Department of Anesthesiology, University of California, San Diego, La Jolla, CA, United States.
Department of Anesthesiology, University of California, San Diego, La Jolla, CA, United States; Division of Biomedical Informatics, University of California, San Diego, La Jolla, CA, United States; Outcomes Research Consortium, Cleveland, OH, United States.
Burns. 2020 Nov;46(7):1565-1570. doi: 10.1016/j.burns.2020.04.021. Epub 2020 Apr 29.
Improvement in the care of burn patients has led to decreased mortality. Length of stay (LOS) has been used as a marker for quality of care in this population. However, the historical association of LOS as correlating only with % burn surface area (BSA) injury has been questioned with retrospective data suggesting other factors may also be associated with LOS. A model to predict prolonged LOS does not exist but could provide important information for clinicians and patients.
Data from January 2014 to December 2016 was used to develop a predictive model utilizing multivariable logistic regression. Prolonged hospital LOS was the outcome used with multiple covariates utilized to identify various associations. Odds ratios (OR) and their associated 95% confidence interval (CI) were reported for each covariate in the final regression model. Model performance in both the training and validation sets was evaluated using area under the receiver operating characteristic (ROC) curve (AUC) for discrimination and the Hosmer-Lemeshow (HL) test for goodness-of-fit.
A total of 441 patients was included in the final analysis, 296 (67.1%) of which were in the training set. Within the training set, the median hospital LOS was 14 days with a range of 4 to 205 days. Patient age (in decades), hypertension, total BSA, involvement of perineum, and abnormal white blood cell count were independent risk factors for prolonged hospital length of stay. When using this separate dataset, the model had an AUC of 0.81 (95% CI 0.74-0.88) and had good calibration based on the HL-test (p=0.10).
Prolonged hospitalization following burns is predicted by patient age (in decades), TBSA, hypertension, perineal involvement, and abnormal white blood cell count.
烧伤患者护理水平的提高导致死亡率降低。住院时间（LOS）已被用作该人群护理质量的指标。然而，随着回顾性数据表明其他因素也可能与 LOS 相关，LOS 仅与烧伤面积百分比（BSA）损伤相关的历史关联受到质疑。目前尚不存在预测 LOS 延长的模型，但它可以为临床医生和患者提供重要信息。
利用多变量逻辑回归，使用 2014 年 1 月至 2016 年 12 月的数据开发预测模型。将延长的住院 LOS 作为结果，利用多个协变量来识别各种关联。报告最终回归模型中每个协变量的优势比（OR）及其相关 95%置信区间（CI）。使用接受者操作特征（ROC）曲线下面积（AUC）评估模型在训练集和验证集中的性能，以进行区分，以及 Hosmer-Lemeshow（HL）检验以评估拟合优度。
共纳入 441 例患者进行最终分析，其中 296 例（67.1%）患者纳入训练集。在训练集中，中位数住院 LOS 为 14 天，范围为 4 至 205 天。患者年龄（以十年计）、高血压、总 BSA、会阴受累和异常白细胞计数是延长住院时间的独立危险因素。当使用这个独立数据集时，该模型的 AUC 为 0.81（95%CI 0.74-0.88），HL 检验显示具有良好的校准度（p=0.10）。
烧伤后住院时间延长的预测因素为患者年龄（以十年计）、TBSA、高血压、会阴受累和异常白细胞计数。
