Development and validation of the AF score for diagnosis of adult-onset Still's disease in fever of unknown origin

Abstract

Objective: To develop and validate a diagnostic score to identify adult-onset Still's disease (AOSD) in fever of unknown origin (FUO). Methods: A single center, retrospective case-control study of inpatients with FUO from January 2018 to December 2021. Using clinical and laboratory data from 178 cases with AOSD and 486 cases with FUO, we developed an AOSD/FUO (AF) score with a Bayesian Model Averaging approach. AF score and Yamaguchi's criteria were evaluated by sensitivity, specificity, accuracy, and positive/negative predictive value for diagnosis of AOSD in developmental and validation samples. Results: Persistent pruritic eruptions (PPEs) in patients with rashes was higher in AOSD group than FUO group (52.3% vs 7.4%; P < 0.01). PPEs yielded a specificity of 97.5% and a sensitivity of 44.9%. AF score = PPEs × 3.795+Evanescent rash × 2.774+Serum ferritin × 1.678+Myalgia × 0.958+Neutrophil count × 0.185+Platelet count × 0.004. A cut-off value ≥ 5.245 revealed the maximizing sensitivity of 88.7% and specificity of 95.8% in discriminating AOSD from FUO in the validation group. And AF score improved the accuracy from 82.6% to 93.3% compared with Yamaguchi's criteria. Conclusions: We developed and validated a new score which can identify AOSD in FUO with higher classification accuracy than Yamaguchi's criteria. Future multi-centric prospective studies need to be designed to confirm the diagnosis value of AF score.

Keywords

• Adult-onset Still's disease
• Fever of unknown origin
• Persistent pruritic eruptions
• Classification criteria
• Bayesian method