The association between the experience of lay responders and response interval to medical emergencies in a rural area: an observational study

Aim The aim of this retrospective observational study was to determine how response intervals correlated to the experience of the community first responders (CFRs) using data collected from the Danish Island of Langeland via a global positioning system (GPS)-based system. Methods All medical emergency calls involving CFRs in the time period from 21st of April 2012 to 31st of December 2017 were included. Each emergency call activated 3 CFRs. Response intervals were calculated using the time from when the system alerted the CFRs to CFR time of arrival at the emergency site measured by GPS. CFRs response intervals were grouped depending on their level of experience according to ≤ 10, 11–24, 25–49, 50–99, ≥ 100 calls accepted and arrived on-site. Results A total of 7273 CFR activations were included. Median response interval for the CFR arriving first on-site (n = 3004) was 4:05 min (IQR 2:42–6:01) and median response interval for the arrival of the CFR with an automated external defibrillator (n = 2594) was 5:46 min (IQR 3:59–8:05). Median response intervals were 5:53 min (3:43–8:29) for ≤ 10 calls (n = 1657), 5:39 min (3:49–8:01) for 11–24 calls (n = 1396), 5:45 min (3:49–8:00) for 25–49 calls (n = 1586), 5:07 min (3:38–7:26) for 50–99 calls (n = 1548) and 4:46 min (3:14–7:32) for ≥ 100 calls (n = 1086) (p < 0.001). There was a significant negative correlation between experience and response intervals (p < 0.001, Spearman’s rho = -0.0914). Conclusion This study found an inverse correlation between CFR experience and response intervals, which could lead to increased survival after a time-critical incident.


Introduction
Critically illness should be treated as fast as possible in case of life-threatening events such as sudden cardiac arrest to improve their odds for better mental outcome and survival. To shorten response intervals in areas with prolonged arrival times for emergency medical services (EMS), it has been proposed that citizens trained as community first responder (CFRs) can be activated to perform cardiopulmonary resuscitation (CPR) before EMS arrival. A randomized study carried out in Sweden showed an absolute increase of 14% in bystander-initiated CPR with the use of a global positioning system (GPS)-based message system [1]. Other lay responder systems have been successfully used in the Netherlands [2,3], Switzerland [4], France [5] and are under development in the United Kingdom [6] and North America [7].
The global resuscitation alliance postulated a frame of survival as an extension of the chain of survival. The frame of survival states that leadership and training are needed to improve the quality of resuscitation and thereby lead to a culture of excellence [8]. One way to improve quality is faster initiation of CPR and we theorize that more experienced and trained CFRs arrive faster at the emergency site. We assume that more experienced CFRs would have increased knowledge of the location of the local AED-cabinets and that they would have established routines to arrive quicker at the site. On the Danish island of Langeland, a GPS-based community first responder system was initiated in 2012 to locate and dispatch CFRs alongside standard emergency medical services response [9].
In this retrospective observational study, our main purpose was to determine a possible correlation between response intervals and experience of the community first responders by using data collected from the island of Langeland.

Study design and approval
This retrospective observational study presents data collected from a GPS-based system on the Danish Island of Langeland. Approval for this study was given by The Danish Data Protection Agency (Journal no. 17/32047) and Danish Patient Safety Authority under the administration of Danish Health Authority (no. 3-3013-2848/1, ref.LOSC) and the volunteers accepted during their registration in the project, that data was collected from their smartphones, stored on the server and used for research purposes. In Denmark, ethical approval is not needed for this kind of study.

Setting
The island of Langeland has a population of approximately 12,000 inhabitants, but during summertime approximately 260,000 tourists visit the holiday island. It is 52 km long and 11 km at its widest and is bridge connected to the mainland [9]. There are one town and four larger villages, and the population is a typical island community, where most of the island's inhabitants have lived most of their life on the island. Langeland has one ambulance and one paramedic vehicle and in case the ambulance is reserved for another duty, response intervals increase drastically as another ambulance needs to arrive from the city of Svendborg placed on the mainland 20-50 km away [9]. The island has 96 AEDs placed strategically with no more than 2 km between each of them.

GPS system
A GPS-based system to activate CFRs was initiated in 2012 on Langeland. Each time an ambulance was dispatched to an emergency (not only cardiac arrest), the system identified the nine closest CFRs in a five km radius from the emergency site via GPS and notified them by an alarm sent to their smartphone. The CFRs could then choose to accept or decline the call. Based on the choices made, the system within 20 s identified the three closest CFRs and sent a map to the location of the emergency site and tasks. Two CFRs were dispatched directly to the emergency site, while the third was sent to pick up an automated external defibrillator (AED) before proceeding to the emergency site. These tasks are selected by the system based on the CFRs GPS location and nearest AED. After each activation, there was a possibility for a debriefing for the CFRs [9,10].

Recruitment of lay responders
CFRs were recruited to the project through local advertisement and were mostly laypersons but included some off-duty healthcare providers. To become CFR certified, the CFRs needed to be at least 18 years of age and complete a 12-hour basic life support training course provided by Langeland AED Association. The CFRs were trained by ERC certified staff. Each CFR is manually approved by an administrator before being able to accept assignments in the system. To remain active as a CFR an annual 3-hour retraining course is required.

Data collection
All medical emergency calls involving CFRs on Langeland in the time period from 21st of April 2012 to 31st of December 2017 were included. CFRs were excluded from analysis if they did not arrive on-site or if their data did not appear to be appropriate (arrived onsite > 2 hours after accepting the call).
The time for arrival at the emergency site was automatically logged by the GPS-based system. Time and date for when the system alerts the CFRs, the emergency site's address and CFRs arrival time at the emergency site were collected from the project-server. Emergency service records were acquired and were used to correct any missing data as far as possible in the CFR information.

Definition of variables
CFR experience was measured as the number of times a CFR was activated and arrived to the emergency site, and was divided into the following groups based on milestones from the Langeland AED Association as the association gives rewards based on the categories: ≤ 10, 11-24, 25-49, 50-99, ≥ 100 calls accepted. We presumed these milestones to be applicable and good indicators for CFR experience. These groups were used to investigate the link between CFRs experience and response interval; e.g. the response intervals of a CFR who was called out 34 times was placed in the category ≤ 10 calls the first 10 times he/she was called out, then placed in the category 11-24 calls for the next 14 calls, and finally placed in the category 25-49 for the last 10 calls. Response intervals were calculated using the time from alerting the CFRs to CFRs time of arrival at the emergency site. In the multiple regression analysis, the analysis was carried out without the described experience groups and instead data were re-assessed with the number of each CFR's responses analyzed as a continuous variable instead of in groups.

Variables of interest
The median response interval could be influenced by multiple variables and was therefore analysed by the effect of a CFR acquiring an AED on the way to the emergency site, the time of day, season and the town of Rudkoebing. Time of day was categorised according to hours at work (7:00 to 15:00), spare time (15:00 to 23:00) and nighttime (23:00 to 7:00). The season was defined meteorologically as winter (December, January, February), spring (March, April, May), summer (June, July, August) and fall (September, October, November). Rudkoebing is the biggest town on Langeland, and therefore potentially had shorter distances to emergencies which may have had an impact on response intervals when compared to the rest of Langeland.

Statistics
The data were analysed using Stata 16 and statistical significance was set to p = 0.05. To validate if data were normally distributed a Q-Q plot was used. Non-normally distributed outcomes were reported as median with interquartile ranges (IQR). Categorical data were described as absolute numbers and percentages. A Spearman's rank-order correlation was used to assess if there was a monotonic relationship between experience and response intervals. To evaluate non-normally distributed outcomes, a Kruskal Wallis test was used if there were three or more unmatched groups and a Mann-Whitney U-test was used to test for statistically significant differences between two groups. A multiple regression analysis was used to analyse the association between CFR response intervals and the independent variables (experience, AED, year, time of day, season, Rudkoebing vs. the rest of Langeland). A post-hoc analysis was used to determine if the correlation between experience and VFR response interval still prevailed when using only data from the rural region of Langeland and with only data from the town of Rudkoebing.

Results
Our study included 3063 emergency calls with activated CFRs resulting in 9189 possible CFR activations, of which 7273 activations were included and 1916 were excluded. We excluded 712 activations with no response intervals generated, 177 activations had no response intervals available because no willing CFRs were available within a five kilometre search radius. The remaining 535 possible activations had no response intervals due to only one or two of the three potential CFRs being available within a five kilometre radius. The CFRs did not arrive on-site in 1192 activations and were thus excluded. We also excluded 12 activations with a response interval over 2 hours (Fig. 1).
At least one CFR arrived at the emergency site in 3004 emergency calls (98.1%). An AED arrived at the emergency site in 2594 emergency calls (84.7%). The annual mean number of alerts for each CFR in the study period was approximately 6.25 alerts divided between 204 CFRs.
While the number of CFRs roughly remained the same, the number of experienced CFRs increased and the number of less experienced CFRs declined over the study period (Fig. 3). A Spearman's correlation was conducted to assess the relationship between experience and response intervals and a significant negative correlation was found (p < 0.001, Spearman's rho = -0.0914).
A multiple regression was run to predict CFR response interval from experience, AED, year, time of day, season, Rudkoebing vs. the rest of Langeland. CFR response interval was significantly inverse correlated with experience. Bringing an AED, nighttime and summer were associated with significantly increased response intervals and the town of Rudkoebing significantly reduced response intervals (Table 3). These variables statistically significantly predicted CFR response interval, F (13, 7259) = 17,71, p < 0.001, R 2 = 0,031 (Appendix Table 4).  Official average ambulance response interval on Langeland was 11:18 min and 12.5% of ambulances arrived after more than 20 min, for comparison the official average ambulance response interval in the Region of Southern Denmark was 7:54 min and 1.2% arrived after 20 min [11].

Discussion
Results from this study of response interval in CFRs show a significant correlation between experience and response intervals. While it seems that there is no substantial difference in median response intervals for CFRs with less than 50 calls accepted and arrived, the most experienced CFRs had median response intervals that were more than one minute shorter when compared to the group with minor experience (Fig. 2). This comparison is strengthened by the good balance of all the group sizes. Increasing experience was negatively correlated to response intervals when analysed according to picking up an AED or time of day through the experience groups. The significant negative correlation between CFR response interval and increasing experience could be replicated when using multiple regression (Table 3). Picking up an AED on the way to the emergency site increased median response intervals by 20-40 s. Increased experience had an impact on median response interval depending on the time of day thus activations at night were reduced by 1:43 min, while working hours was decreased with 1:08 min and spare time with only 46 s when comparing the lowest experienced group with the most experienced. Only the season of summer did impact response intervals significantly, which may be due to increased number of tourists and therefore more traffic and/or because some CFRs are on vacation leaving fewer CFRs on the island to accept the calls. The number of CFR activations is approximately the same (spring 108 pr. month vs. summer 111 pr. month vs fall 105 pr. month vs. winter 101 pr. month). The town of Rudkoebing had a significant shorter CFR response interval compared to the rest of the more rural Langeland. The multiple regression had a low r-squared, indicating that more variables are needed to fully verify the association between experience and response intervals.
In our multiple regression analysis, the town of Rudkoebing was a significant factor in predicting response intervals. Rudkoebing is the most densely populated area and could have interfered with our data since VFRs in Rudkoebing could have had more emergency calls and shorter travel distances. To determine if the negative correlation between experience and response intervals was a result of faster response intervals in Rudkoebing we decided to carry out a multiple regression without the data from the town of Rudkoebing to see if the negative correlation could be replicated in the rural region of Langeland. In the multiple regression without the town of Rudkoebing, the correlation between experience and VFR response intervals is still significant (data shown in Appendix Tables 4, 5 and 6).
More experience could mean that CFRs establish routines to quickly get to the emergency site as they probably are better prepared with established routines and could have an increased local knowledge of AED placements and local addresses. Another possible cause for the decrease in response intervals with increasing numbers of activations could be that the most dedicated CFRs are also the ones with the fastest response intervals. As they possibly are more dedicated, they would probably stay in the project for a longer period meaning that the shorter response intervals found could be a result of the less dedicated CFRs leaving the project before getting experienced. However, we do not believe this is the cause, as a steady decline in unexperienced CFRs over the study period makes it unlikely that a large amount of CFRs left and entered the program (Fig. 3). Also, the total number of CFRs did not change much either while the number of experienced CFRs on Langeland increased over the years which further makes this unlikely (Fig. 3). Leadership, training, quality improvement and culture of excellence are four important elements deemed essential to support the chain of survival. These four elements are described as the frame of survival and made by the global resuscitation alliance as a tool to define a high-quality EMS system [8]. The frame of survival is also reflected in the Utstein Formula for survival, which states that medical science, educational efficiency and local implementation together equal survival [12]. GPS-based programs like the one initiated on Langeland are a way to provide and improve these four elements from the frame of survival. Strong leadership is shown by implementation of systems to reduce response intervals which multiple countries have done over the last years [13]. The CFRs on Langeland are trained by experienced professionals before they can be activated as part of the system. Leadership  No study has explored the correlation between the experience of CFRs and their response intervals but multiple studies have been made with applicationbased systems [2][3][4][5][6][7]. In comparison, the median response interval from when the system alerts the CFR to CFR arrival at the emergency site, the Swedish study from Berglund et al. reported 2:22 min for their lay responders to arrive on-site [14] compared to 5:29 min in our study, although we found a median response interval of 4:05 min if only counting the first CFR on-site. For a lay responder to arrive with an AED, Berglund et al. reported 5:17 min in median response time [14] while an AED arrived on-site after 5:45.5 min in our study. A possible explanation for their faster response times could be their higher population density (347 inhabitants/km 2 in Stockholm County vs 48 inhabitants/km 2 on Langeland). Our results also support this hypothesis as Rudkoebing which is the biggest town on Langeland has shorter response intervals compared to the rest of Langeland. We believe a higher population density results in shorter response intervals caused by shorter CFR travel distances as more CFRs are available in a higher populated area. Another possible explanation for our longer response intervals is a bigger search radius in our application, as their system used a search radius of 1200 m without AED and 2400 m with AED [14], compared to the Langeland GPS system with a maximum search radius of 5000 m. Thirdly, their system did not activate during nighttime which both our study and another study have shown to increase response intervals [4].
On Langeland in the period of 2012 to 2017 a CFR arrived before the ambulance on the emergency site in 85% of cases [10]. Since studies have highlighted the importance of early CPR [15][16][17], this study found that the first CFR was on-sight after 4:05 min and thereby describes an effective way to counter the effect of a long EMS response interval. We also found it as a reliable system since a CFR arrived in 98.1% of the calls but this could be extraordinary as a result of the island's community since Berglund et al. reported 58% [14] and Derkenne et al. 18% [5] of CFRs to arrive at the emergency site. Further studies are needed to enlighten the attendance of CFRs to arrive on-site in GPS based systems, but GPS based systems could be efficient at providing community first responders at emergency sites.
Our data suggest that CFR experience is an important factor in improving response intervals. This could potentially increase survival after a time-critical incident such as an out of hospital cardiac arrest, especially in a rural area with prolonged ambulance response intervals like Langeland. More experienced CFRs would probably also possess better first aid skills and thereby improve the outcome. Continued training and focus on keeping the CFRs active should therefore not be underestimated.

Limitations
Since the island of Langeland is a small rural part of Denmark, data may not be applicable to other regions or countries. Given no other studies to our knowledge have explored the correlation between experience and response intervals more studies are needed to substantiate our theory. Some CFRs arrived further than 50 m away from the destination resulting in the GPS system not registering them as arrived despite them helping at the location [18]. If a first responder for example, stayed at the cross-section to indicate the direction for the ambulance, they may not register as arrived at the destination. Some data may also have been lost as the system does not register the CFR arrival time if the GPS signal was missing. This could result in some data missing, shown the system as less reliable than in reality and affected CFRs experience and made some CFRs less experienced than they actually were.

Conclusion
This retrospective study found a significant inverse correlation of CFR experience and response intervals. Using the Langeland GPS system, at least one CFR arrived on-site in 98.1% of all calls and the response interval for the first arriving CFR was 4:05 min. To our knowledge, no other studies have explored the correlation between CFR response intervals and their experience, thus further research is needed to confirm our findings as it could lead to increased survival after a time-critical incident.