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

Area of Science:

• Dermatology
• Surgical Oncology
• Biostatistics

Background:

• Skin grafting is a common procedure following skin cancer excision.
• Predicting graft failure preoperatively can optimize patient care and reduce costs.
• A validated risk assessment tool is needed for stratifying patients undergoing skin grafting.

Purpose of the Study:

• To develop and internally validate the skin Graft Risk Assessment of Failure Tool (GRAFT).
• To aid in the risk stratification of patients undergoing skin grafting after skin cancer excision.

Main Methods:

• Retrospective cohort study of 162 patients undergoing skin grafting post-excision.
• Univariate and multivariable logistic regression analysis to identify predictors of graft failure.
• Internal validation using bootstrapping, assessing model discrimination (AUC) and calibration (Hosmer-Lemeshow test).

Main Results:

• Anticoagulation (OR 6.40) and lower body graft location (OR 3.19) were independently associated with graft failure.
• The GRAFT score demonstrated acceptable discrimination (AUC 0.72) and good calibration.
• The tool stratified patients into low (7.5%), intermediate (26.0%), and high (60.0%) risk groups for graft failure.

Conclusions:

• The GRAFT score is a validated, practical bedside tool for estimating skin graft failure risk.
• It supports clinical decision-making and patient counseling after skin cancer excision.
• External validation of the GRAFT score is recommended.