To evaluate the performance and specificity of the Neurological Pupil Index (NPi ≤2) for predicting neurologic outcomes after cardiac arrest.

Quantitative pupillometry, specifically a Neurological Pupil index (NPi) of ≤2, is highly specific for predicting poor functional outcomes after resuscitation from cardiac arrest (CA). We aimed to evaluate the performance of this threshold within a U.S. post-CA population. 

We conducted an observational cohort study including patients resuscitated from both in-hospital and out-of-hospital CA (September 2022 - February 2024). We excluded patients with arrest due to neurological/traumatic causes, those following commands within six hours of resuscitation, and those without pupillometry measurements within 72 hours. Trained nurses performed pupillometry as clinical care; the lowest NPi was recorded. We abstracted the lowest NPi within 0–24, 24–48, and 48–72 hours after arrest. Primary outcomes were death and poor modified Rankin Scale (mRS) (≥3) at discharge. We calculated sensitivity, specificity, and false positive rates (with 95% confidence intervals) for NPi ≤2 for death and poor mRS at each time epoch.

Of 468 patients treated, 104 subjects were included. Subjects were a median of 62 years old [Inter Quartile Range (IQR) 47-71]; 41% were female; 26% had a shockable initial rhythm; and median CPR duration was 20 minutes [IQR 12-31]. Eighty-nine (86%) subjects died, and of the 15 survivors, 9 (60%) had poor mRS. In the 0-24hr epoch (84 subjects), there were 3 false-positives when predicting death and 1 when predicting poor mRS. All false positives had NPi ≤2 within 8 hours of arrest and followed commands during their hospital stay. Specificity for death was 0.70 (95% CI 0.35 – 0.93), and 0.75 (95% CI 0.19 – 0.99) for poor mRS. NPi remained perfectly specific at other time points.

An NPi threshold of ≤2 was not perfectly specific early after resuscitation from cardiac arrest in our cohort.