Development and validation of a predictive nomogram for surgical site infection among general surgery patients in Amhara region Ethiopia

Abstract

Surgical site infections (SSIs) remain a significant cause of morbidity, prolonged hospital stays, and increased healthcare costs, particularly in low-resource settings such as Ethiopia. While SSIs are widely recognized as preventable, the burden of these infections remains high, especially in sub-Saharan Africa, where healthcare resources are limited, and surgical care may not consistently meet recommended standards. In Ethiopia, as in many similar settings, the lack of robust, context-specific predictive tools limits the ability of healthcare providers to proactively manage SSI risks. Current predictive models and nomograms for SSI risk are generally developed in high-resource settings and may not accurately capture the unique risk factors in Ethiopia. The aim of this study was Development and validation of nomogram for Surgical Site Infection Prediction Among General Surgery Patients in Amhara. A prospective follow-up study was conducted involving general surgery patients at referral hospitals in the Amhara region. Predictors of SSIs were identified through logistic regression analysis, and a nomogram was constructed based on these predictors. The model was internally validated using bootstrapping techniques to assess the accuracy and reliability of the risk estimates. Model performance was evaluated in terms of discrimination, measured by the area under the receiver operating characteristic (ROC) curve (AUC), and calibration, using calibration plots. The incidence of SSI was 39.6%. The key prognostic predictors of this model were: sex, age, diabetes mellitus (DM), wound classification, wound care, American Society of Anesthesiologists (ASA) score, residence, surgery duration, preoperative hospital stays, alcohol consumption, and prior surgical history. The model's discrimination power was 90.1% with 95% CI (87-93%) and its calibration is well fitted with 45 degrees. The bootstrapped model produced consistent β coefficients, supporting the stability and robustness of the model. The nomogram was developed with key predictors of SSI and demonstrated excellent discrimination ability, with an AUC of 0.87 (95% CI: 0.84-0.91). Calibration plots showed a strong agreement between predicted and observed probabilities, indicating the model is well-calibrated. The incidence of SSI was notably high. American Society of Anesthesiologists (ASA) score, sex, age, diabetes mellitus (DM), wound classification, wound care practices, patient residence, surgery duration, preoperative hospital stay, alcohol use, and history of previous surgeries were key prognostic predictors. The validated model had an excellent discrimination power with well fitted calibration. The developed nomogram accurately predicts the risk of SSIs among general surgery patients. It might serve as a practical tool for identifying high-risk patients, enabling healthcare providers to implement targeted preventive measures, improving patient outcomes and reducing the burden of SSIs in Ethiopian healthcare settings. Further external validation is recommended to confirm the model's applicability across different settings.

Keywords: Amhara region; Ethiopia; General surgery; Nomogram; Predictive model; Surgical site infection.

Fig. 1

Receiver operating characteristics curve for prediction model of the original model.

Fig. 2

Calibration belt for the original model: predicted probability versus observed probability of SSI.

Fig. 3

Prediction density plot for surgical site infection risk: probability threshold distribution (n = 441).

Fig. 4

Calibration plot of the validated model: predicted probability versus actual probability of SSI.

Fig. 5

A plot of decision curve showing the net benefit of the model against threshold probability and corresponding cost-benefit ratio of SSI.

Fig. 6

Nomogram for predicting the probability of developing surgical site infections (SSIs).