Vasculitis & polymyalgia rheumatica
Comparison of four clinical risk prediction models for the diagnosis of giant cell arteritis: results from the ongoing PREDICT-GCA study
Authors' conclusion
Clinical prediction models for the diagnosis of GCA differ in their diagnostic accuracy. Clinically relevant discordances of individual risk prediction are common.
Study context
- Population
- Patients enrolled in the ongoing prospective PREDICT giant cell arteritis (GCA) study, all undergoing temporal artery biopsy plus detailed clinical and sonographic workupanalysed
- Participants
- 70 patients
- Comparison
- Four clinical prediction rules for GCA: Ing et al model, Southend GCA Probability Score, Bhavsar-Khalidi Score, PREDICT score
- Outcome
- Diagnostic accuracy (2×2 contingency tables and ROC analysis) vs final diagnosis based on 6-month follow-up (ambiguous cases judged by independent expert panel)
- Follow-up
- 6 months
Methodological detailsprospective, ongoing
- Prospective
- Ongoing
Performance
As reported in the abstract
Diagnostic accuracy (AUC, NPV)
| Test or model | AUC | NPV |
|---|---|---|
| Ing et al model | 0.84 | 86.7% |
| Southend GCA Probability Score | 0.74 | 78.6% |
| Bhavsar-Khalidi Score | 0.75 | 73.7% |
| PREDICT score | 0.70 | 77.8% |
AUC
Ing et al model0.84
Southend GCA Probability Score0.74
Bhavsar-Khalidi Score0.75
PREDICT score0.70
NPV
Ing et al model86.7%
Southend GCA Probability Score78.6%
Bhavsar-Khalidi Score73.7%
PREDICT score77.8%
Further findings
Clinically significant discordance in risk categorisation (low vs non-low risk by different scores)
- Patients with discordant categorisation
- 37.5%
- Discordant patients less frequently had clinical symptoms and signs of cranial GCA (all p<0.01)
- Rate of visual impairment did not differ from patients with consistent risk estimation
Original publication
Comparison of four clinical risk prediction models for the diagnosis of giant cell arteritis: results from the ongoing PREDICT-GCA study. RMD Open. Published 25 September 2026. doi:10.
Composed by AI from the published abstract and checked against it. How Briefs are made →