Development and validation of an incidence risk prediction model for early foot ulcer in diabetes based on a high evidence systematic review and meta-analysis. (October 2021)

• Record Type: Journal Article
• Authors: Chen, Dong; Wang, Meijun; Shang, Xin; Liu, Xixi; Liu, Xinbang; Ge, Tiantian; Ren, Qiuyue; Ren, Xiaoxia; Song, Xin; Xu, Hongmei; Sun, Mingyan; Zhou, Hongmei; Chang, Bai
• Is Part Of: Diabetes research and clinical practice. Volume 180(2021)
• Publication Date: 2021-10
• Subjects: Diabetic foot ulcer -- Clinical prediction model -- Meta-analysis -- Cohort study; Diabetes -- Periodicals; Diabetes Mellitus -- Periodicals; 616.462
• DOI: 10.1016/j.diabres.2021.109040
• Languages: English
• ISSNs: 0168-8227

Abstract

Abstract: Objectives: To develop and validate a model for predicting the risk of early diabetic foot ulcer (DFU) based on systematic review and meta -analysis. Methods: Data were analyzed from the risk factors of DFU with their corresponding risk ratio (RR) by meta -analysis. The DFU prediction model included statistically significant risk factors from the meta -analysis, all of which were scored by its weightings, and the prediction model was externally validated using a validation cohort from China. The occurrence of early DFU was defined as patients with type 2 diabetes who were free of DFU at baseline and diagnosed with DFU at follow-up. Evaluation of model performance was based on the area under the discrimination receiver operating characteristic curve (ROC), with optimal cutoff point determined by calculation of sensitivity and specificity. Kaplan–Meier curve were performed to compare the cumulative risk of different groups. Results: Our meta -analysis confirmed a cumulative incidence of approximately 6.0% in 46, 521 patients with diabetes. The final risk prediction model included Sex, BMI, HbA1c, Smoker, DN, DR, DPN, Intermittent Claudication, Foot care, and their RRs were 1.87, 1.08, 1.21, 1.77, 2.97, 2.98, 2.76, 3.77, 0.38, respectively. The total score of all risk factors was 80 points according to their weightings. The prediction model showed good discrimination with AUC = 0.798 (95 %CI 0.738–0.858). At the optimal cut-off value of 46.5 points, the sensitivity, specificity and Youden index were 0.769, 0.798 and 0.567, respectively. The final model stratified the validation cohort into low, low-intermediate, high-intermediate and high-risk groups; Compared with low-risk group, the RR with 95 %CI of developing DFU in high-intermediate and high-risk group were 17.23 (5.12–58.02), p < 0.01 and 46.11 (5.16–91.74), p < 0.01, respectively. Conclusion: We have developed a simple tool to facilitates early identification of patients with diabetes at high risk of developing DFU based on scores. This simple tool may improve clinical decision-making and potentially guide early intervention.