Risk Score to Predict Hospital-Acquired Pneumonia After Spontaneous Intracerebral Hemorrhage
Ruijun Ji,Haipeng Shen,Yuesong Pan,Wanliang Du,Penglian Wang,Gaifen Liu,Yilong Wang,Hao Li,Xingquan Zhao,Yongjun Wang,Q. Bi,Weiwei Zhang,L. Cui,Yu-heng Sun,Mao-lin He,Dongsheng Fan,X. Ji,Jimei Li,Fang Zhang,K. Feng,Xiaojun Zhang,Yansheng Li,Shaoshi Wang,Wei Fan,Zhenguo Liu,Xiao-Jiang Sun,Wei Li,Jianrong Liu,Xu Chen,Q. Bai,D. Gu,Xin Li,Q. Dong,Yan-xue Cheng,Lan Yu,Bin Li,Tongyu Wang,Kun Zhao,Chao-dong Zhang,Dingbo Tao,L. Yin,Fang Qu,Jingbo Zhang,Jianfeng Wang,Ying Lian,Jun Fan,Yingjian Gao,Min Cheng,Jiang Wu,Hua-Yu Sun,Jinying Li,Guozhong Li,Yulan Zhu,Z. Yang,Feng Yang,Jun Zhou,M. Guo,Zhengyi Li,Q. Ma,Renbin Huang,Bo Xiao,Kangning Chen,Xinyue Qin,Changlin Hu,Li Gao,J. Zeng,Anding Xu,Xiong Zhang,Minghua Shao,Feng Qi,Weimin Xiao,S. Zhang,Xiaoping Pan,Suyue Pan,Yefeng Cai,Qi Wan,Yunfei Xu,Kaifu Ke,Yue-nan Kong,Qing Di,Fengyang Shao,Yajun Jiang,Daming Wang,Li Guo,Wen-Jun Xue
DOI: https://doi.org/10.1161/STROKEAHA.114.005023
IF: 10.17
2014-09-01
Stroke
Abstract:Background and Purpose— We aimed to develop a risk score (intracerebral hemorrhage–associated pneumonia score, ICH-APS) for predicting hospital-acquired stroke-associated pneumonia (SAP) after ICH. Methods— The ICH-APS was developed based on the China National Stroke Registry (CNSR), in which eligible patients were randomly divided into derivation (60%) and validation (40%) cohorts. Variables routinely collected at presentation were used for predicting SAP after ICH. For testing the added value of hematoma volume measure, we separately developed 2 models with (ICH-APS-B) and without (ICH-APS-A) hematoma volume included. Multivariable logistic regression was performed to identify independent predictors. The area under the receiver operating characteristic curve (AUROC), Hosmer–Lemeshow goodness-of-fit test, and integrated discrimination index were used to assess model discrimination, calibration, and reclassification, respectively. Results— The SAP was 16.4% and 17.7% in the overall derivation (n=2998) and validation (n=2000) cohorts, respectively. A 23-point ICH-APS-A was developed based on a set of predictors and showed good discrimination in the overall derivation (AUROC, 0.75; 95% confidence interval, 0.72–0.77) and validation (AUROC, 0.76; 95% confidence interval, 0.71–0.79) cohorts. The ICH-APS-A was more sensitive for patients with length of stay >48 hours (AUROC, 0.78; 95% confidence interval, 0.75–0.81) than those with length of stay <48 hours (AUROC, 0.64; 95% confidence interval, 0.55–0.73). The ICH-APS-A was well calibrated (Hosmer–Lemeshow test) in the derivation (P=0.20) and validation (P=0.66) cohorts. Similarly, a 26-point ICH-APS-B was established. The ICH-APS-A and ICH-APS-B were not significantly different in discrimination and reclassification for SAP after ICH. Conclusion— The ICH-APSs are valid risk scores for predicting SAP after ICH, especially for patients with length of stay >48 hours.
Medicine