Background: Elevated intracranial pressure (ICP) is a serious complication in brain injury. Because of the risks involved, ICP is not monitored in all patients at risk. Noninvasive screening tools to identify patients with elevated ICP are needed. Anisocoria, abnormal pupillary size, and abnormal pupillary light reflex are signs of high ICP, but manual pupillometry is arbitrary and subject to interrater variability. We have evaluated quantitative pupillometry as a screening tool for elevated ICP.
Methods: We performed a retrospective observational study of the association between Neurological Pupil index (NPi), measured with the Neuroptics NPi-200 pupillometer, and ICP in patients routinely monitored with invasive ICP measurement in the intensive care unit. We performed a nonparametric receiver operator curve analysis for ICP ≥ 20 mm Hg with NPi as a classification variable. We performed a Youden analysis for the optimal NPi cutoff value and recorded sensitivity and specificity for this cutoff value. We also performed a logistic regression with elevated ICP as the dependent variable and NPi as the independent variable.
Results: We included 65 patients with invasive ICP monitoring. A total of 2,705 measurements were analyzed. Using NPi as a screening tool for elevated ICP yielded an area under receiver operator curve of 0.72. The optimal mean NPi cutoff value to rule out elevated ICP was ≥ 3.9. The probability of elevated ICP decreased with increasing NPi, with an odds ratio of 0.55 (0.50, 0.61).
Conclusions: Screening with NPi may inform high stakes clinical decisions by ruling out elevated ICP with a high degree of certainty. It may also aid in estimating probabilities of elevated ICP. This can help to weigh the risks of initiating invasive ICP monitoring against the risks of not doing so. Because of its ease of use and excellent interrater reliability, we suggest further studies of NPi as a screening tool for elevated ICP.