Survival, prognostication and secondary prevention in cardiac arrest

Abstract

Background: Sudden Cardiac Arrest (SCA) is a highly lethal medical emergency, caused by an abrupt cessation of cardiac function. Despite extensive research efforts, increased public awareness and systematic healthcare improvements, overall survival is still very low. Individuals suffering an out-of-hospital cardiac arrest (OHCA) exhibit the lowest survival rates, with only 10% surviving to 30 days. Highly stochastic factors, such as the time and place of the cardiac arrest, exert a fundamental impact on survival. The emergence of technical advancement, accompanied by artificial intelligence, holds new opportunities that may enable the prediction, and thus prevention, of SCA, in addition to reducing the delay to evidence-based interventions.

Methods: Study I aimed to characterise the impact of the COVID-19 pandemic on the epidemiology of in- and out-of-hospital cardiac arrests in Sweden. Study II aimed to characterise long-term trends in in-hospital cardiac arrest (IHCA) and OHCA, highlighting temporal differences in epidemiology, management and survival. Study III explored the potential benefit of having an implantable cardioverter defibrillator (ICD) at discharge among patients who survived an OHCA with an initial shockable rhythm. Study IV aimed to improve the Swedish Cardiac Arrest Risk Score-1 (SCARS-1), by optimising the machine learning strategy. All studies utilised data from the Swedish Cardiopulmonary Resuscitation Registry (SCRR).

Results: Study I demonstrated a notable shift in epidemiology in IHCA and OHCA during the pandemic, with a 2.3- and 3.4-fold increase in mortality rates, respectively. Patients with COVID-19 and cardiac arrest were much more likely to suffer from respiratory insufficiency prior to SCA, typically with a non-shockable rhythm. Study II revealed a 2.2-fold increase in survival rates in OHCA during 1990–2020, albeit without any improvements during the final decade. For the same time period, we observed more than a two-fold increase from ambulance dispatch to arrival. On the other hand, survival rates in IHCA increased by 47% during 2011–2020, with unabated improvement in survival. For both OHCA and IHCA a cardiac aetiology causing the cardiac arrest decreased, 55% and 70%, respectively. Study III suggested OHCA cases with a cardiac aetiology and initial shockable rhythm discharged with an ICD had superior survival rates compared to patients without. Specifically, the study suggested a 60% reduction in the probability of recurrent SCA or death among those discharged with an ICD. Study IV showed we were able to enhance our previous prediction model, SCARS-1, creating a parsimonious model with excellent calibration and discriminatory performance, suitable for evaluation in future clinical trials. The final model displayed a AUC ROC score of 0.96.

Conclusion: Using both classical epidemiological- and cutting-edge machine learning methods, we have elucidated important epidemiological and clinical gaps in knowledge. We provided novel insights into the COVID-19 pandemic, long-term trends in survival and the potential benefit of broader use of ICDs. We have also demonstrated that survival can be predicted with excellent predictive performance by the time the patient arrives in the emergency department. Importantly, this thesis highlights that future resuscitation practices will face many challenges, ranging from reducing ambulance response times, finding new means of treating the increasing number of cases with non-cardiac causes, and identifying patients who would benefit from secondary preventive measures.