Development and Internal Validation of a Prognostic Model for Delayed Ambulation Recovery After Major Abdominal Surgery in Northwest Ethiopia. A Prospective Multicenter Study

Abstract Background Delayed ambulation recovery following major abdominal surgery presents a significant prognostic challenge, contributes to increased morbidity, mortality, and prolonged hospital stay. Despite recommendations advocating early mobilization, many patients, especially in low-resource settings, experience delayed recovery, and practical risk prediction models are lacking to identify high-risk patients to implement targeted interventions and personalized care. Objective To develop and internally validate a prognostic model predicting delayed ambulation recovery on postoperative day one after major abdominal surgery. Methods A prospective follow-up study was conducted on 645 patients who underwent major abdominal surgery between April to July 2025. Data were collected by Kobo Toolbox and analyzed using R software version 4.4.3. Least Absolute Shrinkage and Selection Operator (LASSO) regression and logistic regression were used for model development. Model performance was evaluated via Receiver Operating Characteristic-Area Under the Curve (ROC-AUC), calibration plots, and Hosmer–Lemeshow tests. Internal validation was done via bootstrapping. Clinical utility was assessed by decision curve analysis, and results were presented in a nomogram. Results Delayed ambulation recovery was observed in 48.4% (95%CI: 44.5–52.2) of patients. The model identified ten predictors (comorbidity, pain, NG tube, drain, general anesthesia, lower abdominal surgery, fluid amount > 2000 ml, blood loss > 500 ml, surgical duration > 2 h and open surgery). The model shows strong discrimination with an AUC of 0.85 (95% CI: 0.82–0.88) and calibration (Hosmer–Lemeshow p = 0.209). Internal validation revealed AUC of 0.84, with minimal overfitting. The model yielded sensitivity of 79.4%, specificity of 82.6%, and overall accuracy of 81.2%( 95% CI: 78%, 84%). The decision curve analysis offers greater net benefit than treating all or none across a wide range of risk thresholds probabilities. Conclusion A practical prognostic model was developed using easily accessible clinical and surgical predictors and displayed in a nomogram. The model demonstrated a good predictive ability for delayed ambulation recovery after major abdominal surgery. Applying the model helps clinicians with perioperative screening and targeted management of surgical patients in this population. We recommend that other researchers to externally validate the model for transportability.