심장수술 환자의 섬망예측을 위한 PRE-Deliric 모델과 E-PRE-Deliric 모델의 타당도 비교

Abstract

Purpose The purpose of this study was to identify characteristics related to the incidence of delirium after cardiac surgery and to analyze the validity of the PRE-DELIRIC model and E-PRE-DELIRIC model.

Method 713 patients aged 18 years or older who were admitted to the cardiac surgery. Data were collected through electronic medical records. Using CAM-ICU, we divided the patients into delirium and non-delirium groups and analyzed their characteristics. Continuous variables were analyzed using t-test, and categorical variables were analyzed using x2-test. The presence or absence of delirium was confirmed by dividing the risk group by the model’s score, and sensitivity, specificity, positive prediction, and negative prediction were verified using the Youden index. The ROC curves of the PRE-DELIRIC model and E-PRE-DELIRIC model were drawn to calculate the area under the curves, and the 95% confidence interval (CI) was presented.

Result There were 205 delirium patients and significant variables were gender (p=.005), age (p=<.001), hospital stay (p<.001), postoperative hospital stay (p=<.001), intensive care unit stay (p=<.001), hypertension (p=<.001), diabetes (p=<.001), alcoholism (p=<.001), kidney failure (p=.010) and BUN (p=.001). The average scores of the PRE-DELIRIC model and E-PRE-DELIRIC Model were 49.72±29.91% and 18.14±0.41%, respectively. In the PRE-DELIRIC model, which is the cutoff point of the Youden index, sensitivity, specificity, positive prediction, negative prediction were 64.88%, 83.27%, 61.01%, 85.45% (cut-off point of 14.05%). In the E-PRE-DELIRIC model, corresponding indicators were 73.17%, 51.57%, 37.88%, 82.65% (cut-off point of 64.01%), respectively. The areas under the curve for the PRE-DELIRIC model and the E-PRE-DELIRIC model were 0.77 (95% CI 0.74~0.82) and 0.65 (95% CI 0.61~0.69), respectively

Conclusion Although the two models showed a good and a poor level of validities, the PRE-DELIRIC model was more valid than the E-PRE-DELIRIC model. The E-PRE-DELIRIC model can easily and quickly detect delirium risk groups, however, it has limitations of not being able to identify delirium risk groups. Therefore, it may be helpful to mix E-PRE-DELIRIC with the PRE-DELIRIC model. The two prediction models are expected to help improve the quality of nursing for early detection and prevention of delirium risk groups so that intensive observation and intervention can be induced for patients at high risk of delirium.