Surgical timing and mortality in acute traumatic subdural hematoma: is the four-hour threshold confounded by injury...

Early surgical evacuation within four hours has long been advocated for acute traumatic subdural hematoma (ASDH), yet the supporting evidence is largely observational and unadjusted for injury severity. We examined whether surgical timing is independently associated with in-hospital mortality after accounting for baseline severity.

We retrospectively analyzed 91 consecutive patients who underwent surgical evacuation for ASDH between January 2014 and December 2024. The primary outcome was in-hospital mortality; the secondary outcome was functional status at discharge measured by the Glasgow Outcome Scale (GOS). Surgical timing (≤ 4 vs. >4 h) was assessed through unadjusted comparison, stratification by Glasgow Coma Scale (GCS) score, multivariable Firth penalized logistic regression with 1000-iteration bootstrap internal validation, and sensitivity analyses including alternative timing thresholds (1-12 h), restricted cubic spline modeling, and timing-by-severity interaction testing.

In-hospital mortality was 33.0% (30/91; 95% Wilson CI 24.2-43.1%). Crude analysis showed higher mortality with early surgery (50.0% vs. 15.6%, p = 0.001), but these patients had lower admission GCS and higher rates of CT-defined traumatic axonal injury (TAI) and cerebral contusion. After stratification, no statistically detectable mortality difference was observed within the severe traumatic brain injury (TBI) subgroup (GCS ≤ 8: 69.6% vs. 66.7%, p = 1.000), although the late-surgery stratum contained only six patients. On multivariable Firth penalized analysis adjusting for the four-hour threshold, admission GCS, traumatic axonal injury (TAI), and age, the four-hour threshold was not independently associated with mortality (adjusted OR 1.46, 95% CI 0.41-5.04); admission GCS (OR 0.75 per point, 95% CI 0.63-0.87) and TAI (OR 7.56, 95% CI 1.81-38.04) were the only independent predictors. The model demonstrated strong discrimination (apparent AUC 0.920; bootstrap 95% CI 0.856-0.974; optimism-corrected AUC 0.902). Calibration was also satisfactory (Brier score 0.112). Threshold sensitivity analyses (1, 2, 3, 6, 8, and 12 h) yielded no threshold with an independent association after severity adjustment. Good functional outcome (GOS 4-5) was achieved by 56 patients (61.5%); the parallel timing distribution (early 41.3% vs. late 82.2%) reflected the same severity-driven pattern observed for mortality.

In surgically treated ASDH, no statistically significant independent association between the four-hour threshold and in-hospital mortality was identified after adjustment for available severity markers. These findings should not be interpreted as support for delaying surgery in clinically indicated ASDH: ASDH with progressive neurological deterioration and significant intracranial mass effect remains a true neurosurgical emergency requiring prompt surgical evacuation. The four-hour benchmark should be applied in conjunction with severity-stratified prognostication rather than as a standalone criterion, with admission GCS and TAI as the principal determinants of outcome.