Multimodal approach for neurologic prognostication of out-of-hospital cardiac arrest patients undergoing targeted temperature management

Abstract

AIM: Since the introduction of targeted temperature management (TTM), the accuracy and timing of prognostic tests for post-cardiac arrest patients have changed. Although previous studies have demonstrated the effectiveness of a multimodal approach in assessing the prognosis of TTM patients, few studies have investigated an optimised strategy that sequentially combines different prognostic modalities. This study identified an optimal sequential combination of prognostic modalities to predict poor neurologic outcomes in patients undergoing TTM.

METHODS: We performed a retrospective analysis using TTM management registry data. All patients underwent an identical sequence of prognostic tests at fixed timings. The sequence included brain computed tomography (CT), serum neuron-specific enolase (NSE), electrophysiological examination, neurologic examination, and diffusion-weighted imaging. We used hierarchical classification and regression tree analysis to find the optimal prognostic model. The primary measure was a poor neurologic outcome at one month after cardiac arrest.

RESULTS: A total of 192 patients were included and 103 patients (53.6%) had poor neurologic outcomes. The final model consisted of brain CT, serum NSE, electroencephalogram, somatosensory-evoked potentials, and pupil light reflex. Our model predicted poor outcomes with a 0% false positive rate. Moreover, our model had an area under the receiver operating characteristic curve value of 0.911 (95% confidence interval, 0.872-0.950), which was significantly higher than that of each prognostic modality alone.

CONCLUSIONS: Our stepwise model showed excellent prognostic ability to predict poor outcomes at one month after cardiac arrest and may be used to minimise the risk of false pessimistic predictions in patients undergoing TTM.