Mortality Risk Stratification Research Articles

Abstract 16883: Validation of a Deep Neural Network Electrocardiographic-Age as a Mortality Predictor: The CODE Study

Introduction: Aging affects the electrocardiogram (ECG) with a higher incidence of abnormalities in older patients. ECG-age can be predicted by artificial intelligence (AI) and can be used as a measure of cardiovascular health. Hypothesis: ECG-age predicted by AI is a risk factor for overall mortality. Methods: The Clinical Outcomes in Digital Electrocardiography (CODE) study is a retrospective cohort with a mean follow-up of 3.67 years.The dataset consists of Brazilian patients, mainly from primary care centers. Two established cohorts, ELSA-Brasil, of Brazilian public servants, and SaMi-Trop, of Chagas disease patients, were used for external validation. 2,322,513 ECGs from 1,558,421 patients over 16 years old that underwent an ECG from 2010 to 2017 were included. A deep convolutional neural network was trained in order to predict the age of the patient based solely on ECG 12-lead tracings. The ECG database was split into 85-15% training and test datasets, respectively. Death was ascertained using probabilistic linkage with Brazil′s mortality information data. The Cox regression model, adjusted by age and sex, was used for statistical analysis. The model was validated in two cohorts: ELSA-Brasil (n=14,263) and SaMi-Trop (n=1,631). Results: he mean predicted ECG-age was 52.0 years (±18.7) with a mean absolute error of 8.38 (±7.0) years. Patients with ECG-age >8y older than chronological age had higher mortality rate (HR 1.79, 95%CI 1.69-1.90; p<0.001), whereas those ECG-age >8y younger than chronological age were associated with a lower mortality rate (HR 0.78, 95%CI 0.74-0.83; p<0.001). These results were similar in ELSA-Brasil and SaMi-Trop external validation cohorts (HR 1.75, 95%CI 1.35-2.27; p<0.001;HR 2.42, 95%CI 1.53-3.83; p<0.001 for >8y difference, retrospectively; HR 0.74, 95%CI 0.63-0.88; p<0.001;HR 0.89, 95%CI 0.52-1.54; p=0.68 for <8y difference, respectively). Conclusions: ECG-age, predicted by AI, can be useful as a tool for risk stratification of mortality.

Abstract 13008: Combined Assessment of American Heart Association Get With the Guidelines-heart Failure Risk Score With Plasma D-dimer and N-terminal Pro-b-type Natriuretic Peptide Levels Improves Prediction of Mortality in Patients Hospitalized for Worsening Heart Failure

American Heart Association Get With the Guidelines-Heart Failure (GWTG-HF) risk score is a useful multivariable score model to predict mortality in patients with acutely decompensated heart failure. We prospectively investigated the prognostic value of combined assessment of GWTG-HF risk score with D-dimer and N-terminal pro-B-type natriuretic peptide (NT-proBNP). Methods: On admission, we measured plasma D-dimer and NT-proBNP levels in 1735 patients (mean age, 75 yrs) hospitalized for worsening heart failure. GWTG-HF risk score, including race, age, systolic blood pressure, heart rate, blood urea nitrogen level, sodium concentration, and presence of chronic obstructive pulmonary disease, was calculated at admission. Results: During a follow-up period of 12 month after admission, 371 (21%) deaths occurred. Patients who died were older (mean: 78 vs. 74 years; P<0.001), had a higher frequency of NYHA functional class IV (63 vs. 49%; P<0.001), higher values of GWTG-HF risk score (43 vs. 38; P<0.001), D-dimer (2.90 vs. 1.30 μg/mL; P<0.001), and NT-proBNP (7202 vs. 3689 pg/mL; P<0.001) than survivors. In the multivariate logistic analysis, elevation (more than the highest tertile value) in D-dimer (>2.5 μg/mL; P<0.001), NT-proBNP (>7399 pg/mL; P<0.001), and GWTG-HF risk score (>42 point; P<0.001) were independently associated with mortality within 12 months. Furthermore, adding D-dimer and NT-proBNP to the GWTG-HF risk score improved the C-index (P<0.01), net reclassification improvement (P<0.001), and integrated discrimination improvement (P<0.001) greater than GWTG-HF risk score alone. The number of elevation in GWTG-HF risk score, D-dimer, and NT-proBNP was strongly associated with a higher risk of mortality within 12 months ( Figure ). Conclusion: The combined assessment of GWTG-HF risk score with D-dimer and NT-proBNP facilitates the risk stratification for mortality within 12 months in patients hospitalized for worsening heart failure.

Abstract 15752: Acute Kidney Injury and Shock Severity for Mortality Risk Stratification in Cardiac Intensive Care Unit Patients

Introduction: Acute kidney injury (AKI) is a highly prevalent risk factor for mortality among patients with cardiogenic shock (CS). We sought to assess the incidence and prognostic relevance of AKI as a function of shock severity in unselected Cardiac Intensive Care Unit (CICU) patients, as measured by the Society for the Cardiovascular Angiography and Interventions (SCAI) shock stage. Methods: We retrospectively reviewed admissions to the Mayo Clinic from 2007 to 2015 and stratified patients by the SCAI shock stage. AKI was defined and staged based on changes in serum creatinine during hospitalization as per KDIGO guidelines. Predictors of in-hospital mortality were analyzed using Kaplan-Meier survival analysis, and one-year mortality was analyzed using Cox proportional-hazards analysis. Results: The final study population included 10,004 unique patients with a mean age of 67 years and 37% females. The percentage of patients with SCAI shock stages A, B, C, D, and E were 47%, 30%, 15%, 7%, and 1%, respectively. AKI of any severity occurred in 51% of patients during hospitalization, including severe (stage 2/3) AKI in 16%. The incidence of AKI and severe AKI increased with the SCAI shock stage. Hospital mortality occurred in 8% of patients and increased as a function of the AKI stage and SCAI shock stage. AKI was associated with increased hospital mortality after multivariable adjustment (adjusted OR per AKI stage 1.17, 95% CI 1.05-1.30, p=0.005). Twenty-one percent of patients died within one year of CICU admission, and worse AKI was associated with increased one-year mortality (adjusted HR per AKI stage 1.11, 95% CI 1.05-1.18, p=<0.001). Hospital survivors with AKI of any severity had higher mortality compared with patients who did not have AKI (p<0.001). Conclusions: AKI was increasingly common in CICU patients with higher shock severity. In-hospital and one-year mortality risk increased as a function of the severity of AKI and the SCAI shock stage. This analysis emphasizes the importance of AKI as a complication of shock and as a predictor of adverse outcomes in CICU patients.

High-Sensitivity Cardiac Troponin I for Risk Stratification in Older Adults.

Traditional cardiovascular risk factors are less predictive in older age. High-sensitivity cardiac troponin I (hs-cTnI) is a marker of subclinical cardiomyocyte damage associated with cardiovascular risk in middle-aged adults. We hypothesized hs-cTnI would be indicative of mortality and cardiovascular risk beyond traditional cardiovascular risk factors in older adults and may be more discriminatory compared to hs-troponin T (hs-cTnT). Prospective cohort study. Population-based Atherosclerosis Risk in Communities (ARIC) Study. We included 5,876 ARIC participants at Visit 5 (2011-2013). We used Cox regression for the association of hs-cTnI categories (women: <4, 4-<10, ≥10 ng/ml; men: <6, 6-<12, ≥12 ng/ml, prevalent cardiovascular disease (CVD)) with mortality and incident CVD (atherosclerotic CVD [ASCVD]: coronary heart disease or stroke, or heart failure). Participants were ages 66 to 90, 23% black, 42% male, and 24% had prevalent CVD. There were 1,053 (321 CVD) deaths (median follow-up 6.3 years). Participants with elevated hs-cTnI and no CVD (7% of participants) had mortality risk similar to those with a history of CVD (55.6 vs 55.7 deaths/1,000 person-years, P = .99). After adjustment, elevated hs-cTnI and no CVD (hazard ratio (HR) = 2.38, 95% confidence interval (CI) = 1.85-3.06) and prevalent CVD (HR = 2.21, 95% CI = 1.90-2.57) remained associated with mortality, compared to low hs-cTnI and no CVD. Elevated hs-cTnI was independently associated with incident CVD (HR = 3.41, 95% CI = 2.58-4.51), ASCVD (HR = 2.02, 95% CI = 1.36-2.98), and heart failure (HR = 6.16, 95% CI = 4.24-8.95). The addition of hs-cTnI significantly improved C-statistics for all outcomes and added greater discrimination than hs-cTnT for cardiovascular mortality and incident heart failure. Hs-cTnI improves mortality and CVD risk stratification in older adults beyond traditional risk factors and improved model discrimination more than hs-cTnT for certain outcomes. Elevated hs-cTnI without CVD identifies a high-risk group with comparable mortality risk as those with a history of clinical CVD.

Phenotypic difference of sarcopenia in the prediction of mortality for elderly patients with heart failure: from CURE-HF registry

Abstract Background Sarcopenia and malnutrition are associated with mortality in elderly patients with heart failure (HF). However, impact of phenotypic difference of sarcopenia on mortality is not well investigated. We hypothesized that evaluation using both body mass index (BMI) and free-fat mass index (FFMI) enabled distinction of phenotypic difference of sarcopenia and risk stratification for mortality in elderly patients with heart failure. Methods In 800 patients who discharged after treatment for HF were prospectively enrolled from 13 medical centers. Body mass index and FFMI was evaluated. Free-fat mass index was calculated dividing the square of the patients' heights in meters into lean body mass. All-cause mortality was followed-up. Patients were divided into 3 subgroups according to BMI and FFMI values, and compared incidence of mortality among them. Results In our study cohort (median age, 78 [range 72–87]), BMI was 21.1 [18.9, 23.8] and FFMI was 16.7 [15.2, 18.0]. 211 patients were experienced all-cause mortality during 631 [266, 983] days follow-up. In Kaplan-Meier analysis, lower BMI and lower FFMI was associated with all-cause mortality (Log-rank p&amp;lt;0.001, p&amp;lt;0.001, respectively). Furthermore, FFMI was an independent predictor of adverse events after adjustment for age, gender, albumin, hemoglobin, creatinine, brain natriuretic peptide, and left ventricular ejection fraction (HR 95% CI: 0.841 (0.745–0.944), p=0.004). In subgroup analysis, comparing with low-BMI and low-FFMI subgroup, better prognosis was observed in the other 2 subgroups (Log-rank p&amp;lt;0.001, p=0.022, Figure 1). Conclusions Phenotypic difference was evident in elderly patients with HF. A combination of BMI and FFMI would be useful for risk stratification of mortality in those patients. Figure 1 Funding Acknowledgement Type of funding source: None

Prognostic role of left ventricular diastolic function assessed by speckle tracking echocardiography in patients after surgical ventricular repair

Abstract Background Assessment of left ventricular (LV) diastolic function brings important prognostic information for patients with heart failure and could be evaluated by speckle tracking echocardiography (STE). Less known about its role in patients planned for surgical ventricular repair due to LV aneurysm. Purpose The aim of this study was to evaluate the prognostic role of STE parameters of LV diastolic function for prediction of all-cause mortality in patients after surgical ventricular repair. Methods We retrospectively evaluated data of 163 consecutive pts (mean age 62.3±11.5 years, 74.8% males) with anteroapical LV aneurysm who underwent surgical ventricular repair combined with coronary artery bypass surgery (71.8%) Prognostic role for prediction of all-cause mortality was assessed for various STE parameters, including left atrial strain (LAS) measured as peak reservoir strain and for the ratio of early to late global longitudinal strain rate (GLSRe/GLSRa). Results During a median follow-up of 4.7 years (IQR: 1.6–8.9 years) there were 65 deaths, 5 year survival rate was 73.8 (95% CI 67–79%). Baseline ejection fraction, end-diastolic and end-systolic volumes of LV did not differ between pts who died and survived at 5 year after the surgery, whereas LAS was significantly higher and GLSRe/GLSRa was significantly lower in survivors. Cox proportional hazard model adjusted to demographic and clinical variables demonstrated that LAS and GLSRe/GLSRa were independent predictors of all-cause death, with HR of 0.79 (95% CI 0.66–0.95, p=0.012) for each 5% increase of LAS and HR of 1.24 (95% CI 1.1–1.4, p=0.001) for each 0.5 increase of GLSRe/GLSRa. Moreover, GLSRe/GLSRa remained an independent predictor after additional adjustment for LV end-systolic volume, sphericity index and presence of mitral insufficiency of grade 2 and higher. A significant difference in median survival time was demonstrated according to the following cut-offs: LAS ≥16.7% (12.1 vs. 6.4 years, p=0.01), GLSRe/GLSRa ratio ≥2.3 (3.3 years vs. 10.2 years, p=0.0005) (Figure). The classification and regression tree analysis with the application of all two-dimensional, Doppler and various speckle-tracking echocardiographic parameters revealed that GLSRe/GLSRa and LAS were the most important echocardiographic variables for risk stratification for 5-year mortality. Conclusion This study demonstrates that STE parameters of LV diastolic function are important predictors of all-cause mortality after surgical ventricular repair due to anteroapical aneurysm of LV and could be used in the preoperative decision-making process. Funding Acknowledgement Type of funding source: None

Prognosis of acute kidney injury during acute heart failure: the role of diuretics

Abstract Background Acute kidney injury (AKI) frequently occurs after diuretic treatment initiation during acute heart failure (AHF). Treatment-induced hemoconcentration seems associated with improved prognosis. Transient AKI, with or without hemoconcentration, is of unsettled prognosis. Purpose We aimed to determine the independent prognostic values of transient AKI, persistent AKI and hemoconcentration in the context of hospitalized AHF. Methods Data were obtained from our institution's Clinical Data Warehouse. Patients that visited our unit at least once were screened. All hospitalizations in our institution were examined (&amp;gt;30 hospitals). Inclusion criteria were: ≥1 hospitalization with ≥1 recorded furosemide administration and ≥1 AHF ICD-10 code. Only the first hospitalization fulfilling these criteria was considered. AKI during 1–13 days following first furosemide administration was defined based on Kidney Disease Improving Global Outcome guidelines. Hemoconcentration was defined as an increase in serum proteins ≥5 g/l during the same period. We performed multivariate logistic regression to determine which characteristics were predictive of AKI. We used Cox regression of 100-days all-cause mortality using several confounders to determine the prognostic values of transient AKI (lasting &amp;lt;14 days), persistent AKI (lasting ≥14 days) and hemoconcentration. To account for immortality bias, AKI and hemoconcentration were treated as time-dependent covariates. Results We included 579 patients in the study. Median follow-up was 114 days. AKI following furosemide initiation occurred in 234 patients (40.4%). Patients that experienced AKI more frequently suffered from chronic kidney disease (43.6% vs. 33%, p=0.01) or presented with right ventricular dilatation (12% vs. 6.7%, p=0.04). Independent predictors of AKI were arterial hypertension (adjusted OR: 1.86 [1.08–3.22]), elevated serum creatinine at baseline (adjusted OR: 1.07 [1.01–1.14] per 10 μmol/l increase) and initial intravenous furosemide (adjusted OR: 2.42 [1.39–4.29]). Death during follow-up occurred in 35% of patients in the AKI group compared to 21% in the non-AKI group (p&amp;lt;0.001). In Cox regression, persistent AKI was independently associated with increased mortality in a period of 100 days following furosemide initiation (adjusted HR: 2.31 [1.07–4.99]). Transient AKI was not significantly associated with mortality (adjusted HR: 0.64 [0.34–1.19]). Hemoconcentration was independently associated with decreased mortality (adjusted HR: 0.46 [0.27–0.79]). Conclusion In the context of hospitalized AHF, AKI that developed 1–13 days after furosemide initiation and that lasted ≥14 days was independently associated with decreased 100 days survival. Hemoconcentration, using a clinically relevant definition, was independently associated with improved survival. These findings show that serum creatinine and proteins, routinely used and with limited cost, accurately stratify mortality risk during AHF. Kaplan-Meier curves Funding Acknowledgement Type of funding source: None

Higher BNP/NT-proBNP levels stratify prognosis in patients with coronary artery disease but without heart failure

Abstract Background Natriuretic peptides (BNP/NT-proBNP) are routinely used for the diagnosis of heart failure and predicts outcome in patients with both heart failure with preserved and reduced ejection fraction. In addition, natriuretic peptides are associated with incident cardiovascular disease manifestation in primary prevention cohorts. Whether the assessment of BNP/NT-proBNP is of value in patients with coronary artery disease but without heart failure has not been investigated in detail. We here evaluate the association of BNP/NT-pro BNP with mortality patients with coronary artery disease but without known chronic heart failure. Methods The present analysis is based on the ECAD registry of patients undergoing conventional coronary angiography at the Department of Cardiology and Vascular Medicine between 2004 and 2019. For this analysis, we excluded all patients with a diagnosis of heart failure or with elevated BNP/NT-proBNP values at baseline (&amp;gt;100pg/nl for BNP, &amp;gt;400pg/nl for NTproBNP). Moreover, patients with missing follow-up information or without BNP/NT-proBNP levels at admission were excluded. As either BNP or NT-proBNP was available for singular patients, we standardized BNP and NT pro BNP levels based on percentile rank in levels from 0 to 99. Cox regression analysis was used to determine the association of BNP/NT-proBNP with morality in unadjusted and risk factor adjusted models with effect sizes depicted per one standard deviation change in BNP/NT-proBNP rank. Results Overall, 3738 patients (mean age: 62.8±12.6 years, 71% male) were included in our analysis. During a mean follow-up of 2.6±3.5 years, 172 deaths of any cause occurred. Patients without fatal events had significantly lower BNP/NT-prBNP values compared to patients who died (48.4±28.8 vs. 58.4±27.5, p&amp;lt;0.0001). In unadjusted cox regression analysis, BNP/NT-proBNP increase by one standard deviation was associated with a 47% increased risk of morality (HR (95% CI): 1.47 (1.25–1.72), p&amp;lt;0.0001). Upon adjustment for cardiovascular risk factors, the significant link between BNP/NT-proBNP levels and morality remained (HR (95% CI): 1.38 (1.14–1.66). Effect sizes were similar for patients receiving coronary revascularization therapy as part of the coronary angiography (1.32 [1.03–1.70], p=0.03) as well as for patients with purely diagnostic procedures (1.58 [1.28–1.94], p&amp;lt;0.0001). Conclusion In patients without heart failure undergoing coronary angiography, BNP/NT-proBNP levels stratify mortality risk independently of traditional cardiovascular risk factors. Our results support the routine assessment of natriuretic peptides also in patients without heart failure to identify patients at increased risk. Funding Acknowledgement Type of funding source: None

The predictive value of preimplant pulmonary function testing in LVAD patients

The predictive value of preoperative pulmonary function testing (PFT) in left ventricular assist device (LVAD) patients remains unknown. This study evaluates the relationship between abnormal PFTs and postimplant outcomes in LVAD patients. LVAD implants from January 2004 to December 2018 at a single institution were included. Patients were stratified based on the presence of abnormal preoperative PFTs, and the primary outcome was respiratory adverse events (AE). Secondary outcomes included 1-year overall postimplant survival, and complications including bleeding, renal failure, thromboembolism, and device malfunction. The total of 333 patients underwent LVAD implant, 46.5% (n = 155) with normal PFTs and 53.5% (n = 178) with abnormal PFTs. Patients with abnormal PFTs were noted to have higher rates of respiratory AEs (25.9% vs. 15.1%, p = .049). In multivariable analysis, the impact of PFTs was most significant when forced expiratory volume in 1 s/forced expiratory volume (FEV1/FVC) ratio was less than 0.5 (hazard ratio [HR] 16.32, 95% confidence interval [CI], 1.70-156.78). The rates of other AEs including bleeding, renal failure, right heart failure, and device malfunction were similar. One-year overall postimplant survival was comparable between the groups (56.8% vs. 68.8%, p = .3183), though patients in the lowest strata of FEV1 (<60% predicted) and FEV1/FVC (<0.5) had elevated risk-adjusted hazards for mortality (HR 2.63, 95% CI, 1.51-4.60 and HR 18.92, 95% CI, 2.10-170.40, respectively). The presence of abnormal preoperative PFTs is not prohibitory for LVAD implantation although it can be used for risk stratification for respiratory AEs and mortality, particularly in patients with severely reduced metrics. The importance of careful patient selection should be underscored in this higher risk patient subset.

Cystatin C-based CKD-EPI estimated glomerular filtration rate equations as a better strategy for mortality stratification in acute heart failure

An increasing number of studies outline renal function as an important risk marker for mortality in acute heart failure (AHF). However, routine estimation of glomerular filtration rate (eGFR) based on serum creatinine is imprecise.This study aims to compare the prognostic impact of CKD-EPI creatinine based equation (eGFRcr), cystatin C based equation (eGFRcyst), and creatinine-cystatin C equation (eGFRcrcyst) for the mortality stratification in AHF.A total of 354 Patients with AHF were prospectively included between January 2012 and June 2016. Creatinine and cystatin C were measured using the same blood sample tube on admission. We quantified eGFR by the eGFRcr, eGFRcyst, and eGFRcrcyst equations. The continuous net reclassification improvement (cNRI) and integrated discrimination improvement (IDI) were calculated to compare the discriminative prognostic value of different CKD-EPI formula.After a median follow-up of 35 months, 161 patients (45.5%) died. Reduced eGFRcyst and eGFRcrcyst remained significant association with death after adjustment. eGFRcyst showed the best area under the curve value (0.706) for the prediction of all-cause mortality. Considering mortality reclassification, both eGFRcyst (IDI = 7.3%, P < .001; cNRI = 19.6%, P = .012) and eGFRcrcyst (IDI = 4.3%, P < .001; cNRI = 8.7%, P = .138) showed its tendency in improving risk prediction compared to eGFRcr. Compared to eGFRcrcyst showed, eGFRcyst further improved mortality stratification (IDI = 3%, P = .049; cNRI = 11.1%, P = .036).In patients with AHF, our study demonstrates the eGFR calculated by CKD-EPI cystatin C-based equation improved the risk stratification of mortality over both creatinine-based and creatinine/cystatin C-based equations.

Qualitative and quantitative chest CT parameters as predictors of specific mortality in COVID-19 patients.

PurposeTo test the association between death and both qualitative and quantitative CT parameters obtained visually and by software in coronavirus disease (COVID-19) early outbreak.MethodsThe study analyzed retrospectively patients underwent chest CT at hospital admission for COVID-19 pneumonia suspicion, between February 21 and March 6, 2020. CT was performed in case of hypoxemia or moderate-to-severe dyspnea. CT scans were analyzed for quantitative and qualitative features obtained visually and by software. Cox proportional hazards regression analysis examined the association between variables and overall survival (OS). Three models were built for stratification of mortality risk: clinical, clinical/visual CT evaluation, and clinical/software-based CT assessment. AUC for each model was used to assess performance in predicting death.ResultsThe study included 248 patients (70% males, median age 68 years). Death occurred in 78/248 (32%) patients. Visual pneumonia extent > 40% (HR 2.15, 95% CI 1.2–3.85, P = 0.01), %high attenuation area – 700 HU > 35% (HR 2.17, 95% CI 1.2–3.94, P = 0.01), exudative consolidations (HR 2.85–2.93, 95% CI 1.61–5.05/1.66–5.16, P < 0.001), visual CAC score > 1 (HR 2.76–3.32, 95% CI 1.4–5.45/1.71–6.46, P < 0.01/P < 0.001), and CT classified as COVID-19 and other disease (HR 1.92–2.03, 95% CI 1.01–3.67/1.06–3.9, P = 0.04/P = 0.03) were significantly associated with shorter OS. Models including CT parameters (AUC 0.911–0.913, 95% CI 0.873–0.95/0.875–0.952) were better predictors of death as compared to clinical model (AUC 0.869, 95% CI 0.816–0.922; P = 0.04 for both models).ConclusionsIn COVID-19 patients, qualitative and quantitative chest CT parameters obtained visually or by software are predictors of mortality. Predictive models including CT metrics were better predictors of death in comparison to clinical model.Supplementary InformationThe online version contains supplementary material available at 10.1007/s10140-020-01867-1.

Lactate and hypotension as predictors of mortality after in-hospital cardiac arrest

AimGuidance on post-cardiac arrest prognostication is largely based on data from out-of-hospital cardiac arrest (OHCA), despite clear differences between the OHCA and in-hospital cardiac arrest (IHCA) populations. Early prediction of mortality after IHCA would be useful to help make decisions about post-arrest care. We evaluated the ability of lactate and need for vasopressors after IHCA to predict hospital mortality. MethodsSingle center retrospective observational study of adult IHCA patients who achieved sustained return of spontaneous circulation (ROSC), required mechanical ventilation peri-arrest and had a lactate checked within 2 h after ROSC. We evaluated the association of post-ROSC lactate and need for vasopressors with mortality using multivariate logistic regression. ResultsA total of 364 patients were included. Patients who received vasopressors within 3 h after ROSC had significantly higher mortality compared to patients who did not receive vasopressors (58% vs. 43%, p = 0.03). Elevated lactate level was associated with mortality (44% if lactate <5 mmol/L, 58% if lactate 5–10 mmol/L, and 73% if lactate >10 mmol/L, p < 0.01). A multivariable model with lactate group and post-ROSC vasopressor use as predictors demonstrated moderate discrimination (AUC 0.64 [95%CI:0.59−0.70]). Including other variables, the most parsimonious model included lactate, age, body mass index, race, and history of arrhythmia, cancer and/or liver disease (AUC 0.70 [95% CI: 0.64−0.75]). ConclusionPost-ROSC lactate and need for vasopressors may be helpful in stratifying mortality risk in patients requiring mechanical ventilation after IHCA.

Mortality risk stratification using artificial intelligence-augmented electrocardiogram in cardiac intensive care unit patients.

An artificial intelligence-augmented electrocardiogram (AI-ECG) algorithm can identify left ventricular systolic dysfunction (LVSD). We sought to determine whether this AI-ECG algorithm could stratify mortality risk in cardiac intensive care unit (CICU) patients, independent of the presence of LVSD by transthoracic echocardiography (TTE). We included 11266 unique Mayo Clinic CICU patients admitted from 2007 to 2018 who underwent AI-ECG after CICU admission. Left ventricular ejection fraction (LVEF) data were extracted for patients with a TTE during hospitalization. Hospital mortality was analysed using multivariable logistic regression. Mean age was 68 ± 15 years, including 37% females. Higher AI-ECG probability of LVSD remained associated with higher hospital mortality [adjusted odds ratio (OR) 1.05 per 0.1 higher, 95% confidence interval (CI) 1.02-1.08, P = 0.003] after adjustment for LVEF, which itself was inversely related with the risk of hospital mortality (adjusted OR 0.96 per 5% higher, 95% CI 0.93-0.99, P = 0.02). Patients with available LVEF data (n = 8242) were divided based on the presence of predicted (by AI-ECG) vs. observed (by TTE) LVSD (defined as LVEF ≤ 35%), using TTE as the gold standard. A stepwise increase in hospital mortality was observed for patients with a true negative, false positive, false negative, and true positive AI-ECG. The AI-ECG prediction of LVSD is associated with hospital mortality in CICU patients, affording risk stratification in addition to that provided by echocardiographic LVEF. Our results emphasize the prognostic value of electrocardiographic patterns reflecting underlying myocardial disease that are recognized by the AI-ECG.

Machine learning based early warning system enables accurate mortality risk prediction for COVID-19

Soaring cases of coronavirus disease (COVID-19) are pummeling the global health system. Overwhelmed health facilities have endeavored to mitigate the pandemic, but mortality of COVID-19 continues to increase. Here, we present a mortality risk prediction model for COVID-19 (MRPMC) that uses patients’ clinical data on admission to stratify patients by mortality risk, which enables prediction of physiological deterioration and death up to 20 days in advance. This ensemble model is built using four machine learning methods including Logistic Regression, Support Vector Machine, Gradient Boosted Decision Tree, and Neural Network. We validate MRPMC in an internal validation cohort and two external validation cohorts, where it achieves an AUC of 0.9621 (95% CI: 0.9464–0.9778), 0.9760 (0.9613–0.9906), and 0.9246 (0.8763–0.9729), respectively. This model enables expeditious and accurate mortality risk stratification of patients with COVID-19, and potentially facilitates more responsive health systems that are conducive to high risk COVID-19 patients.

Risk stratification for mortality in cardiovascular disease survivors: A survival conditional inference tree analysis

Background and aimsEfficient analysis strategies for complex network with cardiovascular disease (CVD) risk stratification remain lacking. We sought to identify an optimized model to study CVD prognosis using survival conditional inference tree (SCTREE), a machine-learning method. Methods and resultsWe identified 5379 new onset CVD from 2006 (baseline) to May, 2017 in the Kailuan I study including 101,510 participants (the training dataset). The second cohort composing 1,287 CVD survivors was used to validate the algorithm (the Kailuan II study, n = 57,511). All variables (e.g., age, sex, family history of CVD, metabolic risk factors, renal function indexes, heart rate, atrial fibrillation, and high sensitivity C-reactive protein) were measured at baseline and biennially during the follow-up period. Up to December 2017, we documented 1,104 deaths after CVD in the Kailuan I study and 170 deaths in the Kailuan II study. Older age, hyperglycemia and proteinuria were identified by the SCTREE as main predictors of post-CVD mortality. CVD survivors in the high risk group (presence of 2–3 of these top risk factors), had higher mortality risk in the training dataset (hazard ratio (HR): 5.41; 95% confidence Interval (CI): 4.49–6.52) and in the validation dataset (HR: 6.04; 95%CI: 3.59–10.2), than those in the lowest risk group (presence of 0–1 of these factors). ConclusionOlder age, hyperglycemia and proteinuria were the main predictors of post-CVD mortality. Trial registrationChiCTR-TNRC-11001489.

TCT CONNECT-319 Risk Stratification of All-Cause Mortality Using Resistive Reserve Ratio in Patients With No Obstructive Coronary Artery Disease

TCT CONNECT-319 Risk Stratification of All-Cause Mortality Using Resistive Reserve Ratio in Patients With No Obstructive Coronary Artery Disease

Tailoring the Beers Criteria for Mortality Risk Stratification Among Older Adults Initiating Hemodialysis

Tailoring the Beers Criteria for Mortality Risk Stratification Among Older Adults Initiating Hemodialysis

Estratificação de risco para predição de disseminação e gravidade da Covid-19 no Brasil

Objetivo: realizar estratificação de risco para disseminação e gravidade da Covid-19 nas unidades da federação (UF) brasileiras a partir de características apontadas como situações de risco. Métodos: foram selecionados alguns indicadores sociais, demográficos e de saúde e submetidos à análise de componentes principais. Em seguida foi possível dividir as UF por análise de cluster. A partir da carga fatorial dos componentes, obtivemos um escore para as UF, que foram estratificadas quanto ao risco de disseminação e mortalidade da Covid-19. Resultados: os componentes referem-se às condições assistenciais, de saúde (incluindo fatores de risco), demográficas e sociais. Estes componentes permitiram a classificação final das 27 UF, com diferença na ordem quanto ao potencial de disseminação e a mortalidade. Conclusão: espera-se que a estratificação de risco possa ser uma medida de apoio à saúde pública, definindo áreas com maior potencial de dano, no sentido de subsidiar a criação de estratégias de intervenção prioritárias.

Association of obesity indices with in-hospital and 1-year mortality following acute coronary syndrome.

According to the "obesity paradox", adults with obesity have a survival advantage following acute coronary syndrome, compared with those without obesity. Previous studies focused on peripheral obesity and whether this advantage is conferred by central obesity is unknown. The objective of this study was to describe the association of peripheral and central obesity indices with risk of in-hospital and 1-year mortality following acute coronary syndrome (ACS). Gulf COAST is a prospective ACS registry that enrolled 4044 patients age ≥18 years from January 2012 through January 2013, across 29 hospitals in four Middle Eastern countries. Associations of indices of peripheral obesity (body-mass index, [BMI]) and central obesity (waist circumference [WC] and waist-to-height ratio [WHtR]) with mortality following ACS were analyzed in logistic regression models (odds ratio, 95% CI) with and without adjustment for Global Registry of Acute Coronary Events risk score. Of 3882 patients analyzed (mean age: 60 years; 33.3% women [n = 1294]), the prevalence of obesity was 34.5% (BMI ≥ 30.0 kg/m2), 72.2% (WC ≥ 94.0 cm [men] or ≥80.0 cm [women]) and 90.0% (WHtR ≥ 0.5). In adjusted models, deciles of obesity indices showed higher risk of mortality at extreme versus intermediate deciles (U-shaped). When defined by conventional cut-offs, peripheral obesity (BMI ≥ 30.0 versus 18.5-29.9 kg/m2) showed inverse association with risk of in-hospital mortality (0.64; 95% CI, 0.42-0.99; P = 0.04; central obesity showed trend toward reduced mortality). In contrast, for risk of 1-year mortality, all indices showed inverse association. Obesity, defined by presence of all three indices, versus nonobesity showed inverse association with risk of 1-year mortality (0.52; 95% CI, 0.35-0.75; P = 0.001). Results were similar among men and women. The degree of obesity paradox following ACS depends on the obesity index and follow-up time. Obesity indices may aid in risk stratification of mortality following ACS.

Changes in Stage Distribution and Disease-Specific Survival in Differentiated Thyroid Cancer with Transition to American Joint Committee on Cancer 8th Edition: A Systematic Review and Meta-Analysis.

Recent revision significantly changed the American Joint Committee on Cancer (AJCC) staging criteria for differentiated thyroid cancer (DTC). To quantitatively evaluate resulting changes in patient stage distribution and the associated disease-specific survival (DSS) incorporating diverse populations, we performed a meta-analysis of studies comparing the AJCC 7th edition (AJCC-7) with 8th edition (AJCC-8) staging for DTC. After PROSPERO registration (#CRD42019123657), publications in English reporting DSS of DTC with AJCC-7 and AJCC-8 from inception to June 2019 were identified by search of MEDLINE and PubMed. Random-effects meta-analyses were conducted to compare differences in survival between AJCC-7 and AJCC-8. Pooled hazard ratios, 10-year DSS, and corresponding interval estimates were calculated for AJCC subgroups. Differences in survival between editions were assessed using subgroup analysis with nonoverlapping confidence intervals indicating statistical significance. Final analysis included six studies with 10,850 subjects and median follow-up from 55 to 148 months. Use of AJCC-8 shifted classification to earlier stages: stage I, from 60% to 81%; stage II, from 5% to 13%; stage III, from 21% to 2%; stage IV, from 10% to 3%. Ten-year DSS was significantly lower in AJCC-8 versus AJCC-7 in patients with stage II (88.6%, 95% confidence interval [CI] 82.7-94.6% vs. 98.1%, 95% CI 96.6-99.6%, respectively) and stage III disease (70.5%, 95% CI 59.1-83.9% vs. 96.8%, 95% CI 94.1-99.64%, respectively). Meta-analysis of revised AJCC staging for DTC, incorporating diverse populations, demonstrates redistribution of patients toward earlier clinical stages and better stratification of disease-specific mortality risk, specifically among patients now classified with stage II and III disease. This study provides updated estimates of disease-specific survival for patients with differentiated thyroid cancer determined by the American Joint Committee on Cancer staging system that are generalizable to broader populations and support improved stratification using the recently revised criteria.