医学
比例危险模型
回顾性队列研究
内科学
危险系数
生存分析
四分位数
队列
纵隔气肿
病历
并发症
外科
置信区间
作者
Can Li,Mei’e Liang,Hui Jiang,Jiuliang Zhao,Chanyuan Wu,Qian Wang,Liyun Zhang,Yan Zhao
出处
期刊:Rheumatology
[Oxford University Press]
日期:2020-08-21
卷期号:60 (5): 2288-2295
被引量:10
标识
DOI:10.1093/rheumatology/keaa582
摘要
Pneumomediastinum (PnM) is a rare but life-threatening complication of DM. The present study aims to characterize the long-term prognosis and prognostic factors of DM-associated PnM. Inpatients with DM-associated PnM were retrospectively enrolled from two tertiary referral centres for rheumatic disease. The enrolled patients were divided into survivors or non-survivors. Information about the demographics, clinical manifestations, CT scan features, laboratory findings and outcomes were collected from their medical records. A least absolute shrinkage and selection operator regularized Cox regression model was used to select the most relevant factors. Prognosis was analysed using a Kaplan–Meier curve. A Cox proportional hazards model was used to identify independent predictive factors for long-term survival. A total of 62 patients (26 women) with DM-associated PnM were enrolled. The mean age was 44.3 years (s.d. 11.7). The median follow-up duration was 17 days (quartiles 7, 266.5). Thirty-five patients died during follow-up. The survival rates were 75.4% at 1 week, 46.2% at 3 months and 41.9% at 1 year. The Cox proportional hazards model identified the development of fever [hazard ratio (HR) 3.23 (95% CI 1.25, 8.35), P = 0.02] and a decrease in the number of lymphocytes [HR 2.19 (95% CI 1.10, 4.39), P = 0.03] as independent risk factors for death. The results suggest poor overall survival among patients with DM-associated PnM. Survival during the first 3 months is crucial for long-term survival. Meanwhile, the development of fever and a decrease in the number of lymphocytes were associated with long-term mortality. Early recognition and prompt treatment of this high-risk group of DM patients is therefore important.
科研通智能强力驱动
Strongly Powered by AbleSci AI