BACKGROUND Pupillary reflex dilation is a reliable indicator of response to noxious stimulation. In a proof of concept study, we investigated the performance of pupillary pain index, a new score derived from pupillary reflex dilation measurements, to predict nociceptive response to endotracheal suctioning in sedated critically ill patients. METHODS Twenty brain-injured and 20 non-brain-injured patients were studied within 48 hours of admission (T1) in the intensive care unit and at 48-72 hours later (T2). Video-based pupillometer was used to determine pupillary reflex dilation during tetanic stimulation. The tetanic stimulation (100 Hz) was applied to the skin area innervated by the ulnar nerve and was stepwise increased from 10 to 60 mA until pupil size had increased by 13% compared to baseline. The maximum intensity value allowed the determination of a pupillary pain index score ranging from 1 (no nociception) to 9 (high nociception). The Behavioral Pain Scale response to endotracheal suctioning was measured thereafter. RESULTS Behavioral Pain Scale responses to endotracheal suctioning and pupillary pain index scores were positively correlated at T1 and T2 (both P < .01). After adjustments for repeated measurements and group of patients, the area under the receiver operating characteristic curve of pupillary pain index to predict Behavioral Pain Scale response to endotracheal suctioning was of 0.862 (95% CI, 0.714-0.954). In the combined set of patients, a pupillary pain index score of ≤4 could predict no nociceptive response to endotracheal suctioning with a sensitivity of 88% (95% CI, 68%-97%) and a specificity of 79% (95% CI, 66%-88%). By contrast with endotracheal suctioning, tetanic stimulation had no effect on intracranial pressure in the brain-injured group. CONCLUSIONS These results are a proof of concept. The nociceptive response to endotracheal suctioning could be accurately predicted using the determination of pupillary pain index score in sedated critically ill patients whether they have brain injury or not.
