To determine if automated pupillometry (AP) changes occur prior to clinical change and quantify outcomes in patients with large hemispheric infarctions (LHI).

LHI patients often develop cerebral edema and may have limited neurologic exams. AP may be utilized to monitor for worsening cerebral edema and may lead to improved outcomes.

Retrospective study of patients admitted to the neurosciences intensive care unit with the diagnosis of ischemic stroke from 1/2012-12/2018. Inclusion criteria included imaging with infarct size ≥2/3 of the middle cerebral artery territory, use of AP every four hours, and hourly Glasgow Coma Scale (GCS). Neurological pupil index (NPi) changes were recorded, including NPi<3 and a difference of NPi>0.7 between pupils. Clinical change was defined as GCS decrease of ≥2. Baseline demographics, need for mechanical ventilation, hyperosmolar therapy, decompressive hemicraniectomy, and mortality data were collected. 

Of the patients screened, 380 patients met initial imaging criteria, and 52 had AP readings. Mean age was 60; 61.5% were female. Mean admission NIHSS was 22; GCS was 9. There were 39 patients (75%) who had a GCS change; 30 patients (57.7%) had NPi<3 with a mean 12.4 hours prior to GCS change, and 36 patients (69.2%) had a difference of NPi>0.7 between pupils with a mean 17.2 hours prior to GCS change. The mean post stroke days (PSD) of NPi<3 was 1.6, NPi>0.7 was 1.3, and GCS change was 1.8. There were 40 patients (76.9%) that required hyperosmolar therapy; 21 patients (40.4%) underwent decompressive hemicraniectomy with a mean PSD 2. Mortality was 34.6%. 

NPi changes occurred several hours prior to GCS changes in patients with LHI. AP may be considered to monitor patients for escalation in management of cerebral edema. Further research is required to determine if earlier notification of worsening cerebral edema can improve outcomes.