Abstract 12008: Quantitative Pupillometry for Neuroprognostication in Post-Cardiac Arrest: A Prospective Validation Study

Abstract

Background: Quantitative pupillometry is a new tool in multimodal neuroprognostication of comatose patients after out-of-hospital cardiac arrest, but the reliability and precision of quantitative pupillometry is not fully understood. Methods: We conducted a prospective blinded validation study with repeated patient assessment at our cardiac intensive care unit. To investigate intra- and interobserver variability of pupil size (mm), each assessment was performed on the same patient by two staff nurses within 2 minutes. The sequence consisted of a series of measurements with quantitative-, and manual pupillometry, performed subsequently to each other. Bland-Altman analysis with 95% limits of agreement (LoA) was used. Results: We included 14 sedated, comatose patients (6 females) with a mean age of 70 ± 12 years, resulting in 28 individual assessments. Quantitative pupillometry presented intraobserver variability with mean difference for maximum pupil size of 0.04 ± 0.13 with LoA of -0.21 to 0.28, and interobserver variability with mean difference of 0.03 ± 0.17 and of -0.31 to 0.36. For comparison, manual pupillometry revealed a mean difference of -0.02 ± 0.31 with LoA of -0.58 to 0.63, and -0.14 ± 0.44 with LoA of -1.00 to 0.71 for intra-, and interobserver variability, respectively. When comparing pupil size for left and right eyes, quantitative pupillometry resulted in LoA at -0.19, 0.25 (left) and -0.23, 0.32 (right), and LoA at -0.33, 0.39 (left) and -0.30, 0.34 (right) for intra-, and interobserver variability, respectively. Manual pupillometry presented LoA at -0.59, 0.63 (for both left and right), and LoA at -1.10, 0.79 (left) and -0.95, 0.83 (right) for intra-, and interobserver variability, respectively. Conclusion: Results from quantitative pupillometry are significantly more reproducible than manual measurements, suggesting it can be safely implemented in clinical practice for multimodal neuroprognostication of post-cardiac arrest patients.