DOI: 10.1097/pcc.0000000000004015 ISSN: 1529-7535

Neurological Pupil Index and Outcome After Cardiac Arrest in Children: Single-Center, Retrospective Cohort, 2018–2023

Maya R. Silver, Hongyan Liu, Alexis Z. Tomlinson, Ayush Lacoul, Kathryn Graham, Ryan W. Morgan, Danielle Traynor, Amy Kleinschmidt, Kenya Agarwal, Kristen Lourie, Robert A. Berg, Nadir Yehya, Alexis Topjian, Matthew P. Kirschen

Objectives:

To determine the association between abnormal pupillary reactivity using automated infrared pupillometry and unfavorable outcome after pediatric cardiac arrest (CA).

Design:

Single-center, retrospective, cohort study, 2018–2023.

Setting:

Large tertiary PICU in the United States.

Patients:

We identified 127 patients 18 years old or younger with 8464 pupillary assessments, greater than or equal to 2 measured within 72 hours after return of circulation (ROC).

Interventions:

None.

Measurements and Main Results:

Abnormal pupillary reactivity was defined as greater than or equal to 2 neurological pupil index (NPi) values less than 3 in either eye. Unfavorable outcome was hospital discharge Pediatric Cerebral Performance Category of 4–6 and increase greater than or equal to 1 from pre-CA status. We explored the diagnostic performance of abnormal NPi for categorizing unfavorable outcome in the first 72 hours post-ROC, and at the time periods of 0–12, 12–24, 24–48, and 48–72 hours. We used a mixed-effects linear regression model to evaluate whether NPi values were associated with favorable vs. unfavorable outcome, after adjusting for clinical variables. Group-based trajectory modeling (GBTM) characterized NPi trajectories over time. Abnormal pupillary reactivity had a sensitivity of 66%, specificity 79%, positive predictive value of 82%, negative predictive value of 63%, and area under the receiver operating characteristic curve of 0.73 (95% CI, 0.65–0.80) for unfavorable outcome. Specificity did not differ between time periods. In the mixed-effects model, unfavorable outcome was associated with lower NPi values that decreased over time (interaction of time from ROC and outcome β = –0.02; p < 0.001). Through GBTM we identified a subgroup of patients with favorable outcome whose NPi was initially abnormal but subsequently improved over time.

Conclusions:

In our 2018–2023 experience, the presence of at least two NPi values less than 3 post-CA was associated with unfavorable outcome, with an acceptable level of diagnostic characterization. Of note, some patients with initially lower NPi values improved over time and had favorable outcome.

More from our Archive