Optimum Baseline Clinical Severity Scale Cut Points for Prognosticating Intracerebral Hemorrhage: INTERACT Studies

BACKGROUND: The optimal cut point of baseline National Institutes of Health Stroke Scale (NIHSS) and Glasgow Coma Scale scores for prognosticating acute intracerebral hemorrhage (ICH) is unknown. METHODS: Secondary analyses of participant data are from the INTERACT (Intensive Blood Pressure Reduction in Acute Intracerebral Hemorrhage Trials) 1 and 2 studies. Receiver operating characteristic analyses were used to compare the predictive performance of baseline NIHSS and Glasgow Coma Scale scores, ICH score, and max-ICH score. Optimal cut points for predicting 90-day clinical outcomes (death or major disability [defined as modified Rankin Scale scores 3–6], major disability [defined as modified Rankin Scale scores 3–5], and death alone) were determined using the Youden index. Logistic regression models were adjusted for age, sex, hematoma volume, and other known risk factors for poor prognosis. We validated our findings in the INTERACT1 database. RESULTS: There were 2829 INTERACT2 patients (age, 63.5±12.9 years; male, 62.9%; ICH volume, 10.96 [5.77–19.49] mL) included in the main analyses. The baseline NIHSS score (area under the curve, 0.796) had better prognostic utility for predicting death or major disability than the Glasgow Coma Scale score (area under the curve, 0.650) and ICH score (area under the curve, 0.674) and was comparable to max-ICH score (area under the curve, 0.789). Similar findings were observed when assessing the outcome of major disability. A cut point of 10 on baseline NIHSS optimally (sensitivity, 77.5%; specificity, 69.2%) predicted death or major disability (adjusted odds ratio, 4.50 [95% CI, 3.60–5.63]). The baseline NIHSS cut points that optimally predicted major disability and death alone were 10 and 12, respectively. The predictive effect of NIHSS≥10 for poor functional outcomes was consistent in all subgroups including age and baseline hematoma volume. Results were consistent when analyzed in the independent INTERACT1 validation database. CONCLUSIONS: In patients with mild-to-moderate ICH, a baseline NIHSS score of ≥10 was optimal for predicting poor outcomes at 90 days. Prediction based on baseline NIHSS is better than baseline Glasgow Coma Scale score. REGISTRATION: URL: https://www.clinicaltrials.gov ; Unique identifiers: NCT00226096 and NCT00716079.