Amputation Risk Prediction Tools for Patients With Diabetic Foot: A Systematic Review

OBJECTIVE: To systematically evaluate the research on amputation risk prediction models for patients with diabetic foot. DATA SOURCES: PubMed, Embase, Web of Science, China National Knowledge Infrastructure (CNKI), Wanfang Database, China Science and Technology Journal Database (VIP), and SinoMed were searched for studies on risk prediction tools for amputation in patients with diabetic foot. STUDY SELECTION: Two researchers screened 2134 articles, and 9 met the inclusion criteria. DATA EXTRACTION: Included study basic information, model construction methods, predictive factors, and model efficacy indicators (AUC, sensitivity, specificity). DATA SYNTHESIS: The 9 studies included in the review described the construction of 15 prediction models and one prediction tool. Eight of the 9 studies had a high overall risk of bias; 2 had poor applicability in the field of predictive factors, and 6 had good applicability in all fields and overall. The main predictive factors included in the models were diabetes duration, glycated hemoglobin, white blood cell count, fibrinogen, and infection. The most common predictive factors were duration of diabetes (odds ratio=2.79; 95% CI: 1.65-3.93), white blood cell count (odds ratio=1.88; 95% CI: 0.78-2.98), and fibrinogen (odds ratio=0.10; 95% CI: 0.06-0.14). CONCLUSIONS: The predictive performance of current amputation risk prediction tools for patients with diabetic foot is good, but the literature has a high risk of bias and needs improved clinical applicability. Researchers should further validate and calibrate the existing tools or develop risk prediction tools with low bias risk and high clinical applicability based on local data.