Incorporating conditional survival into prognostication for gunshot wounds to the head

Area of Science:

• Neurosurgery
• Trauma Surgery
• Prognostic Modeling

Background:

• Prognosis for gunshot wounds to the head (GSWH) is typically assessed on hospital admission.
• The prognostic accuracy may change for patients surviving the acute phase of hospitalization.

Purpose of the Study:

• To evaluate the accuracy of the Baylor score in predicting outcomes for GSWH patients who have survived the initial 48 hours.
• To generate conditional outcome curves for GSWH patients throughout their hospital stay.

Main Methods:

• A retrospective review of GSWH patients with dural penetration from January 2009 to June 2019.
• Calculation of the Baylor score and generation of conditional survival and functional outcome curves.
• Assessment of Baylor score accuracy using receiver operating characteristic curves and area under the curve (AUC) among acute-phase survivors.

Main Results:

• 129 of 297 patients survived the acute phase (>48 hours), showing lower mortality (32.6%) and higher good functional outcome rates (48.1%).
• The Baylor score demonstrated good accuracy in predicting mortality (AUC = 0.807) and functional outcome (AUC = 0.837) in acute-phase survivors.
• The score tended to overestimate mortality and underestimate good functional outcomes; hospital day 18 was an inflection point for decreasing good functional outcome probability.

Conclusions:

• Survival beyond 48 hours significantly improves survival and functional outcomes for GSWH patients.
• The Baylor score remains reasonably accurate for acute-phase survivors but requires calibration for over/underestimation.
• Future prognostic models should integrate conditional survival data for enhanced accuracy post-acute phase.