Quantitative pupillary light reflex assessment for prognosis of carbon monoxide poisoning

BACKGROUND: A non-reactive pupil in standard pupillary light reflex (sPLR) is regarded as a factor predicting neurological sequelae at 1-month after carbon monoxide (CO) poisoning. An automated pupillometer is used in the intensive care unit to quantitatively assess PLR. Quantitative PLR (qPLR) was superior to sPLR using penlight for prognosis of various neurological diseases. Therefore, this study aimed to analyze whether quantitative pupillary variables (neurological Pupil index [NPi] and qPLR) are superior to sPLR in predicting 1-month neurocognitive sequelae after acute CO poisoning. METHODS: We performed a prospective observational study of consecutive patients with acute CO poisoning admitted to an emergency department (ED) between August 2019 and December 2020 in a single academic medical center. sPLR and pupillometer examinations (qPLR and NPi) were performed by emergency physicians at the ED on hospital days 0–2. The lowest values among those recorded within 24 h and during the total measurement period were considered the 24-h and total lowest values, respectively. Global Deterioration Scale scores were measured at 1 month as an outcome and were dichotomized into favorable (1–4) or poor (5–7) outcomes. RESULTS: We analyzed the data of 104 adult patients with acute CO poisoning. qPLR was significantly higher in the favorable outcome group than in the poor outcome group 24-h and total lowest values (21.2% vs. 15.0%, p = 0.006 and 21.0% vs. 14.8%, p = 0.006). qPLR <18% had fair predictive power for poor neurocognitive outcomes [area under the curve (AUC), 0.70; 95% confidence interval (0.60–0.78)]. Among the patients with decreased mental status (Glasgow Coma Scale ≤12), the power of NPi and qPLR increased [AUC, 0.72 and AUC, 0.80]. NPi < 1 and qPLR <18% showed sensitivity (9.5% vs. 76.2%) and specificity (98.8% vs. 67.5%) for the prediction of poor outcomes. qPLR was significantly superior to sPLR in predicting poor neurocognitive outcomes at 1 month after CO poisoning (p = 0.007). CONCLUSION: qPLR and NPi were superior to sPLR in terms of predicting poor neurocognitive outcomes. qPLR and NPi measured from hospital days 0–2 may be valuable in predicting neurocognitive outcome.