Bismil Ramprasad, Dudek Nancy L, Wood Timothy J
Department of Psychiatry, University of Ottawa, Ottawa, Ontario, Canada.
Med Educ. 2014 Jul;48(7):724-32. doi: 10.1111/medu.12490.
In-training evaluation (ITE) is used to assess resident competencies in clinical settings. This assessment is documented on an evaluation report (In-Training Evaluation Report [ITER]). Unfortunately, the quality of these reports can be questionable. Therefore, training programmes to improve report quality are common. The Completed Clinical Evaluation Report Rating (CCERR) was developed to assess completed report quality and has been shown to do so in a reliable manner, thus enabling the evaluation of these programmes. The CCERR is a resource-intensive instrument, which may limit its use. The purpose of this study was to create a screening measure (Proxy-CCERR) that can predict the CCERR outcome in a less resource-intensive manner.
Using multiple regression, the authors analysed a dataset of 269 ITERs to create a model that can predict the associated CCERR scores. The resulting predictive model was tested on the CCERR scores for an additional sample of 300 ITERs.
The quality of an ITER, as measured by the CCERR, can be predicted using a model involving only three variables (R(2) = 0.61). The predictive variables included the total number of words in the comments, the variability of the ratings and the proportion of comment boxes completed on the form.
It is possible to model CCERR scores in a highly predictive manner. The predictive variables can be easily extracted in an automated process. Because this model is less resource-intensive than the CCERR, it makes it possible to provide feedback from ITER training programmes to large groups of supervisors and institutions, and even to create automated feedback systems using Proxy-CCERR scores.
培训期间评估（ITE）用于评估住院医师在临床环境中的能力。该评估记录在一份评估报告（培训期间评估报告[ITER]）中。不幸的是，这些报告的质量可能存在问题。因此，提高报告质量的培训项目很常见。已开发出完整临床评估报告评分（CCERR）来评估完整报告的质量，并已证明其评估方式可靠，从而能够对这些项目进行评估。CCERR是一种资源密集型工具，这可能会限制其使用。本研究的目的是创建一种筛选措施（Proxy-CCERR），能够以资源消耗较少的方式预测CCERR结果。
作者使用多元回归分析了269份ITER的数据集，以创建一个能够预测相关CCERR分数的模型。在另外300份ITER样本的CCERR分数上对所得预测模型进行了测试。
使用仅包含三个变量的模型（R² = 0.61），可以预测由CCERR衡量的ITER质量。预测变量包括评论中的单词总数、评分的可变性以及表格上填写的评论框比例。
以高度预测性的方式对CCERR分数进行建模是可能的。预测变量可以在自动化过程中轻松提取。由于该模型比CCERR资源消耗少，因此有可能向大量的主管和机构提供来自ITER培训项目的反馈，甚至使用Proxy-CCERR分数创建自动反馈系统。
