Predicting which children with juvenile idiopathic arthritis will not attain early remission with conventional treatment: Results from the Reacch-out cohort

Objective. To estimate the probability of early remission with conventional treatment for each child with juvenile idiopathic arthritis (JIA). Children with a low chance of remission may be candidates for initial treatment with biologics or triple disease-modifying antirheumatic drugs (DMARD). Methods. We used data from 1074 subjects in the Research in Arthritis in Canadian Children emphasizing Outcomes (ReACCh-Out) cohort. The predicted outcome was clinically inactive disease for ≥ 6 months starting within 1 year of JIA diagnosis in patients who did not receive early biologic agents or triple DMARD. Models were developed in 200 random splits of 75% of the cohort and tested on the remaining 25% of subjects, calculating expected and observed frequencies of remission and c-index values. Results. Our best Cox logistic model combining 18 clinical variables a median of 2 days after diagnosis had a c-index of 0.69 (95% CI 0.67-0.71), better than using JIA category alone (0.59, 95% CI 0.56-0.63). Children in the lowest probability decile had a 20% chance of remission and 21% attained remission; children in the highest decile had a 69% chance of remission and 73% attained remission. Compared to 5% of subjects identified by JIA category alone, the model identified 14% of subjects as low chance of remission (probability < 0.25), of whom 77% failed to attain remission. Conclusion. Although the model did not meet our a priori performance threshold (c-index > 0.70), it identified 3 times more subjects with low chance of remission than did JIA category alone, and it may serve as a benchmark for assessing value added by future laboratory/imaging biomarkers.