Avaliação do índice de choque como preditor de mortalidade e desfechos clínicos em pacientes pós-parada cardiorrespiratória: uma coorte retrospectiva

Abstract

Introduction: the Cardiac Arrest (CA) is a critical event associated with high mortality and risk of neurological impairment. The use of prognostic scores such as the Shock Index (SI), the Charlson Comorbidity Index (CCI), and NEWS2 can assist in the early risk stratification, although their clinical applicability still requires validation. Objective: to evaluate the accuracy of the shock index as a predictive tool for clinical outcomes in patients post-cardiac arrest hospitalized in hospitals in the state of Sergipe. Materials and methodology: an observational, retrospective cohort, and analytical study that included 135 adult patients who suffered cardiac arrest in non-critical care units. Demographic, clinical, and hospital evolution variables were collected. The association between SI ≥ 0.9 and clinical conditions was evaluated using logistic regression. For mortality prediction, calibration analyses (calibration curves), accuracy (ROC curves), Cox regression, and Kaplan-Meier curves were applied. Neurological outcomes were measured using the CPC score (Cerebral Performance Category). The variables were analyzed using descriptive and inferential statistics, employing ROC curves, logistic regression, survival analysis (Kaplan-Meier and Cox regression), and association tests (chi-square, Fisher, and Mann-Whitney). The significance level adopted was 5%. The study was approved by the Ethics Committee of the Federal University of Sergipe (CAAE: 60839422.6.1001.0217; opinion no. 5.794.971). Results: the majority of patients were female (57.8%), mixed race (65.4%), and admitted for clinical conditions (54.5%). The main identified causes of cardiac arrest were hypoxia (15.6%) and acidosis (14.9%), although 60.7% of cases had no defined etiology. The NEWS2 score had the largest area under the curve (0.64), followed by the ICC (0.60) and SI (0.49), but there was no statistical difference between them. Acidosis was independently associated with SI ≥ 0.9 (OR = 2.83; p = 0.031), while the absence of a clinical cause was a protective factor (OR = 0.47; p = 0.049). The survival analysis showed a median of 12 days (SI ≥ 0.9) and 14 days (SI < 0.9), with no statistical significance (p = 0.873). Conclusion: the assessed scores showed low accuracy in the isolated prediction of in-hospital mortality. The SI demonstrated an association with acidosis and the absence of reported etiology, but did not correlate significantly with survival time. These findings reinforce the need for multivariate approaches for risk stratification in post-CPR patients.