Effect of a feedback system on the quality of 2-minute chest compression-only cardiopulmonary resuscitation: a randomised crossover simulation study.

OBJECTIVE: We evaluated the quality of 2-minute continuous chest compressions (CCCs) performed by emergency staff in 30-second intervals to determine the effect of a feedback system on maintaining the quality of CCCs.
METHODS: Two hundred three physicians and nurses were randomised into two groups. Each participant performed 2-minute CCCs both with and without feedback. Group A performed CCCs under the guidance of a feedback device followed by performance without feedback, and Group B performed these tasks in reverse order. The primary outcome was the proportion of optimal compressions; i.e., compressions at both the correct rate (100-120 beats/minute) and correct depth (5-6 cm).
RESULTS: During 2-minute CCCs, the proportion of optimal compressions was poor in personnel without feedback. The proportion of optimal compressions was unchanged and low from 2.4% (interquartile range, 0.0%-32.8%) in the first 30 seconds to 3.3% (0.0%-47.7%) in the last 30 seconds of the 2-minute period. Use of the feedback device significantly improved and maintained the quality of compressions from the first 30 seconds (53.3%; 29.2%-70.4%) to the last 30 seconds (82.8%; 50.8%-96.2%).
CONCLUSION: Use of the feedback device was helpful for maintaining the quality of CCCs.

RCT Entities:   Population Interventions Outcomes

Introduction

The compression rate and depth are important parameters affecting the overall quality of chest compressions (CCs) during cardiopulmonary resuscitation (CPR). These parameters are responsible for maintaining blood flow and oxygen delivery to the heart and brain; therefore, they play a key role in increasing the rate of restoration of spontaneous circulation and neurologically intact survival at hospital discharge.[1,2] However, it is difficult to reach and maintain the optimal CC rate and depth, even for well-trained health care professionals.[3,4] Rescuers are generally unaware of the quality of CCs being delivered or the time at which the quality of CCs begins to decrease.

To ensure high-quality CCs, the 2005 American Heart Association (AHA) guidelines for CPR and emergency cardiovascular care recommended that rescuers rotate the compressor role every 2 minutes to prevent compressor fatigue and deterioration in the quality of CCs.[5] Nishiyama et al.[6] subsequently reported that the CPR quality decreased significantly faster when performing continuous CCs (CCCs) than when performing CCs in a 30:2 ratio and recommended that CPR providers change their roles every 1 minute during compression-only CPR. However, their study did not provide sufficient evidence to warrant changes in standard practices because of the limited experimental conditions.

Feedback devices can track compression metrics in real time as a means of evaluating the CC quality in real CPR or CPR training courses; moreover, they can help to close the gap between current CPR metrics and established guidelines and increase the likelihood of high-quality CPR in terms of minimal interruptions and no-flow time.[7-18] Although the positive effects of audiovisual feedback during CPR or CPR training have already been confirmed by numerous studies, the results remain controversial because of differences in the individual participants, research designs, outcome metrics, or CC criteria.[7-18]

In this study, we investigated the quality of 2-minute CCCs performed by healthcare professionals according to the 2015 AHA guidelines. The aim of this study was to determine the effect of a feedback system on maintaining the quality of CCCs.

Methods

Study setting and population

This study was conducted on 5 June 2018. The participants were physicians and nurses from a major academic conference on resuscitation and hypothermia therapy attended by national emergency health staff in Hangzhou, China. Upon enrolment, the participants were randomised into two groups. Each group performed 2 minutes of CPR with and without guidance from a feedback device. We excluded medical staff with health problems (e.g., spine or wrist injury), pregnancy, or other conditions that may have prevented proper performance of CPR as well as staff who did not complete the study.

Feedback system

We used an R Series Monitor/Defibrillator (ZOLL Medical Corporation, Chelmsford, MA, USA), which is an all-in-one electrode with an integrated accelerometer and a sternal back mode that can detect sternal movement. The instrument provides visual feedback (displaying the current compression depth and rate and the compression recoil) and verbal feedback (prompts of ‘press deeper’ and ‘chest compressions are better’) according the 2015 AHA guidelines (correct depth: 5–6 cm; correct rate: 100–120 beats/minute). The participants were instructed to perform continuous CPR (no rescue breathing) on an adult bust manikin (Little Anne manikin; Laerdal Medical, Stavanger, Norway). A chest pad was positioned on the manikin to provide information on the compression depth and rate. The manikin was placed on a hard floor, and the participants conducted CCs on the right side of the manikin in a kneeling position. When performing CPR without feedback, the participant received no verbal or visual feedback. This feedback device is compatible with defibrillators and can be kept on the patient during defibrillation.

Monitor-generated data from every participant were accessible via an SD card and measured using RescueNet Code Review software (ZOLL Medical Corporation). During the preliminary analysis, two researchers tested three different intervals (10-, 20-, and 30-second intervals) on the RescueNet Code Review software. Considering the accuracy and feasibility of these data, we chose the 30-second interval to evaluate the chronological change in the quality of CCs.

Study protocol

This was a phase 2b, crossover, randomised, active-controlled study. The flow chart is shown in Figure 1. After receiving a short instruction regarding the use of the feedback system, the participants were randomly assigned to Group A or B. Group assignment was performed as a simple (complete) randomisation. A research staff member (JY) at the study site opened the next sequentially numbered opaque envelope that contained the assignment. Another individual (AQ) unassociated with the clinical portion of the study prepared the envelopes. The randomisation sequence was created using IBM SPSS version 22.0 statistical software (IBM Corp., Armonk, NY, USA) and was stratified by centre with a 1:1 allocation. In the first part of this study, the participants Group A performed 2 minutes of CCCs with the feedback device (display visible and alert turned on). The participants could adjust the compression rate and depth according to the audiovisual information. After a 1-hour rest (washout period), the second session was conducted, and CCCs were performed without the feedback device (display covered and alert muted). The participants in Group B performed the same tasks in the reverse order.

Figure 1.

Flow chart of study design and participant recruitment. CC = chest compressions.

Sample size calculation

The sample size was calculated based on published data with the percentage of optimal CCs (relative to all CCs) as the primary outcome.[19] According to previous data,[10] the percentage of optimal CCs in the group blinded to feedback was estimated to be 40%. The hypothesised margin of superiority for those performing CCs with versus without the feedback device was set at 15%. These parameters were found to achieve a 5% significance level at 80% power in a sample size of 170 subjects per group, as a result of the online program (http://www.powerandsamplesize.com) for the two compared proportions. Considering the possibility of data loss and the substantial bias caused by the simple (complete) randomisation,[20] we enrolled 203 participants.

Statistical analysis

Statistical analyses were performed using IBM SPSS version 22.0 for Windows software (IBM Corp.). Continuous variables are presented as mean ± standard deviation for normally distributed data and as median (interquartile range [IQR]) for non-normally distributed data. The normality of the data was tested using the Kolmogorov–Smirnov test. The normally distributed data were analysed by a t-test, and the non-normally distributed data were analysed by the Wilcoxon signed-rank test. The categorical variables were analysed by the χ test or Fisher’s exact test. Changes in the percentage of optimal compressions, the percentage of compressions at the target depth, the percentage of compressions at the target rate, and the mean compression depth and rate over time were compared using repeated-measures analysis. Statistical significance was set at p < 0.05.

Primary and secondary endpoints

The primary endpoint was the absolute percentage of optimal compressions (compressions at a rate of 100–120 beats/minute and depth of 5–6 cm, simultaneously) per person. The secondary endpoint was the percentage of compressions with the target depth (5–6 cm), percentage of compressions with the target frequency (100–120 beats/min), number of ‘effective’ trials, mean compression depth, and mean rate. All performances with a mean rate of 100 to 120 beats/minute and >80% of compressions with a depth 5 to 6 cm were classified as ‘effective’ trials according to the published literature.[9] All performances were done under the direct supervision of two researchers (CW and SL). At the end of the study, the data were input into Microsoft Office Excel (Microsoft Corp., Redmond, WA, USA) by another researcher (LY) for further analysis.

Results

Study population

This study included 203 subjects who were randomly divided into Group A (n = 102) and Group B (n = 101). Four participants in Group A and eight in Group B did not complete the second part of the study; therefore, the analysis was performed on data from 191 subjects. There were no differences in sex, profession, professional title, department, last training time, or estimated frequency of CPR provision between Groups A and B (Table 1).

Table 1.

Participant demographics in Groups A and B.

	Group A (n = 98)	Group B (n = 93)	p*
Sex			0.683
Male	31 (31.6)	32 (34.4)
Female	67 (68.4)	61 (65.6)
Profession			0.764
Physician	35 (35.7)	38 (40.9)
Nurse	60 (61.2)	51 (54.8)
Other	3 (3.1)	4 (4.3)
Professional title			0.258
Senior	3 (3.1)	6 (6.5)
Semi-senior	23 (23.5)	19 (20.4)
Junior	61 (62.2)	50 (53.8)
Other	11 (11.2)	18 (19.4)
Department			0.373
Emergency	59 (60.2)	57 (61.3)
Intensive care unit	23 (23.5)	18 (19.4)
Prehospital	3 (3.1)	0 (0.0)
Other	13 (13.3)	18 (19.4)
Last training time			0.273
<0.5 years	55 (56.1)	46 (49.5)
0.5–1.0 year	33 (33.7)	34 (36.5)
<2 years	8 (8.2)	6 (6.5)
≥2 years	2 (2.0)	7 (7.5)
Frequency of CPR provision			0.296
Once per week	17 (17.3)	10 (10.8)
Once per month	26 (26.5)	34 (36.6)
Once per 3 months	30 (30.6)	23 (24.7)
Less than once per 3 months	25 (25.5)	26 (28.0)

Data are presented as n (%). Professional title: Senior, associate chief or chief/associate chief of nursing or head nurse; Semi-senior, attending doctor/nurse-in-charge; Junior, resident doctor/nurse; Other: intern/nursing assistant. CPR, cardiopulmonary resuscitation. *Chi-square χ2.

Compression quality during whole 2-minute performance

Without feedback, the proportion of optimal CCs was 6.9% (IQR, 0.4%–39.6%), the proportion of CCs at the target depth was 42.9% (IQR, 13.1%–66.5%), the proportion of CCs at the target rate was 55.2% (IQR, 7.3%–89.9%), and the proportion of effective CPR trials was 5.2%. While under the guidance of the feedback device, these proportions were 70.4% (IQR, 47.9%–85.1%), 83.8% (IQR, 68.7%–91.4%), 88.3% (IQR, 72.2%–95.8%), and 51.3%, respectively (all p < 0.001 compared with no feedback) (Figure 2). The mean CC depth and rate without the feedback device were 5.7 ± 0.8 cm and 118.0 ± 14.1 beats/minute, respectively, and those with the feedback device were 5.5 ± 0.3 cm and 112.6 ± 6.3 beats/minute, respectively.

Figure 2.

Percentages of adequate depth, adequate frequency, and optimal compressions per person in chest compressions with and without feedback and the percentages of effective CPR with and without feedback. Effective CPR: CPR with a mean rate of 100 to 120 compressions/minute and >80% of compressions of adequate depth. CPR = cardiopulmonary resuscitation.

Chronological quality of compressions with and without feedback

Table 2 shows the chronological change in the quality of CCs with and without feedback guidance. The mean CC depth and rate gradually decreased over time (p < 0.001) both with and without feedback guidance. The proportion of optimal CCs was 2.4% (IQR, 0.0%–32.8%) to 3.3% (IQR, 0.0%–47.7%) from 0–30 to 90–120 seconds without feedback guidance. The proportion of optimal CCs in the fourth 30-second interval was significantly higher than that in the first 30-second interval under guidance from the feedback device: 82.8% (IQR, 50.8%–96.2%) vs. 53.3% (IQR, 29.2%–70.4%) (p < 0.001). Without feedback guidance, the proportions of CCs at the target depth and rate were also stable and low. The corresponding parameters under feedback guidance were 71.2% (IQR, 50.0%–84.9%) to 91.3% (IQR, 66.2%–98.2%) (p < 0.001) and 75.9% (IQR, 53.3%–92.6%) from the first 30-second interval to 96.0% (IQR, 79.1%–100.0%) for the fourth 30-second interval (p < 0.001).

Table 2.

Chronological change in quality of chest compressions.

	0–30 seconds	30–60 seconds	60–90 seconds	90–120 seconds	p
Percentage of optimal compressions (%)
Feedback	53.3 (29.2–70.4)	79.6 (44.6–93.6)	83.3 (44.1–95.9)	82.8 (50.8–96.2)	<0.001#
No feedback	2.4 (0.0–32.8)	3.2 (0.0–50.8)	3.6 (0.0–43.4)	3.3 (0.0–47.7)	0.515
p	<0.001*	<0.001*	<0.001*	<0.001*
Percentage at the target depth (%)
Feedback	71.2 (50.0–84.9)	90.4 (69.4–98.1)	93.0 (72.1–100.0)	91.3 (66.1–98.2)	<0.001#
No feedback	35.4 (2.9–72.3)	48.7 (1.8–86.0)	34.6 (0.0–89.6)	30.77 (0.0–80.9)	0.630
p	<0.001*	<0.001*	<0.001*	<0.001*
Percentage at the target rate (%)
Feedback	75.9 (53.3–92.6)	92.9 (77.6–98.4)	96.1 (80.7–98.3)	96.0 (79.1–100.0)	<0.001#
No feedback	44.9 (2.9–87.7)	52.1 (2.9–96.6)	59.0 (3.6–96.2)	64.6 (4.5–96.3)	0.330
p	<0.001*	<0.001*	<0.001*	<0.001*
Mean compression depth (cm)
Feedback	5.6 ± 0.5	5.5 ± 0.4	5.4 ± 0.4	5.4 ± 0.4	<0.001#
No feedback	6.0 ± 0.8	5.7 ± 0.8	5.5 ± 0.9	5.4 ± 0.9	<0.001#
p	<0.001*	0.001*	0.225*	0.811*
Mean compression rate (beats/minute)
Feedback	113.9 ± 9.4	112.1 ± 7.3	111.8 ± 6.5	111.8 ± 6.49	<0.001#
No feedback	120.5 ± 14.3	118.1 ± 14.6	117.2 ± 14.3	116.3 ± 14.2	<0.001#
p	<0.001*	<0.001*	<0.001*	<0.001*

Values are presented as median (interquartile range) or mean ± standard deviation.

*p < 0.05 versus no feedback; #p < 0.05 with time effect.

Optimal compressions: compressions with both correct rate of 100 to 120 beats/minute and depth of 5 to 6 cm.

Discussion

In a sample of Chinese physicians and nurses, the overall CPR performance of medical workers was poor within the recommended 2-minute CC scenario in adults. When analysed in 30-second intervals, the feedback device significantly improved the endpoints and maintained the quality of compressions during 2-minute CCCs.

The Institute of Medicine, AHA, and International Liaison Committee on Resuscitation have all emphasised that high-quality CPR plays a key role in increasing the likelihood of survival. It is especially important to meet specific compression depth and rate targets. In adults, the greatest benefit is attained when the compression depth is 5 to 6 cm and the rate is 100 to 120 beats/minute in accordance with the 2015 AHA guidelines. The rates of restoration of spontaneous circulation and survival decline at values above or below these ranges.[21] In the present study, CPR performance without the feedback device was suboptimal, which is consistent with a previous study.[22] Although health practitioners are familiar with current AHA guidelines, the performance of CPR in clinical settings was poor.

Various outcome indexes have been used in different studies to evaluate the compression quality. Metrics such as the mean compression rate and depth are common endpoints; however, such a measurement cannot accurately determine whether compression falls within the target depth or target rate at any time of the CC trial. In other words, if 50% of CCs are above the target range and 50% of CCs are below the target range, the mean value is still within the recommendations. In the present study, the mean compression rate and depth with or without feedback were within the normal range. This distinction is important because guidelines have emphasised the need for CCs at both the correct rate and depth simultaneously. Therefore, the percentage of CCs with the correct rate and depth (‘optimal compressions’ in this study) and the percentages of CCs at the target depth and within the target rate are more appropriate for the evaluation of CC quality. In the present study, the intergroup differences in the proportion of optimal compressions, compressions at the target depth, and compressions at the target rate were statistically significant across all CPR periods. In addition, as an endpoint for evaluation of the entire 2-minute performance, the number of effective CPR trials was also used in our study to evaluate the quality of CPR.

Based on the analyses of 30-second intervals during 2-minute CCCs, we wanted to obtain detailed information regarding the chronological changes of CCs with and without feedback devices. In the present study, the quality of CCs without feedback was poor, which is similar to the findings of previous studies.[23,24] However, the difference is that the proportion of optimal compressions at the target depth and rate remained steady over time in our 2-minute CPR trial; further studies are needed for a proper evaluation of this finding. The compression quality with feedback increased over the first 30 seconds and was maintained at a high level thereafter, rather than exhibiting a decline.[25] Buléon et al.[10] analysed 10-minute continuous CCs minute by minute and found that the compression quality with feedback improved significantly every minute and remained stable over time, consistent with our results. A subsequent study[26] showed that CPR with real-time feedback was independently associated with improved CPR quality, increased survival, and more favourable functional outcomes after out-of-hospital cardiac arrest. A final study[27] revealed that the use of feedback not only improves the rate and depth of CCs but also allows for correct chest relaxation. Other researchers have confirmed the effects of a feedback system on CC skills and suggested that feedback systems may allow practitioners to continue delivering high-quality CPR after passing the recommended 2-minute window.[17,28] The use of a real-time audiovisual feedback device improved the quality of CCs; however, one remaining challenge is how to choose between, or unify, currently available devices, which differ in precision. Further studies will be necessary to properly evaluate such parameters.

This study had several limitations. First, selection bias might have been present during the recruitment of participants in the regional healthcare-related conference. To reduce the potential for such bias, we recruited more participants than deemed necessary based on the sample size calculations. Second, the crossover study design may have affected the results. To reduce the potential for learning effects, we implemented a 1-hour rest before the CCs in the next phase, which is longer than in previous studies.[11,29,30] Third, providing CPR to a dummy lying down on a hard floor is inconsistent with patient interactions in most clinical settings. Fourth, the compression quality may be reduced by other important interventions (e.g., intubation, defibrillation) during stressful real-life cardiac arrest situations. Finally, incomplete chest wall recoil leads to ineffective CPR, which was not assessed in our study.

Conclusion

During a recommended 2-minute CCC scenario in adults, although the mean CC rate and depth were maintained over time, the overall CC quality was poor when feedback devices were not used. Use of a real-time audiovisual feedback device is helpful for maintaining a high quality of CCs.