We aimed to determine whether urinary neutrophil gelatinase-associated lipocalin (uNGAL) can accurately predict persistent AKI, major adverse kidney events at 30 days (MAKE30) and 365 days (MAKE365) in hospitalized AKI patients. This is a retrospective study of adult patients who were admitted at King Chulalongkorn Memorial Hospital. We performed multivariable logistic regression for persistent AKI, MAKE30, and MAKE365. We developed equations for predicting MAKE30 and MAKE365 and divided the dataset into derivation and validation cohorts. uNGAL performance and predictive models were assessed using the area under the receiver operating characteristic curve (AROC). Among 1,322 patients with AKI, 76.9%, 45.1%, and 61.7% had persistent AKI, MAKE30, and MAKE365. The AROC were 0.75 (95% confidence interval[CI] 0.70–0.80), 0.66 (95%CI 0.61–0.71), and 0.64 (95%CI 0.59–0.70) for prediction of persistent AKI, MAKE30, and MAKE365 by uNGAL. The AROC in the validation dataset combining uNGAL with clinical covariates were 0.74 (95%CI 0.69–0.79) and 0.72 (95%CI 0.67–0.77) for MAKE30 and MAKE365. We demonstrated an association between uNGAL and persistent AKI, MAKE30, and MAKE365. Prediction models combining uNGAL can modestly predict MAKE30 and MAKE365. Therefore, uNGAL is a useful tool for improving AKI risk stratification.

我们旨在确定尿中性粒细胞明胶酶相关的脂蛋白（uNGAL）是否可以准确预测住院AKI患者在30天（MAKE30）和365天（MAKE365）的持续性AKI，主要不良肾脏事件。这是一项对朱拉隆功国王纪念医院收治的成年患者的回顾性研究。我们对持久性AKI，MAKE30和MAKE365执行了多变量logistic回归。我们开发了用于预测MAKE30和MAKE365的方程式，并将数据集分为推导和验证队列。使用接收器工作特性曲线（AROC）下的面积评估uNGAL性能和预测模型。在1,322例AKI患者中，有76.9％，45.1％和61.7％的患者持续存在AKI，MAKE30和MAKE365。AROC分别为0.75（95％置信区间[CI] 0.70-0.80），0.66（95％CI 0.61-0.71）和0。64（95％CI 0.59–0.70）用于通过uNGAL预测持久性AKI，MAKE30和MAKE365。对于MAKE30和MAKE365，将uNGAL与临床协变量相结合的验证数据集的AROC分别为0.74（95％CI 0.69–0.79）和0.72（95％CI 0.67–0.77）。我们展示了uNGAL与持久性AKI，MAKE30和MAKE365之间的关联。结合uNGAL的预测模型可以适度地预测MAKE30和MAKE365。因此，uNGAL是改善AKI风险分层的有用工具。结合uNGAL的预测模型可以适度地预测MAKE30和MAKE365。因此，uNGAL是改善AKI风险分层的有用工具。结合uNGAL的预测模型可以适度地预测MAKE30和MAKE365。因此，uNGAL是改善AKI风险分层的有用工具。