Multivariable Models to Predict a Diagnosis of Giant Cell Arteritis: Systematic Review and Metaanalysis.

Area of Science:

• Rheumatology
• Clinical Epidemiology
• Diagnostic Accuracy Studies

Background:

• Giant cell arteritis (GCA) is a systemic vasculitis requiring timely diagnosis.
• Numerous multivariable models have been developed to aid clinicians in diagnosing GCA.
• The performance and variables of these predictive models require systematic evaluation.

Purpose of the Study:

• To conduct a systematic review and meta-analysis of variables used in GCA diagnostic models.
• To assess the performance and risk of bias of existing GCA predictive models.

Main Methods:

• Systematic search of PubMed, Embase, and Cochrane Library (1990-2024).
• Inclusion of studies using multivariable models for GCA diagnosis.
• Meta-analysis of individual signs/symptoms and assessment of predictor certainty using GRADE.
• Risk of bias assessment using the Prediction Model Risk of Bias Assessment Tool (PROBAST).

Main Results:

• 44 studies including 15,409 patients and 4340 GCA diagnoses were analyzed.
• Predictors with high certainty and large effect size: jaw claudication, elevated C-reactive protein (>24.5 mg/L), thrombocytosis (>400 × 10^9/L), positive temporal artery ultrasound, and synovitis (predictive of non-GCA).
• Classical GCA symptoms (vision loss, polymyalgia rheumatica, headache) had lower certainty of effect.
• Most models exhibited a high risk of bias.

Conclusions:

• Key predictors for GCA diagnosis are consistent across models.
• Existing GCA predictive models are generally of poor methodological quality.
• Future development of GCA diagnostic models requires enhanced methodological rigor.