Benefit of Active Compression-Decompression Cardiopulmonary Resuscitation as a Prehospital Advanced Cardiac Life Support
A Randomized Multicenter Study
Background We compared short-term prognosis of active compression-decompression (ACD) and standard (STD) cardiopulmonary resuscitation (CPR) in out-of-hospital cardiac arrests.
Methods and Results We randomized advanced cardiac life support (ACLS) with ACD ACLS CPR on odd days and STD ACLS CPR on even days. We measured the rates of return of spontaneous circulation (ROSC), survival at 1 hour (H1), at 24 hours (H24), and at 1 month (D30); hospital discharge (HD); neurological outcome; and complications. Mean times from collapse to basic cardiac life support CPR was 9 minutes and from collapse to ACLS CPR was 21 minutes. Compared with the STD ACLS patients (n=258), ACD ACLS patients (n=254) had higher survival rates (ROSC, 44.9% versus 29.8%, P=.0004; H1, 36.6% versus 24.8%, P=.003; H24, 26% versus 13.6%, P=.002; HD without neurological impairment, 5.5% versus 1.9%, P=.03) and a trend for improvement in neurological outcome at D30 (Glasgow-Pittsburgh Outcome Categories=1.6±0.8 versus 2.3±1.1, P=.09). Sternal dislodgments (2.9% versus 0.4%, P=.03) and hemoptysis (5.4% versus 1.3%, P=.01) were more frequent in the ACD ACLS group.
Conclusions Despite long time intervals, ACD significantly improved short-term survival rates in out-of-hospital cardiac arrests compared with STD CPR.
The technique of CPR has changed very little since the original description by Kouwenhoven et al in 1960.1 2 Some changes have been proposed to improve cardiopulmonary circulation during resuscitation.3 4 5 6 However, none except perhaps intermittent abdominal compression7 has gained acceptance as a preferred method of CPR.
From an anecdotal report,8 ACD CPR with a suction device has been proposed recently. Compressions are similar to STD CPR, but the negative intrathoracic pressure induced by active chest decompression9 has been shown to improve hemodynamics10 11 12 and ventilation.13 14 Two in-hospital studies found that ACD CPR led to a better initial resuscitation, 24-hour survival rate, and neurological outcome than STD CPR,15 16 but two out-of-hospital studies, collecting data from paramedic teams, did not find any significant difference in outcome.17 18 No difference in rate of HD was found between the two methods.
The aim of the present study was to analyze, in the French prehospital emergency medical system, ROSC, quality of initial neurological recovery, survival rate, and potential complications of ACD CPR compared with STD CPR in a large number of out-of-hospital cardiac arrest patients.
The French Emergency Medical System
Out-of-hospital cardiac arrest management depends on the French prehospital system for emergencies called SAMU (Service d'Aide Me´dicale Urgente).19 France is covered by 95 regional SAMUs organized as follows: one dispatching center involving switchboard operators and two dispatching physicians, situated in a major hospital covering a given medical region, and several units named SMURs (Service Mobile d'Urgence et de Re´animation), based in different hospitals of this region, depend on the dispatching center. Each SMUR is equipped with one or several MICUs, which are medical resuscitation ambulances that always include a physician (anesthesiologist or general practitioner trained in emergency medicine), a nurse or a medical student, and one or two specially trained ambulance drivers. The telephone number, “15,” is an emergency call-for-help number. Switchboard operators, available 24 hours per day, receive all calls relative to cardiac arrest in the dispatching centers and forward them to the dispatching physician, who decides the simultaneous deployment of an EMT team and an MICU. Because of a greater number of EMTs covering a given region, they are frequently closer to the patient and can start BCLS before the arrival of the MICU. BCLS consists of bag valve mask ventilation with O2 and external chest compressions but not defibrillation. MICUs provide all rescue techniques and medical advice, which is termed ACLS CPR. ACLS CPR, according to American Heart Association guidelines, consists of defibrillation, systematic intubation of the patient, and administration of drug therapies until ROSC is observed or a decision is made by the physician of the MICU to stop CPR. There are at least three EMTs in a vehicle with a chief. They work together with the members of the MICU. The physician of the MICU and the chief EMT always supervise the rescuers. ACD CPR was introduced to the Fire Brigade of Paris 2 years before the study was begun. Consequently, in Paris and its suburbs, most of the EMTs are firefighters who often perform BCLS with the use of the suction device.
In summary, a large proportion of BCLS CPR is associated with the use of ACD. A second-tier ACLS CPR is displaced by the SAMU in all cases when BCLS CPR is initiated. Therefore, ACLS CPR is often started as a continuation of BCLS CPR or is sometimes started as first CPR with no BCLS when the MICU reaches the patient before the EMTs. A first level of patient selection is done by EMTs, who eliminate those with irreversible death (decapitation, incineration, decomposition, rigor mortis, or dependent cyanosis) or do-not-resuscitate orders.20 The second level of selection depends on the senior physician (MICU), who eliminates patients with known terminal illness or in whom the delay between cardiac arrest and basic CPR exceeds 30 minutes. All the patients are immediately orotracheally intubated by the physician and are ventilated with a 100% inspired fraction of oxygen. The rescue team stays on the scene until ROSC or decision to stop CPR. No patient is transported to the hospital during CPR except in the event of profound hypothermia, major trauma, gunshot wound, or massive bleeding.
Since training for STD and essentially for the new technique of ACD CPR is crucial, this aspect will be explained briefly. For firemen and SAMU personnel, the training program, performed by physicians or instructors, was similar. After theoretical background was given to explain the major physiological differences between STD and ACD CPR, practical training was performed and special attention was given to the proper position of the suction cup on the chest, the adequate posture of the rescuer, and the active decompression phase. Then, each student used the suction device on specially modified mannequins as recommended by Cohen et al,11 with special attention to the correct positioning of the device on the chest, efficiency of the grip, correct level of compression and decompression, adequacy of rhythm and duty cycle, and time at which fatigue occurred (usually within 3 minutes). The gauge allowed the instructor to evaluate the performance. Rescuers had refresher courses once per week for both techniques and were tested once per month for management of cardiac arrest.
All ACD CPR (BCLS and ACLS) was performed with the Ambu CardioPump (Ambu International Inc) as previously described by Cohen et al,11 using the same rhythm as conventional CPR. Compressions were not interrupted for ventilation when ACLS CPR was performed, according to AHA guidelines.2 Active decompressions were performed in alternation with compressions by lifting the circular handle up to −13.5 kg (−30 lb). CPR was always performed with the patient in the supine position, lying on the floor. Rescuers performing CPR kneeled beside the thorax of the patients with a blanket or a pillow under the knees to lift them up when the suction device was used.
The study was approved by the Consultative Council for the Protection of Persons Volunteering for Biomedical Research of our institution. This trial concerned 5 of 11 medical units covering a part of Paris and its suburbs, corresponding to an approximate population of one million, with daytime population estimated to two million.
From November 1993 to January 1995, patients with prehospital cardiac arrests confirmed by ECG monitoring were enrolled in the study when their age was >18 years and when ACLS CPR was attempted under the supervision of the MICU physician. Patients who recovered a spontaneous palpable carotid or femoral pulse simply with BCLS were not enrolled. The study used a prospective, randomized, parallel-group design by intention to treat. Only the ACLS CPR phase of treatment was randomized (not the BCLS phase). The randomization was based on the calendar date (ACLS ACD CPR on odd days and ACLS STD CPR on even days).
Data were collected for each patient enrolled in the trial according to the recommendations of the Utstein consensus20 : (1) baseline characteristics such as sex, age, place where cardiac arrest occurred, suspected etiology, initial cardiac arrest electrical rhythm (documented as the first rhythm seen by the physician of the MICU on the cardiac monitor [Defigard 2000, ODAM France]) classified as ventricular fibrillation, ventricular tachycardia, asystole, or other rhythms including electromechanical dissociation; (2) incidence of witnessed cardiac arrests (collapse seen or heard by a bystander or emergency personnel); (3) percentage of CPR performed by bystanders; (4) type of BCLS CPR (no BCLS, STD BCLS, ACD BCLS); (5) principal event-to-event-intervals; (6) total doses of epinephrine in the two groups; (7) survival rates at different periods of time: ROSC, defined as a spontaneous palpable carotid or femoral pulse detected >5 minutes; first-hour survival; ICU admission; 24-hour survival; 1-month survival; HD without neurological impairment; (8) neurological outcome evaluated by the Glasgow-Pittsburgh Outcome Categories (GPOC); (9) complications, diagnosed by clinical examination on the scene, by chest radiograph in the ICU, or by autopsy.
Sample Size Considerations
Rate of HD, calculated from 9 French medical theses studying a total of 2490 cardiac arrests under recovery conditions similar to the control group of the present study (references available on request), was ≈2%. Compared with this theoretical rate in the STD ACLS group, using calculations based on Fleiss et al,21 we considered that 252 patients had to be enrolled in each group in order to detect a rate of HD equal to 10% in the ACD ACLS group with a bilateral procedure 5% significance test and with a power of 95%.
Demographic and clinical data in the groups assigned to ACD ACLS CPR and STD ACLS CPR were compared using χ2 analysis for qualitative data and independent sample t test for continuous data. A global comparison of the effect of STD ACLS CPR or ACD ACLS CPR on the outcomes (ROSC, ICU admission, first-hour survival, 24-hour survival, 1-month survival, and HD without neurological impairment) was first performed by standard χ2 analysis. Odds ratios were calculated and their asymptotic standard errors were used to estimate their confidence intervals. A second analysis of these outcomes was performed after stratification according to either the type of BCLS or the classification of cardiac arrest. Mantel-Haenszel statistics were used to obtain the estimate of odds ratio in each strata and to test the null hypothesis of no difference between STD ACLS and ACD ACLS. The homogeneity of the odds ratio between the different strata was tested by the appropriate χ2 test. All tests were bilateral, and significance level was fixed at 5%. Calculations were made using Biomedical Data Package software (University of California at Los Angeles).
A total of 738 emergency calls for cardiac arrest were recorded during the study period. Of these 738 emergency calls, according to study design, 226 were excluded (Table 1⇓): (1) patients who recovered during the BCLS phase (n=6; 4 received ACD BCLS CPR and 2 received STD BCLS CPR); (2) patients with irreversible death or do-not-resuscitate order (n=76; 10.7%); and (3) patients with terminal illness, anoxia lasting >30 minutes without any basic CPR (n=144; 19.5%). The remaining 512 patients were enrolled in the study. Forty-six (9%) patients treated with a CPR technique that did not correspond to the odd/even-day basis were not excluded. Twenty-six received STD ACLS CPR on an odd day; 20 patients received ACD ACLS CPR on an even day. A total of 254 patients (49.6%) received ACD ACLS CPR and 258 (51.4%) received STD ACLS CPR. Demographic data and baseline characteristics of ACD ACLS and STD ACLS groups are described in Table 2⇓. No significant difference was observed between the study and control groups except for the duration of ACLS CPR, which was shorter in the ACD group than in the STD group (26±15.4 versus 29.9±19.9 minutes, respectively, P=.01).
Patients assigned to ACD ACLS CPR had higher rates of ROSC (44.9% versus 29.8%; P=.0004; odds ratio, 0.62 [CI, 0.36 to 0.75]), of survival after 1 hour (36.6% versus 24.8%; P=.003; odds ratio, 0.57 [CI, 0.39 to 0.84]), and of survival after 24 hours (26% versus 13.6%; P=.002; odds ratio, 0.45 [CI, 0.28 to 0.7]) compared with patients assigned to STD ACLS CPR. A significant difference was also observed between the two groups for the rate of ICU admission (33.5% in the ACD ACLS group versus 23.6% in the STD ACLS group; P=.01; odds ratio, 0.61 [CI, 0.42 to 0.91]). The probability value for the difference in survival rate after 1 month was .089 (6.3% in ACD ACLS group versus 3.1% in STD ACLS; odds ratio, 0.48 [CI, 0.2 to 1.14]). At day 30, the number of patients discharged from the hospital without neurological impairment (GPOC=1 or 2) was significantly higher in the ACD ACLS group than in the STD ACLS group (5.5% versus 1.9%; P=.03; odds ratio, 0.34 [CI, 0.2 to 1.14]). These data are summarized in the Figure⇓.
Table 3⇓ shows that the distribution of the initial cardiac rhythm of the discharged patients was not different in both groups. Compared with the outcome of the overall population, Table 4⇓ shows the results of unwitnessed, witnessed, and witnessed patients with or without suspected cardiac etiology subgroups. For the different time-related outcomes (ROSC, 1-hour survival, ICU admission, 24-hour survival, 1-month survival, and HD), no significant difference was found between odds ratio among the cardiac arrest strata considered (ie, unwitnessed, witnessed patients with or without suspected cardiac etiology). For all the outcomes studied, probability levels associated with the test for homogeneity of odds ratio were P>.5.
Before ACLS, patients of the present study could have received the following: no BCLS; STD BCLS; or ACD BCLS CPR. Table 5⇓ shows that ACD similarly improved outcome whatever the three pre-ACLS conditions. No significant difference was found between odds ratios when patients were stratified according to the pre-ACLS received. For all the outcomes studied, probabilities associated with the test for homogeneity of odds ratio were P>.3.
Moreover, even though it was not significant, a trend of GPOC improvement 1 month after ICU admission in ACD ACLS CPR versus STD ACLS CPR should be mentioned (1.6±0.8 and 2.3±1.1, respectively; P=.09).
Complications are shown in Table 6⇓. Hematomas or superficial erythemas at the point of chest compression, sternal dislodgments, and hemoptysis were more frequent with ACD ACLS, but the incidence of rib fractures was similar.
This study demonstrated that prehospital ACD ACLS CPR improved rates of initial resuscitation, short-term survival, 24-hour survival, and HD without neurological impairment. Similar results were observed in the witnessed cardiac arrest subgroup.
In the present study, survival rate with STD CPR was similar to previously published studies15 17 18 despite a longer delay for initiating CPR and a higher incidence of asystole (81%). This result was initially surprising because delay is a major component of prognosis.22 This longer delay resulted in part from the traffic in Paris, which impedes the rescue, and also from the poor education of the French population (<5%) to diagnose cardiac arrest and to rapidly alert the dispatching center by the emergency phone number. Consistent with this long delay, the proportion of asystole in our study was higher than other studies,15 17 18 an element usually associated with less success.22 23 In the present study, of the five STD ACLS patients discharged from the hospital without neurological impairment, two of them initially suffered from a rhythm other than ventricular tachycardia or ventricular fibrillation (see Table 3⇑). Despite these patient characteristics (delay, asystole incidence), survival rate did not differ from previously published results.15 17 18 This might be related to the specificity of the French EMS system. First, exclusion criteria of ACLS CPR are used, such as a long time interval from collapse to initiation of CPR, including cases of incorrect technique of CPR by bystanders. These aspects are rapidly checked on the scene by the physician. This time interval has been fixed at 30 minutes, as in Silfvast's study.24 Mullie et al25 showed in 3083 cardiac arrest cases that a mean duration of complete cardiac arrest of 10.3 minutes was associated with a CPR failure. Second, on the scene, two teams are often working together: the EMTs and the MICU personnel, grouping at least 6 persons. This is very important for the first minutes of ACLS CPR, during which the patient is rapidly undressed, initial cardiac rhythm is monitored, the patient is intubated and ventilated, and drugs and solutions are prepared. This number of rescuers allows a frequent turnover for chest compression, avoiding individual fatigue, and allows the physician to concentrate on the medical aspects of the resuscitation including assessment of the adequate performance of CPR. In our system, the rescue team stays on the scene until ROSC or the decision is made to stop CPR. The later decision is made by the physician when resuscitation fails or when the cardiac arrest occurs in a terminally ill patient, as in the course of a severe comorbid untreatable disease. As a consequence, in the present study, successfully resuscitated patients were transported only when hemodynamics were stable, often with intravenous vasoactive drugs. Therefore, the characteristics of ACLS CPR should be an important point, as stated by Pepe et al26 in a prospective outcome study of 2404 cardiac arrests presenting with rhythms other than ventricular fibrillation: “Therefore, this study provides reasonable proof that some aspects of advanced cardiac life-support measures have lifesaving merit.”
The present study is the first, to our knowledge, to show a positive effect of ACD CPR not only on ROSC, ICU admission, and 24-hour survival but also on HD of patients without neurological impairment. Schwab et al18 did not observe any difference, whereas Lurie et al17 only found a nonsignificant tendency of survival rate improvement. In the Lurie et al study, the statistical significance concerned only ROSC and ICU admission rate after ACD CPR in witnessed arrests with a delay for starting CPR <10 minutes. The benefit observed in the ACD ACLS group might be explained by the characteristics of the French EMS system discussed above. First, the greater number of rescuers limited the risk of fatigue (an important point of ACD CPR27 ) and allowed a supervision of the ACD technique, particularly focusing on the quality of the decompression phase. This was not done in the Schwab et al study.18 Second, the training program could have influenced the present results. Schwab et al18 have pointed out this aspect, detailing the 1-hour teaching program regularly repeated. This differed from the present study, in which the teaching program was longer (2 hours) and focused on the gauge for compression and decompression as performed on the scene. One original quality of this study was the control of an outstanding technical practice before participation in the trial. On the one hand, this might limit ACD CPR to selected rescuers. On the other hand, this is an important prerequisite to test a therapeutic mean in order to improve the “signal-to-noise ratio.”
The Glasgow Coma Score was not calculated after ROSC because most patients were sedated before and during transport for postanoxic cerebral protection and adaptation to the ventilator. After 1 month, the mean levels of the GPOC were similar in both groups. However, the percentage of patients without neurological impairment (GPOC=1 or 2) and already discharged from the hospital at day 30 was significantly higher in the ACD ACLS group.
It is important to note that considering BCLS alone (which included ACD CPR in ≈50% of cases), the rate of ROSC was minimal (n=6). Furthermore, in a given ACLS group (STD or ACD), outcome results were not significantly different whatever the type of BCLS performed. However, in each group, and especially in the STD ACLS group, an insignificant trend favoring ACD BCLS CPR was observed in the first few hours of evolution (ROSC, 1-hour survival, and ICU admission). One might infer that in patients in asystole, ACD BCLS could better stabilize hemodynamics than STD BCLS, potentially allowing ROSC during ACLS CPR. However, it might be difficult to maintain favorable conditions for a long time because of the poor pre-CPR conditions (long period of anoxia before starting CPR). Unfortunately, the number of patients in the subgroups is too low to yield a definitive answer. Within the limits of the study, this implies that ROSC can better be achieved when ACD CPR is used in the ACLS context.
Considering the in-hospital period, the observation that survival rates at 24 hours in the ICU were significantly higher in the ACD ACLS group than in the STD ACLS group suggests a better preservation of organ function with ACD ACLS CPR. This is also consistent with the higher percentage of patients discharged from the hospital without any neurological impairment in the ACD ACLS group. It may be noted that the number of patients who survived at HD in the present study was relatively low but identical to or higher than some of the recently published studies from the United States.15 18
Complications after CPR were observed in both groups. Hematomas were observed especially at the sternal contact point in the ACD CPR group, without clinical consequences. This cutaneous impact was observed much less frequently with the newly designed suction device that had a cushion inserted between the chest and the applicator region compressing the chest. The incidence of rib fractures was similar in the STD CPR and in the ACD CPR groups. Sternal dislodgments (lesion of the sternal-rib junction) and hemorrhagic exudates from the endotracheal tube were more frequently observed in the ACD CPR group and are reported for the first time in this study. There were no clinical sequelae secondary to sternal dislodgments. The hemorrhagic effusions may result from pulmonary hemorrhage secondary to active thoracic decompression, with traction of the anterior thoracic wall leading to large swings in transmural pressure, and/or may be secondary to increased friction between the endotracheal tube and the tracheal endothelium, leading to some venous blood loss. The suction device had been used by firemen for 2 years with regular refresher courses, and the medical teams involved in this study had been extensively trained before starting the protocol. One might thus rule out an incorrect technique in the occurrence of these complications. The hemorrhagic pulmonary exudates were observed only after >20 minutes of ACD CPR. None of those patients had a ROSC, which raises the question of the impact of such a complication on short-term survival. Finally, one case of gastric rupture was observed in a patient who received ACD BCLS, then ACD ACLS CPR. This kind of complication has been reported previously in patients receiving STD CPR.28
There are some limitations of this study. First, EMTs and medical personnel were obviously not blinded to the treatment method. Therefore, enthusiasm of the rescuers to the new technique could represent a potential bias of this study. However, the duration of ACLS CPR of patients who did not recover was similar in STD CPR and ACD CPR groups, suggesting a similar implication of the teams for both techniques. Second, BCLS CPR was not randomized between the two groups. Furthermore, the number of patients in each type of BCLS CPR was too small to perform an adequate analysis of its specific impact on subsequent outcome.
In a two-tiered EMS system including a physician as part of the MICU team, ACD CPR improved short-term survival rates in out-of-hospital cardiac arrest patients compared with STD CPR. However, the particularities of the French EMS system might have influenced the impact of ACD in CPR outcome. Before MICU arrival, the long delay from collapse to initiation of CPR or the nonavailability of early defibrillation might have negated the benefit of ACD performed at that time. However, after MICU arrival, in association with ACLS, ACD improved rates of ROSC, survival at 1 hour, ICU admission, survival at 24 hours, and HD without neurological impairment. Sternal dislodgments and hemorrhagic pulmonary exudates were more frequent with ACD CPR. Further studies should be performed to understand the discrepancies between outcome results of clinical trials performed in different EMS systems, to address the cause of complications of ACD CPR, and to analyze long-term (1-year) survival.
Selected Abbreviations and Acronyms
|ACLS||=||advanced cardiac life support|
|BCLS||=||basic cardiac life support|
|EMS||=||emergency medical services|
|EMT||=||emergency medical technician|
|ICU||=||intensive care unit|
|ROSC||=||return of spontaneous circulation|
This study was supported in part by Institutional Grant Program of “Universite´ Paris VII; UFR Lariboisie`re Saint-Louis,” 1993 to 1994. We thank all participating personnel for their active cooperation in this study. We are indebted to Dr Sadek Beloucif for editing the manuscript.
Presented in part at the 36th Annual National Congress of the French Society of Anesthesiology and Critical Care, Paris, France, September 1994.
- Received April 29, 1996.
- Revision received September 30, 1996.
- Accepted October 7, 1996.
- Copyright © 1997 by American Heart Association
Kouwenhoven WB, Jude JR, Knickerbocker GG. Closed-chest cardiac massage. JAMA. 1960;173:1064-1067.
Bronsvelt W, Groeneveld ABJ, Thijs LG. Intermittent abdominal compression during cardiopulmonary resuscitation. Lancet. 1984;1:1076-1077.
Crawford FA, Pass HI, Taliaferro WE. External cardiopulmonary resuscitation augmented by the military antishock trousers. Am J Surg. 1983;49:651-654.
Lurie KG, Lindo C, Chin J. CPR. The P stands for plumber's helper. JAMA. 1990;264:1661.
Lindner KH, Pfenninger EG, Lurie KG, Schurmann W, Lindner IM, Ahnefeld FW. Effects of active compression-decompression resuscitation on myocardial and cerebral blood flow in pigs. Circulation. 1993;88:1254-1263.
Shultz JJ, Coffeen P, Sweeney M, Detloff B, Kehler C, Pineda E, Yakshe P, Adler SW, Chang M, Lurie KG. Evaluation of STD and active compression-decompression CPR in an acute human model of ventricular fibrillation. Circulation. 1994;89:684-693.
Cohen TJ, Goldner BG, Maccaro PC, Ardito AP, Trazzera S, Cohen MB, Dibs SR. A comparison of active compression-decompression cardiopulmonary resuscitation with standard cardiopulmonary resuscitation for cardiac arrests occurring in the hospital. N Engl J Med. 1993;26:1918-1921.
Schwab TM, Callaham ML, Madsen CD, Utecht TA. A randomized clinical trial of active compression-decompression CPR vs standard CPR in out-of-hospital cardiac arrest in two cities. JAMA. 1995;16:1261-1268.
Carli PA, Riou B, Barriot P, eds. Textbook of Trauma Anesthesia and Critical Care. St Louis, Mo: Mosby; 1993:199-204.
Cummins RO, Chamberlain DA, Abramson NS, Allen M, Baskett P, Becker L, Bossaert L, Delooz H, Dick W, Eisenberg M, Evans T, Holmberg S, Kerber R, Mullie A, Ornato JP, Snadoe E, Skulberg A, Tunstall-Pedoe H, Swanson R, Theis WH. Recommended guidelines for uniform reporting of data from out-of-hospital cardiac arrest: the Utstein style. Ann Emerg Med. 1991;20:861-874.
Eisenberg MS, Hallstrom A, Bergner L. Long-term survival after out-of-hospital cardiac arrest. N Engl J Med. 1982;22:1340-1343.
Mullie A, Van Hoeyweghen R, Quests A, and Cerebral Resuscitation Study Group. Influence of time intervals on outcome of CPR. Resuscitation. 1989;17:S23-S33.