Article Text

Download PDFPDF

Cardiopulmonary resuscitation: the science behind the hands
  1. Andrew W Harris,
  2. Peter J Kudenchuk
  1. University of Washington, Division of Cardiology, Seattle, Washington, USA
  1. Correspondence to Dr Peter J Kudenchuk, University of Washington, Division of Cardiology, Box 346422, 1959 NE Pacific Street, Seattle, WA USA; pkudenchuk{at}


Sudden cardiac arrest is a leading cause of death worldwide. Despite significant advances in resuscitation science since the initial use of external chest compressions in humans nearly 60 years ago, there continues to be wide variability in rates of successful resuscitation across communities. The American Heart Association (AHA) and European Resuscitation Council emphasise the importance of high-quality chest compressions as the foundation of resuscitation care. We review the physiological basis for the association between chest compression quality and clinical outcomes and the scientific basis for the AHA’s key metrics for high-quality cardiopulmonary resuscitation. Finally, we highlight that implementation of strategies that promote effective chest compressions can improve outcomes in all patients with cardiac arrest.

  • cardiac arrest

Statistics from

Request Permissions

If you wish to reuse any or all of this article please use the link below which will take you to the Copyright Clearance Center’s RightsLink service. You will be able to get a quick price and instant permission to reuse the content in many different ways.


Sudden cardiac arrest is a leading cause of death worldwide.1 2 Significant scientific advances have transformed the care of the cardiac arrest victim in the 57 years since Kouwenhoven and colleagues3 first reported the successful clinical use of closed-chest compressions. The American Heart Association (AHA) and European Resuscitation Council each published guidelines in 2015 that reflect the current state of the science of resuscitation. These guidelines highlight the importance of high-quality cardiopulmonary resuscitation (CPR) and emphasise the importance of effective chest compressions as the foundation of resuscitation care.1 2 This article will summarise the evidence supporting these recommendations and the improved outcomes achievable with high-quality CPR.

Importance of chest compressions

Defibrillation is regarded as the definitive intervention for cardiac arrest due to ventricular fibrillation (VF) or pulseless ventricular tachycardia (VT). Although success is high during the initial few minutes after collapse, survival from witnessed VF/VT cardiac arrest declines by approximately 10% per minute that defibrillation is delayed. In such circumstances, chest compressions are essential for generating blood flow to the myocardium, brain and other vital organs. Initial canine experiments of external defibrillation in 1943 showed that applying an electrical defibrillatory shock within 1–1.5 min of onset of VF was highly successful in restoring normal sinus rhythm, but less so after longer durations of the arrhythmia. Conversely, when cardiac massage was administered prior to defibrillation, the duration of VF preceding a successful shock could be extended.4 There is an apparent synergistic relationship between CPR and defibrillation clinically, in that the rate of decline in survival with each passing minute from collapse to defibrillation is cut in half when CPR is performed.

A possible explanation for this phenomenon arises from experimental animal work that serially assayed myocardial concentrations of ATP—a high energy compound that drives cellular metabolism—during protracted periods of VF in the absence of CPR. In this model, ATP concentrations declined significantly over time, reaching concentrations that were less than half of normal after only 5 min of unsupported VF (figure 1).5 A corresponding time-dependent diminution in the amplitude and apparent coarseness of the VF waveform has also been observed in clinical VF, such that in its late stages the arrhythmia is often indistinguishable from asystole (figure 2). It is likely that the declining vigour of the VF waveform is a result of consumption of high-energy phosphate molecules resulting in a progressively energy-depleted myocardium. Another possible expression of this energy-depleted state is the heart’s poor contractile function after a protracted period of VF has been terminated by shock. This often results in asystole or pulseless electrical activity, and in fact serves as the basis for creating such rhythms in experimental models.6 7 By providing blood flow to the heart during cardiac arrest, might CPR mitigate or even reverse the loss of myocardial energetics associated with these changes in the VF waveform, poor contractile function after shock, and ultimately death? There is evidence to suggest this might be the case.

Figure 1

Bar graphs of myocardial adenine nucleotide concentrations during normal sinus rhythm (NSR) and at various durations of untreated ventricular fibrillation (VF). All tissue concentrations expressed in nmol/mg protein (from Neumar et al 5). Reproduced with permission from Elsevier.

Figure 2

Typical changes in ventricular fibrillation (VF) waveform in untreated VF (after 1 and 10 min) and after 3 min of cardiopulmonary resuscitation-first (13 min) in swine. AMP, amplitude (in mV); MF, VF median frequency (in Hz) (from Berg et al 8). Reproduced with permission from Elsevier.

In one such study, animals were subjected to 10 min of VF without CPR, then randomised to either immediate defibrillation or to receive 3 min of CPR before shock. VF amplitude and frequency characteristics were sampled in all animals after 1 min of VF (without CPR), at 10 min (without CPR) and again at 13 min (after a 3 min period of CPR) in the group of animals that were randomised to receive preshock CPR. These characteristics of VF expectedly deteriorated during the untreated period of VF, with the waveform becoming visually (and numerically) less coarse at 10 min than at 1 min into the VF episode. Notably, these characteristics improved significantly following 3 min of chest compressions, which virtually brought these parameters and the visual appearance of VF back to its characteristics measured after 1 min of VF (figure 2). Furthermore, after up to three shocks, none of the animals in the group not receiving CPR were resuscitated, compared with 50% of those receiving chest compressions prior to defibrillation.8 Similar findings have emerged from clinical studies that showed improved survival to hospital discharge when CPR was provided for a prescribed time period before defibrillation rather than an immediate shock approach in patients with out-of-hospital cardiac arrest (OHCA) due to VF.9

CPR mechanisms

Conventional CPR occurs in two phases, compression and decompression (figure 3). External chest compression results in arterial blood flow due to phasic changes in intrathoracic pressure produced by the force generated by the hands. Importantly, in closed-chest CPR, the increased intrathoracic pressure and reduced thoracic volume created by chest compressions squeeze the heart itself and all blood-containing structures in the chest, including the lungs and great vessels. Thus, conventional CPR takes advantage of the entire chest, not just the heart, as its ‘pump’. Chest compression leads to a relatively uniform rise in the pressure of all intrathoracic vascular structures, and it is this pressure that ejects blood from the thorax during this phase of CPR. Thoracic pressure is transmitted to the body via the great arteries, but fortunately not via the great veins where this is prevented by the closure of venous valves at the thoracic outlet. Were this not the case, high venous pressures could neutralise the necessary gradient between arterial and venous beds that is critical to organ perfusion.10 All told, the greater the vigour of these chest compressions, the greater the resulting cardiac (actually thoracic) output and magnitude of cerebral blood flow.11

Figure 3

Haemodynamic effects of compression and decompression phases of cardiopulmonary resuscitation. Compression phase creates organ perfusion pressure (difference between aortic and extrathoracic vein pressure). Decompression phase creates myocardial perfusion pressure (difference between aortic and right atrial pressure). Data adapted from Criley et al. Circulation. 1986; 74 (Suppl IV): 42-50.

‘Diastole’ occurs during the decompression phase of the CPR cycle—when the thorax is permitted to rebound to its normal fully expanded configuration. While seemingly passive and unimportant, CPR decompression may be even more important than its compression phase. During this phase, closure of the aortic valve maintains an aortic pressure that is higher than intracardiac pressures, which fall precipitously beneath the closed valve, driven by the ‘vacuum’ effect created by the recoiling thoracic cage. This intrathoracic vacuum is what draws blood to return back into the chest from the periphery, filling the heart, lungs and great vessels in preparation for the next chest compression. The result is that the better the decompression, the stronger the vacuum, the better the refilling, and ultimately the more thoracic blood available for subsequent compression.

CPR diastole serves a second useful purpose that is often unappreciated. All organs of the body are perfused during the compression phase of CPR except, ironically, the heart itself, which is not perfused by blood within its chambers, but rather by the coronary arteries. Coronary blood flow is estimated by coronary perfusion pressure (CPP), defined as the difference between pressure in the aorta (where the coronary arteries originate) and in the right atrium (where coronary venous blood ultimately returns).12 During the compression phase of CPR, all intrathoracic pressures including the aorta and all chambers of the heart equalise with the compressive force of the hands, resulting in no coronary blood flow between the aorta and right atrium. Conversely, in CPR diastole, the higher aortic pressure above its closed valve compared with the falling intrathoracic pressure results in a positive CPP. It is this combination of compressive force producing higher aortic pressures both in systole and in diastole, along with its release causing a fall in intrathoracic pressure, that results in optimal organ perfusion, including the heart itself.

Saving the heart is an essential first step in resuscitation, without which there can be no return of spontaneous circulation (ROSC). Animal studies have shown that there is excellent correlation between CPP and myocardial blood flow during CPR.13 Furthermore, it has been noted that CPP during CPR has significant prognostic implications on outcomes after resuscitation. Among 100 patients in whom invasive aortic and right atrial pressure measurements were obtained during CPR, CPP was directly associated with the likelihood of achieving ROSC. The 24 patients who achieved ROSC had a maximal CPP of 25.6 mm Hg vs 8.4 mm Hg (P<0.0001); among those not achieving a CPP of at least 15 mm Hg, none survived.14

These findings provide the physiological rationale for optimising haemodynamics during CPR, in which the goal is to improve cerebral and myocardial blood flow, achieve ROSC, and ultimately increase survival with good neurological outcome.

Acute coronary occlusion and/or severe coronary stenoses are commonly found in patients with cardiac arrest. These add to the challenge of achieving adequate coronary perfusion during CPR and further underscore the importance of efforts to optimise CPP. In some cases, emergent coronary revascularisation and/or extracorporeal CPR may be required for successful resuscitation, a discussion of which is beyond the scope of this article.

Key metrics for quality CPR

The AHA has identified five key metrics for high-quality CPR on the basis of improved haemodynamics during CPR and improved clinical outcomes: (1) chest compression rate, (2) chest compression depth, (3) chest compression fraction and minimising pauses, (4) allowing full chest recoil, and 5) controlled ventilation with avoidance of hyperventilation.1

Chest compression rate (AHA recommendation: 100–120/min)

Early studies of chest compression rates in dogs showed higher CPP, mean arterial pressure and improved rates of 24-hour survival when randomised to receive chest compressions at a rate of 120/min compared with 60/min.15 A recent study from the Resuscitation Outcomes Consortium (ROC) corroborated this association between chest compression rates with outcome in 3098 patients with OHCA, among whom the probability of survival to hospital discharge was highest at ~110–120 compressions/min, with a notable decline in survival on each side of this range (figure 4).16 The optimal compression rate represents a balance that attempts to maximise key physiological parameters during CPR. Because stroke volume is lower during CPR, a higher compression rate is necessary to achieve a sufficient cardiac output for organ perfusion. However, excessive chest compression rates may also lead to reduced cardiac output in shortening the diastolic time interval (‘CPR diastole’) required for thoracic refilling and for myocardial perfusion. Furthermore, an inverse association has been observed between chest compression rate and depth, especially at higher rates. In one study, the mean chest compression depth deteriorated (<38 mm) in more than 80% of patients when compression rates were >140/min.16 Thus, maintaining rates within the recommended range helps to optimise key physiological parameters and other important CPR metrics.

Figure 4

Adjusted cubic spline of the relationship between chest compression rates and the probability of survival to hospital discharge. The adjusted model includes sex, age, bystander witnessed arrest, Emergency Medical Services (EMS) witnessed arrest, first known EMS rhythm, attempted bystander cardiopulmonary resuscitation, public location and site location (y-axis). Probability of survival versus average chest compression rate when other covariates are equal to the population average. A histogram of the compression rates and number of patients is included. Dashed lines show 95% CIs (from Idris et al 16). Reproduced with permission from Wolters Kluwer Health.

Chest compression depth (AHA recommendation: 5–6 cm)

Like chest compression rate, compression depth is also associated with clinical outcomes. In a large observational study of 9136 patients with OHCA from ROC, increasing chest compression depth was associated with improved rates of ROSC and survival to hospital discharge. The adjusted OR (95% CI) for survival was 1.04 (1.00 to 1.08) per 5 mm increase in chest compression depth and 1.45 (1.20 to 1.76) for cases that were >38 mm more than 60% of time, as compared with those outside this range. There also appeared to be a plateau association between compression depth and survival, with best outcomes seen at depths between 40 and 54 mm (figure 5), and support current guideline recommendations.17

Figure 5

Covariate-adjusted survival to discharge by compression depth, with 95% CIs. Red vertical dashed line represents highest survival, and grey vertical dashed lines represent 15 mm interval with highest survival (from Stiell et al 17). Reproduced with permission from Wolters Kluwer Health.

Chest compression fraction and minimising pauses (AHA recommendation: chest compression fraction ≥60%, minimise pauses in chest compressions)

Interruptions in chest compressions have been shown to have significant adverse effects on haemodynamics during resuscitation. After pausing chest compressions, CPP quickly decreases to zero and can take upwards of 15 or more compressions to achieve the same magnitude to that present before the pause.18 19 The duration of pauses in CPR can also influence outcome. This was demonstrated in a large observational study of 2006 patients with OHCA due to a shockable arrhythmia in whom the adjusted odds of survival were 7% lower for every 5 s increase in a preshock pause and 6% lower for every 5 s increase in perishock (preshock and postshock) pause.20 Another study found the durations of the longest perishock pause, longest non-shock pause and longest pause for any reason were all associated with decreased survival to hospital discharge, therefore demonstrating that pauses at any point in resuscitation can be detrimental (figure 6).21 Together, these studies indicate that decreasing both the frequency and duration of interruptions in CPR can improve outcomes.

Figure 6

Survival to hospital discharge and duration of longest overall pause in chest compressions (black), longest perishock pause (grey) and longest non-shock pause (white). Numbers in bars represent the number of patients. P<0.01 for each pause category (from Brouwer et al 21). Reproduced with permission from Wolters Kluwer Health.

Chest compression fraction represents the percentage of time that chest compressions are performed in a pulseless patient with cardiac arrest. Either frequent and/or longer pauses in chest compressions manifest as a lower chest compression fraction. In a prospective study of 506 patients with OHCA arrest due to a shockable arrhythmia, a higher chest compression fraction prior to the first defibrillation attempt was associated with improved survival to hospital discharge. Patients with a chest compression fraction of >60% had the highest survival, with an adjusted OR for survival to hospital discharge of 1.11 (1.01 to 1.21) per 10% increase in chest compression fraction (figure 7).22 Chest compression fraction is also a complex measure because of its potential interaction with other CPR metrics such that its interpretation in isolation from these other factors can prove to be misleading. For example, a high or low chest compression fraction does not necessarily trump the other qualities of CPR that are essential to good outcome.23 This is why the AHA recommends targeting a chest compression fraction of greater than 60% in the context of the totality of CPR metrics.24

Figure 7

Smoothing spline representing the incremental probability of survival corresponding to a linear increase in chest compression fraction (CCF) (from Christenson et al 22). Reproduced with permission from Wolters Kluwer Health.

Chest recoil (AHA recommendation: allow full chest recoil)

Permitting complete chest recoil during the decompression phase of CPR is essential for refilling the chest and for adequate myocardial perfusion. Incomplete chest recoil or leaving a residual pressure on the chest (‘leaning’) results in an increase in intrathoracic pressure when it needs to be at its minimum. Among a group of animals that received standard chest compressions with full chest recoil, followed by chest compressions permitting only 75% of full chest recoil, CPP decreased by more than a third during the period of incomplete chest recoil and cerebral perfusion pressures by more than 50% (both P<0.05).25

Inattention to full recoil is an all too common problem during resuscitation. The problem often arises as a result of fatigue, whereby the chest compressor (who is usually leaning over the patient while performing compressions) unintentionally uses the patient’s chest as a resting table during the decompression phase of CPR, rather than permit its full recoil. An observational study of 108 adults with inhospital cardiac arrest at the University of Pennsylvania used force-sensing devices during resuscitation to detect 5 pounds of residual pressure (‘lean’) left on the chest during the decompression phase of CPR. By this definition, leaning was observed in 91% of the resuscitations, underscoring its all-too-common occurrence.26

Controlled ventilation, avoiding hyperventilation (30:2 chest compression:ventilation ratio prior to advanced airway; 10 breaths/min with advanced airway)

Prior to the placement of an advanced airway, the provision of rescue breaths is associated with a trade-off between cardiac output and blood oxygenation due to interruptions in chest compressions.19 This is why interruptions for ventilation during 30:2 (30 chest compressions followed by 2 breaths) CPR need to be brief—at most 2 s in duration. Care needs to be taken to avoid hyperventilation particularly after placement of an advanced airway, when the impression that patients in arrest need more ventilation frequently results in hyperventilation, a potentially lethal error.27 Each positive pressure breath, while providing needed oxygen and clearance of carbon dioxide, also increases intrathoracic pressure and compromises chest refilling. Excessive ventilation can also alter pH balance, creating alkalosis and cerebral vasoconstriction. The impact these effects might have on outcome was studied in an experimental model in which a group of intubated animals undergoing CPR for VF were randomised to a ventilation rate of 12 vs 30 breaths/min (supplemented with carbon dioxide in a subgroup to maintain pH balance). Expectedly, the mean intrathoracic pressure was significantly lower among the animals ventilated at a rate of 12 vs 30/min (7.1 vs 17.5 mm Hg, respectively; P<0.001). This higher intrathoracic pressure resulted in an increase in right atrial diastolic pressure and 28% decrease in CPP in the hyperventilated animals (P<0.05). Notably, survival was significantly lower in the hyperventilated animals (with or without receipt of supplemental carbon dioxide).27 The AHA cautions that the lower cardiac output during CPR requires less ventilation than normal—and that 10 breaths per minute, each sufficient to make the chest rise, are adequate. This approach offers the best balance between meeting the patient’s ventilation needs without compromising their haemodynamic demands.

Improved CPR quality and outcomes

The AHA’s guidelines for CPR focus on optimising the key performance metrics discussed above. Large, observational studies have shown that the incorporation of these recommendations can lead to improved outcomes for patients with cardiac arrest. In 2005, the AHA recommended several changes to resuscitation protocols to increase the proportion of time devoted to chest compressions. In advance of these changes, resuscitation protocols for King County, Washington, were instituted that closely corresponded with these new AHA guidelines. In two subsequent observational studies from King County, significant improvements in survival associated with this change in protocol were seen among patients with cardiac arrest due to either shockable or non-shockable rhythms. For example, among 509 patients with OHCA due to bystander-witnessed VF, significant reductions in postshock pauses in CPR, along with an increase in chest compression fraction, were observed between the time periods before and after the change in CPR protocol, resulting in significantly improved survival (adjusted OR (95% CI) 1.75 (1.14 to 2.69)).24 Similarly, among 3960 patients with OHCA due to a non-shockable rhythm, survival to hospital discharge increased from 4.6% to 6.8% (P=0.004) and 1-year survival from 2.7% to 4.9% (P=0.001) (figure 8), in association with similar improvements in ‘hands-on’ time during the intervention period.28 Such findings indicate that ‘CPR works’ but that ‘better CPR works even better’.

Figure 8

Outcomes according to calendar year. Shown are the frequency of return of spontaneous circulation (ROSC), survival to hospital discharge and 1-year survival during each year during the control and intervention periods, with 95% CIs. No temporal trends were seen within each study period but rather improved outcomes in the intervention period as compared with the control period (from Kudenchuk et al 28). Reproduced with permission from Wolters Kluwer Health.

Implications on life-saving

Cardiac arrest remains a significant public health issue. Multiple clinical trials of advanced interventions for cardiac arrest have been attempted with variable success. It has become clear that high-quality CPR with effective chest compressions is the foundation of all successful resuscitation efforts. While survival after cardiac arrest in many communities remains poor, this problem is clearly remediable. Improvements in CPR practice can lead to substantial improvement in outcome.24 28 By incorporating clinical guidelines into practice, thousands of lives can be saved each year with little more than correctly applying human hands. 



  • AWH and PJK contributed equally.

  • Contributors The authors contributed equally to the writing of this manuscript.

  • Competing interests None declared.

  • Provenance and peer review Not commissioned; externally peer reviewed.