Bystander Cardiopulmonary Resuscitation: Training, Delivery, And Measurement Error

Bystander-delivered cardiopulmonary resuscitation (B-CPR) is an essential treatment for sudden cardiac arrest (SCA), yet less than one-third of victims receive B-CPR. Few studies have examined disparities in either layperson CPR training or B-CPR delivery. Furthermore, the association between CPR training and B-CPR delivery, and the potential impact of Dispatch-assisted CPR (D-CPR), has been inadequately quantified, partially due to limited observational datasets. We performed a nationally representative survey to measure estimated CPR training prevalence in the United States. We acquired clinical SCA and B-CPR data from the Resuscitation Outcomes Consortium (ROC) national registry and from Seattle King County (SKC) Emergency Medical Services (EMS) to enable inquiry into D-CPR and missing data within ROC. We assessed the differences in estimated CPR training prevalence, disparities in B-CPR, the association of community-level CPR training and B-CPR, and the impact of missing data. Aim 1: Between 09/2015-11/2015, 9,022 individuals completed the national CPR training survey; 18% reported current training in CPR, and 65% reported prior training. Aim 2: In the ROC cohort, 19,331 out-of-hospital cardiac arrests were assessed. In public locations, 39% (272/694) of females and 45% (1,170/2,600) of males received B-CPR (p<0.01), whereas in private settings, 35% (2,198/6,328) of females and 36% (3,364/9,449) of males received B-CPR (p=ns). Aim 3: From survey and ROC data analysis (n=17,883), increased community CPR training was associated with B-CPR delivery (OR: 1.21(95% CI: 1.04-1.39)), but this relationship was modified by site (p=ns). Aim 4: The ROC D-CPR variable had 80% missingness; multiple imputation (MI) was used and provided comparable results to the complete SKC dataset on the association of D-CPR on B-CPR (ROC MI RR: 3.84 (95% CI: 2.97-4.98) vs EMS complete case RR: 3.51 (95% CI: 3.22-3.83)). MI was verified in Missing Completely at Random and Missing at Random simulated datasets. In conclusion, rates of public CPR training and B-CPR delivery were low in the US. Males were more likely to receive B-CPR in public locations. The association of community-level CPR training on B-CPR delivery was modified by site. Future work is required to understand the role of D-CPR in encouraging CPR delivery.