Predicting Discharge Outcomes from In-Hospital Characteristics after Cerebral Arterial Aneurysm Rupture: A Single Institutional Experience from Ukraine

Abstract

Most existing predictive models developed to assess outcomes after aneurysmal subarachnoid hemorrhage (aSAH) use a dichotomized approach. However, this strategy does not reflect the full clinical spectrum of patient status after aSAH and leads to underestimation of differences in rehabilitation needs and associated economic costs. Based on the Hospital Assessment Scale (HAS), we developed and validated an ordinal predictive model to predict in-hospital outcomes after aSAH. We conducted a single-center retrospective cohort study in Ukraine including 489 patients with aSAH. The outcome was categorized into four HAS scores. Candidate predictors included demographic and medical history variables, clinical and radiological characteristics, including cerebral arterial aneurysm (CAA) features, treatment strategy, and complications. Missing data were handled using multiple imputation. An ordinal logistic regression model with proportional odds assumption and penalized maximum likelihood estimation was applied. Internal validation was performed using bootstrap resampling (B = 200). Independent predictors of worse HAS outcome included aneurysm re-rupture, larger aneurysm size, limb paresis at admission, conservative treatment, cerebral vasospasm, and hospital-acquired pneumonia. The final ordinal predictive model was constructed using 9 variables: sex, age groups, presence of CAA re-rupture and limb paresis at admission, size of the CAA, mWFNS grade at admission, treatment strategy, and the occurrence of cerebral vasospasm and hospital-acquired pneumonia. The model demonstrated good discrimination and calibration after internal bootstrap validation (C-index≈0.82). We developed and internally validated an ordinal predictive model to estimate HAS outcomes after aSAH. The proposed model integrates baseline clinical characteristics and dynamic changes occurring throughout hospitalization.