Limits...
DELirium Prediction Based on Hospital Information (Delphi) in General Surgery Patients

View Article: PubMed Central - PubMed

ABSTRACT

To develop a simple and accurate delirium prediction score that would allow identification of individuals with a high probability of postoperative delirium on the basis of preoperative and immediate postoperative data.

Postoperative delirium, although transient, is associated with adverse outcomes after surgery. However, there has been no appropriate tool to predict postoperative delirium.

This was a prospective observational single-center study, which consisted of the development of the DELirium Prediction based on Hospital Information (Delphi) score (n = 561) and its validation (n = 533). We collected potential risk factors for postoperative delirium, which were identified by conducting a comprehensive review of the literatures.

Age, low physical activity, hearing impairment, heavy alcoholism, history of prior delirium, intensive care unit (ICU) admission, emergency surgery, open surgery, and increased preoperative C-reactive protein were identified as independent predictors of postoperative delirium. The Delphi score was generated using logistic regression coefficients. The maximum Delphi score was 15 and the optimal cut-off point identified with the Youden index was 6.5. Generated area under the (AUC) of the receiver operating characteristic (ROC) curve was 0.911 (95% CI: 0.88–0.94). In the validation study, the calculated AUC of the ROC curve based on the Delphi score was 0.938 (95% Cl: 0.91–0.97). We divided the validation cohort into the low-risk group (Delphi score 0–6) and high-risk group (7–15). Sensitivity of Delphi score was 80.8% and specificity 92.5%.

Our proposed Delphi score could help health-care provider to predict the development of delirium and make possible targeted intervention to prevent delirium in high-risk surgery patients.

No MeSH data available.


Related in: MedlinePlus

Receiver operating characteristic (ROC) curves and calculated area under the curves (AUC). (A) Development study of the Delphi score (dotted line, ROC curve of the logistic regression model; solid line, ROC curve of the Delphi score). (B) Validation study of the Delphi score. Delphi score is useful to distinguish patients with high risk of postoperative delirium from those with low risk. The cut-off value of the Delphi score corresponding to the optimal trade-off between sensitivity and specificity is 6.5. AUC = area under the curves; ROC = receiver operating characteristic.
© Copyright Policy - open-access
Related In: Results  -  Collection

License
getmorefigures.php?uid=PMC4998372&req=5

Figure 3: Receiver operating characteristic (ROC) curves and calculated area under the curves (AUC). (A) Development study of the Delphi score (dotted line, ROC curve of the logistic regression model; solid line, ROC curve of the Delphi score). (B) Validation study of the Delphi score. Delphi score is useful to distinguish patients with high risk of postoperative delirium from those with low risk. The cut-off value of the Delphi score corresponding to the optimal trade-off between sensitivity and specificity is 6.5. AUC = area under the curves; ROC = receiver operating characteristic.

Mentions: We generated a ROC curve with predicted probabilities from the logistic regression model; the AUC was 0.918 (95% CI: 0.89–0.95) (Figure 3A). The maximum score was 15 (Table 3). A new AUC was estimated based on the Delphi score to compare it with the original AUC. The new AUC was 0.911 (95% CI: 0.88–0.94), which was similar to the original AUC (Figure 3A). The optimal cut-off point to discriminate between high and low probability of postoperative delirium was 6.5. If the Delphi score was 7 or more, the patient was classified as having a high risk of postoperative delirium, and vice versa.


DELirium Prediction Based on Hospital Information (Delphi) in General Surgery Patients
Receiver operating characteristic (ROC) curves and calculated area under the curves (AUC). (A) Development study of the Delphi score (dotted line, ROC curve of the logistic regression model; solid line, ROC curve of the Delphi score). (B) Validation study of the Delphi score. Delphi score is useful to distinguish patients with high risk of postoperative delirium from those with low risk. The cut-off value of the Delphi score corresponding to the optimal trade-off between sensitivity and specificity is 6.5. AUC = area under the curves; ROC = receiver operating characteristic.
© Copyright Policy - open-access
Related In: Results  -  Collection

License
Show All Figures
getmorefigures.php?uid=PMC4998372&req=5

Figure 3: Receiver operating characteristic (ROC) curves and calculated area under the curves (AUC). (A) Development study of the Delphi score (dotted line, ROC curve of the logistic regression model; solid line, ROC curve of the Delphi score). (B) Validation study of the Delphi score. Delphi score is useful to distinguish patients with high risk of postoperative delirium from those with low risk. The cut-off value of the Delphi score corresponding to the optimal trade-off between sensitivity and specificity is 6.5. AUC = area under the curves; ROC = receiver operating characteristic.
Mentions: We generated a ROC curve with predicted probabilities from the logistic regression model; the AUC was 0.918 (95% CI: 0.89–0.95) (Figure 3A). The maximum score was 15 (Table 3). A new AUC was estimated based on the Delphi score to compare it with the original AUC. The new AUC was 0.911 (95% CI: 0.88–0.94), which was similar to the original AUC (Figure 3A). The optimal cut-off point to discriminate between high and low probability of postoperative delirium was 6.5. If the Delphi score was 7 or more, the patient was classified as having a high risk of postoperative delirium, and vice versa.

View Article: PubMed Central - PubMed

ABSTRACT

To develop a simple and accurate delirium prediction score that would allow identification of individuals with a high probability of postoperative delirium on the basis of preoperative and immediate postoperative data.

Postoperative delirium, although transient, is associated with adverse outcomes after surgery. However, there has been no appropriate tool to predict postoperative delirium.

This was a prospective observational single-center study, which consisted of the development of the DELirium Prediction based on Hospital Information (Delphi) score (n = 561) and its validation (n = 533). We collected potential risk factors for postoperative delirium, which were identified by conducting a comprehensive review of the literatures.

Age, low physical activity, hearing impairment, heavy alcoholism, history of prior delirium, intensive care unit (ICU) admission, emergency surgery, open surgery, and increased preoperative C-reactive protein were identified as independent predictors of postoperative delirium. The Delphi score was generated using logistic regression coefficients. The maximum Delphi score was 15 and the optimal cut-off point identified with the Youden index was 6.5. Generated area under the (AUC) of the receiver operating characteristic (ROC) curve was 0.911 (95% CI: 0.88–0.94). In the validation study, the calculated AUC of the ROC curve based on the Delphi score was 0.938 (95% Cl: 0.91–0.97). We divided the validation cohort into the low-risk group (Delphi score 0–6) and high-risk group (7–15). Sensitivity of Delphi score was 80.8% and specificity 92.5%.

Our proposed Delphi score could help health-care provider to predict the development of delirium and make possible targeted intervention to prevent delirium in high-risk surgery patients.

No MeSH data available.


Related in: MedlinePlus