[Feasibility of using amplitude-integrated electroencephalogram to identify epileptic seizures by pediatric intensive care unit medical staff independently].

Objective: To evaluate the feasibility of using amplitude-integrated electroencephalogram (aEEG) to identify epileptic seizures by physicians and nurses in pediatric intensive care unit (PICU) independently. Method: Six testees (two PICU physicians versus one EEG physician and two PICU nurses versus one EEG technician) accepted a short-term training, then interpreted aEEG in a single blinded way. These aEEG recordings with synchronous VEEG monitoring were done from January 2013 to May 2015 in PICU. The testees should recognize and mark both the seizure type and the seizure duration from the two-channel recorder (C3/C4) of aEEG (short-term seizure or status epilepticus (SE)). Using raw VEEG monitoring as a gold standard to determine a seizure, the accuracy, missing and error rate of each testees were confirmed, and the reasons of the latter two situations were analyzed by rank sum test and inter-testee agreement (k) . Result: Eighty-two aEEG recordings from 56 patients were interpreted. Thirty-two recordings had 141 epileptic seizures confirmed by VEEG, including 119 short-term seizures and 22 SE. There were 50 recordings without seizure. As for the short-term seizures, the average accuracy of 6 testees by aEEG alone was (66±4)%. The accuracy for SE was 100% in three testees and 95% in the other three. Missing rate of the seizures were 24.1%-32.6% in all 6 testees. Those missed seizures were all short-term (duration less than 20 seconds) but one SE. The average error rate was (19±9) times (P=0.000). These false interpretations were misunderstanding, many kinds of artifacts were regarded as epileptic seizures. The accuracy and missing rate among the testees had no significance(P=0.930, 0.996), but the error rate had(P=0.000). The inter-testee agreement (k) between two physicians in PICU and the EEG doctor were 0.700 and 0.687 respectively (P<0.01), which is good. As for two nurses and the EEG technician, the inter-testee agreement (k) was 0.705 and 0.396 respectively (P<0.01). Conclusion: Most of the seizures especially status epilepticus can be detected by PICU staff after short term training. The accuracy of identification of epileptic seizures was similar among observers from PICU and EEG, although some short-term seizures may be missed, and artifacts are mistaken.It's necessary to communicate with EEG doctors and compare with the row VEEG when physicians in PICU find suspicious events.