“…Firstly, we estimated a PS using a logistic regression model that adjusted for the following 14 variables: year (2014–2019), sex (male or female), age (continuous value), witness status (no, yes), bystander cardiopulmonary resuscitation (no, yes), adrenaline administration at pre-hospital (no, yes), tracheal intubation at pre-hospital (no, yes), duration from call to hospital arrival (continuous value), first documented rhythm after hospital arrival (shockable, non-shockable, presence of pulse), adrenaline administration in-hospital (no, yes), coronary angiography (no, yes), target temperature management (no, yes), extracorporeal membrane oxygenation (no, yes), and cause of arrest (cardiac, noncardiac). We chose these variables, which potentially affect the probability of treatment assignment, based on clinical knowledge and previous studies [ 8 ]. We performed receiver operating characteristic curve analysis with an area under the curve of PS for predicting IABP use in patients with OHCA.…”