation (Figure 3E). These results indicate improved hemodynamics and peripheral oxygen supply-demand balance²³ and can thus explain the observed increase in organ failure–free days and improved survival in this severe sepsislike syndrome.^6,23-25

According to post hoc analysis, our new CPR-drug combination resulted in a 2.2-fold increase in the frequency of rapid successful resuscitation. This was associated with halving of the death risk, thus implying an additional potential mechanism for survival improvement. Also, the treatment of postresuscitation shock with a full course of hydrocortisone resulted in a 6.7-fold reduction of death risk, suggesting combined benefit of vasopressin-epinephrine and corticosteroids in refractory cardiac arrest followed by postresuscitation shock.

The use of postarrest therapeutic hypothermia was limited mainly to ventricular fibrillation cardiac arrest⁸ and was similar in the control group and the study group (15% vs 18% of successfully resuscitated patients; P > .99). Finally, our results are most likely generalizable because (1) our experimental treatment comprises the addition of widely available and widely used drugs during and after CPR; (2) the studied population had a broad case mix with primarily cardiovascular disease (Table 1)²; and (3) major periarrest factors (ie, frequency of primary cardiac causes of cardiac arrest and of witnessed arrest, resuscitation team response times, and leading initial cardiac rhythm) were similar in this trial and a preceding 3-center trial of in-hospital cardiac arrest.²

Thirty study group patients (as opposed to just 21 controls) were successfully resuscitated without additional epinephrine. This could be regarded as a between-group imbalance biasing the study results. However, our post hoc analyses showed that, during advanced life support, the study group patients who received no additional epinephrine had more potentially reversible major disorders⁸ than did controls. These disorders (eg, hypoxemia, hyperkalemia, and hypovolemia) are considered causes of failed or prolonged resuscitation.⁸ Consequently, the aforementioned imbalance was probably due to a more rapid and favorable response of more severely ill study group patients to a superior treatment.

The contribution of the subgroup who received additional epinephrine to the positive study results was relatively minor: only 2 of 17 study group patients (12%) survived, with moderate to severe neurologic deficits. For this subgroup (n = 46), the determination of an experimental treatment–related rise in survival from 2% to 8% (with α = .05 and power = 0.80) would require 86 patients or more, corresponding to a total study population of more than 180.

Results could have been similar if hydrocortisone had been used instead of methylprednisolone during CPR. We chose methylprednisolone on the basis of contemporary literature.¹³ Finally, for reasons of protocol feasibility, we did not determine baseline stress hormone concentrations.

In conclusion, the results of this trial suggest that the combined use of vasopressin, epinephrine, and corticosteroids may improve by a factor of 4.5 the long-term survival after refractory in-hospital cardiac arrest. This result is supported and explained by the more frequent successful resuscitation, increased postarrest mean arterial pressure and central venous oxygen saturation, and attenuated postarrest systemic inflammatory response and organ dysfunction in the study group.

Correspondence: Spyros D. Mentzelopoulos, MD, PhD, Department of Intensive Care Medicine, Evaggelismos General Hospital, 45-47 Ipsilandou St, GR-10675 Athens, Greece (sdmentzelopoulos@yahoo.com).

Accepted for Publication: June 8, 2008.

Author Contributions: All authors had full access to all the data in the study and take responsibility for the integrity of the data and the accuracy of the data analysis. Dr Mentzelopoulos was the principal investigator; Dr Zakynthinos, the study director; and Dr Roussos, the study chair. Study concept: Mentzelopoulos. Study design: Mentzelopoulos, Zakynthinos, and Roussos. Acquisition of data: Katsios, Papastylianou, Gkisioti, Stathopoulos, Kollintza, and Stamataki. Analysis and interpretation of data: Mentzelopoulos, Zakynthinos, and Roussos. Drafting of the manuscript: Mentzelopoulos. Critical revision of the manuscript for important intellectual content: all authors. Statistical analysis: Mentzelopoulos and Tzoufi. Obtained funding: Mentzelopoulos and Zakynthinos. Administrative, technical, and material support: Mentzelopoulos, Zakynthinos, Stamataki, and Roussos. Study supervision: Mentzelopoulos, Zakynthinos, and Roussos.

Financial Disclosure: None reported.

Funding/Support: This study was funded by the Thorax Research Foundation, Athens, Greece, and the Greek Society of Intensive Care Medicine. All funding was used for the acquisition of necessary laboratory materials.

Previous Presentations: This study was presented in part at the 11th State-of-the-Art Interdisciplinary Review Course; April 21, 2007; Athens, Greece; and at the 20th Annual Congress of the European Society of Intensive Care Medicine; October 9, 2007; Berlin, Germany. The abstract of the latter presentation was published in Intensive Care Med. 2007;33(suppl 2):S186.

Additional Contributions: Christina Sotiropoulou, PhD (Thorax Research Foundation, University of Athens Medical School), acted as the study's statistical supervisor and John Portolos, PhD (Director of Pharmacy, Evaggelismos General Hospital), as the study's pharmacist; Marinos Pitaridis, MD, and Vassiliki Markaki, MD (First Department of Intensive Care Medicine, University of Athens Medical School), and Sotirios Malachias, MD (Department of Anesthesiology, Evaggelismos General Hospital), constituted the independent main end point and safety monitoring committee; John Portolos, PhD, Marinos Pitaridis, MD, and Sotirios Malachias, MD, provided quality assurance and data management; and John Koutsourelakis, MD, and Sotiris Sourlas, MD (First Department of Intensive Care Medicine, University of Athens Medical School), performed data acquisition for the retrospective analyses.