To construct a risk factor model of chemotherapy-induced nausea and vomiting (CINV) in patients with pancreatic cancer and verify its application effect.

A retrospective collection was made of 138 patients with pancreatic cancer admitted to First Affiliated Hospital of Air Force Medical University of the People's Liberation Army from February 2022 to February 2024. The patients were divided into the CINV group (n=57) and the non-CINV group (n=81) based on whether CINV occurred, the differences in clinical data between the two groups were analyzed, and statistically significant variables were screened through univariate analysis, it was further verified through multivariate Logistic regression. The patients were divided into a training group (n=104) and a validation group (n=34) at a 3∶1 ratio. A nomogram prediction model was constructed, and the receiver operating characteristic (ROC) curve and decision curve analysis (DCA) were used to verify the predictive efficacy and calibration degree of the model.

The terms of age, history of motion sickness, history of nausea and vomiting during pregnancy, sleep status (Pittsburgh Sleep Quality Index, PSQI score), anxiety level (Self-Rating Anxiety Scale, SAS score), number of chemotherapy cycles, gastrointestinal diseases, psychological anticipation of CINV, hypokalemia, hypomagnesemia, and malnutrition between the CINV group and the non-CINV group were significantly different (P<0.05). Logistic multivariate analysis showed that age 18-50 years old, previous motion sickness, history of nausea and vomiting during pregnancy, PSQI score ≥8 points, SAS score ≥50 points, chemotherapy frequency <3 times, combined gastrointestinal diseases, psychological anticipation of CINV, hypokalemia, hypomagnesemia, and malnutrition were all independent risk factors affecting the occurrence of CINV in pancreatic cancer (P<0.05). The areas under the curve of the model applied in the training group and the verification group were 0.843 and 0.838 respectively, and the sensitivity and specificity applied in the verification group were 82.70% and 83.19% respectively, and DCA analysis showed that the clinical net rate of return of the model was relatively high. Through calibration curve analysis and Hosmer-Lemeshow test (P=0.372) verification, the predicted probability of the model was highly consistent with the actual observed probability, suggesting a good calibration degree.

Patients with pancreatic cancer have a relatively high risk of CINV occurrence, and it is related to factors such as age, previous motion sickness, history of nausea and vomiting during pregnancy, PSQI score, SAS score, number of chemotherapy sessions, gastrointestinal diseases, psychological anticipation of CINV occurrence, hypokalemia, hypomagnesemia, and malnutrition, and the prediction model established based on the above factors can provide a reliable reference for the risk assessment of reducing the occurrence of CINV in patients with pancreatic cancer.