Proposal of a prediction score for hematoma expansion after intracerebral hemorrhage

To propose and validate a prediction score for intracerebral hemorrhage (ICH) patients at risk of hematoma expansion (HE).

A retrospective observational study was designed to propose and validate the score.

Sanxiang Road branch and Xuguan branch belonging to the Second Affiliated Hospital of Soochow University (China).

A total of 317 ICH patients in Sanxiang Road branch were registered as the development cohort, and 109 ICH patients in Xuguan branch were enrolled as the validation cohort.

Independent risk factors for HE were identified using multiple logistic regression analysis. A prediction score was then proposed based on β coefficients and preliminarily verified in the validation cohort.

All clinical data of the patients were compiled from the electronic medical records. Hematoma expansion was defined as an increase in hematoma volume >33% or absolute hematoma growth >6ml from the initial scan. Specific non-contrast CT(NCCT) signs were identified by two observers independently.

Our score demonstrated satisfactory discrimination ability for HE (area under the ROC curve 0.854 in the development cohort versus 0.893 in the validation cohort). Appropriate calibration was found in the development cohort, whereas calibration in the validation cohort was slightly lower but still within the accuracy range (maximum deviation, average deviation and P were 0.070, 0.028, 0.773, respectively, versus 0.114, 0.056, 0.156). Decision curve analysis of the score from two samples were both far from the curve of treat all and curve of treat none, which verified its security and reliability. Patients with a total score ≥4.5 were at greatest risk of HE.

The score may provide some reference and help in accurately identifying individuals at high risk of HE, allowing rapid guidance of clinical management and also serving as an aid in clinical trials.