Professional Documents
Culture Documents
PII: S0300-9572(15)00248-8
DOI: http://dx.doi.org/doi:10.1016/j.resuscitation.2015.06.004
Reference: RESUS 6440
Please cite this article as: Yannopoulos D, Aufderheide TP, Abella BS, Duval S,
Frascone RJ, Goodloe JM, Mahoney BD, Nadkarni VM, Halperin HR, O’Connor R,
Idris AH, Becker LB, Pepe PE, Quality of CPR: An Important Effect Modifier in Cardiac
Arrest Clinical Outcomes and Intervention Effectiveness Trials, Resuscitation (2015),
http://dx.doi.org/10.1016/j.resuscitation.2015.06.004
This is a PDF file of an unedited manuscript that has been accepted for publication.
As a service to our customers we are providing this early version of the manuscript.
The manuscript will undergo copyediting, typesetting, and review of the resulting proof
before it is published in its final form. Please note that during the production process
errors may be discovered which could affect the content, and all legal disclaimers that
apply to the journal pertain.
1 Quality of CPR: An Important Effect Modifier in Cardiac Arrest Clinical
3
4
5 Demetris Yannopoulos, MD (Corresponding author)
t
6 University of Minnesota, 420 Delaware Street SE, Minneapolis, MN 55455
ip
7 Phone: 612-626-1382
8 Email: yanno001@umn.edu
cr
9
10 Tom P. Aufderheide, MD, MS, FACEP, FACC, FAHA
11 Medical College of Wisconsin, 9200 West Wisconsin, Milwaukee, WI 53226
us
12
13 Benjamin S. Abella, MD, FACEP.
14 University of Pennsylvania, 3400 Spruce Street, Philadelphia, PA 19104
15
16
17
Sue Duval, PhD, FAHA
an
University of Minnesota, 420 Delaware Street SE, Minneapolis, MN 55455
18
M
19 Ralph J. Frascone, MD, FACEP.
20 University of Minnesota, 420 Delaware Street SE, Minneapolis, MN 55455
21
ed
32
33 Henry R. Halperin, MD, MA, FAHA, FHRS
34 Johns Hopkins Hospital, 600 North Wolfe Street, Baltimore, MD 21287
35
36 Robert O’Connor, MD, MPH, FACEP.
37 University of Virginia, P.O. Box 800699, Charlottesville, VA 22908
38
39 Ahamed H. Idris, MD, FACEP
40 University of Texas Southwestern Medical Center, 5323 Harry Hines Boulevard, Dallas,
41 TX 75390
Page 1 of 34
42
43 Lance B. Becker, MD, FACEP
44 University of Pennsylvania, 3400 Spruce Street, Philadelphia, PA 19104
45
46 Paul E. Pepe, MD. MPH, FACEP, MACP, FCCM.
47 University of Texas Southwestern Medical Center, 5323 Harry Hines Boulevard, Dallas,
48 TX 75390-8579
t
49
ip
50
51
52 ADDRESS FOR CORRESPONDENCE and REPRINTS:
cr
53
54 Demetris Yannopoulos, MD (Corresponding author)
us
55 University of Minnesota, 420 Delaware Street SE, Minneapolis, MN 55455
56 Phone: 612-626-1382
57 Email: yanno001@umn.edu
58
59
60
an
61 LOCATION WHERE WORK WAS CONDUCTED:
M
62
63 Multi-center North American network of Resuscitation Research Centers study sites
64 comprising the National Institutes of Health (NIH) Resuscitation Outcomes Consortium
65 (ROC) -- https://roc.uwctc.org/tiki/tiki-index.php?page=roc-public-home
ed
66
67
68
FUNDING SOURCES:
pt
69
70
71 No funding was used for the current study. The ROC Study database was funded and
ce
76
77
78 Text Word Count: 3571 (preferred range 2000 to 4000 words)
79
80
81
Page 2 of 34
82 Abstract
83 Objectives: To determine if the quality of CPR had a significant interaction with the
t
85 Design: The public access database from the NIH PRIMED trial was accessed to
ip
86 determine if there was an interaction between quality of CPR performance,
cr
87 intervention, and outcome (survival to hospital discharge with modified Rankin Score
us
88 (mRS) ≤3).
96 Measurements and Main Results: Overall, 848 and 827 patients, respectively, in the
Ac
97 active and sham-ITD groups had “acceptable” CPR quality performed (n=1,675). There
98 was a significant interaction between the active and sham-ITD and compression rate,
99 depth and fraction as well as their combinations. The strongest interaction was seen
100 with all three parameters combined (unadjusted and adjusted interaction p-value,
101 <0·001). For all presenting rhythms, when “acceptable” quality of CPR was performed,
Page 3 of 34
102 use of an active-ITD increased survival to hospital discharge with mRS ≤3 compared to
103 sham (61/848 [7·2%] versus 34/827 [4·1%], respectively; p=0·006). The opposite was
104 true for patients that did not receive “acceptable” quality of CPR. In those patients, use
105 of an active –ITD led to significantly worse survival to hospital discharge with mRS≤3
t
ip
106 compared to sham (34/1012 [3.4%] versus 62/1061 [5.8%], p=0.007).
cr
107 Conclusions: There was a statistically significant interaction between the quality of CPR
108 provided, intervention, and survival to hospital discharge with mRS≤3 in the NIH
us
109 PRIMED trial. Quality of CPR delivered can be an underestimated effect modifier in CPR
115
ce
116
Ac
Page 4 of 34
116 INTRODUCTION
117 Interventions that create circulation and restore a stable heart rhythm during the first
118 minutes of cardiopulmonary resuscitation (CPR) are critical for achieving survival after
t
out-of-hospital cardiac arrest (OHCA), a leading cause of death in adults.1–4 The
ip
119
120 American Heart Association (AHA) guidelines for CPR have recommended prompt
cr
121 delivery of chest compressions to limit the time of no or low flow and now have
us
122 underscored the importance of recognizing, delivering and even measuring quality
123 compression parameters to alter the outcomes from OHCA positively .1–5
124
an
The predominate determinants of blood flow generation during CPR that have
125 been identified are chest compression rate, depth, and fraction (percentage of CPR time
M
126 when chest compressions are being delivered without interruption).6–11 Accordingly, as
ed
127 chest compressions (and the chest compression-decompression cycle) are performed
128 primarily to generate blood flow, survival outcomes would likely depend upon the
pt
130 Other than timely defibrillation, there has been limited progress in confirming
ce
131 the effectiveness of other interventions that can improve outcomes beyond
Ac
132 conventional closed chest CPR in the clinical setting.1,4 Despite robust laboratory
133 findings and positive results in smaller-scaled human trials, most highly-promising
134 cardiac arrest therapies have not been demonstrated to have clinical benefit when
Page 5 of 34
136 The National Institutes of Health (NIH) sponsored Resuscitation Outcomes
137 Consortium (ROC) PRIMED trial was an example of such a trial.13,14 The PRIMED trial
139 standard, manual CPR in a multi-center, randomized clinical trial (RCT).14 Despite
t
ip
140 positive outcomes in preclinical and smaller clinical trials, however, no differences in
141 survival to hospital discharge with favorable neurological function were identified when
cr
142 patients were randomized to receive active versus sham ITD devices in this large,
us
143 effectiveness RCT.14
144 Since the design, implementation and completion of the PRIMED trial, evolving
145
an
data and broader concerns regarding the interaction of CPR quality and outcomes have
M
146 raised the issue of sub-optimal CPR performance as being a potential unrecognized
147 effect modifier in resuscitation research and patient outcomes.5 In turn, this concern
ed
148 raised the speculation that the effect of the ITD may have been masked by the quality of
149 CPR performed. Few data exist to confirm the potential modifier effect of CPR quality on
pt
151 Based upon the fundamental physiology of closed chest CPR and the ITD itself,
152 the participating authors here, including independent investigators not involved in the
Ac
153 PRIMED trial, hypothesized that there would be a significant and clinically important
154 interaction between outcome and the respective individual CPR quality parameters of
155 chest compression rate, depth, and fraction, as well as the overall combination of all
Page 6 of 34
157 Although, traditionally, secondary analyses of study data have been viewed with
158 skepticism and often dismissed without consideration, the value of this function, even
160 important in the realm of RCT data.15 For example, a recent study published in the
t
ip
161 Journal of the American Medical Association highlighted the notion that re-analysis of
162 RCT data may help the scientific community assess the validity of reported trial results,
cr
163 particularly when the re-analysis involved prospectively-collected blinded data (RCT) as
us
164 well as the addition of independent expert authors not involved in the original trial. In
165 several cases of such re-analyses, the conclusions and subsequent medical practices
168 prospectively-collected data, including CPR quality parameters from numerous centers,
ed
169 it was also an NIH-supervised multi-center trial with study groups and other
171 resuscitation trials with well-matched groups and implementation, effect modifiers,
ce
172 such as ventilatory parameters, well-controlled in the laboratory, may not be recognized
173 and controlled in the clinical setting.12,16,17 Thus, in clinical trials, such modifiers may
Ac
174 mask the effects of a study intervention on outcomes, making their recognition
176 To test the hypothesis that quality of CPR was an outcome modifier in the ROC
177 PRIMED Trial, investigators evaluated if there was a statistically significant interaction
178 between the three different components of chest compression quality (rate, depth, and
Page 7 of 34
179 fraction) and their combinations, the tested interventions, and the original primary
180 study outcome, survival to hospital discharge with a Modified Rankin Scale Score (mRS)
182
t
ip
183 MATERIALS AND METHODS
cr
184 Ethics Approval, Data Source and Investigators
us
185 The ROC PRIMED database was acquired from the NIH under the NIH Data Sharing
186 Policy, three years following publication.14 This de-identified database was re-analyzed
187
an
independently from the ROC by the authors of the current study, including
190 ROC PRIMED Trial. The authors had no identifiable conflict of interest with the current
ed
191 investigation. The study was exempted from review by the University of Minnesota
192
193
ce
195 Detailed rationale and methodologies for the ROC PRIMED trial have been published.12,14
Ac
196 The study was a prospective, randomized, double-blind, multi-center North American
197 effectiveness trial performed by EMS personnel under an FDA Investigational Device
198 Exception. Adult subjects found in OHCA were randomly assigned and treated with
199 either a sham (un-activated) or active ITD (-16 cm resistance; manufacturer, Advanced
Page 8 of 34
201 Sham and active devices were indistinguishable to rescuers. The first qualified
202 EMS provider placed the ITD between the ventilation bag and the patient’s airway
203 device. All other resuscitative measures followed individual EMS agency standard
204 operating procedures,13,14 Participating EMS agencies generally followed the concurrent
t
ip
205 2005 AHA Guidelines for CPR and Emergency Cardiovascular Care with occasional local
cr
207
us
208 Data Collection
209 All participating EMS agencies implemented an electronic system for monitoring
210
an
individual components of CPR during the trial. CPR quality parameters were
M
211 electronically recorded in subjects immediately following application of the defibrillation
212 electrodes using thoracic impedance recorded from external defibrillation electrodes or
ed
213 via an accelerometer interface between the rescuer’s hands and the patient’s chest
214 using commercially available defibrillators.14 Electronic capture of the quality of CPR
pt
215 performed included chest compression rate, depth, and fraction. These data were used
ce
216 in a run-in phase to determine when sites qualified to start enrollment in the pivotal
217 phase of the study. Although CPR quality was also monitored during the pivotal phase of
Ac
218 the study, sites were not required to discontinue patient entry if they did not meet CPR
219 quality benchmarks. Quality of CPR data linked to individual cardiac arrest subject
220 records from the pivotal phase of the ROC study were available to the current authors
222
Page 9 of 34
223 Parameters of the Current Study
224 Following acquisition of the PRIMED database, key study results were evaluated
225 to determine the validity of the database. Patient counts and outcomes in the sham and
226 active ITD groups were assessed for rates of return of spontaneous circulation (ROSC) on
t
ip
227 arrival to the emergency department, survival to hospital admission, and survival to
228 hospital discharge, as well as the primary outcome, survival to hospital discharge with
cr
229 mRS ≤3. 14
us
230 In the present study, the acceptable rate of chest compressions was defined
232
an
AHA guidelines at the time of the study.3 Therefore, 100 ccpm 20% (80-120) was used
M
233 to categorize an “acceptable” range for rate. The reported rate was averaged over the
234 first five recorded minutes of CPR as previously defined in the NIH database dictionary.
ed
236 compression depth was defined as 5 cm 20% (4-6 cm or approximately 1·6 to2·4
pt
238 CPR chest compression fraction (CCF) was calculated in the usual manner as:
239 [total seconds with active compressions][total seconds with interpretable signal and no
Ac
240 evidence of spontaneous circulation]. The “acceptable” CPR CCF was defined as ≥0·5
241 averaged over five minutes, as previously specified in the ROC PRIMED protocol.1,14 The
242 AHA guidelines at the time did not refer to chest compression fraction but encouraged
10
Page 10 of 34
244 Ventilation rate, although initially considered in the PRIMED study as a metric to
245 be collected, was, eventually, not collected and there were no data on chest wall recoil
247 The first five minutes of electronically recorded CPR were evaluated in the ROC
t
ip
248 PRIMED study because it has been demonstrated to reliably correlate with the quality of
249 CPR for the entire resuscitation episode.18 The value for each CPR parameter was
cr
250 determined for each minute of the first five minutes of electronically recorded CPR and
us
251 an average for that five minutes ascertained. Thus, CPR was prospectively defined as
253
an
average compression rate of 100±20 ccpm, depth of 51 cm, and a compression fraction
M
254 ≥50% over the first five minutes of the recordings.
255
ed
256 Endpoints
257 The primary endpoint of our study was to determine the interaction between the three
pt
258 individual components used to assess the quality of CPR provided (rate, depth and
ce
259 fraction) and their combinations, the intervention (sham or active ITD); and survival to
261
263 Variables included in this study were determined before beginning the analysis and
264 were limited by the data collected in the parent ROC PRIMED trial. Continuous variables
265 are reported as mean and standard deviation or median with 25th and 75th percentiles, and
11
Page 11 of 34
266 categorical variables as frequency and percentage. The difference between proportions
267 experiencing the primary endpoint was calculated and compared between intervention
t
ip
270 accordance with previous practice.19 To test for interaction between intervention group
271 and CPR quality measures in their association with the primary endpoint (survival to
cr
272 discharge with mRS ≤3), separate regression models with fixed effects for treatment
us
273 group, individual CPR quality parameters or their combination, and treatment by quality
274 interaction, were fitted. We further fit a model with all three parameters and their
275
an
combinations, and their interactions with treatment, to further investigate the most
M
276 critical parameter combinations. The most parsimonious model containing significant
277 interactions with treatment was then derived. Tests for interaction were deemed
ed
278 statistically significant when the p-value was <0.05. All analyses were performed using
279 Stata Version 13 (StataCorp. 2013. Stata Statistical Software: Release 13. College
pt
281
282 RESULTS
Ac
284 The pivotal phase ROC PRIMED database included 8,719 patients with 4,374 randomly
285 assigned to treatment with an active ITD and 4,345 with a sham device. We note that
286 the database used in this study contains one less patient than was originally reported.14
12
Page 12 of 34
287 The current investigators’ independent analysis exactly matched rates of return of
289 admission, survival to hospital discharge, and the primary outcome (survival to hospital
t
ip
291 Although CPR quality measurements were attempted in all patients, they were
292 not required during the pivotal phase of study. Among the 8,719 total enrolled and
cr
293 randomized subjects, compression rate data were available in 6,199 (71·1%) subjects,
us
294 compression depth data in 3,750 (43·0%) subjects, and compression fraction data in
296
an
Of those subjects who had electronically documented CPR performance,
M
297 approximately 10% did not receive CPR within “acceptable” chest compression fraction
298 parameters, one-fourth did not receive the “acceptable” rate and over one-third did not
ed
300 There were no differences in patient characteristics between the sham and
pt
301 active ITD groups for the patients who received documented acceptable quality CPR
ce
302 (Table 2). The demographic and cardiac arrest related characteristics of the patients
303 who did not have CPR quality parameters recorded were also compared to those who
Ac
304 did (the patient population that was used for the analysis). The two groups also
306
13
Page 13 of 34
308 Interaction p values are reported as unadjusted and adjusted for age, public location,
309 witnessed arrest, bystander CPR, time interval from EMS dispatch to first EMS arrival,
311
t
ip
312 Chest Compression Rate
313 Chest compression rates were available in 3121 and 3078 patients who were treated
cr
314 with a sham and active ITD respectively. Of those, 74% had documented rates between
us
315 80-120 ccpm. The unadjusted and adjusted test for interaction between the sham and
316 active ITD, chest compression rate of 100±20 compressions/minute, and survival to
317
an
hospital discharge with mRS ≤3 showed p-values of 0·090 and 0·046, respectively (Table
M
318 4).
319
ed
321 Data for chest compression depth were available in 1888 and 1862 patients treated with
pt
322 a sham and active ITD respectively. Of those, 61% had recorded depth of 4-6cm. The
ce
323 unadjusted and adjusted test for interaction between the sham and active ITD,
324 compression depth of 4-6 cm, and survival to hospital discharge with mRS ≤3 showed p-
Ac
326
328 Data on chest compression fraction were available in 2791 and 2743 patients treated
329 with the sham and active ITD, respectively. Of those, 88% had a compression fraction
14
Page 14 of 34
330 ≥50%. The unadjusted and adjusted test for interaction between the sham and active
331 ITD, compression fraction of ≥50%, and survival to hospital discharge with mRS ≤3
333
t
ip
334 Chest Compression Rate and Depth Combined
335 Data for both chest compression rate and depth were available in 1888 and 1861
cr
336 patients treated with a sham and active ITD respectively. Of those, 50% had a recorded
us
337 rate between 80-120 and a depth of 4-6cm. The unadjusted and adjusted test for
338 interaction between sham and active ITD, rate between 80-120 and compression depth
339
an
of 4-6 cm, and survival to hospital discharge with mRS ≤3 showed p-values of 0·006 and
M
340 0.041 respectively (Table 4).
341
ed
343 Data for chest compression rate, depth, and fraction were available in 1888 and 1861
pt
344 patients treated with a sham and active ITD respectively. Of those, 45% had a recorded
ce
346 The unadjusted and adjusted test for interaction between sham and active ITD,
Ac
347 compression rate, depth, and fraction within “acceptable” boundaries, and survival to
348 hospital discharge with mRS ≤3 showed p-values of <0·001 and <0.001 respectively.
350
351 Active and Sham ITD Outcomes When Acceptable CPR Quality Was Performed
15
Page 15 of 34
352 A total of 1,675 patients, 848 patients in the active ITD and 827 patients in the sham ITD
353 group, had documented CPR performed adequately within all three recommended
354 performance parameters (rate, depth, and fraction). Among all patients in this cohort,
355 regardless of the presenting rhythm, use of an active ITD significantly increased the
t
ip
356 chances of survival to hospital discharge with mRS ≤3 (61/848 [7·2%%]) compared with
357 the sham ITD (34/827 [4·1%]); p=0·006; OR: 1·81; 95% CI: 1·17, 2·78). ROSC rates were
cr
358 similar between the two groups (Table 5).
us
359 In contrast, when standard CPR was performed outside the boundaries of “acceptable”
360 CPR the addition of an active –ITD significantly decreased the chances of survival to
361
an
hospital discharge with mRS ≤3 compared to the sham-ITD [34/1013(3·4%) for the
M
362 active-ITD versus 62/1061(5·8%) for the sham-ITD, p=0.0071)
363 Finally, in statistical models where we tested all interactions of the CPR quality
ed
364 parameters and their combinations with treatment, we found that the combination of
365 compression depth and fraction was the most potent effect modifier of the treatment
pt
366 effect (p for interaction=0.006) followed by the combination of compression rate and
ce
368
Ac
369 DISCUSSION
370 This analysis supports the notion that the quality of CPR needs to be taken into account
371 during randomized controlled trials of interventions for cardiac arrest. Our analysis of
16
Page 16 of 34
373 significant and clinically important interactions between the quality of CPR provided, the
374 study interventions, and survival to hospital discharge with favorable neurological
t
ip
377 interaction with each individual CPR quality parameter which was then further
378 strengthened (greater statistical significance) as CPR parameters were combined and
cr
379 the quality of CPR delivered improved. The higher degree of statistical significance with
us
380 the addition of each CPR quality parameter and the degree of statistical significance
381 seen, limit the possibility these findings are the result of unknown bias or chance
382
an
because they are in concordance with the physiological principals of blood flow
M
383 generation during CPR.
384 In a broader context this independent analysis of the ROC PRIMED dataset
ed
386 parameters during the performance of CPR could represent a pivotal modifier in
pt
387 determining potential benefit of promising and novel CPR therapies as well as the
ce
388 chances for survival to hospital discharge with favorable neurological function following
389 any cardiac arrest. 15,20 It also corroborates the assertion that well-designed follow-up
Ac
390 analyses of randomized clinical trials may be an important activity in terms of assisting
391 the scientific community in better assessing the outcome of such studies.15
393 fraction outside “acceptable” CPR quality boundaries were common and likely modified
394 the effect on the overall outcome observed from this “real life” effectiveness clinical
17
Page 17 of 34
395 trial. As such, interaction with CPR quality may have masked an important clinical
396 benefit of the active ITD.14 The findings from the current study highlight the imperative
397 to monitor, record, control, and ultimately achieve consistent high-quality CPR
398 performance, not only during interventional cardiac arrest clinical trials designed to
t
ip
399 improve blood flow during CPR, but also in routine clinical practice. 5,14,19,20 The evidence
400 from this study supports the contention that unrecognized or poorly controlled
cr
401 variables may under power a clinical trial’s ability to detect potential benefits of an
us
402 intervention.16
404
an
effectiveness, this analysis also demonstrated that when the quality of chest
M
405 compressions was provided outside the studied boundaries, use of an active ITD have
406 resulted in lower survival to hospital discharge with mRS ≤3 compared with standard
ed
407 CPR and the sham ITD. (Table 4) Therefore, ITD use with standard CPR outside the
408 “acceptable” range appears harmful. Accordingly, to maximize survival and minimize
pt
409 any potential harm, it may be prudent to make every effort to perform CPR within
ce
410 guidelines when using the ITD and electronically record CPR performance. Further,
411 understanding the effect of individual CPR quality parameters on the type of CPR used
Ac
412 and consistently providing them within the proper boundaries may be crucial to
414 Understanding the effect of individual CPR quality components and their
415 combined effect is supported by the findings from earlier computer modeling and
416 laboratory studies with the ITD. Since the active ITD predominantly functions during the
18
Page 18 of 34
417 decompression phase of CPR, its effectiveness may be lost at higher compression rates
418 as the decompression period is decreased. 8.9 In addition, the effect of the active ITD on
419 coronary perfusion pressure has been shown to be sensitive to inadequate rate and
420 depth. Therefore, the current findings are also consistent with results of another trial
t
ip
421 that compared active compression decompression (ACD) CPR plus the ITD versus
422 standard CPR. 21 In that positive trial, the ACD CPR device not only provided essential
cr
423 active chest wall recoil, but also helped to provide rate and depth feedback to rescue
us
424 personnel.
425 This analysis has limitations. It was a post-hoc analysis and thus subject to the
426
an
typical concerns for such analyses, including type I error. However, unlike many post-
M
427 hoc analyses, the data for this NIH-supervised RCT were collected prospectively and
429 Furthermore, all data parameters were prospectively defined and recorded and were
430 under careful review by a data safety monitoring board for this NIH trial.
pt
431 In addition, the study was not performed as a random retrospective review of
ce
432 the data, but rather a targeted a priori question asked by the investigators. The
433 rationale for using this comprehensive database was also based on growing general
Ac
434 concerns articulated since completion of the study that quality of CPR may significantly
435 affect outcomes and act as an effect modifier in interventional studies.5,16 Outcomes
436 from the ROC PRIMED database had not been previously assessed based upon CPR
437 quality metrics even though all of the target data had been collected in a prospective
438 and well-defined manner. For these reasons, the current study is also subject to the
19
Page 19 of 34
439 potential benefits of post-hoc analyses, including reassessment of outcomes based upon
440 initial assumptions that subsequently may have proven to be erroneous (e.g. there is no
441 interaction between quality of CPR and the effect of the study intervention). 15
442 One other potential limitation is that the technological equipment and methods
t
ip
443 used to acquire rate, depth, and compression fraction data in the ROC PRIMED database
444 did vary from site to site and the CPR quality data were not available to analyze on a
cr
445 site-to-site basis. Also full release of chest compressions (recoil) was not reported, and
us
446 that parameter may be another important CPR quality metric when considering the
447 physiology of the ITD intervention. Nonetheless, these factors were all recognized and
448
an
prospectively defined as part of the trial itself. The final data set evaluated was balanced
M
449 in terms of subject characteristics for those treated with an active and sham ITD. The
450 same was true for the patients who did not have quality parameters recorded. Their
ed
451 population characteristics were similar to the population studied and therefore it is
452 reasonable to assume that the results could be extrapolated to the entire ROC PRIMED
pt
453 population. The fact that each individual compression parameter had a favorable
ce
454 interaction with the use of the ITD and was the strongest with all three parameters
455 combined lends credence to the veracity of the observations made in this analysis. Thus,
Ac
456 the findings provide important insights into interaction between CPR quality and this
457 physiology-based cardiac arrest intervention which could influence the design of future
458 cardiac arrest clinical trials and interpretation of past clinical trials.
459
460
20
Page 20 of 34
461 CONCLUSIONS
462 A statistically significant interaction was demonstrated between the quality of CPR
463 provided, the active or sham ITD, and survival to hospital discharge with favorable
464 neurological outcome (mRS≤3). The quality of CPR delivered to a cardiac arrest patient is
t
ip
465 an important national healthcare issue and may be an underappreciated effect modifier
cr
466 in CPR clinical trials. Consideration of these observations is important for the
467 appropriate design and execution of future resuscitation research, the interpretation of
us
468 previous CPR studies, and clinical resuscitation practice.
469 an
470
M
471
ed
472
473
pt
474
ce
475 ACKNOWLEDGMENTS
Ac
476 The investigators extend their deepest appreciation and respect to the thousands of
477 men and women who comprise the NIH Resuscitation Outcomes Consortium and who
478 have worked tirelessly and with purpose to improve outcomes for those with out-of-
479 hospital cardiac arrest and major trauma for the past decade.
480
21
Page 21 of 34
481 Demetris Yannopoulos, MD: Multiple NIH grants in the area of Resuscitation but
482 nothing in conflict with this study.
483 Tom Aufderheide, MD: current grants to institution from NHLBI Resuscitation
484 Outcomes Consortium, NINDS Neurological Emergency Treatment Trials (NETT)
485 Network, and NNLBI NIH Director’s Transformative Research Award
486 Paul Pepe, MD: current grant to institution from NIH Resuscitation Outcomes
487 Consortium,support for travel from NIH Resuscitation Outcomes Consortium
488 Benjamin Abella, MD: no conflicts of interest for this work
t
489 Sue Duval, PhD: no conflicts of interest for this work
ip
490 Ralph Frascone, MD: no conflicts of interest for this work
491 Jeffery Goodloe, MD: no conflicts of interest for this work
cr
492 Brian Mahoney, MD: no conflicts of interest for this work
493 Vinay Nadkarni, MD: no conflicts of interest for this work
494 Henry Halperin, MD: no conflicts of interest for this work
us
495 Robert O’Connor, MD: no conflicts of interest for this work
496 Ahamed Idris, MD: ongoing grants to institution from NIH/NHLBI and American Heart
497 Association, ongoing support for travel from NIH/NHLBI
498 Lance Becker, MD: grant to institution from NIH
499
an
REFERENCES
M
500
501 1. Field J, Hazinski M, Sayre M, et al: Part 1: Executive summary: 2010 American
ed
503 cardiovascular care. Circulation 2010; 122 (18 Suppl 3): S640–656
pt
504
ce
505 2. Kleinman M, de Caen A, Hameides L, et al: Part 10: Pediatric basic and advanced
509
511 and emergency cardiovascular care. Circulation 2005; 112 (24 Suppl);IV:1–203
22
Page 22 of 34
512
t
ip
516
cr
517 5. Meaney PA, Bobrow BJ, Mancini ME, et al: written on behalf of the CPR Quality
us
519 Care Committee, and the Council on Cardiopulmonary, Critical Care,
520 an
Perioperative and Resuscitation. CPR quality: improving cardiac resuscitation
521 outcomes both inside and outside the hospital: a consensus statement from the
M
522 American Heart Association. Circulation 2013;128:1–19
ed
523
524 6. Halperin H, Tsitlik J, Guerci A, et al: Determinants of blood flow to vital organs
pt
526
530
23
Page 23 of 34
531 8. Sigurdsson G, Yannopoulos D, McKnite S, Lurie K: Cardiorespiratory interactions
532 and blood flow generation during cardiac arrest and other states of low blood
534
t
ip
535 9. Tucker K, Khan J, Idris A, Savitt M: The biphasic mechanism of blood flow during
cr
537 decompression and high-impulse manual external cardiac massage. Ann Emerg
us
538 Med 1994; 24: 895–906.
539
540
an
10. Babbs C, Nadkarni V: Optimizing chest compression to rescue ventilation ratios
M
541 during one-rescuer CPR by professionals and lay persons: children are not just
543
544 11. Berg R, Sanders A, Kern K, et al: Adverse hemodynamic effects of interrupting
pt
547
Ac
548 12. Brown CG, Martin DR, Pepe PE, et al: Standard versus high-dose epinephrine in
549 out-of hospital cardiac arrest - a controlled clinical trial. N Engl J Med 1992;
550 327:1051-1055
551
24
Page 24 of 34
552 13. Stiell I, Nichol G, Leroux B, et al: Early versus later rhythm analysis in patients
553 with out-of-hospital cardiac arrest. N Engl J Med 2011; 365: 787–797
554
555 14. Aufderheide T, Nichol G, Rea T, et al. A trial of an impedance threshold device in
t
ip
556 out-of-hospital cardiac arrest. N Engl J Med 2011; 365(9): 798–806
557
cr
558 15. Ebrahim S, Sohani ZN, Montoya L, Agarwal A, Thorlund K, Mills EJ, Ioannidis JP:
us
559 Reanalyses of randomized clinical trial data. JAMA. 2014 Sep 10;312:1024-1032
561
an
M
562 16. Aufderheide TP, Sigurdsson G, Pirrallo RG, Yannopoulos D, McKnite A, von Briesen C,
563 Sparks CW, Conrad CJ, Provo TA, Lurie KG. Hyperventilation induced hypotension during
ed
565
pt
566 17. Pepe PE, Roppolo LP, Fowler RL: The detrimental effects of ventilation
ce
568
Ac
570 resuscitation during out-of-hospital cardiac arrest. JAMA 2005; 293: 299–304
571
25
Page 25 of 34
572 19. Rothman KJ: No adjustments are needed for multiple comparisons. Epidemiology
574
t
ip
576 resuscitation quality on cardiac arrest outcome: a systematic review and meta-
cr
578
us
579 21. Aufderheide TP, Frascone RJ, Wayne MA et al: Standard cardiopulmonary
581
an
resuscitation with augmentation of negative intrathoracic pressure for out-of-
M
582 hospital cardiac arrest: a randomised trial. Lancet. 2011;377(9762):301-311
583
ed
584
pt
ce
Ac
26
Page 26 of 34
584 Table 1. Patients with available data within pre-defined CPR quality subgroups
Sham Active
Group
(n=4345) (n=4374)
Documented compression rate 3121 3078
Rate of 100±20 ccpm 2274/3121 (72·9%) 2319/3078 (75·3%)
t
Documented compression depth 1888 1862
ip
Depth of 5±1 cm 1146/1888 (60·7%) 1139/1862 (61·2%)
Documented compression fraction 2791 2743
cr
Compression fraction ≥0·5 2471/2791 (88·5%) 2417/2743 (88·1%)
Documented rate, depth, and fraction 1888 1861
us
Rate of 100±20 ccpm, depth of 5±1 827/1888 (43·8%) 848/1861 (45·6%)
cm, and fraction ≥0·5
585 an
M
ed
pt
ce
Ac
27
Page 27 of 34
t
ip
cr
Table 2. Characteristics of patients receiving acceptable quality of CPR (rate 80-120 ccpm and depth 4-6 cm and fraction ≥50%) and those not
receiving acceptable quality CPR (none of rate 80-120 ccpm or depth 4-6 cm or fraction ≥50%), stratified by treatment arm
us
“Acceptable” quality CPR Not “acceptable” quality CPR
Sham Active Sham Active
Characteristic p-value p-value
(n=827) (n=848) (n=1061 ) (n=1013 )
an
Age, years 67.3 ± 14.6 66.9 ± 15.1 0.63 67.9 ± 16.2 67.9 ± 16.5 0.97
Male sex 550 (66.5) 556 (65.6) 0.69 647 (61.0) 662 (65.4) 0.039
Public location 109 (13.2) 117 (13.8) 0.71 145 (13.7) 129 (12.7) 0.53
Witnessed status
M
EMS witnessed 37 (4.5) 49 (5.8) 0.23 77 (7.3) 80 (7.9) 0.58
Bystander witnessed 369 (44.6) 359 (42.3) 0.35 445 (41.9) 392 (38.7) 0.13
Bystander CPR performed (%) 312 (37.7) 321 (37.9) 0.91 385 (38.4) 342 (36.4) 0.38
Time from dispatch to first EMS arrival (mins) 5.8 ± 2.0 5.6 ± 1.9 0.06 5.7 ± 2.1 5.7 ± 2.1 0.74
d
Time from dispatch to first arrival of ALS providers (mins) 8 (6, 11) 8 (6, 10) 0.024 8 (6, 10) 8 (6, 10) 0.71
Treated by ALS providers 825 (99.8) 847 (99.9) 0.62 1053 (99.3) 1011 (99.8) 0.11
First rhythm interpretation
Shockable, VT or VF
Pulseless electrical activity te 228 (27.6)
194 (23.5)
230 (27.1)
187 (22.1) 0.71
247 (23.3)
247 (23.3)
212 (20.9)
302 (29.8) 0.01
ep
Asystole 385 (46.6) 404 (47.6) 535 (50.4) 472 (46.6)
AED used, no shock advised, no recording 20 (2.4) 27 (3.2) 32 (3.0) 27 (2.7)
No. of shocks, if given 2 (1, 4) 3 (1, 4) 0.82 2 (1, 4) 2 (1, 4) 0.68
Prehospital intubation attempted 693 (83.8) 726 (85.6) 0.30 892 (84.1) 851 (84.0) 0.97
c
Successful 626 (75.7) 638 (75.2) 0.83 773 (72.9) 765 (75.5) 0.17
Values are n (%), mean ± SD or median (25th, 75th percentile)
Ac
29
Page 28 of 34
t
ip
cr
Table 3. Characteristics of subpopulations with a) all 3 CPR quality measures recorded and b) no CPR quality measures recorded.
us
Clinical characteristics were similar regardless of the presence or absence of quality parameter recordings.
an
recorded
All 3 None
M
Characteristic (n=3749) (n=2515)
d
Male sex 2415 (64·4) 1634 (65·0)
Public location
1565 (41·7)
345 (13·7)
988 (39·3)
ep
Bystander performed CPR
Yes 1360 (36·3) 892 (35·5)
c
30
Page 29 of 34
t
ip
cr
Shockable VT or VF 917 (24·5) 586 (23·3)
us
Pulseless electrical activity 930 (24·8) 580 (23·1)
Asystole 1796 (47·9) 1012 (40·2)
an
AED used, no shock advised, no recording available 106 (2·8) 319 (12·7)
Unknown or could not be determined 0 (0·0) 31 (1·2)
M
Shocks applied 1523 (40·6) 958 (38·2)
Prehospital intubation
d
Attempted 3162 (84·3) 2121 (84·3)
Successful
te 2802 (74·7) 1951 (77·6)
ep
Drugs administered before hospital arrival
Epinephrine 3407 (90·9) 2173 (86·7)
Atropine 2780 (74·2) 1814 (72·4)
c
31
Page 30 of 34
t
ip
cr
Table 4. Survival to hospital discharge with mRS ≤3 based on “acceptable” CPR quality measures of rate, depth, and fraction. “Acceptable” range
for rate was considered to be between 80-120 ccpm, depth between 4-6 cm, fraction ≥50%. Δunadj represents the unadjusted difference between
us
sham and active ITD; pint represents the p-value for interaction (unadjusted); Δadj represents the difference between sham and active ITD
adjusted for age, public location, witnessed arrest, bystander CPR, time from dispatch to first EMS arrival, shockable rhythm; pint-adj represents
the p-value for interaction (adjusted).
an
All Sham Active Δunadj pint Δadj pint-adj
Rate
M
Acceptable Quality 284/4593 (6.2%) 132/2274(5.8%) 152/2319(6.6%) +0.8% 0.090 +0.9% 0.054
d
Depth
Acceptable Quality
Unacceptable Quality
te 134/2285 (5.9%)
58/1465 (4.0%)
61/1146(5.3%)
35/742(4.7%)
73/1139(6.4%)
23/723(3.2%)
+1.1%
-1.5%
0.075 +1.0%
-1.0%
0.14
ep
Fraction
Acceptable Quality 294/4888 (6.0%) 140/2471(5.7%) 154/2417(6.4%) +0.7% 0.017 +0.8% 0.013
c
Acceptable Quality 116/1890 (6.1%) 48/930(5.2%) 68/960(7.1%) +1.9% 0.006 +1.7% 0.024
32
Page 31 of 34
t
ip
cr
Acceptable Quality 107/2006 (5.3%) 42/1009(4.2%) 65/997(6.5%) +2.3% 0.001 +2.4% <0.001
us
Unacceptable Quality 85/1744 (4.9%) 54/879(6.1%) 31/865(3.6%) -2.5% -2.3%
an
Acceptable Quality 225/3690 (6.1%) 98/1828(5.4%) 127/1862(6.8%) +1.4% 0.004 +1.6% 0.002
M
Rate, Depth & Fraction
Acceptable Quality 95/1675 (5.7%) 34/827(4.1%) 61/848(7.2%) +3.1% <0.001 +3.0% <0.001
d
Quality Monitoring
All 3 recorded
te
191/3749 (5.1%) 96/1888(5.1%) 95/1861(5.1%) 0% 0.57 +0.2% 0.96
ep
Either 1 or 2 recorded 188/2455 (7.7%) 92/1235(7.5%) 96/1220(7.9%) +0.4% +0.4%
33
Page 32 of 34
t
ip
cr
Table 5. Rates of ROSC and survival of patients receiving acceptable quality of CPR (rate 80-120 ccpm, depth 4-6 cm, fraction ≥50%), and those
not receiving acceptable quality CPR (none of rate 80-120 ccpm or depth 4-6 cm or fraction ≥50%), stratified by treatment arm
us
Acceptable quality CPR Unacceptable quality CPR
an
Active Active
Sham p-value Sham p-value
(n=827) (n=848) (n=1,061) (n=1,012)
M
ROSC 213 (25.8) 233 (27.5) 0.43 244 (23.0) 230 (22.7) 0.88
d
Discharged alive with mRS ≤3 34 (4.1) 61 (7.2) 0.0064 62 (5.8) 34 (3.4) 0.0071
34
Page 33 of 34
Figure 1. Difference in survival to hospital discharge with mRS≤3 (unadjusted) for
sham- and active-ITD groups based on the presence or absence of CPR quality
parameters.
t
ip
cr
us
an
M
ed
pt
ce
Ac
35
Page 34 of 34