Abstract 338: Temporal Characteristics of a Smartphone Alert System for CPR in a Mid-Sized Metropolitan Area

Abstract

Introduction: Smartphone alert systems (SAS) mobilize trained volunteers to initiate CPR prior to Emergency Medical Services (EMS) arrival. Among the potential factors affecting the probability of a successful response is time; time of day, day of the week, and season each present plausible barriers to response. We sought to examine the temporal characteristics of an SAS in order to better understand this relationship between time and system performance. Methods: Operational records were obtained from the PulsePoint deployment in Allegheny County, Pennsylvania, USA for the period June 2016 to May 2023. PulsePoint sends alerts including nearby AED locations to volunteer responders within a 400m radius of public cardiac arrest events. Case data included alert time and number of responders and AEDs within range of the event. We associated each PulsePoint event with an 8-hour Timeframe (TF) {1 (00:00-07:59), 2 (08:00-15:59), and 3 (16:00-23:59)}, a day of the week, and a season. For each, we calculated the median number of associated responders and AEDs, and the total number of events. We then compared each across each time variable with a Mann-Whitney U-test or Kruskal-Wallis test. Results: There were a total of 2354 events in the capture period. Alerts were least common on Sundays (12.4%) and most common on Tuesday (15.6%). Alerts most frequently occurred in TF2 (45.6% vs TF1: 18.3, TF3: 36.1) and were similar between seasons. Weekday alerts had a higher median number of responders than weekend alerts {3 (1-5) vs 2 (1-4), p=0.004}. Responder count was lower for TF1 than TF2 and TF3 {2 (1-4) vs 3(1-5) vs 3(1-5), p < 0.001}. Count of proximal AEDs did not vary by any of the temporal characteristics. Conclusion: Both event frequency and number of notified responders were associated with day of the week and time of day. Either feature is more complicated than meets the eye, and may include associations with diurnal population movement cycles, overall patterns in emergency services utilization, and responder device configuration preferences. All could be influential in optimizing such SAS deployments.