Is It Possible to Simplify Risk Stratification Scores for Patients with ST-Segm...

• Autor(es): Timóteo, AT ; Papoila, AL ; Lopes, JP ; Oliveira, JA ; Ferreira, ML ; Cruz Ferreira, R
• Data: 2013
• Identificador Persistente: http://hdl.handle.net/10400.17/1778
• Origem: Repositório do Centro Hospitalar de Lisboa Central, EPE
• Assunto(s): HSM CAR; Síndrome Coronário Agudo; Prognóstico; Risco
• Tipo de Documento: Artigo
• Idioma: Inglês

Descrição

INTRODUCTION: There are several risk scores for stratification of patients with ST-segment elevation myocardial infarction (STEMI), the most widely used of which are the TIMI and GRACE scores. However, these are complex and require several variables. The aim of this study was to obtain a reduced model with fewer variables and similar predictive and discriminative ability. METHODS: We studied 607 patients (age 62 years, SD=13; 76% male) who were admitted with STEMI and underwent successful primary angioplasty. Our endpoints were all-cause in-hospital and 30-day mortality. Considering all variables from the TIMI and GRACE risk scores, multivariate logistic regression models were fitted to the data to identify the variables that best predicted death. RESULTS: Compared to the TIMI score, the GRACE score had better predictive and discriminative performance for in-hospital mortality, with similar results for 30-day mortality. After data modeling, the variables with highest predictive ability were age, serum creatinine, heart failure and the occurrence of cardiac arrest. The new predictive model was compared with the GRACE risk score, after internal validation using 10-fold cross validation. A similar discriminative performance was obtained and some improvement was achieved in estimates of probabilities of death (increased for patients who died and decreased for those who did not). CONCLUSION: It is possible to simplify risk stratification scores for STEMI and primary angioplasty using only four variables (age, serum creatinine, heart failure and cardiac arrest). This simplified model maintained a good predictive and discriminative performance for short-term mortality.