available at www.sciencedirect.com journal homepage: www.elsevier.com/locate/resuscitation

Resuscitation (2008) 79, 424 431 available at www.sciencedirect.com journal homepage: www.elsevier.com/locate/resuscitation CLINICAL PAPER Defibrillation or cardiopulmonary resuscitation first for patients with out-of-hospital cardiac arrests found by paramedics to be in ventricular fibrillation? A randomised control trial, Paul W. Baker a,, Jane Conway a, Chris Cotton a, Dale T. Ashby b, James Smyth c, Richard J. Woodman d, Hugh Grantham a, Clinical Investigators 1 a South Australian Ambulance Service, 216 Greenhill Road, Eastwood, South Australia 5063, Australia b Royal Adelaide Hospital, North Terrace, Adelaide, South Australia 5000, Australia c The Queen Elizabeth Hospital, Woodville Road, Woodville South, South Australia 5011, Australia d Flinders Medical Centre, Flinders Drive, Bedford Park, South Australia 5042, Australia Received 4 February 2008; received in revised form 26 June 2008; accepted 30 July 2008 KEYWORDS Advanced life support (ALS); Ambulance; Basic life support (BLS); Cardiac arrest; Cardiopulmonary resuscitation (CPR); Summary Aim: To determine whether in patients with an ambulance response time of >5 min who were in VF cardiac arrest, 3 min of CPR before the first was more effective than immediate in improving survival to hospital discharge. Methods: This randomised control trial was run by the South Australian Ambulance Service between 1 July, 2005, and 31 July, 2007. Patients in VF arrest were eligible for randomisation. Exclusion criteria were: (i) <18 years of age, (ii) traumatic arrest, (iii) paramedic witnessed arrest, (iv) advanced life support performed before arrival of paramedics and (v) not for resuscitation order or similar directive. The primary outcome was survival to hospital discharge with secondary outcomes being neurological status at discharge, the rate of return of spontaneous circulation (ROSC) and the time from first to ROSC. A Spanish translated version of the summary of this article appears as Appendix in the final online version at doi:10.1016/j.resuscitation.2008.07.017. Trial Registration: This trial was registered with the Australian Clinical Trials Registry, Registration Number: ACTRN12607000401459. URL: http://actr.org.au/default.aspx. Corresponding author at: South Australian Ambulance Service, 216 Greenhill Road, Eastwood, South Australia 5063, Australia. Tel.: +61 08 8274 0432. E-mail address: baker.paul@saambulance.com.au (P.W. Baker). 1 Deb Wood, Evan Everest, Flinders Medical Centre, Flinders Drive, Bedford Park, South Australia 5042, Australia; James Smyth, Christopher Zeitz, Devan Mahadevan, The Queen Elizabeth Hospital, Woodville Road, Woodville South, South Australia 5011, Australia; Michelle McLay, Pauline Whyte, Modbury Hospital, Smart Road, Modbury, South Australia 5092, Australia; Joy Treasure, Lyell McEwin Hospital, Haydown Road, Elizabeth Vale, South Australia 5112, Australia. 0300-9572/$ see front matter 2008 Elsevier Ireland Ltd. All rights reserved. doi:10.1016/j.resuscitation.2008.07.017

Does cardiopulmonary resuscitation before the first make a difference in VF cardiac arrests 425 Defibrillation; Paramedic; Ventricular fibrillation Results: For all response times, no differences were observed between the immediate group and the CPR first group in survival to hospital discharge (17.1% [18/105] vs. 10.3% [10/97]; P = 0.16), the rate of ROSC (53.3% [56/105] vs. 50.5% [49/97]; P = 0.69) or the time from the first to ROSC (12:37 vs. 11:19; P = 0.49). There were also no differences between the immediate group and the CPR first group, for response times of or > 5 min: survival to hospital discharge (50.0% [7/14] vs. 25.0% [4/16]; P = 0.16 or 12.1% [11/91] vs.7.4% [6/81]; P = 0.31, respectively) and the rate of ROSC (71.4% [10/14] vs. 75.0% [12/16]; P = 0.83 or 50.5% [46/91] vs. 45.7% [37/81]; P = 0.54, respectively). No differences were observed in the neurological status of those surviving to hospital discharge. Conclusion: For patient in out-of-hospital VF cardiac arrest we found no evidence to support the use of 3 min of CPR before the first over the accepted practice of immediate. 2008 Elsevier Ireland Ltd. All rights reserved. Introduction Ventricular fibrillation (VF) is a time critical cardiac arrhythmia that results in brain damage and death within minutes, if not converted to a perfusing cardiac rhythm. While immediate is perceived to be the treatment of choice for out-of-hospital VF cardiac arrest, in recent times, based on both scientific 1 and limited out-of-hospital clinical evidence, 2,3 the rationale behind this has been questioned. As survival rates for out-of-hospital VF cardiac arrests have not significantly improved since the introduction of defibrillators to emergency response vehicles this seems appropriate. 2,3 Many laboratory-based animal studies suggest that cardiopulmonary resuscitation (CPR) prior to can sufficiently reverse ischemia allowing successful and a return of spontaneous circulation (ROSC). Recent studies by Cobb 2 and Wik 3 demonstrated that for patients experiencing out-of-hospital VF cardiac arrest with ambulance response times of 4 and >5 min, respectively, an extended period of CPR before all s was associated with an improvement in survival to hospital discharge. A recent randomised control trial by Jacobs et al. 4 however, found no improvement when CPR was performed for 90 s before the first only. Methods Description of local population and criteria for selection of study participants This study was conducted in the metropolitan and suburban areas of Adelaide, and 18 larger rural centres, in the state of South Australia by the South Australian Ambulance Service (SAAS). SAAS is the sole emergency medical response provider in the state servicing a land area of 984,377 km 2 with a population of around 1,558,200 in 2006. 5 Around 71% of the population reside in Adelaide and its immediate suburbs in a land area of around 1826 km 2 (where around 83% of patients were randomised). 5 The larger rural centres in this study had a total population of around 183,000 in 2006. 5 Patients found to be in VF cardiac arrest on arrival of the first ambulance crew were eligible for inclusion. The exclusion criteria were: (i) age <18 years, (ii) arrest of traumatic origin, (iii) arrest witnessed by SAAS paramedics, (iv) advanced life support (ALS) given prior to arrival of paramedics (i.e. medication, or ventilation with oxygen and appropriate airway devices) or (v) the patient was identified as being not for resuscitation. Technical information All emergency (0 0 0) calls for assistance were received in a centralised call centre. Computer aided despatch systems allowed retrieval of call received and scene arrival times, enabling calculation of the response time. Two crews (four officers in total) were despatched to all cardiac arrests one intensive care paramedic as a minimum and three paramedics. Intensive care paramedics have the skill of endotracheal intubation and a broader clinical education, enabling them to administer cardiac specific medications such as adrenaline (epinephrine), atropine, adenosine, amiodarone and lignocaine. Patients were randomised to 3 min of CPR before the first, or routine care (immediate ). Randomisation envelopes were prepared using the sequentially numbered opaque sealed envelopes (SNOPE) technique described by Doig and Simpson. 6 At a confirmed VF cardiac arrest the next envelope in sequence was opened and the randomisation card was followed. Hence, the paramedics were not blinded to treatment. As randomisation was performed in a time critical situation ethics approval was given by the South Australian Government Department of Health to waive informed consent on scene; it was however gained at a later time for all patients who survived to hospital. Aside from 3 min of CPR before the first in the intervention group, all other aspects of arrests were managed in accordance with the Australian Resuscitation Council (ARC) guideline for unwitnessed VF cardiac arrest at that time. In late April 2006 (around 300 days into this trial) SAAS, in line with recommendations from the ARC, following an extensive evaluation of current resuscitation science by the International Liaison Committee on Resuscitation, changed the guideline for unwitnessed VF arrests (the 2006 guideline 7 can be seen at http://www.resus.org.au/). This change was complete by mid July 2006 (around 380 days into the study) after being rolled out to around 750 paramedics across the state. In the pre-2006 guideline patients received a salvo of three s (200, 200 and 360 J monophasic or 3 150 J biphasic) followed by a rhythm check and 1 min of CPR. If after another rhythm check they remained in

426 P.W. Baker et al. VF then a cycle of a salvo of 3 s (either 360 J monophasic or 150 J biphasic) a rhythm check, 1 min of CPR and another rhythm check was repeated until a ROSC was achieved or resuscitation was ceased. CPR was performed with a compression to ventilation ratio of 15:2 at a rate of 80 100 compressions per minute and advanced airway management (i.e. insertion of a laryngeal mask apparatus (LMA) or endotracheal tube (ETT)) took precedence over gaining IV access. In the 2006 guideline patients received a single (360 J monophasic or 200 J biphasic) followed by 2 min of CPR with no rhythm check until this time had elapsed. If the patient was still in VF a cycle of, 2 min of CPR and rhythm checks was performed until a ROSC was achieved or resuscitation was ceased. CPR was performed with a compression to ventilation ratio of 30:2 at a rate of 100 compressions per minute with IV access and drug administration taking precedence over advanced airway management (i.e. LMA or ETT). 7 In both guidelines, (i) if the airway was soiled or obstructed it was managed as a priority, (ii) adrenaline (1 mg, IV) was given every 3 min in support of each cycle, (iii) atropine (1.2 mg, IV) was considered for asystole or bradycardia post-rosc, and (iv) if the rhythm changed to either asystole or there was pulseless electrical activity (PEA), was ceased and CPR was performed. In the 2006 guideline if the patient s VF was refractory after 3 cycles of CPR and then amiodarone was considered (300 mg 1st dose followed by 150 mg at 10 min if still refractory, slow IV push-in the pre-2006 guideline lignocaine was used). Following the arrest, the randomisation card, the patient care record and a printout of critical events from the defibrillator (including the initial rhythm) were forwarded to the study coordinators who then contacted the hospital medical officers to initiate follow-up. The medical officers conducting follow-up and assessment of neurological function were blinded to treatment. Patients who arrested in the metropolitan area and achieved a sustained ROSC were transported to The Royal Adelaide Hospital, Flinders Medical Centre, The Queen Elizabeth Hospital, Modbury Hospital or The Lyell McEwin Hospital. Patients who had an arrest in a regional centre and achieved a sustained ROSC were retrieved to The Royal Adelaide Hospital or Flinders Medical Centre. Approval for access to medical records was gained from the Human Research Ethics Committees at each of these institutions. This research was conducted in accordance with the principles of the Helsinki Declaration. Statistics The study was powered to detect a 5.5-fold increase in survival to hospital discharge in patients with a response time of >5 min receiving 3 min of CPR before, as observed by Wik. 3 A similar increase with 100 subjects in each group and a baseline survival of 3.3% would provide 90% power with a type 1 error rate of alpha = 0.05. All statistical analyses were performed using Stata version 10.0 (Stata Corp, Texas, USA). Differences between groups were assessed using unpaired t-tests for continuous data and either 2 -tests or Fisher s exact test for categorical data. Multinomial logistic regression analysis was used to determine the relative probability of survival between groups at each of five stages (resuscitation ceased on scene, achieved ROSC on scene, survival to hospital, survival to ICU, survival to hospital discharge). All relative risk ratios (RRR) are reported as the ratio of the relative risks of those surviving to hospital discharge versus those where resuscitation was ceased on scene. Results Study participants During the 761 days of this study 3245 patients were found lifeless; resuscitation was not commenced on 1845 of these. On analysis of the SAAS Cardiac Arrest Database, the initial rhythm was VF in 377, VT in 26 and asystole or PEA in 2005. The rhythm was not recorded in 837 patients where significant lividity, rigor, or other obvious signs of death were present. In these patients resuscitation was not commenced. Exclusion criteria were met in 31 patients; age <18 (n = 1), (ii) traumatic arrest (n = 9), (iii) arrest witnessed by paramedics (n = 19) and (v) resuscitation not commenced due to not for resuscitation order or similar directive (n =2) (Figure 1). The randomisation cards were not followed for 5 patients; they were excluded. Of the 341 patients who were eligible only 202 were randomised. Figure 2 (panel A) shows the temporal randomisation of patients into this study. The baseline characteristics of age, gender, proportion of bystander witnessed arrests, proportion of arrests where bystander CPR was performed and the location of the arrest were not significantly different between the groups (Table 1). Specifics of the arrests of study participants Although no difference was observed in response times, the 3 min required to perform CPR before the first in the intervention group resulted in a significant increase in the time from the monitor being turned on to the first (Table 2). These 3 min of CPR however, did not significantly change the time taken from either the monitor being turned on or the first being given to when a ROSC was achieved (Table 2). There were also no significant differences in: the total number of s given, the number of s required to achieve a ROSC, the percentage of patients in whom a ROSC was achieved, the number of patients with a clear airway on initial examination and whom were intubated, the dose of adrenaline administered and the numbers of patients who received adrenaline, atropine, lignocaine or amidarone (Table 2). Survival and neurological function of study participants In this study only 14.9% (30/202) of patients had a response time of 5 min, with 39.3% (11/28) of survivors coming from this subgroup (Table 3). For all response times combined, and for subgroups with response times of or >5 min, there was a tendency for reduced survival to hospital discharge in the CPR before group (lower by 6.8%, 25.0% and 4.7%, respectively), but this was not statistically significant (Table 3) for comparison the temporal rates of survival to discharge can be seen in Figure 2 (panel C). There were

Does cardiopulmonary resuscitation before the first make a difference in VF cardiac arrests 427 Figure 1 Flow diagram of patients assessed for eligibility and randomised into the study. also no differences in the rates of survival to the Emergency Department or admission to the Intensive Care Unit (Table 3). Following introduction of the 2006 guideline, the combined survival increased more than twofold from 8.8% to 18.0% (8/91; days 1 340, vs. 20/111; days 341 760, P = 0.059), on univariate analysis however this fell short of significance. In contrast, the combined rates of ROSC increased only marginally from 50.5% to 53.2% (46/91 vs. 59/111, P = 0.71). A graphical representation of this can be seen in Figure 2 (panels C and B, respectively). Post-hoc subgroup analysis of patients treated with the pre-2006 guideline or the 2006 guideline revealed no significant differences in any of the parameters listed in Tables 1 3 with the exception of the number of patients who received lignocaine (21/91 (23.1%) vs. 7/111 (6.3%), respectively, P < 0.001) or amiodarone (8/91 (8.8%) vs. 48/111 (43.2%), respectively, P < 0.001). Upon multinomial logistic regression analysis the response time was the only covariate tested that was found to influence survival to hospital discharge (RRR 0.68, 95% CI 0.55 0.84, P = 0.001). The relationship between response time and survival to discharge for all patients combined can be seen in Figure 3. No clear relationship was found between survival to hospital discharge and whether the airway was initially soiled or obstructed (RRR 0.70, 95% CI 0.25 1.91, P = 0.49), whether the patient was intubated (RRR 1.13, 95% CI 0.39 3.33, P = 0.81) or the patient age (RRR 1.00, 95% CI 0.97 1.03, P = 0.92). There were however, trends indicating a beneficial effect on survival to hospital discharge from the arrest being witnessed by a bystander (RRR 5.36, 95% CI 0.95 30.2, P = 0.057), whether the 2006 Table 1 Patient demographics and information about bystander involvement and the location of the arrests. Patient demographics No. (%) a P-value b Immediate (n = 105) CPR before (n = 97) Age, median (range), years 66.0 (23 92) 65.0 (31 92) 0.90 Male gender 84 (80.0) 81 (83.5) 0.52 Bystander information Bystander witnessed arrest 83 (79.0) 81 (83.5) 0.42 Bystander CPR performed 61 (58.0) 57 (58.8) 0.92 Location of arrest Home 67 (63.8) 64 (65.0) Public place 30 (28.6) 25 (25.8) 0.90 Other 8 (7.6) 8 (8.2) Urban 88 (83.8) 80 (82.5) 0.80 a All data (except for age) are expressed as the number of patients in a category and their respective percentage (i.e. No. (%)). b Categorical data were analysed as 2 2or2 3 contingency tables using either 2 or Fisher s exact tests while continuous variables were analysed using a two-tailed t-test. Abbreviations: CPR, cardiopulmonary resuscitation.

428 P.W. Baker et al. Table 2 Specific information about times, s, ROSC, airway management and intravenous medications given. Times No. (%) a P value b Immediate (n = 105) CPR before (n = 97) Response time (call received to arrival of 1st Ambulance) 8:14 (7:39 8:49) 7:41 (7:08 8:14) 0.18 Monitor on to 1st 1:56 (1:11 2:41) 4:26 (3:59 4:52) 0.01 Monitor on to ROSC 13:59 (11:22 16:36) 15:53 (13:11 18:36) 0.32 1st to ROSC 12:37 (10:06 15:09) 11:19 (8:40 13:58) 0.49 Defibrillation and ROSC information Total s received, mean (95% CI) 5.02 (4.41 5.63) 5.06 (4.44 5.68) 0.92 Number of s to get ROSC, mean (95% CI) 3.61 (2.98 4.23) 3.47 (2.67 4.27) 0.79 ROSC achieved 56 (53.3) 49 (50.5) 0.69 Airway details Clear airway 64 (60.9) 68 (70.8) 0.14 Intubated with an ETT 83 (79.0) 69 (71.1) 0.19 Intravenous medication given Adrenaline, mean (95% CI) (mg) 4.57 (4.09 5.04) 4.53 (4.01 5.04) 0.91 Adrenaline 95 (90.5) 85 (87.6) 0.52 Atropine 1200 g 37 (35.2) 34 (35.1) 0.98 Lignocaine 100 mg 11 (10.5) 17 (17.5) 0.16 Amiodarone 300 mg 31 (29.5) 25 (25.8) 0.55 Clear airway refers to patients who had a clear airway on initial inspection as distinct from those who had either a soiled or obstructed airway. Intubated with an ETT refers to patients who were intubated as distinct from those who received either an oropharyngeal airway or a laryngeal mask apparatus as their highest airway management. Abbreviations: ROSC, return of spontaneous circulation; ETT, endotracheal tube; CPR, cardiopulmonary resuscitation. a Unless indicated in the first column (or in the case of all the times which are listed as mean (95% CI), min:sec) all data are expressed as the number of patients in a category and their respective percentage (i.e. No. (%)). b Categorical data were analysed as 2 2 contingency tables using either 2 or Fisher s exact tests while continuous variables were analysed using a two-tailed t-test. guideline was used (RRR 2.02, 95% CI 0.75 5.42, P = 0.16) or whether bystander CPR was performed (RRR 1.93, 95% CI 0.71 5.22, P = 0.19), none of these reached significance though. There were also no significant differences in the Cerebral Performance Category (CPC) or Overall Performance Category (OPC) of patients who survived to hospital discharge (Table 4). Discussion We were unable to demonstrate any significant increase in survival to hospital discharge following 3 min of CPR before the first, in patients found to be in VF cardiac arrest by paramedics. This was true for all response times combined, as observed previously by Wik 3 and Jacobs et al. 4 but importantly it was also true when response times were broken down into or >5 min intervals. The lack of a difference upon subgroup analysis is not consistent with the observations of either Cobb 2 or Wik 3 who found significant improvements (10% and 18%, respectively) with response time of 4 and >5 min, respectively. The rates of ROSC (53% vs. 46%), hospital/icu admission (37% vs. 46%) and especially survival to hospital discharge (17% vs. 15%) in the control group in this study are very similar to those observed by Wik. 3 There were also no marked differences between these studies in the time taken to achieve ROSC (11 12 min vs. 13 14 min, respectively) or the patient populations with respect to age (66 vs. 70 years), the proportion of male patients (80% vs. 85%) and the rates of bystander witnessed arrests (80% vs. 91%) and bystander CPR (58% vs. 62%). Finally, the probability of survival to hospital discharge vs. response time is very similar (see Figure 3). Hence, aside from methodological differences discussed below, these studies are comparable in terms of the EMS systems involved, the level of practitioner skill and the population enrolled. The lack of any evidence in favour of an extended period of CPR before the first, in this and the Jacobs et al. 4 study, argues for immediate, as is currently the practice. It may well be that in the Cobb 2 and Wik 3 studies it was the extended period of CPR before subsequent s (not the first ) that had the greatest impact on survival to discharge. This hypothesis is supported by the twofold increase in survival observed in this study, following a doubling of the duration of CPR after from 1 to 2 min, with the introduction of the 2006 guideline for VF cardiac arrest. 7 It should be noted that, in practice, performing CPR after is synonymous with performing it before (e.g., CPR, rhythm check (quick),, CPR, etc.). While the twofold increase in survival (Figure 2, panel C) observed following the guideline change may be a Hawthorne effect, this seems unlikely given that: (i) there was only a small (5%) increase in the rate of ROSC following the change (Figure 2, panel B) and (ii) paramedics

Does cardiopulmonary resuscitation before the first make a difference in VF cardiac arrests 429 Table 3 Rates of survival to discharge, admission to ICU, survival to ED and ROSC. Parameter No. (%) a OR (95% CI) P-value b Immediate (n = 105) CPR before (n = 97) All response times Survival to discharge 18 (17.1) 10 (10.3) 0.56 (0.25 1.25) 0.16 Admission to ICU 39 (37.1) 36 (37.1) 0.99 (0.57 1.76) 0.89 Survival to ED 43 (41.0) 40 (41.2) 1.01 (0.58 1.77) 0.97 ROSC 56 (53.3) 49 (50.5) 0.89 (0.52 1.55) 0.69 Parameter No. (%) a OR (95% CI) P-value b Immediate (n = 14) CPR before (n = 16) Response time 5 min Survival to discharge 7 (50.0) 4 (25.0) 0.33 (0.08 1.49) 0.16 Admission to ICU 8 (57.1) 7 (43.8) 0.58 (0.14 2.41) 0.71 Survival to ED 8 (57.1) 8 (50.0) 0.75 (0.18 3.08) 0.70 ROSC 10 (71.4) 12 (75.0) 1.20 (0.25 5.65) 0.83 Parameter No. (%) a OR (95% CI) P-value b Immediate (n = 91) CPR before (n = 81) Response time >5 min Survival to discharge 11 (12.1) 6 (7.4) 0.58 (0.21 1.60) 0.31 Admission to ICU 31 (34.1) 29 (35.8) 1.08 (0.58 2.02) 0.87 Survival to ED 35 (38.5) 32 (39.5) 1.05 (0.57 1.93) 0.89 ROSC 46 (50.6) 37 (45.7) 0.82 (0.45 1.50) 0.54 Abbreviations: ROSC, return of spontaneous circulation; CPR, cardiopulmonary resuscitation; ED, Emergency Department; ICU, Intensive Care Unit. a All data are expressed as the number of patients in a category and their respective percentage (i.e. No. (%)). b All data were analysed as 2 2 contingency tables using either 2 or Fisher s exact tests. Table 4 Cerebral performance categories and overall performance categories of patients who survived to hospital discharge. Outcome All response times a Response time 5 min a Response time >5 min a Immediate CPR before Immediate CPR before Immediate CPR before CPC 1 12 7 5 3 7 4 2 3 1 1 0 2 1 3 3 2 1 1 2 1 4 0 0 0 0 0 0 5 (dead) 87 87 7 12 80 75 Total 105 97 14 16 91 81 OPC 1 9 7 3 3 6 4 2 6 2 3 1 3 1 3 3 1 1 0 2 1 4 0 0 0 0 0 0 5 (dead) 87 87 7 12 80 75 Total 105 97 14 16 91 81 Abbreviations: CPC, cerebral performance category; OPC, overall performance category; CPR, cardiopulmonary resuscitation. a All data were analysed as 2 4 contingency tables using either 2 or Fisher s exact tests. No statistically significant differences were found in the CPCs or OPCs between the groups in either of the response time categories.

430 P.W. Baker et al. Figure 2 Temporal representation of the rates of randomisation of patients into the study (panel A), ROSC (panel B) and survival to hospital discharge (panel C); immediate group (thick line) and CPR before group (thin line). The vertical line shows the mid point in the 80 days (day 340) required to introduction the 2006 guideline. this argument, the rates of survival of patients receiving either lignocaine or amiodarone were also not significantly different (1/25 (4.0%) vs. 4/52 (7.7%), respectively, P = 0.54). It is important to note that the single policy of the 2006 guideline, 7 instead of a salvo of 3 s as in pre-2006 guideline, may also have played a part in improving survival post introduction of the 2006 guideline. Significantly, this study may contain the first indication of the effectiveness of the ARC 2006 guideline for VF cardiac arrest. It should be remembered however, that while both the univariate and multivariate analyses of the effect of the 2006 guideline on survival to hospital discharge showed a strong trend towards the 2006 guideline being beneficial, this did not reach statistical significance. Aside from an unavoidable guideline change during this study, a further limitation is that 139 patients, or 37%, who were found in VF were not considered for randomisation into the study as it was inadvertently overlooked by the paramedics for comparison in the Wik 3 and Jacobs et al. 4 studies around 14 15% of patients were not randomised. Analysis of the characteristics of all patients randomised into the treatment groups however, indicate that they were very well matched for all variables (Table 1). Hence, we do not expect the patients who were not randomised to have significantly influenced the outcome given the effectiveness of the randomisation process. For patients with a response time of >5 min in the Wik 3 study, who received 3 min of CPR before all s, there was an 18% increase in survival to hospital discharge. This study was terminated early (according to the stopping rules of Wald 8 for binary outcomes) as after 2 years, and randomisation of a similar number of patients to the Wik 3 study, there were no significant increases in survival to discharge when analysing all response times or subgroups with response times of or >5 min there were actually non-significant reductions following 3 min of CPR for these response times (6.8%, 25.0% and 4.7%, respectively). The larger number of patients in the > 5 min response time subgroup in this study, compared to that of Wik 3 (172 vs. 81), are already well trained in resuscitations they are required to perform under pressure on a regular basis. Given that the vast majority of patients surviving to discharge in the pre-2006 guideline or 2006 guideline groups received neither lignocaine (7/8 (87.5%) nor amiodarone (16/20 (80.0%), respectively, it also seems unlikely that the change from lignocaine to amiodarone lead to such a marked increase in survival. As these agents are indicated for refractory VF their use is more likely to be associated with a poor outcome. Indeed, the rate of survival of patient receiving lignocaine and/or amiodarone compared to those receiving neither was significantly lower (5/80 (6.25%) vs. 23/122 (18.6%), P < 0.01). Finally, in support of Figure 3 Estimated probability of survival to hospital discharge versus response time. Line generated by logistic regression analysis survival (%) = 49.2 18.5 ln(response time), R 2 = 0.86.

Does cardiopulmonary resuscitation before the first make a difference in VF cardiac arrests 431 gave it a 90% power to detect an 18% increase in survival following CPR. Conclusion This study found no benefit from CPR before the first in unwitnessed out-of-hospital VF cardiac arrest, therefore we recommend the current practice of immediate be continued. This study may contain the first data on the effectiveness of the Australian Resuscitation Councils 2006 guidelines for VF cardiac arrest, as evidenced by a twofold increase in survival to hospital discharge following its introduction, however this requires independent confirmation. Conflict of interest statement None. Acknowledgements The following individuals from the SA Ambulance Service: Kate Biggins, Michael Bohrnsen, Julie Parker, Clare Beard, Nan Gladdis, Jocelyn Graham, Harish Kumar and Dave Norman. Dr. Adrian Estermann (bio-statistician) from Flinders Medical Centre. Appendix A. Supplementary data Supplementary data associated with this article can be found, in the online version, at doi:10.1016/j.resuscitation. 2008.07.017. References 1. Weisfeldt M, Becker L. Resuscitation after cardiac arrest. A 3- phase time-sensitive model. JAMA 2002;288:3035 8. 2. Cobb L, Fahrenbruch C, Walsh T, et al. Influence of cardiopulmonary resuscitation prior to in patients with out-of-hospital ventricular fibrillation. JAMA 1999;281: 1182 8. 3. Wik L, Hansen T, Fylling F, et al. Delaying to give basic cardiopulmonary resuscitation to patients with outof-hospital ventricular fibrillation. A randomised trial. JAMA 2003;289:1389 95. 4. Jacobs I, Finn J, Oxer H, Jelinek G. CPR before in out-of-hospital cardiac arrest: a randomised trial. Emerg Med Australas 2005;17:39 45. 5. The Australian Bureau of Statistics: http://www.abs.gov.au/. 6. Doig G, Simpson F. Randomization and allocation concealment: a practical guide for researchers. J Crit Care 2005;20:187 91. 7. Australian Resuscitation Council. Adult advanced life support: Australian resuscitation Council Guidelines 2006. Emerg Med Australas 2006;18:337 56. 8. Wald A. Sequential analysis. New York: Wiley; 1947.