Background/Purpose

Predicting the mortality in patients admitted to the ICU is important for determining a treatment strategy and public health policy. Although many scores have been developed to predict the mortality, these scores were based on Caucasian population. We aimed to develop a new prognostic index, the New nutritional index (NNI), to predict 90-days mortality after ICU admission based on Korean population.

Methods

Patients (1453) who admitted intensive care unit (ICU) of the Gangnam Severance hospital were analyzed. After exclusion, 984 patients were randomly divided into internal (n = 702) and external validation (n = 282) data set. The new nutritional index (NNI) was developed using univariate and multivariate logistic regression with backward selection of predictors. Receiver operating characteristic (ROC) curve analysis and comparison of the area under the curve (AUC) verified the better predictor of 90 days-mortality after ICU admission.

Results

The NNI better predicted 90 days-mortality compared to modified NUTRIC score, APACHE II scores, SOFA scores, CRP, glucose, total protein, and albumin level in internal and external data sets, with AUC of 0.862 (SE: 0.017, 95% CI: 0.829–0.895) and 0.858 (SE: 0.015, 95% CI: 0.829–0.887), respectively. The calibration plots using external data set for validation showed a close approximation to the logistic calibration of each nomogram, and p-value of Hosmer and Lemeshow test was 0.1804.

Conclusion

The NNI has advantages as a predictor of 90 days mortality based on nutritional status in the Korean population.

中文翻译：

---

推导并验证了一种新的营养指数，该营养指数可预测韩国人群中ICU入院90天后的死亡率。

背景/目的

预测入住ICU的患者的死亡率对于确定治疗策略和公共卫生政策很重要。尽管已开发出许多评分来预测死亡率，但这些评分是基于白种人人口的。我们旨在开发一种新的预后指数，即新营养指数（NNI），以根据韩国人口预测ICU入院后90天的死亡率。

方法

分析了江南遣散医院重症监护病房（ICU）的患者（1453）。排除后，将984位患者随机分为内部（n = 702）和外部验证（n = 282）数据集。新的营养指数（NNI）是使用单变量和多变量logistic回归并向后选择预测因子来开发的。接受者工作特征（ROC）曲线分析和曲线下面积比较（AUC）证实了ICU入院90天死亡率的更好预测指标。

结果

与修改后的NUTRIC评分，APACHE II评分，SOFA评分，CRP，葡萄糖，总蛋白和白蛋白水平相比，NNI更好地预测了90天死亡率，AUC为0.862（SE：0.017，95％CI ：0.829–0.895）和0.858（SE：0.015，95％CI：0.829–0.887）。使用外部数据集进行验证的校准图显示出与每个列线图的逻辑校准非常接近，Hosmer和Lemeshow测试的p值为0.1804。

结论

根据韩国人口的营养状况，NNI具有预测90天死亡率的优势。