Chest compression quality and patient outcomes with the use of a CPR feedback device: A retrospective study

Feedback devices were developed to guide resuscitations as targets recommended by various guidelines are difficult to achieve. Yet, there is limited evidence to support their use for in-hospital cardiac arrests (IHCA), and they did not correlate with patient outcomes. Therefore, this study has investigated the compression quality and patient outcomes in IHCA with the use of a feedback device via a retrospective study of inpatient code blue activations in a Singapore hospital over one year. The primary outcome was compression quality and secondary outcomes were survival, downtime and neurological status. 64 of 110 (58.2%) cases were included. Most resuscitations (71.9%) met the recommended chest compression fraction (CCF, defined as the proportion of time spent on compressions during resuscitation) despite overall quality being suboptimal. Greater survival to discharge and better neurological status in resuscitated patients respectively correlated with higher median CCF (p = 0.040 and 0.026 respectively) and shorter downtime (p < 0.001 and 0.001 respectively); independently, a higher CCF correlated with a shorter downtime (p = 0.014). Overall, this study demonstrated that reducing interruptions is crucial for good outcomes in IHCA. However, compression quality remained suboptimal despite feedback device implementation, possibly requiring further simulation training and coaching. Future multicentre studies incorporating these measures should be explored.


Patient population
Waiver for informed consent was granted by the SingHealth Centralised IRB (Reference Number: 2019/2496).All patients in SKH and Sengkang Community Hospital (SKCH) who had a cardiopulmonary collapse in areas covered by the inpatient CBT were included.These areas included the general ward, high dependency unit, endoscopy suite, dialysis centre, cardiovascular laboratory and SKCH inpatient wards.
Patients excluded were those who: (1) had achieved ROSC upon the CBT arrival; or (2) were on a Do Not Resuscitate (DNR) order.Peri-arrest activations which ultimately did not go into cardiac arrest were excluded from the analysis.

Measured outcomes
Primary outcome variables were CCF, proportion of compressions in target, mean compression rate and proportion of compressions in target rate, mean compression depth and proportion of compressions in target depth, and mean release velocity (RV).CCF was defined as proportion of time during resuscitation that chest compressions were performed.Compressions in target were defined as compressions that met both target rate and depth concurrently.RV was used as a surrogate marker for chest recoil.The compression targets were assessed using recommendations by AHA 5 .
Secondary outcome variables were downtime, survival and neurological outcomes.Downtime was defined as the interval from cardiopulmonary collapse to sustained ROSC, in which spontaneous circulation had been achieved without the need for further chest compressions for at least 20 min 25 .Survival of patients was recorded at hospital discharge.
A patient's neurological function was assessed using the Cerebral Performance Categories (CPC).The CPC is a commonly used modality to prognosticate long-term neurological function after a cardiac arrest.The categories are: (1) good cerebral performance; (2) moderate cerebral disability; (3) severe cerebral disability; (4) coma, vegetative state; and (5) dead 26 .

Statistical analysis
Data was recorded using Microsoft® Excel® 365 software (Microsoft Corporation, Redmond, WA, United States of America).Missing data were reconciled where possible.Data which were unable to be reconciled were excluded from the analysis.Analysis was performed using IBM SPSS Statistics 28 software (IBM Corporation, Armonk, NY, United States of America).Shapiro-Wilk test was used to determine if continuous variables were normally distributed.
Due to the small sample size within subgroups, continuous variables were presented as median with interquartile range (IQR).Categorical variables were presented as numbers or percentages (%).
Mann-Whitney U and Kruskal-Wallis tests were performed to determine if there were significant differences between 2 or more than 2 outcome categories measured on a continuous scale.Measurements for compression parameters and downtime were collected.The data were analysed using the Kruskal-Wallis test to detect significant differences in medians of parameters and downtime for all survival outcome categories.Post-hoc tests were

Patient prognosis after ROSC correlated with CCF and downtime
Next, we investigated if resuscitation parameters were associated with patient survival.The patients who attained ROSC were categorised into those who: (1) did not survive till hospital discharge; and (2) survived till hospital discharge.
The overall median CCF and downtime for patients who attained ROSC were 86.6% (77.0-92.3%)and 15.0 (7.5-20.3)minutes respectively.Amongst these patients who attained ROSC, the median CCF was higher in those who "survived till hospital discharge" (92.3%) than those who "did not survive till hospital discharge" (86.8%).Further, the median downtime was also shorter in the former, by a difference of 11.5 min.These differences in CCF and downtime, in terms of patients who survival till hospital discharge or not, were statistically different (p = 0.04 and < 0.001 respectively).However, no statistically significant difference was found for depth, rate and RV for these survival categories (p = 0.69, 0.96 and 0.79 respectively) (Table 3).

CPC correlated with CCF and downtime respectively
To investigate the extent in which neurological status is affected after resuscitation, we explored how the compression parameters correlated with CPC.We observed statistically significant differences in median CCF and downtime with CPC at discharge (p = 0.03 and 0.001 respectively).Median CCF (92.3%) was higher in patients with CPC = 1 than in patients with CPC = 5 (86.4%).Moreover, median downtime (6.0 min) was also shorter in patients with CPC = 1 than in patients with CPC = 5 (17.5 min).However, no statistically significant difference was observed in compression depth, rate and RV with CPC at discharge (p = 0.68, 0.97 and 0.77 respectively) (Table 4).

Discussion
This study examined CPR quality with a feedback device and how CPR performance correlated with outcomes of IHCA patients.Herein, under the guidance of the CPR feedback device, we identified that increased CCF and shorter downtime were associated with improved CPR survivability and neurological prognosis of patients.This study provided nascent insights and thorough evaluation on the execution of CPR.7][8] .Similarly, the former reported an average depth of 4.3 cm while the latter observed an average depth of 3.4 cm, which improved to 3.8 cm with feedback [6][7][8] .Compression depth in target also improved from 28 to 53% with feedback, as demonstrated by studies in 2005 and 2006 respectively 7,8 ..Based on the abovementioned study and AHA 2015 guidelines, the CPR quality in this study was suboptimal.There was inadequate chest recoil and the median compression rate was also above target 5 .Overall, there was a low percentage of compressions in target.These occurred despite real-time feedback.Achieving full chest recoil is difficult 28 .High levels of self-reported stress were also reported to be associated with decreased CPR performance 29 .Even with the implementation of the feedback device, quality was still inadequate and calls for further intervention to improve CPR quality.

Compressions required further guidance
Thus, one solution is for CBT to introduce the role of a CPR coach.The CPR coach deduces compression quality based on the feedback generated from the CPR feedback device and first-hand observation to guide the provider in making adjustments to achieve the CPR targets.The CPR coach may be the CBT leader or assigned to another CBT member.Over the years, some had attempted to review the effectiveness of CPR coaching in resuscitation.While these studies were mostly limited to OHCA and paediatric IHCA, they demonstrated that CPR coaching improved adherence to resuscitation guidelines [30][31][32][33] .Most pertinently, a 2022 South Korean study which looked into a novel OHCA resuscitation protocol (which included CPR coaching) showed substantial improvements in the rates of prehospital ROSC, survival till hospital discharge and favourable neurological status 34 .It is crucial that future reviews of CPR data explore whether the introduction of a CPR coach, especially in adult resuscitation with prolonged downtime, leads to improved compression parameters and if such improvements correlate with better survival and neurological outcomes.
Moreover, with the variability of CPR quality between providers, more training to standardise quality is likely necessary.Simulation training was reported to be most effective in training to improve CPR quality 16,35 .Therefore, training sessions should also include the CPR coach in simulations to enable the CBT to incorporate this role while mimicking stressful real-life conditions.

Minimising interruptions during CPR is essential for good outcomes in IHCA patients
This study demonstrated that a higher CCF was associated with shorter downtime.Current evidence corroborates with our findings where a higher CCF reduces mortality and morbidity in patients who suffered from a cardiac arrest [36][37][38] .Additionally, a shorter downtime was also associated with better survival to hospital discharge and neurological status, supporting our findings 39,40 .
It is noteworthy that the current AHA recommendation to minimise interruptions during CPR are based on papers which studied OHCA patients 5,41 .Past studies also did not find reduced interruptions to be associated with improved survival and neurological status in IHCA patients 38 .Hence, it is reasonable to postulate that achieving uninterrupted compressions is important for good outcomes in IHCA patients, and this further strengthens AHA recommendation to minimise interruptions during CPR for OHCA and now IHCA patients (Fig. 1).
With a median CCF of above 80%, this study has demonstrated that such a target is achievable within the hospital setting when performing resuscitations on IHCA patients.This supports current expert consensus 41 .This is an important finding since high CCF is associated with better patient outcomes.
However, it is currently unclear whether CCF decreases over time with prolonged CPR due to fatigue, or whether having a high CCF directly reduces downtime.Therefore, future studies can further explore this association.

Other confounders
The use of analysis of means difference does not exclude other confounders from contributing to the differences.Therefore, spurious correlations may be generated.For example, the quality of post-resuscitation care may be different among various patients, which is an important factor that affects the prognosis of patients 42 .It is thus included as one of the steps in the chain of survival for all cardiac arrest patients 43,44 .Moreover, non-modifiable variables such as the patient's gender, age and prior comorbidities affect outcomes, but cannot be reliably excluded for comparison and discussion 45

Limitations
Firstly, the sample size was small (n = 64).As such, this study is potentially underpowered to conclude the effects of compression parameters.A future multicentre study can increase the sample size.Secondly, there was significant data loss (n = 34; 31%).Data loss over wifi is being rectified for future evaluations.Thirdly, this study was unable to verify self-reported variables like patient downtime and CPC.Such variables were entered as a free-text note.Unlike retrieving compression parameters via the manufacturer's software, a retrospective accuracy check for free-text notes was not possible.Lastly, this study was unable to capture pauses during CPR (for example, pulse check, perishock and intubation) and time from collapse to CPR.These affect survival and patient outcomes 3,[46][47][48] .The data collection methodology will be revised to include these in future studies.This study was ultimately a retrospective one.It cannot control for all confounders or support causal relationships 49 .

Areas for further research
Only a few studies utilising a feedback device have been conducted on humans 6,8,[19][20][21]50 . Fewwere conducted on IHCA 6,[19][20][21] .This is perhaps the first study to review the effectiveness of a CPR feedback device which provided depth, rate and RV feedback during IHCA. At the ime of writing, there are ongoing studies using real-time feedback on these parameters conducted in Japan and England 51,52 .

Conclusions
CPR quality was suboptimal despite using a CPR feedback device to guide resuscitation.CPR providers and the CBT included in this study require further training with on-site CPR coaching.Despite its limitations, this study revealed that achieving higher CCF is achievable and essential for good outcomes in IHCA patients who achieve ROSC, further strengthening AHA recommendation to minimise interruptions during CPR.Nonetheless, future multicentre studies with a greater sample size exploring the clinical effects of CPR feedback device implementation will be helpful.However, it should be emphasised that providing high-quality CPR is merely one factor which impacts the outcomes of IHCA patients.Patient outcomes are also dependent on a robust system involving training, protocols and post-resuscitation care 16,35,43,44 .Therefore, optimisation of post-resuscitation care and review of current systems are also essential in improving the prognosis of cardiac arrest patients 43,44,53 .

Table 1 .
Characteristics of patients included in this study.SD: Standard Deviation; IQR: Interquartile Range.

Table 4 .
1ssociations between the various chest compression parameters and downtime, and CPC.CPC: Cerebral Performance Categories; IQR: Interquartile Range.1Mann-WhitneyU test.Significant values are in bold.

Table 5 .
AssociationsKovacs et al. found that an RV equal to or exceeding 40 cm/s was associated with improved survival and neurological outcomes 2etween the various chest compression parameters and downtime for patients who attained ROSC.ROSC: Return of Spontaneous Circulation; IQR: Interquartile Range.1Kruskal-Wallistest.2Bonferroni correction was performed for multiple tests between groups.There was statistically significant difference in CCF for those who had a short downtime than those who had a long downtime (p = 0.012).Significant value is in bold.Vol:.(1234567890)Scientific Reports | (2023) 13:19852 | https://doi.org/10.1038/s41598-023-46862-xwww.nature.com/scientificreports/ .