Construction and validation of a predictive model for postoperative psychological distress in oral cancer patients

Technical Summary: Construction and Validation of a Predictive Model for Postoperative Psychological Distress in Oral Cancer Patients

Problem Statement
Oral cancer, a common head and neck malignancy, imposes significant psychological burdens on patients due to the diagnosis itself, potential facial disfigurement, and functional impairments (dysphagia, speech disorders) following treatment. While psychological distress is recognized as the "sixth vital sign" in oncology, existing literature predominantly focuses on cross-sectional surveys and factor analysis. There is a notable absence of validated predictive models specifically for postoperative psychological distress in oral cancer patients, unlike models available for breast, lung, and gastric cancers. This gap hinders the ability of healthcare professionals to proactively identify high-risk patients and implement timely interventions.

Methodology
The study employed a convenience sampling method to recruit 351 patients with pathologically confirmed oral cancer from a tertiary hospital in Tianjin, China, between July 2024 and January 2026.

• Data Collection: Participants completed a General Information Questionnaire and the Distress Management Screening Measure (DMSM), which includes the Distress Thermometer (DT) and a 40-item Problem List (PL). Data were collected on postoperative day 7.
• Statistical Analysis:
  • Variable Screening: Univariate logistic regression was used to screen potential predictors (P ≤ 0.1).
  • Model Construction: Significant variables were entered into a stepwise multivariate logistic regression analysis (P ≤ 0.05) to identify independent risk factors.
  • Visualization: A nomogram was constructed using R 4.4.1 to visually represent the predictive model.
  • Validation and Evaluation: The model's performance was assessed using the Receiver Operating Characteristic (ROC) curve, the Hosmer-Lemeshow test for goodness-of-fit, and Bootstrap resampling (1,000 iterations) for internal validation. Clinical utility was evaluated via Decision Curve Analysis (DCA).

Key Contributions and Results
The study successfully identified six independent risk factors for postoperative psychological distress and constructed a predictive tool based on them:

1. Identified Risk Factors:

   • Gender: Female patients showed a higher risk (OR = 3.664).
   • Age: Younger age was associated with higher risk (OR = 0.960 per year increase).
   • Tracheostomy Status: Presence of a tracheostomy significantly increased risk (OR = 3.531).
   • Emotional State: The presence of sadness was a strong predictor (OR = 5.913).
   • Sleep: Sleep disturbances increased risk (OR = 7.654).
   • Appearance: Concerns regarding body image/appearance were significant (OR = 3.800).

2. Model Performance:

   • Discrimination: The Area Under the Curve (AUC) was 0.818 (95% CI: 0.775–0.862), indicating moderate-to-good predictive performance.
   • Optimal Cutoff: The optimal cutoff value for the predicted probability was 0.571, yielding a sensitivity of 0.710 and specificity of 0.795.
   • Calibration: The calibration curve demonstrated a slope close to 1, indicating strong agreement between predicted and observed risks. The Hosmer-Lemeshow test confirmed good model fit (P = 0.118).
   • Clinical Utility: The Decision Curve Analysis (DCA) indicated that the model provides a net clinical benefit over treating all or no patients across a range of threshold probabilities.

Significance and Claims
The authors claim that the constructed nomogram model offers good predictive performance and clinical utility. Its primary significance lies in assisting healthcare professionals to:

• Rapidly Identify High-Risk Patients: By inputting the six specific variables, clinicians can calculate a patient's probability of experiencing postoperative psychological distress.
• Facilitate Early Intervention: Patients with a predicted probability ≥ 0.571 can be flagged for timely, targeted psychological interventions and preventive management measures.
• Address a Clinical Gap: The study provides the first reported predictive model specifically for postoperative psychological distress in oral cancer patients, moving beyond general factor analysis to a practical clinical tool.

The authors maintain a modest tone regarding the study's limitations, acknowledging that it is a single-center investigation with a relatively small sample size and that the model has only undergone internal validation. They explicitly state that future multicenter studies are required to perform external validation to confirm the model's generalizability.