“…A mixed-effects regression model (Stata command xtmelogit) similar to our prior prehospital variation analysis 7 was then created for each outcome of interest: TTM, PCI, LHC, survival to discharge, and survival with good CPC. 7 We modeled outcome and hospital as random intercepts and adjusted each analysis for age, gender, race, bystander witnessed, initial shockable rhythm witnessed cardiac arrest, bystander CPR, and bystander AED use. As LHC and PCI are supplemental data elements, we excluded 23 hospitals treating 91 cardiac arrests from the analysis of LHC and PCI that had no reported LHC or PCI during the study period.…”