Mechanical Chest Compressions and Simultaneous Defibrillation vs Conventional Cardiopulmonary Resuscitation in Out-of-Hospital Cardiac Arrest: The LINC Randomized Trial

Importance  A strategy using mechanical chest compressions might improve the poor outcome in out-of-hospital cardiac arrest, but such a strategy has not been tested in large clinical trials.

Objective  To determine whether administering mechanical chest compressions with defibrillation during ongoing compressions (mechanical CPR), compared with manual cardiopulmonary resuscitation (manual CPR), according to guidelines, would improve 4-hour survival.

Design, Setting, and Participants  Multicenter randomized clinical trial of 2589 patients with out-of-hospital cardiac arrest conducted between January 2008 and February 2013 in 4 Swedish, 1 British, and 1 Dutch ambulance services and their referring hospitals. Duration of follow-up was 6 months.

Interventions  Patients were randomized to receive either mechanical chest compressions (LUCAS Chest Compression System, Physio-Control/Jolife AB) combined with defibrillation during ongoing compressions (n = 1300) or to manual CPR according to guidelines (n = 1289).

Main Outcomes and Measures  Four-hour survival, with secondary end points of survival up to 6 months with good neurological outcome using the Cerebral Performance Category (CPC) score. A CPC score of 1 or 2 was classified as a good outcome.

Results  Four-hour survival was achieved in 307 patients (23.6%) with mechanical CPR and 305 (23.7%) with manual CPR (risk difference, –0.05%; 95% CI, –3.3% to 3.2%; P > .99). Survival with a CPC score of 1 or 2 occurred in 98 (7.5%) vs 82 (6.4%) (risk difference, 1.18%; 95% CI, –0.78% to 3.1%) at intensive care unit discharge, in 108 (8.3%) vs 100 (7.8%) (risk difference, 0.55%; 95% CI, –1.5% to 2.6%) at hospital discharge, in 105 (8.1%) vs 94 (7.3%) (risk difference, 0.78%; 95% CI, –1.3% to 2.8%) at 1 month, and in 110 (8.5%) vs 98 (7.6%) (risk difference, 0.86%; 95% CI, –1.2% to 3.0%) at 6 months with mechanical CPR and manual CPR, respectively. Among patients surviving at 6 months, 99% in the mechanical CPR group and 94% in the manual CPR group had CPC scores of 1 or 2.

Conclusions and Relevance  Among adults with out-of-hospital cardiac arrest, there was no significant difference in 4-hour survival between patients treated with the mechanical CPR algorithm or those treated with guideline-adherent manual CPR. The vast majority of survivors in both groups had good neurological outcomes by 6 months. In clinical practice, mechanical CPR using the presented algorithm did not result in improved effectiveness compared with manual CPR.

Trial Registration  clinicaltrials.gov Identifier: NCT00609778

Many factors affect the chances of survival after cardiac arrest, including early recognition of arrest, effective cardiopulmonary resuscitation (CPR) and defibrillation, and postresuscitation care. One important link is the delivery of high-quality chest compressions to achieve restoration of spontaneous circulation (ROSC).^1-4

The effectiveness of manual chest compressions depends on the endurance and skills of rescuers, and manual compressions provide only approximately 30% of normal cardiac output.^5,6 Manual CPR is also limited by prolonged hands-off time, and its quality is particularly poor when it is administered during patient transport.^7,8 Mechanical chest compression devices have therefore been developed to improve CPR.

Experimental studies with the mechanical chest compression device used in this study have shown improved organ perfusion pressures, enhanced cerebral blood flow, and higher end-tidal CO₂ compared with manual CPR, with the latter also supported by clinical data.^9-11 This device sustains adequate circulation during percutaneous coronary intervention and has been used in cases of hypothermia/drowning.^12,13

Two randomized pilot studies (N = 328 and N = 149) of out-of-hospital cardiac arrest compared manual and mechanical chest compressions using this device and did not find any outcome differences.^14,15 To date, there has been no evidence from large randomized trials about the effectiveness and safety of this mechanical device compared with manual CPR.

The LINC (LUCAS in Cardiac Arrest) study was designed to evaluate the effectiveness and safety of an algorithm using mechanical chest compressions combined with defibrillation during ongoing compressions (mechanical CPR) compared with manual CPR according to guidelines.^16,17 The rationale for this design of the algorithm with mechanical chest compressions was based on studies suggesting the importance of compressions before defibrillation and a minimal hands-off interval.^4,18,19 The primary objective was to assess whether treatment with mechanical CPR would result in superior 4-hour survival in patients with out-of-hospital cardiac arrest compared with treatment with manual CPR.

The LINC study was initiated by Uppsala University and sponsored by Physio-Control/Jolife AB. The study was approved by the regional ethical review board in Uppsala, Sweden, the research ethics committee in the United Kingdom, and the United Human Subjects Research Committees in the Netherlands. It was conducted in accordance with regulatory requirements, Good Clinical Practices, and the ethical principles of the Declaration of Helsinki. All survivors with sufficient mental capacity were given information about the study. If further participation was agreed on, written consent was obtained. If survivors did not have sufficient mental capacity, information was presented to family, who provided written consent if they decided to further participate. Consent was waived for included nonsurvivors by the ethical committees.

This multicenter randomized clinical trial enrolled patients from January 2008 to August 2012 in 6 advanced life support emergency medical systems (EMS): Gävle, Malmö, Västerås, and Uppsala in Sweden, Utrecht in the Netherlands, and Dorset in the United Kingdom. Its protocol has been described in detail.¹⁷ For inclusion, patients had to be adults with unexpected out-of-hospital cardiac arrest for whom an attempt of resuscitation was considered appropriate. Exclusion criteria were traumatic cardiac arrest (including hanging), age younger than 18 years, known pregnancy, and a body size too large or small to fit the chest compression device. Patients undergoing defibrillation before the device arrived on scene and patients with crew-witnessed cardiac arrest who achieved ROSC after immediate defibrillation were not eligible for the study.

The LUCAS Chest Compression System (Physio-Control/Jolife AB) is a mechanical CPR device with an integrated suction cup designed to deliver compressions according to resuscitation guidelines. The device and a randomization envelope were placed on all ambulances and brought to all patients with dispatch codes of sudden cardiac arrest or unconsciousness and when called for by local guidelines.¹⁷ Enrollment was performed on scene immediately when the EMS recognized a cardiac arrest. Manual CPR was started and patients who met the eligibility criteria were immediately randomized in 1:1 balance using sealed opaque envelopes at the patient’s side.

Patients randomized to the mechanical CPR algorithm (Figure 1) were immediately treated with manual chest compressions until the device was deployed. Mechanical compressions were initiated and continued for 3 minutes; first defibrillation shock was delivered during ongoing compressions, without pausing to check the heart rhythm, 90 seconds into the first 3-minute cycle. Heart rhythm was checked after each 3-minute cycle; if a shockable rhythm was observed, a new 3-minute cycle was started and a countershock was delivered after 90 seconds of compressions without pausing. If no shockable rhythm was observed, a 3-minute cycle without interruption started. Patients randomized to receive manual CPR were treated in accordance with the 2005 European Resuscitation Council guidelines.¹⁶ In both groups, ventilation and drugs were given according to guidelines.¹⁶

To ensure adherence to the study design, all EMS personnel were trained in both study algorithms before starting the study and were retrained every 6 months during the entire study period. For randomly chosen EMS personnel, skill level and adherence to the algorithms were evaluated using manikins by 2 supervisors visiting the sites once a year. Feedback of their skills was given by the supervisors.

The primary outcome was 4-hour survival after successful ROSC. Secondary outcomes included ROSC defined as spontaneous palpable pulse, arrival to the emergency department with a spontaneous palpable pulse, and survival with good neurological outcome to intensive care unit discharge, to hospital discharge, and at 1 and 6 months. Cerebral Performance Category (CPC) scores were used in survivors to define neurological outcome, with CPC scores of 1 or 2 indicating good outcome and CPC scores of 3 or 4 indicating poor outcome (Box).²⁰ This was done by the on-site responsible nurse or physician who had access to the study documentation. Follow-up after hospital discharge was performed by telephone or visits to the clinic at 1 and 6 months after the cardiac arrest. To monitor the clinical safety of the device, adverse device events and serious adverse events were recorded by the EMS and hospital personnel for each individual patient. An interim analysis of the primary end point of 4-hour survival was performed during spring 2011 by an independent safety committee within the Scandinavian Society of Anaesthesiology and Intensive Care Medicine. The committee recommended continuing the study.

Box Section Ref ID

Patients with ROSC were treated with mild hypothermia to 32°C to 34°C (89°F-93°F) for 24 hours, regardless of initial electocardiogram (ECG) rhythm, if no contraindications were present. Acute coronary angiography was considered during the first 48 hours and, if indicated, including ST-segment elevation on a 12-lead ECG, a percutaneous coronary intervention was performed.

Monitoring, database management, and all statistical analyses were coordinated independently by Uppsala Clinical Research Center, Uppsala, Sweden. All predefined analyses were performed in accordance with the protocol on the intention-to-treat population, comprising all randomized patients except surviving patients who refused participation in the trial. All outcomes were analyzed using Wald 95% confidence intervals for the difference in proportions and a 2-sided Fisher exact test. Missing values were imputed as the worst outcome, as predefined, so that for the CPC end points patients lost to follow-up were analyzed as not being alive with CPC scores of 1 or 2. Statistical significance for the primary variable was defined as P < .048 in accordance with the interim analysis plan. It was assumed that in the manual CPR group, the proportion of 4-hour survival would be 25% and with mechanical CPR at least 31%. To detect the anticipated difference of at least 6% with a power of 90% in the final analysis, the study required a total of 2500 patients; ie,1250 patients in each treatment group in the intention-to-treat population. Further details of the group-sequential design and sample size calculation are described elsewhere.¹⁷ All statistical analyses were performed using SAS version 9.3, SAS Institute.

During the study, 4998 cases of cardiac arrest were screened, of which 2593 were included in randomization. Four patients were excluded because of withdrawn informed consent, resulting in 2589 included patients in the intention-to-treat population, with 1300 patients in the mechanical CPR group and 1289 patients in the manual CPR group. After randomization, 116 patients were found either to meet exclusion criteria or not to meet inclusion criteria; for this intention-to-treat analysis, however, all patients were analyzed in the group they were randomized to regardless of this or eventual crossover or other protocol deviations (Figure 2).

Cardiac arrest background variables and events are described in Table 1. The notable differences between groups were the number of defibrillations delivered by the EMS crew and time to first defibrillation, which was delivered 1.5 minutes later in the mechanical CPR group, a possible result of the difference between the 2 treatment algorithms.

For the primary outcome, there was no significant difference in 4-hour survival between the mechanical CPR group and the manual CPR group (307/1300 [23.6%] vs 305/1289 [23.7%]; risk difference, −0.05%; 95% CI, –3.3% to 3.2%; P<.99). Similarly, there was no significant difference between groups in any of the secondary outcomes (Table 2).

Among the surviving patients in the mechanical CPR vs manual CPR groups, 62% vs 54% had CPC scores of 1 or 2 at intensive care unit discharge, 92% vs 86% had such scores at hospital discharge, 94% vs 88% at 1 month, and 99% vs 94% at 6 months after cardiac arrest. The CPC scores of surviving patients are shown in Table 2.

Among patients admitted to the hospital after ROSC, 198 (63%) were treated with hypothermia in the mechanical CPR group vs 214 (66%) in the manual CPR group (risk difference, –3.4%; 95% CI, –10.8% to 4.0%). Median duration of the treatment was 24.0 and 24.5 hours, respectively. Coronary angiography was performed in 118 patients (37%) in the mechanical CPR group and in 130 (40%) in the manual CPR group (risk difference, –2.8%; 95% CI, –10.3% to 4.8%); 75 mechanical CPR patients (24%) and 87 manual CPR patients (27%) (risk difference, –3.1%; 95% CI, –9.9% to 3.6%) were treated with percutaneous coronary intervention, respectively.

Twenty-three device-related adverse events were reported among 1282 uses of mechanical CPR. Of these, 8 cases involved a device malfunction in which the use of the mechanical device was discontinued. In the remaining 15 cases, mechanical CPR could be continued; in 7 of these cases, the device had to be repositioned and in 8 cases, minor technical issues were reported.

There were 7 reported serious adverse events in the mechanical CPR group and 3 in the manual CPR group. In the mechanical CPR group the following were reported: 1 case of possible airway bleeding; 1 case of suspected rupture of the spleen seen on computed tomography that was not confirmed when an autopsy was done; 1 case of pneumothorax; 1 case of a fractured thoracic vertebra in which bystander CPR was provided in the patient’s bed followed by mechanical CPR; 1 flail chest (noted before deploying mechanical chest compressions); 1 migration of the device due to mucus on the chest, leading to subsequent removal of the device; and 1 case of preexisting stomach distension preventing the device from being properly applied. In the manual CPR group, 1 case of flail chest and abdominal aortic aneurysm, 1 case of flail chest, and 1 case of pneumothorax were reported.

In this large, randomized, multicenter trial, an algorithm combining mechanical chest compressions and defibrillation during ongoing compressions provided no survival advantage over manual CPR administered according to guidelines. No difference in survival or neurological outcome was seen for up to 6 months after the cardiac arrest as, by then, the vast majority of survivors had CPC scores of 1 or 2, and most patients with initial CPC scores of 3 or 4 had either improved or died. The numbers of serious adverse events and device-related adverse events were low.

We chose 4-hour survival as the primary end point to study the effect of the 2 prehospital interventions because it would minimize any influence of expected variations in postresuscitation care. However, postresuscitation care was similar between the groups, supporting the validity of the observed similarity in secondary end points at time points up to 6 months. The current sample size has a 95% confidence interval for the 4-hour survival ranging from −3.3% to +3.2%. Translated another way, while the point estimate for treatment effect was near 0.0, our study could not rule out the possibility of a 3.2% benefit or a similarly sized harm from mechanical CPR relative to standard CPR. Similar considerations will affect the interpretation of the secondary outcomes of survival (ie, survival with good neurological outcome up to 6 months), which may be an even more relevant measurement of treatment outcome.

Rather than simply replacing manual compressions with mechanical ones, the mechanical CPR algorithm bundled several other changes to the resuscitation algorithm. Most notably, a first countershock was to be delivered to each patient in this group regardless of the presenting rhythm during ongoing compressions. This provided a continuous period of mechanical compressions leading up to the shock, eliminating the usual preshock pause to assess rhythm, and thereby potentially improving outcomes for patients presenting in ventricular fibrillation. As a consequence, many patients with nonshockable initial rhythms received an unnecessary shock. Inappropriate shocks have previously been shown to be relatively common during manual defibrillator use, and there is little or no evidence that they are harmful.²¹ The consensus of the steering committee designing the study was that this initial shock without analysis in the mechanical CPR group had more potential for benefit than harm. Also of note, the mechanical CPR algorithm used 3-minute CPR periods rather than conventional 2-minute periods. With mechanical devices, compressions can be delivered for 3 minutes without concern about rescuer fatigue, and the approach might improve outcomes by increasing chest compression fraction.²²

If the mechanical CPR group received consistently good chest compressions with few pauses, we can only speculate about why our hypothesis was not supported. Perhaps manual chest compressions were also consistently good or the delay of defibrillation in the mechanical CPR group caused by the specific algorithm was detrimental. By specifying initial defibrillation without any prior ECG rhythm analysis in that group, the aim was to minimize any delays to mechanical compressions and to the first defibrillation. However, the first defibrillation occurred 1.5 minutes later in the mechanical CPR group than in the manual CPR group (Table 1). By protocol, the first countershock was to be delivered 90 seconds after starting mechanical compressions; if it had been delivered at the start of mechanical compressions instead, time to defibrillation could have been similar in the 2 groups. This adjustment to the protocol might improve survival in the mechanical CPR group by several percent.²³ However, it is also possible that the additional compressions before defibrillation were beneficial.

Except for the difference in the number of defibrillations provided, which reflects the 2 different algorithms, background and demographic variables did not differ between the groups. This, together with the dropout of only 4 patients not willing to provide informed consent, supports the robustness of our results.

To better evaluate the mechanical CPR algorithm, our design excluded patients treated with defibrillation before arrival of the EMS crew (n = 337) and patients with crew-witnessed cardiac arrest achieving ROSC after the first defibrillation (numbers unknown). Because these excluded patients have relatively high survivability, the survival rate across all treated cardiac arrests in the participating communities is probably higher than the survival rate observed in our study.

Good clinical outcomes with a medical device depend in part on the usability and reliability of the device. This study documented a low rate of device malfunctions (<1%). Before randomization, 1.5% of the patients were deemed to be too large or too small to fit and were excluded. After randomization, 3.5% of the patients randomized to mechanical CPR did not fit the device; 2.3% were too big and 1.2% too small. This suggests the device can be expected to fit about 95% of cardiac arrest patients.

There were some limitations. The adherence to the 2 different algorithms was not evaluated on scene but is reflected in the number of defibrillations delivered. Even if monitoring on scene would have been performed, available technology allowed recordings only of compression rate and pauses and not of correct depth or optimal positioning of hand or suction cup on the chest. In approximately 10% of the patients, impedance data (Code-Stat, Physio-Control Inc) was recorded and showed a chest compression fraction of 0.78 in the manual CPR group vs 0.84 in the mechanical CPR group. The mechanical CPR algorithm called for a first shock to all patients, without any prior cardiac rhythm analysis. At least 1 defibrillation was delivered to 75% of the patients in the mechanical CPR group vs 45% in the manual CPR group (Table 1). With 1% of patients receiving an unknown number of defibrillation shocks, we believe that responders did not fully adhere to the mechanical CPR algorithm in 24% of the cases. Of those, 93% were nonshockable rhythms; therefore, we suspect that some of the responders looked at the ECG before the shock. But we do not know if the responders registered the first rhythm as being the rhythm seen after the first defibrillation or before the first defibrillation in the mechanical CPR group. However, the distribution of the initial rhythms is similar in the 2 groups and similar to that in other large randomized trials within this patient population.^24-26

We cannot tell to what degree the unique components of the 2 different algorithms or to what degree the mechanical and manual chest compressions alone have influenced the results. The question of whether this mechanical CPR device should replace manual chest compressions while maintaining other components of the guideline-directed resuscitation algorithm has to be investigated separately. However, we studied mechanical CPR implemented in an algorithm expected to work in most EMS organizations without unreasonable requests for resources.

In patients with out-of-hospital cardiac arrest, mechanical chest compressions in combination with defibrillation during ongoing compressions provided no improved 4-hour survival vs manual CPR according to guidelines. There was a good neurological outcome in the vast majority of survivors in both groups, and neurological outcomes improved over time. Thus, in clinical practice, CPR with this mechanical device using the presented algorithm can be delivered without major complications but did not result in improved outcomes compared with manual chest compressions.

Corresponding Author: Sten Rubertsson, MD, PhD, Department of Surgical Sciences/Anaestesiology and Intensive Care, Uppsala University, Uppsala University Hospital, SE 75185 Uppsala, Sweden (sten.rubertsson@akademiska.se).

Published Online: November 17, 2013. doi:10.1001/jama.2013.282538.

Author Contributions: Dr Rubertsson had full access to all of the data in the study and takes responsibility for the integrity of the data and the accuracy of the data analysis.

Study concept and design: Rubertsson, Östlund, Silfverstolpe, Lichtveld, Boomars, Kastberg, Halliwell, Box, Herlitz, Karlsten.

Acquisition of data: Rubertsson, Lindgren, Smekal, Silfverstolpe, Lichtveld, Boomars, Ahlstedt, Skoog, Kastberg, Halliwell, Box.

Analysis and interpretation of data: Rubertsson, Lindgren, Östlund, Silfverstolpe, Boomars, Kastberg, Herlitz, Karlsten.

Drafting of the manuscript: Rubertsson, Lindgren, Östlund, Lichtveld, Halliwell, Herlitz, Karlsten.

Critical revision of the manuscript for important intellectual content: Rubertsson, Lindgren, Smekal, Östlund, Silfverstolpe, Lichtveld, Boomars, Ahlstedt, Skoog, Kastberg, Box, Herlitz, Karlsten.

Statistical analysis: Rubertsson, Östlund.

Obtained funding: Rubertsson.

Administrative, technical, or material support: Rubertsson, Lindgren, Smekal, Östlund, Silfverstolpe, Lichtveld, Boomars, Ahlstedt, Skoog, Kastberg, Halliwell, Box, Herlitz, Karlsten.

Study supervision: Rubertsson, Lindgren, Smekal, Östlund, Silfverstolpe, Lichtveld, Boomars, Ahlstedt, Kastberg, Box, Herlitz, Karlsten.

Conflict of Interest Disclosures: All authors have completed and submitted the ICMJE Form for Disclosure of Potential Conflicts of Interest. Dr Rubertsson reports receiving funding for consultation from Physio-Control. No other disclosures were reported.

Funding/Support: This work was supported by institutional grants from Uppsala University and by Physio-Control/Jolife AB.

Role of the Sponsor: Physio-Control/Jolife AB provided the mechanical device used in the study, assisted with training, funded a fee per patient enrolled at each site, and funded the CRO-Uppsala Clinical Research Center. Two persons from Physio-Control/Jolife AB were nonvoting members of the steering committee of the study. Physio-Control/Jolife AB had no role in the design and conduct of the study; collection, management, analysis and interpretation of the data; preparation or approval of the manuscript. Before submission the sponsor was allowed to review and comment on the manuscript. The investigators were under no obligation to incorporate any such input. The decision to submit for publication was by the authors and not by the sponsor.

Previous Presentation: An abstract of the study was presented at the European Society of Cardiology annual conference; September 1, 2013; Amsterdam, the Netherlands.

Additional Contributions: We thank the members of the LINC study group: Liselott Rehn, RN, and Tomas Nyman, RN, Region Skåne Prehospital Center, Skåne University Hospital, Lund, Sweden; Wendy Bruins, PhD, and Anja Radstok, RN, Regional Ambulance Service Utrecht, Utrecht, the Netherlands; Helena Puggioli, RN, Västerås Central Hospital, Västerås, Sweden; Anna Lindblad, RN, Gävle Hospital, Gävle, Sweden; Douglas Chamberlain, MD, Institute of Primary Care and Public Health, Cardiff University School of Medicine, Cardiff, Wales; and Fredrik Arnwald and Bjarne Madsen Hardig, RN, PhD, Physio-Control/Jolife AB, Lund, Sweden. We also thank the Scandinavian Society of Anaesthesiology and Intensive Care Medicine research board for performing the interim analysis of the data; all involved EMS personnel, ambulance services, and hospital staff in Uppsala, Gävle, Västerås, and Malmö, Sweden, the Dorset region in England, and Utrecht in the Netherlands; and Uppsala Clinical Research Center for handling the database and monitoring the study.