EVD-Associated Infections in Subarachnoid Hemorrhage: Risk Factors and Clinical Predictions-A Retrospective...

Background and Objectives: External ventricular drain (EVD)-associated infections are a serious complication in subarachnoid hemorrhage (SAH) patients, with reported incidence rates of 1-45%. Existing prediction models show limited performance and focus on the static risk factors assessed at insertion, failing to examine how infection risk changes over time. We sought to identify the independent predictors of EVD infections in SAH patients and develop a practical clinical prediction model. Materials and Methods: We retrospectively analyzed 198 SAH patients with EVDs treated at our center between January 2022 and April 2025, collecting 4757 laboratory observations throughout their hospital stay. Univariate and multivariate logistic regression analyses were performed to identify the independent risk factors and develop a clinical prediction model. Results: Of 198 patients undergoing EVD insertion for SAH, 49 developed associated infections (24.7%). Univariate analysis identified several significant risk factors, including EVD duration (Cohen's d = 1.00, p < 0.001), EVD revisions (d = 1.11, p < 0.001), Hunt & Hess grade ≥ 4 (p = 0.011), and peak laboratory values, including CSF lactate (d = 0.53, AUC = 0.79), CSF protein (d = 0.52, AUC = 0.74), CSF glucose (d = 0.63, AUC = 0.73), and procalcitonin (d = 0.58, AUC = 0.75). However, multivariate analysis revealed that only EVD duration retained statistical significance (adjusted OR = 3.50 per continuous day; 95% CI: 2.11-5.78; p < 0.000001); note that continuous daily scale modeling implies exponential risk escalation (3.5-fold increase per single day). For clinical interpretation, categorical duration analysis provides more conservative estimates: 8-14 days versus ≤7-day reference OR = 1.92 (p = 0.013), and >14 days versus ≤7-day reference OR = 3.70 (p < 0.001). All other variables lost statistical independence after mutual adjustment. Infection rates demonstrated a dose-response relationship with EVD duration: 11.1% for ≤7 days, 19.3% for 8-14 days, and 31.6% for >14 days. The final prediction model achieved good discrimination (AUC = 0.737). Conclusions: EVD duration emerged as the dominant predictor of infection risk in SAH patients, which is a traditional factor. These findings support clinical protocols that prioritize minimizing drain duration whenever medically appropriate, shifting focus from complex risk scoring to time-based management strategies.