Validation of the Premise Score after Addition of Recurrent Stroke Variable to Predicting Early Mortality in Acute Ischemic Stroke Patients

Abstract

BACKGROUND: Predicting Early Mortality of Acute Ischemic Stroke (PREMISE) score is a recent scoring derived from the Austrian Stroke Unit Registry. Internal validation assigned by Haji Adam Malik General Hospital Medan in 2020 stated that the performance of its discrimination was not strong enough and leads to a further necessity for an improvement by an addition of another variable. AIM: This study aims to assess the calibration and discrimination performance of the PREMISE score after addition of recurrent stroke variable to predicting early mortality in acute ischemic stroke patients. METHODS: Retrospective cohort study was performed in a population of ischemic stroke patients admitted to Haji Adam Malik General Hospital Medan from January to December 2019. The score’s calibration was calculated by the Hosmer–Lemeshow goodness-of-fit test and its discriminatory power by calculating the area under curve (AUC) after adding the recurrent stroke variable. RESULTS: In total, 223 ischemic stroke patients were included in the analysis, there were 69 patients with recurrent stroke (30.9%). The PREMISE score showed good calibration (Hosmer–Lemeshow p = 0.331) and a strong enough discrimination power with an AUC value of 0.783. The addition of the recurrent stroke variable to the PREMISE score showed an increase score’s calibration with Hosmer–Lemeshow p = 0.575 and discrimination power with an AUC value of 0.806. CONCLUSION: The PREMISE score after the addition of the recurrent stroke variable showed an increase calibration and discrimination performance to predicting early mortality in acute ischemic stroke patients.