Prove it: Real-time CPR feedback devices during out-of-hospital resuscitation improve ROSC rates

• Introduction
• The review
• Results
• What it means for you
• Limitations

Introduction

Engine 42 and Medic 14 respond to a report of an unconscious person. After a short response interval, they arrive to find bystanders performing CPR on a middle-aged man lying in a warehouse break room. About one minute before EMS arrival, the patient’s co-workers delivered one shock from an AED.

Firefighters take over CPR from the bystanders while one medic attaches the manual monitor/defibrillator. This new monitor is equipped with both audible and visual real-time feedback CPR cues, although no one at the station has had the opportunity to use it yet.

The device immediately begins instructing the firefighters to slow down the rate of chest compressions, push deeper and allow maximum chest recoil.

Two minutes after the last defibrillation, the patient remains in ventricular fibrillation. The medics provide a second shock, and the firefighters resume chest compressions.

During this two-minute cycle of CPR, the medics establish intraosseous access and administer 1 milligram of epinephrine. At the two-minute mark, the medics notice an organized rhythm on the monitor and confirm the presence of a pulse.

The patient remains unconscious during the post-cardiac arrest period, and the medics elect to insert an endotracheal tube. The patient’s blood pressure is 86/40 mm Hg.

The patient shows no signs of waking up, so the medics attempt to normalize the blood pressure while simultaneously instituting therapeutic hypothermia by quickly infusing two liters of chilled saline.

The on-scene 12-lead ECG reveals an evolving STEMI of the inferior left ventricular wall.

By the time the patient arrives in the ED, the cardiac catheterization team is ready. The ED physician comments on the great job that the rescue team performed. The station captain states that the CPR feedback cues from the monitor/defibrillator made all the difference in the world.

Review

Researchers in the Resuscitation Outcomes Consortium tested the hypothesis that CPR guided by a real-time feedback device would improve ROSC rates when compared to CPR without the device (Hostler et al., 2011).

They designed a prospective cluster-randomized trial using three sites in the United States and Canada. The study population included adult patients over the age of 20 who suffered an out-of-hospital cardiac arrest and received rescue shocks or chest compressions by a participating agency equipped with a monitor/defibrillator capable of providing real-time CPR feedback. The study population excluded any patient who was incarcerated by law enforcement, pregnant, had a traumatic cardiac arrest or had a “do not resuscitate” order.

The researchers measured CPR quality with the Philips MRx monitor-defibrillator equipped with Q-CPR software. This permitted measurement of the depth, rate and recoil of each chest compression. Changes in the electrical resistance across the chest cavity or waveform capnography measurements helped evaluate the assisted ventilation’s quality.

Measuring CPR quality is important in cardiac arrest research because it allows researchers to determine whether paramedics perform CPR differently for one group of patients compared to another, which could obviously influence the outcome of the study.

The feedback feature of this device provides both audible and visual cues about rate, depth and no-compression (hands-off) intervals. At the beginning of the study, a randomization procedure allowed a program administrator to lock each monitor/defibrillator as either feedback-on or feedback-off. Paramedics assigned to the ambulances knew whether their particular device was providing feedback.

Although paramedics could mute the voice prompts, they could not turn off the feedback mechanism if it was on, nor could they activate it if it was off.

Each machine remained in its assigned mode for two to seven months after being locked. After that, an administrator reprogrammed each machine to the opposite assignment. After a similar period, randomization determined a new assignment, helping to distribute any measurement errors evenly between the two groups of patients.

In addition to CPR quality data, the researchers used standardized definitions to collect patient demographic data and information about the circumstances of each cardiac arrest, the electronic defibrillator recordings, type of care provided and the outcome at hospital discharge for each patient.

The primary outcome variable for the study was ROSC in the prehospital environment. Secondary outcome variables included survival to hospital discharge, neurological status at discharge and the CPR quality variables.

The research team thought they would find a 10 percent absolute difference in ROSC rate between the intervention group and the control group and a 20 percent difference in ROSC rates at any point during the resuscitation attempt.

Based on this estimate, the researchers performed a power calculation, a mathematical determination of how many patients they would need to enroll in the study to be able to detect that difference.

Results

The ROC enrolled 1586 consecutive patients. The randomization process placed 771 patients into the feedback-on group and 815 patients into the feedback-off group.

There was no significant difference between the two groups with respect to patient demographics, arrest circumstances or EMS response intervals.

However, there were significant differences in CPR quality between the two groups (Table 1). Having the feedback mechanism on achieved the following:

• Reduced the mean number of chest compressions per minute by 4.7 percent (108.0 vs. 103.1; p < .001),
• Reduced the mean percentage of chest compressions with incomplete recoil by 3.4 percent (14.6 vs. 10.4; p < .001),
• Increased the mean CPR fraction (percentage of time that someone is actually performing chest compressions) by 1.9 percent (64.0 vs. 65.9; p = .016)
• Increased the mean chest compression depth by 1.6 percent (37.8 vs. 39.6; p = .005) when compared to having the feedback mechanism off

The mean number of ventilations per minute was not significantly different between the two groups (feedback-off 4.9 vs. feedback-on 5.2; p = .501).

Despite the fact that CPR quality improved when the feedback mechanism was on, there was no statistically significant difference in the number of patients who achieved prehospital ROSC (primary outcome variable) between the feedback-on group and the feedback-off group (361 vs. 345, respectively; p = .962).

Table 2 summarizes the outcome data. Comparing the number of patients with feedback-off vs. feedback-on revealed no statistically significant difference in the number of patients who arrived at the hospital with a pulse present (243 vs. 260, respectively; p = .713), survival for one day (213 vs. 234; p = .862), survival to hospital discharge (96 vs. 92; p = .206) or discharge from the hospital awake (78 vs. 84; p = .855).

What it means for you

Real-time CPR feedback devices help medical personnel provide individual CPR components as close to the guidelines as possible (Dine et al., 2008; Handley & Handley, 2003; Hostler, Wang, Parrish, Platt, & Guimond, 2005).

Two previous investigations (Abella, 2007; Kramer-Johansen et al., 2006) could not demonstrate improvements in clinical outcome resulting from feedback-induced improvements in CPR quality. However, neither of those investigations was sufficiently powered to detect clinically relevant outcome improvements.

On the other hand, investigators sufficiently powered this study to detect a 10 percent improvement in ROSC rates related to real-time changes in CPR quality, but the investigation could not detect such changes.

The results seem counterintuitive. One might expect that CPR quality improvements would result in better outcomes for patients. There may, however, be important reasons for why this study could not demonstrate higher ROSC rates.

First, although feedback helped paramedics perform CPR that was closer to the established guidelines, there is no guarantee that the established guidelines are optimal for survival. The feedback mechanism used in this study helped paramedics reduce the compression rate from 108 to 103 compressions per minute, which is closer to the current AHA recommendations for compression rate (Berg, 2010).

Is it reasonable to expect significant survival benefits with a reduction in the mean number of chest compressions by an average of five per minute when both the intervention and the control groups have compression rates that are already within the recommendations?

In fact, data from an out-of-hospital observational study suggests that improved survival to hospital discharge rates may still be possible if medics deliver 120 chest compressions per minute (Christenson, 2009).

Patient enrollment for this study occurred during a period when the American Heart Association (2005) recommended a chest compression depth of 1.5-2 inches (roughly 38-50 millimeters). Data suggests that improving chest compression depth to meet that recommendation increases defibrillation success (Edelson, 2006) and results in higher rates of short-term survival (Kramer-Johansen, 2006).

Both the control and intervention groups in this study met the 2005 recommendation. In 2010, however the AHA increased its recommendation for compression depth to at least 2 inches (Berg, 2010), which is about 0.5 inches deeper than the intervention group achieved in the present study.

The ROC recently found a strong association between increased compression depth and survival outcomes but no clear evidence indicating that the 2010 compression depth recommendation of the AHA was any better or worse than the 2005 recommendations (Stiell, 2012).

Finally, the feedback provided by the Q-CPR software was corrective, meaning that the medic had to do something wrong before being cued to perform correctly. These corrective cues may have distracted the resuscitation team from quickly performing other important tasks that affect ROSC rates. It is possible that prescriptive prompts, such as the cues provided by a metronome, would result in a different outcome.

Limitations

One limitation of this study is that the researchers could not, for obvious reasons, blind the paramedics to the intervention. Medics always knew whether or not they were receiving feedback. Those with preconceived ideas about the value of feedback could have inadvertently performed resuscitation skills differently for one group compared to another.

The authors of the study did not report on time interval from the time the dispatch center received the emergency call until the rescue team delivered the first shock for the 24 percent of patients in each group who presented with ventricular fibrillation.

On average, both groups received the same amount of epinephrine during the arrest period, but the authors did not report the time interval between EMS arrival on the scene and when the medics administered the first epinephrine.

Although randomization helps to distribute measurement error evenly, the inability to blind the rescuers to the intervention introduces a variable that could be responsible for the study results.

This study could not demonstrate survival advantages resulting from the use of a real-time CPR feedback device despite improving CPR quality measures. However, even before the study began, the participating EMS agencies dedicated the resources and personnel necessary for an aggressive CPR quality improvement program.

Perhaps the survival advantages provided by CPR feedback devices come in helping agencies with poor CPR performance achieve CPR component parameters that are more closely aligned with established guidelines.

References

Abella, B. S., Edelson, D. P., Kim, S., Retzer, E., Myklebust, H., Barry, A. M., O’Hearn, N., Hoek, T. L., & Becker, L. B. (2007). CPR quality improvement during in-hospital cardiac arrest using a real-time audiovisual feedback system. Resuscitation, 73(1), 54-61. doi:10.1016/j.resuscitation.2006.10.027

American Heart Association. (2005). Part 2: Adult basic life support. Circulation, 112(suppl), III-5-III-16. doi:10.1161/CIRCULATIONAHA.105.166472

Berg, R. A., Hemphill, R., Abella, B. A., Aufderheide, T. P., Cave, D. M., Hazinski, M. F., Lerner, E. B., Rea, T. D., Sayre, M. R., & Swor, R. A. (2010). Part 5: Adult basic life support: 2010 American Heart Association guidelines for cardiopulmonary resuscitation and emergency cardiovascular care. Circulation, 122(suppl 3), S685-S705. doi:10.1161/CIRCULATIONAHA.110.970939

Christenson, J., Andrusiek, D., Everson-Stewart, S., Kudenchuk, P., Hostler, D., Powell, J., Callaway, C. W., Bishop, D., Vaillancourt, C., Davis, D., Aufderheide, T. P., Idris, A., Stouffer, J. A., Stiell, I., & Berg, R. (2009). Chest compression fraction determines survival in patients with out-of-hospital ventricular fibrillation. Circulation, 120(13), 1241–1247. doi:10.1161/CIRCULATIONAHA.109.852202

Dine, C. J., Gersh, R. E., Leary, M., Riegel, B. J., Bellini, L. M., & Abella, B. S. (2008). Improving cardiopulmonary resuscitation quality and resuscitation training by combining audiovisual feedback and debriefing. Critical Care Medicine, 36(10), 2817-2822. doi:10.1097/CCM.0b013e318186fe37

Edelson, D. P., Abella, B. S., Kramer-Johansen, J., Wik, L., Myklebust, H., Barry, A. M., Merchant, R. M., Vanden Hoek, T. L., Steen, P. A., & Becker, L. B. (2006). Effects of compression depth and pre-shock pauses predict defibrillation failure during cardiac arrest. Resuscitation, 71, 137—145. doi:10.1016/j.resuscitation.2006.04.008

Handley, A. J., & Handley, S. A. J. (2003). Improving CPR performance using an audible feedback system suitable for incorporation into an external defibrillator. Resuscitation, 57(1), 57-62. doi:10.1016/S0300-9572(02)00400-8

Hostler, D., Everson-Stewart, S., Rea, T. D., Stiell, I. G., Callaway, C. W., Kudenchuk, P. J., Sears, G. K., Emerson, S. M., Nichol, G., & the Resuscitation Outcomes Consortium Investigators. (2011). Effect of real-time feedback during cardiopulmonary resuscitation outside hospital: Prospective, cluster-randomised trial. British Medical Journal, 342, d512. doi:10.1136/bmj.d512

Hostler, D., Wang, H., Parrish, K., Platt, T. E., & Guimond, G. (2005). The effect of a voice assist manikin (VAM) system on CPR quality among prehospital providers. Prehospital Emergency Care, 9(1), 53-60. doi:10.1080/10903120590891660

Kramer-Johansen, J., Myklebust, H., Wik, L., Fellows, B., Svensson, L., Sorebo, H., & Steen, P. A. (2006). Quality of out-of-hospital cardiopulmonary resuscitation with real time automated feedback: A prospective interventional study. Resuscitation, 71(3), 283-292. doi:10.1016/j.resuscitation.2006.05.011

Stiell, I. G., Brown, S. P., Christenson, J., Cheskes, S., Nichol, G., Powell, J., Bigham, B., Morrison, L. J., Larsen, J., Hess, E., Vaillancourt, C., Davis, D. P., Callaway, C. W., & the Resuscitation Outcomes Consortium (ROC) Investigators. (2012). What is the role of chest compression depth during out-of-hospital cardiac arrest resuscitation? Critical Care Medicine, 40(4), 1192-1198. doi:10.1097/CCM.0b013e31823bc8bb