Development and validation of a risk calculator to differentiate flares from infections in systemic lupus erythematosus patients with fever

• Published in: Autoimmunity reviews 2015-07, Vol.14 (7), p.586-593
• Main Authors: Beça, Sara, Rodríguez-Pintó, Ignasi, Alba, Marco A, Cervera, Ricard, Espinosa, Gerard
• Format: Article
• Language: English
• Subjects: Algorithm; Allergy and Immunology; Fever; Flare; Infection; Systemic lupus erythematosus
• ISSN: 1568-9972; 1568-9972
• DOI: 10.1016/j.autrev.2015.02.005

Abstract Objective To develop and validate a predictive risk calculator algorithm that assesses the probability of flare versus infection in febrile patients with systemic lupus erythematosus (SLE). Methods We evaluated SLE patients admitted because of fever in the Department of Autoimmune Diseases of our Hospital between January 2000 and February 2013. Included patients were those with final diagnosis of infection, SLE flare or both. Data on clinical manifestations, treatment and laboratory results were collected. Variables considered clinically relevant were used to build up all possible logistic regression models to differentiate flares from infections. Best predictive variables for SLE relapses based on their higher area under the receiver operating characteristic (ROC) curve (AUC) were selected to be included in the calculator. The algorithm was developed in a random sample of 60% the cohort and validated in the remaining 40%. Results One hundred and thirty SLE patients presented 210 episodes of fever. Fever was attributed to SLE activity and to infection in 45% and 48% of the cases, respectively. Three independent variables for prediction of flares were consistently selected by multivariate analysis: days of fever, anti-dsDNA antibody titres and C-reactive protein levels. Combination of these variables resulted in an algorithm with calculated AUC of 0.92 (95% CI: 0.87 to 0.97). The AUC for the validation cohort was of 0.79 (95% CI: 0.68 to 0.91). Conclusion The proposed flare risk predictive calculator could be a useful diagnostic tool for differentiation between flares and infections in febrile SLE patients.