Risk prediction models for postherpetic neuralgia: a systematic review and meta-analysis.

OBJECTIVE

This study conducted a systematic review and meta-analysis of risk prediction models for postherpetic neuralgia (PHN), aiming to provide a reference for Chinese scholars to develop higher-quality risk prediction models.

METHODS

This study systematically searched the China National Knowledge Infrastructure (CNKI), Wanfang Data Knowledge Service Platform, VIP Chinese Science and Technology Journal Database, Chinese Biomedical Literature Database (CBM), PubMed, Web of Science, Embase, and Cochrane Library for studies on risk prediction models for postherpetic neuralgia. The search period for all databases was from inception to March 1, 2026. Two researchers independently screened the literature and extracted information. The Prediction Model Risk of Bias Assessment Tool (PROBAST) was used to assess the risk of bias and applicability of the included studies. R 4.5.1 software was used to perform meta-analyses of the area under the curve (AUC) values and predictive factors of the models.

RESULTS

A total of 25 studies were ultimately included in this study, with sample sizes ranging from 90 to 8,878 cases and PHN incidence rates ranging from 6.2% to 52.9%. Among them, 18 studies performed internal validation and 4 studies performed external validation. The literature quality assessment results indicated high risk of bias and good applicability in all studies. The area under the receiver operating characteristic curve (AUC) of the models ranged from 0.71 to 0.98. Meta-analysis results showed that the pooled AUC was 0.86 (0.82-0.90), indicating good predictive performance. In addition, Age, VAS, rash site, Prodromal pain, and Extent of Rash were common predictive factors for the occurrence of postherpetic neuralgia.

CONCLUSION

Research on risk prediction models for postherpetic neuralgia is still at an early stage, with an overall high risk of bias and a lack of clinical application. In the future, scholars may develop high-quality risk prediction models with high accuracy and strong generalizability based on machine learning methods and multicenter, large-sample prospective studies.

SYSTEMATIC REVIEW REGISTRATION: https://www.crd.york.ac.uk/PROSPERO/view/CRD420261354649, Identifier CRD420261354649.