Beyond dichotomy: patterns and amplitudes of SSEPs and neurological outcomes after cardiac arrest

BACKGROUND: We hypothesized that the absence of P25 and the N20–P25 amplitude in somatosensory evoked potentials (SSEPs) have higher sensitivity than the absence of N20 for poor neurological outcomes, and we evaluated the ability of SSEPs to predict long-term outcomes using pattern and amplitude analyses. METHODS: Using prospectively collected therapeutic hypothermia registry data, we evaluated whether cortical SSEPs contained a negative or positive short-latency wave (N20 or P25). The N20–P25 amplitude was defined as the largest difference in amplitude between the N20 and P25 peaks. A good or poor outcome was defined as a Glasgow-Pittsburgh Cerebral Performance Category (CPC) score of 1–2 or 3–5, respectively, 6 months after cardiac arrest. RESULTS: A total of 192 SSEP recordings were included. In all patients with a good outcome (n = 51), both N20 and P25 were present. Compared to the absence of N20, the absence of N20–P25 component improved the sensitivity for predicting a poor outcome from 30.5% (95% confidence interval [CI], 23.0–38.8%) to 71.6% (95% CI, 63.4–78.9%), while maintaining a specificity of 100% (93.0–100.0%). Using an amplitude < 0.64 μV, i.e., the lowest N20–P25 amplitude in the good outcome group, as the threshold, the sensitivity for predicting a poor neurological outcome was 74.5% (95% CI, 66.5–81.4%). Using the highest N20–P25 amplitude in the CPC 4 group (2.31 μV) as the threshold for predicting a good outcome, the sensitivity and specificity were 52.9% (95% CI, 38.5–67.1%) and 96.5% (95% CI, 91.9–98.8%), respectively. The predictive performance of the N20–P25 amplitude was good, with an area under the receiver operating characteristic curve (AUC) of 0.94 (95% CI, 0.90–0.97). The absence of N20 was statistically inferior regarding outcome prediction (p < 0.05), and amplitude analysis yielded significantly higher AUC values than did the pattern analysis (p < 0.05). CONCLUSIONS: The simple pattern analysis of whether the N20–P25 component was present had a sensitivity comparable to that of the N20–P25 amplitude for predicting a poor outcome. Amplitude analysis was also capable of predicting a good outcome. ELECTRONIC SUPPLEMENTARY MATERIAL: The online version of this article (10.1186/s13054-019-2510-x) contains supplementary material, which is available to authorized users.