Preoperative nutritional status, frailty, and postoperative delirium in patients undergoing pancreatic cancer surgery: a prospective observational study.

Abstract

Postoperative delirium (POD) is a common neuropsychiatric complication after major abdominal surgery and is associated with adverse clinical outcomes. Patients undergoing pancreatic cancer surgery are particularly vulnerable to nutritional impairment and physiological decline, both of which may increase susceptibility to POD. Frailty reflects reduced physiological reserve and may be associated with both poor nutritional status and postoperative cognitive vulnerability. However, whether frailty helps explain the association between preoperative nutritional status and POD remains unclear. This study aimed to investigate the association between preoperative nutritional status and POD and to explore the role of frailty in this relationship. A prospective observational study was conducted among patients who underwent elective pancreatic cancer surgery at a tertiary hospital in Jiangsu Province, China, between August 2023 and March 2024. Preoperative nutritional status was assessed using the prognostic nutritional index (PNI), and frailty was evaluated using the FRAIL scale. POD was assessed twice daily from postoperative day 1 to postoperative day 4 using the Confusion Assessment Method (CAM). Multivariable logistic regression was performed to identify factors independently associated with POD. Mediation analysis was conducted using PROCESS macro version 5.0 (Model 4) with 5,000 bootstrap resamples to evaluate the potential mediating role of frailty in the association between PNI and POD. Receiver operating characteristic (ROC) curve analysis was carried out to assess the predictive performance of PNI, frailty score, and their combined model. A total of 293 patients were included, among whom 57 (19.5%) developed POD. In the multivariable logistic regression model, higher frailty score (OR = 2.451, 95% CI: 1.442-4.164, p = 0.001), older age (OR = 1.108, 95% CI: 1.038-1.182, p = 0.002), history of cerebral infarction (OR = 3.774, 95% CI: 1.003-14.195, p = 0.049), longer preoperative fasting time (OR = 2.440, 95% CI: 1.693-3.518, p < 0.001), and intraoperative hypotension (OR = 2.868, 95% CI: 1.155-7.122, p = 0.023) were independently associated with POD. PNI was significantly associated with frailty score (B = -0.0476, SE = 0.0097, p < 0.001). Mediation analysis showed a significant indirect effect of PNI on POD through frailty (effect = -0.0430, 95% bootstrap CI: -0.0866 to -0.0187), whereas the direct effect was not significant (B = -0.0039, SE = 0.0439, p = 0.9295), indicating an indirect-only mediation pattern. ROC analysis showed that the area under the curve (AUC) was 0.661 for PNI, 0.829 for frailty score, and 0.835 for the combined model. The AUC of frailty score was significantly higher than that of PNI (DeLong test, p < 0.001), whereas the combined model did not significantly improve discrimination compared with frailty score alone ( p = 0.473). Frailty was independently associated with postoperative delirium in patients undergoing pancreatic cancer surgery and serves as a potential pathway linking poor preoperative nutritional status to POD. Poorer nutritional status may be associated with POD susceptibility partly through frailty. Although adding PNI to frailty slightly increased the AUC, it did not provide statistically significant incremental discriminative value beyond frailty alone.