Quality of Cardiopulmonary Resuscitation Among Highly Trained Staff in an Emergency Department Setting FREE

Sudden cardiac death is a major clinical and public health problem,¹ and survival rates remain poor.² Although, to our knowledge, there have been no randomized controlled trials, observational studies in humans^3- 4 and experimental models^5- 6 indicating the importance of high-quality cardiopulmonary resuscitation (CPR) to improve patient outcomes. Avoidance of hyperventilation and interruption of chest compressions have been identified as the hallmarks of successful CPR and beneficial survival.^5- 9 However, despite international guidelines¹⁰ and intensive training programs, the quality of CPR seems to remain poor and correlates significantly with poor postresuscitation survival rates as reported in studies evaluating bystander CPR.^11- 14

Nevertheless, bystanders perform basic life support rarely and often insufficiently,^15- 16 and recent studies^17- 18 have demonstrated that the CPR provided by health care professionals outside the hospital is suboptimal; eg, paramedics provide chest compressions only half of the time during their resuscitation efforts. Data available from in-hospital cardiac arrests show hands-off times to be about 24%, with chest compression rates that are too low and ventilation rates that are too high.^3,19 Thus, improvements seem to be needed even for health care professionals.

Inadequate data were available concerning CPR quality in an optimal setting with sufficient, highly qualified medical staff and monitoring capabilities. Therefore, we assessed CPR quality markers and identified what quality levels can be achieved during the daily clinical routine in the emergency department of a tertiary care university hospital.

This prospective, observational survey was performed from June 1, 2002, to August 30, 2005. All patients with nontraumatic cardiac arrest were eligible for study participation within this time frame. All study-related procedures were carried out in accordance with the Helsinki Declaration, and ethics approval was obtained from the local review board.

The study was performed at the emergency department of a tertiary care university hospital with an annual census of 75 000 patients. Within this 2000-bed hospital, the emergency department is an academically and administratively independent unit that provides care for life- and non–life-threatening emergencies of all medical specialties except trauma. Patients with unstable life-threatening diseases are transported immediately to the acute care unit treatment areas, which are fully equipped with monitors, ventilators, and complete nursing facilities. Capabilities of the acute care unit include installation of extracorporeal devices (eg, for cardiopulmonary emergency bypass and hemodialysis) and performance of diagnostic procedures, including portable radiography, fluoroscopy, sonography, echocardiography, pulmonary artery catheterization, electroencephalography, measurement of evoked potentials, intracranial pressure monitoring, and endoscopy. Intensive care medical support, including controlled mechanical ventilation and/or advanced cardiac life support, was provided in accordance with a standard protocol.^20- 22 Thirty physicians and 60 nurses are employed full-time in the emergency department, and at least 4 physicians and 6 nurses are on duty around the clock throughout the week. The physicians are critical care–trained internists, emergency-trained practitioners, or fellows of all medical subspecialties excluding traumatology with more than 10 years of acute care clinical experience. Training and education to achieve this level of high-quality CPR performance followed the established European Resuscitation Council course in basic and advanced life support.²² Most of our team members are instructors in such courses several times a year and participate on a yearly basis in the educator master classes. As a part of routine care, patients in cardiac arrest are monitored via 3-lead electrocardiography (ECG), pulse oximetry, and invasive arterial pressure as early as possible. The airway is secured with an endotracheal tube, and patients are routinely ventilated via a valve-bag system, but sometimes also with a software-supported monitoring system (Servo I ventilator system, version 1.2; Siemens Medical Group, Frankfurt, Germany).

Admission diagnosis and known medical history were routinely assessed, and the data for patients in cardiac arrest encompassed all information required for the international Utstein templates for resuscitation reporting.²³

A specially designed biphasic defibrillator (HeartStart 4000 SP; Laerdal Medical, Stavanger, Norway) was used in manual mode to monitor the quality of artificial ventilation and external chest compressions. This device is certified by the Conformité Européene and has the capability to measure the rate and depth of chest compressions. Ventilation and pulse data were obtained using impedance change measurements captured from the self-adhesive defibrillation pads. External chest compressions were assessed via an additional pad placed on the middle of the sternum of the patient, incorporating an accelerometer sensor (ADXL202e; Analog Devices, Norwood, Mass) and a pressure sensor (22PCCFBG6; Honeywell, Morristown, NJ) that were capable of recording compression depth data.^24- 25 The precision of the compression depth recording may have been attenuated slightly because no backboards were used during the compressions. Each resuscitation episode was recorded by 2 data cards from the Personal Computer Memory Card International Association: One standard card collected ECG signals, time, and shock events, and the other recorded the output signals from the extra chest pad and the thoracic impedance between the defibrillation pads.

Raw data were merged and formatted for final analysis by a blinded investigator. Only cases with at least 1 minute of data collection during CPR were eligible for analysis. An ad hoc–designed software tool (Sister Studio, version 1.4.16; Laerdal Medical) was used to assess the quality of CPR. The ECG was manually annotated by a physician as asystole, pulseless electric activity, ventricular fibrillation, ventricular tachycardia, or rhythm with pulse. All ventilations were scrutinized, and the automatic annotations were corrected if the physician disagreed with the detection decision of the equipment. Episodes were excluded if the ventilation signals were too noisy. Based on these annotations, the software tool was set to report quality markers of CPR for each 30-second interval and for the whole episode. Quality markers included the hands-off time and ratio, compression rate, actual number of compressions performed per minute, compression depth, compression duty cycle, incomplete hand release, and ventilations per minute.

We evaluated the quality of CPR and adherence to the international guidelines applicable within the study period (compression rate, 100 per minute; compression depth, 40-50 mm; ventilation rate, 12 per minute).¹⁰ The primary outcome was the hands-off ratio, defined as the hands-off time divided by the cardiac arrest time. The hands-off time was defined as the total recorded time minus the time with chest compressions or with spontaneous circulation (period of cardiac arrest without compressions being performed). Therefore, the hands-off ratio represents the fraction of time during the resuscitation episode without cerebral or myocardial circulation.

The secondary outcomes included the ventilation rate, chest compression rate, actual chest compressions performed per minute, depth of compressions, compression duty cycle, and compression with incomplete release. The average number of shocks per patient and the hands-off time before and after each shock were also investigated.

The data are presented as number and percentage of patients, as median and 25% to 75% interquartile range (IQR), or as mean and 95% confidence interval (CI) as appropriate. The time-dependent variables were log-transformed and a geometrical mean was calculated. Patients were grouped according to their duration of cardiac arrest as 1 to less than 5, 5 to 15, and more than 15 minutes to yield approximately balanced groups. We used the χ² test for trend to assess whether increasing cardiac arrest time is linearly associated with the hands-off ratio. We further used linear regression to quantify the relation of the cardiac arrest time with the hands-off ratio. The data were collected and processed using commercially available software (Excel 2002 [Microsoft Corp, Redmond, Wash] and Stata, version 8.0 [StataCorp, College Station, Tex]). A 2-sided P<.05 was considered statistically significant.

During the study period of 39 months, 213 patients requiring CPR and eligible for study participation were admitted to our emergency department. Of those patients, 80 were entered into the study. We excluded 133 patients for the following reasons: nonimmediate availability of the device, inappropriate usage of the monitoring capabilities of the defibrillator by not using the chest pad for monitoring compression rate and depth in the heat of the moment, and period of data records of less than 1 minute. Ninety-five data sets from the 80 patients were available for analysis, yielding a total of 1065 minutes of cardiac arrest time (mean, 11.2 minutes; 95% CI, 9.4-13.0 minutes; range, 1.0-39.7 minutes) with chest compression for analysis. For assessment of ventilation rate, 69 patients with 84 episodes were eligible for analysis; of these, 11 episodes could not be analyzed owing to poor signal quality. Basic patient demographic and cardiac arrest data are presented in Table 1.

A summary of measured outcome variables is shown in Table 2. Chest compressions were performed according to guidelines most of the time, with a mean rate of 114 per minute and a mean actual delivered rate of 96 per minute. The mean hands-off time was calculated to be 7.6 (95% CI, 7.2-8.0) seconds per minute. The mean hands-off ratio was 12.7% (95% CI, 12.3%-13.1%). The longest period of hands-off time was 2.7 minutes. The hands-off ratio was linearly associated with the duration of CPR (R² = 0.95; mean, 4.3% increments per 5-10 minutes; P<.001) (Figure and Table 3).

Patients were hyperventilated with a median rate of 18 (IQR, 14-24) ventilations per minute. Correct ventilations with the recommended rate of 12 to 15 per minute were performed 18.0% of the time (192 of 1065 minutes) during CPR.

We subjected the 49 episodes of defibrillation in 16 patients to further detailed analysis. A mean number of 3 shocks were delivered (range, 1-8 shocks). The median hands-off time before each shock was 10.9 (IQR, 9.3-12.8) seconds; after each shock, 6.2 (IQR, 5.2-7.2) seconds. The median hands-off time between consecutively delivered shocks was 12.9 (IQR, 11.4-14.5) seconds. No shock was delivered inappropriately, but 4 episodes of ventricular fibrillation occurring during chest compressions went unrecognized for prolonged periods (2, 3, 4, and 7 minutes).

Although the participants in our study did not meet all of the recommendations of the CPR guidelines, overall performance quality was higher than that reported by previous investigators.^3,18 The hands-off ratio was minimal, and chest compressions were performed at a satisfactory mean rate. Taking the hands-off time into account, participants delivered compressions per minute according to the guidelines recommended.¹⁰ However, the ventilation rate was above the recommended range.

To our knowledge, this is the first report of a prospective observational study evaluating the quality of CPR in the optimal setting of an emergency department with highly trained health care providers and sufficient manpower and equipment that therefore could show the limits that could be achieved for CPR quality.^20- 22 Training and education to achieve this level of high-quality CPR performance followed the established European Resuscitation Council course in basic and advanced life support.²² Most of our team members are instructors in such courses several times a year and participate on a yearly basis in the educator master classes. This high-level, enthusiastic activity of our team in the well-accepted training courses guarantees continuous quality improvement in advanced cardiac life support and postresuscitation care. Even so, the performance of our emergency team is not excellent, because in 4 episodes of ventricular fibrillation, recognition was delayed after prolonged chest compressions, and correct ventilations with the recommended rate were performed only 18.0% of the time during CPR. Therefore, there is still room for improvement in the quality of CPR.

The hands-off time ratio is one of the determinating factors of outcome after CPR.⁸ A hands-off ratio of 48% to 57% was observed in previous out-of-hospital studies of cardiac arrest.^17- 18 In these cases, performance of chest compressions was often interrupted by defibrillation with the automatic external defibrillator (AED); thus, substantial time was also needed to analyze, charge, and shock the patient; to reanalyze the situation and reassess the patient for pulse or rhythm changes; and possibly for intubation or placement of intravenous lines. The implications of the potential detrimental effect of delays caused by the AED are important and should not be dismissed. A randomized trial performed in a tertiary care center comparing AEDs with manual defibrillators with a focus on CPR performance measures could give answers to this important issue.

In our present study, we observed a mean hands-off ratio of 12.7%, corresponding to a mean hands-off time of 7.6 seconds per minute. These findings are clearly better than those reported in the recent literature.^3,18- 19 One reason for this significant difference in the hands-off ratio might be that we use AEDs only in the manual mode rather than in the automatic mode; the latter may delay rhythm diagnosis and defibrillation. Eilevstjonn et al²⁶ reported median hands-off times of about 25 seconds before, during, and after defibrillation with an AED. In contrast, our corresponding mean hands-off times were only 11, 13, and 6 seconds, respectively. After the time for charging is taken into account—about 2 to 4 seconds—the hands-off time arising from defibrillation is quite short.

Abella et al¹⁹ found a mean no-flow ratio of 24% for in-hospital cardiac arrest. This observed difference from our study may be explained by different settings. Abella et al evaluated the quality of CPR performed by cardiac arrest response teams outside the intensive care unit, whereas our study was performed in a specialized, high-volume emergency department of a tertiary care university teaching hospital. As already mentioned, CPR is frequently performed in our department and we have enough manpower and monitoring capabilities at all times. However, we observed a significant time-dependent increase of the hands-off ratio during CPR. This fact might be explained by interruptions for concurrent procedures such as transthoracic or transesophageal echocardiography, insertion of arterial lines, repeated and sometimes too-frequent ECG rhythm and pulse checks during prolonged CPR, exhaustion of medical personnel, perceived futility of progressively prolonged but ineffective CPR, or, indeed, a causal relationship between more hands-off time and unfavorable outcomes.

Interruption of chest compressions for more than 15 seconds critically reduces the coronary perfusion pressure, compromises the outcome of CPR, and increases the severity of postresuscitation myocardial dysfunction.^6,9,25 Also, the chest compression rate has been shown to correlate significantly with the return of spontaneous circulation. Abella et al¹⁹ found that the mean compression rate for initial survivors was 90 per minute, whereas nonsurvivors had a mean rate of 79 per minute.³ We found a mean chest compression rate of 114 per minute, resulting in a mean of 96 actual performed compressions per minute, meeting the guidelines. In out-of-hospital cardiac arrest, the mean rate was 121 compressions per minute, but taking the hands-off time into account, a mean of 64 actual compressions were performed per minute.¹⁸ In our data, compression rate was rarely too slow (<90 per minute in 0%), but despite our mean of 96 compressions achieved per minute (allowing for interruptions between sequences), 16 (17%) of 95 episodes achieved fewer than 80 compressions per minute, and 11 (12%) achieved fewer than 70 per minute. This is in contrast to previously reported data of in-hospital cardiac arrest reporting the chest compression rate but not the actual number of compressions performed, where rates were within 100 per minute for only 30% of the time and were less than 80 per minute 13% to 37% of the time.^3,19

Arterial and coronary blood flow increase with increasing compression force and depth.^27- 29 Chest compression depths have been reported to be too shallow: for out-of-hospital cardiac arrest, the mean compression depth was 34 mm, with only 28% of compressions having a depth within guidelines recommendations.¹⁷ For in-hospital cardiac arrest, Abella et al¹⁹ found that 37% of compressions were too shallow. In contrast, our mean compression depth was 63 mm, and only 7% of the measured compressions were less than 40 mm. However, the actual depth may have been overestimated in our study because we did not use a backboard, thereby altering the precision of the recording device.

Another important issue in CPR concerns ventilation, with even experienced paramedics tending to hyperventilate their patients up to a maximum of 37 ventilations per minute. An excessive ventilation rate during CPR will result in increased positive intrathoracic pressures, decreased coronary perfusion, and decreased survival rates.⁷ Similar to previous in-hospital data,¹⁹ our median ventilation rate was 18 per minute. Despite our patients undergoing CPR in a well-controlled setting, the recommended rate of 12 to 15 ventilations per minute was achieved in only 18.0% of CPR efforts. This difference compared with the data of Wik et al,¹⁸ who reported a mean rate of 11 ventilations per minute in ambulance services, might be owing to the fact that some of our patients underwent inappropriate mechanical ventilation (ie, the pressure limits and trigger were not adjusted). Nevertheless, even if we performed manual ventilation on our patients, they were hyperventilated.

There are several limitations to our study. The intent of this study was to objectively describe multiple variables as surrogate markers of CPR quality during cardiac arrest. It was not designed or powered to find CPR quality differences between survivors and nonsurvivors. Therefore, we are not able to report an association of survival and CPR quality. In addition, extensive confounding by patient management at the out-of-hospital scene, emergency department, and intensive care unit would be present, because we did not know the preceding quality of CPR provided by bystanders or ambulance services. Also, some cardiac arrests were not witnessed and/or received no bystander CPR. Only 13 patients (16%) experienced cardiac arrest after admission to our department, whereas the other patients were admitted during ongoing CPR or had a rearrest in our department. Among the weaknesses, we should note the relatively small sample size (80 patients) and the risk of selection bias. However, selection bias appears to be a minor factor in this study. Some recorded data were excluded from the analysis because of human and technical failures. In addition, it was not possible to blind operators to the use of monitoring devices, and the knowledge of being under investigation might have biased the physicians and nurses toward more focused attention on CPR than usual. Nevertheless, we have shown that, under these conditions, focus can be well maintained for a considerable period.

Compared with previous reports, we found that CPR can be satisfactorily but still not perfectly performed by highly trained professionals in the specialized health care environment of an emergency department. A correct chest compression rate resulting in a mean of 96 compressions per minute and a hands-off ratio of 12.7% could be achieved. However, a deterioration in performance with resuscitation duration was observed. Avoidance of hyperventilation and unnecessary no-flow time by reducing ECG rhythm, pulse checks, and examination times to the minimum should be emphasized and provide room for improvement in the quality of CPR.

Correspondence: Fritz Sterz, MD, Department of Emergency Medicine, Medical University of Vienna, Währingergürtel 18-20/6D, 1090 Vienna, Austria (fritz.sterz@meduniwien.ac.at).

Accepted for Publication: August 24, 2006.

Author Contributions:Study concept and design: Losert and Sterz. Acquisition of data: Losert, Sterz, Köhler, Fleischhackl, Eisenburger, Kliegel, and Myklebust. Analysis and interpretation of data: Losert, Sterz, Sodeck, Fleischhackl, Kliegel, Herkner, Nysæther, and Laggner. Drafting of the manuscript: Losert and Sterz. Critical revision of the manuscript for important intellectual content: Losert, Sterz, Köhler, Sodeck, Fleischhackl, Eisenburger, Kliegel, Herkner, Myklebust, Nysæther, and Laggner. Statistical analysis: Losert, Sodeck, Kliegel, and Herkner. Obtained funding: Sterz. Administrative, technical, and material support: Losert, Sterz, Köhler, Myklebust, Nysæther, and Laggner. Study supervision: Sterz.

Financial Disclosure: Dr Köhler was employed for 12 months at the Department of Emergency Medicine, Medical University of Vienna, with the support of a grant from Laerdal Medical. Laerdal Medical provided travel grants for scientific meetings for Drs Losert and Köhler. Dr Losert received a laptop computer from Laerdal Medical.

Funding/Support: This study was supported in part by Laerdal Medical.

Disclaimer: Nonemployees of Laerdal Medical had unrestricted editing rights, so that the manuscript was as free from corporate bias as possible. The sponsor could not have suppressed publication if the results were negative or detrimental to their product.