Comparación pronóstica entre scores de riesgo y la aplicación de redes neuronales para la predicción de la mortalidad a corto y mediano plazo en pacientes con insuficiencia cardíaca / Prognostic Comparison between Risk Scores and Neural Networks to Predict Short- and Mid-Term Mortality in Acute Heart Failure

RESUMEN Introducción: En el contexto de la insuficiencia cardíaca (IC) existen scores de riesgo para evaluar la mortalidad por cualquier causa durante el primer año, con áreas bajo la curva ROC que oscilan entre 0,59 y 0,80. Objetivo: Desarrollar y validar un modelo basado en algoritmos de redes neuronales (RN) destinado a mejorar el rendimiento de los modelos tradicionales para predecir mortalidad a corto y mediano plazo de pacientes con IC aguda. Material y métodos: Se analizó una base de datos con 181 variables de 483 pacientes con IC aguda en un hospital de comunidad de la Ciudad de Buenos Aires (junio de 2005-junio de 2019). Se utilizaron 25 variables para calcular 5 modelos de riesgo validados para predecir la mortalidad a 30 días, 6 meses y un año: EFFECT, ADHERE, GWTG-HF, 3C-HF y ACUTE-HF. Resultados: La edad media fue 78 ± 11,1años, 58% eran varones, el 35% de las IC eran de etiología isquémico necrótica, y la fracción de eyección media fue 52% (35-60). En término de discriminación a 30 días, fueron mejores el score EFFECT (ROC: 0,68) y el 3C-HF (ROC: 0,67) que el ACUTE- HF (ROC: 0,54). A los 6 meses y al año, el score EFFECT (ROC: 0,69 y 0,69) superó al ADHERE (ROC: 0,53 y 0,56) (p=0,011 y p = 0,003, respectivamente), y los scores EFFECT GWRG-HF (ROC: 0,68 y 0,66) y 3C-HF (ROC: 0,67 y 0,67) superaron al score ACUTE-HF (ROC: 0,53 y 0,56). De los algoritmos de RN los mejores resultados se obtuvieron con un perceptrón multicapa (PMC) con dos capas ocultas. Se usó una RN de arquitectura de capas 24-9-7-2 con los siguientes resultados: ROC: 0,82, valor predictivo negativo (VPN) 93,2% y valor predictivo positivo (VPP) 66,7% para mortalidad a 30 días; ROC: 0,87, VPN: 89,1% y VPP: 78,6% para mortalidad a 6 meses; y ROC: 0,85, VPN: 85,6% y VPP: 78,9% para mortalidad al año. En términos de discriminación, los algoritmos de RN superaron a los scores tradicionales ( p <0,001). Los factores que obtuvieron ≥50% de importancia estandarizada para predecir la mortalidad a los 30 días fueron en orden descendente la creatinina sérica, la hemoglobina, la frecuencia respiratoria, la urea, el sodio, la edad y la presión arterial sistólica. Agregaron capacidad pronóstica la clase III-IV NYHA y la demencia para mortalidad a 6 meses, y la frecuencia cardíaca y la disfunción renal crónica para mortalidad al año. Conclusiones: Los modelos con algoritmos de RN fueron significativamente superiores a los scores de riesgo tradicionales en nuestros pacientes con IC. Estos hallazgos constituyen una hipótesis de trabajo a validar con una mayor muestra de casos y en forma multicéntrica.

ABSTRACT Background: Heart failure (HF) risk scores to assess all-cause mortality during the first year have areas under the ROC curve (AUC) ranging between 0.59 and 0.80 Objective: To develop and validate a neural network (NN) algorithm-based model to improve traditional scores' performance for predicting short- and mid-term mortality of patients with acute HF. Methods: A prospective clinical database was analyzed including 483 patients admitted with diagnosis of acute HF in a coronary care unit community hospital of Buenos Aires, between June 2005 and June 2019. Among 181 demographic, laboratory, treatment and follow-up variables, only 25 were selected to calculate five acute heart failure risk scores aimed to predict 30-day, 6-month and 1-year mortality: EFFECT, ADHERE, GWTG-HF, 3C-HF, and ACUTE-HF. Results: Mean age was 78 ± 11.1 years, 58% were men, 35% had ischemic necrotic HF and median left ventricular ejection fraction was 52% (35-60). At 30 days, the EFFECT score (AUC:0.68) and the 3C-HF score (AUC: 0.68) showed better performance than the ACUTE-HF score (AUC: 0.54). At 6-month and 1-year follow-up, the EFFECT score (ROC: 0.69 and 0.69) outperformed the ADHERE score (AUC: 0.53 and 0.56), and EFFECT (AUC: 0.69 and 0.69), GWRG-HF (AUC = 0.68 and 0.66), and 3C-HF (AUC:0.67 and 0.67) scores outperformed the ACUTE-HF score (AUC:0.53 and 0.56). The best results with NN algorithms were obtained with a two-hidden layer multilayer perceptron. A 24-9-7-2-layer architecture NN was used with the following results: AUC: 0.82, negative predictive value (NPV) 93.2% and positive predictive value (PPV) 66.7% for 30-day mortality; AUC: 0.87, NPV: 89.1% and PPV: 78,6% for 6-month mortality; and AUC: 0.85, NPV: 85.6% and PPV: 78.9% for 1-year mortality. In terms of discrimination, NN algorithms outperformed all the traditional scores (p <0.001). For this algorithm, the most influential factors in descending order that scored ≥50% normalized importance to predict 30-day mortality were serum creatinine, hemoglobin, respiratory rate, blood urea nitrogen, serum sodium, age and systolic blood pressure. Also, NYHA functional class III-IV and dementia added prognostic capacity to 6-month mortality, and heart rate and chronic kidney disease to 1-year mortality. Conclusions: The models with NN algorithms were significantly superior to traditional risk scores in our population of patients with HF. These findings constitute a working hypothesis to be validated with a larger and multicenter sample of cases.

Continuous Monitoring of Coronary Care Mortality Using the Global Registry for Acute Coronary Events (GRACE) Score.

BACKGROUND: The aim was to assess the utility of cumulative sum techniques for continuous monitoring of coronary care outcomes, applied to patients with acute coronary syndrome stratified by the Global Registry for Acute Coronary Events (GRACE) risk score. METHODS: A prospective longitudinal study to assess GRACE score for real-time monitoring of coronary care mortality in a community hospital was conducted between January 2012 and December 2017. An expected-to-observed probability of death chart for individual risk and a variable life-adjusted display were used to monitor the results. RESULTS: A total of 1,255 patients undergoing acute coronary syndrome were included in the analysis. GRACE-based variable life-adjusted plots monitoring in-hospital mortality showed that observed death rates remained in general within the expected 95% confidence limit over time, and these behaviors were similar for ST-segment elevation and non-ST-segment elevation myocardial infarction. In-hospital all-cause mortality was 2.6% for the overall population, and 56% of these cases corresponded to unexpected deaths; conversely, unexpected survival occurred in 5.2% of survivors. CONCLUSIONS: Continuous monitoring of coronary care mortality based on cumulative sum charts and the GRACE score demonstrated the occurrence of series of favorable and unfavorable outcomes on a real-time basis. Additionally, plotting the expected-to-observed probability of death for individual cases was useful to individualize unexpected deaths in low-risk patients. Although overall coronary care performance was adequate according to the GRACE score, we found that there is still some room for improvement, since over half of the deaths occurring in low-risk patients were potentially preventable.

Supervivencia de la cirugía coronaria de múltiples vasos en el seguimiento hasta 20 años / Long-term survival of multi-vessel coronary surgery bypass grafting up to 20 years of follow-up

RESUMEN Introducción: Los resultados del seguimiento a largo plazo constituyen uno de los mejores parámetros para evaluar la calidad de una intervención médica. Objetivos: Analizar la supervivencia global y la supervivencia libre de eventos cardiovasculares a 20 años de la cirugía coronaria en un hospital de comunidad con historia clínica electrónica de larga data. Métodos: Se estudiaron en forma retrospectiva los resultados alejados de las cirugías coronarias aisladas efectuadas en pacientes con enfermedad de múltiples vasos o tronco de coronaria izquierda entre 1999 y 2003 en un hospital de comunidad. El seguimiento hasta 20 años se realizó a través de la historia clínica electrónica. Resultados: Se logró un tiempo de seguimiento medio de 125 meses (rango: 6-268) en 254 pacientes de los 272 operados en ese período (93,4%). El número promedio de puentes fue 3,3 (desvío estándar: 0,97); en 97,6% se usó al menos una arteria mamaria y 59,4% recibieron un puente con arteria radial. Se obtuvo un seguimiento de 2646 pacientes-años, con un riesgo anual de muerte por toda causa de 2,5%. La supervivencia global al seguimiento medio fue de 0,806 (error estándar, EE: 0,03), y la supervivencia libre de eventos cardiovasculares fue de 0,826 (EE: 0,03). Conclusiones: El registro electrónico completo de los afiliados a un hospital de comunidad operados hace más de 15 años permitió analizar la supervivencia global y libre de eventos a largo plazo. Estos resultados servirán como estándar al momento de elegir entre la cirugía y la angioplastia de múltiples vasos.

ABSTRACT Background: The outcomes of long-term follow-up constitute one of the best parameters to assess the quality of a medical intervention. Objectives: To analyze the overall and the free-of-cardiovascular events 20-year survival after coronary surgery in a community hospital with a long-standing electronic medical records. Methods: The results of coronary surgeries in patients with multi-vessel or trunk disease operated between 1999 and 2003 in a community hospital were retrospectively studied. The follow-up up to 20 years was carried out through the electronic medical record. Results: Of 272 patients operated, a mean follow-up of 125 months (range 6-268) was achieved in 254 (93.4%). The average number of bypasses was 3.3 (standard deviation 0.97); in 97.6%, at least one internal mammary artery was used and 59.4% received a radial artery graft. A follow-up of 2646 patient-years was obtained with an annual risk of death from all causes of 2.5%. The overall survival at the mean follow-up time was 0.806 (standard error (SE) 0.03), and the cardiovascular event-free survival 0.826 (SE 0.03). Conclusions: The complete electronic registration of affiliates to a community hospital operated more than 15 years ago allowed us to analyze the overall survival and the freedom of long-term events. These results will serve as a standard when choosing between surgery and multi-vessel angioplasty.

Comparación de dos registros argentinos de infarto de miocardio: SCAR 2011 y ARGEN-IAM ST 2015 / Comparison of Two Argentine Registries on Myocardial Infarction: the 2011 SCAR Registry and the 2015 ARGEN-IAM-ST Registry

RESUMEN Objetivos: Comparar pacientes con infarto agudo de miocardio con elevación del segmento ST (IAMCEST) incluidos en centros participantes de dos registros argentinos. Material y métodos: Se compararon pacientes con IAMCEST incluidos en 54 centros que participaron tanto en el registro SCAR (2011) como en el ARGEN-IAM-ST (2015). Resultados: Se analizaron 676 pacientes con IAMCEST; 222 del SCAR y 454 del ARGEN-IAM-ST No hubo cambios significativos en la edad y el género. Se observó una reducción significativa en el uso de fibrinolíticos, con un incremento de la angioplastia primaria. El shock cardiogénico se redujo a la mitad. No hubo diferencias en la mortalidad y de reinfarto durante la hospitalización. Conclusiones: Se observó una mayor indicación de angioplastia primaria y una disminución en el uso de fibrinolíticos. El shock cardiogénico se redujo significativamente en los últimos 5 años, sin cambios significativos en la mortalidad hospitalaria.

ABSTRACT Objectives: The aim of this study was to compare patientis with ST-segment elevation myocardial infarction (STEMI) included in centers participating of two registries in Argentina. Methods: STEMI patientis included in the 54 centers participating in the SCAR (2011) registry and in the ARGEN-IAM-ST (2015) registry were compared. Resultis: A total of 676 STEMI patientis were analyzed: 222 in the SCAR registry and 454 in the ARGEN-IAM-ST registry. There were no significant differences in age and sex. The use of fibrinolytic agentis was significantly lower and the use of primary percutaneous coronary intervention was significantly increased. The incidence of cardiogenic shock was 50% lower. There were no differences in mortality and reinfarction during hospitalization. Conclusions: The indication of primary percutaneous coronary intervention increased and the use of fibrinolytic agentis decreased. Cardiogenic shock decreased significantly in the last 5 years without significant changes in in-hospital mortality.

Single spot albumin to creatinine ratio: A simple marker of long-term prognosis in non-ST segment elevation acute coronary syndromes.

BACKGROUND: Microalbuminuria is a known risk factor for cardiovascular morbidity and mortality suggesting that it should be a marker of endothelial dysfunction. Albumin to creatinine ratio (ACR) is an available and rapid test for microalbuminuria determination, with a high correlation with the 24-h urine collection method. There is no prospective study that evaluates the prognostic value of ACR in patients with non ST-segment elevation acute coronary syndromes (NSTE-ACS). The purpose of our study was to detect the long-term prognostic value of ACR in patients with NSTE-ACS. METHODS: Albumin to creatinine ratio was estimated in 700 patients with NSTE-ACS at admission. Median follow-up time was 18 months. The best cutoff point of ACR for death or acute myocardial infarction was 20 mg/g. Twenty-two percent of patients had elevated ACR. RESULTS: By multivariable Cox regression analysis, ACR was an independent predictor of the clinical endpoint: odds ratio 5.8 (95% confidence interval [CI] 2-16), log-rank 2 p < 0.0001 in a model including age > 65 years, female gender, diabetes mellitus, creatinine clearance, glucose levels at admission, elevated cardiac markers (troponin T/CK-MB) and ST segment depression. The addition of ACR significantly improved GRACE score C-statistics from 0.69 (95% CI 0.59-0.83) to 0.77 (95% CI 0.65-0.88), SE 0.04, 2 p = 0.03, with a good calibration with both models. CONCLUSIONS: Albumin to creatinine ratio is an independent and accessible predictor of long-term adverse outcomes in NSTE-ACS, providing additional value for risk stratification.

[NT-Pro-BNP levels performance before and after reversion to sinus rhythm in patients with preserved ventricular function]. / Descenso de los niveles de propéptido natriurético de tipo B-N terminal luego de la reversión de pacientes con fibrilación auricular y función ventricular conservada.

Our objective was to evaluate changes of N-terminal pro-BNP (NT-Pro-BNP) levels at baseline and after restoration to sinus rhythm in hemodynamic stable patients with lone atrial fibrillation (LAF) with preserved left ventricular function. NT-Pro-BNP levels were obtained before and after cardioversion in thirty hemodynamic stable patients with LAF and preserved left ventricular function. At baseline levels of NT-Pro BNP levels were significatively higher than a normal control group. NT-Pro-BNP levels decreased significantly following cardioversion from 529 (157-1763) to 318 (98-870) pg/ml, p < 0.0001. Decreasing of N-terminal pro-BNP concentrations was observed after any mode of cardioversion: electrical or pharmacologic, 345 (153-1151) pg/ml to 169 (86-407) pg/ml, p: 0.02 and from 1624 (541-4010) pg/ml to 856 (532-1160) pg/ml, p < 0.001, respectively. N-terminal pro-BNP decreasing was observed mainly in patients with length of LAF longer than 8 hours: 1289 (338-2103) to 410 (169-905) pg/ml, p < 0.001 but no difference was detected when such length was less than 8 hours: 274 (137-2300) to 286 (82-1440), p = NS. Our study showed that baseline levels of NT-pro-BNP decreased shortly after reversion of patients with LAF to sinus rhythm. This performance occurs predominantly in patients with LAF length of at least eight hours.

Descenso de los niveles de propéptido natriurético de tipo B-N terminal luego de la reversión de pacientes con fibrilación auricular y función ventricular conservada / NT-Pro-BNP levels performance before and after reversion to sinus rhythm in patients with preserved ventricular function

El objetivo fue comparar los niveles de Pro Péptido Natriurético tipo B-N terminal (Pro-BNP-NT) basales y post reversión en pacientes con fibrilación auricular solitaria (FAS) de comienzo reciente y con función ventricular izquierda conservada. Se determinaron niveles del Pro BNP NT antes y después de su reversión en treinta pacientes con FAS de comienzo reciente y fracción de eyección de ventrículo izquierdo ≥ 50%. Basalmente, los niveles de Pro BNP NT fueron significativamente más elevados respecto a un grupo control sano: 529 pg/ml (157-1763) versus 31.5 pg/ml (24-76), p < 0.0001. Las concentraciones de Pro BNP NT descendieron significativamente luego de la cardioversión de 529 (157-1763) a 318 (98-870) pg/ml, p < 0.001. Los descensos se objetivaron tanto luego de la cardioversión eléctrica como de la obtenida por vía farmacológica, 345 (153-1151) pg/ml a 169 (86-407) pg/ml, p: 0.02 y de 1624 (541-4010) pg/ml a 856 (532-1160) pg/ml, p < 0.001, respectivamente. Este fenómeno se observó fundamentalmente en aquellos con una FAS con duración mayor a 8 horas: 1289 (338-2103) a 410 (169-905) pg/ml, p < 0.001. No se observó correlación entre los cambios de la frecuencia cardíaca y del Pro BNP NT pre y post cardioversión a ritmo sinusal. Se observaron descensos significativos de Pro BNP NT basalmente y post reversión (tanto farmacológica como eléctrica) en pacientes con FA de reciente aparición y función ventricular conservada. Estos descensos fueron más ostensibles a partir de las 8 horas de duración de la arritmia.

Our objective was to evaluate changes of N-terminal pro-BNP (NT-Pro-BNP) levels at baseline and after restoration to sinus rhythm in hemodynamic stable patients with lone atrial fibrillation (LAF) with preserved left ventricular function. NT-Pro-BNP levels were obtained before and after cardioversion in thirty hemodynamic stable patients with LAF and preserved left ventricular function. At baseline levels of NT-Pro BNP levels were significatively higher than a normal control group. NTPro-BNP levels decreased significantly following cardioversion from 529 (157-1763) to 318 (98-870) pg/ml, p < 0.0001. Decreasing of N-terminal pro-BNP concentrations was observed after any mode of cardioversion: electrical or pharmacologic, 345 (153-1151) pg/ml to 169 (86-407) pg/ml, p: 0.02 and from 1624 (541-4010) pg/ml to 856 (532-1160) pg/ml, p < 0.001, respectively. N-terminal pro-BNP decreasing was observed mainly in patients with length of LAF longer than 8 hours: 1289 (338-2103) to 410 (169-905) pg/ml, p < 0.001 but no difference was detected when such length was less than 8 hours: 274 (137-2300) to 286 (82-1440), p = NS. Our study showed that baseline levels of NT-pro-BNP decreased shortly after reversion of patients with LAF to sinus rhythm. This performance occurs predominantly in patients with LAF length of at least eight hours.