Groot L J J, Gosens N, Vles J S H, Hoogland G, Aldenkamp A P, Rouhl R P W
Maastricht University Medical Center, Dept. of Neurology, PO Box 5800, 6202 AZ Maastricht, The Netherlands.
Maastricht University Medical Center, Dept. of Neurology, PO Box 5800, 6202 AZ Maastricht, The Netherlands; School for Mental Health and Neurosciences, Maastricht University, PO Box 616 (UNS 50, BOX 38), 6200 MD Maastricht, The Netherlands.
Epilepsy Behav. 2015 Jan;42:10-3. doi: 10.1016/j.yebeh.2014.10.030. Epub 2014 Dec 10.
The Racine scale is a 5-point seizure behavior scoring paradigm used in the amygdala kindled rat. Though this scale has been applied widely in experimental epilepsy research, studies of reproducibility are rare. The aim of the current study was, therefore, to assess its interobserver variability and intraobserver variability.
A video database set was acquired in the course of amygdala kindling of 67 Wistar rats. Six blinded observers received scoring instructions and then viewed a set of 15 random videos (session #1). Next, each observer scored 379 to 1048 additional videos (session #2) and finally scored the same set of 15 videos again (session #3). Scores included the occurrence of seizures (yes or no), the total seizure time (start of stimulus until the absence of seizure behavior), and the highest Racine stage. Interobserver variability and intraobserver variability were assessed in and between sessions #1 and #3 using a 2-way mixed intraclass correlation or Cohen's kappa depending on the variable.
Interobserver agreement in session #1 was 0.664 for seizure occurrence, 0.861 for total seizure time, and 0.797 for the highest Racine stage. In session #3, interobserver agreement on seizure occurrence declined to 0.492, total seizure time declined to 0.625, and agreement for the highest Racine stage was 0.725. Interobserver agreement was scored insufficiently on focal R2 seizures in both sessions (0.287 and 0.182). Intraobserver agreement reached >0.80 agreement for seizure occurrence, highest seizure score, and total seizure time in 3 out of 4 observers. Racine's scale stage 2 seizure scores were only 0.135 in one observer but 0.650, 0.810, and 0.635 in the other observers.
Overall, interobserver agreement and intraobserver agreement in scoring with Racine's scale were adequate. However, because interobserver agreement declined after a period of individually scoring videos, we suggest periodic repetition of the standardized instruction in the course of evaluating videos in order to ensure reproducible results.
拉辛量表是一种用于杏仁核点燃大鼠的5分制癫痫发作行为评分范式。尽管该量表已在实验性癫痫研究中广泛应用，但关于其可重复性的研究却很少。因此，本研究的目的是评估其观察者间变异性和观察者内变异性。
在67只Wistar大鼠杏仁核点燃过程中获取了一个视频数据库集。六名不知情的观察者接受评分指导，然后观看一组15个随机视频（第1阶段）。接下来，每位观察者对另外379至1048个视频进行评分（第2阶段），最后再次对同一组15个视频进行评分（第3阶段）。评分包括癫痫发作的发生情况（是或否）、总发作时间（刺激开始至癫痫发作行为消失）以及最高拉辛阶段。根据变量情况，使用双向混合组内相关系数或科恩kappa系数在第1阶段和第3阶段内及之间评估观察者间变异性和观察者内变异性。
在第1阶段，观察者间在癫痫发作发生情况上的一致性为0.664，在总发作时间上为0.861，在最高拉辛阶段上为0.797。在第3阶段，观察者间在癫痫发作发生情况上的一致性降至0.492，总发作时间降至0.625，最高拉辛阶段的一致性为0.725。在两个阶段中，观察者间在局灶性R2癫痫发作上的一致性评分均不足（分别为0.287和0.182）。4名观察者中有3名在癫痫发作发生情况、最高发作评分和总发作时间上的观察者内一致性达到>0.80。在一名观察者中，拉辛量表2期癫痫发作评分仅为0.135，但在其他观察者中分别为0.650、0.810和0.635。
总体而言，使用拉辛量表评分时观察者间一致性和观察者内一致性是足够的。然而，由于在一段时间的单独视频评分后观察者间一致性下降，我们建议在视频评估过程中定期重复标准化指导，以确保结果的可重复性。
