Part 3: Adult Basic Life Support and Automated External Defibrillation: 2015 International Consensus on Cardiopulmonary Resuscitation and Emergency Cardiovascular Care Science With Treatment Recommendations

For the critical outcome of cardiac arrest recognition, we identified very-low-quality evidence (downgraded for risk of bias, indirectness, and imprecision) from 1 cluster RCT,¹² as well as very-low-quality evidence from 26 non-RCTs comprising 8 before-after observational studies,^13–20 9 prospective single-arm observational studies,^13,21–28 8 retrospective single-arm observational studies,^29–36 and 1 case-control study.¹¹ A total of 17 420 patients were included in these 27 studies.

“Cardiac arrest recognition” was reported heterogeneously across the included studies, precluding meta-analysis. Seven observational studies reported the sensitivity of dispatch protocols to recognize cardiac arrest,^{17,18,21,29–32} with results that ranged from 38% to 96.9% and specificity that exceeded 99% in the 2 studies that reported this outcome.^29,30 Recognition rates of cardiac arrest ranged from 18% to 83% where reported.^22,25

The majority of the study dispatch centers used scripted dispatch protocols with questions to identify patients who are unconscious and not breathing or not breathing normally. Four before-after studies^16–18,20 suggested that introducing scripted dispatch protocols or modifying existing protocols can help increase cardiac arrest recognition. One study reported an increase in cardiac arrest recognition¹⁶ while 3 reported an increase in the rates of telephone-assisted CPR after the introduction of scripted dispatch protocols.^17,18,20 One study also reported an increase in “high-acuity” calls after a modification to the seizure protocol.¹⁹

Recognition of unconsciousness with abnormal breathing is central to dispatcher recognition of cardiac arrest. Many terms may be used by callers to describe abnormal breathing: difficulty breathing, poorly breathing, gasping breathing, wheezing breathing, impaired breathing,²² occasional breathing, barely/hardly breathing, heavy breathing, labored or noisy breathing, sighing, and strange breathing.¹¹ Agonal breaths were reported in approximately 30% of cases in 1 study,¹³ which can make obtaining an accurate description of the patient’s breathing pattern challenging for dispatchers. The presence of agonal breaths were mentioned as a factor negatively affecting cardiac arrest recognition in 10 studies,^{13–15,18,22,23,25,33,35,37} with 1 study reporting that agonal breaths were present in 50% of nonidentified cardiac arrest calls.¹⁸ Other terms reported in the studies that may help identify possible cardiac arrest cases include “dead,” “is dead,” “cold and stiff,” “blue,” “gray,” or “pale.”²⁷ The aforementioned descriptions, however, may be limited, owing to cultural influences and language translation limitations.

Three before-after studies suggested that offering dispatchers additional education that specifically addresses agonal breaths can increase the rates of telephone-assisted CPR^14,15 and decrease the number of missed cases.³⁷

There is evidence from 3 studies that failure to recognize cardiac arrest may be associated with failure to follow scripted protocols by omitting specified questions about consciousness and breathing.^23,24,26

We recommend that dispatchers determine if a patient is unconscious with abnormal breathing. If the victim is unconscious with abnormal or absent breathing, it is reasonable to assume that the patient is in cardiac arrest at the time of the call (strong recommendation, very-low-quality evidence).

We recommend that dispatchers be educated to identify unconsciousness with abnormal breathing. This education should include recognition and significance of agonal breaths across a range of clinical presentations and descriptions (strong recommendation, very-low-quality evidence).

In making these recommendations, we placed a higher value on the recognition of cardiac arrest by dispatchers, and we placed a lower value on the potential harms arising from inappropriate CPR and the potential need for increased resources. In this situation, we believe that the benefits associated with increased numbers of cardiac arrest patients receiving timely and appropriate interventions outweigh the undesirable effects (potential for patients not in cardiac arrest to inappropriately receive chest compressions, potential need for increased resources).

We recognize that the evidence in support of these recommendations comes from mainly observational studies of very low quality. Large, high-quality RCTs addressing this question are unlikely to be conducted. We believe that the available evidence shows consistent results favoring scripted dispatch protocols and that education including a description of the presenting signs of cardiac arrest and populations at risk (eg, patients presenting with seizures) enables dispatchers to identify cardiac arrest. We recognize that dispatch protocols for a range of conditions (including but not limited to “seizures,” “breathing problems,” “chest pains,” “falls,” and “unknown problem”) optimized to identify potential cardiac arrest without undue delay may further improve early recognition of cardiac arrest.

High-quality data from RCTs are currently lacking. Further studies are required to determine the following:

Bystander CPR rates remain relatively low in most communities. Dispatcher-assisted telephone CPR instructions have been demonstrated to improve bystander CPR rates.

This review identified 1 meta-analysis,³⁸ 3 randomized trials,^39–41 and 11 observational studies.^{15,17,18,20,27,34,42–46} The population evaluated in most studies were adults with a presumed cardiac cause of their cardiac arrest and excluded traumatic and/or asphyxial causes of cardiac arrest.^{17,20,41,42,44} Two studies evaluated all out-of-hospital arrests,^15,34 although benefit was limited to the cardiac cause subgroup in 1 of these studies.¹⁵ Two studies evaluated telephone-assisted CPR in children.^45,46

Some studies evaluated the survival benefit of dispatch-assisted CPR instructions and compared systems where such instructions can be offered to systems where they were never or infrequently offered.^{15,17,18,20,27,34,43–45} Other studies compared traditional CPR to chest compression–only CPR instructions delivered by telephone.^39–42

For the critical outcome of survival with favorable neurologic outcome, we have identified very-low-quality evidence from 2 RCTs,^39,40 2 cohort studies,^45,46 and 1 before-after study.¹⁵ The level of evidence was downgraded for risk of bias, indirectness, and imprecision. Four studies reported no benefit in neurologic outcomes.^39,40,45,46 The before-after study, which included dispatcher instructions to start compression-only CPR as part of a bundle of interventions used as part of a quality improvement initiative, noted improved neurologic outcomes at 12 months (odds ratio [OR], 1.81; 95% confidence interval [CI], 1.2–2.76).¹⁵

For the critical outcome of survival, we identified very-low-quality evidence from 3 RCTs.^39–41 The level of evidence was downgraded for risk of bias, indirectness, and imprecision. Meta-analysis of these trials found an absolute survival benefit of 2.4% (95% CI, 0.1%–4.9%) in favor of telephone-assisted continuous chest compressions over telephone-assisted traditional CPR (number needed to treat [NNT], 41; 95% CI, 20–1250; relative risk [RR], 1.22; 95% CI, 1.01–1.46).³⁸

We also identified 6 before-after studies.^{15,17,18,20,42,43} One study was inconsistent with the others and found decreased survival, although it was not powered to evaluate survival outcomes.¹⁸ One study showed a survival benefit at 1 year (population of 73 patients) from an educational program for dispatchers on continuous chest compressions and agonal breaths (adjusted OR, 1.81; 95% CI, 1.20–2.76).¹⁵

We also identified 5 cohort studies.^{27,34,44–46} One study showed a survival benefit at 30 days when, after an educational program, telephone-assisted CPR was provided to a pediatric out-of-hospital cardiac arrest (OHCA) population versus not (adjusted OR, 1.46; 95% CI, 1.05–2.03).⁴⁵ A second cohort study in the pediatric (less than 18 years of age) population showed survival benefit at 30 days when telephone-assisted CPR was provided (adjusted OR for group not receiving CPR, 0.70; 95% CI, 0.56–0.88).⁴⁶

For the critical outcome of ROSC, we identified very-low-quality evidence from 1 RCT⁴⁰ and 1 before-after study.¹⁸ The studies were downgraded for indirectness and imprecision. Neither study showed a statistically significant benefit.

For the important outcome of delivery of bystander CPR, we identified very-low-quality evidence from 6 before-after studies: 1 study compared 2 medical priority dispatch system versions,⁴² 3 studies compared telephone-assisted CPR versus not,^17,18,43 and 2 studies^15,20 compared various educational programs. In addition, we identified 1 cohort study.⁴⁵ The level of evidence was downgraded for indirectness and imprecision. All showed a strong association between telephone-assisted CPR and bystander CPR. The cohort study showed increased performed chest compressions (adjusted OR, 6.04; 95% CI, 4.72–7.72) and ventilation (adjusted OR, 3.10; 95% CI, 2.44–3.95) from telephone-assisted CPR, and an absolute increase in bystander CPR rate of 40.9% (95% CI, 36.1–45.5).⁴⁵

For the important outcome of time to commence CPR, we have identified very-low-quality evidence from 4 before-after studies^15,17,20,43 and 1 cohort study.⁴⁴ The level of evidence was downgraded for risk of bias, inconsistency, indirectness, and imprecision. None of these reported a statistically significant benefit.

For the important outcome of CPR parameter, assessed with initial rhythm of ventricular fibrillation (VF)/pulseless ventricular tachycardia (pVT), we have identified very-low-quality evidence from 1 RCT⁴¹ and 1 before-after study.¹⁸ The studies were downgraded for risk of bias, indirectness, and imprecision. Neither study showed a statistically significant benefit.

We recommend that dispatchers provide chest compression–only CPR instructions to callers for adults with suspected OHCA (strong recommendation, low-quality evidence).

In making these recommendations, we placed a higher value on the initiation of bystander CPR and a lower value on the harms of performing CPR on patients who are not in cardiac arrest. We recognize that the evidence in support of these recommendations comes from randomized trials and observational data of variable quality. However, the available evidence consistently favors telephone CPR protocols that use a compression-only CPR instruction set, suggesting a dose effect—that is, quick telephone instructions in chest compressions result in more compressions and faster administration of CPR to the patient.

Opioid overdose is a leading cause of death in many communities and at-risk populations. With widespread implementation of community naloxone distribution programs, there is a need for evaluating the evidence of such initiatives to provide guidance for policy makers.

We did not identify any published studies to determine if adding intranasal or intramuscular naloxone to conventional CPR is superior to conventional CPR alone for the management of adults and children with suspected opioid-associated cardiac or respiratory arrest in the prehospital setting. An additional search was performed to assess available evidence for overdose education and naloxone distribution programs (see BLS 891).

No treatment recommendation can be made for adding naloxone to existing BLS practices for the BLS management of adults and children with suspected opioid-associated cardiac or respiratory arrest in the prehospital setting.

All patients with suspected opioid-associated cardiac or respiratory arrest should receive standard BLS care, with or without the addition of naloxone. In making this recommendation, we place greater value on the potential for lives saved by recommending immediate BLS care and education, with or without naloxone, and lesser value on the costs associated with naloxone administration, distribution, or education.

Opioid overdose events have increased dramatically and become a leading cause of premature, preventable mortality in many communities. With widespread implementation of various community programs, there is a need for evaluating the evidence of such initiatives to provide guidance for policy makers.

For the critical outcome of survival to hospital discharge, we have identified very-low-quality evidence (downgraded for risk of bias, inconsistency, indirectness, and imprecision) from 3 observational before-after studies.^49,50 Only 1 of 3 studies⁵¹ attempted to correct for any confounding factors expected in interventional studies by using historic controls. This study did observe a dose-response gradient with 0.73 (95% CI, 0.57–0.91) and 0.54 (95% CI, 0.39–0.76) adjusted-rate ratios for lethal overdose in communities with low and high implementation, respectively.⁵¹ The remaining 2 observational studies reported reductions in rate ratios for lethal overdose in communities, 0.62 (95% CI, 0.54–0.72)⁵² and 0.70 (95% CI, 0.65–0.74).⁴⁹

We suggest offering opioid overdose response education, with or without naloxone distribution, to persons at risk for opioid overdose in any setting (weak recommendation, very- low-quality evidence).

In making these recommendations, we place greater value on the potential for lives saved by recommending overdose response education, with or without naloxone, and lesser value on the costs associated with naloxone administration, distribution, or education.

This question was initiated in response to a request that ILCOR review the evidence for prognostic factors that predict outcome in relation to a drowning incident. Drowning is the third leading cause of unintentional injury death worldwide, accounting for nearly 400 000 deaths annually. Care of a submersion victim in high-income countries often involves a multiagency approach, with several different organizations being independently responsible for different phases of the victim’s care, from the initial aquatic rescue, on-scene resuscitation, transfer to hospital, and hospital and rehabilitative care. Attempting to rescue a submerged victim has substantial resource implications and may place rescuers at risk themselves.

The systematic review included observational studies with control groups, which presented data enabling us to calculate RRs or reported ORs for the following prognostic factors: (1) age, (2) EMS response time, (3) salinity, (4) submersion duration, and (5) water temperature. The review excluded single case reports and case series less than 5 or without comparator groups. Full details of the search strategy and included articles are summarized in Scientific Evidence Evaluation and Review System (SEERS). It is worth highlighting that, in accordance with GRADE guidelines for prognostic studies, evidence from observational studies starts as high. In reviewing the summaries of evidence, one should note that the studies reviewed extended over a 30-year period. It is possible that outcomes may have changed over time, although this was not observed in the 2 studies that evaluated outcomes over time.^53,54 The populations evaluated varied between studies and included emergency service data, coroners’ registries, emergency department, and intensive care admissions.

We recommend that submersion duration be used as a prognostic indicator when making decisions surrounding search and rescue resource management/operations (strong recommendation, moderate-quality evidence for prognostic significance).

We suggest against the use of age, EMS response time, water type (fresh or salt), water temperature, and witness status when making prognostic decisions (weak recommendation, very-low-quality evidence for prognostic significance).

We acknowledge that this review excluded exceptional and rare case reports that identify good outcomes after prolonged submersion in icy cold water.

In making these recommendations, the task force placed priority on producing simple guidance that may assist those responsible for managing search and rescue operations. The public comments highlighted the difficult moral dilemmas facing the rescuer in these often emotionally charged and fast-moving environments. While it is hoped that the information will also be of interest to those managing the initial resuscitation and intensive care treatment of drowning victims, the factors evaluated were limited to those available at the scene of the drowning incident and excluded factors available after the victim was rescued (eg, patient regurgitation, cardiopulmonary arrest duration, EMS response time, CPR duration, hospitalization^82,83).

The recommendations presented place a relatively high value on the associations demonstrated in the exploratory prognostication modeling but acknowledge that these have not been prospectively validated as clinical decision rules. Submersion durations of less than 10 minutes are associated with a very high chance of favorable outcome, and submersion durations more than 25 minutes are associated with a low chance of favorable outcomes. Given the known difficulties with accurate timing, we suggest the time of the emergency service call as the start point for estimating submersion duration.

This question raised significant debate during the plenary conference session, task force discussion, and public comment. The main areas of controversy related to (1) how ILCOR intended the information presented would be used, and (2) the prognostic value of water temperature. We clarified in the introduction to this section that the review is intended to provide evidence from the published literature to support those responsible for search-and-rescue operations about chances of survival.

In recommending submersion duration as a factor, we acknowledge that definitions of submersion were either absent or varied between studies, and, in many studies, the precise submersion duration was not known. The 2015 Utstein consensus on drowning defines submersion duration as the duration of time that liquid covering the nose and mouth prevents air from entering the lungs.⁸⁴ We suggest that the studies are interpreted as assuming the point from which continuous submersion started (ie, not when the person is struggling and intermittently submerging then drawing breath). Because the underwater interval is seldom documented with a timepiece, estimates can be imprecise. The Utstein consensus recommended cross-referencing submersion point with the emergency call and ambulance arrival times when possible.

The scope of this systematic review was limited to large case series and cohort studies with control groups. The review, therefore, excluded rare and exceptional case reports of survival after prolonged submersion in ice-cold water. One such example is the series of case reports presented by Tipton and Golden, which identified 26 survivors after submersion, 8 reports documented favorable outcomes in victims submerged for longer than 25 minutes in mostly ice-cold water.⁸⁵ A further case series noted 80% of victims surviving after cardiac arrest after immersion in ice-cold water (2°C) for up to 2.5 hours.⁸⁶ This review included more than 1000 drowning victims and produced conflicting results on the role of water temperature. Both studies noted difficulty in estimating water temperature, which is likely to be reflected in real-life rescue situations. Thus, a combination of uncertainty in the published evidence and practical difficulties of measurement led us to suggest against the routine use of water temperature as a prognostic factor.

The task force is very grateful for insightful comments submitted during the public commenting process.

Delivering high-quality chest compressions as early as possible is vital to high-quality CPR and optimizes the chance of ROSC and survival after cardiac arrest. Thus, a major change after publication of the 2010 International Consensus on CPR and ECC Science With Treatment Recommendations was the recommendation that, for adult victims of cardiac arrest, CPR should begin with giving chest compressions rather than opening the airway and delivering rescue breaths. This treatment recommendation is based on a review of science from the perspective of developing a treatment recommendation for adults.

There were no human studies identified in this evidence review, but 4 manikin studies were identified; 1 randomized study⁸⁷ focused on adult resuscitation, 1 randomized study focused on pediatric resuscitation,⁸⁸ and 2 observational studies focused on adult resuscitation.^89,90 Compared with the previous review in 2010, this review also identified 3 new studies that were included for analysis.^87,88,90 Overall, the reviewers had serious concerns for trial methodology of included studies. The nature of comparing 2 different resuscitation protocols meant that all studies suffered from performance and detection bias because healthcare professionals were not blinded to the intervention (C-A-B versus A-B-C).

For the important outcome of time to commencement of chest compressions, we identified very-low-quality evidence from 1 randomized manikin study⁸⁸ representing 155 two-person teams and very-low-quality evidence from 2 observational manikin studies^89,90 representing 40 individual rescuers⁹⁰ and 33 six-person teams.⁸⁹ All studies were downgraded due to risk of bias. All studies found that C-A-B decreased the time to commencement of chest compression. The randomized trial found a statistically significant 24.13-second difference (P<0.05) in favor of C-A-B.⁸⁸ The observational studies found statistically significant decreases of 20.6 seconds (P<0.001)⁹⁰ and 26 seconds (P<0.001),⁸⁹ respectively.

For the important outcome of time to commencement of rescue breaths, we identified very-low-quality evidence from 2 randomized manikin studies^87,88 representing 210 two-person teams. Both studies were downgraded due to risk of bias. Lubrano⁸⁸ found a statistically significant 3.53-second difference (P<0.05) in favor of C-A-B during a respiratory arrest scenario; however, in a cardiac arrest scenario, A-B-C decreased the time to commencement of rescue breaths by 5.74 seconds (P<0.05).⁸⁷ Marsch found that C-A-B decreased time to commencement of rescue breaths by 5 seconds (P=0.003). The clinical significance of these differences is unknown.⁸⁷

For the important outcome of time to completion of first CPR cycle (30 chest compressions and 2 rescue breaths), we identified low-quality evidence from 1 randomized manikin study⁸⁷ representing 55 two-person teams. Marsch⁸⁷ found that C-A-B decreased time to completion of first CPR cycle by 15 seconds (P<0.001). The clinical significance of this difference is unknown.

We suggest commencing CPR with compressions rather than ventilations (weak recommendation, very-low-quality evidence).

In making this recommendation in the absence of human data, we placed a high value on time to specific elements of CPR (chest compressions, rescue breathing, completion of first CPR cycle). In making this recommendation in the absence of human data, given that most cardiac arrests in adults are cardiac in cause, we placed a high value on reducing time to specific elements of CPR (chest compressions and completion of first CPR cycle).

We refer the reader to the systematic review Peds 709 (see “Part 6: Pediatric Basic Life Support and Pediatric Advanced Life Support”) for recommendations in children.

Bystander CPR is a key life-saving factor in the Chain of Survival. This foundational principle was evaluated in the ILCOR 2000 Consensus on Science and was accepted in the 2005, 2010, and 2015 consensuses without reevaluation. The review in 2000 found that CPR before EMS arrival can (1) prevent VF/pVT from deteriorating to asystole, (2) increase the chance of defibrillation, (3) contribute to preservation of heart and brain function, and (4) improve survival.⁹¹ A large systematic review from 79 studies involving 142 740 patients confirmed that bystander CPR improves survival from 3.9% to 16.1%.⁹²

Although the practice of bystander CPR is accepted, a key question is whether bystanders should perform chest compression–only CPR or conventional CPR. Advocates of chest compression–only CPR note that it is easier to teach, remember, and perform compared with chest compressions with assisted ventilation. Others are concerned that chest compressions without assisted ventilation are less effective because of inadequate oxygenation and worse respiratory acidosis. These concerns are especially pertinent in the setting of asphyxial cardiac arrests (and perhaps others with a noncardiac cause) and in the setting of prolonged CPR.⁹³

It is not feasible to conduct RCTs of bystander-initiated, compression-only CPR versus bystander conventional CPR. Therefore, the clinical evidence for this question is derived from 2 sources: observational studies, and RCTs of dispatcher-assisted CPR. The benefits of telephone-assisted CPR are summarized in our EMS dispatch question (see BLS 359).

Further, much of the research on this topic has been done on patients whose arrests are presumed to be of cardiac origin, which would be difficult, if not impossible, to coach bystanders to determine in a brief training session. In addition, the research was often conducted in settings with short EMS response intervals. It is likely that a time threshold exists beyond which the absence of ventilation may be harmful.^94,95 Thus, the generalizability of the findings from the studies to all settings is cautioned. These factors taken together mean that the data available for considering this question are indirect.

When observational studies are conducted for bystander CPR, a key factor in evaluating the data is determining how the investigator determined the type of bystander CPR that was performed. In some cases, providers stayed on the scene and interviewed the bystanders about the care they provided. However, in studies of registry data, EMS providers visually evaluated bystanders’ actions as they took over care during this high-stress, high-risk, and time-intensive event. This issue led to many studies being downgraded for validity because determination of the type of bystander CPR may have been biased.

For the critical outcome of survival with favorable neurologic outcome at 12 months, we identified very-low-quality evidence (downgraded for risk of bias, indirectness, and imprecision) from a single observational study of 1327 adult cardiac arrest victims of a presumed cardiac cause. The study reported no overall difference between compression-only and conventional CPR (OR, 0.98; 95% CI, 0.54–1.77).⁹⁴

For the critical outcome of survival with favorable neurologic outcome at 30 days, we identified very-low-quality evidence (downgraded for risk of bias and indirectness) from 4 observational studies that enrolled 40 646 patients.^93,95–97 These studies reported no overall difference in outcomes.

For the critical outcome of survival with favorable neurologic outcome at discharge, we identified very-low-quality evidence (downgraded for risk of bias, inconsistency, and indirectness) from 1 randomized trial⁴⁰ and 3 observational studies.^98–100 The randomized trial enrolled 1268 patients and reported no difference in outcomes (OR, 1.25; 95% CI, 0.94–1.66). The observational studies enrolled 2195 patients and also found no overall differences between compression-only and conventional CPR.

For the critical outcome of survival at 30 days, we identified very-low-quality evidence (downgraded for risk of bias and indirectness) from 1 randomized trial⁴¹ and 2 observational studies.^101,102 The randomized trial enrolled 1276 patients and found no difference in outcomes (OR, 1.24; 95% CI, 0.85–1.81).⁴¹ The observational studies enrolled 11 444 patients, and found no overall difference between compression-only and conventional CPR.^101,102

For the critical outcome of survival at 14 days, we identified very-low-quality evidence (downgraded for risk of bias and indirectness) from 1 observational study¹⁰³ enrolling 829 patients, which found no difference between compression-only and conventional CPR (OR, 0.76; 95% CI, 0.46–1.24).

For the critical outcome of survival to discharge, we identified very-low-quality evidence (downgraded for risk of bias, inconsistency, and indirectness) from 1 randomized trial¹⁰⁴ and 2 observational studies.^105,106 The randomized trial enrolled 520 patients and found no difference in outcomes (OR, 1.4; 95% CI, 0.88–2.22).¹⁰⁴ The observational studied enrolled 2486 patients and reported no difference between compression-only and conventional CPR.

We recommend that chest compressions should be performed for all patients in cardiac arrest (strong recommendation, very-low-quality evidence).

We suggest that those who are trained and willing to give rescue breaths do so for all adult patients in cardiac arrest (weak recommendation, very-low-quality evidence).

In making these recommendations, the task force strongly endorsed the 2010 ILCOR Consensus on Science that all rescuers should perform chest compressions for all patients in cardiac arrest.^107,108 We also highlight the 2015 dispatcher CPR recommendation that “dispatchers should provide chest compression–only CPR instructions to callers for adults with suspected OHCA.”

The task force draws attention to the potential gains from the simplicity of teaching compression-only CPR.

The task force further acknowledges the potential additional benefits of conventional CPR when delivered by trained laypersons, particularly in settings where EMS response intervals are long and for asphyxial causes of cardiac arrest.

We refer the reader to Ped 414 systematic review (see “Part 6: Pediatric Basic Life Support and Pediatric Advanced Life Support”) for recommendations in children.

The optimal initial approach to a patient found in VF outside of the hospital has been unclear. Observational studies have supported a short period of CPR followed by early analysis of cardiac rhythm and administration of a shock, if indicated. Other studies support a longer period of CPR before administration of a shock.

Our literature review retained 13 articles. These included 5 RCTs,^109–113 4 observational cohort studies,^114–117 3 meta-analyses,^118–120 and 1 subgroup analysis of data reported in the RCT by Rea et al.¹²¹ For the purposes of this evidence review, the GRADE table is limited to pooled data from the 5 RCTs. All of the studies were conducted in the out-of-hospital setting.

The intervention assessed was a short period of chest compressions before attempted defibrillation with a longer period of chest compressions, defined as between 90 and 180 seconds, before attempted defibrillation. In all of the RCTs reviewed, chest compressions were performed before initial analysis while the defibrillator equipment was being applied. The exact duration of this period was documented precisely in only 1 RCT¹¹² and was noted to be between 30 and 60 seconds.

For the critical outcome of survival to 1 year with favorable neurologic outcome (Cerebral Performance Category [CPC] of 2 or less), we identified low-quality evidence (downgraded for bias and imprecision) from a single randomized trial that showed no benefit from a short period of CPR before shock delivery (OR, 1.18; 95% CI, 0.522–2.667).¹¹³

For the critical outcome of survival to hospital discharge with favorable neurologic outcome (defined as CPC score of 2 or less, modified Rankin Scale score of 3 or less), we identified low-quality evidence (downgraded for inconsistency and imprecision) from 4 RCTs that showed no benefit from a short period of CPR before shock delivery (OR, 0.95; 95% CI, 0.786–1.15).^{109,111–113}

For the critical outcome of survival to 1 year, we identified low-quality evidence (downgraded for bias and imprecision) from 2 RCTs that showed no benefit from a short period of CPR before shock delivery (OR, 1.15; 95% CI, 0.625–2.115).^110,113

For the critical outcome of survival to hospital discharge, we identified low-quality evidence (downgraded for bias and imprecision) from 4 RCTs that showed no benefit from a short period of CPR before shock delivery (OR, 1.095; 95% CI, 0.695–1.725).^{109–111,113}

With respect to ROSC, we identified low-quality evidence (downgraded for bias and imprecision) from 4 RCTs that showed no benefit from a short period of CPR before shock delivery (OR, 1.193; 95% CI, 0.871–1.634).^{109–111,113}

During an unmonitored cardiac arrest, we suggest a short period of CPR until the defibrillator is ready for analysis and, if indicated, defibrillation.

In making these recommendations, we placed a higher value on the delivery of early defibrillation and a lower value on the as-yet-unproven benefits of performing CPR for a longer period of time. We recognize that the evidence in support of these recommendations comes from randomized trials of variable quality conducted in several countries with a variety of EMS system configurations. The available evidence suggests a minimal effect size overall, while recognizing that it remains possible that, in systems with higher baseline survival rates, a longer period of CPR may be superior.

The task force notes that these recommendations apply to unmonitored victims in cardiac arrest. In witnessed, monitored VF/pVT arrest where a patient is attached to a defibrillator, shock delivery should not be delayed.

Hand position is just one of several components of chest compressions that can alter effectiveness. In making this recommendation, we considered the evidence in an attempt to define the optimal compression method. We balanced this against the current recommendation for using the lower half of the sternum¹⁰⁷ and the resource implications of changing current recommendations.

The task force also noted previous recommendations that the lower half of the sternum could be identified by instructing a rescuer, “Place the heel of your hand in the center of the chest with the other hand on top.”^122,123 This instruction should be accompanied by a demonstration of placing the hands on the lower half of the sternum.¹²³

This review focused on studies reporting clinical or physiologic outcomes related to hand position during chest compression. The scope differed from the 2010 CoSTR review, which also included computed tomographic, echocardiographic, and manikin studies reporting on the anatomic structures that would be compressed with different hand positions and the efficiency of different instructional techniques for locating hand position.

There were no studies reporting the critical outcomes of favorable neurologic outcome, survival, or ROSC.

For the important outcome of physiologic end points, we identified 3 very-low-quality studies (downgraded for risk of bias, indirectness, and imprecision).^124–126 One crossover study in 17 adults with prolonged resuscitation from nontraumatic cardiac arrest observed improved peak arterial pressure during compression systole (114 ± 51 mm Hg versus 95 ± 42 mm Hg) and end-tidal carbon dioxide (ETCO₂; 11.0 ± 6.7 mm Hg versus 9.6 ± 6.9 mm Hg) when compressions were performed in the lower third of the sternum compared with the center of the chest, whereas arterial pressure during compression recoil, peak right atrial pressure and coronary perfusion pressure did not differ.¹²⁴ A second crossover study in 30 adults observed no difference between ETCO₂ values and hand placement.¹²⁵ A further crossover study in 10 children observed higher peak systolic pressure and higher mean arterial blood pressure when compressions were performed on the lower third of the sternum compared with the middle of the sternum.¹²⁶

We suggest performing chest compressions on the lower half of the sternum on adults in cardiac arrest (weak recommendation, very-low-quality evidence).

In making this recommendation, we place a high value on consistency with current treatment recommendations in the absence of compelling data suggesting the need to change the recommended approach.

Chest compression rate can be defined as the actual rate used during each continuous period of chest compressions over 1 minute, excluding any pauses. It differs from the number of chest compressions actually delivered in 1 minute, which takes into account any interruptions in chest compressions.

In the 2010 CoSTR, we recommended a manual chest compression rate of at least 100/min in adults. We noted the absence of a specific upper limit for compression rate.¹⁰⁷ This review notes the publication of important new observational studies in humans that suggest the need to limit the upper rate of chest compressions.^127,128

No studies addressed the critical outcome of favorable neurologic outcome.

For the critical outcome of survival to hospital discharge, we identified very-low-quality evidence from 2 observational studies^127,128 representing 13 469 adult patients. Both studies were downgraded due to risk of bias.¹²⁷ They compared chest compression rates of greater than 140/min, 120 to 139/min, less than 80/min, and 80 to 99/min with the control rate of 100 to 119/min. When compared with the control chest compression rate of 100 to 119/min, there was a

The study found a significant relationship between chest compression rate categories and survival without adjustment and when adjusted for covariates, including CPR quality measures such as compression depth and fraction (global test, P=0.02). The study showed chest compression depth declined with increasing chest compression rate. The relationship of reduced compression depth at different compression rates was as follows: for a compression rate of 100 to119/min, 35% of compressions had a depth less than 3.8 cm; for a compression rate of 120 to 139/min, 50% of compressions had a depth less than 3.8 cm; and for a compression rate of 140/min or greater, 70% of the compressions had a depth less than 3.8 cm.

In the second study,¹²⁸ there was a 4.1% decrease in survival to hospital discharge with chest compression rates of greater than 140/min and a 1.9% increase in survival to hospital discharge with rates of less than 80/min when compared with the control rate of 80 to 140/min. The adjusted ORs for survival to hospital discharge were 0.61 (P=0.18) for rates of greater than 140/min and 1.32 (P=0.42) for rates of less than 80/min and, therefore, showed no significant difference in survival to hospital discharge.

For the critical outcome of ROSC, we identified very-low-quality evidence from 3 observational studies^127–129 representing 13 566 adult patients. All studies were downgraded due to risk of bias. All studies had different interventions and different control chest compression rates: 100 to 119/min,¹²⁷ 80 to 140/min,¹²⁸ and 80 to 119/min.¹²⁹

We recommend a manual chest compression rate of 100 to 120/min (strong recommendation, very-low-quality evidence).

In making this recommendation, we place a high value on compatibility with the previous guidelines recommendation of a lower compression threshold of at least 100/min to minimize additional training and equipment costs (eg, reprogramming feedback devices, educational programs). We consider the new evidence that has emerged since 2010 CoSTR as sufficient to suggest that the upper threshold should be limited to no more than 120/min.

In 2010, we recommended that it is reasonable to compress the sternum at least 5 cm (2 inches) for all adult cardiac arrest victims. We stated that there was insufficient evidence to recommend a specific upper limit for chest compression depth. Important new data have emerged since 2010, which has prompted the revision of our treatment recommendation.

The reader is reminded of our 2010 recommendation that CPR should be performed on a firm surface when possible. Air-filled mattresses should be routinely deflated during CPR. There was insufficient evidence for or against the use of backboards during CPR. If a backboard is used, rescuers should minimize delay in initiation of chest compressions, minimize interruption in chest compressions, and take care to avoid dislodging catheters and tubes during backboard placement.

For the critical outcome of survival with good neurologic outcome (CPC 1-2), we found low-quality evidence (downgraded for imprecision, upgraded for a dose-response gradient), from 1 observational study¹³² suggesting that a compression depth in adults of more than 5 cm (2 inches) is better than all other compression depths during manual CPR. Adjusted OR for each 5 mm increase in mean chest compression depth was 1.33 (1.03-1.71) for favorable functional outcome. Upon review by the evidence evaluation experts during the final iterative ILCOR process (see Part 2), 1 citation was excluded from the final consensus on science.¹³³

For the critical outcome of survival to hospital discharge, we found very-low-quality evidence (downgraded for imprecision), from 3 observational studies^132,134,135 suggesting that survival may improve with increasing compression depth. The adjusted ORs for survival to hospital discharge per 5 mm increase in mean chest compression depth were 1.09 (95% CI, 0.94 to 1.27),¹³⁴ 1.04 (95% CI, 1.00 to 1.08),¹³⁵ and 1.30 (95% CI, 1.03 to 1.65).¹³² In the largest study (9136 patients) a covariate-adjusted spline analysis showed a maximum survival at a mean depth of 4.0 to 5.5 cm (1.6 to 2.2 inches), with a peak at 4.6 cm (1.8 inches).¹³⁵

For the critical outcome of ROSC, we found low-quality evidence (downgraded for imprecision, upgraded for a dose-response gradient) from 4 observational studies^{134,135,137,140} suggesting that a compression depth of more than 5 cm (2 inches) in adults is better than all other compression depths during manual CPR. The largest study reported that ROSC increased with each 5 mm increment (adjusted OR 1.06 [95% CI: 1.04 to 1.08, P<0.001]) and that the adjusted OR for ROSC for patients receiving chest compressions with a depth of 3.8 to 5.1 cm (1.5 to 2 inches) compared with more than 5.1 cm (more than 2 inches) was 0.86 (95% CI, 0.75-0.97).¹³⁵ Upon review by the evidence evaluation experts during the final iterative ILCOR process (see Part 2), 4 citations were excluded from the final consensus on science.^{133,136,138,139}

For the important outcome of injury, we found very-low-quality evidence (downgraded for serious risk of bias, imprecision, and very serious indirectness) from 1 observational study suggesting that a compression depth of more than 6 cm (2.4 inches) is associated with an increased rate of injury in adults when compared with compression depths of 5 to 6 cm (2 to 2.4 inches) during manual CPR. This study included 170 of 353 patients (183 excluded for incomplete data), and injuries were reported in 63% with compression depth more than 6 cm (more than 2.4 inches) and 31% with compression depth less than 6 cm. Further, injuries were reported in 28%, 27%, and 49% with compression depths less than 5 cm (less than 2 inches), 5 to 6 cm (2 to 2.4 inches), and more than 6 cm (more than 2.4 inches), respectively.¹⁴¹

We recommend a chest compression depth of approximately 5 cm (2 inches) (strong recommendation, low-quality evidence) while avoiding excessive chest compression depths (greater than 6 cm [greater than 2.4 inches] in an average adult) (weak recommendation, low-quality evidence) during manual CPR.

In making this recommendation, we place a high value on the consistency with our previous recommendations given the resource implications (eg, training, reprogramming CPR devices) of making a change, and consistency in data showing harm from compressions that are too shallow. In addition, we note new data from the US and Canadian Resuscitation Outcomes Consortium group reporting a “sweet spot” for compression depth between 4.03 and 5.53 cm (between 1.59 and 2.2 inches; peak, 4.56 cm [1.8 inches]) and harm from excessive compression depths.¹³⁵ We used the term approximately 5 cm (approximately 2 inches) to reflect these findings plus the known variation in patient shapes and sizes around the world.

We refer the reader to Ped 394 systematic review (see “Part 6: Pediatric Basic Life Support and Pediatric Advanced Life Support”) for recommendations in children.

Critical to hemodynamically effective CPR is blood returning to the chest between compressions. Venous return is in part influenced by the pressure gradient between extrathoracic and intrathoracic veins. Leaning on the chest wall between compressions, precluding full chest wall recoil, could raise intrathoracic pressure and reduce right heart filling, coronary perfusion pressure, and myocardial blood flow.^142,143 Observational studies indicate that leaning is common during CPR in adults and children.^143,144 This question sought to examine the effect of chest wall leaning during standard manual CPR on outcome.

For the critical outcomes of ROSC, survival at hospital discharge, and survival with favorable neurologic/functional outcome, we found no evidence to address the question.

For the important outcome of coronary perfusion pressure, we found 3 observational studies: 2 in animal models^142,145 and 1 in anesthetized children not in cardiac arrest,¹⁴⁶ which provided very-low-quality evidence, downgraded for serious risk of bias and very serious indirectness. All 3 studies reported a reduced coronary perfusion pressure with incomplete chest recoil. In anesthetized children undergoing mechanical ventilation during cardiac catheterization, Glatz et al analyzed the effect of leaning by applying a force on the chest corresponding to 10% and 20% of body weight; this resulted in a proportional reduction in coronary perfusion pressure.¹⁴⁶ Yannopoulos et al and Zuercher et al reported in swine models of VF that leaning on the chest precluding full chest recoil reduced the coronary perfusion pressure in a dose-dependent manner.^142,145

For the important outcome of cardiac output/cardiac index, we found 2 observational studies (1 in an animal model and 1 in anesthetized children not in cardiac arrest) also representing very-low-quality evidence downgraded for serious risk of bias and very serious indirectness.^142,146 The study in animals reported a proportional reduction in cardiac index when 10% and 20% of the forces applied during compression remained between compressions.¹⁴² In contrast, Glatz et al found that leaning forces had no effect on cardiac output.¹⁴⁶

We suggest that rescuers performing manual CPR avoid leaning on the chest between compressions to allow full chest wall recoil (weak recommendation, very-low-quality evidence).

In making this recommendation, the task force placed high value on consistency with previous recommendations and in ensuring that a clear recommendation is provided for CPR training and practice. We acknowledge that some studies have reported a leaning threshold below which there are possibly no adverse hemodynamic effects, but the task force anticipates that this would be difficult to measure and teach.

For adults in cardiac arrest without an advanced airway, such as an endotracheal tube, chest compressions are often briefly paused to allow for ventilation. Some CPR guidelines suggest that the duration of pauses for ventilation should not exceed 5 seconds. However, forceful insufflations to comply with the guideline carry a risk of gastric insufflation, and may not be feasible for mouth-to-mouth ventilation.

Preshock intervals include the time required for assessment of the rhythm, charging, and actually delivering a shock. Postshock intervals reflect the time from shock delivery to resumption of chest compression. Achieving short preshock and postshock pauses requires awareness of the importance of minimizing the pause, attention during training, and an excellent interplay among the rescuers working together during a resuscitation attempt. In this systematic review, we examine the possible consequences of interruptions of chest compressions on various critical and important outcomes.

For the critical outcome of favorable neurologic outcome, we found 1 low-quality observational study (downgraded for imprecision)¹⁴⁷ enrolling 199 patients. This study compared survival against a reference ventilation range of 5 to 6 seconds and found no difference with longer ranges of time for 2 breaths delivered by lay rescuers, ranging over 10 to 12 seconds (adjusted OR, 1.30; 95% CI, 0.29–5.97) and 13 seconds or greater (adjusted OR, 2.38; 95% CI, 0.46–12.1). We found no studies addressing pauses for rhythm analysis and shock.

For the critical outcome of survival to hospital discharge, there were no studies examining duration required to deliver 2 breaths. For perishock pauses, we identified moderate-quality evidence (downgraded for indirectness) from 1 RCT that compared 2 AED algorithms.¹⁴⁸ The study enrolled 845 patients but found no benefit (OR, 0.81; 95% CI, 0.33–2.01) of reducing preshock and postshock pauses. We found moderate-quality evidence from 3 observational studies (upgraded for dose-response gradient)^149–151 including 3327 patients showing a strong relationship with shorter preshock and postshock pauses (less so for postshock pauses) or higher chest compression fraction.

For the critical outcome of ROSC, we found no studies addressing the duration required to deliver 2 breaths. For perishock pause, we found 1 very-low-quality observational study¹⁵² (downgraded for risk of bias and imprecision) including 35 patients, indicating benefit from limiting preshock and postshock pauses and 1 very-low-quality study (downgraded for risk of bias)¹⁵³ including 2103 patients, suggesting benefit from achieving chest compression fractions (ie, total CPR time devoted to compressions) greater than 40%.

For the important outcome of shock success, we found 1 very-low-quality observational study (downgraded for imprecision)¹³⁸ including 60 patients, indicating benefit of shorter preshock pauses: OR of 1.86 (95% CI, 1.10–3.15) for every 5 seconds.

We suggest that in adult patients receiving CPR with no advanced airway, the interruption of chest compressions for delivery of 2 breaths should be less than 10 seconds (weak recommendation, low-quality evidence).

We recommend that total preshock and postshock pauses in chest compressions be as short as possible. For manual defibrillation, we suggest that preshock pauses be as short as possible and no more than 10 seconds (strong recommendation, low-quality evidence).

We suggest during conventional CPR that chest compression fraction (ie, total CPR time devoted to compressions) should be as high as possible and at least 60% (weak recommendation, low-quality evidence).

In making these recommendations, we place a high priority on minimizing interruptions for chest compressions. We seek to achieve this overall objective by balancing it with the practicalities of delivering 2 effective breaths between cycles of chest compressions to the patient without an advanced airway.

For adults in cardiac arrest without an advanced airway, such as an endotracheal tube, chest compressions are often briefly paused to allow for ventilation. In 2005, many guidelines for adults in cardiac arrest were changed from a compression-ventilation ratio of 15:2 to a ratio of 30:2. This structured review identified 4 observational cohort before-after studies, all conducted in the out-of-hospital setting, that evaluated a bundle of care interventions implemented after the 2005 guidelines and used a 30:2 compression-ventilation ratio, compared with care before the 2005 guidelines that used a 15:2 compression-ventilation ratio.^154–157 No studies were identified that compared a 30:2 compression-ventilation ratio to a compression-ventilation ratio other than 15:2. One observational cohort study,¹⁵⁸ although meeting our inclusion criteria, was excluded due to concerns regarding the study design, analytical technique, and challenges with data reporting and abstraction.

For the critical outcome of survival with favorable neurologic outcome at discharge, we found very-low-quality evidence from 2 observational studies^154,155 that were downgraded for risk of bias and indirectness. Of the 1711 patients included, those who were treated under the 2005 guidelines with a compression-ventilation ratio of 30:2 had slightly higher survival than those patients treated under the 2000 guidelines with a compression-ventilation ratio of 15:2 (8.9% versus 6.5%; RR 1.37 [0.98–1.91]).

For the critical outcome of survival to hospital discharge, we identified very-low-quality evidence from 4 observational studies.^154–157 The level of evidence was downgraded for risk of bias and indirectness. Of the 4183 patients included, those who were treated under the 2005 guidelines with a compression-ventilation ratio of 30:2 had slightly higher survival than those patients treated under the 2000 guidelines with a compression-ventilation ratio of 15:2 (11.0% versus 7.0%; RR 1.75 [1.32–2.04]).

For the critical outcome of survival to 30 days, we identified very-low-quality evidence from 1 observational study¹⁵⁷ that was downgraded for risk of bias and indirectness. Patients treated under the 2005 guidelines had slightly higher survival than those patients treated under the 2000 guidelines (16.0% versus 8.3%; RR 1.92 [1.28-2.87]).

For the critical outcome of any ROSC, we identified very-low-quality evidence from 2 observational studies.^154,155 The studies were downgraded for risk of bias and indirectness. Patients treated under the 2005 guidelines had a ROSC more often than those patients treated under the 2000 guidelines (38.7% versus 30.0%; RR 1.30 [1.14–1.49]).

For the critical outcome of ROSC at hospital admission, we identified very-low-quality evidence from 2 observational studies.^155,157 The studies were downgraded for risk of bias and indirectness. Of the 1708 patients included, those treated under the 2005 guidelines had ROSC at hospital admission more often than those patients treated under the 2000 guidelines (34.5% versus 17.1%; RR 2.02 [1.69–2.41]).

For the important outcome of hands-off time, we identified very-low-quality evidence from 2 observational studies^155,156 that were downgraded for risk of bias and indirectness. Patients who were treated with the use of the 2005 guidelines had less hands-off time than those patients treated under the 2000 guidelines.

We suggest a compression-ventilation ratio of 30:2 compared with any other compression-ventilation ratio in patients in cardiac arrest (weak recommendation, low-quality evidence).

In making this recommendation, we placed a high priority on consistency with our 2005 and 2010 treatment recommendations and the findings identified in this review, which suggest that the bundle of care (which included changing to a compression to ventilation ratio of 30:2 from 15:2) resulted in more lives being saved. We note that there would likely be substantial resource implications (eg, reprogramming, retraining) associated with a change in recommendation, and an absence of any data addressing our critical outcomes to suggest our current recommendation should be changed.

The 2005 and 2010 CoSTR publications recommended that pausing chest compressions to undertake a rhythm check should occur every 2 minutes. This recommendation is supported by indirect evidence that rescuer fatigue occurs by about 2 minutes, and a rhythm check is a natural time point, when possible, to change the compressor.

Feedback from the public comment period for this publication supported maintaining consistency with previous recommendations.

There are currently no studies that directly address the question of optimal CPR intervals and their effect on the identified critical outcomes of survival with favorable neurologic or functional outcome at discharge or survival only at discharge or the important outcomes of ROSC, coronary perfusion pressure, cardiac output.

We suggest pausing chest compressions every 2 minutes to assess the cardiac rhythm (weak recommendation, low-quality evidence).

In making this recommendation, we placed a high priority on consistency with previous recommendations and the absence of contradictory evidence to prompt a change. We placed value on simplifying resuscitation logistics by coordinating rhythm and pulse checks with standard recommendations for rotating the provider performing chest compressions every 2 minutes.

In the 2015 PICO development process, additional questions were developed to address knowledge gaps identified in 2010. As a result of significant similarities in other BLS PICO questions, very rigid study inclusion criteria were applied. Only comparative human studies assessing the PICO listed outcomes were considered.

Of the 654 articles found during the search, and a follow-up search performed in early 2015 identifying a potential additional 112 studies, none were found to relate to the specific question.

Outside of the ALS environment where invasive monitoring is available, there is insufficient data around the value of a pulse check while performing CPR. We therefore do not make a treatment recommendation regarding the value of a pulse check.

Emphasis should remain on minimizing interruptions in chest compressions and avoiding pausing for a pulse check without strong suspicion of ROSC (eg, clinically or by hemodynamic monitoring).

The use of CPR feedback or prompt devices during CPR in clinical practice or CPR training is intended to improve CPR quality as a means to improving ROSC and survival. The forms of CPR feedback or prompt devices include audio and visual components such as voice prompts, metronomes, visual dials, numerical displays, waveforms, verbal prompts, and visual alarms. Visual displays enable the rescuer to see compression-to-compression quality parameters, including compression depth and rate, in real time. All audio prompts may guide CPR rate (eg, metronome) and may offer verbal prompts to rescuers (eg, “push harder,” “good compressions”).

The use of CPR feedback or prompt devices should be considered as part of a broader system of care that should include comprehensive CPR quality improvement initiatives, rather than an isolated intervention. The reader is referred to the relevant sections of the Education, Implementation, and Teams Task Force recommendations, EIT 640: Measuring Quality of Systems, EIT 645: Debriefing of Resuscitation Performance, and EIT 648: CPR Feedback Devices During Training (see “Part 8: Education, Implementation, and Teams”).

This review identified 12 studies, of which 2 studies were randomized studies^133,137 and 10 studies were observational of before-after design.^{131,140,143,159–165} The included studies were 9 studies in 3716 adults^{131,133,137,140,159,161–164} and 3 studies of 34 pediatric patients.^143,160,165 Four studies included patients with in-hospital cardiac arrest,^{143,159,160,165} 7 studies with OHCA^{133,137,140,161–164} and 1 study¹³¹ had a mixture of patients from in and out-of-hospital settings.

Feedback devices examined included accelerometer-based devices^{133,137,140,143,159,161,163–165} and audiotape of prompts.^131,160,162 Compared with the previous evidence review performed in 2010, this review identified 8 new studies that were included for analysis.^{133,137,140,143,161,163–165} The nature of using feedback or prompt devices meant that all studies suffered from performance and detection bias because healthcare professionals were not blinded to intervention (feedback or no feedback).

For the critical outcome of favorable neurologic outcome, we identified moderate-quality evidence from 1 cluster-randomized study¹³³ representing 1586 patients, and very-low-quality evidence from 2 observational studies in adults^161,164 representing 670 patients. All studies were downgraded due to risk of bias. The randomized trial found no difference in the number of patients who achieved favorable neurologic outcome (control 10.1% versus 10.3%, P=0.855). No studies showed a statistically significant difference in favorable neurologic outcome with the use of CPR feedback. Effect of CPR feedback on survival with favorable neurologic outcome ranged from −0.8 to 5.8%.

For the critical outcome of survival to hospital discharge, we identified moderate-quality evidence from 1 cluster-randomized study¹³³ representing 1586 patients, and very-low-quality evidence from 4 observational studies in adults,^{140,159,161,164} and 1 observational study in children¹⁶⁵ representing 1192 patients. All studies were downgraded due to risk of bias. The randomized trial found no difference in the number of patients who achieved survival to hospital discharge (control 44.7% versus 44.3%, P=0.962). No studies showed a statistically significant difference in survival to hospital discharge with the use of CPR feedback. The effect of CPR feedback on survival to hospital discharge ranged from −0.9 to 5.2.

For the critical outcome of ROSC, we identified moderate-quality evidence from 2 randomized studies^133,137 representing 1886 patients, and very-low-quality evidence from 7 observational studies in adults^{137,140,159,161–164} and 1 observational study in children¹⁶⁵ representing 3447 patients. All studies were downgraded due to risk of bias. The randomized trial found no difference in the number of patients who achieved ROSC (control 44.7% versus 44.3%, P=0.962). Only 1 study¹⁶⁴ showed a statistically significant difference in ROSC with the use of feedback; however, in this study, feedback was activated at the discretion of the physician, and no details were provided regarding the decision-making process to activate or not activate feedback. Effect of CPR feedback on ROSC ranged from −4.4% to 17.5%: 1 study demonstrated a 50% increase in ROSC with CPR feedback; however, this small study had only 4 patients in each group.¹⁶⁵

For the important outcome of chest compression rate, we identified moderate-quality evidence from 2 randomized studies^133,137 representing 1474 patients, and very-low-quality evidence from 4 observational studies: 3 in adults^140,159,161 representing 777 patients, and 1 in children¹⁶⁵ representing 8 patients. All studies were downgraded due to risk of bias. The cluster RCT¹³³ found a significant difference of −4.7/min (95% CI, −6.4 to −3.0/min) when feedback was used, and the prospective randomized trial¹³⁷ showed no difference in compression rates with and without feedback. In both randomized trials, compression rates were all close to international recommendations of 100/min. One observational study¹⁵⁹ showed no difference in chest compression rates with and without feedback, and, again, all compression rates were close to international recommendations of 100/min. The other 2 observational studies^140,161 showed lower compression rates in the group with CPR feedback: 128 to 106/min (difference, −23; 95% CI, −26 to −19)¹⁶¹ and 121 to 109/min (difference, −12; 95% CI, −16 to −9).¹⁴⁰ The pediatric study¹⁶⁵ found a median difference of −10/min with feedback, and, again, the chest compression rate in the control group exceeded 120/min. The use of CPR feedback devices may be effective in limiting compression rates that are too fast.

For the important outcome of chest compression depth, we identified moderate-quality evidence from 2 randomized studies^133,137 representing 1296 patients, and very-low-quality evidence from 4 observational studies: 3 in adults^140,159,161 representing 777 patients and 1 in children¹⁶⁵ representing 8 patients. All studies were downgraded due to risk of bias. The cluster RCT¹³³ found a significant +1.6 mm (95% CI, 0.5–2.7) (cluster adjusted) difference in chest compression depth with feedback. However, this is of questionable clinical significance, and the average compression depths in both arms were less than international recommendations of 5 cm (2 inches) in adults (3.96 cm [1.55 inches] and 3.87 cm [1.52 inches]). The prospective randomized trial¹³⁷ found no significant difference in compression depth with and without feedback, and all compression depths were close to, but slightly less than, international recommendations of 5 cm (2 inches) in adults. One observational study¹⁵⁹ showed no difference in chest compression depth with and without feedback, and all compression rates were close to, but less than, international recommendations of 5 cm (2 inches) in adult patients (4.4 and 4.3 cm or 1.7 inches). Two observational studies^140,161 showed significantly deeper chest compressions in the groups with CPR feedback: Bobrow et al¹⁶¹ found a 1.06 cm (0.42 inches) increase with feedback (5.46 versus 4.52 cm, or 2.15 versus 1.78 inches) (mean difference, 0.97 cm; 95% CI, 0.71–1.19 cm), while the findings by Kramer-Johansen¹⁴⁰ were more modest (increase from 3.4 to 3.88 cm, or [from 1.3 to 1.5 inches]) (mean difference, 0.4 cm; 95% CI, 0.2–0.6). The pediatric study¹⁶⁵ found no median difference in compression depth. The use of CPR feedback devices did not seem to make an appreciable difference in chest compression depth.

For the important outcome of chest compression fraction, we identified moderate-quality evidence from 1 randomized study¹³³ and very-low-quality evidence from 3 observational studies in adults^140,159,161 and 1 in children.¹⁶⁵ All studies were downgraded due to risk of bias. The randomized study found a cluster adjusted difference of +1.9% (65.9% versus 64.0%; P=0.016) when CPR prompt devices were used. Although statistically significant, such a small difference has questionable clinical significance. The adult studies found no significant difference between groups, and the sample size of the pediatric study was too small to enable inferential statistical analysis. The use of CPR feedback devices did not seem to make an appreciable difference in chest compression fraction.

For the important outcome of ventilation rate, we identified very-low-quality evidence from 3 observational studies in adults^140,159,161 representing 532 patients. All studies were downgraded due to risk of bias. None of the studies showed a significant difference in ventilation rate with and without CPR feedback.

For the important outcome of ETCO₂, we identified very-low-quality evidence from 2 observational studies in adults^131,161 representing 131 patients. All studies were downgraded due to risk of bias. Kern¹³¹ found that the ETCO₂ was significantly higher when CPR feedback was used (+6.3 mm Hg with compression rate feedback of 120/min and +4.3 mm Hg with compression rate feedback of 80/min). Bobrow¹⁶¹ found an absolute difference of −2.2 mm Hg with CPR feedback. The clinical significance of these differences is questionable.

For the important outcome of leaning force during chest compressions, we identified very-low-quality evidence from 1 observational study in children¹⁴³ representing 20 patients. This study was downgraded due to risk of bias. Leaning force was decreased by 0.9 kg with the use of feedback.

We suggest the use of real-time audiovisual feedback and prompt devices during CPR in clinical practice as part of a comprehensive system for care for cardiac arrest (weak recommendation, very-low-quality evidence).

We suggest against the use of real-time audiovisual feedback and prompt devices in isolation (ie, not part of a comprehensive system of care) (weak recommendation, very-low-quality evidence).

In making this recommendation, we place a higher value on development of systems of care with continuous quality improvement than on cost. Resource-poor environments may choose not to adopt this technology in favor of allocating resources to other system developments. Devices that provide real-time CPR feedback also document CPR metrics that may be used to debrief and inform strategies aimed at improving CPR quality. Currently available audiovisual feedback devices provide information on key CPR parameters such as compressions and ventilation; however, the optimal targets and the relationships among different targets have not been fully defined.

The treatment of a patient with OHCA is extremely complex from several perspectives. Operationally, there is very significant heterogeneity in EMS systems design (eg, emergency medical responder versus emergency medical technician versus paramedic versus physician based) and resource availability (eg, number of rescuers, equipment, evidence-based protocols, quality improvement programs). Clinically, there is extreme patient and arrest location variability, with patients located in urban, rural, and remote patient settings and in unpredictable and diverse environments with variable rates of bystander CPR and AED use. Logistically, the initial EMS care of an OHCA victim involves several concurrent goals with complex interactions of scene safety, patient assessment, patient care, communication, extrication, and transport.

This systematic review found no studies that directly addressed the question of EMS compression-only CPR compared with conventional CPR. Four North American observational studies were identified, which adopted a bundled intervention for adult patients with a presumed cardiac cause of their cardiac arrest.^166–169 EMS response intervals in these settings were generally within 5 to 6 minutes. The bundled interventions were broadly similar and comprised 200 initial chest compressions, a single shock rather than stacked shocks, and immediate resumption of a further 200 chest compressions before rhythm/pulse check. Epinephrine was given early and endotracheal intubation delayed. Basic airway adjuncts were used with either passive oxygen insufflation or bag-mask ventilation (ventilation rate 8/min). The findings from the review provide evidence about the effects of the bundled intervention rather than delayed ventilation in isolation.

For the critical outcome of survival to hospital discharge with favorable neurologic outcome in all OHCAs, we have identified very-low-quality evidence (downgraded for risk of bias and indirectness) from 1 observational trial¹⁶⁸ that enrolled 1019 patients showing no benefit (unadjusted OR, 1.07; 95% CI, 1.41–8.79).

For the critical outcome of survival with favorable neurologic outcome in the subset of witnessed arrest/shockable rhythm OHCA, we identified very-low-quality evidence (downgraded for risk of bias and indirectness) from 3 observational studies^166–168 that enrolled 1325 patients showing benefit: OR, 3.6 (95% CI, 1.77–7.35)¹⁶⁶; OR, 5.24 (95% CI, 2.16–12.75)¹⁶⁷; and adjusted OR, 2.5 (95% CI, 1.3–4.6).¹⁶⁸

For the critical outcome of survival to hospital discharge in the subgroup of all OHCAs, we identified very-low-quality evidence (downgraded for risk of bias, indirectness, and imprecision) from 3 observational studies^169,170 showing benefit: OR, 3.26 (95% CI, 2.46–4.34)¹⁶⁹; OR, 2.50 (95% CI, 1.75–3.58; cohort)¹⁷⁰; and OR, 3.05 (95% CI, 1.07–8.66; before-after).¹⁷⁰

For the critical outcome of survival to hospital discharge in the subgroup of witnessed arrest/shockable rhythm cardiac arrest, we identified very-low-quality evidence (downgraded for indirectness and imprecision) in 3 observational studies^166,167,170 that showed benefit: OR, 3.67 (95% CI, 1.98–7.12)¹⁶⁶; OR, 5.58 (95% CI, 2.36–13.20)¹⁶⁷; OR, 2.94 (95% CI, 1.82–4.74); and OR, 4.3 (95% CI, 0.98–19.35).¹⁷⁰

For the critical outcome of ROSC in all out of hospital cardiac arrest patients, we identified very-low-quality evidence (downgraded for risk of bias and indirectness) in 1 observational study¹⁶⁸ showing no benefit (OR, 0.85; 95% CI, 0.64–1.11) to EMS provision of chest compressions with delayed ventilation.

We suggest that where EMS systems* have adopted bundles of care involving minimally interrupted cardiac resuscitation†, the bundle of care is a reasonable alternative to conventional CPR for witnessed shockable out of hospital cardiac arrest (weak recommendation, very-low-quality evidence).

This recommendation places a relatively high value on (1) the importance of provision of high-quality chest compressions and (2) simplifying resuscitation logistics in the out-of-hospital setting in a defined EMS system with demonstrated clinical benefit, and a relatively low value on the uncertain effectiveness, acceptability, feasibility, and resource use in different EMS systems compared with those in this CoSTR.

We acknowledge the pending results of the important and very large clinical trial NCT01372748 with a primary aim to compare survival at hospital discharge after continuous chest compressions versus conventional CPR with interrupted chest compressions in patients with OHCA.

The following knowledge gaps currently exist:

During chest compression–only CPR in the out of hospital setting, some described EMS systems have chosen to provide passive ventilation in the form of an airway maneuver and/or device combined with an oxygen-delivery mask. No studies were found describing this in the lay rescuer setting.

Three studies were identified. Two compared intermittent positive-pressure ventilation via an endotracheal tube with continuous insufflation of oxygen through a modified endotracheal tube.^171,172 The third study compared placement of an oropharyngeal airway and administration of oxygen by nonrebreather mask or by bag-mask ventilation during a bundle of care involving 200 continuous chest compressions and delayed intubation.¹⁶⁸

For the critical outcome of favorable neurologic outcome, we identified very-low-quality evidence (downgraded due to serious risk of bias and indirectness) from 1 retrospective study, which involved 1019 patients that showed no difference between passive (nonrebreather mask) and active (bag-mask) ventilation¹⁶⁸ (adjusted OR, 1.2; 95% CI, 0.8–1.9).

For the critical outcome of survival, we found very-low-quality evidence from a single retrospective study (downgraded for serious indirectness and risk of bias).¹⁶⁸ This study reported no significant difference in survival (RR, 1.1; 95% CI, 0.72–1.54).

For the critical outcome of ROSC, we found very-low-quality evidence (downgraded for serious indirectness and risk of bias) from 2 RCTs^171,172 and 1 observational study.¹⁶⁸ None of the studies showed any significant difference: OR, 0.88 (95% CI, 0.6–1.3)¹⁷²; OR, 0.8 (95% CI, 0.7–1.0)¹⁶⁸; and RR, 1.27 (95% CI, 0.6–2.61).¹⁷¹

We suggest against the routine use of passive ventilation techniques during conventional CPR (weak recommendation, very-low-quality evidence).

We suggest that where EMS systems have adopted bundles of care involving continuous chest compressions, the use of passive ventilation techniques may be considered as part of that bundle for patients in OHCA (weak recommendation, very-low-quality evidence).

In making this recommendation, we place priority on consistency with our previous recommendations in the absence of compelling evidence for improvement in any of our critical outcomes. We acknowledge that where EMS systems have adopted a bundle of care that includes passive ventilation, it is reasonable to continue in the absence of compelling evidence to the contrary.

Many lay rescuers are concerned that delivering chest compressions to a person who is not in cardiac arrest could lead to serious complications and, thus, are reluctant to initiate CPR even when a person is actually in cardiac arrest. Studies reporting harm from CPR to persons not in cardiac arrest were reviewed.

For the important outcome of “harm,” we identified very-low-quality evidence (downgraded for risk of bias and imprecision) from 4 observational studies enrolling 762 patients who were not in cardiac arrest and received CPR by lay rescuers outside the hospital.^173–176 Three of the studies^173–175 reviewed the medical records to identify harm, and 1 included follow-up telephone interviews.¹⁷³ Pooled data from these 3 studies, encompassing 345 patients, found an incidence of bone fracture (ribs and clavicle) of 1.7% (95% CI, 0.4%–3.1%), pain in the area of chest compression of 8.7% (95% CI, 5.7%–11.7%), and no clinically relevant visceral injury. The fourth study¹⁷⁶ relied on fire department observations at the scene, and there were no reported injuries in 417 patients.

We recommend that laypersons initiate CPR for presumed cardiac arrest without concerns of harm to patients not in cardiac arrest (strong recommendation, very-low-quality evidence).

In making this recommendation, we place a higher value on the survival benefit of CPR initiated by laypersons for patients in cardiac arrest against the low risk of injury in patients not in cardiac arrest.

Population (eg, rates of witnessed arrest) and EMS program (eg, response intervals) characteristics affect survival and can vary considerably. The concept of early defibrillation is well established in improving outcome from cardiac arrest. This review identified 15 relevant studies (1 RCT and 14 observational studies) spanning the years 2002–2013, with the associated variations in recommended practice of bystander CPR during these periods. The authors recognize that some studies may involve repeat analysis and reporting of the same cardiac arrest population, which limits the ability to provide a summative effect measure in the reported analyses.

For the critical outcome of survival to 1 year with favorable neurologic outcome, we identified very-low-quality evidence (downgraded for risk of bias) from 1 observational trial¹⁷⁷ enrolling 1394 patients showing improved outcomes with public-access defibrillation (unadjusted OR, 3.53; 95% CI, 1.41–8.79).

For the critical outcome of survival 30 days with favorable neurologic outcome, we identified very-low-quality evidence (downgraded for inconsistency and indirectness) from 3 observational studies^178–180 enrolling 182 119 patients demonstrating improved survival (range, 31.6%–55%) with public-access defibrillation compared with no program (range, 3%–37%).

For the critical outcome of survival to discharge with favorable neurologic outcome, we identified very-low-quality evidence (downgraded for risk of bias, inconsistency, and imprecision) from 1 randomized trial¹⁸¹ and 3 observational studies.^177,182,183 The randomized trial enrolled 235 patients and found no difference in favorable neurologic outcomes (CPC, 1–2; RR, 1.73; 95% CI, 0.95–3.19). The observational studies included 4581 patients demonstrating improved survival (range, 4.1%–50%) with public-access defibrillation compared with no program (1.4%–14.8%), and 1 observational pilot study (20 patients)¹⁸⁴ showing reduced survival (0% versus 30.7) with public-access defibrillation compared with no program.

For survival to 30 days, we identified very-low-quality evidence (downgraded for indirectness) from 3 observational studies^178,180,185 enrolling 14 135 patients demonstrating improved survival (range, 37.2%–65.5%) with public-access defibrillation compared with no program (23.3%–48.5%). If combined in a formal meta-analysis, a summary effect measure of OR 1.63 (95% CI, 1.41–1.88) would be generated. However, we recognize the limitations of significant heterogeneity in the study populations and the fact that some patient data were reported in more than 1 publication.

For survival to discharge, we identified very-low-quality evidence (downgraded for risk of bias, indirectness, and imprecision) from 1 randomized trial¹⁸¹ and 9 observational studies.^{177,182,183,186–191} The randomized trial enrolled 235 participants and observed improved survival (adjusted RR, 2.0; 95% CI, 1.07–3.77). The observational studies enrolled 46 070 patients demonstrating improved survival (range, 4.4%–51%) with public -access defibrillation compared with no program (1.4%–25.0%) and 1 observation pilot study (20 patients)¹⁸⁴ showing reduced survival (0% versus 30.7%) when public-access defibrillation programs were present.

We recommend the implementation of public-access defibrillation programs for patients with OHCAs (strong recommendation, low-quality evidence).

In making this recommendation, we considered the societal impact of delayed defibrillation and balanced this against the costs of setting up a comprehensive public-access defibrillation program. We place a higher value on a single randomized trial supported by multiple large-scale, international observational studies. Together, these indicate that the magnitude of change on outcome may vary based on the setting or community within which programs are introduced. Public sites with large population densities may benefit the most from public-access defibrillation programs.

The following knowledge gaps currently exist:

The 2010 CoSTR stated that interruptions in chest compressions during CPR must be minimized. Legitimate reasons for the interruption of CPR were highlighted as the needs to ventilate, to assess the rhythm or ROSC, and to defibrillate. This question sought to identify the optimal timing of rhythm checks in relation to attempted defibrillation. Public comments on this question during the consultation process highlighted concerns about drug administration without first confirming whether ROSC had been achieved after shock delivery. This latter concern fell outside the remit of this question but has been highlighted as an area requiring future research.

This review identified 5 observational studies relevant to this question. In each case, the studies evaluated omitting a rhythm check immediately after CPR as part of a bundle of interventions (eg, elimination of 3 stacked shocks and postshock rhythm and pulse checks). Thus, the evidence presented in this review must be considered as indirect evidence with respect to this narrow question.

For the critical outcome of survival with favorable neurologic outcome at discharge, we identified very-low-quality evidence (downgraded for serious risk of bias, indirectness, and imprecision) from 3 observational studies enrolling 763 OHCAs showing a harmful effect for interrupting chest compressions to check rhythm immediately after shock delivery (RR, 0.62; 95% CI, 0.51–0.75).^167,170,192

For the critical outcome of survival hospital discharge, we identified low-quality evidence from 1 RCT enrolling 845 OHCAs showing no benefit for interrupting chest compressions to check rhythm immediately after shock delivery (RR, 0.80; 95% CI, 0.55–1.15)¹⁴⁸ and very-low-quality evidence (downgraded for serious risk of bias and indirectness) from 3 observational studies enrolling 3094 OHCAs showing a harm effect for checking rhythm immediately after defibrillation (RR, 0.55; 95% CI, 0.45–0.67).^167,170,192 In addition, for the same outcome, we identified very-low-quality evidence from 1 observational study of 528 victims of OHCA showing potential harm for interrupting chest compressions to check rhythm immediately after shock delivery (RR, 0.42; 95% CI, 0.29–0.61).¹⁷⁰

For the critical outcome of survival to hospital admission, we identified low-quality evidence from 1 RCT enrolling 845 victims of OHCA showing no benefit for interrupting chest compressions to check rhythm immediately after shock delivery (RR, 0.99; 95% CI, 0.85–1.15).¹⁴⁸

For the critical outcome of ROSC, we identified very-low-quality evidence (downgraded for serious risk of bias and indirectness) from 2 observational studies enrolling 2969 victims of OHCA showing a harm effect for interrupting chest compressions to check rhythm immediately after shock delivery (RR, 0.69; 95% CI, 0.61–0.78).^170,192

For the important outcome of recurrence of VF, we identified low-quality evidence from 1 RCT, enrolling 136 OHCAs showing no benefit for interrupting chest compressions to check rhythm immediately after shock delivery (RR, 1.00; 95% CI, 0.81–1.23).¹⁹³

We suggest immediate resumption of chest compressions after shock delivery for adults in cardiac arrest in any setting (weak recommendation, very-low-quality evidence).

If there is alternative physiologic evidence of ROSC (eg, arterial waveform or rapid rise in ETCO₂), chest compressions can be paused briefly for rhythm analysis.

In making this recommendation, we place a higher value on avoiding interruptions in chest compressions for an intervention showing no benefit for critical and important outcomes. Also, this recommendation assumes that shocks are generally effective and that a perfusing rhythm is generally not present immediately after elimination of VF.

Motion artifacts effectively preclude the possibility of reliably assessing the heart rhythm during chest compressions. This has 2 undesirable consequences: first, it forces the rescuer to stop chest compressions to assess the rhythm to determine if a shock (or another shock) is required. Second, during chest compressions, possible recurrence of VF cannot be recognized, eliminating the possible beneficial effect of immediate defibrillation in case of refibrillation. Some modern defibrillators contain filtering modalities that allow visual or automated rhythm analysis during chest compressions. This review sought to examine the use of such technology to determine if it leads to better clinically meaningful outcomes in human cardiac arrest.

There are currently no human studies that address the identified critical outcomes criteria of favorable neurologic outcome, survival, or ROSC or the important outcomes criteria of CPR quality, time to commencing CPR, or time to first shock.

We suggest against the introduction of artifact-filtering algorithms for analysis of electrocardiographic rhythm during CPR unless as part of a research program.

We suggest that where EMS systems have already integrated artifact-filtering algorithms into clinical practice, it is reasonable to continue with their use.

In making this recommendation, we placed priority on avoiding the costs of introducing a new technology where the effectiveness or harm on patient outcomes remains to be determined. Where such technologies have already been implemented, we place priority on avoiding the likely costs and inconvenience of their withdrawal from practice. We encourage such systems to report on their experiences to build the evidence base regarding the use of these technologies in clinical practice.