Accuracy of existing diagnosis criteria for Behcet's disease (BD).

Abstract: There are 5 major sets of diagnosis criteria for BD: Mason & Barnes, O'Duffy, Dilsen, Japan revised criteria, and the International criteria (the latest one). To evaluate their accu­racy we analysed their sensitivity and specificity in 2069 BD and 1519 control patients. The accuracy was calculated by the sum of BD patients fulfilling the criteria and the control patients correctly classified as non BD, divided by the sum of the total BD and control patients. A Confidence Interval (CI) at 95% was calculated for each accuracy to allow their comparison. The sensitivity of the 5 criteria was: Mason & Barnes 74.2%, O'Duffy 80.3%, Diasen 91.7%, Japan 93.2%, and the International 86.2%. The specificity was: Mason & Barnes 99,4%; O'Duffy 97.6%, Dilsen 89.3%, Japan 96.4%, and the international 97.5%. The highest sensitivity was for the Japan revised criteria, and the highest specificity for Mason & Barnes. The analysis of the 2 most sensitive criteria showed that their high sensitivity is due to the use of an ex­tremely simplified linear discriminant function charac­teristic of BD symptoms (giving different diagnostic value to each symptom). For the Japan criteria ophthalmologic manifestations are given 2 points instead of 1 for other major manifestations. For Dilsen criteria the pathergy phenomenon is given 2 points instead of 1. The accuracy of the 5 sets of criteria and their CI was: Mason & Barnes 84.9% (CI = 0.7), O'Duffy 87.6% (CI=1.1), Dilsen 90.7% (CI = 0.9), Japan 94.7% (CI = 0.7), and the International 91% (CI = 0.9). The most accurate criteria was the revised Japan criteria. We analysed the structure of the International criteria to see if some modifications could increase its sensitivity (which is very low) without loosing too much on specificity, or complicating too much its calculation as for the Japan or Dilsen criteria. After multiple analysis we came up with the following modifications: Omit the obligation for oral aph­thosis and give 2 points for the ophthalmological manifesta­tions. Oral aphthosis, genital aphthosis, skin manifestations, and positive pathergy takes 1 point each. Diagnosis is reached with 3 points. The sensitivity increases from 86.2% to 96.7 %, the specificity decreases from 97.5% to 94.5%, and the accuracy improves from 91 % to 95.8% (CI = 0.65).