Interventions to improve the quality of bystander cardiopulmonary resuscitation: A systematic review

Background Performing high-quality bystander cardiopulmonary resuscitation (CPR) improves the clinical outcomes of victims with sudden cardiac arrest. Thus far, no systematic review has been performed to identify interventions associated with improved bystander CPR quality. Methods We searched Ovid MEDLINE, Ovid EMBASE, EBSCO CINAHL, Ovid PsycInfo, Thomson Reuters SCI-EXPANDED, and the Cochrane Central Register of Controlled Trials to retrieve studies published from 1 January 1966 to 5 October 2018 associated with interventions that could improve the quality of bystander CPR. Data regarding participant characteristics, interventions, and design and outcomes of included studies were extracted. Results Of the initially identified 2,703 studies, 42 were included. Of these, 32 were randomized controlled trials. Participants included adults, high school students, and university students with non-medical professional majors. Interventions improving bystander CPR quality included telephone dispatcher-assisted CPR (DA-CPR) with simplified or more concrete instructions, compression-only CPR, and other on-scene interventions, such as four-hand CPR for elderly rescuers, kneel on opposite sides for two-person CPR, and CPR with heels for a tired rescuer. Devices providing real-time feedback and mobile devices containing CPR applications or software were also found to be beneficial in improving the quality of bystander CPR. However, using mobile devices for improving CPR quality or for assisting DA-CPR might cause rescuers to delay starting CPR. Conclusions To further improve the clinical outcomes of victims with cardiac arrest, these effective interventions may be included in the guidelines for bystander CPR.


Results
Of the initially identified 2,703 studies, 42 were included. Of these, 32 were randomized controlled trials. Participants included adults, high school students, and university students with non-medical professional majors. Interventions improving bystander CPR quality included telephone dispatcher-assisted CPR (DA-CPR) with simplified or more concrete instructions, compression-only CPR, and other on-scene interventions, such as four-hand CPR for elderly rescuers, kneel on opposite sides for two-person CPR, and CPR with heels for a tired rescuer. Devices providing real-time feedback and mobile devices containing CPR applications or software were also found to be beneficial in improving the quality of bystander CPR. However, using mobile devices for improving CPR quality or for assisting DA-CPR might cause rescuers to delay starting CPR. PLOS

Introduction
Sudden cardiac arrest (SCA) poses a significant threat to our community in the industrialized world and is responsible for 420,000 and 275,000 deaths per year in the US and Europe, respectively [1,2]. It has been proved that bystander cardiopulmonary resuscitation (CPR) improves the survival rate of victims with SCA [3]. Therefore, numerous strategies have been implemented to increase the rate of bystander CPR. For example, many CPR training courses have been held to enable more laypersons to perform CPR because wider dissemination of CPR training might ultimately increase the rate of bystander CPR. Additionally, dispatcherassisted CPR programs had also been shown to increase bystander CPR rates and the clinical outcomes of victims with SCA [4,5], and were recommended to be integrated into the system of care for prehospital cardiac arrest [6][7][8]. In addition to improving the rate of performing bystander CPR, the quality of CPR is also vital for the outcome of victims with SCA. Highquality CPR is associated with survival to emergency department arrival [9], survival to hospital admission [10], survival to hospital discharge [11,12], and favourable functional outcome [12]. To further improve the outcomes of victims with SCA, it is necessary to explore which interventions can improve the quality of bystander CPR. Therefore, the aim of our study was to perform a systematic review to identify interventions that could improve the quality of bystander CPR.

Population, intervention, comparator and outcome question
We conducted a systematic review by using a predetermined protocol and reported it according to the PRISMA (Preferred Reporting Items for Systematic Reviews and Meta-analyses) guidelines [13]. The Population, Intervention, Comparator, Outcome (PICO) question of our study was: for laypersons who perform bystander CPR either in a real resuscitation situation or in a simulation setting (P), what interventions or methods except education (I), compared with no such interventions or methods (C), improved the quality of CPR (O)?

Eligibility criteria
To answer our PICO question, the inclusion criteria for our systematic review were as follows: (1) studies addressing the question "can the intervention improve the quality of bystander CPR?"; (2) studies that were randomized controlled trials (RCTs), quasi-experimental studies, before-and-after interventional studies, crossover studies, and prospective/retrospective observational studies; (3) studies that were published after 1966; (4) original articles, articles in press or short communications; (5) studies in which the participants were laypersons; (6) studies in which the setting was a real resuscitation situation or a simulation setting; (7) studies in which the outcome measures had at least one CPR quality parameter (the CPR quality parameters considered in this study were described below: chest compression depth, chest compression rate, number of chest compression, chest recoil, interruption time during CPR, ventilation volume, time to first compression/ventilation, and correct hand positioning); (8) studies focusing on the correlation between the intervention and bystander CPR quality; and (9) studies written in English. The exclusion criteria were as follows: (1) conference abstracts or articles, reviews, editorials, erratum, letters, case studies, and case reports; (2) non-human studies; (3) studies in which the participants included both laypersons and non-laypersons; (4) studies that provided survival outcomes instead of CPR quality data, which were required for this review.
The layperson in our study was defined as an individual who did not work at a hospital. Therefore, we did not include studies whose participants were doctors, dentists, nurses, emergency medical technicians, and pharmacists. We also excluded studies in which the participants were students who majored in medical, nursing or associated medical professionals. In addition, studies in which participants whose duty is to save lives, such as lifeguards and first responders in a public place, were also excluded.

Information sources
We searched Ovid MEDLINE, Ovid EMBASE, EBSCO CINAHL, Ovid PsycInfo, Thomson Reuters SCI-EXPANDED, and the Cochrane Central Register of Controlled Trials (CEN-TRAL) to acquire studies, which could answer our PICO question. The time range was set from 1966 because 1966 was the year when the American Heart Association (AHA) published the first guidelines for CPR [14]. The last time we searched was 5 October, 2018. We also checked the reference lists of studies included for additional relevant articles.
Search. Our search strategy consisted of three key concepts, including cardiac arrest, bystander/layperson, and the quality of CPR. Analogous terms for each were also used. A full search strategy is provided in S1 Table in Supplementary materials.

Study selection
Two reviewers (KYC and MJH) performed the database searching and screened out papers that were potentially relevant by reviewing titles and abstracts independently.
The article would receive full-text assessment if one of the reviewers determined that it was needed. If two reviewers had different opinions during the process of determining which article ought to be included in the final analysis, they reached an agreement after full discussion.

Data collection process and data items
After identifying the final papers included, we collected the data, using a standard data extraction form specifically adapted for this review. Extracted data included author(s), publication year, nation, identity of the participants, date of enrolment, study design, study group, outcome, evaluation methods, and funding sources.
The outcomes among studies varied considerably. The AHA 2015 guidelines recommended high-quality CPR required adequate chest compression depth (50-60 mm), adequate chest compression rate (100-120/min), full chest wall recoil, minimal pauses in chest compressions, correct hand position during compressions, and avoidance of hyperventilation [6]. In order to focus on the essential parameters of CPR quality which the guidelines recommended, we only extracted data on compression depth, chest compression rate, number of chest compressions, chest recoil, time of interruption, ventilation, time to first compression/ventilation, and correct hand positioning.

Risk of bias for individual studies
We used two tools to assess studies included. We used the "Cochrane Collaboration's tool for assessing risk of bias" to assess the quality of studies for randomized controlled studies [15] and Newcastle-Ottawa Scale (NOS) for non-randomized controlled studies [16].

Study selection
After the initial database searching, a total of 4,524 records were retrieved. After removing 1,821 duplicates, 2,703 articles were screened by reviewing the titles and abstracts. We then found 152 potentially relevant articles. After we reviewed the full text of these articles and reached an agreement between the two reviewers, 42 articles were finally included in our study . The reasons for exclusion included papers that were not in English (9), review articles (19), those included participants who were not laypersons (37), publication types that do not meet the inclusion criteria (26), and those articles with outcomes which were not CPR quality parameters considered in this study (19). The flow diagram of the articles included is shown in Fig 1. The different study designs, interventions and types of data presentations in included articles precluded further meta-analysis. For example, the authors tried to evaluate whether changing some instructional content of telephone dispatcher-assisted CPR would improve bystander CPR quality in some included studies, but those changes were different among studies [22,26,27,[33][34][35]37,40,41,52]. In other studies, the authors tried to compare the effect of compression-only CPR with that of conventional CPR on bystander CPR quality, but they presented different types of outcome data [18,25,32]. Therefore, a narrative review was performed instead.

Results of the individual studies
We grouped the interventions of studies included into three groups: (A) modifications to dispatcher-assisted CPR (DA-CPR), (B) Different methods to perform CPR, and (C) additional aids to bystander CPR. The summary of interventions in studies included is shown in Table 4.
Modified telephone DA-CPR instructions. There were ten studies that employed interventions of modified telephone DA-CPR instructions [22,26,27,[33][34][35]37,40,41,52]. Among them, seven studies found that the interventions improved the quality of CPR or shorten time to start chest compressions [27,33,34,37,40,41,52]. These interventions included adding    RCT (A) CPR with no additional instruction (n = 9) (B) CPR with receiving telephone instruction (n = 10) (C) CPR with advice over a videolink (n = 10) (D) CPR with advice by an instructor standing beside them (n = 9) The CPR standard was determined by 2 observers and by computerized analysis of manikin recordings.  [41], eliminating the instruction to remove the victim's clothing [52], and a novel instructional protocol with changing instructional content of hand position, compression depth and compression rate at the same time [33]. The other three studies showed no significant improvement in the quality of CPR, including the added instruction of "put the phone down" [22], repeated or intensified wording to remind or emphasize the importance of chest compression depth [26], and for paediatric CPR using "push as hard as you can." [35] Video-assisted DA-CPR. Six studies discussed whether video-assisted DA-CPR improved CPR quality when compared to telephone DA-CPR [17,20,28,39,46,50]. Among those, five studies used video-conferencing DA-CPR [17,20,28,39,46] and one study asked the participants to watch a video on cellular phone when performing CPR [50]. The comparison between video-assisted and telephone DA-CPR for different components of CPR quality in studies is shown in Table 5.
The conclusions of five studies were in favor of video-assisted DA-CPR [17,20,28,39,50]. Only one study demonstrated video communication was unlikely to improve telephone CPR significantly [46]. However, three studies showed that video-conferencing DA-CPR has increased time to first chest compression/rescuer breathing when compared to audio-assisted DA-CPR [20,28,39]. The other two studies had the opposite result, showing that video-conferencing DA-CPR did not have significantly longer time to first chest compression [17,46]. One study in which the participants watched a video on cellular phone during CPR had a shorter time starting chest compression when compared with audio-assisted DA-CPR [50]. Among the six studies, video-assisted DA-CPR is superior to, at least equivalent to telephone DA-CPR on the performance of correct chest compression depth, correct chest compression rate and correct hand position (Table 5).
Playing metronome sounds to the rescuer. In one study, metronome sounds were played to the rescuers through their mobile phone when performing DA-CPR, and it showed no Interventions for bystander CPR quality   improvement in the overall CPR quality even though it improved the chest compression rates [21], but was associated more with shallow compressions than the conventional telephone dispatcher-assisted compression-only CPR. Another study showed the rescuers receiving emergency medical dispatchers' instructions with metronome assistance performed better with correct compression rate than those receiving instructions without metronome assistance [58]. However, the compression depth tended to be shallower under metronome assistance. (B) Different methods of performing CPR. There were nine studies comparing different methods of performing CPR [18,24,25,30,32,36,42,44,47].

Modified telephone DA-CPR instructions
Added instructions with speakerphone activation, removing obstacles and continuous instruction [27] Improved CPR quality, but longer time to first chest compression Simplified compression only-CPR protocols [34,40] Improved CPR quality and shorter time to first chest compression Instructions with"push as hard as you can" [37] Improved compression depth compared to instruction with "push down firmly 2 inches" Modified instructions using arm and nipple line [41] Improved hand position Elimination of the instruction to remove the victim's clothing [52] Shortened time to first compression without affecting CPR quality Novel protocols with changing instructional content of hand position, compression depth and compression rate at the same time [33] Improved CPR quality compared to standard protocol Added instructions with "put the phone down" [22] Similar CPR quality compared to instruction without "put the phone down" Basic life support software programs on a mobile phone with a metronome function [43] Improved CPR quality, but longer time to first chest compression compared to no such software program.
Newly-developed CPR support applications on a mobile phone [19] Increased number of total chest compressions, but longer time to start compressions or ventilations compared to no such application Compression-only CPR vs. conventional CPR. There were five studies comparing the quality of CPR between compression-only CPR and conventional CPR with compression: breath ratio as 30:2 [18,25,32,42,44]. All of these studies showed that compression-only CPR had better CPR quality than the conventional CPR since compression-only CPR got more chest compressions, less hands-off time, and less time to first compression. However, two studies pointed out that chest compressions with appropriate depth decreased more rapidly in groups with compression-only CPR than those with conventional CPR due to increased physical fatigue [18,32]. A study performed in Japan even found that appropriate chest compression depth decreased significantly one minute after starting compression-only CPR [32]. On the other hand, two studies performed in the UK and Norway showed no difference in CPRrelated exhaustion between the compression-only CPR group and the conventional CPR group during the 10-minute test [25,42], and one study performed in the United States also revealed no difference in perceived fatigue in both groups after performing CPR for 3-minutes [44].
Dominant vs. non-dominant hands. One study explored whether there were any differences in CPR quality when laypersons compressed the victim's chest with their dominant or non-dominant hand against the chest wall [36]. This study demonstrated that, although there was a trend towards increased incidence of correct chest compressions when the dominant hand was positioned in contact with the sternum, it did not have statistical significance for a 5-minute-long CPR session.
Opposite sides vs. the same side. One study compared CPR quality between two rescuers being on the same side and being on the opposite sides in a two-rescuer situation [47]. The study showed that changing compression from the opposite sides reduced hands-off time compared to changing on the same side in prehospital hands-only CPR provided by two bystanders. The other parameters such as CPR quality were similar between the two groups.
Four hands vs. two hands. One study compared the quality of four-hand CPR to that of a two-hand CPR [30]. The study showed that four-hand chest compression during the simulated DA-CPR significantly improved the chest compression depth without affecting the compression rate among older female rescuers.
Heels vs. hands. One study compared the quality of CPR when chest compressions were applied with heels or hands [24]. The study results showed significantly more compressions meeting guidelines and fewer compressions without adequate depth in the heel group. The study concluded that heel compressions were useful in situations where a lone rescuer could not get down on the floor, could not compress the chest to adequate depth because of an Interventions for bystander CPR quality infirmity or lack of weight, or when the rescuer became too tired to continue manual compressions.
Telephone DA-CPR. There were five studies including telephone DA-CPR intervention [23,29,31,48,49]. Three of them were RCTs [23,29,31]. All of them demonstrated telephone DA-CPR improved the overall quality of CPR. One study showed that dispatcher-assisted compression-only CPR had better CPR quality than dispatcher-assisted conventional 30:2 CPR [31]. Another non-RCT showed that participants with CPR training before receiving telephone instructions had better CPR performance than those without CPR training before receiving the same telephone instructions [49].
Simple basic life support flowchart. One study demonstrated that the quality of bystander CPR could be improved significantly by a simple basic life support (BLS) flowchart offered to the rescuer [51].
Assistance via mobile phone. One study demonstrated that the BLS software program on a mobile phone, which had a metronome function, contributed to a better overall performance [43]. Nevertheless, participants with the mobile program took a longer time to call the dispatch centre and to start chest compressions. Another study showed that the newly-developed CPR support application (app) for smartphones resulted in an increased number of total chest compressions performed [19]. However, the participants with the new app delayed starting compressions or ventilations.
Assistance via real-time feedback device. Certain kinds of real-time feedback devices for CPR performance improved the quality of bystander CPR in seven studies. One study showed that the group with visual real-time feedback from PC Skillreporter performed more compressions and had a higher rate of chest compression with less hands-off time when compared with those without such feedback [42]. One study revealed that the group with visual real-time feedback from CPR meter, a device put between the victim's chest and the rescuer's hands when performing CPR, significantly improved chest compression quality [45]. Another study showed that, with feedback on compression rate from the test device, the rescuer could perform higher quality of CPR with higher compression rate and without compromising compression depth [38]. The other 4 studies also found that real-time feedback devices improved the quality of bystander CPR in a simulation setting [54][55][56][57].
Usage of M730 (a pneumatically powered transport ventilator). One study showed that M730 ventilator yielded better ventilation quality than bag-valve mask, resulting in lower delivered airway flow rate, lower airway pressure and lower volume of gas entering the stomach per breath [53].

Discussion
Our review showed that telephone DA-CPR seemed to improve the overall quality of bystander CPR. Further studies revealed that telephone DA-CPR with simplified or more concrete instructional protocols might further improve the quality of bystander CPR. It suggested that more efforts might be needed in the future not only for dispatchers to identify patients with cardiac arrest and instruct the bystander to perform CPR, but also to build up an effective instructional protocol to let bystanders perform high-quality CPR. Including effective onscene interventions into instructional protocols, such as four-hand CPR for elderly rescuers, kneel on opposite sides for two-person CPR, and CPR with heels for a tired rescuer, could improve the overall quality of CPR. In addition, it had been shown that, among participants who received the same telephone instruction, the participants receiving CPR training before had better CPR performance than those without receiving any CPR training [49]. It seemed that prior CPR education and telephone DA-CPR had a synergistic effect on the quality of bystander CPR. Therefore, to improve the survival of victims with SCA by improving the quality of bystander CPR, the importance of CPR education and the effective strategy of telephone DA-CPR should be emphasized to the community. More people who receive CPR training would improve the overall quality of bystander CPR. It could translate into better outcomes for patients with out-of-hospital cardiac arrest if an effective, evidence-based protocol of telephone DA-CPR is implemented in the community.
Although dispatcher-used video devices, including mobile phones, can be used to communicate with bystanders to improve some parameters of CPR quality, it is crucial to make the rescuer to perform CPR as fast as possible. Some studies revealed that video-assisted DA-CPR had more time starting chest compression or rescue breathing than audio-assisted DA-CPR. The increased non-flow time might compromise the clinical outcomes of patients with SCA. Further studies on video-assisted DA-CPR are needed to overcome this problem.
It is also worth noting that additional assistance from electronic devices has been shown to improve the quality of bystander CPR. Various electronic devices with apps and software are currently being developed. These devices could provide readily available instructions, sensing and feedback, which might improve the quality of bystander CPR. However, using such mobile devices also caused rescuers to delay starting chest compressions or ventilations in some studies [19,43]. An app or software on mobile devices with clear and easy-to-understand content that can be activated easily is helpful for rescuers to start CPR quickly.
Our study also revealed that compression-only CPR had better CPR quality and less time spent starting chest compression compared with conventional CPR. It has been shown that, by skipping the breathing part, compression-only CPR not only increased the compression rate of bystander CPR, but also improved the clinical outcomes of adult victims [59]. Additionally, one study showed that the quality of dispatcher-assisted compression-only CPR was better than that of dispatcher-assisted conventional CPR [31]. It was also recommended that dispatchers should provide compression-only CPR instructions to callers for adults with suspected out-of-hospital cardiac arrest by the guidelines of International Liaison Committee of Resuscitation, European Resuscitation Council and American Heart Association [6][7][8]. However, some studies pointed out the possibility of compression-only CPR exhausting the first rescuer quickly and compromising the quality of CPR before the medical personnel arrived [32,60]. In our review, there were different results among studies about whether compressiononly CPR exhausted rescuers more quickly. One study showed that the quality of chest compressions rapidly declined in compression-only CPR when compared with performing conventional CPR for only one minute [32]. Another study involving elderly volunteers also showed that the conventional CPR group had significantly more adequate depth of chest compressions than the compression-only CPR group one minute after starting CPR [60]. However, other studies revealed no differences between the two methods [25, 42,44]. The different results hinted that compression-only CPR might cause rescuers with less muscle strength, low body weight or older age to become fatigued quickly [61]. These types of rescuers may change their roles sooner than every 2 minutes to maintain the quality of chest compressions during compression-only CPR if there are two or more bystanders at the scene.
In our study, we only included studies whose participants were laypersons. Although medical personnel might also be at the scene of the event of a cardiac arrest and perform bystander CPR, most of the cardiac arrest events happened outside health care institutions, and it is reasonable to assume that bystander CPR was performed by laypersons in most cases. Hence, we only selected laypersons in our study. In addition, the interventions found to be effective in improving the quality of CPR performed by laypersons could reasonably be speculated to improve of the quality of CPR performed by the medical personnel.
From our review, strategies to improve the quality of bystander CPR were proposed as follows. Telephone DA-CPR with effective, evidence-based instructions to instruct callers to perform chest compression-only CPR for adults with SCA may be implemented first. If only one rescuer performs CPR on the scene and feels tired, CPR with heels may be suggested. If more than one rescuer performs CPR, a second rescuer is suggested on the opposite side of the first rescuer. The rescuers with less muscle strength, low body weight or older age may change their roles sooner than every 2 minutes. Four-hand CPR may be suggested to elderly rescuers. Realtime feedback devices are considered to be used during CPR if available. Easy-to-understand information about how to perform high-quality CPR via mobile devices may be provided to rescuers immediately. Finally, before dispatchers use video devices, such as mobile phones, to communicate to rescuers, a simplified communication protocol to minimize time to start resuscitation should be designed and proved effective first.
In our study, we did not perform meta-analysis due to high inconsistencies among included studies in study designs, interventions and types of data presentations. Therefore, our study cannot report any conclusive results and can only give a clue about that what interventions might be helpful in improving bystander CPR quality. Further evaluation will be needed before an intervention is implemented in a community.

Limitations
There were some limitations in our study. First, during our search for studies on interventions which could improve the quality of bystander CPR, we could not find a study performed in a real-life resuscitation situation. Whether such interventions associated with higher quality of bystander CPR could be translated to better victims' outcomes remained unknown. However, several studies have already shown that high-quality CPR could be translated to good outcomes for the victims [9][10][11][12]. Second, the inconsistencies in CPR quality measurement among studies were high. There were different parameters that were recorded in studies included, and how the researchers measured CPR quality was also different. In addition, there were inherent biases when the instructors assessed CPR performance when using video recordings or checklists by visual assessment [62]. Yet, we extracted the most commonly recognized parameters in CPR quality measurement [63]. Finally, we found that all of studies included had a high risk of bias in the blinding of participants and personnel while performing risk of bias assessment. Because the participants knew the interventions they were performing during evaluation, it might have affected their self-confidence somewhat, influencing CPR performance. This might be another cause for some bias.

Conclusion
In our systematic review, telephone DA-CPR with simplified or more concrete instructional protocols was shown to improve the quality of bystander CPR. Compression-only CPR and other on-scene interventions also seemed to improve CPR quality. Devices providing real-time feedback and mobile devices containing a CPR app or software were also found to be beneficial to CPR quality. However, using mobile devices for improving CPR quality or for assisting DA-CPR might cause rescuers to spend more time starting CPR. Additional efforts are needed to build up an effective protocol to organize these interventions to improve bystander CPR quality, further improving the clinical outcomes of cardiac arrest victims.
Supporting information S1