Dispatcher-Assisted Telephone Cardiopulmonary Resuscitation Using a French-Language Compression-Ventilation Pediatric Protocol

Objective Out-of-hospital cardiac arrest (OHCA) in pediatrics is a devastating event associated with poor survival rates. Although telephone dispatcher-assisted cardiopulmonary resuscitation (CPR; T-CPR) instructions improve the frequency and quality of bystander CPR for OHCA in adults, this support remains undeveloped in children. Our objective was to assess the effectiveness of a pediatric T-CPR protocol in untrained and trained bystanders. Secondarily, we sought to determine the feasibility and the effectiveness of ventilation in such a protocol. Methods Eligible adults with no CPR experience were recruited in a movie theater in Liege, as well as bachelor nursing students in Liege. All volunteers were randomly assigned either to T-CPR or to no–T-CPR using randomization. The volunteers were exposed to a pediatric manikin model cardiac arrest. On the basis of Cardiff evaluation test, data were collected to evaluate CPR performance. Results A total of 115 volunteers were assigned to 4 groups: untrained nonguided group (n = 27), untrained guided group (n = 32), trained nonguided group (n = 26), and trained guided group (n = 30). We found an improvement in CPR performance in the guided groups. Most volunteers (81.2%) in untrained guided group and 83.3% in the trained guided group were able to give 2 ventilations after each compressions cycle. Conclusions In a pediatric manikin model of OHCA, T-CPR instructions including mouth-to-mouth ventilations and chest compressions produced a significant increase in resuscitation performance not only among previously untrained but also among trained volunteers.

A lthough uncommon in children, out-of-hospital cardiac arrest (OHCA) has devastating complications. 1 Indeed, the survival rate at hospital discharge rarely exceeds 8% 2 despite care being conducted in accordance with chain of survival concept. 3 In adults, early bystander-initiated cardiopulmonary resuscitation (CPR) has been related to better neurological outcomes and higher survival rates. 4 However, regardless of population, most OHCAs still do not receive any CPR before the arrival of emergency medical services. 1,5 As a consequence, only 37% of infants at best actually receive early bystander CPR, whereas most OHCAs occur at home. 3 The lack of CPR knowledge or skills has been advocated to explain this low rate of CPR and subsequent poorer prognosis for OHCA occurring in private places as compared with the ones taking place in public locations. 6,7 Besides, stress, reluctance to practice mouth-to-mouth ventilation, and fear of communicable disease are other major barriers for bystander CPR practice, although these bystanders are most frequently the parents.
In adults, dispatcher-assisted compression-only CPR has been demonstrated to enhance the rate of CPR attempts. This observation was the rationale for current guidelines recommending bystander chest compression-only CPR in adults. 8 However, such recommendations raise major concerns in children 7,8 because pediatric OHCA most commonly results from primary respiratory failure. As a consequence, ventilations have always been considered as a key element of CPR sequence. 4,[9][10][11][12] Therefore, the feasibility and potential impact of T-CPR in children have been questioned in view of the significant increase in bystanders CPR rate, quality, and survival outcome demonstrated in adults. 8,13,14 Indeed, preliminary results from a large observational study of pediatric OHCA in Japan have revealed that telephone dispatcher-assisted CPR (T-CPR) might increase bystander CPR rate and survival at 1 month, without real neurological improvement though. 15 However, dispatcher-assisted CPR instructions for pediatric patients have not been documented, and there is currently no validated T-CPR protocol adapted for infant OHCA.
In our study, we aim to develop a T-CPR algorithm for infant (aged <1 year) and investigate in a manikin model of cardiac arrest whether or not dispatcher instructions could enhance the resuscitation performance of untrained and trained volunteers using a protocol combining chest compressions and ventilations as currently recommended. 16,17

Study Design
This was a prospective randomized study approved by the ethics committee of the University Hospital of Liege, Belgium (project 2014/193).

Participants and Randomization
Untrained volunteers were recruited in a movie theater in Liège. Inclusion criteria were as follows: adults 23 years or older and adults 40 years or younger (corresponding to mean age of children's parents ± SD according to Belgian statistics). Exclusion criteria were health care providers, previous basic life support (BLS) training, non-French speaking, physical handicap, or significant cardiopulmonary disease.
Previously trained volunteers were recruited among seconddegree bachelor nursing students from 2 high schools in Liège district. According to the current official study program, these students have been trained to adult and infant BLS.
As a consequence, we constituted the following 4 study groups: untrained nonguided (U-NG) or guided (U-G) groups and trained nonguided (T-NG) or guided (T-G) groups.

Interventions
On the basis of 2010 pediatric CPR guidelines in place at the time of the study 10,11 and the previously described ALERT telephone CPR protocol for adults, 18 a group of experts developed a new French CPR protocol for infant, named ALERT PEDIA (Algorithme Liègeois d'Encadrement à la Réanimation Téléphonique PEDIAtrique).
Seven dispatchers from the same center were specifically trained to this protocol and its proper use through simulated case scenario. The original ALERT PEDIA protocol and its English translation are available at http://alertpedia.phonecpr.be/.
Volunteers were randomly assigned to T-CPR vs no-T-CPR using a randomization system by drawing lots of opaque envelopes that contained a number associated to the group. The volunteers were blinded to the results of the randomization before CPR started.
According to the standardized scenario, volunteers were exposed to a 6-minute cardiac arrest scenario using a baby mannequin. For the guided groups, the dispatchers were located in 112 dispatching centers.
Data recordings began as soon as script acknowledgment ended.

Data Collection
This study used a Resusci Baby QCPR (Laerdal Medical, Stavanger, Norway), representing an approximately 4-to 6-monthold infant. The manikin was connected to a Simpad SkillReporter (Laerdal Medical, Stavanger, Norway) collecting the data that were recorded onto a laptop computer. As previously described, data performance collection grid was constructed based on the Cardiff Test version 3.1. to assess BLS performances in a standardized and objective way, with further analysis including the audio and video recordings. 19 A global performance score based on 12 binary variables was also determined. 19,20 Finally, time required for each CPR step was collected: speaker activation, check responsiveness, open airway, check breathing, first rescue breath, and first chest compression (no-flow time).

Statistical Methods
Results were expressed as means and SDs for quantitative variables with a normal distribution and as medians and 25th to 75th percentiles for asymmetric quantitative variables. Categorical variables were expressed using numbers and percentages. The normality of our variables was tested using the Shapiro-Wilk W test. Quantitative variables were compared between the 4 groups using an analysis of variance followed by a multiple comparison test or Kruskal-Wallis test. χ 2 Tests for categorical variables were performed to compare the characteristics or outcomes between groups. Multivariate analyses were performed through multiple regression to assess the relationship among CPR performances, sample characteristics, and call dispatcher features.
Statistical significance was defined as P < 0.05. All statistical analyses were conducted using Statistica software version 10.
There also was a statistical difference in sex, with a significant proportion of women among the nurses (73% for T-NG and 83% for T-G as compared with the previously untrained groups (10% in U-NG and 50% in U-G; P = 0.001; Table 1).

Initial Check for Responsiveness
There were significant differences in assessing for response (P < 0.001). The worst performance was observed in the U-NG group (0%) and T-NG group (3.9%), whereas the best performances were noted in the guided groups: T-G (43.3%) and U-G (37.5%).

Airway Opening and Check for Breathing
Airway management quality was different between the groups (P < 0.001). No participant of the unguided groups managed the airway opening and look-listen-feel procedures for normal breathing adequately, whereas 46.9% (15/32) of the U-G group and 56.7% (17/30) of the T-G group did (Fig. 2).
Mistakenly, 3 volunteers of the U-G group reported that the manikin was breathing.

First rescue breaths
The initiation of the first 5 rescue breaths was higher (P < 0.001) in the guided groups (100% in the U-G group and 93.3% in the TG group) than that in the unguided groups (34.6% in the T-NG group and 22.2% in the U-NG group, respectively).
Similarly, the effective administration of the first 5 rescue breathes was higher in the guided groups (62.5% in the U-G group and 73.3% in the TG group), whereas only 1 participant (3.9%) in the T-NG group and none of the U-NG group gave 5 effective breaths (P < 0.001). The average volume delivered by these rescue breaths did not differ between groups (U-G, 45.2 ± 15.9 mL; T-NG, 55.8 ± 16.3 mL; T-G, 53 ± 14.4 mL; not significant). Rescue breathes during CPR Mouth-to-mouth ventilation during CPR was attempted in 100% of the U-G group, 93.3% of the T-G group, and 96% of the T-NG group, but only 66.7% of the U-NG group (P < 0.001). However, the proportion of volunteers who managed to deliver 2 rescue breaths/cycle was greater (P < 0.001) in the guided groups (81.2% in the U-G group and 83.3% in the T-G group) than in the unguided groups (0% in the U-NG group and 46.1% in the T-NG group; Fig. 2). Failure to deliver proper ventilations resulted from improperly opened airway (U-NG, 48.1%; U-G, 12.5%; T-NG, 30.8%; T-G, 0%), and from ventilation less than twice per cycle (U-NG, 14.8%; U-G, 6.2%; T-NG, 19.2%; T-G, 3.3%).
As a consequence, proper ventilation rate (including 2 breathes delivered/cycle with optimal volume) remained equally low in the 4 groups

Chest Compressions
Detailed chest compression performance is supplied in Table 2. The proportion of participants performing chest compressions rate in the range of 90 to 120 per minute was higher in the guided groups (87.6% participants in the U-G group and 96.7% in the T-G group) as compared with the nonguided groups (7.4% in the U-NG group and 50% in the T-NG group; P < 0.001).

Global Performance Score
The median global CPR score was significantly different be- Multivariate regression analysis assessing the effect of sample characteristic on that global score is depicted in Table 3. All the variables from Table 3 explained 75% (R 2 = 0.7505) of the overall variability of the CPR score. This score was higher (P < 0.001) when the participant belonged to 1 of these 3 groups (as compared with the U-NG group), with a higher increase in the guided groups.

Timing
Timeline analysis as represented in Figure 3 illustrates the impact of dispatcher's guidance on CPR sequence.

DISCUSSION
Out-of-hospital cardiac arrest in pediatrics is a catastrophic event associated with low survival rates, partly because of lack of CPR undertaken by witnesses. 3 After creating an original infant T-CPR protocol, we assessed its effectiveness to guide either trained or untrained volunteer in a manikin model of cardiac arrest. Under these conditions, the ALERT PEDIA protocol increased the number of bystanders initiating structured BLS efforts and significantly improved resuscitation performance among previously untrained and trained volunteers. Although a previous study on ALERT algorithm in adults indicated a cumulative improvement of guidance and previous BLS training, in children, the guided groups seemed to perform equally whether they had previous training or not. 18 Assessing breathing in unconscious patients is difficult not only for unprofessional but also for professional rescuers. 7 In the present study and despite T-CPR assistance, airway management required a significant time to resume (U-G group, 48.5 seconds; T-G group, 44.5 seconds) and remained difficult, and hardly half of the volunteers reached current requirements (U-G group, 46.9%; T-G group, 56.9%). As previously described in other  studies, 21 3 participants erroneously reported that the manikin was breathing. Unfortunately, such failure to assess and control the airway was not a surprise. Indeed, previous work on training sessions reported inadequate airway control in 65% of pediatric intensive care professionals. 22 Dispatcher-assisted recognition of cardiac arrest is complex and associated with deleterious consequences in case of a mistake. 13,23 In these situations, agonal breathing, a common event occurring in the first moments of cardiac arrest, 7 is often misleading and delays the initiation of bystander CPR. 24 As a consequence, the American Heart Association has claimed for a simplification of the procedures involved in the evaluation of normal breathing. 10 Such a simplified approach might significantly increase OHCA detection by telephone, 25 with few adverse effects in the case of CPR undertaken in patients who are not in cardiac arrest. 26 Adopting this simplified procedure in our protocol could increase OHCA detection by telephone, with first chest compressions and ventilations that could occur earlier.
Our data concerning ventilation revealed that T-CPR increased the number of untrained volunteers aiming to initiate the first 5 rescue breathes and trying to perform mouth-to-mouth ventilation. However, although dispatchers' guidance increased its quality, the percentage of successful maneuvers remained relatively low (62.5% and 73.3% of successful 5 rescue breathes, and 81% to 82% of successful rescue breaths during CPR in the U-G and T-G groups, respectively) raising question about the actual efficacy and ventilation quality. In adequacy with previous studies, 22,27 we observed that an excessive tidal volume was delivered in all groups. Such hyperventilation was associated in a porcine model with serious adverse effects: increased positive intrathoracic pressures, decreased coronary perfusion, and decreased survival rates. 28 Finally, we have to notice that because of T-CPR instructions, the fraction of time required to perform ventilations was significantly longer in the guided groups (69% for the U-G group and 73.3% for the T-G group; P < 0.001), whereas it is essential to minimize interruptions in chest compressions less than 20% to generate blood flow. 16 These results should be considered in view of the preponderance of respiratory etiologies of pediatric OHCA, where ventilation seems as an essential component of CPR. Combined with chest compressions, rescue breath has been related to better outcomes and better survival results. 4,9,12 Despite the lack of evidence, the American Heart Association implemented the Circulation, Airway, Breathing sequence rather than Airway, Breathing, Circulation sequence. 10,16 This approach allows for detection of cardiac arrest and starting of cardiac massage 24 seconds earlier. 29 Because of an intermediate achievement of the first 5 ventilations in our protocol, combined with high median time to perform it (52.5 seconds), we question the benefit of this step in the T-CPR protocol. Adopting the Circulation, Airway, Breathing method could substantially reduce the no-flow time and delay the first ventilation by only a few seconds.
Almost all volunteers attempted chest compression, with the exception of 25.9% volunteers in the unguided, previously untrained group. This high rate should be discussed in view of reallife studies results showing that 15% to 40% of pediatric cardiac arrests did not actually receive early bystander CPR. 1,3,5 Among the volunteers attempting chest compressions, the delay was significantly higher in groups with dispatcher assistance (U-G, 176 seconds; T-G, 160 seconds) than in the unguided groups (U-NG, 17 seconds; T-NG, 16.5 seconds).
Current guidelines recommend chest compressions at a rate of 100 to 120 per minute. 16,17 Inadequate chest compression rates have been widely reported during pediatric and adult-simulated CPR. 18,23,30,31 Whereas optimal compression rate has been reached by only 50% of the T-NG group, the participants with telephone assistance met the target for 87.5% of the U-G group and 96.7% of the T-G group. These better results may be explained by the metronome guidance introduced in our protocol to help in performing adequate compression rate. 32 Chest compression depth is an essential CPR quality parameter. 10,11,16,33,34 However, the recommended depth of chest compression is not achieved during manikin model simulations, even by pediatric BLS providers. 35 In our study, the number of participants who accomplished depth target did not vary between the groups (40.7% for the U-NG group, 59.4% for the U-G group, 50% for the T-NG group, and 43.3% for the T-G group) but remained higher than in other studies. 21,30 Nonetheless, these results must be viewed with a manikin limitation, which was not designed to support compression depth greater than 44 mm and therefore reproduced with poor fidelity real thorax stiffness. 36 Current guidelines recommend the delivery of chest compressions to be performed with the tips of 2 fingers for single rescuer and with encircling technique in case of multiple rescuers. 16,17 The chest compression quality may be influenced by the technique we adopted in our protocol. Indeed, several studies on trained volunteers without T-CPR showed that the 2-thumb technique not only improved chest compression depth compared with the 2-finger technique 30,35,37,38 but also reduced the erratic positioning of the fingers. 38 However, these results must be viewed in a T-CPR context where the 2-finger technique seemed more easily understandable by telephone for a lay bystander.
This study has several key limitations. First, it was based on manikin simulations in a controlled environment and under less stressful conditions than in reality. Stress constitutes a cause of failure of T-CPR, 13,39 and the effect of our protocol on actual performance therefore remains uncertain. Second, the manikin properties reproduce less faithfully the real chest compliance characteristics and therefore can affect chest compression evaluation. 36 Next, duty cycle is an important factor to assess the quality of CPR. 40 Unfortunately, we were not been able to collect these data, as well as the time for each ventilation delivering.
Last, we chose to select persons with recent BLS training from 2 different schools. It allowed us to constitute homogeneous guided groups with a minimum of bias related to teaching, approaching the level of ideal rescuers and minimizing other bystanders' factors. Aware that this sample does not reflect the characteristics of the population, we hypothesized that if the T-CPR was useful for this newly trained population, it could also be useful for persons less recently trained.

CONCLUSIONS
The ALERT PEDIA algorithm of T-CPR instructions including mouth-to-mouth ventilations and chest compressions significantly improved resuscitation performance among previously trained or untrained volunteers. Although successful mouth-tomouth ventilations increased, it was at the cost of major interruptions in chest compressions and an overall hyperventilation. We believe that this valid algorithm should be evaluated in real-life conditions to determine its benefit in terms of quality of resuscitation and clinical outcome.