Cardiac arrest: prediction models in the early phase of hospitalization

Relationship between initial red cell distribution width and ΔRDW and mortality in cardiac arrest patients

AIMS

There has been a lack of research examining the relationship between red cell distribution width (RDW) and the prognosis of cardiac arrest (CA) patients. The prognostic value of the changes in RDW during intensive care unit (ICU) hospitalization for CA patients has not been investigated. This study aims to investigate the correlation between RDW measures at ICU admission and RDW changes during ICU hospitalization and the prognosis of CA patients and then develop a nomogram that predicts the risk of mortality of these patients.

METHODS AND RESULTS

A retrospective cohort study is used to collect clinical characteristics of CA patients (>18 years) that are on their first admission to ICU with RDW data measured from the Medical Information Mart for Intensive Care IV Version 2.0 database. Patients are randomly divided into a development cohort (75%) and a validation cohort (25%). The primary outcome is 30 and 360 day all-cause mortality. ΔRDW is defined as the RDW on ICU discharge minus RDW on ICU admission. A multivariate Cox regression model is applied to test whether the RDW represents an independent risk factor that affects the all-cause mortality of these patients. Meanwhile, the dose-response relationship between the RDW and the mortality is described by restricted cubic spine (RCS). A prediction model is constructed using a nomogram, which is then assessed using receiver operating characteristic curves, calibration curves, and decision curve analysis (DCA). A total of 1278 adult CA patients are included in this study. We found that non-survivors have a higher level of RDW and ΔRDW compared with survivors, and the mortality rate is higher in the high RDW group than in the normal RDW group. The Kaplan-Meier survival curve indicates that patients in the normal RDW group had a higher cumulative survival rate at 30 and 360 days than those in the high RDW group (log-rank test, χ2  = 36.710, χ2  = 54.960, both P values <0.05). The multivariate Cox regression analysis shows that elevated RDW at ICU admission (>15.50%) is an independent predictor of 30 [hazard ratio = 1.451, 95% confidence interval (CI) = 1.181-1.782, P < 0.001] and 360 day (hazard ratio = 1.393, 95% CI = 1.160-1.671, P < 0.001) all-cause mortality among CA patients, and an increase in RDW during ICU hospitalization (ΔRDW ≥ 0.4%) can serve as an independent predictor of mortality among these patients. A non-linear relationship between the RDW measured at ICU admission and the increased risk of mortality rate of these patients is shown by the RCS. This study established and validated a nomogram based on six variables, anion gap, first-day Sequential Organ Failure Assessment score, cerebrovascular disease, malignant tumour, norepinephrine use, and RDW, to predict mortality risk in CA patients. The consistency indices of 30 and 360 day mortality of CA patients in the validation cohort are 0.721 and 0.725, respectively. The nomogram proved to be well calibrated in the validation cohort. DCA curves indicated that the nomogram provided a higher net benefit over a wide, reasonable range of threshold probabilities for predicting mortality in CA patients and could be adapted for clinical decision-making.

CONCLUSIONS

Elevated RDW levels on ICU admission and rising RDW during ICU hospitalization are powerful predictors of all-cause mortality for CA patients at 30 and 360 days, and they can be used as potential clinical biomarkers to predict the bad prognosis of these patients. The newly developed nomogram, which includes RDW, demonstrates high efficacy in predicting the mortality of CA patients.

Variation in coronary angiography use in Out-of-Hospital cardiac arrest

INTRODUCTION

Multiple studies have examined the association of early coronary angiography (CAG) among out-of-hospital cardiac arrest (OHCA) patients with conflicting results. However, patterns of use of CAG among OHCA patients in real-world settings are not well-described.

METHODS

Utilizing data from the Resuscitation Outcomes Consortium's Continuous Chest Compressions trial for our analysis, we stratified patients based on initial arrest rhythm and ST-elevation on initial post-resuscitation electrocardiogram (ECG) and examined the rates of CAG in resuscitated patients. We also examined the rates of CAG across different trial clusters in the overall study population as well as in pre-specified patient subgroups RESULTS: Of 26,148 patients in the CCC trial, 5,608 survived to hospital admission and were enrolled in the study. Among them, 26 % underwent CAG. Patients with ST-elevation underwent CAG at a significantly higher rate than patients presenting without ST-elevation (70 % vs 31 %, p < 0.001). Similarly, patients presenting with shockable rhythms underwent CAG more frequently compared with patients with non-shockable rhythms (28 % vs 5 %, p < 0.001). There was marked variation in CAG frequency across different trial clusters with the proportion of patients within a trial cluster receiving CAG ranging from 4 % - 41 %. The proportion varied more among patients with ST-elevation (16 % - 82 %) or initial shockable rhythm (11 % - 75 %) compared with no ST-elevation (2 % - 28 %) or initial non-shockable rhythm (0 % - 19 %).

CONCLUSION

Among a national cohort of OHCA patients, large variation in the use of CAG exists, highlighting the existing uncertainty regarding perceived benefit from early CAG in OHCA.

Machine Learning-Based Cardiac Arrest Prediction for Early Warning System

The early warning system detects early and responds quickly to emergencies in high-risk patients, such as cardiac arrest in hospitalized patients. However, traditional early warning systems have the problem of frequent false alarms due to low positive predictive value and sensitivity. We conducted early prediction research on cardiac arrest using time-series data such as biosignal and laboratory data. To derive the data attributes that affect the occurrence of cardiac arrest, we performed a correlation analysis between the occurrence of cardiac arrest and the biosignal data and laboratory data. To improve the positive predictive value and sensitivity of early cardiac arrest prediction, we evaluated the performance according to the length of the time series of measured biosignal data, laboratory data, and patient data range. We propose a machine learning and deep learning algorithm: the decision tree, random forest, logistic regression, long short-term memory (LSTM), gated recurrent unit (GRU) model, and the LSTM–GRU hybrid model. We evaluated cardiac arrest prediction models. In the case of our proposed LSTM model, the positive predictive value was 85.92% and the sensitivity was 89.70%.

Intensive care-treated cardiac arrest: a retrospective study on the impact of extended age on mortality, neurological outcome, received treatments and healthcare-associated costs

Background

Cardiac arrest (CA) is a leading cause of death worldwide. As population ages, the need for research focusing on CA in elderly increases. This study investigated treatment intensity, 12-month neurological outcome, mortality and healthcare-associated costs for patients aged over 75 years treated for CA in an intensive care unit (ICU) of a tertiary hospital.

Methods

This single-centre retrospective study included adult CA patients treated in a Finnish tertiary hospital’s ICU between 2005 and 2013. We stratified the study population into two age groups: <75 and \documentclass[12pt]{minimal} \usepackage{amsmath} \usepackage{wasysym} \usepackage{amsfonts} \usepackage{amssymb} \usepackage{amsbsy} \usepackage{mathrsfs} \usepackage{upgreek} \setlength{\oddsidemargin}{-69pt} \begin{document}$$\ge$$\end{document}≥75 years. We compared interventions defined by the median daily therapeutic scoring system (TISS-76) between the age groups to find differences in treatment intensity. We calculated cost-effectiveness by dividing the total one-year healthcare-associated costs of all patients by the number of survivors with a favourable neurological outcome. Favourable outcome was defined as a cerebral performance category (CPC) of 1–2 at 12 months after cardiac arrest. Logistic regression analysis was used to identify independent associations between age group, mortality and neurological outcome.

Results

This study included a total of 1,285 patients, of which 212 (16 %) were \documentclass[12pt]{minimal} \usepackage{amsmath} \usepackage{wasysym} \usepackage{amsfonts} \usepackage{amssymb} \usepackage{amsbsy} \usepackage{mathrsfs} \usepackage{upgreek} \setlength{\oddsidemargin}{-69pt} \begin{document}$$\ge$$\end{document}≥75 years of age. Treatment intensity was lower for the elderly compared to the younger group, with median TISS scores of 116 and 147, respectively (p < 0.001). The effective cost in euros for patients with a good one-year neurological outcome was €168,000 for the elderly and €120,000 for the younger group. At 12 months after CA 24 % of the patients in the elderly group and 47 % of the patients in the younger group had a CPC of 1–2 (p < 0.001). Age was an independent predictor of mortality (multivariate OR = 2.90, 95 % CI: 1.94–4.31, p < 0.001) and neurological outcome (multivariate OR = 3.15, 95 % CI: 2.04–4.86, p < 0.001).

Conclusions

The elderly ICU-treated CA patients in this study had worse neurological outcomes, higher mortality and lower cost-effectiveness than younger patients. Elderly received less intense treatment. Further efforts are needed to recognize the tools for assessing which elderly patients benefit from a more aggressive treatment approach in order to improve the cost-effectiveness of post-CA management.

Supplementary Information

The online version contains supplementary material available at 10.1186/s13049-021-00923-0.

Prediction of In-Hospital Cardiac Arrest Using Shallow and Deep Learning

Sudden cardiac arrest can leave serious brain damage or lead to death, so it is very important to predict before a cardiac arrest occurs. However, early warning score systems including the National Early Warning Score, are associated with low sensitivity and false positives. We applied shallow and deep learning to predict cardiac arrest to overcome these limitations. We evaluated the performance of the Synthetic Minority Oversampling Technique Ratio. We evaluated the performance using a Decision Tree, a Random Forest, Logistic Regression, Long Short-Term Memory model, Gated Recurrent Unit model, and LSTM–GRU hybrid models. Our proposed Logistic Regression demonstrated a higher positive predictive value and sensitivity than traditional early warning systems.

[Rational therapeutic decisions in intensive care patients]

Decisions about the initiation, continuation and termination of life-supporting treatments are a permanent challenge in intensive care units (ICUs). Decisions should be based on patient preferences and the medical indication. The medical indication is mainly the result of an assessment of the patient's prognosis and the applicable therapeutic options. Factors influencing the short term prognosis are mostly the severity of the acute leading disease, the number and severity of other organ failures and the response to initial treatment. Long term prognosis is dominated by the severity and number of comorbidities, age and the resulting frailty. Because in many patients all these informations are not available at the time of admission, in these cases a time-limited trial is often justified to gather all this information before a decision is made. These principles of decision making can also applied to situations in which ICU-capacities are limited (e. g. COVID-19 pandemic).

Association of Initial Illness Severity and Outcomes After Cardiac Arrest With Targeted Temperature Management at 36 °C or 33 °C

IMPORTANCE

It is uncertain what the optimal target temperature is for targeted temperature management (TTM) in patients who are comatose following cardiac arrest.

OBJECTIVE

To examine whether illness severity is associated with changes in the association between target temperature and patient outcome.

DESIGN, SETTING, AND PARTICIPANTS

This cohort study compared outcomes for 1319 patients who were comatose after cardiac arrest at a single center in Pittsburgh, Pennsylvania, from January 2010 to December 2018. Initial illness severity was based on coma and organ failure scores, presence of severe cerebral edema, and presence of highly malignant electroencephalogram (EEG) after resuscitation.

EXPOSURE

TTM at 36 °C or 33 °C.

MAIN OUTCOMES AND MEASURES

Primary outcome was survival to hospital discharge, and secondary outcomes were modified Rankin Scale and cerebral performance category.

RESULTS

Among 1319 patients, 728 (55.2%) had TTM at 33 °C (451 [62.0%] men; median [interquartile range] age, 61 [50-72] years) and 591 (44.8%) had TTM at 36 °C (353 [59.7%] men; median [interquartile range] age, 59 [48-69] years). Overall, 184 of 187 patients (98.4%) with severe cerebral edema died and 234 of 243 patients (96.3%) with highly malignant EEG died regardless of TTM strategy. Comparing TTM at 33 °C with TTM at 36 °C in 911 patients (69.1%) with neither severe cerebral edema nor highly malignant EEG, survival was lower in patients with mild to moderate coma and no shock (risk difference, -13.8%; 95% CI, -24.4% to -3.2%) but higher in patients with mild to moderate coma and cardiopulmonary failure (risk difference, 21.8%; 95% CI, 5.4% to 38.2%) or with severe coma (risk difference, 9.7%; 95% CI, 4.0% to 15.3%). Interactions were similar for functional outcomes. Most deaths (633 of 968 [65.4%]) resulted after withdrawal of life-sustaining therapies.

CONCLUSIONS AND RELEVANCE

In this study, TTM at 33 °C was associated with better survival than TTM at 36 °C among patients with the most severe post-cardiac arrest illness but without severe cerebral edema or malignant EEG. However, TTM at 36 °C was associated with better survival among patients with mild- to moderate-severity illness.

Ethical considerations about artificial intelligence for prognostication in intensive care

Background

Prognosticating the course of diseases to inform decision-making is a key component of intensive care medicine. For several applications in medicine, new methods from the field of artificial intelligence (AI) and machine learning have already outperformed conventional prediction models. Due to their technical characteristics, these methods will present new ethical challenges to the intensivist.

Results

In addition to the standards of data stewardship in medicine, the selection of datasets and algorithms to create AI prognostication models must involve extensive scrutiny to avoid biases and, consequently, injustice against individuals or groups of patients. Assessment of these models for compliance with the ethical principles of beneficence and non-maleficence should also include quantification of predictive uncertainty. Respect for patients’ autonomy during decision-making requires transparency of the data processing by AI models to explain the predictions derived from these models. Moreover, a system of continuous oversight can help to maintain public trust in this technology. Based on these considerations as well as recent guidelines, we propose a pathway to an ethical implementation of AI-based prognostication. It includes a checklist for new AI models that deals with medical and technical topics as well as patient- and system-centered issues.

Conclusion

AI models for prognostication will become valuable tools in intensive care. However, they require technical refinement and a careful implementation according to the standards of medical ethics.

Identifying out-of-hospital cardiac arrest patients with no chance of survival: An independent validation of prediction rules

BACKGROUND

The Basic life support (BLS) and Advanced life support (ALS) are known prediction rules for termination of resuscitation (TOR) in out-of-hospital cardiac arrest (OHCA). Recently, a new rule was developed by Jabre et al. We aimed to independently validate and compare the predictive accuracy of these rules.

METHODS

OHCA cases in Iceland from 2008 to 2017 from a population-based, prospectively registered database. Primary outcome was survival to discharge among patients that met all conditions of abovementioned rules: BLS (not witnessed by EMS personnel, no defibrillation nor ROSC before transport), ALS (BLS criteria plus not witnessed nor CPR by bystander) and Jabre (not witnessed by EMS personnel, initial rhythm non-shockable, no sustainable ROSC before third dose of adrenaline).

RESULTS

Overall, 568 OHCA patients were included in validation of TOR rules. Mean age 67, males 74%, witnessed by EMS 11%, by bystander 66% that attempted CPR in 50%, transported to hospital 60%, overall survival 20%. All rules had high specificity for mortality, 99.6-100% (95%CI 95-100). The Jabre and BLS rules had similar sensitivity 47% (43-52) vs. 44% (40-49), respectively, the sensitivity of ALS was lower, 8% (5-11). Combined use of positive BLS and Jabre rules performed the best, identifying 88/226 (39%) of futile cases transported to hospital, specificity 100% (97-100) and sensitivity 59% (55-64).

CONCLUSIONS

The accuracy of the BLS and Jabre TOR rules to predict mortality after OHCA is very good and their combined use may be superior to the use of either one.

The second information revolution: digitalization brings opportunities and concerns for public health

The spread of the written word, facilitated by the introduction of the printing press, was an information revolution with profound implications for European society. Now, a second information revolution is underway, a digital transformation that is shaping the way Europeans live and interact with each other and the world around them. We are confronted with an unprecedented expansion in ways to share and access information and experiences, to express ourselves and communicate. Yet while these changes have undoubtedly provided many benefits for health, from information sharing to improved surveillance and diagnostics, they also open up many potential threats. These come in many forms. Here we review some the pressing issues of concern; discrimination; breaches of privacy; iatrogenesis; disinformation and misinformation or 'fake news' and cyber-attacks. These have the potential to impact negatively on the health and wellbeing of individuals as well as entire communities and nations. We call for a concerted European response to maximize the benefits of the digital revolution while minimizing the harms, arguably one of the greatest challenges facing the public health community today.