Content uploaded by Paul Crawford
Author content
All content in this area was uploaded by Paul Crawford on May 01, 2019
Content may be subject to copyright.
CPR Guideline Chest Compression Depths May Exceed Requirements for
Optimal Physiological Response
Olibhéar McAlister1,3, Daniel Guldenring2, Dewar D Finlay1, Raymond R Bond1, Hannah Torney1,3,
Ben McCartney3, Laura Davis3, Paul Crawford4, Adam Harvey3
1Ulster University, Newtownabbey, UK
2University of Applied Sciences HTW Berlin, Berlin, Germany
3HeartSine Technologies Ltd, Belfast, UK
4Veterinary Anaesthesia Consultancy, Larne, UK
Abstract
A twelve-animal porcine study dataset was
retrospectively analyzed to assess associations between
chest compression (CC) depth, systolic blood pressure
(SBP) and end-tidal carbon dioxide (EtCO2). Manual CCs
were applied for 7 two-minute episodes, at CC depths
between 10mm-55mm. A rolling 15s analysis window was
applied to the continuous signals. Mean peak values were
calculated for each window. Correlation analysis was
applied to assess strength of association. Optimal CC
depth to achieve physiological targets was determined via
cut-off analysis.
A total of 672 observations for each variable were
available for analysis. Pearson correlations (95%
confidence interval; p-value) between CC depth and both
SBP and ETCO2 were 0.84 (0.82, 0.86; p < 0.001) and 0.75
(0.71, 0.78; p < 0.001) respectively. Optimal CC depth cut-
off (sensitivity, specificity) to achieve SBP ≥ 100mmHg and
EtCO2 ≥ 10mmHg was 33 mm (98.29%, 88.94%) and 20
mm (95.08%, 78.30%) respectively.
A reasonable relationship between CC depth and
physiological response was observed. Optimal SBP and
EtCO2 cut-offs were achieved significantly below guideline
depths. Furthermore, cut-off analysis suggests a disparity
between CC depth and physiological targets.
1. Introduction
Cardiac arrest is a leading cause of premature death
worldwide. To increase survival rates, early defibrillation
and effective cardiopulmonary resuscitation (CPR) are
crucial. American Heart Association (AHA) and European
Resuscitation Council (ERC) basic life support guidelines
consider chest compressions (CCs) at a rate of 100 to
120 CCs min-1 and a depth of 50 to 60 mm to be effective,
amongst an adult population [1], [2].
Guideline CC depth between 50 and 60 mm has been
proven to marginally increase survival to hospital
admission, compared to previous guideline depths of 40 to
50 mm [3], [4]. While associated survival improves with
deeper CCs so does the risk of causing injury to the patient
[5]. It has been repeatedly reported that CPR performance
is poor for both professional and lay rescuers, over several
revisions of the basic life support guidelines [6], [7]. The
low incidence of CC depth meeting the minimum guideline
depth of 50 mm may be due to the target not being suitable
for the entire adult population or early onset of fatigue [8],
[9].
Research into patient response as an indicator of CPR
quality is in its infancy with few physiological endpoints
and cut-offs established. Advanced life support guidelines
do suggest an alternative indicator of measuring CC
quality. Observing a patient EtCO2 response of < 10 mmHg
is associated with mortality and efforts to improve CPR
quality should be made. Supporting evidence suggests that
continuous CCs between 100 and 120 compressions min-1
maintains ideal blood pressure [10]. Friess et al.
investigated the use of SBP as an indicator of CPR quality
and used a physiological cut-off of 100 mmHg [11].
2. Methods
The purpose of this analysis is to determine the optimal
CC depth cut-off to achieve EtCO2 ≥ 10 mmHg and SBP ≥
100 mmHg. A retrospective analysis was conducted on a
porcine dataset. The dataset included continuous time-
series data for CC depth, EtCO2 and SBP.
2.1. Study Design
All experiments were performed in accordance with the
Home Office Guidance on the Operation of the Animals
(Scientific Procedures) Act 1986 (UK).
Twelve (n = 12) commercial pigs, aged approximately
Computing in Cardiology 2018; Vol 45 Page 1 ISSN: 2325-887X DOI: 10.22489/CinC.2018.116
9 to 10 weeks and weighing between 30 to 35 kg, were
used in the study. Ventricular fibrillation (VF) was induced
electrically, and the animals were left untreated for 3-
minutes. During the untreated period animals were
ventilated at an approximate rate of 23 ventilations min-1.
Each animal had 7 episodes of continuous CCs applied
at a rate of 110 compressions min-1. The initial 4 episodes
of CPR were applied to achieve an EtCO2 response of
< 15 mmHg. The remaining 3 episodes targeted an EtCO2
response of ≥ 15 mmHg. There was a rest period of at least
10-seconds between CC episodes.
2.2. Signal Data
A HeartStart Mrx (Philips, USA) coupled with Q-CPR
technology (Laerdal Medical, Norway) was used to record
CC depth data. Depth signal data was captured at a sample
frequency of 50 Hz and a resolution of 0.01 mm per least
significant bit (LSB).
Physiological signals were recorded using a Datex-
Ohmeda S/3 Anesthesia Monitor (GE Healthcare, USA)
using VitalSignsCapture [12]. Side-stream capnograph
was used to measure EtCO2 at a sampling frequency of
25 Hz. Arterial blood pressure (BP) was captured from the
carotid artery and sampled at a rate of 100 Hz. Outputs
from the anesthesia were recorded in mmHg and did not
require scaling prior to processing.
2.3. Data Processing
An annotation review was conducted on the CC depth
and BP signals, by study personnel, to identify the
beginning of each CPR episode. The point of VF induction
was also annotated on each BP signal. Annotated BP
timepoints were transferable to the capnogram, as the same
device was used to capture both signals. Episodes were
segmented into 15-second epochs. An analysis widow was
applied to each epoch to determine the amplitude of the
signal.
Local minima were identified in the CC depth signal
for each CC. The absolute value of the mean of the
identified local minima, within a CC depth epoch,
represented the mean CC depth.
Capnograph and BP signals were analyzed by peak
envelope. The mean of the upper envelope in the
capnograph and BP signals were taken as the
representative values of EtCO2 and SBP for a given epoch,
respectively.
Additional processing was applied to the capnograph
signals. A rolling, non-overlapping, analysis widow was
applied to each signal starting at the point of VF induction.
The analysis window had a fixed duration of 12-seconds
which terminated after 180-seconds of signal had been
processed. The amplitude of the capnogram was calculated
for each analysis window.
2.4. Data Analyses
Data was audited by independent review prior to
analysis. R for statistical computing version 3.5.1 was used
for all analyses.
Between-subject, within-subject and Pearson
correlation analyses were applied to each combination of
CC depth, EtCO2 and SBP [13], [14].
Cutoff analyses were applied to the data to determine
the probabilistic CC depth cutoffs for EtCO2 ≥ 10 mmHg
and peak BP ≥ 100 mmHg. Depth cutoffs increased in
increments of 1 mm and accuracy, sensitivity, specificity
and Youden index were calculated for each CC depth
cutoff. Cutoffs which are associated with maximum
accuracy and maximum Youden index were reported.
The decay of EtCO2 post VF induction was
characterized by applying a log-log regression model to
EtCO2 and time data.
3. Results
A total of 672 observations (12 animals x 7 episodes x
8 analysis windows) of EtCO2 and SBP were processed.
There were 13 missing observations for CC depth due to
the administration of shallow CCs.
There were non-significant, between-subject
correlations observed for all combinations of CC depth,
EtCO2 and SBP (Table 1).
Strong within-subject correlations were observed for all
combinations of the study endpoints; EtCO2 and CC depth
(0.83), EtCO2 and SBP (0.86) and SBP and CC depth
(0.89). Additionally, lower, yet strong Pearson correlations
were observed between all combinations of study
endpoints. Further details of all correlation analyses are
listed in Table 1.
The maximum accuracy cut-off for CC depth to predict
EtCO2 ≥ 10 mmHg was 20 mm. This provided an accuracy
of 89.68% (sensitivity = 0.95; specificity = 0.78; Youden
index = 0.73). The optimal depth cut-off to classify
SBP ≥ 100 mmHg was 33 mm with an accuracy of 92.26%
(sensitivity = 0.98; specificity = 0.89; Youden
Index = 0.87).
Adjusting this analysis in favor of maximum Youden
index the optimal cut-off for EtCO2 ≥ 10 mmHg increases
to 21 mm with an accuracy of 87.86% (sensitivity = 0.90;
specificity = 0.84; Youden index = 0.74). There was no
change to the depth-cut off after adjusting for maximum
Youden index.
The deterioration of EtCO2 post VF induction
resembled characteristics indicative of exponential decay.
Data obtained for each 12-second analysis window during
the untreated duration of VF were not considered to be
Page 2
normally distributed (Figure 1). The median value for each
timepoint was applied to a regression model to trend the
decay. A log-log model was trained after applying a natural
log transform to both EtCO2 and time data. This provides
the following model satisfied in terms of EtCO2 () as a
response of time ():
The coefficient of determination for the model (R2) was
0.996, suggesting the model is a good fit for the data.
4. Discussion
The analysis conducted could not identify a relationship
between animals which received higher CC depths and
those which produced higher EtCO2 or SBP responses.
This may be consequential to the damage which occurred
to one animal’s thorax or that each animal received the
same treatment.
The Pearson correlation analysis suggests there is a
moderate relationship between CC depth and EtCO2
(0.75). Comparisons of CC depth and SBP show a stronger
relationship (0.84). Comparing the Pearson and within-
subject correlation analyses, however, suggests there is an
improvement in the relationship, between the covariates,
when fitting for each animal.
The binary outcome for EtCO2 ≥ 10 mmHg was
predicted accurately using a CC depth cut-off of 20 mm;
Accuracy (89.68%) and Youden index (0.73). Similarly,
binary outcome for SBP was accurately predicted with a
depth cut-off of 33 mm; Accuracy (92.26%) and Youden
index (0.87).
Depths required to meet the physiological targets
however, are considerably lower than the current guideline
recommended depths of 50 to 60 mm. It is also worth
noting that depths required to achieve the defined
physiological cut-offs were also not in agreeance. This
reflects the conclusion of Steill et al, that an increase in CC
depths is associated with better patient outcomes, however,
the optimal depth of CC is still unknown [9].
One animal suffered extensive thoracic trauma as a
result of receiving CC depths in excess of 45 mm.
Complications due to this treatment included bleeding into
the thorax, bleeding into the pericardial sac, bruised
myocardium, bullae on the lungs and most severely
ruptured atria. This highlights that a balanced approach to
delivering safe and efficacious CC depth targets is
required. This study demonstrated depths greater than
45 mm were acceptable for 11 of the 12 enrolled animals,
however, one displayed irreversible damage as a result of
this treatment. At what point are the CC depth targets
considered safe for use on the human population?
Upon visual inspection of the synchronized signals, it is
apparent that EtCO2 is a slow response variable, which
does not reflect sudden changes in applied CC depth
5. Limitations
The investigation was a retrospective analysis of a
previously obtained dataset. The objective of the original
study did not match the objective of this post hoc analysis.
Treatment was not randomized during the study. Depth of
CC increased with each episode which may have an
indirect impact on physiological response of the animal. As
observed as part of the analysis, EtCO2 had a decay artefact
and requires a considerable amount of time to baseline,
perhaps the 10-second interval between CC episodes
would need extended to accommodate this.
Figure 1 Boxplot series displaying the exponential
decay of EtCO2 as function of time, after induction of
VF. (N = 144; 12 observations per time point x
12 timepoints)
Table 1 Pearson, between-subject and within-subject
correlation analyses for each combination of EtCO2 (n
= 672, SBP (n = 672) and CC depth (n = 659).
Covariates
r
95% CI
p-value
Between-Subjects
EtCO2 and Depth
-0.28
(-0.77, 0.43)
0.386
SBP and Depth
-0.22
(-0.75, 0.48)
0.491
SBP and EtCO2
0.35
(-0.35, 0.80)
0.258
Within-Subjects
EtCO2 and Depth
0.83
(0.80, 0.85)
< 0.001
SBP and Depth
0.89
(0.87, 0.91)
< 0.001
SBP and EtCO2
0.86
(0.84, 0.88)
< 0.001
Pearson
EtCO2 and Depth
0.75
(0.71, 0.78)
< 0.001
SBP and Depth
0.84
(0.82, 0.86)
< 0.001
SBP and EtCO2
0.80
(0.78, 0.83)
< 0.001
Page 3
6. Conclusions
This investigation provides encouraging preliminary
results indicating that CC depths recommended by AHA
and ERC guidelines may be excessive. As this is a
retrospective analysis further research is required to
establish the relationship between animals for CC depth
and the physiological endpoints of EtCO2 and BP.
Conflicts of Interest
Olibhéar McAlister, Hannah Torney, Ben McCartney,
Laura Davis and Adam Harvey are employees of HeartSine
Technologies Ltd. Paul Crawford is a consultant veterinary
anesthetist contracted by HeartSine Technologies Ltd.
Acknowledgements
The authors of this paper would like to thank the staff at
the Roslin Institute (The University of Edinburgh) for their
help collecting the presented data.
References
[1] M. E. Kleinman et al., “Part 5: Adult basic life support
and cardiopulmonary resuscitation quality: 2015
American Heart Association guidelines update for
cardiopulmonary resuscitation and emergency
cardiovascular care,” Circulation, vol. 132, no. 18, pp.
S414–S435, 2015.
[2] J. Soar et al., “European Resuscitation Council
Guidelines for Resuscitation 2015. Section 3. Adult
advanced life support.,” Resuscitation, vol. 95, pp. 100–
147, 2015.
[3] T. Vadeboncoeur et al., “Chest compression depth and
survival in out-of-hospital cardiac arrest,” Resuscitation,
vol. 85, no. 2, pp. 182–188, Feb. 2014.
[4] C. H. Wang et al., “Outcomes of adults with in-hospital
cardiac arrest after implementation of the 2010
resuscitation guidelines,” Int. J. Cardiol., vol. 249, pp.
214–219, 2017.
[5] H. Hellevuo et al., “Deeper chest compression – More
complications for cardiac arrest patients?,”
Resuscitation, vol. 84, no. 6, pp. 760–765, Jun. 2013.
[6] B. S. Abella, “Quality of Cardiopulmonary
Resuscitation During In-Hospital Cardiac Arrest,”
Jama, vol. 293, no. 3, p. 305, Jan. 2005.
[7] T. Gyllenborg, A. Granfeldt, F. Lippert, I. S. Riddervold,
and F. Folke, “Quality of bystander cardiopulmonary
resuscitation during real-life out-of-hospital cardiac
arrest,” Resuscitation, vol. 120, pp. 63–70, Nov. 2017.
[8] T.-G. Kampmeier et al., “Chest compression depth after
change in CPR guidelines—Improved but not
sufficient,” Resuscitation, vol. 85, no. 4, pp. 503–508,
Apr. 2014.
[9] I. G. Stiell et al., “What Is the Optimal Chest
Compression Depth During Out-of-Hospital Cardiac
Arrest Resuscitation of Adult Patients?,” Circulation,
vol. 130, no. 22, pp. 1962–1970, Nov. 2014.
[10] L. M. Cunningham, A. Mattu, R. E. O’Connor, and W.
J. Brady, “Cardiopulmonary resuscitation for cardiac
arrest: the importance of uninterrupted chest
compressions in cardiac arrest resuscitation,” Am. J.
Emerg. Med., vol. 30, no. 8, pp. 1630–1638, Oct. 2012.
[11] S. H. Friess et al., “Hemodynamic directed CPR
improves cerebral perfusion pressure and brain tissue
oxygenation,” Resuscitation, vol. 85, no. 9, pp. 1298–
1303, Sep. 2014.
[12] J. Karippacheril and T. Ho, “Data acquisition from S/5
GE Datex anesthesia monitor using VSCapture: An
open source.NET/Mono tool,” J. Anaesthesiol. Clin.
Pharmacol., vol. 29, no. 3, p. 423, 2013.
[13] J. M. Bland and D. G. Altman, “Statistics notes:
Calculating correlation coefficients with repeated
observations: Part 1--correlation within subjects,” BMJ,
vol. 310, no. 6977, pp. 446–446, Feb. 1995.
[14] J. M. Bland and D. G. Altman, “Statistics notes:
Calculating correlation coefficients with repeated
observations: Part 2--correlation between subjects,”
BMJ, vol. 310, no. 6980, pp. 633–633, Mar. 1995.
Address for correspondence.
Olibhéar McAlister
Ulster University, NIBEC, Newtownabbey, BT37 OQB
McAlister-o2@ulster.ac.uk
Figure 2 Representative time series plots of CC depth,
BP and EtCO2, demonstrating the slow response of
EtCO2 to sudden changes in CC depth
Page 4