|Home | About | Journals | Submit | Contact Us | Français|
Quality CPR contributes to cardiac arrest survival. The proportion of time in which chest compressions are performed in each minute of CPR is an important modifiable aspect of quality CPR. We sought to estimate the effect of an increasing proportion of time spent performing chest compressions during cardiac arrest on survival to hospital discharge in patients with out-of hospital ventricular fibrillation or pulseless ventricular tachycardia.
This is a prospective observational cohort study of adult patients from the Resuscitation Outcomes Consortium Cardiac Arrest Epistry with confirmed ventricular fibrillation or ventricular tachycardia, no defibrillation prior to emergency medical services arrival, electronically recorded cardiopulmonary resuscitation prior to the first shock and a confirmed outcome. Patients were followed to discharge from hospital or death. In the 506 cases, the mean age was 64 years, 80% were male, 71% were witnessed by a bystander, 51% received bystander cardiopulmonary resuscitation, 34% occurred in a public location, and 23% survived. After adjustment for age, gender, location, bystander cardiopulmonary resuscitation, bystander witness status, and response time the odds ratios of surviving to hospital discharge in the two highest categories of chest compression fraction compared to the reference category were 3.01 (95% CI, 1.37, 6.58) and 2.33 (95% CI, 0.96, 5.63). The estimated adjusted linear effect on odds ratio of survival for a 10% change in chest compression fraction was 1.11 (95% CI, 1.01, 1.21).
Increased chest compression fraction is independently predictive of better survival in patients suffering a prehospital ventricular fibrillation/tachycardia cardiac arrest.
Out-of-hospital cardiac arrest is a leading cause of premature death throughout the world. 1 Survival from out-of-hospital cardiac arrest is variable and often less than 5% .2 3 Survival depends on effective cardiopulmonary resuscitation and early defibrillation. 4 To improve survival, it is important to understand and then optimize modifiable predictors of outcome.
The quality of cardiopulmonary resuscitation is likely an important contributor to successful outcome. One of the most important aspects of quality cardiopulmonary resuscitation is thought to be the proportion of time spent performing chest compressions. Interruptions in chest compressions are common during treatment of cardiac arrest. 5–7 Animal studies demonstrate that interruptions in chest compressions decrease coronary and cerebral blood flow, resulting in worse survival outcomes. 8, 9 Emergency medical service providers typically perform chest compressions only 50% of the time during their resuscitative efforts. 7 The clinical consequences of interruptions in chest compressions on cardiac arrest survival have yet to be determined. Based on these clinical and laboratory observations and the rationale that minimizing blood flow disruptions during cardiopulmonary resuscitation should improve survival, the 2005 American Heart Association and the European Resuscitation Council Guidelines for Cardiopulmonary Resuscitation recommended increasing the proportion of time delivering chest compressions. 10, 11
The objective of this multicenter cohort study was to estimate the independent effect of chest compression fraction (proportion of time delivering chest compressions during cardiopulmonary resuscitation) on survival to hospital discharge in a cohort of patients with out-of-hospital ventricular fibrillation or pulseless ventricular tachycardia.
The Resuscitation Outcomes Consortium consists of 11 geographically distinct regional clinical centers across North America created to study promising out-of-hospital therapies for cardiac arrest and significant traumatic injury. 12 The 11 regional centers are Ottawa, Toronto and British Columbia in Canada and Iowa, Pittsburgh, Dallas, Milwaukee, Alabama, Seattle/King County, Portland, and San Diego in the United States, and include over 260 separate emergency medical service agencies. Since December 2005, the Resuscitation Outcomes Consortium Cardiac Arrest Epistry, 13 has prospectively gathered data on out-of-hospital cardiac arrest cases attended by a participating emergency medical services agency. Pre-specified data related to out-of-hospital treatments and outcomes were collected using standardized operational definitions, including initial cardiac rhythms, response times, descriptions of professional responders, timing of cardiopulmonary resuscitation and defibrillation, response to interventions, return of spontaneous circulation, and survival to hospital discharge. The Resuscitation Outcomes Consortium encouraged collection of digital, electronic recordings of rhythm and chest compressions. All data were managed by a central data coordinating center. Seven sites and 78 agencies contributed cases for this report.
Between December 2005 and March 2007, all patients who suffered a cardiac arrest prior to emergency medical services arrival with a first recorded rhythm of ventricular fibrillation/tachycardia and who were not enrolled in a concurrent clinical trial were eligible for this study. The initial rhythm was determined to be ventricular fibrillation/tachycardia if the initial automated external defibrillator analysis recommended a shock or if the emergency medical service provider interpreted the initial rhythm as ventricular fibrillation/tachycardia. Rhythm diagnosis was confirmed as ventricular fibrillation/tachycardia by research staff. We excluded patients receiving public access defibrillation prior to emergency medical services arrival, patients without at least one minute of digitally-recorded cardiopulmonary resuscitation before or during the minute of the first shock, and patients in whom the outcome was unknown.
The presence and frequency of chest compressions were measured indirectly either by changes in thoracic impedance recorded from external defibrillation electrodes as described by Valenzuela 6 or via an accelerometer interface between the rescuer and the patient’s chest using commercially available defibrillators. The chest compression fraction was defined as the proportion of resuscitation time without spontaneous circulation during which chest compressions were administered. This was calculated by automated external defibrillator analytic software which permitted identification of all interruptions greater than two seconds (Phillips, ZOLL devices) or three seconds (Medtronic devices). These pauses were defined as time without chest compressions. Tracings were acquired and downloaded from Medtronic defibrillators (482), Zoll defibrillators (18) and Phillips defibrillators (6). Each case included the minute interval during which the first analysis was performed (including some time before and after the first shock) and all recorded minute intervals prior to the first analysis. The chest compression fraction values for all minute intervals were averaged for each patient. Trained research staff reviewed the automated calculation of chest compression fraction at each site prior to entering chest compression fraction values. The prospectively selected primary outcome measure was survival to hospital discharge.
All statistical analyses were performed with a commercially available statistical package (SAS, version 9.1.3, Cary, NC; R, version 2.5.1, Vienna, Austria). Summary results are presented as mean (±SD) or median (IQR). We categorized chest compression fraction (from 0 to 100%) into five groups based on the average chest compression fraction delivered to the patient over all minutes with available data: 0–20%, 21–40%, 41–60%, 61–80%, and 81–100%. These groups corresponded to receiving cardiopulmonary resuscitation, on average, for 0–12, 13–24, 25–36, 37–48, and 49–60 seconds per minute, respectively, over all analyzed minutes of data. Potential confounding variables identified a priori included: age, gender, location of cardiac arrest (public place or private residence), bystander cardiopulmonary resuscitation, bystander witnessed cardiac arrest and the time interval from receipt of the emergency call to emergency medical services arrival at the scene. We calculated descriptive and bivariate statistics and used logistic regression to estimate the unadjusted and adjusted odds ratio of survival for each category of chest compression fraction relative to the lowest category (0–20%). The adjusted model was repeated including Resuscitation Outcomes Consortium site as a covariate to determine whether other unknown local influences affected the relationship. A secondary multivariable linear-regression analysis estimated the effect of a 10% change in chest compression fraction.
As an exploratory analysis, we fit a penalized cubic smoothing spline curve to further characterize the nature of the relationship between chest compression fraction and survival. 14
A total of 14,090 cardiac arrest cases occurred prior to emergency medical services arrival and were treated by emergency medical service responders. Of these, 3170 patients had an initial rhythm of ventricular fibrillation/tachycardia, and 506 were eligible for analysis. The remaining 2664 patients were excluded primarily because a shock was delivered prior to initiation of cardiopulmonary resuscitation (1114) or chest compression fraction data were not obtained (1550). (Figure 1) Two sites with pre-existing capacity to download these electronic data contributed 79% of the eligible cases. The number of cases contributed by site is shown in Table 1.
Clinical characteristics and outcome of the excluded patients with an initial rhythm of ventricular fibrillation/tachycardia but less than 1 minute of recorded compression data before the first shock are compared in Table 1. Analyzed patients had a higher proportion of males, more arrests in public places, more frequent bystander CPR and a higher proportion with return of spontaneous circulation. Overall 117 subjects (23%) in the cohort survived to hospital discharge. The mean age of patients was 64 ± 15 years old, 80% were male, 34% arrested in a public location, 71% were witnessed by bystanders, and 51% received bystander cardiopulmonary resuscitation.
The demographics by chest compression fraction category are shown in Table 2. The percentage of patients with return of spontaneous circulation was 58%, 73%, 76%, 73% and 79% respectively in the five categories of increasing chest compression fraction. Survival to hospital discharge in these five categories of increasing chest compression fraction was 12.0%, 22.9%, 24.8%, 28.7% and 25.0%.
We did not identify any imbalances in demographic or arrest characteristics in patients among the categories. The association between chest compression fraction categories and the probability of survival is shown in Figure 2. Unadjusted and adjusted odds ratios of survival for the pre-selected factors potentially associated with survival are shown in Table 3. The effect of an increasing chest compression fraction on survival remained significant after adjustment for possible and known determinants of survival including age, gender, bystander cardiopulmonary resuscitation, bystander witnessed arrest, emergency medical services response time and location. The estimated adjusted linear effect on odds ratio of survival for a 10% change in chest compression fraction was 1.11 (95% CI, 1.01, 1.21).
When the model was extended to include site, the recalculated odds ratios for survival to hospital discharge were 2.71 (95% CI, 1.18, 6.26) for the 61–80% category and 2.02 (95% CI, 0.78, 5.20) for the 81–100% category. The odds ratio for survival for an average 10% increase in chest compression fraction was 1.08 (95% CI, 0.98, 1.20).
Since two sites contributed such a large percentage of cases, we ran a post hoc analysis using only cases from those two sites. The Odds Ratio (95% CI) of survival to discharge in the 5 ascending categories of CCF were very similar to the whole cohort and the conclusions are unchanged.
A spline smoother was fit to visually explore the relationship between chest compression fraction and survival changes over the range of chest compression fraction. (Figure 3)
The results of this large, observational multi-center study demonstrate an association between the proportion of resuscitation time that chest compressions are performed prior to the first defibrillation and survival to hospital discharge after out-of-hospital cardiac arrest due to ventricular fibrillation/tachycardia. The relationship of chest compression fraction and survival was independent of other known predictors. This observation is important and provides a rationale for relatively simple changes to resuscitation training and practice that if implemented is likely to improve survival.
These clinical findings strongly support animal study observations that minimizing coronary and cerebral blood flow disruptions during resuscitation improves survival from cardiac arrest. 8, 15–20 Our findings build on previous but smaller clinical studies. Edelson et al observed that an increased pause in compressions just prior to the first shock was associated with a lower rate of successful conversion of ventricular fibrillation among 60 hospital and prehospital patients. 21 Ko et al reviewed electronic ECG tracings in 52 cases of out-of-hospital witnessed ventricular fibrillation and demonstrated a positive correlation between the quality of cardiopulmonary resuscitation and survival. In his study, the number of chest compressions delivered was one component of quality cardiopulmonary resuscitation. 22 Eftestol et al observed that increasing hands off time (the reciprocal of chest compression fraction) just prior to out-of-hospital defibrillation in 156 cases, correlated with a lower rate of return of spontaneous circulation. 23 Our results also support the large before and after study by Bobrow et al where they demonstrated an increase in survival in an EMS system after training to a protocol of 200 uninterrupted compressions before and after the first analysis. 24
The current study is the largest clinical investigation evaluating the independent association between chest compression fraction in the minutes prior to the first attempted defibrillation with survival to hospital discharge.
There are several unique aspects of this Resuscitation Outcomes Consortium investigation. It includes a large cohort of cardiac arrest patients with outcome data and cardiopulmonary resuscitation process information that was collected prospectively using standardized operational definitions. The primary analysis was determined a priori without the bias of preliminary data exploration. These data represent a diverse group of emergency medical services providers within North America, including large cities, rural areas, and a multitude of systems and populations all using a common database. Since sites had varying levels of ECG upload and monitoring sophistication, the majority of cases were contributed by two sites with pre-existing ability to analyze ECG recordings. For part of the collection period, the cardiac arrest protocol at these two sites recommended that emergency medical services providers deliver continuous chest compressions with superimposed ventilations at 8–10 per minute prior to and after endotracheal intubation. The post hoc analysis of the adjusted odds ratios of survival did not differ substantively whether all sites were included or only the two most prominent sites.
These data suggest that increasing chest compression fraction is an effective approach to improve sudden cardiac arrest outcomes. This is an important finding, relatively easily implemented, and widely generalizable. The optimal level of chest compression fraction that defines a practical goal for emergency medical services training and quality improvement however, cannot be established by this study.
A curious finding in this analysis is the modest reduction in the point estimate of survival in the highest category of CCF (81–100%) compared with the next highest category (61–80%). The most likely reason is due to the small sample size and wide confidence limits. Other possible reasons include a true plateau effect of CCF above 80%, association of better performance in patients who are perceived unlikely to survive or chance inclusion of patients in this group with variables associated with poor survival that were not included in our model. It is noted that some of the characteristics in this group were unusual compared with other groups including; slightly prolonged time to first shock, higher proportion of cases with ALS on scene first, longer period of electronic tracing prior to defibrillation which may suggest inordinately long CPR prior to defibrillation, and a higher probability of epinephrine use. It is unknown but possible that one or more of these variables are associated with a reduction in survival.
The spline graph (Figure 3) depicts how a survival curve related to CCF may appear over a more complete range of values. Moving from low levels of chest compression fraction to intermediate levels provides clinically significant benefit. Above the mid range, incremental benefit continues but is less dramatic. More research is required to better define the optimum target for chest compression fraction. Nevertheless, this spline curve based on clinical data supports the evidence that increasing pre-shock coronary and cerebral blood flow can improve outcome.
These findings are especially important in clinical practice. Chest compression fraction is often poor and therefore provides significant opportunity for improvement. Improving the chest compression fraction is a matter of education and subsequent behaviour change. Utilizing new technology, such as that employed in this study, the behaviour change can be measured and appropriate feedback given to EMS personnel. It is also possible to provide direct, real-time feedback to emergency medical services personnel during actual cardiopulmonary resuscitation. The impact on quality CPR and training programs is under evaluation. We believe that the survival benefit of increasing chest compression fraction is real and that improving care by increasing chest compression fraction is a relatively easy and inexpensive intervention.
This study has several limitations. First, because this is an observational cohort study, it can only establish an association between chest compression fraction and survival rather than a causal relationship. It is possible that chest compression fraction is correlated with an unmeasured determinant of survival [e.g., rescuer commitment to the resuscitation or expectation of patient survival]. Nevertheless, we believe a causal relationship is likely. Our findings are biologically plausible, an increase in the intervention incrementally improved the clinical outcome, pre-clinical experimental trials in animals support a direct causal relationship 19, 20, 25, and no differential effect from chest compression fraction was observed across sites. A second limitation is the possibility of selection bias introduced by the exclusion of patients who were less likely to survive. It is possible that these excluded patients represent those who received poorer quality of cardiopulmonary resuscitation, including reduced chest compression fraction, though this is impossible to verify. Should this have been the case, and such patients included in the analysis, it is likely that the number of patients in the two lower categories would increase, resulting in greater precision of our effect estimate. A third limitation is that the majority of cases contributed to this study came from two sites with pre-existing ability to analyze electronic resuscitation recordings. However, inclusion of study site in the multivariate analysis did not substantially alter the relationship observed between chest compression fraction and survival. Likewise, limiting the analysis to the two sites providing the largest number of cases resulted in similar results.
Chest compression fraction appears to be an important determinant of survival from cardiac arrest but many questions remain unanswered. These include whether uninterrupted chest compressions are more important during specific periods before and after the shock than at others times, 20, 26,21 the optimal duration of pre-shock cardiopulmonary resuscitation for prolonged ventricular fibrillation, 27,28 and the relative importance of chest compression fraction immediately after the shock, which may be a critical time for stabilizing an organized electrical activity and re-establishing adequate myocardial contractility. Increasing chest compression fraction during out-of-hospital resuscitation of patients with ventricular fibrillation/tachycardia patients is an independent determinant of survival to hospital discharge. These data strongly support the contention that more time spent performing chest compressions in the early phase of resuscitation substantially impacts on survival to hospital discharge. Implementation of strategies to alter resuscitation practices to maximize chest compression fraction are likely to result in a real and sustainable increase in survival from cardiac arrest.
Sudden cardiac death is an epidemic with an extremely high mortality rate. Recent attention provides hope that better clinical management even after the heart arrests can have a very important influence on survival. Quality of CPR is important. The proportion of time in which chest compressions are performed in each minute of CPR is one key modifiable aspect of quality CPR. This observational cohort study of patients from the Resuscitation Outcomes Consortium Cardiac Arrest Epistry estimated the effect of an increasing proportion of time spent performing chest compressions during cardiac arrest on survival to hospital discharge in patients with out-of hospital ventricular fibrillation. Overall, 23% of patients survived. After adjustment for age, gender, location, bystander cardiopulmonary resuscitation, bystander witness status, and response time the odds ratios of surviving to hospital discharge in the two highest categories of chest compression fraction compared to the reference category were 3.01 (95% CI, 1.37, 6.58) and 2.33 (95% CI, 0.96, 5.63). The estimated adjusted linear effect on odds ratio of survival for a 10% change in chest compression fraction was 1.11 (95% CI, 1.01, 1.21).
This study confirms that increasing chest compression fraction (hands-on time) during out-of-hospital resuscitation of patients with ventricular fibrillation/tachycardia patients is an independent determinant of survival to hospital discharge. Devising CPR protocols that take advantage of this simple fact can save thousands of lives each year and are extremely inexpensive to implement.
To the paramedics and first responders who worked hard to resuscitate each of these patients and who submitted all the data. To the research teams at the sites who coordinated and encouraged the effort and who entered all the data. To the strengthening relationship between the two.
Sources of Funding
The ROC is supported by a series of cooperative agreements to 10 regional clinical centers and one data Coordinating Center (5U01 HL077863, HL077881, HL077871 HL077872, HL077866, HL077908, HL077867, HL077885, HL077887, HL077873, HL077865) from the National Heart, Lung and Blood Institute in partnership with the National Institute of Neurological Disorders and Stroke, U.S. Army Medical Research & Material Command, The Canadian Institutes of Health Research (CIHR) - Institute of Circulatory and Respiratory Health, Defence Research and Development Canada, the Heart and Stroke Foundation of Canada, and the American Heart Association.