Out-of-hospital cardiac arrest: predict and then protect!

Medicine is a science of uncertainty and an art of probabilityWilliam Osler Despite advances in resuscitation and post-resuscitation care, out of hospital cardiac arrest (OHCA) remains a dramatic condition associated with high mortality, to the point that, if optimally managed, only 20% of patients survive.1Rea T. Kudenchuk P.J. Sayre M.R. Doll A. Eisenberg M. Out of hospital cardiac arrest: past, present, and future.Resuscitation. 2021; 165: 101-109https://doi.org/10.1016/j.resuscitation.2021.06.010Summary Full Text Full Text PDF PubMed Scopus (15) Google Scholar Indeed, the time-dependent nature of this condition, the initial management by non-healthcare subjects (when witnessed) and the critical baseline condition (mostly of cardiac origin) lead affected subjects to poor outcomes in terms of survival and neurological drawbacks.1Rea T. Kudenchuk P.J. Sayre M.R. Doll A. Eisenberg M. Out of hospital cardiac arrest: past, present, and future.Resuscitation. 2021; 165: 101-109https://doi.org/10.1016/j.resuscitation.2021.06.010Summary Full Text Full Text PDF PubMed Scopus (15) Google Scholar Consequently, it is not always worth to attempt or continue cardiopulmonary resuscitation (CPR) in this setting. The decision to undertake, and if undertaken to discontinue, CPR is challenging and there is no uniformity of management in clinical practice, even though indications are provided by current guidelines.1Rea T. Kudenchuk P.J. Sayre M.R. Doll A. Eisenberg M. Out of hospital cardiac arrest: past, present, and future.Resuscitation. 2021; 165: 101-109https://doi.org/10.1016/j.resuscitation.2021.06.010Summary Full Text Full Text PDF PubMed Scopus (15) Google Scholar,2Hessulf F. Bhatt T.D.L. Engdahl D. et al.Predicting survival and neurological outcome in out-of-hospital cardiac arrest using machine learning: the SCARS model.eBioMedicine. 2023; 89104464https://doi.org/10.1016/j.ebiom.2023.104464Summary Full Text Full Text PDF PubMed Scopus (2) Google Scholar One could argue that it would be appropriate to attempt resuscitation in all patients, yet there are at least two major sets of reasons for identifying patients in whom CPR should not be undertaken or terminated. Firstly, the deployment of resources to assist patients with OHCA is remarkable in terms of economic cost and healthcare personnel involvement. Rephrased, transporting patients with near zero chances of survival to the emergency department (ED) reduces the viability for critical patients but with a better prognosis.3Ewy G.A. Cardiac resuscitation--when is enough enough?.N Engl J Med. 2006; 355: 510-512https://doi.org/10.1056/NEJMe068090Crossref PubMed Scopus (14) Google Scholar Secondly there is an ethical issue: if a patient's living will is available this should be taken into great consideration, otherwise, if not available, therapeutic obstinacy is a matter of debate from the bioethical point of view.3Ewy G.A. Cardiac resuscitation--when is enough enough?.N Engl J Med. 2006; 355: 510-512https://doi.org/10.1056/NEJMe068090Crossref PubMed Scopus (14) Google Scholar There is evidence that patients who match criteria for CPR discontinuation usually do not undergo better outcomes if, conversely, resuscitation is prolonged: in a retrospective cohort study by Yates et al. it emerged how there were no survivors between patients who met criteria for CPR discontinuation, in absence of reversible causes of OHCA, but were equally conducted to ED to continue advanced CPR.4Yates E.J. Schmidbauer S. Smyth A.M. et al.Out-of-hospital cardiac arrest termination of resuscitation with ongoing CPR: an observational study.Resuscitation. 2018; 130: 21-27https://doi.org/10.1016/j.resuscitation.2018.06.021Summary Full Text Full Text PDF PubMed Scopus (16) Google Scholar Given the clinical, socio-economic and bioethical relevance of the problem, Hessulf et al. present us a promising new tool to predict survival and neurological outcomes in patients with OHCA.2Hessulf F. Bhatt T.D.L. Engdahl D. et al.Predicting survival and neurological outcome in out-of-hospital cardiac arrest using machine learning: the SCARS model.eBioMedicine. 2023; 89104464https://doi.org/10.1016/j.ebiom.2023.104464Summary Full Text Full Text PDF PubMed Scopus (2) Google Scholar It is a machine learning-derived model, called SCARS model and developed as a web application, based on data retrieved from a Swedish national registry, including data of 55,615 OHCAs from 2010 to 2020 (Fig. 1).2Hessulf F. Bhatt T.D.L. Engdahl D. et al.Predicting survival and neurological outcome in out-of-hospital cardiac arrest using machine learning: the SCARS model.eBioMedicine. 2023; 89104464https://doi.org/10.1016/j.ebiom.2023.104464Summary Full Text Full Text PDF PubMed Scopus (2) Google Scholar The primary outcome for which the model was tested was 30-day survival, the secondary outcome was neurological function using the CPC score at discharge.2Hessulf F. Bhatt T.D.L. Engdahl D. et al.Predicting survival and neurological outcome in out-of-hospital cardiac arrest using machine learning: the SCARS model.eBioMedicine. 2023; 89104464https://doi.org/10.1016/j.ebiom.2023.104464Summary Full Text Full Text PDF PubMed Scopus (2) Google Scholar As many as 393 candidate predictors were tested, of which the top 20 best performers were selected in order to establish the final model, which showed excellent accuracy (the model had an area under the receiver operating curve of 0.95, with optimal calibration) in predicting outcomes.2Hessulf F. Bhatt T.D.L. Engdahl D. et al.Predicting survival and neurological outcome in out-of-hospital cardiac arrest using machine learning: the SCARS model.eBioMedicine. 2023; 89104464https://doi.org/10.1016/j.ebiom.2023.104464Summary Full Text Full Text PDF PubMed Scopus (2) Google Scholar Among predictors, they included data on initial presentation to the ED, age, pre-hospital interventions (such as adrenaline use) and critical time intervals. These variables are easily available for clinicians in the emergency setting, indeed less than 15 s are required to apply the SCARS model.2Hessulf F. Bhatt T.D.L. Engdahl D. et al.Predicting survival and neurological outcome in out-of-hospital cardiac arrest using machine learning: the SCARS model.eBioMedicine. 2023; 89104464https://doi.org/10.1016/j.ebiom.2023.104464Summary Full Text Full Text PDF PubMed Scopus (2) Google Scholar As highlighted by the authors in the discussion section of their paper, other researchers have already tried to identify a predictive model in patients with OHCA. Chen Ji et al. in England have proposed a model that has been shown to have good accuracy, including factors such as age, gender, witness/bystander and initial rhythm.5Ji C. Brown T.P. Booth S.J. et al.Risk prediction models for out-of-hospital cardiac arrest outcomes in England.Eur Heart J Qual Care Clin Outcomes. 2021; 7: 198-207https://doi.org/10.1093/ehjqcco/qcaa019Crossref PubMed Scopus (10) Google Scholar Instead, Kim et al. have developed TACOM, an original tool capable of predicting minute-by-minute outcomes in terms of survival and residual neurological function.6Kim J.W. Ha J. Kim T. et al.Developing a time-adaptive prediction model for out-of-hospital cardiac arrest: nationwide cohort study in Korea.J Med Internet Res. 2021; 23e28361https://doi.org/10.2196/28361Crossref Scopus (3) Google Scholar Another captivating study, even if the clinical setting was slightly different from that of our discussion, was performed by Bunch et al., in which ECG was carefully evaluated in OHCA survivors as predictor of outcomes: the duration of the QRS complex was predictive of greater mortality or the need for an implant of ICDs rather than PR interval or other ECG features.7Bunch T.J. White R.D. Bruce G.K. et al.Prediction of short- and long-term outcomes by electrocardiography in survivors of out-of-hospital cardiac arrest.Resuscitation. 2004; 63: 137-143https://doi.org/10.1016/j.resuscitation.2004.05.008Summary Full Text Full Text PDF PubMed Scopus (15) Google Scholar Of all the models mentioned above, and in the discussion of the paper by Hessulf et al., SCARS is certainly the most manageable and reliable one, being based on rapidly accessible variables and validated on a larger number of patients compared to the others. Indeed, we believe that decision-making in the field of OHCA management is of utmost importance: more tools should be provided to clinicians to avoid futile resuscitation attempts. This holds true, above all, in those cases of OHCA in which we are stuck in gray areas: an example is the management of OHCA in elderly patients loaded by many comorbidities, in whom it is difficult to establish the appropriateness of resuscitative maneuvers.8Einav S. Cortegiani A. Marcus E.L. Cardiac arrest in older adult patients.Curr Opin Anaesthesiol. 2021; 34: 40-47https://doi.org/10.1097/ACO.0000000000000942Crossref PubMed Scopus (3) Google Scholar Moreover, another gray area is represented by OHCA patients with concomitant hypothermia, in whom CPR should be significantly prolonged compared to normothermic patients, due to the greater chances of survival and good neurological outcomes, but still there is some reticence in daily clinical practice towards this concept.9Paal P. Pasquier M. Darocha T. et al.Accidental hypothermia: 2021 update.Int J Environ Res Public Health. 2022; 19https://doi.org/10.3390/ijerph19010501Crossref Scopus (29) Google Scholar In addition, a tool such as the SCARS model, combined with new technologies including artificial intelligence and telemedicine, could help to more easily discriminate cases of OHCA caused by reversible causes, in which specific etiological therapy can be crucial.10Lee S. Okubo M. The use of artificial intelligence to predict the on-scene return of spontaneous circulation in the out-of-hospital setting: a time to do more for cardiac arrest?.Ann Emerg Med. 2022; 79: 145-147https://doi.org/10.1016/j.annemergmed.2021.10.016Summary Full Text Full Text PDF PubMed Scopus (2) Google Scholar For instance, having a digital medical record of an OHCA patient, who is clearly not responsive, could help us to have precious medical history information for the identification of any reversible causes. The use of predictive models and new technologies might be in the future accompanied by an intense implementation, on a large scale, of life support special techniques such as extracorporeal membrane oxygenation (ECMO) and targeted temperature management (TTM), which appear to convey a strong preservation from anoxia-related injuries.1Rea T. Kudenchuk P.J. Sayre M.R. Doll A. Eisenberg M. Out of hospital cardiac arrest: past, present, and future.Resuscitation. 2021; 165: 101-109https://doi.org/10.1016/j.resuscitation.2021.06.010Summary Full Text Full Text PDF PubMed Scopus (15) Google Scholar Eventually, future perspectives in OHCA are promising, with the possibility of detecting more accurately patients with OHCA at low probability of surviving or having bad neurological outcomes that seems closer; however further validations of SCARS model are needed before routinely considering it in clinical practice. Moreover, even though SCARS might be incredibly useful, it is always necessary to put medical judgment at the forefront in the management of a condition as complex as OHCA: the interaction of artificial tools with clinical sense must be of support and not of substitution. All three authors, Luigi Spadafora, Giuseppe Biondi-Zoccai, and Marco Bernardi, equally contributed to designing, drafting and eventually approving this manuscript. Giuseppe Biondi-Zoccai has consulted for Amarin, Balmed, Cardionovum, Crannmedical, Endocore Lab, Eukon, Innovheart, Guidotti, Meditrial, Microport, Opsens Medical, Replycare, Teleflex, and Terumo. All other authors report no conflict of interest. Funding: None. Predicting survival and neurological outcome in out-of-hospital cardiac arrest using machine learning: the SCARS modelThirty-day survival and neurological outcome in OHCA can rapidly and reliably be estimated during ongoing cardiopulmonary resuscitation in the emergency room using a machine learning model incorporating widely available variables.