balt2/zhc-ci/zhc-ci/zhc02412/zhc0782-12z xppws S 1 5/14/12 12:11 4/Color Figure(s): F1 Art: 059535 Input-len

June 19, 2012

AQ:1

Relationship Between Chest Compression Rates and Outcomes From Cardiac Arrest
Ahamed H. Idris, MD; Danielle Guffey, BS; Tom P. Aufderheide, MD; Siobhan Brown, PhD; Laurie J. Morrison, MD, MSc; Patrick Nichols, DO; Judy Powell, BSN; Mohamud Daya, MD; Blair L. Bigham, MSc; Dianne L. Atkins, MD; Robert Berg, MD; Dan Davis, MD; Ian Stiell, MD, MSc; George Sopko, MD, MPH; Graham Nichol, MD, MPH; the Resuscitation Outcomes Consortium (ROC) Investigators
BackgroundGuidelines for cardiopulmonary resuscitation recommend a chest compression rate of at least 100 compressions per minute. Animal and human studies have reported that blood flow is greatest with chest compression rates near 120/min, but few have reported rates used during out-of-hospital (OOH) cardiopulmonary resuscitation or the relationship between rate and outcome. The purpose of this study was to describe chest compression rates used by emergency medical services providers to resuscitate patients with OOH cardiac arrest and to determine the relationship between chest compression rate and outcome. Methods and ResultsIncluded were patients aged 20 years with OOH cardiac arrest treated by emergency medical services providers participating in the Resuscitation Outcomes Consortium. Data were abstracted from monitordefibrillator recordings during cardiopulmonary resuscitation. Multiple logistic regression analysis assessed the association between chest compression rate and outcome. From December 2005 to May 2007, 3098 patients with OOH cardiac arrest were included in this study. Mean age was 67 16 years, and 8.6% survived to hospital discharge. Mean compression rate was 112 19/min. A curvilinear association between chest compression rate and return of spontaneous circulation was found in cubic spline models after multivariable adjustment (P 0.012). Return of spontaneous circulation rates peaked at a compression rate of 125/min and then declined. Chest compression rate was not significantly associated with survival to hospital discharge in multivariable categorical or cubic spline models. ConclusionsChest compression rate was associated with return of spontaneous circulation but not with survival to hospital discharge in OOH cardiac arrest. (Circulation. 2012;125:00-00.) Key Words: cardiac arrest cardiopulmonary resuscitation guidelines heart arrest outcomes research

AQ:2

P1-foo

AQ:3

AQ:4

he quality of cardiopulmonary resuscitation (CPR) and chest compressions is thought to be a major determinant of survival from cardiac arrest.1 Over the past 50 years, there has been a progressive increase in the recommended rate of chest compressions during CPR.2 4

Clinical Perspective on p Editorial see p

The current 2010 American Heart Association guidelines for CPR recommend using a chest compression rate of at least 100 compressions per minute.1 Whereas the 2010 European Resuscitation Council CPR guidelines5 recommend an upper

rate limit of 120 chest compressions per minute, the American Heart Association guidelines do not provide a similar recommendation because of a stated lack of evidence from human studies with return of spontaneous circulation (ROSC) or survival as outcomes. In addition, few studies have reported the distribution of actual chest compression rates used to resuscitate patients with out-of-hospital (OOH) cardiac arrest.6 8 Whenever compression rates of emergency medical services (EMS) providers have been measured in the OOH setting, most often the rates have been 100 compressions per minute, usually averaging between 100 and 120 compressions per minute.

AQ:9

Received July 30, 2011; accepted April 16, 2012. From the University of Texas Southwestern Medical Center, Dallas (A.H.I.); University of Washington, Seattle (D.G., S.B., J.P., G.N.); Medical College of Wisconsin, Milwaukee (T.P.A.); University of Toronto, Toronto, Ontario, Canada (L.J.M., B.L.B.); Texas Tech University, El Paso (P.N.); Oregon Health and Science University, Portland (M.D.); University of Iowa, Iowa City (D.L.A.); University of Pennsylvania, Philadelphia (R.B.); University of California, San Diego (D.D.); University of Ottawa, Ottawa, Ontario, Canada (I.S.); and National Heart, Lung, and Blood Institute, National Institutes of Health, Washington, DC (G.S.). Correspondence to Ahamed H. Idris, MD, Division of Emergency Medicine, University of Texas Southwestern Medical Center at Dallas, 5323 Harry Hines Blvd, Dallas, TX 75390-8579. E-mail [email protected] 2012 American Heart Association, Inc. Circulation is available at http://circ.ahajournals.org DOI: 10.1161/CIRCULATIONAHA.111.059535

1
<zjs;Original Article> <zjss;25> <zdoi;10.1161/CIRCULATIONAHA.111.059535>

balt2/zhc-ci/zhc-ci/zhc02412/zhc0782-12z xppws S 1 5/14/12 12:11 4/Color Figure(s): F1 Art: 059535 Input-len

Circulation

June 19, 2012

Figure 1. An example of an electronic recording from a monitor-debrillator showing the electric channel (black line), the bioimpedance channel (green line), and red arrows marking each chest compression.

The objectives of this North American multicenter observational study are as follows: (1) to describe the mean rate of chest compressions and range used by EMS providers, including firefighters and paramedics, to resuscitate patients with OOH cardiac arrest and (2) to determine the relationships between chest compression rate and ROSC and between rate and survival to hospital discharge.

Methods
Setting and Design
The Resuscitation Outcomes Consortium (ROC) is a network of regional research centers in the United States and Canada and a data coordinating center (DCC) in the United States that conducts research focused on cardiac arrest and severe traumatic injury. The ROC established a registry for OOH cardiac arrest in December 2005. An important component of the registry is the collection of defibrillator-monitor electronic files that record the quality of CPR given by EMS providers during treatment of cardiac arrest. This prospectively acquired, retrospectively analyzed, multicenter, EMS-based observational study was approved by institutional review boards of the University of Washington (location of the DCC) and the participating US and Canadian study sites including Ottawa, ON; Toronto, ON; and Vancouver, BC in Canada and Birmingham, AL; Dallas, TX; Des Moines, IA; Pittsburgh, PA; Milwaukee, WI; Portland, OR; Seattle/King County, WA; and San Diego, CA in the United States. Each board waived the requirement for informed consent for this study because it was considered to meet criteria for minimal risk. ROC was established to evaluate strategies for treatment of patients with cardiac arrest or life-threatening traumatic injury with

the primary focus on the OOH emergency setting. It includes 260 separate EMS agencies serving a population of 24 million.9 Since December 2005, the ROC Epidemiological Cardiac Arrest Registry (Epistry),10 a population-based EMS registry, has prospectively collected data on OOH cardiac arrest cases attended by participating EMS agencies. The data are collected across all ROC sites with the use of standardized data element forms and uniform definitions developed by ROC investigators. Data collected on each subject included information related to OOH treatments and outcomes, including initial cardiac rhythms, response times, descriptions of the types of professional responders on scene, timing of CPR and defibrillation, response to interventions, ROSC, survival to hospital discharge, and collection of digital, electronic recordings of rhythm and chest compressions. All data were collected by trained personnel who followed standardized procedures to ensure the validity and reproducibility of the data. Data were managed by the DCC and included error, logic, and cross-form checks. In addition, the DCC audited a proportion of cases to compare data entry with the original source documents to ensure uniformity and quality of data entry across sites. Two of 11 sites were excluded from this study because they contributed 2 cases.

Patient Population
Included were patients aged 20 years with completed case status who had OOH cardiac arrest treated by EMS providers participating in the ROC. Age was entered into the database in years, months, or days. If that information was unavailable, then age was indicated as one of the following categories: infant 1 year, child 1 to 11 years, adolescent 12 to 19 years, adult 20 to 39 years, middle age 40 to 60 years, older 61 to 75 years, and elderly 75 years. In addition, the cohort included in this analysis had electronic recordings of chest compressions available. Patients

balt2/zhc-ci/zhc-ci/zhc02412/zhc0782-12z xppws S 1 5/14/12 12:11 4/Color Figure(s): F1 Art: 059535 Input-len

Idris et al
with traumatic cardiac arrest or other obvious noncardiac causes of arrest were excluded.

Chest Compression Rate and Outcomes

Measurement
F1

Monitor-defibrillators recorded chest compression rates during CPR. The electronic recordings (Figure 1) were reviewed for accuracy, and data from the first 5 minutes of CPR are included in the analysis. The ROC DCC audited a proportion of original recordings from each site to ensure consistency of review and annotation methods. Selected recordings that had very high chest compression rates were reviewed by a panel with reviewers from other ROC sites and additionally by engineers from the defibrillator manufacturers. The presence and frequency of chest compressions were measured either indirectly by changes in thoracic impedance recorded from external defibrillation electrodes11 or directly via an accelerometer interface between the rescuer and the patients chest with the use of commercially available defibrillators. The electronic recordings were reviewed, annotated by trained personnel, and then analyzed with special software that automatically calculated average chest compression rates for each minute of CPR. Chest compression rate was defined as the rate at which chest compressions were performed during an uninterrupted series of chest compressions, in which interruptions in chest compressions (time without chest compressions) are defined as a pause 3 seconds (Medtronic devices [Minneapolis, MN]) or 2 seconds (Philips devices [Andover, MA] and ZOLL devices [Chelmsford, MA]). Chest compression rate was defined as the actual rate used during each set of chest compressions within a 1-minute interval independent of pauses within the 1-minute interval. Thus, the rate is the same whether chest compressions are given during the entire 1-minute interval or during only 10 seconds of the 1-minute interval. On the other hand, we defined delivered chest compressions as the actual number of chest compressions delivered during a 1-minute interval, which is illustrated by the following formula: Average chest compression rate multiplied by the chest compression fraction equals the number of delivered chest compressions per minute, where chest compression fraction is the proportion of time per minute that chest compressions are given. Interruptions in chest compression are inversely proportional to the chest compression fraction (eg, no interruptions in chest compression would result in a chest compression fraction of 1). Thus, the number of delivered chest compressions is dependent on both chest compression rate and chest compression fraction. The monitor-defibrillator software counted the actual number of chest compressions delivered each minute rather than estimating it with a formula.

Figure 2. Study cohort and exclusions. EMS indicates emergency medical services; CPR, cardiopulmonary resuscitation.

Outcomes
The prospectively selected primary outcome measure was survival to hospital discharge; ROSC was a secondary outcome.

Statistical Analysis
All statistical analyses were performed with commercially available statistical packages (SAS, version 9.1.3, Cary, NC; R, version 2.5.1, Vienna, Austria; Stata, version 11, College Station, TX). Descriptive statistics were calculated for average chest compression rate and within-subject variability of chest compression rate over the first 5 minutes of CPR for the overall sample population as well as for each site. ANOVA was used to determine whether there was a difference in average chest compression rate between the different sites. Summary results are presented as mean ( SD) or median (interquartile range). Those with available data were categorized into 3 groups of chest compression rates based on the average chest compression rate over the first 5 minutes: 80, 80 to 140, and 140 compressions per minute. These intervals were selected a priori because animal and human physiological studies suggested that blood flow was maximized in the reference range (80 140/min).1218 Delivered chest compressions were also categorized into 3 groups on the basis of the average number of chest compressions administered each minute over the first 5 minutes: 75, 75 to 100, and 100 compressions delivered each minute. These intervals were selected a priori because

human studies suggested that ROSC19 and survival7 were maximized in the reference range (75100/min). Potential confounding variables identified a priori included age, gender, bystander-witnessed cardiac arrest, EMS-witnessed cardiac arrest, attempted bystander CPR, public location, ROC site, first known EMS rhythm, and chest compression fraction. Logistic regression with robust SEs was used to calculate unadjusted and adjusted odds ratios of the association between chest compression rate and number of delivered chest compressions with both survival and ROSC. As a post hoc exploratory analysis, we fit an adjusted natural cubic spline curve, adjusted for the aforementioned confounding variables (y axis), to further characterize the nature of the relationship between chest compression rate and survival and chest compression rate and ROSC.20 The basis of the curve was a piecewise cubic polynomial with multiple knots. Four knots were chosen because this number produced a curve that appeared adequately smooth. The knots are located at the 5th, 35th, 65th, and 95th percentiles of average chest compression rate. For the adjusted cubic spline graphs, we used a global test that tested the null hypothesis that the spline curve is a horizontal line.

Results
A total of 26 902 OOH cardiac arrest cases were available; 15 876 received CPR, CPR process data were available for 3148 cases (19.8% of those treated), and 3098 cases (19.5% of treated patients) formed the analyzable cohort (Figure 2). The CPR process data were derived from electronic files downloaded from monitor-defibrillators on board emergency response vehicles. During this study period, December 2005 to May 2007, many ROC sites were in the process of developing the EMS infrastructure necessary to collect these files; this accounts for the relatively low proportion of treated cardiac arrest cases with CPR process data. Across all ROC sites, the distribution of defibrillators was as follows: 82% Medtronic, 10% Philips, 7% ZOLL, and 1% Laerdal and missing defibrillator assignment.
AQ:5

F2

balt2/zhc-ci/zhc-ci/zhc02412/zhc0782-12z xppws S 1 5/14/12 12:11 4/Color Figure(s): F1 Art: 059535 Input-len

Circulation

June 19, 2012

Table 2. Comparison of Analyzed Cohort With Patients Excluded From the Study
Analyzed Cohort (n 3098) (%) Age, mean (SD), y Male, No. (%) EMS-witnessed arrest, No. (%) Bystander-witnessed arrest, No. (%) Bystander CPR, No. (%) Public location, No. (%) Presenting rhythm, No. (%) VT/VF Pulseless electric activity 884 (29) 690 (22) 1381 (45) 100 (3) 43 (1) 1082 (35) 265 (9) 39 (32) 32 (26) 45 (37) 2 (2) 4 (3) 46 (38) 13 (11) 794 (29) 620 (22) 1234 (45) 81 (3) 34 (1) 972 (35) 242 (9) 51 (24) 38 (18) 102 (48) 17 (8) 5 (2) 64 (30) 10 (5) Asystole AED, no shock advised Cannot determine/ missing ROSC, No. (%) Survival to discharge, No. (%) 884 (29) 690 (22) 1383 (45) 100 (3) 43 (1) 1082 (35) 265 (8.6) 2889 (23) 2315 (19) 4819 (39) 1300 (10) 1105 (9) 3100 (25) 975 (7.8) 0.001 0.197 66.6 ( 16) 2071 (67) 154 (5) 1267 (41) 1201 (39) 534 (17) Excluded Patients (n 12 428) (%) 67.4 ( 16) 7771 (63) 1203 (10) 4729 (38) 3711 (30) 1875 (15) P ( 2 or t Test) 0.031 0.001 0.001 0.004 0.001 0.003 0.001

Table 1. Demographic Characteristics of Patients and Details of Cardiac Arrest

Chest Compression Rate Categories Patient Characteristics Age, mean (SD), y Male, No. (%) EMS-witnessed arrest, No. (%) Bystander-witnessed arrest, No. (%) Bystander CPR administered, No. (%) Public location, No. (%) Presenting rhythm, No. (%) VT/VF Pulseless electric activity Asystole AED, no shock advised Cannot determine/ missing ROSC, No. (%) Survival to discharge, No. (%) All Patients (n 3098) 66.6 (16) 2071 (67) 154 (5) 1267 (41) 1201 (39) 534 (17) 80 (n 122) 62.8 (17) 88 (72) 8 (7) 47 (39) 50 (41) 21 (17) 80 140 (n 2763) 66.9 (16) 1844 (67) 130 (5) 1142 (41) 1080 (39) 480 (17) 140 (n 213) 65.8 (17) 139 (65) 16 (8) 78 (37) 71 (33) 33 (15)

EMS indicates emergency medical services; CPR, cardiopulmonary resuscitation; VT, ventricular tachycardia; VF, ventricular fibrillation; AED, automatic external defibrillator; and ROSC, return of spontaneous circulation.

EMS indicates emergency medical services; CPR, cardiopulmonary resuscitation; VT, ventricular tachycardia; VF, ventricular fibrillation; AED, automatic external defibrillator; and ROSC, return of spontaneous circulation.
T1,AQ:6

T2

T3

Patient demographic information is summarized in Table 1. Mean age ( SD) was 67 16 years, 1082 had ROSC (35%), and 265 (8.6%) survived to hospital discharge; the distribution of patient characteristics and other key data between 3 chest compression rate categories are also presented in Table 1. As well, patient characteristics and key data for the analyzed cohort and patients excluded from the study are compared in Table 2. Important differences between the analyzed cohort and the excluded group include EMSwitnessed arrest (5% versus 10%), bystander CPR (39% versus 30%), ventricular tachycardia/ventricular fibrillation rhythm (29% versus 23%), and ROSC (35% versus 25%), respectively. Mean chest compression rate was 112 19 compressions per minute (range, 45202), and median was 111 (interquartile range, 100 123) during the first 5 minutes of CPR, with rates between sites significantly different (P 0.0001) (Table 3). Mean within-subject variability of compression rate from minute to minute was 7 8 compressions per minute (median, 5; interquartile range, 39; n 2974). In addition, chest compression rates at each of 5 chest compression fraction quintiles (0 20%, 20 40%, 40 60%, 60 80%, and 80 100%) were not significantly different (mean range, 111113; P 0.40). Furthermore, there was little correlation between chest compression fraction and chest compression rate (correlation coefficient 0.025; P 0.16).

The categorical model showed that patients receiving compression rates 140/min had an unadjusted odds ratio for survival of 0.51 (95% confidence interval [CI], 0.27 0.98; P 0.04) compared with the reference rate of 80 to 140 compressions per minute and an adjusted odds ratio of 0.61 (95% CI, 0.29 1.25; P 0.18) (Table 4). As well, patients receiving compression rates 140/min had an unadjusted odds ratio of 0.79 (95% CI, 0.58 1.07; P 0.13) for ROSC compared with the reference rate and an adjusted odds ratio of 1.01 (95% CI, 0.721.41; P 0.96) (Table 4). One third of cases had a rate of chest compressions 120/min, and 7% of cases had a rate 140/min. Bystander-witnessed cardiac arrest, EMS-witnessed cardiac arrest, first known EMS rhythm, age, and public location had odds ratios suggesting that these variables had important effects on ROSC and survival in the model, whereas attempted bystander CPR did not (Table 4). A sensitivity analysis that included chest compression fraction in the model did not change the estimated relationship between rate and survival. An adjusted natural cubic spline curve described graphically the relationship between chest compression rate and ROSC over the range of chest compression rates (Figure 3). The curve shows that ROSC peaks at a chest compression rate of 125/min and then declines sharply (P 0.012). Another adjusted cubic spline curve (Figure 4) showed the relationship between chest compression rate and survival to hospital discharge (P 0.63). The curves were adjusted for bystanderwitnessed cardiac arrest, EMS-witnessed cardiac arrest, first known EMS rhythm, attempted bystander CPR, age, and public location (y axis).

T4

F3

F4

balt2/zhc-ci/zhc-ci/zhc02412/zhc0782-12z xppws S 1 5/14/12 12:11 4/Color Figure(s): F1 Art: 059535 Input-len

Idris et al

Chest Compression Rate and Outcomes

Table 3. Average Chest Compression Rate by Site (Indicated by Letter) for the First 5 Minutes of Cardiopulmonary Resuscitation
No. of Patients Overall average compression rate for the first 5 min Average compression rate by site for the first 5 min A B C D E F G H I 205 265 81 61 302 68 712 1169 235 106 113 112 100 115 113 111 112 113 22 20 23 17 22 23 16 17 22 60 59 58 62 53 53 45 56 60 91 100 99 91 101 96 102 102 98 105 110 111 98 114 114 110 113 112 119 122 124 105 128 128 121 123 128 187 200 186 159 202 175 167 175 192
T5

Mean 112

SD 19

Minimum 45

25th Percentile 100

Median 111

75th Percentile 123

Maximum 202

3098

The number of chest compressions actually delivered each minute combined the effects of chest compression rate and chest compression fraction. The mean number of delivered chest compressions was 74 23, and the median was 75

(interquartile range, 58 91) (Table 5). Delivered chest compressions was modeled categorically for 75, 75 to 100, and 100 compressions delivered each minute. Patients receiving 75 compressions each minute had adjusted odds ratios of

Table 4. Odds Ratios of Variables Known to Affect Outcome From Cardiac Arrest and Odds Ratio of Chest Compression Rate by Category
Chest Compression Rates, Categories Unadjusted model 0 80/min 80140/min 140/min Adjusted model* 080/min 80140/min 140/min Global test for chest compression rate categories Male sex Age (10-y increase) Bystander-witnessed arrest EMS-witnessed arrest Bystander CPR Public location Rhythm VT/VF Pulseless electric activity Asystole AED, no shock advised Cannot determine Reference 0.43 (0.340.54) 0.22 (0.170.27) 0.30 (0.170.52) 0.59 (0.311.12) 0.001 0.001 0.001 0.11 Reference 0.22 (0.150.32) 0.04 (0.020.07) 0.07 (0.010.52) 0.24 (0.070.85) 0.001 0.001 0.01 0.03 1.18 (0.781.79) Reference 1.01 (0.721.41) vs ROSC 0.72 (0.600.86) 0.96 (0.911.01) 2.05 (1.712.47) 2.04 (1.392.98) 0.93 (0.781.11) 1.25 (1.011.55) 0.96 0.73 0.001 0.1 0.001 0.001 0.41 0.04 0.79 1.32 (0.672.62) Reference 0.61 (0.291.25) vs survival 0.83 (0.601.17) 0.77 (0.710.84) 2.18 (1.543.09) 3.18 (1.686.02) 1.15 (0.851.56) 1.83 (1.342.50) 0.18 0.25 0.29 0.001 0.001 0.001 0.36 0.001 0.42 1.12 (0.771.62) Reference 0.79 (0.581.072) 0.13 0.57 1.24 (0.692.24) Reference 0.51 (0.270.98) 0.04 0.47 ROSC OR (95% CI) P Survival to Discharge OR (95% CI) P

The reference rate category is 80 140 compressions per minute. ROSC indicates return of spontaneous circulation; OR, odds ratio; CI, confidence interval; EMS, emergency medical services; CPR, cardiopulmonary resuscitation; VT, ventricular tachycardia; VF, ventricular fibrillation; and AED, automated external defibrillator. *Model includes sex, age, bystander-witnessed arrest, EMS-witnessed arrest, first known EMS rhythm, attempted bystander CPR, public location, and site location. The numbers of patients in each category of compressions per minute are as follows: 80/min, n 122 (3.9%); 80 140/min, n 2763 (89.2%); 140/min, n 213 (6.9%).

balt2/zhc-ci/zhc-ci/zhc02412/zhc0782-12z xppws S 1 5/14/12 12:11 4/Color Figure(s): F1 Art: 059535 Input-len

Circulation

June 19, 2012

Table 5. Average Number of Delivered Chest Compressions per Minute by Site (Indicated by Letter) for the First 5 Minutes of Cardiopulmonary Resuscitation
No. of Patients Mean SD Minimum Maximum Overall average No. of delivered chest compressions per minute for the first 5 min Average No. of delivered chest compressions by site for the first 5 min A B C D E F G H I 205 265 81 61 302 68 712 1169 235 65 64 64 65 68 64 85 76 72 22 21 24 21 21 28 20 23 23 12 9 9 9 10 11 19 6 14 136 127 113 116 126 162 167 143 135 3098 74 23 6 167

Figure 3. Adjusted cubic spline of the relationship between chest compression rates and the probability of return of spontaneous circulation (ROSC). The adjusted model includes sex, age, bystander-witnessed arrest, emergency medical services witnessed arrest, rst known emergency medical services rhythm, attempted bystander cardiopulmonary resuscitation, public location, and site location (y axis). Probability of ROSC vs average chest compression rate when other covariates are equal to the population average is shown. We used a global test that tested the null hypothesis that the spline curve is a horizontal line (P 0.012). A histogram of the compression rates and numbers of patients is included. Dashed lines show 95% condence intervals.

T6

T7

0.81 (95% CI, 0.68 0.98; P 0.03) for ROSC and 0.78 (95% CI, 0.58 1.06; P 0.11) for survival compared with the reference range of 75 to 100 compressions (Table 6). The global test for delivered chest compressions versus ROSC and versus survival showed values of P 0.01 and P 0.25, respectively. Proportionately more patients receiving 75 compressions delivered each minute also had a chest compression rate 80/min (Table 7).

Chest compression depth was measured in 362 of 3098 patients (11.7%). There was a significant relationship between chest compression rate and depth (P 0.03) (Table 8).21 Depth declined with increasing chest compression rates. However, a sensitivity analysis that included depth in the model did not change the estimated relationship between rate and ROSC or between rate and survival. This large, observational, multicenter study showed that the likelihood of ROSC peaks at a chest compression rate
Table 6. Odds Ratios of Delivered Chest Compressions per Minute by Category
Delivered Chest Compressions, Categories Unadjusted model 0 75/min 75100/min 100/min Adjusted model* 075/min 75100/min 100/min Global test for delivered chest compressions 0.81 (0.680.98) Reference 1.15 (0.891.49) vs ROSC 0.28 0.01 0.03 0.78 (0.581.06) Reference 0.77 (0.471.28) vs survival 0.32 0.25 0.11 0.78 (0.660.92) Reference 0.99 (0.791.26) 0.96 0.003 0.82 (0.631.07) Reference 0.64 (0.420.99) 0.05 0.14 ROSC OR (95% CI) Survival to Discharge OR (95% CI)

T8

Discussion

Figure 4. Adjusted cubic spline of the relationship between chest compression rates and the probability of survival to hospital discharge. The adjusted model includes sex, age, bystanderwitnessed arrest, emergency medical serviceswitnessed arrest, rst known emergency medical services rhythm, attempted bystander cardiopulmonary resuscitation, public location, and site location (y axis). Probability of survival vs average chest compression rate when other covariates are equal to the population average is shown. We used a global test that tested the null hypothesis that the spline curve is a horizontal line (P 0.63). A histogram of the compression rates and numbers of patients is included. Dashed lines show 95% condence intervals.

ROSC indicates return of spontaneous circulation; OR, odds ratio; and CI, confidence interval. *Model includes sex, age, bystander-witnessed arrest, emergency medical serviceswitnessed arrest, first known emergency medical services rhythm, attempted bystander cardiopulmonary resuscitation, public location, and site location. The number of patients in each category of delivered compression per minute are as follows: 75/min, n 1555 (50.2%); 75100/min, n 1131 (36.5%); 100/min, n 412 (13.3%).

balt2/zhc-ci/zhc-ci/zhc02412/zhc0782-12z xppws S 1 5/14/12 12:11 4/Color Figure(s): F1 Art: 059535 Input-len

Idris et al
Table 7. Average Chest Compression Rates vs No. of Delivered Chest Compressions
Average Delivered Compressions per Minute 75 75100 100 Total P 0.001,
2

Chest Compression Rate and Outcomes

Average Compression Rate per Minute, No. of Patients (%) 80 122 (100) 0 0 122 (100) 80 140 (50) 1056 (38) 327 (12) 2763 (100) 140 53 (25) 75 (35) 85 (40) 213 (100) Total 1555 (50) 1131 (37) 412 (13) 3098 (100)

when chest compressions were given 60% to 80% of the time.7 Although analysis of the number of delivered chest compressions per minute provides important insight into overall quality of CPR, this measurement is not displayed on most defibrillators, and therefore use of chest compression rate for feedback and monitoring is more applicable to clinical practice. The use of metronomes as well as real-time feedback during CPR has been shown to be useful and effective in helping rescuers to maintain a desired chest compression rate.8,18

of 125/min during the first 5 minutes of OOH CPR. However, we were unable to confirm a significant relationship between chest compression rate and survival to hospital discharge. The study also demonstrates that EMS rescuers frequently apply rates 100/min, with rates 120/min occurring in one third of cases. Faster than recommended chest compression rates have been reported previously and likely occur frequently.6,8 Animal studies showed that chest compression rates of 120/min were associated with improved blood flow and survival, whereas rates faster than that were associated with decreased blood flow.1214 Similarly, human studies also showed improved blood flow and end-tidal CO2 levels (a surrogate for blood flow) with a chest compression rate of 120/min.17,18,22 In contrast to rate studies, a CPR study of humans with in-hospital cardiac arrest measured and reported the actual number of compressions delivered each minute, which is a product of chest compression rate and chest compression fraction (the proportion of each minute spent doing chest compressions).19 This study found that patients who received 90 delivered chest compressions each minute had a significantly greater rate of ROSC compared with those who received only 79 compressions.19 The present study found that patients receiving 75 chest compressions each minute had a decreased likelihood of ROSC, possibly because it reflected lower chest compression rates and more interruptions in chest compressions (Table 7). Interruptions in chest compressions are commonly observed, even in a high-performance EMS system,6 and an average chest compression fraction of 0.70 to 0.80 may be the best that can be achieved during conventional CPR. A study of OOH CPR found that the proportion of time that chest compression is given during each minute is associated with survival to hospital discharge and that survival was greatest
Table 8. Average Chest Compression Depth vs Chest Compression Rate per Minute
Average Compression Depth, mm 38 38 51 51 Total Average Compression Rate, No. of Patients (%) 80 8 (42) 6 (32) 5 (26) 19 (100) 81140 171 (52) 122 (38) 33 (10) 326 (100) 140 14 (82) 3 (18) 0 (0) 17 (5) Total 193 (53) 131 (36) 38 (11) 362 (100)

Limitations
Certain factors regarding the quality of CPR that are known to affect ROSC and survival were not available for analysis in the present study, including chest compression depth and incomplete recoil (leaning). The amount of leaning was not abstracted in our database, and chest compression depth could not be analyzed completely because of insufficient numbers of electronic files captured from devices capable of recording compression depth. However, in the subset of patients in whom depth was measured, depth was found to decrease significantly with increasing chest compression rates (Table 8).21 Measurement of depth was available for only 12% of cases in the analyzed group, which is an important limitation that could affect results in this study. The percentage of eligible patients who had electronic CPR process files available was only 20% of all treated cases, which may produce a selection bias. We compared the cohort in the analysis with those who had CPR but were not included and found differences in proportions of those whose cardiac arrest was witnessed, those who received bystander CPR, or those whose cardiac arrest occurred in a public location and differences in presenting rhythm and ROSC rate but not survival (Table 2). A larger percentage of patients in the analyzed cohort had ROSC; however, we did not find a significant relationship between ROSC and chest compression rate in the categorical multivariate analysis. If the association between compression rate and outcome is different in the excluded cases than in those included in the analysis, the bias produced by this exclusion would be large. Furthermore, a mandatory condition for participation in the ROC Epistry was that agencies had to use devices capable of recording chest compressions and interruptions in chest compressions. It is possible that non-ROC communities in North America using devices incapable of recording chest compressions could have different results related to differences in socioeconomic status. Although our study suggests a positive association of ROSC with chest compression rates within a relatively defined range, a larger study with fewer excluded patients would help to confirm these findings. Certain monitor-defibrillators are capable of providing feedback with respect to chest compression rate and depth, and 7 of 9 ROC sites had such monitors. The feedback feature was turned on during CPR in 12% of all patients included in this study. A trial of real-time feedback in a similar group of patients showed that average chest compression rates were 104/min with and 109/min without

P 0.03, Fisher exact test (for association).

balt2/zhc-ci/zhc-ci/zhc02412/zhc0782-12z xppws S 1 5/14/12 12:11 4/Color Figure(s): F1 Art: 059535 Input-len

Circulation

June 19, 2012

3. Cardiopulmonary resuscitation: statement by the Ad Hoc Committee on Cardiopulmonary Resuscitation of the Division of Medical Sciences, National Academy of SciencesNational Research Council. JAMA. 1966; 198:138 145. 4. Standards and guidelines for cardiopulmonary resuscitation (CPR) and emergency cardiac care (ECC). JAMA. 1986;255:29052984. 5. Koster RW, Baubin MA, Bossaert LL, Caballero A, Cassa P, Castren M, Granja C, Handley AJ, Monsieurs KG, Perkins GD, Raffay V, Sandroni C. European Resuscitation Council guidelines for resuscitation 2010, section 2: adult basic life support and use of automated external defibrillators. Resuscitation. 2010;81:12771292. 6. Wik L, Kramer-Johansen J, Myklebust H, Sorebo H, Svensson L, Fellows B, Steen PA. Quality of cardiopulmonary resuscitation during out-ofhospital cardiac arrest. JAMA. 2005;293:299 304. 7. Christenson J, Andrusiek D, Everson-Stewart S, Kudenchuk P, Hostler D, Powell J, Callaway CW, Bishop D, Vaillancourt C, Aufderheide TP, Idris A, Stoufer JA, Stiell I, Berg R; Resuscitation Outcomes Consortium Investigators. Chest compression fraction determines survival in patients with out-of-hospital ventricular fibrillation. Circulation. 2009;120: 12411247. 8. Fletcher D, Galloway R, Chamberlain D, Pateman J, Bryant G, Newcombe RG. Basics in advanced life support: a role for download audit and metronomes. Resuscitation. 2008;78:127134. 9. Davis DP, Garberson LA, Andrusiek DL, Hostler D, Daya M, Pirrallo R, Craig A, Stephens S, Larsen J, Drum AF, Fowler R. A descriptive analysis of emergency medical service systems participating in the Resuscitation Outcomes Consortium (ROC) network. Prehosp Emerg Care. 2007;11:369 382. 10. Morrison LJ, Nichol G, Rea TD, Christenson J, Callaway CW, Stephens S, Pirrallo RG, Atkins DL, Davis DP, Idris AH, Newgard C. Rationale, development and implementation of the Resuscitation Outcomes Consortium Epistry-Cardiac Arrest. Resuscitation. 2008;78: 161169. 11. Stecher FS, Olsen JA, Stickney RE, Wik L. Transthoracic impedance used to evaluate performance of cardiopulmonary resuscitation during out of hospital cardiac arrest. Resuscitation. 2008;79:432 437. 12. Harris LC, Kirimli B, Safar P. Ventilation-cardiac compression rates and ratios in cardiopulmonary resuscitation. Anesthesiology. 1967;2: 806 813. 13. Maier GW, Tyson GS Jr, Olsen CO, Kernstein KH, Davis JW, Conn EH, Sabiston DC Jr, Rankin JS. The physiology of external cardiac massage: high-impulse cardiopulmonary resuscitation. Circulation. 1984;70: 86 101. 14. Feneley MP, Maier GW, Kern KB, Gaynor JW, Gall SA Jr, Sanders AB, Raessler K, Muhlbaier LH, Rankin JS, Ewy GA. Influence of compression rate on initial success of resuscitation and 24-hour survival after prolonged manual cardiopulmonary resuscitation in dogs. Circulation. 1988;77:240 250. 15. Fitzgerald KR, Babbs CF, Frissora HA, Davis RW, Silver DI. Cardiac output during cardiopulmonary resuscitation at various compression rates and durations. Am J Physiol. 1981;241:H442H448. 16. Wolfe JA, Maier GW, Newton JR, Glower DD, Tyson GS Jr, Spratt JA, Rankin JS, Olsen CO. Physiologic determinants of coronary blood flow during external cardiac massage. J Thorac Cardiovasc Surg. 1988; 95:523. 17. Swenson RD, Weaver WD, Niskanen RA, Martin J, Dahlberg S. Hemodynamics in humans during conventional and experimental methods of cardiopulmonary resuscitation. Circulation. 1988;78:630 639. 18. Kern KB, Sanders AB, Raife J, Milander MM, Otto CW, Ewy GA. A study of chest compression rates during cardiopulmonary resuscitation in humans: the importance of rate-directed chest compressions. Arch Intern Med. 1992;15:145149. 19. Abella BS, Sandbo N, Vassilatos P, Alvarado JP, OHearn N, Wigder HN, Hoffman P, Tynus K, Vanden Hoek TL, Becker LB. Chest compression rates during cardiopulmonary resuscitation are suboptimal: a prospective study during in-hospital cardiac arrest. Circulation. 2005; 111:428 434. 20. Chambers JM, Hastie T. Statistical Models in S. Pacific Grove, CA: Wadsworth & Brooks/Cole Advanced Books & Software; 1992. 21. Stiell IG, Brown SP, Christenson J, Cheskes S, Nichol G, Powell J, Bigham B, Morrison LJ, Larsen J, Hess E, Vaillancourt C, Davis DP, Callaway CW; Resuscitation Outcomes Consortium (ROC) Investigators. What is the role of chest compression depth during out-of-hospital cardiac arrest resuscitation? Crit Care Med. 2012;40:11921198. AQ:8

feedback turned on, a difference that is probably not clinically important.23 Another limitation was that only the first 5 minutes of CPR were analyzed, but some patients had substantially longer periods of CPR, which could be an effect-modifying variable. A prior study, however, showed that chest compressions during the first 5 minutes of CPR were similar to those in subsequent minutes of CPR.24

Conclusions
The likelihood of ROSC during CPR for OOH cardiac arrest was greatest with use of a chest compression rate of 125 compressions per minute and then declined at higher rates. The association between chest compression rate and survival was not significant after adjustment for confounding variables.

Acknowledgments
We are indebted to the firefighters and paramedics participating in the ROC for their hard work and dedication. We would also like to thank the data coordinators at each ROC site for their extraordinary diligence and focus in abstracting the data for this study.

Sources of Funding
This study was supported by a series of cooperative agreements to 10 regional clinical centers and 1 DCC (5U01 HL077863, University of Washington DCC; HL077865, University of Iowa; HL077866, Medical College of Wisconsin; HL077867, University of Washington; HL077871, University of Pittsburgh; HL077872, St. Michaels Hospital; HL077873, Oregon Health and Science University; HL077881, University of Alabama at Birmingham; HL077885, Ottawa Health Research Institute; HL077887, University of Texas Southwestern Medical Center at Dallas; HL077908, University of California at San Diego) from the National Heart, Lung, and Blood Institute in partnership with the National Institute of Neurological Disorders and Stroke, American Heart Association, US Army Medical Research and Material Command, Canadian Institutes of Health ResearchInstitute of Circulatory and Respiratory Health, Heart and Stroke Foundation of Canada, and Defense Research and Development Canada.

Disclosures
Drs Idris, Aufderheide, Brown, Daya, Morrison, Davis, Stiell, and Nichol, as well as Danielle Guffey and Judy Powell, receive ROC grant funding. Dr Aufderheide receives grant funding from the National Heart, Lung, and Blood Institute and National Institute of Neurological Disorders and Stroke; he is a consultant for Jolife and Medtronics and is a board member for Take Heart America and for the Citizen CPR Foundation. Drs Idris, Aufderheide, Daya, Morrison, Atkins, and Nichol are volunteers for the American Heart Association National Emergency Cardiovascular Care Committee. Drs Daya and Idris are unpaid consultants for Philips Medical Systems. Dr Nichol has research grant funding from the Asmund S. Laerdal Foundation for Acute Medicine and Medtronic, Inc.

AQ:7

References
1. Berg RA, Hemphill R, Abella BS, Aufderheide TP, Cave DM, Hazinski MF, Lerner EB, Rea TD, Sayre MR, Swor RA. Part 5: adult basic life support: 2010 American Heart Association guidelines for cardiopulmonary resuscitation and emergency cardiovascular care. Circulation. 2010; 122:S685S705. 2. Kouwenhoven WB, Jude JR, Knickerbocker GG. Closed-chest cardiac massage. JAMA. 1960;173:1064 1067.

balt2/zhc-ci/zhc-ci/zhc02412/zhc0782-12z xppws S 1 5/14/12 12:11 4/Color Figure(s): F1 Art: 059535 Input-len

Idris et al
22. Idris AH, Staples E, OBrien D, Melker RJ, Rush W, Del Duca KJ, Falk J. End-tidal carbon dioxide during extremely low cardiac output. Ann Emerg Med. 1994;23:568 572. 23. Hostler D, Everson-Stewart S, Rea TD, Stiell G, Callaway CW, Kudenchuk PJ, Sears GK, Emerson SS, Nichol G; Resuscitation Outcomes Consortium Investigators. Effect of real-time feedback during

Chest Compression Rate and Outcomes

cardiopulmonary resuscitation outside hospital: prospective, clusterrandomised trial. BMJ. 2011;342:d512. 24. Valenzuela TD, Kern KB, Clark LL, Berg RA, Berg MD, Berg DD, Hilwig RW, Otto CW, Newburn D, Ewy GA. Interruptions of chest compressions during emergency medical systems resuscitation. Circulation. 2005;112:1259 1265.

CLINICAL PERSPECTIVE
Guidelines for cardiopulmonary resuscitation recommend a chest compression rate of at least 100 compressions per minute but do not provide an upper limit for rate because of lack of evidence from human studies with return of spontaneous circulation or survival as outcomes. In this study, we describe chest compression rates used by emergency medical services providers to resuscitate patients with out-of-hospital cardiac arrest, and we sought to determine the relationship between chest compression rate and return of spontaneous circulation and survival. This study used data from the Resuscitation Outcomes Consortium Epidemiological Cardiac Arrest Registry database. The study included 3000 patients who had out-of-hospital cardiac arrest and cardiopulmonary resuscitation. The investigators found that 75% of rescuers used chest compression rates 100/min, and one third used rates 120/min. Depth of chest compressions is an important factor related to cardiopulmonary resuscitation quality and survival. We found that chest compression depth decreases when chest compression rate is 140/min. This study found that the likelihood of return of spontaneous circulation from out-of-hospital cardiac arrest was greatest with use of a chest compression rate of 125 compressions per minute and then declines sharply with faster rates. By using chest compression rates of at least 100/min but no faster than 125/min, we can further improve return of spontaneous circulation.

JOBNAME: AUTHOR QUERIES PAGE: 1 SESS: 3 OUTPUT: Mon May 14 12:12:13 2012 /balt2/zhc-ci/zhc-ci/zhc02412/zhc0782-12z

AUTHOR QUERIES
AUTHOR PLEASE ANSWER ALL QUERIES
1Please turn to page 3 of your proof and review the running head, which will appear in the upper right-hand margins of odd-numbered pages. Running heads must be 50 or fewer characters in length, including spaces and punctuation. If your original short title was longer than 50 characters, we may have shortened it. Please modify if necessary (but observe our length guidelines). 2Please confirm that all authors are included in the correct order in the byline and that all names are spelled correctly, including special characters, accents, middle initials, and degrees, if applicable. Note that journal style discourages listing American honorary degrees in the byline; such degrees are deleted during editing. 3Key words have been edited to match the US National Library of Medicines Medical Subject Headings (http://www.nlm.nih.gov/MBrowser.html). If they need modification, please refer to this site and limit the total number of key words to 7. 4 If you have color in your proof, please indicate whether you approve the color charge by completing and returning the color form attached to your proof e-mail. Please return the color form at the same time as your corrections (if you have not already returned it at an earlier date). The rate is $653 per 1 printed color page. If you have any questions, please contact [email protected]. 5First sentence of Results: Total number of cases is shown here as 26,902, but in Figure 2, total number of cases is shown as 26,904. Please advise. Also, for the 3098 cases that formed the analyzable cohort, the percentage is shown as 98.4% (of those with CPR process data available) in Figure 2 but as 19.5% (of those treated) in the text. 6 Please review the typeset tables carefully against a copy of the originals to verify accuracy of editing and typesetting. 7Please carefully review any Acknowledgments, Sources of Funding, and/or Disclosures listed at the end of the manuscript (before the References) and confirm that they are accurate and complete for all authors. 8 Reference 21 was updated. 9 Please confirm that all authors institutional affiliations (including city/state/country locations) 1

JOBNAME: AUTHOR QUERIES PAGE: 2 SESS: 3 OUTPUT: Mon May 14 12:12:13 2012 /balt2/zhc-ci/zhc-ci/zhc02412/zhc0782-12z

AUTHOR QUERIES
AUTHOR PLEASE ANSWER ALL QUERIES
are correct as shown in the affiliations footnote. 2