Desarrollo y validación interna de un modelo predictivo clínico simplificado para complicaciones postoperatorias mayores en cirugía oncológica: Análisis retrospectivo

Introduction. Reliable preoperative risk assessment is crucial for anticipating adverse surgical outcomes and optimizing perioperative planning. However, many scoring systems are difficult to apply in resource-limited settings due to their complexity and technical requirements. Methods. This was an observational retrospective cohort study involving predictive modeling of consecutive adult patients undergoing elective or urgent oncological surgery at a tertiary care center. Major postoperative complications were defined as Clavien–Dindo grade ≥ III. Predictors were selected a priori based on clinical relevance and availability. Absolute and relative risks, along with 95% confidence intervals (95% CI), were estimated for dichotomous variables. A multivariable logistic regression model was developed to identify independent predictors. Model performance was assessed via discrimination (AUC) and calibration. Internal validation was performed using bootstrapping (1,000 iterations). A simplified clinical score was derived from the final coefficients. Results. A total of 120 patients were included. The incidence of major complications was 17.5% (21/120). Four variables were independently associated with increased risk: preoperative hypoalbuminemia, higher ASA classification, urgent surgery, and prolonged operative time. The model showed good discrimination (AUC 0.82; 95% CI 0.73–0.91) and adequate calibration. The score stratified patients into three risk categories (4%, 18%, and 55%). Conclusions. The simplified clinical score demonstrated satisfactory internal performance and could serve as a practical tool for estimating perioperative risk in oncological surgery, particularly in resource-limited settings.