Predictor Of 1-year Mortality Research Articles

The clinical frailty scale predicts 1-year mortality in emergency department patients aged 65 years and older.

ObjectiveTo validate the Clinical Frailty Scale (CFS) for prediction of 1‐year all‐cause mortality in the emergency department (ED) and compare its performance to the Emergency Severity Index (ESI).MethodsProspective cohort study at the ED of a tertiary care center in Northwestern Switzerland. All patients aged ≥65 years were included from March 18 to May 20, 2019, after informed consent. Frailty status was assessed using CFS, excluding level 9 (palliative). Acuity level was assessed using ESI. Both CFS and ESI were adjusted for age, sex and presenting condition in multivariable logistic regression. Prognostic performance was assessed for discrimination and calibration separately. Estimates were internally validated by Bootstrapping. Restricted mean survival time (RMST) was determined for all levels of CFS.ResultsIn the final study population of 2191 patients, 1‐year all‐cause mortality was 17% (n = 372). RMST values ranged from 219 days for CFS 8 to 365 days for CFS 1. The adjusted CFS model had an area under receiver operating characteristic of 0.767 (95% confidence interval [CI]: 0.741–0.793), compared to 0.703 (95% CI: 0.673–0.732) for the adjusted ESI model.ConclusionThe CFS predicts 1‐year all‐cause mortality for older ED patients and predicts survival time in a graded manner. The CFS is superior to the ESI when adjusted for age, sex, and presenting condition.

Medium-Term Outcomes in COVID-19.

COVID-19 causes severe illness that results in morbidity and mortality. Electrocardiographic features, including QT prolongation, have been associated with poor acute outcomes; data on the medium-term outcomes remain scarce. This study evaluated the 1-year outcomes of patients who survived the acute COVID-19 infection. Methods and Materials: Data of the 159 patients who survived the COVID-19 illness during the first wave (1 March 2020–18 May 2020) were collected. Patient demographics, laboratory findings and electrocardiography data were evaluated. Patients who subsequently died within 1-year of the index illness were compared to those who remained well. Results: Of the 159 patients who had survived the index illness, 28 (17.6%) subsequently perished within 1-year. In comparison to the patients that were alive after 1-year, the deceased were older (68 vs. 83 years, p < 0.01) and equally male (60.4% vs. 53.6%, p = 0.68), with a similar proportion of hypertension (59.5% vs. 57.1%, p = 0.68), diabetes (25.2% vs. 39.2%, p = 0.096) and ischaemic heart disease (11.5% vs. 7.1%, p = 0.54). The QTc interval for the alive and deceased patients shortened by a similar degree from the illness to post-COVID (−26 ± 33.5 vs. −20.6 ± 30.04 milliseconds, p = 0.5); the post-COVID R-R interval was longer in the alive patients compared to the deceased (818.9 ± 169.3 vs. 761.1 ± 61.2 ms, p = 0.02). A multivariate Cox regression analysis revealed that age (HR1.098 [1.045–1.153], p < 0.01), diabetes (HR3.972 [1.47–10.8], p < 0.01) and the post-COVID R-R interval (HR0.993 [0.989–0.996], p < 0.01) were associated with 1-year mortality. Conclusions: The COVID-19-associated mortality risk extends to the post-COVID period. The QTc does recover following the acute illness and is not associated with outcomes; the R-R interval is a predictor of 1-year mortality.

Prediction of 1-Year Mortality After Heart Transplantation in Adults with Congenital Heart Disease with Machine Learning Models

PurposeMachine learning (ML) can be used assisting clinical decision-making. We developed a ML model for the prediction of 1-year mortality after heart transplantation (HT) in adults with congenital heart disease.MethodsThe United Network for Organ Sharing (UNOS) database was queried from 2000-2020 for adults with congenital heart disease who underwent isolated HT and had at least 1-year follow-up. The cohort was randomly split in derivation (70%) and validation (30%) dataset used to train and test a CatBoost decision tree model respectively. The primary outcome was 1-year mortality. Explainability analysis with Shapley Additive exPlanations (SHAP) was performed.ResultsA total of 1,032 recipients were included in the study (35±13 years, 61% males). At 1-year after HT, there were 205 deaths. After feature selection, area under the curve and predictive accuracy for the final ML model were 0.82 and 77% respectively. The impact of each model variable for each individual prediction in the validation dataset is represented by its SHAP value (Figure 1).ConclusionA ML model developed using data from the UNOS database showed satisfactory predictive accuracy for 1-year mortality after HT in adults with congenital heart disease. Explainability analysis helps interpret the results in a clinical manner. The nature of the selected features along with their effect on the prediction outcome can have significant implications in optimal patient selection, organ allocation and prognostication. Machine learning (ML) can be used assisting clinical decision-making. We developed a ML model for the prediction of 1-year mortality after heart transplantation (HT) in adults with congenital heart disease. The United Network for Organ Sharing (UNOS) database was queried from 2000-2020 for adults with congenital heart disease who underwent isolated HT and had at least 1-year follow-up. The cohort was randomly split in derivation (70%) and validation (30%) dataset used to train and test a CatBoost decision tree model respectively. The primary outcome was 1-year mortality. Explainability analysis with Shapley Additive exPlanations (SHAP) was performed. A total of 1,032 recipients were included in the study (35±13 years, 61% males). At 1-year after HT, there were 205 deaths. After feature selection, area under the curve and predictive accuracy for the final ML model were 0.82 and 77% respectively. The impact of each model variable for each individual prediction in the validation dataset is represented by its SHAP value (Figure 1). A ML model developed using data from the UNOS database showed satisfactory predictive accuracy for 1-year mortality after HT in adults with congenital heart disease. Explainability analysis helps interpret the results in a clinical manner. The nature of the selected features along with their effect on the prediction outcome can have significant implications in optimal patient selection, organ allocation and prognostication.

Explainable machine learning to predict long-term mortality in critically ill ventilated patients: a retrospective study in central Taiwan

BackgroundMachine learning (ML) model is increasingly used to predict short-term outcome in critically ill patients, but the study for long-term outcome is sparse. We used explainable ML approach to establish 30-day, 90-day and 1-year mortality prediction model in critically ill ventilated patients.MethodsWe retrospectively included patients who were admitted to intensive care units during 2015–2018 at a tertiary hospital in central Taiwan and linked with the Taiwanese nationwide death registration data. Three ML models, including extreme gradient boosting (XGBoost), random forest (RF) and logistic regression (LR), were used to establish mortality prediction model. Furthermore, we used feature importance, Shapley Additive exPlanations (SHAP) plot, partial dependence plot (PDP), and local interpretable model-agnostic explanations (LIME) to explain the established model.ResultsWe enrolled 6994 patients and found the accuracy was similar among the three ML models, and the area under the curve value of using XGBoost to predict 30-day, 90-day and 1-year mortality were 0.858, 0.839 and 0.816, respectively. The calibration curve and decision curve analysis further demonstrated accuracy and applicability of models. SHAP summary plot and PDP plot illustrated the discriminative point of APACHE (acute physiology and chronic health exam) II score, haemoglobin and albumin to predict 1-year mortality. The application of LIME and SHAP force plots quantified the probability of 1-year mortality and algorithm of key features at individual patient level.ConclusionsWe used an explainable ML approach, mainly XGBoost, SHAP and LIME plots to establish an explainable 1-year mortality prediction ML model in critically ill ventilated patients.

Machine Learning-Based Short-Term Mortality Prediction Models for Patients With Cancer Using Electronic Health Record Data: Systematic Review and Critical Appraisal.

BackgroundIn the United States, national guidelines suggest that aggressive cancer care should be avoided in the final months of life. However, guideline compliance currently requires clinicians to make judgments based on their experience as to when a patient is nearing the end of their life. Machine learning (ML) algorithms may facilitate improved end-of-life care provision for patients with cancer by identifying patients at risk of short-term mortality.ObjectiveThis study aims to summarize the evidence for applying ML in ≤1-year cancer mortality prediction to assist with the transition to end-of-life care for patients with cancer.MethodsWe searched MEDLINE, Embase, Scopus, Web of Science, and IEEE to identify relevant articles. We included studies describing ML algorithms predicting ≤1-year mortality in patients of oncology. We used the prediction model risk of bias assessment tool to assess the quality of the included studies.ResultsWe included 15 articles involving 110,058 patients in the final synthesis. Of the 15 studies, 12 (80%) had a high or unclear risk of bias. The model performance was good: the area under the receiver operating characteristic curve ranged from 0.72 to 0.92. We identified common issues leading to biased models, including using a single performance metric, incomplete reporting of or inappropriate modeling practice, and small sample size.ConclusionsWe found encouraging signs of ML performance in predicting short-term cancer mortality. Nevertheless, no included ML algorithms are suitable for clinical practice at the current stage because of the high risk of bias and uncertainty regarding real-world performance. Further research is needed to develop ML models using the modern standards of algorithm development and reporting.

Developing machine learning models for prediction of mortality in the medical intensive care unit

Background and objectiveAlert of patient deterioration is essential for prompt medical intervention in the Medical Intensive Care Unit (MICU). Logistic Regression (LR) has been used for the development of most conventional severity-of-illness scoring systems to anticipate the risk of mortality in the MICU. Machine Learning (ML) models such as probabilistic graphical models and Extreme Gradient Boosting (XGB) have demonstrated improved prediction accuracy in patient outcomes compared to LR. The aim was to compare three ML models to the SAPS, SAPS II, SAPS III, SOFA, serial SOFA, LODS, and OASIS for prediction of MICU mortality. MethodsA Bayesian Network (BN), Naïve Bayes network (NB), and a XGB model were developed. 9893 adult MICU-stays from the MIMIC-III database were studied. The primary outcome was MICU mortality prediction and the secondary outcome was 1-year mortality prediction. Data analyzed consisted on routine physiological measurements collected during 5 hours in the MICU, demographic and diagnoses/procedure features. The performance was evaluated by accuracy statistics, discrimination and calibration measures. Limitations of the study were discussed. ResultsThe AUROC for MICU mortality prediction was 0.919 for XGB, 0.905 for BN, and 0.864 for NB, while the conventional systems displayed much lower values with the serial SOFA having the best value (0.814). The Diagnostic Odds Ratio was ≤7.099 for all the conventional systems, reaching values of 30.115 for XGB and 22.648 for BN. The XGB achieved a sensitivity of 0.831 and specificity of 0.86 assuring an acceptable precision (0.528), whose values were much lower for the conventional systems. The Brier score was better for the ML models, except for the NB (0.119), with 0.072 for XGB and 0.081 for BN. ConclusionsThe XGB and BN substantially outperformed the conventional systems for discrimination, calibration and the accuracy statistics assessed. The NB showed inferior performance to the XGB and BN but improved the discrimination and all accuracy statistics of the conventional systems except for an inferior calibration and 1-year mortality discrimination. The XGB showed the best performance among all models. These ML models have the potential to improve the monitoring of MICU patients, which must be evaluated in future studies.

Validation of existing risk scores for mortality prediction after a heart transplant in a Chinese population.

The objectives of this study were to validate 3 existing heart transplant risk scores with a single-centre cohort in China and evaluate the efficacy of the 3 systems in predicting mortality. We retrospectively studied 428 patients from a single centre who underwent heart transplants from January 2015 to December 2019. All patients were scored using the Index for Mortality Prediction After Cardiac Transplantation (IMPACT) and the United Network for Organ Sharing (UNOS) and risk stratification scores (RSSs). We assessed the efficacy of the risk scores by comparing the observed and the predicted 1-year mortality. Binary logistic regression was used to evaluate the predictive accuracy of the 3 risk scores. Model discrimination was assessed by measuring the area under the receiver operating curves. Kaplan-Meier survival analyses were performed after the patients were divided into different risk groups. Based on our cohort, the observed mortality was 6.54%, whereas the predicted mortality of the IMPACT and UNOS scores and the RSSs was 10.59%, 10.74% and 12.89%, respectively. Logistic regression analysis showed that the IMPACT [odds ratio (OR), 1.25; 95% confidence interval (CI), 1.15-1.36; P < 0.001], UNOS (OR, 1.68; 95% CI, 1.37-2.07; P < 0.001) and risk stratification (OR, 1.61; 95% CI, 1.30-2.00; P < 0.001) scores were predictive of 1-year mortality. The discriminative power was numerically higher for the IMPACT score [area under the curve (AUC) of 0.691)] than for the UNOS score (AUC 0.685) and the RSS (AUC 0.648). We validated the IMPACT and UNOS scores and the RSSs as predictors of 1-year mortality after a heart transplant, but all 3 risk scores had unsatisfactory discriminative powers that overestimated the observed mortality for the Chinese cohort.

Thrombolysis in Myocardial Infarction Risk Index Predicts 1-Year Mortality in Patients with Heart Failure: An Analysis of the SELFIE-TR Study

Objective: Predicting outcomes is an essential part of evaluation of patients with heart failure (HF). While there are multiple individual laboratory and imaging variables as well as risk scores available for this purpose, they are seldom useful during the initial evaluation. In this analysis, we aimed to elucidate the predictive usefulness of Thrombolysis in Myocardial Infarction Risk Index (TIMI-RI), a simple index calculated at the bedside with three commonly available variables, using data from a multicenter HF registry. Subjects and Methods: A total of 728 patients from 23 centers were included in this analysis. Data on hospitalizations and mortality were collected by direct interviews, phone calls, and electronic databases. TIMI-RI was calculated as heart rate × (age/10)²/systolic pressure. Patients were divided into three equal tertiles to perform analyses. Results: Rehospitalization for HF was significantly higher in patients within the 3rd tertile, and 33.5% of patients within the 3rd tertile had died within 1-year follow-up as compared to 14.5% of patients within the 1st tertile and 15.6% of patients within the 2nd tertile (p < 0.001, log-rank p < 0.001 for pairwise comparisons). The association between TIMI-RI and mortality remained significant (OR: 1.74, 95% CI: 1.05–2.86, p = 0.036) after adjustment for other variables. A TIMI-RI higher than 33 had a negative predictive value of 84.8% and a positive predictive value of 33.8% for prediction of 1-year mortality. Conclusion: TIMI-RI is a simple index that predicts 1-year mortality in patients with HF; it could be useful for rapid evaluation and triage of HF patients at the time of initial contact.

Bipolar hemiarthroplasty in unstable intertrochanteric fractures in elderly patients. The predictive value of the Charlson Comorbidity Index in 1-year mortality

BackgroundHip fractures have a significant impact on morbidity and mortality in the elderly. Aims: We retrospectively evaluated the predictive role of the Charlson Comorbidity Index (CCI) for 1-year mortality in elderly patients with unstable intertrochanteric hip fractures (ITHF) treated with bipolar hemiarthroplasty. The secondary objective was to identify other relationships, if any, between the variables recorded and mortality. MethodsWe included ≥75-year-old patients with unstable ITHF treated with bipolar hemiarthroplasty. We recorded patient gender, age, Body Mass Index, pre-fracture walking ability (Parker Mobility score, modified Harris Hip Score), America Society of Anesthesiologists (ASA) score, Charlson Comorbidity Index (CCI), time to surgery, time to mobilization, hospital stay, and postoperative complications. Uni- and multivariate logistic regression analysis were performed. Sensitivity and specificity were calculated using a ROC curve. ResultsA total of 135 patients with a mean age of 87.34 ± 5.5 years were included. The overall 1-year mortality rate was 18.5%. The CCI (OR 1.64 CI 95% 1.21–2.23; p 0.00821) and postoperative complications (OR 3.5 CI 95% 1.19–10.23 p 0.0202) were identified as independent predictors of 1-year mortality in the univariate regression and confirmed in the multivariate regression. CCI sensitivity to predict 1-year mortality was 80%. ConclusionCCI has shown acceptable sensitivity in the prediction of 1-year mortality in elderly patients with unstable ITHF treated with bipolar hemiarthroplasty. It is of utmost importance to prevent postoperative complications due to their significant impact on 1-year mortality.

Association of GrimAge DNA methylation components and 2-year mortality in the Health and Retirement Study

DNA methylation (DNAm) patterns related to age and aging phenotypes (i.e., epigenetic clocks) are of growing interest as indicators of biological age and risk of negative health outcomes. We investigated associations between the components of GrimAge, an epigenetic clock estimated from DNAm patterns for seven blood protein levels and smoking pack years, and 2-year mortality in the Health and Retirement Study (HRS) to determine if any of the DNAm subcomponents were driving observed associations. A representative subsample of individuals who participated in the HRS 2016 Venus Blood Study were included in this analysis (N=3430). DNAm was measured with the Infinium Methylation EPIC BeadChip. Deaths that occurred between 2016 and 2018 contributed to 2-year mortality estimates (N=159, 4.5% of the sample). Weighted logistic regression estimated the association first between GrimAge and 2-year mortality and second between the DNAm subcomponents and 2-year mortality. All models were adjusted for age, sex, race/ethnicity, education, current smoking status, smoking pack years and cell composition of the biological sample. The average GrimAge for participants with and without 2-year mortality was 77 years 68 years respectively. A one-year increase in GrimAge was associated with 17% higher odds of 2-year mortality (95% CI: 1.16, 1.17). Two of the seven DNAm blood protein subcomponents of GrimAge (TIMP metallopeptidase inhibitor 1, adrenomedullin) and DNAm smoking pack years were associated with 2-year mortality and DNAm smoking pack years appeared to drive the overall GrimAge association with 2-year mortality. GrimAge was a better predictor of 2-year mortality than the DNAm subcomponents individually.

Independent Prognostic Utility of 11C-Pittsburgh Compound B PET in Patients with Light-Chain Cardiac Amyloidosis.

11C-Pittsburgh compound B (PiB) PET/CT visualizes the amount of myocardial amyloid deposit and can be used to prognosticate patients with amyloid light-chain (AL) cardiac amyloidosis (CA). However, whether 11C-PiB PET/CT has any independent additional prognostic value beyond the commonly used biomarkers remains unknown. Methods: This prospective study was on a cohort of 58 consecutive patients with AL CA who underwent 11C-PiB PET/CT. The patients were stratified into 2 groups on the basis of a visual assessment of whether there was myocardial 11C-PiB uptake on PET/CT. The primary endpoint was 1-y overall mortality. The independent prognostic utility of 11C-PiB PET/CT was analyzed using net reclassification improvement and integrated discrimination improvement. Results: Among the 58 patients enrolled, 35 were positive for myocardial 11C-PiB uptake on PET/CT. Patients with myocardial 11C-PiB PET uptake had a worse 1-y overall survival rate than those without (81.8% vs. 45.5%, P = 0.003 by log-rank test). In the multivariate analysis, positivity for myocardial 11C-PiB uptake on PET/CT was an independent predictor of 1-y mortality (adjusted hazard ratio, 3.382; 95% CI, 1.011-11.316; P = 0.048). In analysis of 3 subgroups of patients-those with a troponin I level of at least 0.1 ng/mL, those with an N-terminal pro-B-type natriuretic peptide (NT-proBNP) level of at least 1,800 pg/mL, and those with a difference of at least 180 mg/L between free light chains (the 3 commonly used biomarkers and their thresholds for staging in AL amyloidosis)-Kaplan-Meier curves showed for all 3 subgroups that patients positive for myocardial 11C-PiB uptake on PET/CT had a worse prognosis than those who were negative. Additionally, when the results of 11C-PiB PET/CT were added to these 3 biomarkers, the performance of 1-y mortality prediction significantly improved by net reclassification improvement (troponin I, 0.861; NT-proBNP, 0.914; difference between free light chains, 0.987) and by integrated discrimination improvement (0.200, 0.156, and 0.108, respectively). Conclusion:11C-PiB PET/CT is a strong independent predictor of 1-y overall mortality and provides incremental prognostic benefits beyond the 3 commonly used biomarkers of AL amyloidosis staging. Considering the recent development of numerous amyloid-targeting molecular imaging agents, further investigations are warranted on whether PET/CT should be included in risk stratification for patients with AL CA.

Clinical and Echocardiographic Parameters Predicting 1- and 2-Year Mortality After Transcatheter Aortic Valve Implantation.

Background: Transcatheter aortic valve implantation (TAVI) has become a standard treatment option for patients with symptomatic aortic stenosis. Elderly high-risk patients treated with TAVI have a high residual mortality due to preexisting comorbidities. Knowledge of factors predicting futility after TAVI is sparse and clinical tools to aid the preoperative evaluation are lacking. The aim of this study was to evaluate if echocardiographic measures, including speckle-tracking analysis, in addition to clinical parameters, could aid in the prediction of mortality beyond 30 days after TAVI.Methods: This prospective observational cohort study included 227 patients treated with TAVI at the University Hospital of North Norway, Tromsø and Oslo University Hospital, Rikshospitalet from February 2010 to June 2013. All the patients underwent preoperative echocardiographic evaluation with retrospective speckle-tracking analysis. Primary endpoints were 1- and 2-year mortality beyond 30 days after TAVI.Results: All-cause 1- and 2-year mortality beyond 30 days after TAVI was 12.1 and 19.5%, respectively. Predictors of 1-year mortality beyond 30 days were body mass index [hazard ratio (HR): 0.88, 95% CI: 0.80–0.98, p = 0.018], previous myocardial infarction (HR: 2.69, 95% CI: 1.14–6.32, p = 0.023), and systolic pulmonary artery pressure ≥ 60 mm Hg (HR: 5.93, 95% CI: 1.67–21.1, p = 0.006). Moderate-to-severe mitral regurgitation (HR: 2.93, 95% CI: 1.53–5.63, p = 0.001), estimated glomerular filtration rate (HR: 0.98, 95% CI: 0.96–0.99, p = 0.002), and chronic obstructive pulmonary disease (HR: 1.9, 95% CI: 1.01–3.58, p = 0.046) were predictors of 2-year mortality.Conclusion: Both the clinical and echocardiographic parameters should be considered when evaluating high-risk patients for TAVI, as both are predictive of 1-and 2-year mortality. Our results support the importance of individual risk assessment using a multidisciplinary, multimodal, and individual approach.

Machine learning-based prediction of 1-year mortality for acute coronary syndrome✰

BackgroundClinical risk assessment with quantitative formal risk scores may add to intuitive physician risk assessment and are advised by the international guidelines for the management of acute coronary syndrome (ACS) patients. Most previous studies have used the binary regression/classification approach (dead/alive) for long-term mortality post-ACS, without considering the time-to-event as in survival analysis. The use of machine learning (ML)-based survival models has yet to be validated.The primary objective was to compare survival prediction performance of 1-year mortality following ACS of two newly developed ML-based models [random survival forest (RSF) and deep learning (DeepSurv)] with the traditional Cox-proportional hazard (CPH) model. The secondary objective was external validation of the findings. MethodsThis was a retrospective, supervised learning data mining study based on the Acute Coronary Syndrome Israeli Survey (ACSIS) and the Myocardial Ischemia National Audit Project (MINAP). The ACSIS data were divided to train/test in a 70/30 fashion. Next, the models were externally validated on the MINAP data. Harrell's C-index, inverse probability of censoring weighting (IPCW), and the Brier-score were used for models’ performance comparison. ResultsRSF performed best among the three models, with Harrell's C-index on training and testing sets reaching 0.953 and 0.924 respectively, followed by CPH multivariate selected model (0.805/0.849), CPH Univariate selected model (0.828/0.806), DeepSurv model (0.801/0.804), and the traditional CPH model (0.826/0.738). The RSF model also had the highest performance on the validation data set with 0.811 for Harrell's C-index, 0.844 for IPCW, and 0.093 for Brier score. The CPH model performance on the validation set had C-index range between 0.689 to 0.790, 0.713 to 0.826 for IPCW, and 0.094 to 0.103 Brier score. ConclusionsRSF survival predictions for long-term mortality post-ACS show improved model performance compared with the classic statistical method. This may benefit patients by allowing better risk stratification and tailored therapy, however further prospective evaluations are required.

How Do Geriatric Scores Predict 1-Year Mortality in Elderly Patients with Suspected Pneumonia?

Background: Pneumonia has an impact on long-term mortality in elderly patients. The risk factors associated with poor long-term outcomes are understated. We aimed to assess the ability of scores that evaluate patients’ comorbidities (cumulative illness rating scale—geriatric, CIRS-G), malnutrition (mini nutritional assessment, MNA) and functionality (functional independence measure, FIM) to predict 1-year mortality in a cohort of older patients having a suspicion of pneumonia. Methods: Our prospective study included consecutive patients over 65 years old and hospitalized with a suspicion of pneumonia enrolled in a monocentric cohort from May 2015 to April 2016. Each score was analysed in univariate and multivariate models and logistic regressions were used to identify contributors to 1-year mortality. Results: 200 patients were included (51% male, mean age 83.8 ± 7.7). Their 1-year mortality rate was 30%. FIM (p < 0.01), CIRS-G (p < 0.001) and MNA (p < 0.001) were strongly associated with poorer long-term outcomes in univariate analysis. CIRS-G (p < 0.05) and MNA (p < 0.05) were significant predictors of 1-year mortality in multivariate analysis. Conclusion: Long-term prognosis of patients hospitalized for pneumonia was poor and we identified that scores assessing comorbidities and malnutrition seem to be important predictors of 1-year mortality. This should be taken into account for evaluating elderly patients’ prognosis, levels and goals of care.

Urgent transcatheter aortic valve implantation in an all-comer population: a single-centre experience

BackgroundWhen compared with older reports of untreated symptomatic aortic valve stenosis (AoS), urgent transcatheter aortic valve implantation (u-TAVI) seems to improve mortality rates. We performed a single centre, retrospective cohort analysis to characterize our u-TAVI population and to identify potential predictors of worse outcomes.MethodsWe performed a retrospective analysis of 631 consecutive TAVI patients between 2013 and 2018. Of these patients, 53 were categorized as u-TAVI. Data was collected from the local electronic database.ResultsUrgent patients had more often a severely decreased left ventricular ejection fraction (LVEF < 30%) and increased creatinine levels (115.5 [88–147] vs 94.5 [78–116] mmol/l; p = 0.001). Urgent patients were hospitalised for 18 [10–28] days before and discharged 6 [4–9] days after the implantation. The incidence of peri-procedural complications and apical implantations was comparable among the study groups. Urgent patients had higher in-hospital (11.3% vs 3.1%; p = 0.011) and 1-year mortality rates (28.2% vs 8.5%, p < 0.001). An increased risk of one-year mortality was associated with urgency (HR 3.5; p < 0.001), apical access (HR 1.9; p = 0.016) and cerebrovascular complications (HR 4.3; p = 0.002). Within the urgent group, the length of pre-hospital admission was the only significant predictor of 1-year mortality (HR 1.037/day; p = 0.003).ConclusionsCompared to elective procedures, u-TAVI led to increased mortality and comparable complication rates. This detrimental effect is most likely related to the length of pre-procedural hospitalisation of urgent patients.

Abstract 13836: Unbiased Deep Learning Approach Utilizing Longitudinal Data in Assessing All-Cause Mortality in Patients With a De Novo or Worsened Heart Failure

Introduction: Heart failure (HF) is a heterogenous syndrome with complex pathophysiology. Biomarkers and clinical risk scores often fail to capture modifications in the treatment continuity and provide suboptimal patient-level precision in the prognostic stratification. Electronic patient records provide necessary granularity yielding opportunities to develop new artificial intelligence (AI) based strategies for comprehensive prognostic re-stratification. Hypothesis: We assessed the hypothesis that, utilizing longitudinal patient data in an AI approach, yields superior performance predicting all-cause mortality in a cohort of patients hospitalized with a de novo or worsened HF, compared to single observational time point predictions. Methods: In a cohort of 2449 HF patients hospitalized between 2011-2017, we utilized 151 451 patient exams from 422 parameters. Features included clinical phenotyping, medication, ECG, laboratory, echocardiography, catheterization data or percutaneous and surgical interventions gathered on a routine clinical basis reflecting standard of care as captured in individual electronic records. AI models were developed, and their performance on the validation set was compared to industry standard clinical scores. Results: AI models yielded performance ranging from 0.83 to 0.89 AUC on the outcome-balanced validation set in predicting all-cause mortality at 30-, 90-, 180-, 360- and 720-day time-limits. The primary endpoint, 1-year mortality prediction model, recorded 0.85 AUC on the validation set compared to 0.7 AUC (Seattle HF model) and 0.73 AUC (MAGGIC HF Score) respectively. Conclusions: Our findings present a novel, patient-level, AI-based risk prediction approach of all-cause mortality in heart failure utilizing all historical data available in electronic health records. This suggests the potential of AI based predictive models in a point-of-care approach to guide clinical risk stratification.

TP9.2.10The risk and predictors of mortality in octogenarians undergoing emergency laparotomy: A multicentre cohort study

Abstract Aims To evaluate the risk and predictors of postoperative mortality in octogenarians undergoing emergency laparotomy. Methods In compliance with STROCSS guideline for observational studies, we conducted a multicentre retrospective cohort study. All consecutive patients aged over 80 with acute abdominal pathology requiring emergency laparotomy between April 2014 and August 2019 were considered eligible for inclusion. The primary outcome measure was 30-day postoperative mortality and the secondary outcome measures were in-hospital mortality and 1-year mortality. Statistical analyses included simple descriptive statistics, binary logistic regression analyses, and Kaplan–Meier survival statistics. Results A total of 523 octogenarians were eligible for inclusion. Emergency laparotomy in octogenarians was associated with 21.8% (95% CI 18.3-25.6%) 30-day postoperative mortality, 22.6% (95% CI 19.0-26.4%) in-hospital mortality, and 40.2% (95% CI 35.9-44.5%) 1-year mortality. Binary logistic regression analysis identified ASA status (OR: 2.49,95% CI 1.82-3.38,P&amp;lt;0.0001) and peritoneal contamination (OR: 2.00, 95% CI 1.30-3.08, P = 0.002) as predictors of 30-day postoperative mortality. The ASA status (OR: 1.92,95% CI 1.50-2.46,P&amp;lt;0.0001), peritoneal contamination (OR: 1.57,95% CI 1.07-2.48,P=0.020) and presence of malignancy (OR: 2.06,95% CI 1.36-3.10,P=0.001) were predictors of 1-year mortality. Log-rank test showed significant difference in postoperative survival rates among patients with different ASA status (P &amp;lt; 0.0001) and between patients with and without peritoneal contamination (P = 0.0011). Conclusions Emergency laparotomies in patients older than 80 years with ASA status more than 3 in the presence of peritoneal contamination carries a high risk of immediate postoperative and 1-year mortality. There is a need to incorporate modern prognosticators into the available preoperative mortality risk assessment tools.

Short- and Long-Term Mortality Prediction in Critically Ill Cancer Patients Admitted to the Intensive Care Unit (CanICU) Using Machine Learning

Background: Although cancer patients are increasingly admitted to the intensive care unit (ICU) for cancer- or treatment-related complications, improved mortality prediction remains a big challenge. Current prognostic models for a general ICU population underestimate the mortality risk in cancer patients. Methods: We developed CanICU, a machine learning-based 28-day and 1-year mortality prediction program in adult cancer patients admitted to ICU from Medical Information Mart for Intensive Care (MIMIC) database in the USA (n=766), Yonsei Cancer Center (YCC, n=3,571), and Samsung Medical Center in Korea (SMC, n=2,563). We constructed the classifier and compared it with a general prognostic model. Findings: A total of 6,900 patients were included, with a 28-day mortality of 10.2%/12.7%/36.6% and 1-year mortality of 30.0%/31.4%/58.5% in the MIMIC, YCC, and SMC cohort, respectively. Nine clinical and laboratory factors were used to construct the classifier using a random forest machine-learning algorithm. CanICU had the area under the receiver operating characteristic (AUROC) of 0.95 (95% CI 0.93-0.97) for 28-day and 0.79 (95% CI 0.76-0.82) for 1-year mortality, showing better performance than a current prognostic model (Sequential Organ Failure Assessment [SOFA], 0.77 [95% CI 0.72-0.81] for 28-day and 0.67 [95% CI 0.64-0.70] for 1-year mortality). Application of CanICU in external data sets across the countries yielded better AUROC for 28-day mortality than SOFA (95% CI, 0.75-0.80 vs. 0.57-0.63 with SOFA in SMC; 95% CI, 0.72-0.79 vs. 0.64-0.72 with SOFA in MIMIC). Interpretation: CanICU offers improved performance for predicting short- and long-term mortality in critically ill cancer patients admitted to ICU. A user-friendly online implementation is available and should be valuable for better mortality risk stratification. This might help physicians to determine to allocate ICU care for patients with cancer. Funding Information: This study was supported in part by the National Research Foundation of Korea (NRF) grant funded by the Korean government (MSIT) (No. 2019R1C1C1006709, 018R1A5A2025079, 2020M3F7A1094093), a grant from the Korea Health Technology R&D Project through the Korea Health Industry Development Institute (KHIDI), funded by the Ministry of Health & Welfare, the Republic of Korea (No. KHIDIHI19C1015010020), and was also supported by the ational R&D Program for Cancer Control through the National Cancer Center (NCC) funded by the Ministry of Health & Welfare, Republic of Korea (HA21C0065). Declaration of Interests: The authors have no potential conflicts of interest to disclose. Ethics Approval Statement: The institutional review boards of all participating hospitals approved this study and waived the requirement for informed consent because of the observational nature of the research. All patient records and data were anonymized and de-identified before analysis.

ADP-receptor antagonists in patients with acute myocardial infarction complicated by cardiogenic shock: a pooled IABP-SHOCK II and CULPRIT-SHOCK trial sub-analysis

Abstract Purpose The purpose of this pooled analysis is to compare the clinical outcome of patients with acute myocardial infarction complicated by cardiogenic shock treated with either clopidogrel or the newer, more potent ADP-receptor antagonists prasugrel or ticagrelor. Patients from the Intraaortic Balloon Pump in Cardiogenic Shock II (IABP-SHOCK II) and Culprit Lesion Only PCI versus Multivessel PCI in Cardiogenic Shock (CULPRIT-SHOCK) trial were included. Methods and results For the current analysis, the primary endpoint was 1-year mortality and the secondary safety endpoint was moderate or severe bleedings until hospital discharge with respect to three different ADP-receptor antagonists. Eight hundred fifty-six patients were eligible for analysis. Of these, five hundred seven patients (59.2%) received clopidogrel, one hundred seventy-eight patients (20.8%) prasugrel and one hundred seventy-one patients (20.0%) ticagrelor as acute antiplatelet therapy. The adjusted rate of mortality after 1-year did not differ between prasugrel and clopidogrel (hazard ratio [HR]: 0.81, 95% confidence interval [CI] 0.60–1.09, padj=0.17) or between ticagrelor and clopidogrel treated patients (HR: 0.86, 95% CI 0.65–1.15, padj=0.31). In-hospital bleeding events were significantly less frequent in patients treated with ticagrelor vs. clopidogrel (HR: 0.37, 95% CI 0.20–0.69, padj=0.002) and not different in patients treated with prasugrel vs. clopidogrel (HR: 0.73, 95% CI 0.43–1.24, padj=0.24), see Table 1. Conclusion This pooled sub-analysis is the largest analysis on safety and efficacy of three oral ADP-receptor antagonists and shows that an acute therapy with either clopidogrel, prasugrel or ticagrelor is no predictor of 1-year mortality. Treatment with ticagrelor seems to be associated with less in-hospital moderate and severe bleeding events in comparison to clopidogrel. Funding Acknowledgement Type of funding sources: Foundation. Main funding source(s): German Heart FoundationEuropean Union 7th Framework Program

Risk prediction in patients with classical Low-flow, low-gradient aortic stenosis undergoing TAVI

Abstract Background Patients with classical low-flow, low-gradient aortic stenosis (LFLG AS) undergoing transcatheter aortic valve implantation (TAVI) are known to suffer from poor prognosis. Risk prediction is difficult and predictors of outcome are scarce in this complex patient subset. Purpose This study aimed to develop a simple score system for risk prediction in patients with classical LFLG AS scheduled for TAVI including baseline clinical, echocardiographic and computed tomography (CT) parameters. Methods Among 9,150 patients undergoing TAVI (2008–2019) at five German centers, patients with classical LFLG AS were screened for availability of sufficient CT data for aortic valve calcification (AVC) quantification. Score variables were defined as all variables included into the best performing multivariable regression model. Score performance was assessed by Kaplan-Meier analysis for the endpoint of 1-year all-cause as well as c-index calculation. Results A total of 718 patients were included into the study. Predictors of all-cause mortality were male sex (HR 1.27 [0.98, 1.65], p=0.068), chronic obstructive pulmonary disease (HR 1.59 [1.21,2.08], p&amp;lt;0.001), atrial fibrillation (HR 1.33 [1.03,1.70], p=0.026), stroke volume index (HR 0.96 [0.94, 0.98], p&amp;lt;0.001), non-transfemoral access (HR 1.44 [1.09,1.91], p=0.011), pulmonary hypertension (HR 1.58 [1.20,2.08], p=0.0012) and low AVC density (HR 1.49 [1.16,1.90], p=0.0016). A score system was developed ranging from 0 points to 11 points. Kaplan-Meier analysis for low (0–2 points), moderate (3–6 points) and high score (&amp;gt;6 points) demonstrated 1-year mortality rates of 20.9%, 28.9% and 64.3% (p&amp;lt;0.001), respectively. C-index for prediction of 1-year mortality was 0.69 (95%-CI 0.65–0.84). The RELiEF TAVI score outperformed classical risk prediction models like the logistic euroSCORE (c-index 0.60 [95%-CI 0.51–0.72], p=0.013), the EuroSCORE II (c-index 0.57 [95%-CI 0.45–0.70], p=0.039) and the STS PROM (c-index 0.60 [0.48–0.70], p=0.035). Conclusions The RELiEF TAVI score is based on simple clinical, echocardiographic and CT parameters and might serve as a helpful tool for risk prediction in the vulnerable subset of patients with classical LFLG AS scheduled for TAVI. Funding Acknowledgement Type of funding sources: None. RELiEF TAVI Score