Simulating blood pressure and end tidal CO2 in a CPR training manikin

Computer Methods and Programs in Biomedicine 180 (2019) 105009

Contents lists available at ScienceDirect

Computer Methods and Programs in Biomedicine journal homepage: www.elsevier.com/locate/cmpb

Simulating blood pressure and end tidal CO2 in a CPR training manikin Nicholas Widmann a,c,∗, Robert Sutton a,b, Newton Buchanan a, Dana E. Niles a, Godfrey Nazareth a, Vinay Nadkarni a,b, Matthew R. Maltese a,b a

Department of Anesthesiology and Critical Care Medicine, The Children’s Hospital of Philadelphia, 34th Street and Civic Center Blvd Philadelphia, PA 19104, USA The Perelman School of Medicine, University of Pennsylvania, 3400 Civic Center Blvd, Philadelphia, PA 19104, USA c Department of Mechanical Engineering, Drexel University, 3141 Chestnut St, Philadelphia, PA 19104, USA b

a r t i c l e

i n f o

Article history: Received 20 May 2019 Accepted 3 August 2019

Keywords: Cardiopulmonary resuscitation Cardiac arrest Training manikin Blood pressure

a b s t r a c t Background and Objective: The American Heart Association supports titrating the mechanics of cardiopulmonary resuscitation (CPR) to blood pressure and end tidal carbon dioxide (ETCO2) thresholds during in-hospital cardiac arrest. However, current CPR manikin training systems do not prepare clinicians to use these metrics to gauge their performance, and currently provide only feedback on hand placement, depth, rate, release, and interruptions of chest compressions. We addressed this training hardware deficiency through development of a novel CPR training manikin that displays simulated blood pressure and ETCO2 waveforms in real time on a simulated clinical monitor visible to the learner, reflecting the mechanics of chest compressions provided to the manikin. Such a manikin could improve clinicians’ CPR technique while also training them to titrate CPR quality to physiologic blood pressure and ETCO2 targets as performance indicators. Methods: We used data and key findings from 4 human and 6 animal studies (including 132 human subjects, 61 pigs, and 16 dogs in total) to develop an algorithm that simulates blood pressure and ETCO2 waveforms based on compression mechanics for a pediatric patient. We modified an off-the-shelf infant manikin to incorporate a microcontroller sufficient to process the aforementioned algorithm, and a tablet computer to wirelessly display the simulated waveform. We recruited clinicians with in-hospital CPR experience to perform compressions with the manikin and complete a post-test survey on their satisfaction with designated elements of the manikin and display. Results: 34 clinicians performed CPR on the prototype manikin system that simulates real-time bedside monitoring of blood pressure and ETCO2. 100% of clinicians surveyed reported “satisfaction” with the blood pressure waveform. 97% said they thought depth was accurately reflected in blood pressure (0% inaccurate, 3% not sure). 88% reported an accurate chest compression rate modification effect on blood pressure and ETCO2 (3% inaccurate, 9% not sure) and 59% an accurate effect of leaning (6% inaccurate, 35% not sure). Most importantly, all 34 respondents responded “yes” when asked if they thought this system would be helpful for CPR training. Conclusion: A CPR manikin that simulates blood pressure and ETCO2 was successfully developed with acceptable relevance, performance and feasibility as a CPR quality training tool. © 2019 Elsevier B.V. All rights reserved.

1. Introduction Each year in the United States, roughly 60 0 0 children receive cardiopulmonary resuscitation (CPR) for in-hospital cardiac arrest

∗ Corresponding author at: Department of Mechanical Engineering, Drexel University, 3141 Chestnut St, Philadelphia, PA 19104, USA. E-mail addresses: [email protected] (N. Widmann), [email protected] (R. Sutton), [email protected] (D.E. Niles), [email protected] (V. Nadkarni), [email protected] (M.R. Maltese).

https://doi.org/10.1016/j.cmpb.2019.105009 0169-2607/© 2019 Elsevier B.V. All rights reserved.

(IHCA) [1]. Even in hospitals, CPR technique relies on the American Heart Association (AHA) guidelines which recommend a rate of 100–120 chest compressions per minute (CCs/min) and a depth of greater than 1/3 of the anteroposterior (AP) diameter of the chest [2–5]. However, published survival rates (39–52% [6–11]) suggest that these targets may not work for every patient and highlight the need to identify other targets of high-quality CPR to improve outcomes. Recent translational research has shown that hemodynamicdirected CPR improves survival rate in both pediatric and adult

2

N. Widmann, R. Sutton and N. Buchanan et al. / Computer Methods and Programs in Biomedicine 180 (2019) 105009

models of cardiac arrest [12–17]. While achieving the standard rate and depth values is associated with improved outcomes, the primary determinant of survival is diastolic blood pressure (DBP), the driving force of coronary blood flow [14,18,19]. In their 2015 study of end tidal carbon dioxide (ETCO2) during CPR, Sheak et al. determined there was no statistically significant difference between average chest compression (CC) rate and depth in survivors and non-survivors, while survivors had an average 12 mmHg greater ETCO2 than non-survivors (38.2 vs. 26.1 mmHg) [20]. The AHA recommended using ETCO2 in 2011 [21] and DBP in 2013 [22] for monitoring CPR quality. Hemodynamic-directed CPR is widely feasible, as over 90% of pediatric IHCA events occur in an intensive care unit (ICU) where patients’ blood pressure (BP) and ETCO2 are being recorded and displayed on a monitor in real-time [23]. Despite this, as of 2016, barely a third of clinicians were using these metrics to gauge their CPR quality [15]. This suggests the need for increased educational efforts toward physiological monitoring during CPR. CPR manikins are effective tools to train CPR providers, as laypersons can be trained to perform adequate CPR according to AHA guidelines in as little as 30 minutes [24,25]. The current state of the art manikins use visual and audio feedback systems to coach learners to achieve the standard depths and rates recommended by the AHA [24,25]. However, they do not include a modality to train clinicians to monitor BP or ETCO2 during CPR and adjust their CPR accordingly. To address this issue, we have developed a CPR training manikin that simulates and displays arterial BP and ETCO2 waveforms. The values for these waveforms are based on studies where actual patient data was analyzed to find relationships between BP and ETCO2 and the mechanics (depth, rate, etc.) of CPR. Factors considered detrimental to achieving ideal BP and ETCO2 values are depths and CC rates outside the AHA recommended ranges, not achieving full recoil on CCs (leaning), and pausing CCs [2]. We incorporated all of these factors in our “BP manikin.” The structure of this article is as follows. Section 2 will explain design considerations, including how the system should be assessed. Section 3 will describe the actual hardware and software of the manikin. Section 3 also includes the algorithms for finding mechanical values and translating them into simulated physiological values. Section 4 describes the results from a pilot study conducted within the Children’s Hospital of Philadelphia to receive preliminary feedback and test the general practicality of the manikin. Section 5 presents discussions, including lessons learned and future plans. 2. Design considerations There are five mechanical factors that are used to determine quality of CPR: depth of CCs, rate of CCs, leaning during CCs, pauses in CCs, and ventilation rate [2]. These factors can all also affect BP and ETCO2 to some degree. The CPR training system presented herein incorporates the first four but does not consider ventilation rate at this writing. Because BP and ETCO2 responses to CPR vary widely [26], our manikin is designed to represent the average human infant response for the purpose of training clinicians to administer CPR guided by hemodynamic metrics. We designed our training manikin system with the following features. The AHA recommends achieving a depth of at least 1/3 AP chest diameter and a rate of 100–120 CCs/min with full recoil and minimal pauses during CPR [2]. In order to teach quality CPR, if the learner achieves these mechanical metrics, they should be provided the systolic blood pressure (SBP), DBP, and ETCO2 values most indicative of survival (SBP/DBP >80/30 mmHg and ETCO2 >30 mmHg) [18,20]. If the learner is not achieving the aforementioned depth and rate targets, this should be reflected through a lower BP and ETCO2. This should teach the learner the same CPR

technique that other training manikins teach while also training them to use physiology to monitor their performance. Translated to a real-life IHCA in an ICU, they will be able to watch the arterial BP and ETCO2 waveforms to try to target the ranges most associated with survival. The current version of our training manikin should be assessed on two criteria. First, it is important that clinicians trained with this manikin perform high-quality CPR, as recommended by the AHA, at least equivalent to clinicians trained with traditional mechanics-based training manikins. Second, clinicians trained with this manikin should use hemodynamics to monitor the quality of their CPR. 3. Description of training manikin 3.1. Hardware The manikin consists of a microcontroller (Arduino Nano and HC-06 Bluetooth module) retrofit into the spring-loaded linear encoder sensor of a Resusci Baby QCPR infant training manikin (Laerdal Medical, Stavanger) (Fig. 1). The microcontroller wirelessly (HC-06 Bluetooth module) transmits compression depth to a tablet computer (Nexus 7, Asus, Taipei) where the data is processed and BP, ETCO2, and depth are displayed to the trainee in real time. 3.2. Software Because BP is simulated in real-time, it is first necessary to extract from the depth (the distance the manikin’s chest has been compressed) data whether a compression is beginning, ending, in process, or at its peak depth. To do this, we use a delay of 3 depth values (roughly 30 milliseconds) to continuously calculate the current slope of depth with respect to time. Positive and negative slopes indicate increasing and decreasing depths, respectively. If the slope changes from negative to positive, it signifies the start of a compression. Similarly, if the slope changes from positive to negative it means the compression has reached its peak depth. Rate (CCs/min) is found by dividing 60 seconds by the time difference between adjacent depth vs. time peaks. “Leaning” is the compression start depth, “absolute depth” is the peak compression depth, and “relative depth” is absolute depth minus leaning (all in millimeters). These four values are all found as an average over the last five compressions. These numbers are then used to simulate BP and ETCO2. Because many factors must be considered when determining quality CPR [2], we account for each one separately, starting with depth. We first calculate SBP and DBP based on depth assuming

Fig. 1. The manikin system consists of a linear encoder and HC-06 Bluetooth receiver attached to an Arduino Nano.

N. Widmann, R. Sutton and N. Buchanan et al. / Computer Methods and Programs in Biomedicine 180 (2019) 105009

3

Fig. 2. Our DBP and SBP regression models alongside raw data collected by Sutton et al. [18].

every other variable (e.g., rate, leaning, etc.) is within guidelines. For the sake of our calculations, we refer to these values as “maximum SBP” and “maximum DBP”, respectively. We then account for the other factors through penalties to these maximum values. In their study on the hemodynamic effects of CC depth and rate during CPR in children and adolescents, Sutton et al. (2013) found that the odds of achieving an SBP of at least 80 mmHg was more than twice as likely (2.02 relative odds) when rate and depth were in line with AHA guidelines than when they were not [18]. Since we assumed rate was perfect for the calculation of maximum SBP, a depth of at least 1/3 AP chest diameter (37 mm in our manikin) yielded an SBP of at least 80 mmHg in our model. Sutton et al. also found that a depth increase of 10 mm corresponded to average increase in SBP of 15.5 mmHg [18]. This relationship was used for the middle of the model (25–35 mm depth shows SBP increase 68–83 mmHg). Because Sutton et al. only recorded data close to the target depth and SBP would be very low when depth of CCs is zero, the rate of SBP increase with depth was assumed to start higher at shallower depths and eventually level out past the target depth of 37 mm. We constructed our model for SBP using Microsoft Excel’s trendline feature. We chose a series of data points that, through experimentation, gave us the desired shape and values for SBP and depth in a second order polynomial regression curve (Eq. (1) where d is depth). Sainio et al. (2005) found that deeper compressions than the AHA recommendations resulted in a higher SBP and might be beneficial [27]. These findings agree with our model in which SBP continues to increase with increasing depth past the target.

SBP = −0.0297d2 + 3.2635d + 5

(1)

Sutton et al. and Sainio et al. both reported strong associations between AHA recommended depths and rates with DBP greater than 30 mmHg [18,26]. However, correlations between changes in depth and changes in DBP are weak [18,26]. Because of these weak correlations, we built our model using a third order polynomial (Eq. (2) where d is depth) so that DBP would be affected very weakly around the target depth and depths close to zero, with a relatively linear increase between these two regions. Like SBP, this model was constructed using a series of points that, through experimentation, gave us the desired shape using Microsoft Excel’s trendline tool. Fig. 2 shows our SBP regression model (Eq. (1)) and DBP regression model (Eq. (2)) alongside data collected by Sutton et al.

DBP = −0.0 0 0451d3 + 0.030743d2 + 0.246959d + 5

(2)

Leaning is accounted for and penalized by considering both relative compression stroke length and residual leaning. Yannopoulos et al. (2005) examined the effects of leaning on BP during CPR in swine [28]. When investigating leaning in this study, absolute compression depth remained constant while 25% leaning was introduced, resulting in a relative stroke length of 75% ideal depth and a residual leaning depth of 25% ideal depth. They found that coronary perfusion pressure (CPP) decreased by 35% (23.3 to 15.1 mmHg), SBP decreased 12.5% (74.6 to 65.3 mmHg), and DBP decreased 26.3% (28.1 to 20.7 mmHg) when recoil was reduced to 75%. Zuercher et al. conducted a similar study on swine in 2010 [29]. For this study, rather than measuring leaning as a depth, they used weights. They initially found 180 Newtons (18 kg acting under gravity) to be the required compression force to maintain an SBP of 80–90 mmHg. They then tested 10% and 20% leaning by placing weights with mass 1.8 kg and 3.6 kg, respectively, on the chests of the pigs. At the beginning of each 3-minute epoch, they titrated their CPR compression force to achieve an SBP of 80–90 mmHg. They measured CPP drops of 14% (22 to 19 mmHg) for 10% leaning and 23% (22 to 17 mmHg) for 20% leaning. Because of the initial titration of compression force, these CPP drops were due to residual leaning and were unaffected by any change in stroke length. We used Microsoft Excel to find a logarithmic regression line to model Zuercher’s results (Eq. (3)). Using this model to forecast to 25% leaning, Zuercher’s results suggest a 26% drop in CPP for 25% residual leaning. The difference between this drop (26%) and that found in Yannopoulos’ study (35%) suggests that leaning penalties to BP must consider both relative compression stroke length as well as residual leaning.

CP P penalty (% ) = 0.067 ∗ ln (leaning + 1 )

1.62

(3)

To account for residual leaning in DBP and SBP, we used the logarithmic model for CPP penalty in Eq. (3) and scaled it based on the relationships Yannopoulos et al. found between drops in CPP, SBP, and DBP due to leaning [28]. This resulted in Eq. (4) and Eq. (5). Fig. 3 shows Zuercher et al.’s CPP drop data along with the logarithmic regression models we determined for CPP penalty (Eq. (3)), SBP penalty (Eq. (4)), and DBP Penalty (Eq. (5)) due to leaning depth.

SBP penalty (% ) = 0.0237 ∗ ln (leaning + 1 )

1.62 1.62

DBP penalty (% ) = 0.0501 ∗ ln (leaning + 1 )

(4) (5)

To account for change in compression stroke length with leaning, the relative depth is used rather than absolute to calculate

4

N. Widmann, R. Sutton and N. Buchanan et al. / Computer Methods and Programs in Biomedicine 180 (2019) 105009

Fig. 3. Residual leaning penalties for CPP, SBP, and DBP calculated based on results from Yannopoulos et al. [28] and Zuercher et al. [29].

Fig. 4. Halperin et al.’s simulated models for cerebral and myocardial blood flow with respect to chest compression rate for intrathoracic pressure fluctuations (ITP) and direct cardiac compression (CC). [30].

maximum SBP and DBP. This is the same way maximum compression depth was determined by Sutton et al. in their study [18]. Because our models for SBP and DBP are largely based on the results of this study, we felt this was the best way to account for this aspect of leaning. Despite the AHA’s recommendation to achieve a rate of 100– 120 CCs/min [2], many studies suggest very little effect on hemodynamics for changing rates between 60–150 CCs/min [18,30,31]. Halperin et al. (1987) developed software models of myocardial and cerebral blood flow for intrathoracic CPR and direct cardiac compression (Fig. 4) [30]. After performing vest CPR and openchest cardiac compression on eight dogs, Halperin et al. validated their models and concluded that blood flow during CPR is due to intrathoracic pressure fluctuations rather than direct cardiac compression. Assuming BP to be proportional to myocardial blood flow, we used their model for myocardial blood flow during intrathoracic CPR (Fig. 4) to estimate a BP penalty for changing rates. By

iteration, we determined that this curve could be represented by a logistic function for rates less than 60 CCs/min (Eq. (6)), a constant for rates in the range 60–150 CCs/min (Eq. (7)), and a negative linear function for rates greater than 150 CCs/min (Eq. (8)). We scaled each region by the inverse of the maximum blood flow to arrive at scale factor functions for SBP and DBP. These functions return a scale factor that we multiply by SBP and DBP to account for rate.



BP Scale F actor = 1/ 1 + e−0.25(rate−40)



For rates < 60 CCs/min (6)

BP Scale F actor = 1

For rates 60 − 150 CCs/min

(7)

BP Scale F actor = −0.0025 ∗ rate + 1.375 For rates > 150 CCs/min

(8)

N. Widmann, R. Sutton and N. Buchanan et al. / Computer Methods and Programs in Biomedicine 180 (2019) 105009

5

Fig. 5. Our blood pressure (BP) penalty due to pauses. Penalty is given as a percent of BP with no pauses. It is calculated as a function of average rate over the last five seconds divided by maximum instantaneous rate throughout the last five seconds. When there are no pauses, average rate will equal max rate and penalty will be 0%. After a pause of 3–5 s, average rate will be very low compared to maximum rate and BP penalty will be 33%.

Hemodynamic penalties due to pauses are based on Berg et al. (2001) [32], which investigated the adverse hemodynamic effects of interrupting chest compressions for rescue breathing during CPR on pigs. They found that when performing CPR with a 15:2 compression:ventilation ratio, a 4 second pause to deliver 2 ventilations resulted in an average CPP drop of 33% (14 vs 21 mmHg) [32]. We used this as the basis for our pause penalty model, along with the assumption that a pause of zero seconds would result in no penalty. Pause time is determined by calculating the average rate throughout the last 5 seconds and dividing by the maximum instantaneous rate from the last 5 seconds. This number (x) is then substituted into our model for pause penalties (Eq. (9)). As can be seen in Fig. 5, the model levels out at 0% and 33%, with a smooth transition in between. The model levels out at 0% because it is not possible to have a negative pause time or a corresponding negative pause penalty to BP. We chose to level out pause penalty at 33% because other studies have found less significant negative correlations between pauses and blood pressure [33]. For the first compression after a pause greater than 5 seconds, where average rate equals zero, there is a BP penalty of 33%. After five seconds of compressions at the same rate, where average rate equals the maximum rate, x equals 1 and BP penalty decreases to 0%.





BP Pause Penalty (% ) = 1/ 125x4 + 3

(9)

ETCO2 is the other physiological indicator of CPR quality. In a study of 583 in and out of hospital cardiac arrests, Sheak et al. (2015) found the average ETCO2 of survivors was 38.2 +/12.9 mmHg and non-survivors was 26.1 +/- 15.2 mmHg [20]. Like BP, ETCO2 has strong correlations with depth and is influenced little by rate [20]. Because the BP calculation already considers all the mechanical factors of CPR and Segal et al. (2017) showed that ETCO2 has a significant correlation with SBP (p = 0.709) during CPR [34], ETCO2 is calculated linearly based on SBP (Eq. (10)). This way, if the learner achieves an SBP that is indicative of survival they will also achieve an ETCO2 that is indicative of survival.

ET CO2 = 0.412 ∗ SBP

(10)

To find instantaneous BP, SBP and DBP are calculated as explained above based on the average depth, rate, and leaning of

the previous three compressions. Instantaneous BP throughout a compression is found as functions of instantaneous depth divided by average peak depth of the last five compressions. Functions for chest compression (systole) and expansion (diastole) are shown in Eqs. (11) and (12) respectively and plotted in Fig. 6. We created these functions for displaying instantaneous BP based on images of actual arterial waveforms. We decided on different elements and constants for the equations by experimenting in Microsoft Excel until the graphs had the desired form. For example, we modeled the dicrotic notch, which coincides with the aortic valve closure, using a sine function. These functions and the shape of the waveform are not affected by the quality of CPR, and likewise cannot be used as an indicator of performance. They are just for display purposes to make the BP waveform more realistic.

Compression: y = −1.1/(10x2 + 1 ) + 1.1, where: y = (BP − DBP )/(SBP − DBP ) x = Depth/(Avg Depth )

(11)

Expansion: y = (A1 + A2 + 0.463 )/1.782, where: y = (BP − DBP )/(SBP − DBP )





A1 = 1/ 150(4.5x + 0.135)4 + 1 A2 = sin (17π (x + 0.029) )/10 x = 1 − Depth/(Avg Depth )

(12)

The ETCO2 waveform is a square wave that is calculated based on the average SBP during all the compressions of the previous breathing cycle. The current default is 10 breaths-per-minute, resulting in a 6-second breathing cycle with four seconds high and two seconds low per cycle. Low is always shown as zero mmHg and high is shown as the calculated ETCO2 based on the previous cycle. The high and low values remain constant for a full cycle before being recalculated. Besides the arterial BP and ETCO2 waveforms, also on the display is a real-time bar chart that shows depth. The height of the single bar varies to display current depth. We created these three charts using MPAndroidChart, a free to use chart library for Android created by Philipp Jahoda [35]. There are also textual indicators for SBP, DBP, average depth (mm), average leaning (mm), ETCO2 (mmHg), and rate (CCs/min).

6

N. Widmann, R. Sutton and N. Buchanan et al. / Computer Methods and Programs in Biomedicine 180 (2019) 105009

Fig. 6. Compression and expansion progression functions used for displaying BP waveform in real-time. The meanings of the x and y axes are described in Eqs. (11) and (12).

Fig. 7. The GUI of the Android app includes graphs for BP and ETCO2 waveforms and depth. It also has textual indicators for SBP (mmHg), DBP (mmHg), average depth (mm), average leaning depth (mm), ETCO2 (mmHg), and rate (BPM, CCs/min).

These can all be seen in Fig. 7, which shows a screenshot of the user interface of the Android app. 4. Pilot testing To test the feasibility and effectiveness of this manikin, we conducted pilot testing at the Children’s Hospital of Philadelphia in October and November of 2017. The test subjects included 34 clinicians who were certified in basic life support and had performed, assisted with, or closely observed CPR in an ICU during an IHCA. Test subjects were given a brief overview of the manikin and its purpose (as previously described). They were presented with the manikin and Android tablet on an industrial plastic rolling cart. The cart was solid and flat so the mattress effect [36] was not a factor. After being given a verbal description of the user interface, test subjects were asked to perform compressions and utilize the simulated BP and ETCO2 feedback to judge their performance. Our testing was exempt from IRB.

The manikin received overwhelmingly positive feedback. After their trial, subjects were asked to fill out a survey, the results of which are shown in Table 1. For the first section (the first eight rows of the table), subjects were asked to gauge their satisfaction with each of the eight listed factors on a scale from 1 (very dissatisfied) to 5 (very satisfied). For the second section, subjects were asked to rate how accurately they felt depth, rate, and leaning affected BP. Options given to the subjects were “Accurate”, “Not Accurate”, and “Not Sure”. For the third section of Table 1, subjects were asked to rate the relevancy, or “importance”, of each of the eight items shown on the display. For each item, subjects were given options “Important”, “Not Important”, and “Not Sure”. The last row in Table 1 shows responses to the final question of the survey where subjects were asked, “Would this be helpful for CPR training?” They were given choices of “Yes”, “No”, and “Other” (where they could enter their own answer). The respondents all chose “Yes” for this question, indicating that they thought this Manikin would be helpful for CPR training.

N. Widmann, R. Sutton and N. Buchanan et al. / Computer Methods and Programs in Biomedicine 180 (2019) 105009

7

Table 1 Results from a pilot study conducted at the Children’s Hospital of Philadelphia to test the manikin’s performance. Satisfied (4-5)

Not Satisfied (1-2)

Neutral (3)

BP Waveform During Compressions DBP Numbers SBP Numbers Compressibility User Interface BP Waveform Scale ETCO2 Waveform Tablet Screen Size

100% 79% 94% 82% 88% 97% 68% 76%

0% 6% 0% 0% 3% 0% 9% 15%

0% 15% 6% 18% 9% 3% 24% 9%

Depth Effect on BP Rate Effect on BP Leaning Effect on BP

Accurate 97% 88% 59%

Not Accurate 0% 3% 6%

Not Sure 3% 9% 35%

BP Numerical Indicator (SBP/DBP) BP Waveform ETCO2 Numerical Indicator ETCO2 Waveform Real-time Compression Depth Graph Rate Numerical Indicator Leaning Numerical Indicator Average Depth Numerical Indicator

Important 100% 94% 88% 74% 94% 97% 79% 79%

Not Important 0% 3% 0% 9% 6% 3% 12% 15%

Not Sure 0% 3% 12% 18% 0% 0% 9% 6%

Would this be helpful for CPR training?

Yes 100%

No 0%

Other 0%

5. Discussion 5.1. Lessons learned Based on the feedback received during pilot testing, a CPR manikin that simulates BP and ETCO2 to train CPR titrated to physiological indicators is very practical. The survey feedback (Table 1) also suggests that the current version of this manikin meets this goal very well. One frequent issue we encountered while designing the manikin software was a lack of data to support the BP and ETCO2 models. Much of our current algorithm was based on animal and adult study data, which had to be interpolated for use in the pediatric model. For more accuracy, more pediatric research would have to be done on the individual and combined effects of depth, rate, leaning, and pauses on BP and ETCO2 during CPR. Lack of data was especially an issue for values outside the AHA guidelines, since it is not ethical to intentionally perform poor CPR on human subjects. One possible solution for this could be found using a mechanical simulation system. Smereka et al. (2017) used a fixed-volume arterial system to test the BP response of changing hand placements during infant CPR [37]. Using such a system, rather than live patients, could be a sufficiently accurate, more efficient, and far more ethical method for gathering BP data across a wide range of combinations of mechanical factors. Because rate does not have a significant effect on BP or ETCO2, rate-related performance cannot easily be judged based on hemodynamics alone. We therefore suggest that clinicians who are training with manikins like ours use an audible metronome to maintain a rate within the AHA recommended guidelines. This has proven to be a cheap and effective method to maintain a target rate [14,21,22,29,32]. 5.2. Future plans We plan to incorporate two other variables into our system: ventilations and epinephrine injections. The default automatic ventilation rate of the current system is 10 breaths per minute yet Sheak et al. demonstrated that ventilation rate can have a significant impact on ETCO2 [20]. Additionally, epinephrine is standard

practice during IHCA, typically delivered once every 3–5 min during adult CPR [21]. Epinephrine provides a temporary increase in BP and typically a drop in ETCO2 [21]. The incorporation of ventilations and epinephrine delivery would make the simulation more realistic. We also plan to consider compression speed (or duration) when calculating BP. In addition to investigating the effects of rate, Halperin et al. also examined how duration of compressions impacted hemodynamic output during CPR [30]. Their mathematical model determined that duration had a significant effect on intrathoracic CPR, so considering duration in our model could make it more realistic. Another factor we did not consider in this manikin is hand placement during CCs. The American Heart Association currently recommends using the 2-thumbs—encircling hands technique for infant CPR [21]. Multiple studies have shown that using this technique can positively affect compression depth and rate [38–40], as well as increase BP and CPP [37,41,42]. Considering hand placement in our algorithm could make the simulation more realistic. Another way to make the manikin more realistic is incorporating return of spontaneous circulation (ROSC). In real life, if CPR is successful, the patient’s heart will start beating on its own without CCs, in what is known as ROSC. In the current version of the training manikin, even if CPR quality is very good, the BP will always flatline when CCs are stopped. In future versions, ROSC could be achieved if the learner performs CPR exceptionally well for a predetermined amount of time. We also have plans to create different size manikins for training for different age patients. The current manikin trains for CPR on infants. Older children, adolescents, and adults will have a different recommended depth and required force because of different AP chest diameters (e.g., an adult is going to require deeper and more forceful CCs than an infant). These different age groups may have different BP and ETCO2 values that are most indicative of survival. Because of this, it would be ideal to have training manikins for a variety of age groups so that clinicians can be ready for many different scenarios. Built on its current platform, this system can be implemented with rolling refresher carts. The purpose of such carts is to give clinicians a refresher of the components of high-quality CPR

8

N. Widmann, R. Sutton and N. Buchanan et al. / Computer Methods and Programs in Biomedicine 180 (2019) 105009

including the correct depth, rate, and how to use the defibrillator. The manikin is used to practice compressions and build muscle memory in case the patient goes into cardiac arrest and they need to perform real CPR. If our manikin was included in those carts, clinicians could get more realistic practice with a simulation of the bedside monitor and the physiological values they would see during actual resuscitation. With future research, it may be possible to plug patient-specific values into the software to simulate the specific mechanical-physiological relationship for each patient allowing the clinician to practice and be prepared for the rate and depth required by a specific patient and the resulting effect on BP and ETCO2. 5.3. Conclusions A CPR manikin that simulates blood pressure and end tidal CO2 was successfully developed with acceptable relevance, performance and feasibility as a CPR quality training tool. Declaration of competing interest None. Acknowledgments Funding: We received an unrestricted donation and an unrestricted educational grant from the American Heart Association, Dallas, TX for this work. References [1] J.D. Knudson, S.R. Neish, A.G. Cabrera, A.W. Lowry, P. Shamszad, D.L.S. Morales, D.E. Graves, E.A. Williams, J.W. Rossano, Prevalence and outcomes of pediatric in-hospital cardiopulmonary resuscitation in the United States, Crit. Care Med. 40 (11) (2012) 2940–2944, doi:10.1097/CCM.0b013e31825feb3f. [2] D.L. Atkins, S. Berger, J.P. Duff, J.C. Gonzales, E.A. Hunt, B.L. Joyner, P.A. Meaney, D.E. Niles, R.A. Samson, S.M. Schexnayder, Part 11: pediatric basic life support and cardiopulmonary resuscitation quality: 2015 American Heart Association guidelines update for cardiopulmonary resuscitation and emergency cardiovascular care, Circulation 132 (18 suppl 2) (2015) S519–S525, doi:10.1161/CIR. 0 0 0 0 0 0 0 0 0 0 0 0 0265. [3] A.R. Caen, M.D. Berg, L. Chameides, C.K. Gooden, R.W. Hickey, H.F. Scott, R.M. Sutton, J.A. Tijssen, A. Topjian, É.W. Van Der Jagt, S.M. Schexnayder, R.A. Samson, Part 12: pediatric advanced life support: 2015. American Heart Association guidelines update for cardiopulmonary resuscitation and emergency cardiovascular care, Circulation 132 (2015) S526–S542, doi:10.1161/CIR. 0 0 0 0 0 0 0 0 0 0 0 0 0266. [4] M.E. Kleinman, E.E. Brennan, Z.D. Goldberger, R.A. Swor, M. Terry, B.J. Bobrow, R.J. Gazmuri, A.H. Travers, T. Rea, Part 5: adult basic life support and cardiopulmonary resuscitation quality: 2015 American Heart Association guidelines update for cardiopulmonary resuscitation and emergency cardiovascular care, Circulation 132 (18 suppl 2) (2015) S414–S435, doi:10.1161/CIR. 0 0 0 0 0 0 0 0 0 0 0 0 0259. [5] M.S. Link, L.C. Berkow, P.J. Kudenchuk, H.R. Halperin, E.P. Hess, V.K. Moitra, R.W. Neumar, B.J. O’Neil, J.H. Paxton, S.M. Silvers, R.D. White, D. Yannopoulos, M.W. Donnino, Part 7: adult advanced cardiovascular life support: 2015 American Heart Association guidelines update for cardiopulmonary resuscitation and emergency cardiovascular care, Circulation 132 (18 suppl 2) (2015) S444–S464, doi:10.1161/CIR.0 0 0 0 0 0 0 0 0 0 0 0 0261. [6] R.A. Berg, V.M. Nadkarni, A.E. Clark, F. Moler, K. Meert, R.E. Harrison, C.J.L. Newth, R.M. Sutton, D.L. Wessel, J.T. Berger, J. Carcillo, H. Dalton, S. Heidemann, T.P. Shanley, A.F. Zuppa, A. Doctor, R.F. Tamburro, T.L. Jenkins, J.M. Dean, R. Holubkov, M.M. Pollack, Incidence and outcomes of cardiopulmonary resuscitation in PICUs, Crit. Care Med. (2015) 1, doi:10.1097/ CCM.0 0 0 0 0 0 0 0 0 0 0 01484. [7] J. Del Castillo, J. López-Herce, S. Cañadas, M. Matamoros, A. Rodríguez-Núnez, A. Rodríguez-Calvo, A. Carrillo, Iberoamerican pediatric cardiac arrest study network RIBEPCI, cardiac arrest and resuscitation in the pediatric intensive care unit: a prospective multicenter multinational study, Resuscitation 85 (2014) 1380–1386, doi:10.1016/j.resuscitation.2014.06.024. [8] S. Girotra, J.A. Spertus, Y. Li, R.A. Berg, V.M. Nadkarni, P.S. Chan, American Heart Association get with the Guidelines–Resuscitation Investigators, Survival trends in pediatric in-hospital cardiac arrests: an analysis from get with the guidelines-resuscitation, Circ. Cardiovasc. Qual. Outcomes 6 (2013) 42–49, doi:10.1161/CIRCOUTCOMES.112.967968.

[9] J. López-Herce, J. del Castillo, M. Matamoros, S. Cañadas, A. Rodriguez-Calvo, C. Cecchetti, A. Rodriguez-Núñez, A.C. Álvarez, I.P.C.A.S.N. RIBEPCI, Factors associated with mortality in pediatric in-hospital cardiac arrest: a prospective multicenter multinational observational study, Intensive Care Med. 39 (2) (2013) 309–318, doi:10.10 07/s0 0134-012- 2709- 7. [10] K.L. Meert, A. Donaldson, V. Nadkarni, K.S. Tieves, C.L. Schleien, R.J. Brilli, R.S.B. Clark, D.H. Shaffner, F. Levy, K. Statler, H.J. Dalton, E.W. van der Jagt, R. Hackbarth, R. Pretzlaff, L. Hernan, J.M. Dean, F.W. Moler, Multicenter cohort study of in-hospital pediatric cardiac arrest∗ , Pediatr. Crit. Care Med. 10 (5) (2009) 544–553, doi:10.1097/PCC.0b013e3181a7045c. [11] H. Wolfe, C. Zebuhr, A.A. Topjian, A. Nishisaki, D.E. Niles, P.A. Meaney, L. Boyle, R.T. Giordano, D. Davis, M. Priestley, M. Apkon, R.A. Berg, V.M. Nadkarni, R.M. Sutton, Interdisciplinary ICU cardiac arrest debriefing improves survival outcomes∗ , Crit. Care Med. 42 (7) (2014) 1688–1695, doi:10.1097/CCM. 0 0 0 0 0 0 0 0 0 0 0 0 0327. [12] S.H. Friess, R.M. Sutton, U. Bhalala, M.R. Maltese, M.Y. Naim, G. Bratinov, T.R. Weiland, M. Garuccio, V.M. Nadkarni, L.B. Becker, R.A. Berg, Hemodynamic directed cardiopulmonary resuscitation improves short-term survival from ventricular fibrillation cardiac arrest∗ , Crit. Care Med. 41 (12) (2013) 2698–2704, doi:10.1097/CCM.0b013e318298ad6b. [13] R.W. Morgan, B. French, T.J. Kilbaugh, M.Y. Naim, H. Wolfe, G. Bratinov, W. Shoap, T.C. Hsieh, V.M. Nadkarni, R.A. Berg, R.M. Sutton, A quantitative comparison of physiologic indicators of cardiopulmonary resuscitation quality: diastolic blood pressure versus end-tidal carbon dioxide, Resuscitation 104 (2016) 6–11, doi:10.1016/j.resuscitation.2016.04.004. [14] R.W. Morgan, T.J. Kilbaugh, W. Shoap, G. Bratinov, Y. Lin, T.C. Hsieh, V.M. Nadkarni, R.A. Berg, R.M. Sutton, A hemodynamic-directed approach to pediatric cardiopulmonary resuscitation (HD-CPR) improves survival, Resuscitation 111 (2017) 41–47, doi:10.1016/j.resuscitation.2016.11.018. [15] R.M. Sutton, B. French, P.A. Meaney, A.A. Topjian, C.S. Parshuram, D.P. Edelson, S. Schexnayder, B.S. Abella, R.M. Merchant, M. Bembea, R.A. Berg, V.M. Nadkarni, Physiologic monitoring of CPR quality during adult cardiac arrest: a propensity-matched cohort study, Resuscitation 106 (2016) 76–82, doi:10.1016/ j.resuscitation.2016.06.018. [16] R.M. Sutton, S.H. Friess, M.Y. Naim, J.W. Lampe, G. Bratinov, T.R. Weiland, M. Garuccio, V.M. Nadkarni, L.B. Becker, R.A. Berg, Patient-centric blood Pressure–targeted cardiopulmonary resuscitation improves survival from cardiac arrest, Am. J. Respir. Crit. Care Med. 190 (11) (2014) 1255–1262, doi:10. 1164/rccm.201407-1343OC. [17] R.M. Sutton, S.H. Friess, U. Bhalala, M.R. Maltese, M.Y. Naim, G. Bratinov, D. Niles, V.M. Nadkarni, L.B. Becker, R.A. Berg, Hemodynamic directed CPR improves short-term survival from asphyxia-associated cardiac arrest, Resuscitation 84 (5) (2013) 696–701, doi:10.1016/j.resuscitation.2012.10.023. [18] R.M. Sutton, B. French, A. Nishisaki, D.E. Niles, M.R. Maltese, L. Boyle, M. Stavland, J. Eilevstjønn, K.B. Arbogast, R.A. Berg, V.M. Nadkarni, American Heart Association cardiopulmonary resuscitation quality targets are associated with improved arterial blood pressure during pediatric cardiac arrest, Resuscitation 84 (2) (2013) 168–172, doi:10.1016/j.resuscitation.2012.08.335. [19] R.A. Berg, R.M. Sutton, R.W. Reeder, J.T. Berger, C.J. Newth, J.A. Carcillo, P.S. McQuillen, K.L. Meert, A.R. Yates, R.E. Harrison, F.W. Moler, M.M. Pollack, T.C. Carpenter, D.L. Wessel, T.L. Jenkins, D.A. Notterman, R. Holubkov, R.F. Tamburro, J.M. Dean, V.M. Nadkarni, Eunice kennedy shriver national institute of child health and human development collaborative pediatric critical care research network (CPCCRN) PICqCPR (Pediatric intensive care quality of cardio-pulmonary Resuscitation) Investigators, association between diastolic blood pressure during pediatric in-hospital cardiopulmonary resuscitation and survival, Circulation 137 (2017) CIRCULATIONAHA.117.032270, doi:10.1161/ CIRCULATIONAHA.117.032270. [20] K.R. Sheak, D.J. Wiebe, M. Leary, S. Babaeizadeh, T.C. Yuen, D. Zive, P.C. Owens, D.P. Edelson, M.R. Daya, A.H. Idris, B.S. Abella, Quantitative relationship between end-tidal carbon dioxide and CPR quality during both in-hospital and out-of-hospital cardiac arrest, Resuscitation 89 (2015) 149–154, doi:10.1016/j. resuscitation.2015.01.026. [21] R.W. Neumar, C.W. Otto, M.S. Link, S.L. Kronick, M. Shuster, C.W. Callaway, P.J. Kudenchuk, J.P. Ornato, B. McNally, S.M. Silvers, R.S. Passman, R.D. White, E.P. Hess, W. Tang, D. Davis, E. Sinz, L.J. Morrison, Part 8: adult advanced cardiovascular life support: 2010 American Heart Association guidelines for cardiopulmonary resuscitation and emergency cardiovascular care, Circulation 122 (18_suppl_3) (2010) S729–S767, doi:10.1161/CIRCULATIONAHA.110.970988. [22] P.A. Meaney, B.J. Bobrow, M.E. Mancini, J. Christenson, A.R. De Caen, F. Bhanji, B.S. Abella, M.E. Kleinman, D.P. Edelson, R.A. Berg, T.P. Aufderheide, V. Menon, M. Leary, Cardiopulmonary resuscitation quality: improving cardiac resuscitation outcomes both inside and outside the hospital: a consensus statement from the American Heart Association, Circulation 128 (4) (2013) 417–435, doi:10.1161/CIR.0b013e31829d8654. [23] R.A. Berg, R.M. Sutton, R. Holubkov, C.E. Nicholson, J.M. Dean, R. Harrison, S. Heidemann, K. Meert, C. Newth, F. Moler, M. Pollack, H. Dalton, A. Doctor, D. Wessel, J. Berger, T. Shanley, J. Carcillo, V.M. Nadkarni, Ratio of PICU versus ward cardiopulmonary resuscitation events is increasing∗ , Crit. Care Med. 41 (10) (2013) 2292–2297, doi:10.1097/CCM.0b013e31828cf0c0. [24] S. Tanaka, A.E. White, R. Sagisaka, G. Chong, E. Ng, J. Seow, N.A. Mj, H. Tanaka, M. Eng, H. Ong, Comparison of quality of chest compressions during training of laypersons using push heart and little anne manikins using blinded CPRcards, Int. J. Emerg. Med. Tanaka Al. Int. J. Emerg. Med. 10 (1) (2017) 10, doi:10.1186/ s12245- 017- 0147- 6.

N. Widmann, R. Sutton and N. Buchanan et al. / Computer Methods and Programs in Biomedicine 180 (2019) 105009 [25] L.P. Roppolo, P.E. Pepe, L. Campbell, K. Ohman, H. Kulkarni, R. Miller, A. Idris, L. Bean, T.N. Bettes, A.H. Idris, Prospective, randomized trial of the effectiveness and retention of 30-min layperson training for cardiopulmonary resuscitation and automated external defibrillators: the american airlines study, Resuscitation 74 (2) (2007) 276–285, doi:10.1016/J.RESUSCITATION.2006.12.017. [26] M. Sainio, S. Hoppu, H. Huhtala, J. Eilevstjønn, K.T. Olkkola, J. Tenhunen, Simultaneous beat-to-beat assessment of arterial blood pressure and quality of cardiopulmonary resuscitation in out-of-hospital and in-hospital settings, Resuscitation 96 (2015) 163–169, doi:10.1016/j.resuscitation.2015.08.004. [27] M. Sainio, S. Hoppu, H. Huhtala, J. Eilevstjønn, K.T. Olkkola, J. Tenhunen, Early continuous invasive blood pressure monitoring during resuscitation attempt in human: comparison of currently recommended and alternative compression depths, Resuscitation 81 (2) (2010) S17. [28] D. Yannopoulos, S. Mcknite, T.P. Aufderheide, G. Sigurdsson, R.G. Pirrallo, D. Benditt, K.G. Lurie, Effects of incomplete chest wall decompression during cardiopulmonary resuscitation on coronary and cerebral perfusion pressures in a porcine model of cardiac arrest, Resuscitation 64 (2005) 363–372, doi:10.1016/j.resuscitation.2004.10.009. [29] M. Zuercher, R.W. Hilwig, J. Ranger-Moore, J. Nysaether, V.M. Nadkarni, M.D. Berg, K.B. Kern, R. Sutton, R.A. Berg, Leaning during chest compressions impairs cardiac output and left ventricular myocardial blood flow in piglet cardiac arrest, Crit. Care Med. 38 (4) (2010) 1141–1146, doi:10.1097/CCM. 0b013e3181ce1fe2. [30] H.R. Halperin, J.E. Tsitlik, R. Beyar, N. Chandra, A.D. Guerci, Intrathoracic pressure fluctuations move blood during CPR: comparison of hemodynamic data with predictions from a mathematical model, Ann. Biomed. Eng. 15 (3– 4) (1987) 385–403 (accessed November 28, 2017). https://link- springer- com. ezproxy2.library.drexel.edu/content/pdf/10.1007%2FBF02584292.pdf. [31] Z. Hevesi, D.N. Thrush, J.B. Downs, R.A. Smith, Hemodynamic function during CPR: effect of chest compression time and rate, Anesthesiology 89 (1998) B16, doi:10.1097/0 0 0 0 0542-1998090 03-0 0 016. [32] R.A. Berg, A.B. Sanders, K.B. Kern, R.W. Hilwig, J.W. Heidenreich, M.E. Porter, G.A. Ewy, Adverse hemodynamic effects of interrupting chest compressions for rescue breathing during cardiopulmonary resuscitation for ventricular fibrillation cardiac arrest, Circulation 104 (20) (2001) 2465–2470, doi:10.1161/hc4501. 098926. [33] S. Hoppu, M. Sainio, H. Huhtala, J. Eilevstj??nn, J. Tenhunen, K.T. Olkkola, Blood pressure during resuscitation in man-The effect of pause during rhythm analysis revisited, Resuscitation 82 (11) (2011) 1460–1463, doi:10.1016/ j.resuscitation.2011.06.005.

9

[34] N. Segal, A.K. Metzger, J.C. Moore, L. India, M.C. Lick, P.S. Berger, W. Tang, D.G. Benditt, K.G. Lurie, Correlation of end tidal carbon dioxide, amplitude spectrum area, and coronary perfusion pressure in a porcine model of cardiac arrest, Physiol. Rep. 5 (17) (2017) e13401, doi:10.14814/phy2.13401. [35] P. Jahoda, MPAndroidChart, (2016). https://github.com/PhilJay/MPAndroidChart. [36] M.F. Hazinski, J.P. Nolan, J.E. Billi, B.W. Böttiger, L. Bossaert, A.R. De Caen, C.D. Deakin, S. Drajer, B. Eigel, R.W. Hickey, I. Jacobs, M.E. Kleinman, W. Kloeck, R.W. Koster, S.H. Lim, M.E. Mancini, W.H. Montgomery, P.T. Morley, L.J. Morrison, V.M. Nadkarni, R.E. O’Connor, K. Okada, J.M. Perlman, M.R. Sayre, M. Shuster, J. Soar, K. Sunde, A.H. Travers, J. Wyllie, D. Zideman, Part 1: executive summary: 2010 international consensus on cardiopulmonary resuscitation and emergency cardiovascular care science with treatment recommendations, Circulation 122 (16_suppl_2) (2010) S250–S275, doi:10.1161/CIRCULATIONAHA. 110.970897. [37] J. Smereka, A. Rodríguez-Núñez, J.R. Ladny, S. Leung, K. Ruetzler, S. Network, A randomized comparison of three chest compression techniques and associated hemodynamic effect during infant CPR: a randomized manikin study, Am. J. Emerg. Med. 35 (10) (2017) 1420–1425, doi:10.1016/j.ajem.2017.04.024. [38] C.C. Whitelaw, B. Slywka, L.J. Goldsmith, Comparison of a two-finger versus two-thumb method for chest compressions by healthcare providers in an infant mechanical model, 20 0 0. (accessed December 4, 2018). www.elsevier.com/ locate/resuscitation. [39] S. Udassi, J.P. Udassi, M.A. Lamb, D.W. Theriaque, J.J. Shuster, A.L. Zaritsky, I.U. Haque, Two-thumb technique is superior to two-finger technique during lone rescuer infant manikin CPR, Resuscitation 81 (6) (2010) 712–717, doi:10.1016/j.resuscitation.2009.12.029. [40] H.Y. Jang, H. Wolfe, T.-.C. Hsieh, M.K. Abbadessa, S. Myers, V. Nadkarni, A. Donoghue, Infant chest compression quality: a video-based comparison of two-thumb versus one-hand technique in the emergency department, Resuscitation 122 (2018) 36–40, doi:10.1016/j.resuscitation.2017.11.044. [41] P.K. Houri, L.R. Frank, J.J. Menegazzi, R. Taylor, A randomized, controlled trial of two-thumb vs. two-finger chest compression in a swine infant model of cardiac arrest, Prehosp. Emerg. Care 1 (2) (1997) 65–67 (accessed December 4, 2018). http://www.ncbi.nlm.nih.gov/pubmed/9709339. [42] J.J. Menegazzi, ∼ Thomas, E. Auble, K.A. Nicklas, G.M. Hosack, L. Rack, J.S. Goode, Two-Thumb versus two-finger chest compression during CPR in a swine infant model of cardiac arrest, 1992. (accessed December 4, 2018). https://ac- els- cdn- com.ezproxy2.library.drexel.edu/S0196064405802124 /1- s2.0- S0196064405802124- main.pdf?_tid=2dfa541c- ce82- 49a9- b189- b462c5 57f014&acdnat=1543971344_2e18d3a064b587c5a9024b50 0 05c5b6b.