Abstract Sat801: Discrepancies in Bystander CPR Documentation: Comparing the Birmingham CARES Data with 9-1-1 Audio Review

Abstract

Background: Previous reports from the Cardiac Arrest Registry to Enhance Survival (CARES) have indicated a bystander cardiopulmonary resuscitation (B-CPR) rate of only 15.5% in Birmingham, AL; approximately 2.6 times below the national average. Dispatch 9-1-1 audio recordings offer an additional source for identifying out-of-hospital cardiac arrest (OHCA) events where B-CPR was provided. We sought to assess the agreement between B-CPR provision as documented by emergency medical services (EMS) in the Birmingham CARES registry and as identified through 9-1-1 audio review. Methods: We conducted a retrospective observational analysis of all adult non-traumatic OHCA cases in Birmingham from January 1 to December 31, 2023. Excluded cases included EMS-witnessed events, those in nursing homes, healthcare facilities, jails/prisons, or involving patients who were conscious during the 9-1-1 call. B-CPR provision was classified as “yes” or “no” in CARES based on EMS documentation and compared to B-CPR status determined through quality improvement review of the corresponding 9-1-1 audio. Agreement between sources was assessed using percent agreement, Cohen’s kappa, Gwet’s AC, and McNemar’s test. Results: Of 236 cases, 180 (76.3%) showed concordant classification: 19 cases where both sources indicated B-CPR was performed and 161 where both indicated it was not. Discrepancies occurred in 56 cases (23.7%), including 46 instances where 9-1-1 audio identified B-CPR but CARES did not, and 10 where CARES documented B-CPR but audio review did not. Among the 46 audio-confirmed cases not captured by CARES, most involved B-CPR that ended before EMS arrival (e.g., B-CPR was discontinued by the caller), and 7 appeared to be CARES misclassifications. In the 10 cases where CARES documented B-CPR but audio did not, all involved calls that ended prior to EMS arrival without recognition of OHCA or CPR instruction. Overall agreement was fair to moderate: Cohen’s kappa = 0.28 [95%CI 0.15, 0.42], Gwet’s AC = 0.65 [95%CI 0.56, 0.75]), and McNemar’s test showed significant asymmetry in classification, p<0.001. Conclusion: B-CPR documentation differed in nearly 25% of OHCA cases between the Birmingham CARES and 9-1-1 audio review. Most discrepancies were due to early B-CPR discontinuation by the caller before EMS arrival. A smaller number were due to apparent CARES misclassification. These findings highlight the need for continued telecommunicator CPR instruction through EMS arrival.