Prehosp Emerg Care. 2021 Jan-Feb;25(1):67-75. doi: 10.1080/10903127.2020.1725198. Epub 2020 Feb 25.
Various screening tools, ranging in complexity, have been developed to predict large vessel occlusion (LVO) stroke in the prehospital setting. Our objective was to determine whether newly-developed LVO stroke scales offer a clinically-meaningful advantage over the Cincinnati Prehospital Stroke Scale (CPSS).
We retrospectively analyzed prehospital patient care records linked with hospital data from 151 EMS agencies in the United States, between January 1, 2018 and December 31, 2018. We compared the CPSS to the Rapid Arterial Occlusion Evaluation (RACE), Los Angeles Motor Scale (LAMS), and the Vision, Aphasia, Neglect (VAN) assessment for LVO prediction. For each stroke scale, we used the intersection of sensitivity and specificity curves to determine optimal prediction cut-points. We used area under the ROC curve and 95% confidence intervals to assess for differences in discriminative ability between scales.
We identified 13,596 prehospital records with one or more documented stroke scales of interest. Among these, 4,228 patients were diagnosed with stroke. Over half (57%, n = 2,415) of patients diagnosed with stroke experienced an acute ischemic stroke. Of patients with ischemic stroke, 26% (n = 628) were diagnosed with LVO. A CPSS score of 2 or higher demonstrated sensitivity = 69% and specificity = 78% for LVO. A RACE score of 4 or higher demonstrated sensitivity = 63%, specificity = 73%. A LAMS score of 3 or higher demonstrated sensitivity = 63%, specificity = 72% and a positive VAN score demonstrated sensitivity = 86%, specificity = 65%. Comparing the area under the ROC curve for each scale revealed no statistically significant differences in discriminative ability for LVO stroke.
In this large sample of real-world prehospital patient encounters, the CPSS demonstrated similar predictive performance characteristics compared to the RACE, LAMS, and VAN for detecting LVO stroke. Prior to implementing a specific screening tool, EMS agencies should evaluate ease of use and associated implementation costs. Scored 0-3, the simple, widely-used CPSS may serve as a favorable prehospital screening instrument for LVO detection with a cut-point of 2 or higher maximizing the tradeoff between sensitivity and specificity.
为了在院前环境中预测大血管闭塞（LVO）卒中，已经开发出各种复杂程度不同的筛查工具。我们的目的是确定新开发的 LVO 卒中量表是否比辛辛那提院前卒中量表（CPSS）具有临床意义上的优势。
我们回顾性分析了美国 151 个急救机构的院前患者护理记录与医院数据之间的关联，时间范围为 2018 年 1 月 1 日至 2018 年 12 月 31 日。我们将 CPSS 与快速动脉闭塞评估（RACE）、洛杉矶运动量表（LAMS）和视觉、失语、忽视（VAN）评估进行比较，以预测 LVO。对于每个卒中量表，我们使用敏感性和特异性曲线的交点来确定最佳预测切点。我们使用 ROC 曲线下面积和 95%置信区间来评估量表之间的鉴别能力差异。
我们共确定了 13596 份记录有一个或多个感兴趣的卒中量表的院前记录。在这些患者中，4228 例被诊断为卒中。超过一半（57%，n=2415）的卒中患者经历了急性缺血性卒中。在缺血性卒中患者中，26%（n=628）被诊断为 LVO。CPSS 评分≥2 分对 LVO 的敏感性为 69%，特异性为 78%。RACE 评分≥4 分的敏感性为 63%，特异性为 73%。LAMS 评分≥3 分的敏感性为 63%，特异性为 72%，阳性 VAN 评分的敏感性为 86%，特异性为 65%。比较每个量表的 ROC 曲线下面积，发现 LVO 卒中的鉴别能力没有统计学意义上的差异。
在这个真实世界的大量院前患者就诊样本中，CPSS 在检测 LVO 卒中方面与 RACE、LAMS 和 VAN 相比，具有相似的预测性能特征。在实施特定的筛查工具之前，EMS 机构应评估其易用性和相关实施成本。评分 0-3 的简单、广泛使用的 CPSS 可以作为 LVO 检测的有利院前筛查工具，其切点为 2 或更高，在敏感性和特异性之间取得最佳权衡。
