Ch 28 (s28.5, s28.16); Ch 23
The only bedside instrument in Chapter 28 for which a published series reports no false positives, and the only one that is cheap, repeatable, needs no neurophysiologist and works through sedation. Use it as a twice-daily trend; do not act on a value inside the first 24 hours (adrenaline makes the pupils unreliable during resuscitation); and treat a normal index as uninformative rather than reassuring, since sensitivity was 60 percent at best.
Index trends in survivors versus non-survivors; index alone versus index plus the 12-hour PREDICT VA-ECMO score
DOI 10.1016/j.chest.2019.11.037
Automated infrared pupillometry (NPi-200) every 12 hours during the first 72 hours after ECMO insertion, generating the Neurological Pupil index
Single centre; 100 patients; threshold derived in the same dataset in which it was tested; the exact confidence interval for 100 percent specificity in 57 non-survivors has a lower bound well below 100 and is reported only for the combined score; the outcome is 90-day mortality rather than neurological outcome, in a population where mortality is heavily determined by cardiac and multi-organ factors; self-fulfilling prophecy is not excluded and cannot be excluded by this design.
Non-survivors (n=57) had significantly lower index values than survivors at every time point (all p under 0.01). An abnormal index, below 3 at any time from 24 to 72 hours, was 100 percent specific for 90-day mortality with 0 percent false positives. Index plus the 12-hour PREDICT VA-ECMO score gave specificity 100 percent (95 percent CI 92 to 100), sensitivity 60 percent (46 to 72), AUC 0.82.
100 consecutive sedated, mechanically ventilated adult venoarterial ECMO patients at a single European centre — 51 refractory cardiogenic shock, 49 refractory cardiac arrest. Observed 90-day survival 43 percent.