Clinical grading and the effects of scaling.
Clinical grading and the effects of scaling.
复制标题
DOI：
--
复制
发表时间：
1991-02
影响因子：
4.4
通讯作者：
I. Bailey;M. Bullimore;T. Raasch;Hugh R. Taylor
中科院分区：
文献类型：
--
作者：
I. Bailey;M. Bullimore;T. Raasch;Hugh R. Taylor
在临床实践中，需要对检查中评估的功能和质量的大小或严重程度进行分级。流行的是使用四步分级量表来对临床发现的严重程度进行分类。作者讨论了临床分级量表及其对临床医生检测变化能力的影响。这些原则已被应用于从客观测量仪器、主观测试或临床医生做出主观判断的技术中得出的等级或测量。一个假设的数据集被用来显示与使用太粗糙的分级尺度相关的问题。作者提出了一个数学模型，有助于估计使用更精细尺度的好处。数据来自两个独立的研究，一个关于视力测量，另一个关于分级核混浊，以显示使用更精细的尺度来提高临床测量的灵敏度的优势。独立观察结果之间的高度一致性表明分级量表过于粗糙，并且这些量表不必要地降低了临床医生检测评估参数变化的能力。对于中等灵敏度，量表增量的大小不应超过差异的一个标准差，以便配对比较的一致性不超过37%。对于良好的临床灵敏度，量表增量的大小不应超过差异标准差的三分之一，在这种情况下，配对比较的一致性不应超过13%。这里提出的理论和证据可能会促使重新评估临床分级的常用方法。
In clinical practice, there has been a need to grade the magnitude or the severity of the functions and qualities that are assessed in the examination. It is popular to use a four-step grading scale to categorize the severity of clinical findings. The authors discuss clinical grading scales and their influence on the clinician's ability to detect change. These principles have been applied to grades or measures derived from either objective measuring instruments, subjective tests, or techniques in which the clinician makes subjective judgments. A hypothetical data set was used to show the problems associated with using grading scales that are too coarse. The authors presented a mathematic model that helps to estimate the benefits of using use of a finer scale. Data were presented from two separate studies, one on visual acuity measurement and the other on grading nuclear opacity, to show the advantages of using finer scales to enhance the sensitivity of clinical measurement. High levels of concordance between independent observations indicated that the grading scale was too coarse and that these scales needlessly reduced the clinician's ability to detect change in the parameter being assessed. For moderate sensitivity, the size of the scale increments should not exceed one standard deviation of the discrepancy so that the concordance of paired comparisons would not exceed 37%. For fine clinical sensitivity, the size of the scale increments should not exceed one third of the standard deviation of the discrepancy, in which case the concordance of paired comparisons would not exceed 13%. The theory and evidence presented here could prompt re-evaluations of common methods of clinical grading.