Abstract
BACKGROUND: The trend of risk prediction models for diabetic peripheral neuropathy (DPN) is increasing, but few studies focus on the quality of the model and its practical application. AIM: To conduct a comprehensive systematic review and rigorous evaluation of prediction models for DPN. METHODS: A meticulous search was conducted in PubMed, EMBASE, Cochrane, CNKI, Wang Fang DATA, and VIP Database to identify studies published until October 2023. The included and excluded criteria were applied by the researchers to screen the literature. Two investigators independently extracted data and assessed the quality using a data extraction form and a bias risk assessment tool. Disagreements were resolved through consultation with a third investigator. Data from the included studies were extracted utilizing the Checklist for Critical Appraisal and Data Extraction for Systematic Reviews of Prediction Modelling Studies. Additionally, the bias risk and applicability of the models were evaluated by the Prediction Model Risk of Bias Assessment Tool. RESULTS: The systematic review included 14 studies with a total of 26 models. The area under the receiver operating characteristic curve of the 26 models was 0.629-0.938. All studies had high risks of bias, mainly due to participants, outcomes, and analysis. The most common predictors included glycated hemoglobin, age, duration of diabetes, lipid abnormalities, and fasting blood glucose. CONCLUSION: The predictor model presented good differentiation, calibration, but there were significant methodological flaws and high risk of bias. Future studies should focus on improving the study design and study report, updating the model and verifying its adaptability and feasibility in clinical practice.