Feasibility of Biosignal-guided Chest Compression During Cardiopulmonary Resuscitation: A Proof of Concept

BRIEF REPORT Feasibility of Biosignal-guided Chest Compression During Cardiopulmonary Resuscitation: A Proof of Concept Matthew L. Sundermann, MS, David D. Salcido, PhD, Allison C. Koller, and James J. Menegazzi, PhD Abstract Objectives: Cardiac arrest is one of the leading causes of death in the United States and is treated by cardiopulmonary resuscitation (CPR). CPR involves both chest compressions and positive pressure ventilations when given by medical providers. Mechanical chest compression devices automate chest compressions and are beginning to be adopted by emergency medical services with the intent of providing high-quality, consistent chest compressions that are not limited by human providers who can become fatigued. Biosignals acquired from cardiac arrest patients have been characterized in their ability to track the effect of CPR on the patient. The authors investigated the feasibility and appropriate response of a biosignal-guided mechanical chest compression device in a swine model of cardiac arrest. Methods: After a custom signal-guided chest compression device was engineered, its ability to respond to biosignal changes in a swine model of cardiac arrest was tested. In a preliminary series of six swine, two biosignals were used: mean arterial pressure (MAP) and a mathematical derivative of the electrocardiogram waveform, median slope (MS). How these biosignals changed was observed when chest compression rate and depth were adjusted by the signal-guided chest compression device, independent of the user. Chest compression rate and depth were adjusted by the signal-guided chest compression device according to a preset threshold algorithm until either of the biosignals improved to satisfy a set threshold or until the chest compression rate and depth achieved maximum values. Defibrillation was attempted at the end of each resuscitation in an effort to achieve return of spontaneous circulation (ROSC). Results: The signal-guided chest compression device responded appropriately to biosignals by changing its rate and depth. All animals exhibited positive improvements in their biosignals. During the course of the resuscitation, three of the six animals improved their MS biosignal to reach the MS threshold, while two of the six animals improved their MAP biosignal to reach the MAP threshold. In the six experiments conducted, defibrillation was attempted on five animals, and two animals achieved ROSC. Conclusions: In this proof-of-concept study, a signal-guided chest compression device was demonstrated to be capable of responding to biosignal input and delivering chest compressions with a broad range of rates and depths. ACADEMIC EMERGENCY MEDICINE 2016;23:93 97 2015 by the Society for Academic Emergency Medicine Out-of-hospital cardiac arrest (OHCA) is one of the leading causes of death in the United States and is treated by cardiopulmonary resuscitation (CPR), a technique that aims to restore oxygenated blood flow through chest compressions and rescue breaths. 1 4 The American Heart Association recommends chest compressions be given at a depth of at least 2 inches and a rate of at least 100 compressions per minute, regardless of patient chest size, patient age, or cardiac arrest etiology. 4 This simplified approach From the Department of Emergency Medicine (MLS, DDS, ACK, JJM) and the Department of Bioengineering (MLS, JJM), University of Pittsburgh, Pittsburgh, PA. Received June 9, 2015; revisions received August 21, 2015 and August 16, 2015; accepted August 27, 2015. Presented at the American Heart Association RESS, Chicago, IL, November 2014. Funding from the University of Pittsburgh Center for Medical Innovation Grants F_070-2013 and F_036-2012 MLS, DDS, and JJM have a patent pending titled: Automatic Chest Compression Systems That Incorporates Biological Feedback, PCT/US2014/071544. Supervising Editor: David C. Cone, MD. Address for correspondence and reprints: Matthew L. Sundermann, MS; e-mail: sundermannml@upmc.edu. 2015 by the Society for Academic Emergency Medicine ISSN 1069-6563 93 doi: 10.1111/acem.12844 PII ISSN 1069-6563583 93

94 Sundermann et al. SIGNAL-GUIDED FEEDBACK CPR DEVICE ignores the possibility that all patients may not benefit from the same type of CPR. Intuitively, larger patients may require deeper chest compression depths; pediatric patients may require faster chest compression rates. In addition, the quality of chest compressions during CPR has been linked to more favorable outcomes, necessitating a consistent, high-quality chest compression delivery. 5,6 Mechanical chest compression devices automate compressions and are beginning to be adopted with the intent of providing better, high-quality chest compressions that are not limited by fatigued or inconsistent human providers. 7,8 Mechanical chest compression devices are a promising innovation in cardiac arrest treatment, but are limited by their use of fixed rate and depth settings. Due to the nature of mechanical chest compression devices being electronically driven, these devices have the potential to be linked to biosignal acquisition devices commonly used during resuscitation, such as the defibrillator monitor. We think this could be a useful link. Common biosignals recorded on the defibrillator monitor, such as quantitative electrocardiogram (qecg) and end-tidal carbon dioxide, have been well characterized in their ability to effectively track the perfusion status of patients during cardiac arrest. 9 12 Biosignals during cardiac arrest that relate to CPR performance include invasive mean arterial pressure (MAP) and the qecg metric median slope (MS), a noninvasive biosignal that characterizes the ventricular fibrillation (VF) waveform. Higher MS values correlate with greater success of rescue shock conversion and a more perfused myocardium, whereas lower MS values indicate the opposite. 13 If mechanical chest compression devices could respond to changing biosignals from the patient, CPR parameters such as rate and depth could be tailored to each patient. Similar approaches have already been demonstrated with strictly hemodynamic goal-oriented approaches. 14,15 We describe here a preliminary report of a biosignal-guided mechanical chest compression device in a swine model of cardiac arrest as a proof of concept. METHODS Study Design The Institutional Animal Care and Use Committee at the University of Pittsburgh approved this study (Protocol 14033392). The primary goal of the study was to assess the feasibility of the signal-guided chest compression device to modify chest compressions in response to changing biosignals. Animal Subjects and Handling Female mixed-breed domestic swine (Sus scrofa), mean mass of 24.1 kg, were prepared in a standardized fashion. Animals were sedated (10 mg/kg ketamine/4 mg/kg xylazine), anesthetized (fentanyl, 50 lg/kg loading dose/50 lg/kg/hr infusion), and paralyzed (4-mg bolus of vecuronium/2-mg additional boluses as needed), and then following endotracheal intubation, they were mechanically ventilated (Ohmeda 7000, GE Healthcare, Little Chalfront, UK) with room air. Central arterial and venous pressure monitoring was instrumented according to our standard laboratory procedure described elsewhere. 11 Standard Lead II ECG was recorded (Dual Bio AMP FE 135, AD Instruments, Colorado Springs, CO) and sampled at 1000 Hz. Study Protocol Device Creation. We built a custom, electromechanically controlled, signal-guided chest compression device. This system included three principal components: a computer central processing unit (CPU)-coupled linear actuator responsible for piston movement, a signal acquisition unit that collected biosignals, and a main control computer that coordinated feedback signals and commands between the actuator and biosignal acquisition unit (Powerlab 16/30 Model ML880, AD Instruments) recording at 1000 Hz. Chest compressions were delivered by the actuator piston (UltraMotion, Inc., Cutchogue, NY; and Moog, Inc., Elma, NY). Guidance signals were then transmitted to MATLAB (Mathworks, Inc.) through a custom cross-platform memory sharing script provided by AD Instruments. The layout of this design is shown in Figure 1. The signal-guided chest compression device, with experience from previous swine models, was limited to a maximum depth of 2 inches and a maximum rate of 130 compressions/minute. The signal-guided chest compression device was programmed to make adjustments independent of the user. Adjustments to depth and rate would be made only when a preset physiologic threshold was not met during any given chest compression pause, and CPR parameters were changed until the threshold was reached or until the maximum rate and depth were reached. Experimental Protocol. The general experimental timeline is displayed in Figure 2. VF was induced with a 3-second 100 ma transthoracic shock and left untreated Data acquisi on unit Biosignals Piston CPR Control control Algorithm algorithm Figure 1. Layout design of the biosignal-guided chest compression device. CPR = cardiopulmonary resuscitation.

ACADEMIC EMERGENCY MEDICINE January 2016, Vol. 23, No. 1 www.aemj.org 95 Simulated ambulance response time CPR until either the signal threshold is met OR when max parameters met at 14 min MAP biosignal compared to MAP threshold every 30 sec Defibrillation Attempt ROSC 0 min 6 min If signal threshold met at anytime during CPR: MS biosignal compared to MS threshold every 30 sec 14 min: Max CPR Parameters Reached ROSC + 60 min Figure 2. Experiment approach to biosignal-guided CPR. CPR = cardiopulmonary resuscitation; MAP = mean arterial pressure; MS = median slope; ROSC = return of spontaneous circulation. for 6 minutes to simulate an OHCA. After 6 minutes, CPR was initiated using the signal-guided chest compression system, with 3-second pauses every 30 seconds for rhythm analysis. If the physiologic threshold was met or the maximum CPR parameters were met, a rescue shock (150J, Zoll M-Series, Zoll, Chelmsford, MA) was attempted after the subsequent 30 seconds. In the event of a failed rescue shock, no further shocks were attempted. If return of spontaneous circulation (ROSC) was achieved, the swine were monitored briefly. While still under general anesthesia, surviving animals were euthanized with a rapid IV infusion of 40 meq of potassium chloride. Our signal-guided chest compression approach is displayed in Data Supplement S1 (available as supporting information in the online version of this paper). The biosignal thresholds in each experiment were set based on initial recordings of the biosignals during the baseline VF period, defined as the 30 seconds before initiation of CPR. The magnitude of the MS threshold was chosen based on analysis of representative traces from previous, unrelated experiments, as well as what was observed in previous animals within this experimental series. Due to the lack of a priori knowledge of the performance characteristics of the device, we made an educated guess at the beginning of the series, followed by refinement to reflect the over- or undershoot of our initial guess. This was held for MAP as well. Measures The biosignals we investigated were MAP and quantitative MS. These biosignals were checked during 3-second chest compression pauses every 30 seconds throughout the resuscitation so as to reduce signal artifact from the chest compressions. MAP was calculated as the mean central aortic pressure over a 3-second window of a continuous 1000-Hz sample. The MS of the ECG was used as the qecg metric. MS was calculated as the median of the collective sample to sample differences taken from a down-sampled, 250-Hz 3-second window of VF smoothed with a 10-point moving average filter. The primary outcome we assessed was appropriate response of the signal-guided chest compression device to biosignals above and below our biosignal threshold. We defined appropriate response as the signal-guided chest compression device, independent of the user, adjusting compression rate and depth when the biosignals were below threshold, and keeping compression rate and depth constant, when above threshold. Data Analysis Data are reported without statistical analysis due to the pilot size and proof of concept design. RESULTS Our custom device responded appropriately to biosignals by changing its rate and depth. All animals exhibited positive improvements in their biosignals. The animals that did not reach either of the two biosignal thresholds had their chest compression parameters adjusted until they reached the maximum parameter level (depth = 2 inches, rate = 130 compression/min). In the six experiments conducted, defibrillation was attempted on five animals, and two animals achieved ROSC. A summary of the results can be seen in Data Supplement S2 (available as supporting information in the online version of this paper). Data Supplement S3 (available as supporting information in the online version of this paper) shows an example experiment for an animal that reached a MS biosignal threshold and was successfully defibrillated. Data Supplement S4 (available as supporting information in the online version of this paper) shows an animal that reached the MAP biosignal threshold and was not successfully defibrillated. During the course of the resuscitation, three of the six animals improved their biosignals to reach the MS threshold, while two of the six animals reached the MAP threshold. The maximum depth of 2 inches was met in all experiments. DISCUSSION The treatment of cardiac arrest remains a black box in terms of providing optimized care for patients. Patients may have different chest sizes and different cardiac arrest etiology and may respond to chest compressions and drug treatment in different ways. 16 The variable nature of patients necessitates an adaptive approach to CPR. Mechanical chest compression devices, when

96 Sundermann et al. SIGNAL-GUIDED FEEDBACK CPR DEVICE coupled to a smart CPU capable of receiving biosignal input, are adaptable to the variable nature of patients. Our approach very much depends on how accurately our adaptive device is able to acquire and interpret biosignals and in turn relate those biosignals to the quality of perfusion generated by the chest compressions. In a prehospital setting, however, it would be required that these biosignals are noninvasively measured. Biosignal metrics used in our study such as MS qecg are noninvasive and readily available and are derived from the ECG, which is ubiquitously recorded in ambulances for every cardiac arrest patient. A link could therefore be established between existing defibrillator monitors and a biosignal-driven CPR device. Our findings provide preliminary evidence that a beneficial increase in perfusion may be achieved in cardiac arrest patients by linking the delivery of CPR to invasive and noninvasive biosignals. With more optimal CPR given to patients, patient survival outcomes could be improved. While we have focused on using the biosignals to modify the chest compressions provided by a mechanical CPR device, this feedback could likely also be given to EMS personnel providing manual compressions. LIMITATIONS The study was limited by the use of a swine model that may not exhibit the same etiologies and physiology as human cardiac arrest patients. Our study is also limited in that only a small sampling of biosignals were investigated, MAP and MS. The MS biosignal is also limited in that it has only been demonstrated so far to have utility for analyzing VF. It may be unable, for example, to be employed as an analysis tool in a pulseless electrical activity or asystole cardiac arrest patient. The study was limited in its small sample size and the absence of randomization or blinding. To progress past feasibility tests, further studies will have a control group that receives standard, 2-inch compressions at a rate of 100/ min CPR with our device. CONCLUSIONS We demonstrated a proof of concept that a signal-guided chest compression device, capable of responding to biosignal input and delivering chest compressions with a broad range of rates and depths, is feasible. References 1. McNally B, Mehta M, Vellano K, et al. Out-hospital cardiac arrest surveillance: Cardiac Arrest Registry to Enhance Survival (CARES), United States, October 1, 2005-December 31, 2010. MMWR Surveill Summ 2011;60:1 19. 2. Sasson C, Rogers MA, Dahl J, et al. Predictors of survival from out-of-hospital cardiac arrest. Circulation 2010;3:63 81. 3. Nichol G, Thomas E, Callaway CW, et al. Regional variation in out-of-hospital cardiac arrest incidence and outcome. JAMA 2008;300:1423 31. 4. Field JM, Hazinski MF, Sayre MR, et al. Part 1: Executive summary 2010 American Heart Association Guidelines for Cardiopulmonary Resuscitation and Emergency Cardiovascular Care. Circulation 2010;122:640 S656. 5. Wik L, Steen PA, Bircher NG. Quality of bystander cardiopulmonary resuscitation influences outcome after prehospital cardiac arrest. Resuscitation 1994;28:195 203. 6. Gallagher EJ, G Lombardi, Gennis P. Effectiveness of bystander cardiopulmonary resuscitation and survival following out-of-hospital cardiac arrest. JAMA 1995;274:1922 5. 7. Perkins GD, Gates S. Mechanical chest-compression devices: current and future roles. Curr Opin Crit Care 2010;16:203 10. 8. Rubertsson S, Lindgren E, Smekal D, et al. Mechanical chest compressions and simultaneous defibrillation vs conventional cardiopulmonary resuscitation in out-of-hospital cardiac arrest. The LINC Randomized Trial. JAMA 2014;311:53 61. 9. Weil MH, Bisera J, Trevino RP, et al. Cardiac output and end-tidal carbon dioxide. Crit Care Med 1985;13:881 999. 10. Schoene P, Coult J, Murphy L, et al. Course of quantitative ventricular fibrillation waveform measure and outcome following out-of-hospital cardiac arrest. Heart Rhythm 2013;11:230 6. 11. Salcido DD, Kim YM, Sherman LD, et al. Quantitative waveform measures of the electrocardiogram as continuous physiologic feedback during resuscitation with cardiopulmonary bypass. Resuscitation 2012;83:505 10. 12. Alonso E, Aramendi E, Kramer-Johansen J, et al. Beyond ventricular fibrillation analysis: comprehensive waveform analysis for all cardiac rhythms occurring during resuscitation. Resuscitation 2014;85:1541 8. 13. Neurater A, Eftestøl T, Kramer-Johansen J, et al. Prediction of countershock success using single features from multiple ventricular fibrillation frequency bands and feature combinations using neural networks. Resuscitation 2007;73:253 63. 14. Sutton RM, Friess SH, Stuart H, et al. Hemodynamic directed CPR improves short-term survival from asphyxia-associated cardiac arrest. Resuscitation 2013;84:696 701. 15. Friess SH, Stuart H, Sutton RM, et al. Hemodynamic directed CPR improves cerebral perfusion pressure and brain tissue oxygenation. Resuscitation 2014;85:1298 303. 16. Jain R. Nallamothu BK, Chan PS, American Heart Association National Registry of Cardiopulmonary Resuscitation (NRCPR) Investigators. Body mass index and survival after in-hospital cardiac arrest. Circ Cardiovasc Qual Outcomes 2010;3: 490 7. Supporting Information The following supporting information is available in the online version of this paper:

ACADEMIC EMERGENCY MEDICINE January 2016, Vol. 23, No. 1 www.aemj.org 97 Data Supplement S1. Example algorithm used for biosignal feedback-guided CPR. Data Supplement S2. Summary of the experimental results. Data Supplement S3. Example experiment for an animal that reached a median slope biosignal threshold and was successfully defibrillated. Data Supplement S4. An example animal that reached the mean arterial pressure biosignal threshold and was NOT successfully defibrillated.