Risk factors and nomogram for postoperative delirium in elderly orthopedic patients in the Intensive Care Unit: Retrospective study based on MIMIC–IV

Abstract

Background

This study aimed to identify predictors of postoperative delirium (POD) in older adult orthopedic ICU patients and develop a nomogram for predicting POD.

Methods

Patient data were collected from the Medical Information Mart for Intensive Care (MIMIC–IV). Independent predictors for POD were identified using least absolute shrinkage and selection operator and logistic regression. The nomogram was developed using selected predictors and assessed by the bootstrap method. Area under the receiver operating characteristic curves (AUC) and calibration curve were used to determine the discrimination performance and calibration ability. Clinical usefulness was assessed using decision curve analysis (DCA) and clinical impact curve (CIC).

Results

A total of 430 older adult orthopedic patients were screened out, with an incidence of POD of 36.2 %. Four independent predictors were identified, including Braden skin score (OR = 0.886; 95 % CI: 0.800–0.982; p = 0.021), mechanical ventilation (OR = 1.883; 95 % CI: 1.179–3.008; p = 0.008), cerebrovascular disease (OR = 2.113; 95 % CI: 1.165–3.831; p = 0.014) and acute physiology score III (APS III) (OR = 1.029; 95 % CI: 1.017–1.042; p < 0.001). The AUC of our nomogram was 0.724 (95 % CI: 0.674–0.774), and internal validation revealed an AUC of 0.713, which suggested moderate predictive performance. The DCA and CIC analysis suggested good clinical utility.

Conclusion

Braden skin score, mechanical ventilation, cerebrovascular disease and APS III are significant predictors of POD in older adult orthopedic patients in the ICU, and the nomogram would be helpful in supporting nursing management of POD.