Background/Purpose:
The Diagnostic and Classification Criteria for Vasculitis (DCVAS) Study is a multinational observational study to develop diagnostic criteria and to update classification criteria for the primary systemic vasculitides.
By 2015 the database will include clinical, laboratory and radiology data from over 2000 patients with vasculitis and 1500 comparator patients who present with features similar to vasculitis. To avoid the inherent circularity of using the submitting physician diagnosis as the gold standard, a reference diagnosis for each patient will be established using a combination of expert panel opinion and data-driven methods (e.g. machine learning algorithms).
The aim of this analysis was to evaluate the methodology by which the expert panel will assess individual patient data to establish the reference diagnosis.
Methods:
By November 2012, 1662 patients had been recruited; 391 had complete 6 month follow-up data. Forty cases were randomly extracted and developed into clinical vignettes (CVs). The CVs were assessed for diagnoses by 6 independent experts using an online platform. Ten patients were assessed by all experts and the other 30 were each assessed by 2 of the 6 experts, randomly chosen. The experts first chose between primary vasculitis, secondary vasculitis, or other illness; then the respective major class (small-, medium-, or large-vessel vasculitis, or no predominant size vasculitis); and then the subtype or the specific disease for each category. For each answer a level of certainty (unlikely, possible, probable, definitive, or unknown) was provided. The diagnoses of the expert panel and the submitting physician were compared.
Results:
The 40 clinical vignettes represented 26 women and 14 men, with a mean age of 62.5 ± 20.3 years (range 20-86 years). Data from all 120 CV reviews were available for analysis. Treating clinicians submitted a diagnosis of primary vasculitis in 32 patients (17 small-vessel, 1 medium-vessel, 13 large-vessel and 1 with no predominant size); secondary vasculitis in 2, and other illness in 6. The expert panel agreed with the submitted diagnosis of primary vasculitis in 97% of the cases (definite 54%, probable 35% and possible 8%). However, only 78% of the submitted patients with primary vasculitis were classified as having the same sub-type of vasculitis when compared with the expert panel diagnosis (9% could not be sub-typed within the correct major class of vasculitis, 9% classified the cases with another sub-type of the same major class, 3% chose another major class or diagnosis and 2% selected the unknown option). There was an intraclass correlation coefficient of 0.82 (confidence interval 0.57-0.95) in the 10 CVs assessed by the 6 experts, indicating low variability between evaluators.
Conclusion:
An expert panel agreed with the individual submitting physician regarding a diagnosis of some form of primary vasculitis in nearly all cases, but disagreement about the exact form of vasculitis occurred in 22% of cases. Physician-based opinion may be more reliable for defining general categories of vasculitis than for defining specific subtypes. This exercise highlights the potential for diagnostic bias when using physician-opinion to define the gold standard diagnosis.
Disclosure:
C. Ponte,
None;
A. Craven,
None;
J. Robson,
None;
P. C. Grayson,
None;
R. Suppiah,
None;
R. A. Watts,
None;
P. A. Merkel,
None;
R. A. Luqmani,
None.
« Back to 2013 ACR/ARHP Annual Meeting
ACR Meeting Abstracts - https://acrabstracts.org/abstract/review-of-the-expert-panel-methodology-in-the-diagnostic-and-classification-criteria-for-vasculitis-study-a-pilot-study/