Riordan John P, Dell Wayne L, Patrie James T
Department of Emergency Medicine, University of Virginia, Charlottesville, Virginia.
Public Health Sciences, University of Virginia, Charlottesville, Virginia.
J Emerg Med. 2017 May;52(5):769-779. doi: 10.1016/j.jemermed.2016.11.018. Epub 2016 Dec 21.
Emergency department crowding has led to innovative "front end" care models to safely and efficiently care for medium and lower acuity patients. In the United States, most treatment algorithms rely on the emergency severity index (ESI) triage tool to sort patients. However, there are no objective criteria used to differentiate ESI 3 patients.
We seek to derive and validate a model capable of predicting patient discharge disposition (DD) using variables present on arrival to the emergency department for ESI 3 patients.
Our retrospective cohort study included adult patients with an ESI triage designation 3 treated in an academic emergency department over the course of 2 successive years (2013-2015). The main outcome was DD. Two datasets were used in the modeling process. One dataset, the derivation dataset (n = 25,119), was used to develop the statistical model, while the second dataset, the validation dataset (n = 24,639), was used to evaluate the statistical model's prediction performance.
All variables included in the derivation model were uniquely associated with DD status (p < 0.001). We assessed multivariate adjusted odds ratios (ORs) and 95% confidence intervals (CIs) for age (2.50 [95% CI 2.35-2.65]), arrival mode (1.85 [95% CI 1.74-1.96]), heart rate (1.31 [95% CI 1.26-1.37]), sex (1.35 [95% CI 1.28-1.43]), oxygen saturation (1.06 [95% CI 1.01-1.10]), temperature (1.10 [95% CI 1.06-1.15]), systolic blood pressure (1.18 [95% CI 1.12-1.25]), diastolic blood pressure (1.16 [95% CI 1.09-1.22]), respiratory rate (1.05 [95% CI 1.01-1.10]), and pain score (1.13 [95% CI 1.06-1.21]). The validation C-statistic was 0.73.
We derived and validated a model and created a nomogram with acceptable discrimination of ESI 3 patients on arrival for purposes of predicting DD. Incorporating these variables into the care of these patients could improve patient flow by identifying patients who are likely to be discharged.
急诊科拥挤促使创新的“前端”护理模式出现，以安全、高效地护理中低急症患者。在美国，大多数治疗算法依靠急诊严重程度指数（ESI）分诊工具对患者进行分类。然而，尚无用于区分ESI 3级患者的客观标准。
我们试图推导并验证一种模型，该模型能够使用急诊室接诊时ESI 3级患者的现有变量来预测患者出院处置情况（DD）。
我们的回顾性队列研究纳入了连续两年（2013 - 2015年）在一所学术性急诊科接受治疗的ESI分诊为3级的成年患者。主要结局是DD。建模过程中使用了两个数据集。一个数据集，即推导数据集（n = 25119），用于建立统计模型，而第二个数据集，即验证数据集（n = 24639），用于评估统计模型的预测性能。
推导模型中包含所有变量均与DD状态存在独特关联（p < 0.001）。我们评估了年龄（2.50 [95% CI 2.35 - 2.65]）、就诊方式（1.85 [95% CI 1.74 - 1.96]）、心率（1.31 [95% CI 1.26 - 1.37]）、性别（1.35 [95% CI 1.28 - 1.43]）、血氧饱和度（1.06 [95% CI 1.01 - 1.10]）、体温（1.10 [95% CI 1.06 - 1.15]）、收缩压（1.18 [95% CI 1.12 - 1.25]）、舒张压（1.16 [95% CI 1.09 - 1.22]）、呼吸频率（1.05 [95% CI 1.01 - 1.10]）和疼痛评分（1.13 [95% CI 1.06 - 1.21]）的多变量调整比值比（OR）和95%置信区间（CI）。验证C统计量为0.73。
我们推导并验证了一种模型，并创建了列线图，该列线图对急诊室接诊时的ESI 3级患者具有可接受的区分度，用于预测DD。将这些变量纳入这些患者的护理中，可通过识别可能出院的患者来改善患者流动情况。
