Development and internal validation of a multivariable risk stratification model for preoperative anxiety in surgical...

Preoperative anxiety is common among surgical patients and is associated with adverse perioperative outcomes. Early identification of patients at elevated risk of clinically significant preoperative anxiety may facilitate timely psychological screening, support, and targeted interventions. This study aimed to identify factors associated with preoperative anxiety and to develop and internally validate a multivariable risk stratification model for individualized preoperative anxiety assessment.

This retrospective observational study included all consecutively eligible adult patients who underwent elective surgery at a tertiary teaching hospital in Southwest China between January 2021 and October 2024 and had a routinely documented preoperative psychological assessment in the medical record. Preoperative anxiety was retrospectively determined based on standardized cut-off values of validated anxiety assessment instruments documented in preoperative records. Demographic, clinical, perioperative, psychological, behavioral, and social variables were extracted from electronic medical records and routine preoperative assessments. Univariate and multivariable logistic regression analyses were performed to identify independent predictors of preoperative anxiety. Model discrimination was evaluated using receiver operating characteristic (ROC) curve analysis, calibration was assessed using calibration plots with bias correction, and a nomogram was constructed based on the final multivariable model.

Among 425 eligible patients, 168 (39.5%) met the criteria for preoperative anxiety. Multivariable analysis identified female sex, lower body mass index, higher ASA physical status (III-IV), longer expected operative time, poorer sleep quality, higher depressive symptom scores, night-time smartphone use for at least 1 h, longer daily smartphone use duration, and higher trait anxiety scores as independent risk factors for preoperative anxiety, whereas greater social support was independently protective. The prediction model demonstrated good discriminative performance, with an area under the ROC curve of 0.848 (95% confidence interval, 0.811-0.884). At the optimal cut-off value, the model achieved a sensitivity of 0.73, a specificity of 0.82, and an overall accuracy of 0.76. Calibration analysis showed good agreement between predicted and observed risks. A nomogram was developed to facilitate individualized risk prediction.

This multivariable risk stratification model showed good discrimination and calibration for identifying surgical patients at risk of preoperative anxiety. The nomogram provides a practical tool for individualized preoperative risk stratification and may support early psychological screening and targeted perioperative support within similar elective surgical settings. External validation in independent multicenter cohorts is warranted before broader implementation.