Background: To develop a risk model (ICH Progression Score) to predict in-hospital deterioration after ICH using routinely collected variables at presentation.
Methods: The ICH Progression Score was developed based on the China National Stroke Registry, in which eligible patients were randomly divided into derivation (60%) and validation (40%) cohorts. Variables routinely collected at presentation were used for predicting in-hospital deterioration after ICH. For testing the added value of hematoma volume measure, we separately developed two models with (ICH Progression Score-A) and without (ICH Progression Score-B) hematoma volume included. In-hospital deterioration after ICH was defined as the episode in which a patient experienced a persistent increase in NIHSS score of 4 or more or death during hospitalization. Multivariable logistic regression was performed to identify independent predictors. The area under the receiver operating characteristic curve (AUROC) and Hosmer-Lemeshow goodness-of-fit test was used to assess model discrimination and calibration, respectively.
Results: In-hospital deterioration was 13.2% and 11.6% in the derivation (n=2870) and validation (n=1921) cohorts, respectively. A 23-point ICH Progression Score-A was developed based on a set of predictors and showed good discrimination in the derivation (AUROC=0.82; 95%CI=0.80-0.84) and validation (AUROC=0.79; 95%CI=0.76-0.82) cohorts. The ICH Progression Score-A was well calibrated (Hosmer-Lemeshow test) in the derivation (P=0.62) and validation (P=0.36) cohorts. Similarly, a 25-point ICH Progression Score-B was established. The ICH Progression Score-A and -B were not significantly different in discrimination for in-hospital deterioration after ICH.
Conclusion: the ICH Progression Score are valid clinical grading scales for predicting in-hospital deterioration after ICH at presentation.