155 Classification of Bystander CPR Using 911-Call Review Versus Field Report

Bystander CPR in out-of-hospital cardiac arrest (OOHCA) has been associated with improved neurologically intact survival to hospital discharge, and the presence of layperson CPR before arrival of first responders is a key component of the international Utstein guidelines for comparison of community cardiac arrest metrics. Bystander CPR is usually captured by paramedic observation on scene, though some out-of-hospital systems are starting to classify this based on 911 call review. The objective of this study was to compare rates of bystander CPR using review of 911-calls for OOHCA to reports by paramedics on scene. This was a retrospective review of Los Angeles Fire Department (LAFD) 911-calls and electronic health records for cases of OOHCA from January - March 2014, as well as January - March 2015. This included all cases of OOHCA handled by LAFD paramedics with attempted resuscitation. Exclusion criteria were: EMS-witnessed arrests, traumatic arrests, arrests occurring in a medical clinic or nursing home, or incidents handled by non-LAFD dispatch centers. Trained non-LAFD abstractors listened to all recorded calls, and documented if chest compressions were initiated by the caller. Chest compressions were deemed to have occurred if the caller or other rescuer began to count compressions out loud, or if the caller verbally confirmed that compressions were being done. Field personnel are supposed to document if bystander chest compressions were initiated either through bystander interrogation or direct observation upon arrival on scene. The primary outcome was inter-rater agreement (Cohen’s Kappa test) between 911-call reviewers and field personnel for the presence of bystander CPR. Secondary outcomes included rates of discordance among select pre-defined subgroups. Of 1027 calls during the study period, 13 recordings were unavailable for review, and 372 calls met one or more exclusion criteria, leaving 642 calls for analysis. The overall bystander CPR rate on 911-call review was 59.3%, while it was 52.1% on field care reports, for a primary outcome kappa value for inter-rater agreement of 0.37 (95% CI 0.29-0.44), indicating fair agreement between the two data sources. On further examination, in 182 cases (31.4%) the records from 911 call review and field report were discordant regarding the presence of bystander CPR: 12.1% of cases had no bystander CPR on 911 call review but were reported by field personnel to have bystander CPR, and 19.3% of cases had bystander CPR on 911 call review but no bystander CPR reported by field personnel. A similar total percentage of discordance was noted among both English-speaking callers (31.3%, =0.35) and limited-English proficiency callers (32.8%, =0.37), patients in residential (30.0%, =0.40) versus public settings (31.8%, =0.39), patients under 65 (34.5%, =0.30) and over 65 years old (28.4%, =0.43), and in minority (33.8%, =0.33) and white non-Hispanic patients (22.2%, =0.52). Inter-rater agreement on the presence of bystander CPR using 911-call review and field report is only fair, and up to one-third of cases may be misclassified. Given the extraordinary resources dedicated promoting bystander-CPR, clearer consensus should be developed on how to accurately measure community bystander CPR rates.