A practical nomogram incorporating inflammatory markers for predicting short-term prognosis in conservatively managed...

We investigated whether systemic inflammatory indices improve prediction of short-term outcomes in conservatively treated spontaneous basal ganglia hemorrhage (BGH).

We retrospectively analyzed 137 patients (2020-2023) and classified 90-day outcomes as good (mRS 0-2, n = 73) or poor (mRS 3-5, n = 64). A Basic model (age, hematoma volume, NIHSS, IVH) and an Inflammation model (Basic + CAR + NLR) were developed. Discrimination (ROC/DeLong) and calibration (Hosmer-Lemeshow) were assessed; Patients with mRS = 6 were excluded from the primary analysis but included in a prespecified sensitivity analysis. A nomogram was internally validated (1,000 bootstrap) with Brier score and decision curve analysis (DCA).

The Inflammation model outperformed the Basic model (AUC 0.823 [95%CI 0.754-0.892] vs 0.658 [0.568-0.749]; DeLong Z = 3.375, P = 0.001) with acceptable calibration (HL P = 0.896 vs 0.451). In sensitivity analyses including deaths, performance remained superior (AUC 0.815 [0.747-0.882] vs 0.656 [0.568-0.745]; Z = 3.238, P = 0.001). The nomogram showed good internal validity (C-index 0.794; Brier 0.1899) and net benefit across thresholds 0.05-0.95.

Adding CAR and NLR to a clinical model significantly improves short-term outcome prediction after conservatively treated spontaneous BGH and supports individualized risk stratification.