University Hospital Basel, Emergency Department, Switzerland.
Ann Emerg Med. 2012 Sep;60(3):317-25.e3. doi: 10.1016/j.annemergmed.2011.12.013. Epub 2012 Mar 7.
We test predictive validity, interrater reliability, and diagnostic accuracy of the Emergency Severity Index in older emergency department (ED) patients and identify reasons for inadequate triage.
We analyzed data of patients aged 65 years or older who were included in a prospective, single-center cohort study. Predictive validity was assessed by investigating associations of resources, disposition, length of stay, and mortality with Emergency Severity Index levels. Diagnostic accuracy was tested by calculating sensitivity and specificity of Emergency Severity Index level 1 for the prediction of a lifesaving intervention. For the assessment of interrater reliability, 2 experts independently reviewed the triage nurses' notes. Agreement was estimated as raw agreement and as Cohen's weighted κ.
In total, 519 older patients were included. Emergency Severity Index level was associated with resource consumption (Spearman's ρ=-0.449; 95% confidence interval [CI] -0.519 to -0.379), disposition (Kendall's τ=-0.452; 95% CI -0.516 to -0.387), ED length of stay (Kruskal-Wallis χ(2)=92.5; df=4; P<.001), and mortality (log-rank χ(2)=37.04; df=3; P<.001). The sensitivity of the Emergency Severity Index to predict lifesaving interventions was 0.462 (95% CI 0.232 to 0.709), and the specificity was 0.998 (95% CI 0.989 to 1.000). Interrater reliability between experts was high (raw agreement 0.917, 95% CI 0.894 to 0.944; Cohen's weighted κ(w)=0.934, 95% CI 0.913 to 0.954). Undertriage occurred in 117 cases. Main reasons were neglect of high-risk situations and failure to appropriately interpret vital signs.
In our study, older patients were at risk for undertriage. However, our results suggest that the Emergency Severity Index is reliable and valid for triage of older patients.
我们测试了紧急严重程度指数（Emergency Severity Index，ESI）在老年急诊科患者中的预测有效性、评分者间信度和诊断准确性，并确定了分诊不足的原因。
我们分析了一项前瞻性、单中心队列研究中纳入的年龄在 65 岁或以上的患者数据。通过调查资源使用、处置、住院时间和死亡率与 ESI 水平的关联来评估预测有效性。通过计算 ESI 1 级对救生干预的预测的敏感性和特异性来测试诊断准确性。为了评估评分者间信度，2 位专家独立审查了分诊护士的记录。采用原始一致性和 Cohen 的加权 κ（Cohen's weighted κ）来估计一致性。
共纳入 519 例老年患者。ESI 水平与资源消耗（Spearman's ρ=-0.449；95%置信区间[CI] -0.519 至 -0.379）、处置（Kendall's τ=-0.452；95% CI -0.516 至 -0.387）、急诊科住院时间（Kruskal-Wallis χ²=92.5；df=4；P<.001）和死亡率（对数秩 χ²=37.04；df=3；P<.001）相关。ESI 预测救生干预的敏感性为 0.462（95% CI 0.232 至 0.709），特异性为 0.998（95% CI 0.989 至 1.000）。专家间的评分者间信度较高（原始一致性 0.917，95% CI 0.894 至 0.944；Cohen 的加权 κ（Cohen's weighted κ）=0.934，95% CI 0.913 至 0.954）。117 例患者分诊不足。主要原因是忽视高危情况和未能正确解读生命体征。
在我们的研究中，老年患者有分诊不足的风险。然而，我们的结果表明，ESI 对于老年患者的分诊是可靠和有效的。
