The Role of Drones in Out-of-Hospital Cardiac Arrest: A Scoping Review

Drones may be able to deliver automated external defibrillators (AEDs) directly to bystanders of out-of-hospital cardiac arrest (OHCA) events, improving survival outcomes by facilitating early defibrillation. We aimed to provide an overview of the available literature on the role and impact of drones in AED delivery in OHCA. We conducted this scoping review using the PRISMA-ScR and Arksey and O’Malley framework, and systematically searched five bibliographical databases (Medline, EMBASE, Cochrane CENTRAL, PsychInfo and Scopus) from inception until 28 February 2022. After excluding duplicate articles, title/abstract screening followed by full text review was conducted by three independent authors. Data from the included articles were abstracted and analysed, with a focus on potential time savings of drone networks in delivering AEDs in OHCA, and factors that influence its implementation. Out of the 26 included studies, 23 conducted simulations or physical trials to optimise drone network configuration and evaluate time savings from drone delivery of AEDs, compared to the current emergency medical services (EMS), along with 1 prospective trial conducted in Sweden and 2 qualitative studies. Improvements in response times varied across the studies, with greater time savings in rural areas. However, emergency call to AED attachment time was not reduced in the sole prospective study and a South Korean study that accounted for weather and topography. With growing interest in drones and their potential use in AED delivery spurring new research in the field, our included studies demonstrate the potential advantages of unmanned aerial vehicle (UAV) network implementation in controlled environments to deliver AEDs faster than current EMS. However, for these time savings to translate to reduced times to defibrillation and improvement in OHCA outcomes, careful evaluation and addressing of real-world delays, challenges, and barriers to drone use in AED delivery is required.


Introduction Background
Out-of-hospital cardiac arrest (OHCA) is the most time-critical medical emergency. With an incidence of 147 per 100,000 ED presentations [1] and 8.8% of these surviving to discharge [2], OHCA exerts a significant disease burden globally. Successful resuscitation can potentially avert certain death and may allow patients to return to an active life in the community [3]. Early defibrillation significantly improves survival in OHCA [4,5], as the odds of survival after bystander defibrillation raised by 2.30 times in Singapore. An unsolved challenge is how to optimise public access defibrillation (PAD) programs, as automated external defibrillators (AED) must be strategically placed to provide timely access to large populations, while remaining cost-effective [6,7]. Drones, or unmanned aerial vehicles, have emerged as a potential solution.
Early defibrillation is one of the main factors in improving survival outcomes [8]. In OHCA with shockable rhythms, the odds of survival decrease by 10% with each minute that passes without defibrillation [7,9]. PAD enables bystanders to initiate early defibrillation before the arrival of emergency medical services and improves outcomes. However, PAD programs face challenges, as the majority of OHCAs occur in private areas, such as residences [10], or at timings where the AED may not be accessible [11]. The occurrence of the coronavirus disease (COVID-19) pandemic has also resulted in a significant reduction in bystander AED use [12,13]. Combined with increased EMS response times, mortality rates of OHCA events have significantly increased [14][15][16]. As such, novel methods to optimise PAD programs are needed.
Drones are small aircrafts that can be operated remotely without human crew on board. Development in drone technology has increased its capabilities, expanding its use from the military to other areas, such as aerial surveillance, cargo transport, humanitarian relief and healthcare [17,18]. The use of UAVs to deliver life-saving medical devices such as AEDs has been a growing field of interest over the past decade. Drones may be able to deliver the AEDs directly to bystanders before local EMS arrival, thus expediting the time to defibrillation. This may also minimise time required by bystanders to perform ground search for AED. However, an assessment of the overall impact of AED-drone delivery in OHCA events and evaluation of the current literature on the role of drones in OHCA have not yet been carried out. Thus, the aim of this review is to provide a scope of the potential impacts and current state of AED drone delivery research.

Methods
This scoping review adhered to the Preferred Reporting Items for Systematic Reviews and Meta-analyses Extension for Scoping Reviews (PRISMA-ScR) and the methodological framework for scoping reviews proposed by Arksey and O'Malley [19]. Given the heterogeneity of studies that explore the use of drones in OHCA, the decision was made to synthesise the existing literature through a scoping review. This review aims to map the range of the existing literature, identify key research findings or gaps in existing knowledge, and highlight future research directions.

Search Strategy
We systematically searched five bibliographical databases (Medline, EMBASE, Cochrane CENTRAL, PsychInfo, and Scopus) from inception until 28 February 2022. The search strategy and choice of databases was designed in consultation with a medical information specialist (Medical Library, National University of Singapore, Singapore). To retrieve relevant articles, we used keywords and MeSH terms, including "Drones", "Unmanned Aerial Vehicles", "Automated External Defibrillator Delivery", "Out-of-hospital Cardiac Arrest" and their synonyms. We also hand-searched the bibliographies of reviews that addressed related topics, such as the role of drones for health purposes, to identify further relevant articles. We also consulted subject matter experts to identify additional relevant articles. After removing duplicates, we conducted an article assessment using Endnote X9 (Clarivate, Philadelphia, PA, USA) to assess the titles and abstracts of the retrieved articles. Full texts were assessed for articles of interest. The detailed search strategy may be found in Supplementary Materials Figure S1.

Inclusion and Exclusion Criteria
Three authors (JCLL, HHL, NL) conducted the article assessment using predefined criteria. Each article was reviewed by at least two authors and the decision to include or exclude each article was blinded among them. Disputes were resolved through discussion and consensus with a senior author (AFWH). Articles that addressed the application of drones in OHCA were considered eligible for inclusion, where drones were defined as remotely controlled aircraft without any humans on board. Interventional trials, retrospective cohort, or prospective cohort study designs were included, as well as qualitative studies, or studies using simulation or mathematical models. We included conference abstracts to comprehensively assess the literature, referencing them as such. Studies were excluded if there was no primary datum, such as systematic reviews and meta-analyses, narrative reviews, and protocols. They were also excluded if they were non-English and without an English translation.

Data Abstraction
Three authors (JCLL, HHL, NL) abstracted data using a predetermined data collection form. The data abstraction process was independent and blinded among the study authors, and disputes were resolved through consensus with the senior author (AFWH). Article information (author, year of publication, country), methodology (interventional, simulation, qualitative), drone specifications (drone type, maximum range, maximum velocity), and relevant quantitative or qualitative results were abstracted. We presented continuous data in mean and standard deviation (SD) or median and interquartile range (IQR) and categorical data in percentages. Specific outcomes of interest included time saved due to AED delivery by drones compared to EMS, defined as the difference in response time between drones and EMS, the number and distribution of drone base locations according to geographical information systems (GIS) modelling, with respect to the maximum coverage location problem (MCLP), and qualitative barriers or enablers towards using drones in OHCA among bystanders and EMS personnel. Data on survival with good neurological outcome or survival alone to the time points of discharge or 30 days were also extracted whenever possible.

Literature Retrieval and Summary of Included Articles
The literature search retrieved 970 articles. After removal of 297 duplicate articles, 626 articles were excluded based on their title and abstract. A further 24 articles were excluded based on the full text review. Finally, 26 articles were eligible for qualitative synthesis. The study selection process and reasons for exclusion is presented in the PRISMA-P 2020 Flow Diagram (Supplementary Materials S2). Nine papers investigated the outcomes of different drone modelling approaches, seven papers investigated time saving of AED delivery via drones vs. current EMS and one study compared response times of AED delivery via drones vs. ground search for public AEDs. Four papers investigated the effects of meteorological conditions on drone delivery of AED, six papers examined the factors that impact the feasibility of drone delivery, and five papers examined the cost-effectiveness of drone network implementation. Finally, five papers qualitatively assessed user experience and stakeholder attitudes to implementation of AED delivery via drones.

Types of Drones
A total of 14 studies reported on the brand and specifications of drones used in their studies. The drones were fixed wing or multirotor systems, with maximum velocities ranging from 48.3 km/h to 100 km/h, and maximum ranges from 6 km to 80.5 km. The summary of drone models and specifications may be found in Supplementary Materials S3.

Time Saving with AED-Delivery Using Drones
The primary application of drones in all the included studies was to deliver an AED to the site of an OHCA patient, while the most frequently measured outcome was time savings. A total of 7 studies investigated time savings of AED delivery compared to current emergency medical services. The summary of results may be found in Table 2.  The long-term maintenance cost is assumed to be 20% of the drone purchase price annually (Mackle, Bauer)

Feasibility 6
The presence of variable terrains, technical issues, and legal restrictions impact feasibility of drone delivery. Meteorological conditions also significantly impact drone dispatch and response times. Multiple simulated studies concluded that there was an overall reduction in response times when drones were used in rural areas. The extent of improvement varied widely among papers [21,24,29]. The sole prospective trial found that drones arrived prior to EMS in 64% of cases, but also that AED shock was not delivered before EMS arrival on scene in any of the cases [30].
However, response times in urban areas did not improve by a significant margin. Drones arrived before EMS services in 32% of cases (urban) vs. 93% (rural) and reported less improvement in response times [27,34].
Drone delivery was found to be favourable in the ground search for AEDs where public AEDs were not easily accessible, but ground search was favourable when public AEDs were easily accessible [41].

Optimal Drone Positioning
Nine studies investigated and compared the outcomes of different optimisation models to determine the placement and number of bases and drones in an implemented drone network. Use of these models to optimise drone placement at new sites improved response times compared to placement at the existing EMS sites from 80.1% to 90.3% [36].
Drone delivery of AED from regionally optimised models reduced travelling distance and time to arrival compared to EMS [21]. Integrated location-queuing optimisation models, when compared to region-specific models, required less bases and drones, and improved median time to AED [20].
Selective activation of drones via the dispatch rule allowed maintained improvements in response times, with up to 30% fewer dispatches with high accuracy [23].
Increasing the weightage of backup coverage that allowed EMS facilities to respond to a second event in its service area required a greater number of drones and reduced primary coverage [37]. A backup weight of 0.2 was found to minimise this loss of primary coverage, while increasing backup coverage significantly.

Feasibility and Cost-Effectiveness
Six studies assessed the practicality of use of drones for AED delivery in the real world and their barriers, which include legislation, technical issues and variability in terrain. Four studies emphasised the importance of weather and meteorological conditions in the operation of drone networks. The summary of factors that impact feasibility may be found in Table 2.
Five studies examined the expenditure required to purchase and maintain a drone network. The most cost-efficient way to achieve 90% response within 1 min was to station drones at 39 EMS sites and 12 new locations for a total of 51 sites, costing SGD 2,010,000 [36]. In addition to the initial cost, annual maintenance fees may cost 20% of the initial sum [34,38].

Perception of Drone Use in AED Delivery
A total of 5 studies conducted practical simulations with AED delivery via drones to participants performing CPR on a manikin, following up with qualitative interviews. Another study [22] evaluated public perception and acceptability of a drone AED delivery program 'AED on the Fly' in the town of Caledon, Ontario. The qualitative results from the abovementioned studies are documented in Table 3.
These studies all reported positive community attitudes to the delivery of AEDs via drones. Participants and key stakeholders alike perceived value in the potential advantages of this drone delivery system in reducing response times in OHCA events, especially in less accessible locations [26,[41][42][43][44]. However, some participants reported neutral feelings, uncertainty, and anxiety towards interacting with the drone, as well as safety concerns [44].
Challenges and barriers to successful drone AED deployment were highlighted. AED and CPR usage, in addition to technology literacy limitations, were cited as significant obstacles; users reported technical difficulties in electrode attachment and placement and even in mobile phone usage [28]. Key stakeholders cited logistical, financial, legal and safety challenges, as well as concern regarding public use of AEDs [42].
Suggestions for improvement encompassed methods to facilitate bystander use, thereby decreasing the time to defibrillation. Visual and audio indicators may decrease the time required to locate the drone-delivered AED [28]. Upon delivery and location of the AED, short and clear dispatcher instructions may improve technique and compliance in AED operation [44]. Successful community engagement and implementation requires clear and consultative communication with the community during the development of the AED drone network programme [22]. Important considerations also include solidifying partnerships with relevant stakeholders, such as EMS and fire services, and identifying stable funding, as well as learning from existing drone models [42]. Table 3. Summary of Qualitative Results.

Author
Year Key Results

Sanfridsson 2019
Participant attitude and experience in drone delivery of AEDs Difficulties faced by participants, and concerns • Wariness and hesitation due to poor understanding of current paramedic services; concerned regarding possibility of drone program replacing paramedic services • Lack of CPR and AED literacy • Desire to be made aware of all stages of testing of the project, and for in-person demonstrations Table 3. Cont.

Starks 2020
Stakeholder attitude towards drone delivery of AEDs • Broad support for the drone network-value perceived in reduced response times and to access of hard-to-reach areas Challenges raised by stakeholders • Operationalisation of autonomous drone AED network and financial liabilities • Privacy and safety concerns; current legal and regulatory requirements • Public buy-in and concern of public actually using an AED • Need for research on treatment and cost-effectiveness Facilitators of drone network development • Solidification of key partnerships, e.g., EMS and fire services • Identification of viable funding • Learning from existing drone models.

Baumgarten 2021
Participant attitude and experience in drone delivery of AEDs • Bystanders and community first responders were able to collect the AED without any safety concerns • A total of 8.9% of bystanders reported hesitancy to collect the AED and 2.2% found it cumbersome; none of the community first responders expressed problems • A total of 95.6% of bystanders and 100% of community first responders supported the implementation of UAS-based AED delivery systems

Discussion
This scoping review, conducted with five bibliographical databases, yielding 26 relevant articles, provided an overview of the available literature on the use of drones in AED delivery in OHCAs. The majority of studies conducted utilised software simulations to assess the time to delivery of AEDs delivered by drones against conventional EMS, using a variety of location models to determine the positioning and distribution of drones and drone bases. This scoping review is the most current representation of the varied published literature, including quantitative and qualitative studies, regarding drone delivery of AEDs in OHCA. It is an important contribution that may guide the interventional trials needed to confirm the effectiveness of drones for delivering AEDs, which remains a promising venture, as care for OHCA patients develops in the future.
All the included studies demonstrated varying, but significant, time reductions in AED delivery via drones as compared to existing EMS and may decrease time to defibrillation to increase survivability in OHCA events. Studies conducted by Bogle et al. and Pulver et al. estimate the cost of establishing a drone network at 1.3 million every 4 years and 2.01 million, respectively. The former study calculated the cost per QALY to be SGD 1937, far below the Institute for Clinical and Economic Review's valuation of SGD 50,000-SGD 150,000 per QALY; suggesting the cost effectiveness of the implementation of a drone network. Furthermore, advancements in drone technology are promising, with more recent models of the DJI drones being used in three of the studies that showcased twice the flight time, infrared cameras with rangefinder functionality, and expanded operating temperatures of −20 to 50 degrees Celsius.
Most notably, the sole prospective trial found that in 64% of drone dispatches, the drone was able to arrive prior to EMS, with a median time difference of 1:52 min [30]. This trial provided unique insight into real-world integration of a drone network into existing EMS, accounting for the full complement of delays and challenges. While these results are a promising representation of the benefits regarding time to availability of AED, the trial noted no AEDs attached prior to ambulance arrival. Furthermore, the study found 74% of total OHCA cases ineligible for drone dispatch, due to reasons including adverse weather conditions, no-fly zones and technical difficulties. This highlights that for the effective utilisation of drones for AED delivery in OHCA, determinants that can be broadly classified under the following subcategories must first be identified and then solved.
First, appropriate community education and literacy in AED and CPR is paramount, as is adequate knowledge and acceptance of the use of drone technology in the provision of lifesaving aid in OHCA. Currently, key barriers to public access defibrillation include the following: the majority of OHCAs occur in locations unsuitable for timely public-access AED deployment, and lack of bystander literacy and confidence in AED usage [45]. The former is the limiting factor against early bystander defibrillation in OHCA events [46,47]. Drone delivery may allow AEDs to be available at the site of OHCAs more promptly, but concurrent improvements in poor AED literacy rates [28,43] among bystanders is necessary to minimise bottlenecking of early bystander defibrillation rates by low AED usage once delivered. This may be executed through a range of methods, including lay instructors, self-directed learning and brief training. A European study found heterogeneity in AED infrastructure and legislation across different countries in Europe, which was reflected in corresponding differences in AED use and OHCA survival [48]. This highlights not only the challenges in ensuring the use of AEDs upon delivery on-scene across different locations, but that improvements in AED literacy and integration may translate to increased bystander-performed AED resuscitation.
Second, civil aviation regulations regarding drone flight must be conducive to the use of a drone network in the delivery of AEDs. Legal restrictions, including prohibition of flight out of line of sight and no-fly zones as reported by Bauer et al., are prohibitive to the development of a drone network to deliver AEDs to OHCAs. Recently, new regulations in the Aviation Law Act in Poland, for example, introduce standardised requirements for drone use, including registration of drones, weight and height limitations, anti-collision and emergency procedural technology. A challenge that remains is the requirement to obtain airspace clearance and permission before flight and its associated delays, especially in emergency use.
Third, practical considerations in operation and maintenance of a drone network include prohibitive effects of weather effects on drone flight reported in the prospective trial conducted by Schierbeck et al. Rain, specifically, prohibited drone dispatch in 8 out of 53 OHCA cases. Furthermore, technical issues, such as maintenance and battery charging, may increase the number of drones required per base for dispatch. This may be solved via a queuing model proposed by Boutilier et al., or the backup coverage location problem proposed by Pulver et al. Especially in urban contexts, high-rise buildings pose a challenge to drone delivery of AEDs as a physical obstacle to flight paths, adding complexity and delays in AED delivery to OHCAs that occur within their premises.
Other important factors to increase the effectiveness of drone-delivered AEDs were also suggested, including, but not limited to, the use of short encouraging instructions from dispatchers and increasing AED visibility via installing headlights on the drone hovering over the AED.
Finally, the most appropriate location model and cost breakdown must be determined to allow maximum utility of the drone network with the most effective coverage of the required area. This includes more recent variations in these models, including those that consider queuing, backup coverage and evaluation of whether drones should be dispatched in each scenario. This changes depending on factors such as the distribution of OHCAs, with Boutilier et al. finding that integrated networks require fewer bases and drones to achieve the same reduction in the 90th percentile of time to AED arrival as region-specific models, but may result in a loss of rural coverage.

Strengths and Limitations
This review is the most current representation of the varied published evidence related to the time savings, implementation methods, location models and challenges in the use of drones to deliver AEDs in OHCA. The strengths of this review are the systematic search technique, precise inclusion and exclusion criteria, and careful data extraction and representation process. Notably, this review included both quantitative and qualitative studies to deliver a clearer image of the state of the current literature on this novel topic.
Limitations include the inclusion of only English language literature in the review. Due to the limited number of studies performed and the heterogeneous nature of the studies in their methodologies, simulation models and outcomes, a scoping rather than systematic review was conducted. Hence, risk of bias and quality assessments of the included studies were not performed, and we were unable to draw comparisons across studies.
As drone delivery of AEDs remains a novel intervention strategy, the majority of studies conducted were simulations, with only one prospective trial conducted. There are, therefore, limited data from real-world implementation of drone networks, which is imperative in obtaining an understanding and drawing concrete conclusions regarding the multiple factors that influence the effectiveness of this solution.

Conclusions
In this scoping review of the available literature on the use of drones to facilitate AED delivery in OHCAs, the simulations and trials conducted have provided evidence that in a controlled environment, drones can deliver an AED faster than the current EMS services, thereby decreasing time to defibrillation and improving OHCA outcomes. However, drone delivery of AEDs can only be effective if supporting factors such as local aviation regulations, community education and AED literacy, for example, are evaluated together in a cohesive manner.