Ultrasonographic Pupil Assessment in Patients With Severe Traumatic Brain Injury
Qian Shi, Rujing Bai, Yan Xu, Bo Feng, Ling Yuan, Niannian Liu, Danjiang DongBackground
In patients with severe traumatic brain injury, accurate pupillary assessment is critical for neurological monitoring but is often hindered by periorbital edema, trauma, or dressings that preclude direct visualization.
Objective
To validate the feasibility and accuracy of ultrasonographic pupil assessment performed through the closed eyelid as an alternative to direct pupillometry in patients with traumatic brain injury.
Methods
This prospective study enrolled 45 intensive care unit patients (36 male; mean [SD] age, 67.1 [12.9] years). Both direct pupillometry and ultrasonographic pupil assessment were sequentially completed within a 3-minute interval under standardized low-light conditions for both eyes. Direct pupillometry involved video recording of the exposed pupil with a calibrated reference scale. Ultrasonographic pupil assessment was performed by using a high-frequency linear probe placed on the closed eyelid. The primary outcome was the agreement between methods for calculating the percentage of pupillary constriction. Secondary outcomes included agreement on resting and minimum pupil diameters.
Results
Ultrasonographic pupil assessment showed excellent agreement with direct pupillometry for percent pupillary constriction (median [IQR] difference, 0.44% [−2.78% to 2.83%]; Wilcoxon P ≥ .48). Bland-Altman analysis showed narrow limits of agreement. Strong correlations were also noted for resting and minimum pupil diameters (0.6 ≤ r2 ≤ 0.873).
Conclusion
Ultrasonographic pupil assessment is a reliable and accurate method for objective pupillary evaluation when direct visualization is obstructed. By enabling quantitative measurement of the pupillary light reflex through the closed eyelid, ultrasonographic pupil assessment addresses an important monitoring gap in the management of patients with severe traumatic brain injury and periorbital swelling.