Resuscitation 83 (2012) Contents lists available at SciVerse ScienceDirect. Resuscitation

Transcription

1 Resuscitation 83 (2012) Contents lists available at SciVerse ScienceDirect Resuscitation jo u rn al hom epage : Clinical Paper Outcome when adrenaline (epinephrine) was actually given vs. not given post hoc analysis of a randomized clinical trial Theresa M. Olasveengen a,, Lars Wik b, Kjetil Sunde c, Petter A. Steen d a Department of Anaesthesiology and Institute for Experimental Medical Research, Oslo University Hospital, PB 4956 Nydalen, N-0424 Oslo, Norway b Department of Anaesthesiology and National Centre for Prehospital Emergency Medicine, Oslo University Hospital, PB 4956 Nydalen, N-0424 Oslo, Norway c Department of Anaesthesiology, Oslo University Hospital, PB 4956 Nydalen, N-0424 Oslo, Norway d University of Oslo, Faculty Division OUH and Ambulance Department, Oslo University Hospital, PB 4956 Nydalen, N-0424 Oslo, Norway a r t i c l e i n f o Article history: Received 29 August 2011 Received in revised form 15 November 2011 Accepted 15 November 2011 Keywords: Advanced Life Support (ALS) Cardiac arrest Cardiopulmonary resuscitation (CPR) Chest compression Emergency medical services Out-of-hospital CPR Outcome Drugs a b s t r a c t Purpose of the study: IV line insertion and drugs did not affect long-term survival in an out-of-hospital cardiac arrest (OHCA) randomized clinical trial (RCT). In a previous large registry study adrenaline was negatively associated with survival from OHCA. The present post hoc analysis on the RCT data compares outcomes for patients actually receiving adrenaline to those not receiving adrenaline. Materials and methods: : Patients from a RCT performed May 2003 to April 2008 were included. Three patients from the original intention-to-treat analysis were excluded due to insufficient documentation of adrenaline administration. Quality of cardiopulmonary resuscitation (CPR) and clinical outcomes were compared. Results: Clinical characteristics were similar and CPR quality comparable and within guideline recommendations for 367 patients receiving adrenaline and 481 patients not receiving adrenaline. Odds ratio (OR) for being admitted to hospital, being discharged from hospital and surviving with favourable neurological outcome for the adrenaline vs. no-adrenaline group was 2.5 (CI 1.9, 3.4), 0.5 (CI 0.3, 0.8) and 0.4 (CI 0.2, 0.7), respectively. Ventricular fibrillation, response interval, witnessed arrest, gender, age and endotracheal intubation were confounders in multivariate logistic regression analysis. OR for survival for adrenaline vs. no-adrenaline adjusted for confounders was 0.52 (95% CI: 0.29, 0.92). Conclusion: Receiving adrenaline was associated with improved short-term survival, but decreased survival to hospital discharge and survival with favourable neurological outcome after OHCA. This post hoc survival analysis is in contrast to the previous intention-to-treat analysis of the same data, but agrees with previous non-randomized registry data. This shows limitations of non-randomized or non-intention-totreat analyses Elsevier Ireland Ltd. All rights reserved. 1. Introduction Drugs like epinephrine and amiodarone are still recommended in the current international guidelines of Advanced Life Support (ALS) during cardiac arrest, 1,2 although their outcome benefit on survival to hospital discharge is debated. Especially, the use of adrenaline is questioned. 3 In intention-to-treat analysis in a randomized controlled out-of-hospital cardiac arrest (OHCA) study, the intravenous line insertion and drug administration group had improved short term outcome without improved survival rate to A Spanish translated version of the summary of this article appears as Appendix in the final online version at doi: /j.resuscitation Corresponding author at: Department of Anaesthesiology, Oslo University Hospital, PB 4956 Nydalen, N-0424 Oslo, Norway. Tel.: / ; fax: address: t.m.olasveengen@medisin.uio.no (T.M. Olasveengen). hospital discharge. 4 A before and after Canadian study of OHCA similarly failed to find any change in outcome after introducing intravenous drug administration and endotracheal intubation into local resuscitation protocols. 5 In a large Swedish OHCA registry study, which included patients where some ambulance personnel were allowed to give adrenaline and some were not, patients receiving adrenaline were 57% less likely to be alive after one month in a multivariate logistic regression analysis adjusting for all known confounders compared to those who had not received adrenaline. 6 This apparent contrast in results could be due to various factors. Non-randomized registry studies do not intend to prove causality, and there might be unknown factors not adjusted for in regression analysis. Patients with rapid return of spontaneous circulation (ROSC) such as those with ventricular fibrillation (VF) and ROSC after the first defibrillation attempt might never have had time for adrenaline injection. These patients, with very good prognosis, would thereby end in the no-drug group /$ see front matter 2011 Elsevier Ireland Ltd. All rights reserved. doi: /j.resuscitation

2 328 T.M. Olasveengen et al. / Resuscitation 83 (2012) impacting on the data interpretation. Our randomized study was analyzed on an intention-to-treat basis. 4 As expected; some patients in the intravenous group had achieved ROSC before adrenaline could be given, while some in the no-intravenous group received adrenaline for different reasons. For example, it was permitted to place the IV line 5 min after ROSC. If re-arrest occurred, adrenaline could be administered if indicated by the CPR guidelines. 7 Non-randomized, observational registry data from before and after studies are often used to explore therapeutic issues in cardiac arrest. These studies always compare when a certain therapy was actually administered vs. when it was not. Although many confounding factors may be identified in clinical registry data, significant unknown factors may exist. We have therefore performed a post hoc analysis of our previously published data 4 comparing outcomes for patients who actually received adrenaline to those who did not, and included a multivariate regression analysis as in the Swedish registry study Materials and methods We conducted a prospective cohort study using clinical trial population. 4 The randomized trial was designed to evaluate the effect of intravenous access and medication in cardiac arrest resuscitation Description of emergency medical services (EMS) and in-hospital treatment Until January 2006 the one-tiered Oslo EMS system with paramedics and one physician-manned ambulance followed the International ALS Guidelines with the modification that patients with VF received 3 min of CPR before first shock and between unsuccessful series of shocks. 9 Guidelines were implemented January 2006 with the same modification of 3 min periods of CPR. Manual mode defibrillation and endotracheal intubation were standard procedures. Two ambulances are routinely dispatched when cardiac arrest is suspected. The physician manned ambulance is dispatched whenever available. All hospitals in Oslo have goal directed post-resuscitation protocols including therapeutic hypothermia for all actively treated patients. 10 If coronary angiography is indicated from prehospital 12 lead electrocardiogram (ECG) for possible percutaneous coronary intervention (PCI), patients are transported to one of two university hospitals with this capacity 24 h/day Study design and recruitment All patients older than 18 years with non-traumatic, nonambulance witnessed OHCA of all causes in Oslo between May 1st 2003 and April 28th 2008 were randomized to receive either ALS with intravenous access and drugs (IV group) or ALS without intravenous access and drugs (No-IV group). Intravenous access was to be established 5 min after ROSC regardless of randomization, and drugs could then be given if indicated. In the present post hoc study patients who actually received adrenaline were allocated to the adrenaline group and those who did not receive adrenaline were allocated to the no adrenaline group using information from ambulance run sheets. The original study was registered at clinicaltrials.gov (NCT ) and approved by the regional ethics committee. Informed consent for inclusion was waived as decided by this committee, but was required from survivors with 1-year follow-up Equipment and data collection Standard LIFEPAK 12 defibrillators (Physio-Control, Medtronic, Redmond, WA, USA) were used, and ECGs with transthoracic impedance signals were transferred to a hospital server. Data were documented according to Utstein style as abstracted from Utstein cardiac arrest forms, 11 ambulance run sheets and hospital records. Automated, computer based dispatch centre time records supplement ambulance run sheets with regards to response intervals. Administrations of all relevant intravenous drugs including doses administered were documented on the ambulance run sheets. Primary end point was survival to hospital discharge. Secondary endpoints were return of spontaneous circulation, survival to hospital admission, and neurologic outcome at hospital discharge for survivors (using cerebral performance categories 1 4). 11 The study was externally monitored Data processing Ventilations and chest compressions were determined from changes in transthoracic impedance using CODE-STAT TM 7.0 (Physio-Control, Redmond, WA, USA) and added to the written information from patient report forms. Time without spontaneous circulation, time without compressions during time without spontaneous circulation (hands-off time), pre-shock pauses, compression rate and actual number of compressions and ventilations per minute were calculated for each episode. Hands-off ratio is defined as hands-off time divided by total time without spontaneous circulation Statistical analysis Demographic and clinical data are presented as means ± standard deviation (SD), medians with 25 percentile and 75 percentile (25p, 75p) or proportions. Crude associations between the two trial arms and survival were quantified by odds ratio (OR) with 95% confidence interval (CI). For continuous variables, Student s t test was used for normally distributed data and Mann Whitney U test for not normally distributed data. Potential confounders were identified and adjusted for using logistic regression. An investigation of the correlation between potential confounders was performed. P-values less than 0.05 were considered significant 3. Results 3.1. Patient allocation (Fig. 1). Resuscitation was attempted in 1183 patients, and 851 of 946 those eligible were successfully randomized. Thirty-seven of 433 patients randomized to the No-IV arm received adrenaline and 85 of 418 patients randomized to the IV arm did not receive adrenaline. Three cases were excluded as we were unable to determine drug administration, leaving 367 patients in the adrenaline group and 481 patients in the no-adrenaline group. Table Baseline demographics and CPR quality (Table 1). There was no difference in proportion of patients presenting with VF between the two post hoc study groups, while fewer patients presenting with asystole received adrenaline (OR 0.7 CI 0.6, 1.0 vs. the no-adrenaline group) and more patients presenting with PEA tended to receive adrenaline (OR 1.4 CI 1.0, 2.0 vs. the noadrenaline group). Quality of CPR was similar and within guideline

3 T.M. Olasveengen et al. / Resuscitation 83 (2012) Table 1 Demographics and quality of cardiopulmonary resuscitation (CPR). No Adrenaline (n = 481) Adrenaline (n = 367) Odds ratio (95% CI) p-value Age (years) 66 (54, 78) 66 (54, 78) 0.45 Males (%) 337 (70) 266 (73) 1.1 (0.8, 1.5) Cardiac aetiology (%) 337 (70) 265 (72) 1.1 (0.8, 1.5) 0.55 Location of arrest Home 268 (56) 205 (56) 1.0 (0.8, 1.3) 1.00 Public 173 (36) 129 (35) 1.0 (0.7, 1.3) 0.86 Other 38 (8) 33 (9) 1.2 (0.7, 1.9) 0.66 Bystander witnessed 313 (65) 241 (66) 1.0 (0.8, 1.4) 0.91 Bystander BLS 300 (62) 233 (64) 1.1 (0.8, 1.4) 0.79 Initial Rhythm VF/VT 156 (32) 128 (35) 1.1 (0.8, 1.5) 0.50 Asystole 253 (53) 166 (46) 0.7 (0.6, 1.0) 0.04 PEA 72 (15) 73 (20) 1.4 (1.0, 2.0) 0.07 Physician manned ambulance present 171 (36) 145 (40) 0.28 Response interval (min) 9 (7, 12) 10 (7, 12) 0.88 Intubation 387 (81) 341 (93) 3.2 (2.0, 5.0) < Other IV drugs during resuscitation -Atropine 10 (2) 204 (56) 59.7 (30.9, 115.4) < Amiodarone 9 (2) 77 (21) 14.0 (6.9, 28.4) <0.001 Defibrillation 171 (36) 182 (50) 1.8 (1.4, 2.4) <0.001 No. of Shocks 0 (0,1) 1 (0, 3) <0.001 No. of Shocks when defibrillated 2 (1, 4) 3 (2, 6) <0.001 ECG available for analysis 352 (73) 288 (79) CPR duration (min) 16 (11, 21) 25 (15, 31) < Hands off ratio.16 (.09,.023).17 (.09,.23) Compression rate 117 (110, 122) 118 (111, 124) Compressions min 1 95 (86, 103) 94 (87, 104) Ventilations min 1 11 (8, 13) 11 (9, 14) Pre-shock pause (sec) 12 (3, 18) 12 (3, 21) 0.35 VF = ventricular fibrillation. VT = pulseless ventricular tachycardia, PEA = pulseless electrical activity, IV = intravenous, ECG = electrocardiogram, Hands-off = proportion of time without chest compressions during the resuscitation effort, compression rate = rate of compressions when delivered, compressions min 1 = average number of compressions actually given per minute during resuscitation effort. All categorical variables given as numbers (percentages in parenthesis). Continuous variables are given as medians with interquartile range. Differences between groups were analyzed using chi-squared with continuity correction and odds ratio with 95% confidence intervals for categorical data, and Students t-tests or Mann Whitney U-tests for continuous data as appropriate. recommendations for both groups, but patients in the adrenaline group were more likely to be intubated (OR 3.2 CI 2.0, 5.0) and defibrillated (OR 1.8 CI 1.4, 2.4) and had longer resuscitation efforts (25 vs. 16 min, p < 0.001) than those not receiving adrenaline In-hospital treatment and outcome (Tables 2 4). Patients in the adrenaline group were more likely to be admitted to hospital and an intensive care unit compared to the no-adrenaline group (OR 2.5 CI 1.9, 3.4 and OR 1.4 CI 1.0, 1.9, respectively). In the adrenaline group only one patient was awake on hospital admission compared to 13% in the no-adrenaline group (p = 0.002). Patients in the adrenaline group received angiography/percutaneous coronary intervention less often than the no-adrenaline group (OR 0.1 CI 0.0, 0.5 and OR 0.4 CI 0.2, 0.7, respectively). Regarding outcome, patients in the adrenaline group were less likely to be discharged from hospital, less likely to be discharged with favourable neurological outcome and less likely to be alive one year after cardiac arrest (odds ratios for adrenaline vs. no-adrenaline groups were 0.5 CI 0.3; 0.8, 0.4 CI 0.2, 0.7; and 0.5 CI 0.3, 0.8, respectively). (Table 2) Patients were divided into two predefined subgroups based on their initial rhythms (Table 3). Patients in the adrenaline group presenting with VF or pulseless VT were less likely to be admitted to intensive care (OR 0.6 CI 0.4, 1.0), discharged alive (OR 0.3, CI 0.2, 0.6), and discharged with favourable neurological outcome (OR 0.3, CI 0.1, 0.5) compared to the no-adrenaline group. Patients in the adrenaline group presenting with asystole or PEA were more likely to achieve ROSC (OR 5.1 CI 3.2, 8.1) and be admitted to an intensive care unit (OR 3.3 CI 2.0, 5.4), but were no more likely to be discharged alive compared to the no-adrenaline group (OR 0.9 CI 0.3, 3.2). (Table 3). Ventricular fibrillation, response interval, witnessed arrest, gender, age and endotracheal intubation were identified as confounding factors in the multivariate logistic regression analysis. After adjusting for confounders, patients in the adrenaline group had a 48% lower chance of survival with adjusted odds ratio 0.52 (CI 0.29, 0.92) compared to the no-adrenaline group. (Table 4). 4. Discussion The results from our previously published randomized controlled trial of intravenous access and drugs administration during OHCA would suggest that an EMS system opting to include intravenous drug administration in their cardiac arrest treatment protocol could expect increased survival to hospital admission, but no increase in survival to hospital discharge. In this post hoc analysis the actual use of adrenaline was associated with increased short-term survival, but with 48% less survival to hospital discharge. This negative association with survival is very similar to the multivariate analysis of observational Swedish registry data where patients receiving adrenaline were 57% less likely to be alive after one month. 6 Our findings are also consistent with previous studies evaluating effects of amiodarone, 12 atropine 13, adrenaline 14 and high-dose adrenaline, 15 all of which indicated improved short term effects such as survival to hospital or intensive care unit (ICU) admission without improved long-term outcome. An important lesson learned from the present analysis is the limitations of such post hoc analysis when attempting to elucidate causal relationships from non-randomized data, and even more so in registry studies as appropriately discussed in the previous Swedish paper. 6 Some patients randomized to adrenaline never received it as they had ROSC before the drug could be given, thus

4 330 T.M. Olasveengen et al. / Resuscitation 83 (2012) Table 2 In-hospital treatment and outcome. No Adrenaline (n = 481) Adrenaline (n = 367) Odds ratio (95% CI) p-value Admitted to hospital 128 (27) 175 (48) 2.5 (1.9, 3.4) < with ROSC 115 (24) 106 (29) 1.3 (0.9, 1.8) with ongoing CPR 13 (3) 69 (19) 8.3 (4.5, 15.3) <0.001 Admitted to ICU 108 (23) 104 (28) 1.4 (1.0, 1.9) Awake on admission ICU 14 (13) 1 (1) 0.1 (0.0, 0.5) Therapeutic hypothermia 74 (69) 77 (74) 1.3 (0.7, 2.4) Angiography/PCI 58 (54) 35 (33) 0.4 (0.2, 0.7) No. of days in ICU a 5 (3, 9) 4 (2, 7) Cause of death in ICU: b -brain death 29 (71) 52 (69) 0.9 (0.4, 2.2) cardiac death 6 (15) 14 (19) 1.3 (0.5, 3.8) multi-organ failure 6 (15) 9 (12) 0.8 (0.3, 2.4) 0.91 Discharged alive 60 (13) 24 (7) 0.5 (0.3, 0.8) Discharged with CPC (11) 19 (5) 0.4 (0.2, 0.7) CPC CPC CPC CPC Discharged if admitted ICU 56% 23% 0.2 (0.1, 0.4) <0001 Alive 1 year after cardiac arrest c 56 (12) 21 (6) 0.5 (0.3, 0.8) IV = intravenous, ROSC = return of spontaneous circulation, CPR = cardiopulmonary resuscitation, ICU = intensive care unit, PCI = percutaneous coronary intervention, CPC = cerebral performance score. CPC 1: good cerebral performance, CPC 2: moderate cerebral disability, CPC 3: severe cerebral disability, CPC 4: coma or vegetative state. All categorical variables given as numbers (percentages in parenthesis). No. of days in ICU is not normally distributed, and is given as median interquartile range. Differences between groups were analyzed using chi-squared with continuity correction and odds ratio with 95% confidence intervals for categorical data, and Mann Whitney U-test for no. of days in ICU. Proportions of awake on admission to ICU, therapeutic hypothermia, Angiography/PCI and number of days in ICU are reported for patients admitted ICU only. Proportions brain death, cardiac death and multi-organ failure reported for patients who died in ICU only. a Data missing for 6 patients, 2 in no-adrenaline and 4 in adrenaline group. b Data missing for 12 patients, 7 in no-adrenaline and 5 in adrenaline group. c 3 patients lost to 1 year follow up, 2 patients in the no adrenaline group. Table 3 Outcome for VF/pulseless VT (n = 284) and Non-VF/pulseless VT subgroups (n = 564). First rhythm VF/pulseless VT No adrenaline (n = 156) Adrenaline (n = 128) Odds ratio (95% CI) p-value Admitted to hospital 92 (59) 80 (63) 1.2 (0.7, 1.9) 0.63 Admitted to ICU 82 (53) 51 (40) 0.6 (0.4, 1.0) 0.04 Discharged alive 53 (34) 18 (14) 0.3 (0.2, 0.6) <0.001 Discharged with CPC (33) 15 (12) 0.3 (0.1, 0.5) <0.001 Discharged if admitted ICU 65% 35% 0.3 (0.1, 0.6) First rhythm asystole or pulseless electrical activity No adrenaline (n = 325) Adrenaline (n = 239) Odds ratio (95% CI) p-value Admitted to hospital 36 (11) 95 (40) 5.3 (3.4, 8.1) <0.001 Admitted to ICU 26 (8) 53 (22) 3.3 (2.0, 5.4) <0.001 Discharged alive 7 (2.2) 6 (2.5) 1.2 (0.4, 3.5) 1.00 Discharged with CPC (1.8) 4 (1.7) 0.9 (0.3, 3.2) 1.00 Discharged if admitted ICU 27% 11% 0.3 (0.1, 1.2) 0.15 VF = ventricular fibrillation, VT = pulseless ventricular tachycardia, IV = intravenous, ICU = intensive care Unit, CPC = cerebral performance score. All categorical variables given as numbers (percentages in parenthesis). Differences between groups were analyzed using chi-squared with continuity corrections. yielding a selection bias with the most easily resuscitated patients in the post hoc no-adrenaline group. On the other hand, Fig. 1 illustrates that at least 1 of 5 patients randomized to receive IV access and drugs did not receive adrenaline as it was regarded futile or it was impossible to gain intravenous access. At the same time 1 of 10 patients randomized to not receive drugs received adrenaline after they had regained spontaneous circulation for > 5 min. In a single centre double blinded randomized study Jacobs et al. recently reported increased rate of ROSC with adrenaline vs. placebo (24% vs. 8%) with no difference in hospital discharge rate Table 4 Multivariate logistic regression analysis: Adjusted effect on survival when controlling for confounders. Odds Ratio Survival 95% CI p-value Initial VF/VT , <0.001 Witnessed arrest , Male gender , Age (per increased year) , 0.98 <0.001 Response interval (per increased minute) , 0.88 <0.001 Adrenaline , Intubation , VF = ventricular fibrillation. VT = pulseless ventricular tachycardia. 95% CI = confidence interval.

5 T.M. Olasveengen et al. / Resuscitation 83 (2012) Fig. 1. Patient allocation. IV = intravenous. ROSC = return of spontaneous circulation. (4% vs. 2%). The low rates of ROSC (especially in the placebo group) and survival in the Australian study with rates of initial shockable rhythm and ambulance response times relatively similar to the present study makes a comparison difficult, and no data on quality of CPR are reported. 16 Quality of CPR is a known confounder when studying drug effects during cardiac arrest in animal models. 17 As quality was similar and adequate for both groups, this is less likely to have influences our results. Adrenaline is well known to facilitate return of spontaneous circulation, but is also associated with increased post-resuscitation myocardial dysfunction in animals. 18 Negative post-resuscitation effects of adrenaline are also reported to be more prominent after longer, more clinically relevant arrest periods (4 6 min) than short arrest periods (2 min) in experimental models. 19 These experimental data might partly explain our clinical observations. We believe that intention-to-treat analysis of randomized trials is the gold standard when creating clinical guidelines. This does not implicate that continuous quality improvement models cannot be extremely helpful as recently pointed out by Sanders, 20 or that large registry databases do not provide helpful information. All approaches have advantages and limitations; the present report highlights some limitations of post hoc analysis. With the present post-rosc treatment regimes, the general use of adrenaline during CPR does not seem important for long-term outcome. While more patients had improved short-term survival with adrenaline both by intention-to-treat analysis and this post hoc analysis of who actually received adrenaline, more patients did not survive to hospital discharge. In fact, even among the patients randomized to receive drugs, only 43% of the survivors received adrenaline. The actual use of adrenaline may be a surrogate marker for patients with bad prognosis, but that has previously only been published from studies without a group randomized to not receiving drugs. 21 Should patients presenting with initial shockable or nonshockable rhythms be treated differently? Effects of adrenaline seem to vary between these groups, both in magnitude and direction. While there is no negative association between adrenaline and short-term outcome in patients with shockable rhythms, long-term survival was halved in those who received adrenaline compared to those who did not. In a previous non-randomized study of VF patients in Gothenburg, where only some ambulances were allowed to use adrenaline, there was a higher rate of ROSC with adrenaline, but no difference in long-term outcome. 11 For nonshockable rhythms in the present study, there was a 2 3 fold improvements in short-term outcome with no difference in longterm outcome. Several studies have identified dissimilar aetiologies in shockable and non-shockable sub-groups, and it seems reasonable that differences in treatment strategies will emerge. 25 As a limitation, we do not have reliable time points for IV line establishment, adrenaline administration or ROSC. The first two time points were not systematically registered, and we believe the time points for spontaneous circulation to be too inaccurately recorded. As the small ALS team is busy treating the patient, such data are usually manually recorded with significant delay, and defibrillator recordings were not synchronized with the automatic dispatch and ambulance recordings. If such time points had been accurate, we might have been able to adjust for more confounders such as patients who did not receive IV drugs due to rapid establishment of ROSC. A subgroup study of patients who required more than three defibrillation attempts would also be unreliable due to the limited number of patients plus the change in defibrillation strategy from guidelines 2000 to guidelines 2005 in the middle of the study. Early administration of adrenaline, as recently advocated, 26,27 must be evaluated in systems with shorter ambulance response times or other drug regimes and priorities than present guidelines. Further, this is a single centre study and the results may not be generalized to other EMS systems with different training, infrastructure and treatment protocols.

6 332 T.M. Olasveengen et al. / Resuscitation 83 (2012) Conclusions Receiving adrenaline was associated with improved shortterm survival, but decreased survival to hospital discharge and survival with favourable neurological outcome after OHCA. This post hoc survival analysis is in contrast to the previous intention-to-treat analysis of the same data, but agrees with previous non-randomized registry data. This shows limitations of non-randomized or non-intention-to-treat analyses. Conflict of interest Olasveengen and Sunde have no conflicts to declare. Steen is a member of the board of directors for Laerdal Medical and The Norwegian Air Ambulance. Wik is the principle investigator for a multi-centre mechanical chest compression device study sponsored by Zoll. Acknowledgements We thank all physicians and paramedics working in the Oslo EMS Service. The study was supported by grants from South-Eastern Norway Regional Health Authority, Oslo University Hospital, Norwegian Air Ambulance Foundation, Laerdal Foundation for Acute Medicine and Anders Jahres Fund. References 1. Deakin CD, Nolan JP, Soar J, et al. European resuscitation council guidelines for resuscitation 2010 Section 4. adult advanced life support. Resuscitation 2010;81: 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:S Deakin CD, Morrison LJ, Morley PT, et al. Part 8: advanced life support: 2010 international consensus on cardiopulmonary resuscitation and emergency cardiovascular care science with treatment recommendations. Resuscitation 2010;81:e Olasveengen TM, Sunde K, Brunborg C, Thowsen J, Steen PA, Wik L. Intravenous drug administration during out-of-hospital cardiac arrest: a randomized trial. JAMA 2009;302: Stiell IG, Wells GA, Field B, et al. Advanced cardiac life support in out-of-hospital cardiac arrest. N Engl J Med 2004;351: Holmberg M, Holmberg S, Herlitz J. Low chance of survival among patients requiring adrenaline (epinephrine) or intubation after out-of-hospital cardiac arrest in Sweden. Resuscitation 2002;54: Nolan JP, Deakin CD, Soar J, Bottiger BW, Smith G. European Resuscitation Council guidelines for resuscitation Section 4. Adult advanced life support. Resuscitation 2005;67:S Part 6: Advanced Cardiovascular Life Support. Resuscitation 2000;46: Wik L, Hansen TB, Fylling F, et al. Delaying defibrillation to give basic cardiopulmonary resuscitation to patients with out-of-hospital ventricular fibrillation: a randomized trial. JAMA 2003;289: Sunde K, Pytte M, Jacobsen D, et al. Implementation of a standardised treatment protocol for post resuscitation care after out-of-hospital cardiac arrest. Resuscitation 2007;73: Jacobs I, Nadkarni V, Bahr J, et al. Cardiac arrest and cardiopulmonary resuscitation outcome reports: update and simplification of the Utstein templates for resuscitation registries: a statement for healthcare professionals from a task force of the International Liaison Committee on Resuscitation (American Heart Association, European Resuscitation Council, Australian Resuscitation Council, New Zealand Resuscitation Council, Heart and Stroke Foundation of Canada, InterAmerican Heart Foundation, Resuscitation Councils of Southern Africa). Circulation 2004;110: Kudenchuk PJ, Cobb LA, Copass MK, et al. Amiodarone for resuscitation after out-of-hospital cardiac arrest due to ventricular fibrillation. N Engl J Med 1999;341: Stueven HA, Tonsfeldt DJ, Thompson BM, Whitcomb J, Kastenson E, Aprahamian C. Atropine in asystole: human studies. Ann Emerg Med 1984;13: Herlitz J, Ekstrom L, Wennerblom B, Axelsson A, Bang A, Holmberg S. Adrenaline in out-of-hospital ventricular fibrillation. Does it make any difference? Resuscitation 1995;29: Vandycke C, Martens P. High dose versus standard dose epinephrine in cardiac arrest: a meta-analysis. Resuscitation 2000;45: Jacobs IG, Finn JC, Jelinek GA, Oxer HF, Thompson PL. Effect of adrenaline on survival in out-of-hospital cardiac arrest: a randomised double-blind placebocontrolled trial. Resuscitation 2011;82: Pytte M, Kramer-Johansen J, Eilevstjonn J, et al. Haemodynamic effects of adrenaline (epinephrine) depend on chest compression quality during cardiopulmonary resuscitation in pigs. Resuscitation 2006;71: Tang W, Weil MH, Sun S, Noc M, Yang L, Gazmuri RJ. Epinephrine increases the severity of postresuscitation myocardial dysfunction. Circulation 1995;92: Angelos MG, Butke RL, Panchal AR, et al. Cardiovascular response to epinephrine varies with increasing duration of cardiac arrest. Resuscitation 2008;77: Sanders AB. Cardiac arrest and the limitations of clinical trials. N Engl J Med 2011;365: Bunch TJ, White RD, Gersh BJ, et al. Long-term outcomes of out-of-hospital cardiac arrest after successful early defibrillation. N Engl J Med 2003;348: Silfvast T. Cause of death in unsuccessful prehospital resuscitation. J Intern Med 1991;229: Virkkunen I, Paasio L, Ryynanen S, et al. Pulseless electrical activity and unsuccessful out-of-hospital resuscitation: what is the cause of death? Resuscitation 2008;77: Spaulding CM, Joly LM, Rosenberg A, et al. Immediate coronary angiography in survivors of out-of-hospital cardiac arrest. N Engl J Med 1997;336: Wenzel V, Krismer AC, Arntz HR, et al. A comparison of vasopressin and epinephrine for out-of-hospital cardiopulmonary resuscitation. N Engl J Med 2004;350: Ewy GA, Kern KB. Recent advances in cardiopulmonary resuscitation: cardiocerebral resuscitation. J Am Coll Cardiol 2009;53: Morris DC. Vasopressors in cardiopulmonary resuscitation. N Engl J Med 2008;359:1624.