An original model to predict Intensive Care Unit length-of stay after cardiac surgery in a competing risk framework

详细信息    查看全文

• 作者：Fabio Barili ; Nicoletta Barzaghi ; Faisal H. Cheema ; Antonio Capo ; Jeffrey Jiang ; Enrico Ardemagni ; Michael Argenziano ; Claudio Grossi
• 关键词：Risk factors ; Intensive care unit ; Cardiac surgery ; Length of stay
• 刊名：International Journal of Cardiology
• 出版年：2013
• 出版时间：20 September, 2013
• 年：2013
• 卷：168
• 期：1
• 页码：219-225
• 全文大小：612 K

文摘

Background

The aim of the study is to design a specific Intensive Care Unit length-of-stay risk model based on the preoperative factors and surgeries utilizing modeling strategies for time-to-event data in a prospective observational clinical study.

Methods

From January 2004 to April 2011 data on 3861 consecutive heart surgery patients were prospectively collected. ICU length of stay was analyzed as a time-to-event variable in a competing risk framework with death as competing risk.

Results

The median ICU-LOS was one day. All factors considered but gender was included in the multivariable modeling. In the final model, factors that mostly affected time-to-discharge from ICU were critical preoperative state (Relative Risk 0.41; 95 % Confidence Interval: 0.29-0.58), emergency (0.41; 0.32-0.53), poor left ventricular dysfunction (0.50; 0.44-0.57) and serum creatinine > 200 ¦Ìmol/L (0.54; 0.46-0.65). Most of the predictors had a time-dependent effect that decreased in the first fifteen days and was constant thereafter. After the plateau, the risk profile was changed as most of the factors were no longer significant, Conversely, the time-to-ICU death model included only two variables, critical perioperative state and serum creatinine > 200 ¦Ìmol/L, with a constant RR of 9.1 and 3.37 respectively.

Conclusions

ICU-LOS can be predicted by preoperative data and type of surgeries. The derived ICU-LOS prediction model is dynamic and most predictors have an effect that decreases with time. The algorithm can preoperatively predict ICU-LOS curves and could have a major role in the decision making-behavior of clinicians, resources' allocation and maximization of care for high-risk patients.