A simple post hoc transformation that improves the metric properties of the BBB scale for rats with moderate to severe spinal cord injury
A simple post hoc transformation that improves the metric properties of the BBB scale for rats with moderate to severe spinal cord injury
复制标题
DOI：
10.1089/neu.2004.21.1601
复制
发表时间：
2004-11-01
影响因子：
4.2
通讯作者：
Grau, JW
中科院分区：
文献类型：
--
作者：
Ferguson, AR;Hook, MA;Grau, JW
Basso、Beattie、Bresnahan (BBB) 开放场运动量表是脊髓损伤 (SCI) 后功能恢复的常用测量方法。为了检查该量表的度量特性，我们对来自两个不同实验室的 643 只中度和重度 SCI 大鼠（12.5、25 或 50 mm MASCIS）的 BBB 评分进行了详细分析。分析表明，BBB 量表在量表中最常用的部分中是有序的。数据集中并不经常分配较高的分数（14 或更高），因为没有对轻度受伤的动物进行采样，使得量表上端的序数性质难以评估。在这个范围内分配的罕见分数会不成比例地增加方差。在这些条件下，将 14 以上的分数合并为一类会增加效应大小。对量表下部区域的分析表明，很少分配某些分数（2 和 3），这意味着量表存在不连续性。量表下部的不连续性质给参数和非参数统计分析带来了问题。合并分数 2/3/4 消除了差距，增强了量表的度量属性。在评估的损伤条件下，转换有助于确保数据的连续性和有序性。此外，间隔持续时间在转换尺度的整个范围内具有可比性，从而允许应用参数统计技术。在很少分数落在量表上部的情况下，应以事后方式应用转换，以减少变异性并提高功效。
The Basso, Beattie, Bresnahan (BBB) open field locomotor scale is a popular measure of functional recovery following spinal cord injury (SCI). To examine the metric properties of the scale, we performed detailed analyses of BBB scores from 643 rats with moderate and severe SCI (12.5, 25, or 50 mm MASCIS) from two different laboratories. The analyses revealed that the BBB scale is ordinal in the most frequently used portion of the scale. Higher scores (14 and greater) were not frequently assigned in the dataset as animals with mild injuries were not sampled, making the ordinal nature of the upper end of the scale difficult to assess. The rare scores assigned in this range disproportionately increased variance. Under these conditions collapsing scores above 14 into one category increased effect size. Analysis of the lower region of the scale revealed that some scores (2 and 3) were rarely assigned, implying a discontinuity in the scale. The discontinuous nature of the lower portion of the scale presents a problem for both parametric and nonparametric statistical analyses. Pooling scores 2/3/4 eliminated the gap, enhancing the metric properties of the scale. Under the injury conditions evaluated, the transformation helped assure that the data were continuous and ordered. Further, interval durations were comparable across the entire range of the transformed scale, allowing application of parametric statistical techniques. The transformation should be applied in a post hoc fashion to reduce variability and increase power in cases where few scores fall in upper portion of the scale.