Original Article
Impact and predictive value of time-dependent acute physiology and chronic health evaluation Ⅱ score on death risk in patients with severe stroke
data analysis based on MIMIC-Ⅲ
Cheng Yi, Chen Qi, Wu Shengyong, Zhu Ronghui, Wu Cheng
Published 2021-10-28
Cite as Chin Crit Care Med, 2021, 33(10): 1237-1242. DOI: 10.3760/cma.j.cn121430-20210409-00533
Abstract
ObjectiveTo study the influence of time-dependent acute physiology and chronic health evaluation Ⅱ (APACHE Ⅱ) score on 14-day death risk in patients with severe stroke, and to provide reference for clinical diagnosis and treatment.
MethodsData of 3 229 patients with severe stroke were enrolled from Medical Information Mart for Intensive Care-Ⅲ (MIMIC-Ⅲ). According to the main types of stroke, the patients were divided into subarachnoid hemorrhage (SAH), intracerebral hemorrhage (ICH), ischemic stroke (IS) and other groups. According to age, patients were divided into > 60 years old and ≤ 60 years old subgroups. According to the baseline of sequential organ failure assessment (SOFA) score, they were divided into subgroups of > 3 and ≤ 3. The daily measured values of APACHE Ⅱ scores in each patient were recorded. And all-cause death within 14 days after admission to intensive care unit (ICU) was used as the outcome index to obtain the survival status and survival time of patients. Joint models for longitudinal and time-to-event data were established to evaluate the effect of APACHE Ⅱ score measured at multiple time points on the death risk of patients, and a subgroup analysis was performed.
ResultsAmong the joint models, the one which include APACHE Ⅱ score, and the interaction items between APACHE Ⅱ and age showed the better fitting. Further analysis showed that APACHE Ⅱ score was affected by age, gender, hospital admission, baseline SOFA score and smoking history. After controlling for these confounding factors, APACHE Ⅱ score was significantly associated with 14-day all-cause death in patients with severe stroke [hazard ratio (HR) = 1.48, 95% confidence interval (95%CI) was 1.31-1.66, P < 0.001], which indicated that the risk of death increased by 48% (95%CI was 31%-66%) for each 1-point increase in APACHE Ⅱ score. Subgroup analysis showed that for different types of severe stroke patients, APACHE Ⅱ score had a greater impact on the risk of 14-day death in SAH patients (HR = 1.43, 95%CI was 1.10-1.85), but had a smaller impact on ICH and IS groups [HR (95%CI) was 1.37 (1.15-1.64) and 1.35 (1.06-1.71), respectively]. There was no significant difference in APACHE Ⅱ score on the risk of 14-day death between the patients aged > 60 years old and those aged ≤ 60 years old [HR (95%CI): 1.37 (1.08-1.72) vs. 1.35 (1.07-1.70), respectively]. Compared with patients with SOFA score > 3, APACHE Ⅱ score had a greater impact on the risk of 14-day death in patients with SOFA score ≤ 3 [HR (95%CI): 1.40 (1.16-1.70) vs. 1.34 (1.16-1.55)].
ConclusionTime-dependent APACHE Ⅱ score is an important indicator to evaluate the risk of death in patients with severe stroke.
Key words:
Endogenous time-dependent covariate; Acute physiology and chronic health evaluation Ⅱ; Severe stroke; Joint model
Contributor Information
Cheng Yi
Department of Military Health Statistics, Department of Health Services, Naval Medical University, Shanghai 200433, China
Cheng Yi is working on the Health Team, Support Department, 94969 Army of PLA, Shanghai 200131, China
Chen Qi
Department of Military Health Statistics, Department of Health Services, Naval Medical University, Shanghai 200433, China
Wu Shengyong
Department of Military Health Statistics, Department of Health Services, Naval Medical University, Shanghai 200433, China
Zhu Ronghui
Department of Military Health Statistics, Department of Health Services, Naval Medical University, Shanghai 200433, China
Wu Cheng
Department of Military Health Statistics, Department of Health Services, Naval Medical University, Shanghai 200433, China
