By Joshua Moss, MD

Associate Professor of Clinical Medicine, Cardiac Electrophysiology, Division of Cardiology, University of California, San Francisco

Dr. Moss reports he is a consultant for Abbott, Boston Scientific, and Medtronic.

SYNOPSIS: A study of the location of out-of-hospital cardiac arrests (OHCA) compared to the location of automatic external defibrillators (AED) in Denmark showed that if the placements of AEDs more closely matched the location of OHCAs, bystander defibrillation and 30-day survival would improve.

SOURCES: Sun CLF, Karlsson L, Torp-Pedersen C, et al. In silico trial of optimized versus actual public defibrillator locations. J Am Coll Cardiol 2019;74:1557-1567.

Stecker EC, Reinier K, Howell SJ. Improving resuscitation outcomes with AEDs: Location, location, location. J Am Coll Cardiol 2019;74:1568-1569.

Although highly effective when used for out-of-hospital cardiac arrest (OHCA), automated external defibrillators (AEDs) often are placed in areas of low risk and limited temporal availability. Investigators from Denmark performed a retrospective study of OHCA events and AED location in Copenhagen to determine the optimal location for AEDs to maximize the coverage of OHCAs. Emergency medical service data were used. Only OHCAs of presumed cardiac cause were included. Those due to trauma, drug overdose, or with late signs of death were excluded. OHCAs through 1994 were the training models, and those from 2007-2016 were the study period, since AED placement was registered after that point. Two models were studied: where availability was limited to a building’s hours of operation (model 1) and where AEDs were available 24/7 (model 2). The primary outcome was OHCA coverage defined as an OHCA occurring within 100 m of an AED. Also, bystander defibrillator and 30-day survival were estimated.

During the study period, there were 673 public OHCAs and 1,573 AEDs in Copenhagen. Real AED placements covered 22% of the observed OHCAs. Model 1 would have covered 33%, and model 2 would have covered 43%, respectively, for relative gains of 52% and 96% (P < 0.001). Bystander defibrillation would have increased from the actual 15% to 23% and 27% with the two models, for relative increases of 53% and 84% (P < 0.001). Also, 30-day survival would improve from 31% to 35% for both models, for a relative increase of 11% to 13% (P < 0.001).

The authors concluded that optimal AED placement based on the location of OHCAs in the past would increase coverage by 50% to 100% compared to real AED locations, and significantly affect predicted bystander defibrillation and 30-day survival.


This study represents the first in silico trial of a public AED intervention, comparing AED coverage of OHCAs in real life to coverage that could be achieved using mathematical modeling (based on prior OHCA data). Perhaps it is not surprising that placing new AEDs based on a centralized source of data about historical OHCA locations would result in better bystander accessibility. The data confirm the assumption that there are some spatial and temporal patterns to these events within a given metropolitan area that lend some degree of predictability.

The major remaining question is whether using such modeling techniques will translate into higher survival statistics that we all seek, and that the authors indeed predicted using multivariate logistic regression models. In the most optimistic scenario, one in which mathematically optimized AED locations with 24/7 availability were chosen, only 43% of cardiac arrests occurred within 100 meters of an AED, despite the passage of nine years since the first AED was placed and optimization began. Additionally, while the readily available (and functional) AED could be a critical tool for improving survival in those 43% of OHCAs, it is ultimately only one piece of many needed. The presence of a bystander who is ready, willing, and able to use the AED (as well as perform CPR when necessary), a shockable rhythm that is not refractory to defibrillation attempts, and EMS services that can transport a victim to a hospital with facilities and services to manage them are equally critical.

Overall, this study presents an intriguing method for improving public AED availability and, in turn, rapid treatment of OHCA and better chances of survival. Assuming, as the authors did, that additional AEDs will continue to be added month by month, there would seem to be little downside in using validated prediction models to help choose the locations for those AEDs, as long as privately funded devices are not discouraged. Parallel investment must be made in improving bystander education and response.