Development of a Prediction Model for Skin Graft Failure Following Skin Cancer Excision: The GRAFT Score

Abstract Background Skin grafting after skin cancer excision is a commonly performed procedure. Preoperative identification and optimization of patients at risk of graft failure can improve outcomes and healthcare cost effectiveness. This study aimed to develop and internally validate a risk assessment tool, the skin Graft Risk Assessment of Failure Tool (GRAFT) to aid risk stratification of patients undergoing skin grafting after cancer excision. Methods A single-center retrospective cohort of 162 patients who underwent skin grafting after skin cancer excision was included. Relevant variables were assessed using univariate logistic regression. Variables demonstrating significant association were entered into multivariable modeling. Model discrimination and calibration were respectively assessed using the area under the receiver operating characteristic curve (AUC) and Hosmer–Lemeshow goodness-of-fit test. Bootstrapping was used for internal validation. Results Anticoagulation and cancer location demonstrated significant association with graft failure in the univariate analysis. In multivariable analysis, anticoagulation (odds ratio OR 6.40; 95% confidence interval CI 2.25–18.24) and lower body graft location (OR 3.19; 95% CI 1.21–8.45) remained independently associated with graft failure. The final model demonstrated acceptable discrimination (AUC 0.72), and the Hosmer–Lemeshow test showed no evidence of lack of fit ( p = 0.87), with stable performance on bootstrap validation. The GRAFT score stratified patients into low-, intermediate-, and high-risk groups with observed failure rates of 7.5%, 26.0%, and 60.0%, respectively. Conclusion The GRAFT score is an internally validated, simple, practical, bedside tool that estimates risks of skin graft failure after skin cancer excision, supporting clinical decision-making and patient counseling. External validation of this model is warranted.