Validated Prediction Models May Identify Severe Disease Course in Juvenile Idiopathic Arthritis

doctor at desk
Stock exchange market concept, stock broker looking at graph working and analyzing with display screen, pointing on the data presented and deal on a exchange, Businessman trading stocks online.
Researchers validated the performance of a Canadian prediction model in predicting severe disease course and the nonachievement of remission in a Nordic cohort with juvenile idiopathic arthritis.

To improve outcomes in juvenile idiopathic arthritis (JIA), validated prediction models based on clinical assessment may be useful in identifying severe disease course and ability to achieve remission, according to study results published in Arthritis Research & Therapy.1

Investigators of this study sought to validate the performance of the Canadian model developed by Guzman et al2 and the Nordic model derived from Rypdal et al3 to predict severe disease course and nonachievement of remission in patients with JIA.

The validation cohort included the data of 440 Nordic patients with JIA who had a baseline visit within 6 months of disease onset and were followed-up at regular visits up to 8 years. Investigators evaluated the Canadian prediction model in the Nordic cohort to predict severe disease course using cumulative active joint count, remission status, the Childhood Health Assessment Questionnaire disability index, and the Physical Summary score.

The Canadian model was first tested as published and then tested again following fine-tuning of model coefficients. Investigators further performed an internal validation of the Nordic prediction model by repeated partitioning of the cohort into multiple training and validation sets. External validation of the Canadian model’s predictive performance was assessed by varying the probability threshold to obtain sensitivity and specificity values, which were then reported as receiver operating characteristic curves and C-indices (C-index >0.7 was considered useful for clinical prediction).

Related Articles

In the Nordic cohort, 98 of the 440 patients (22%) had severe disease course, which was consistent with the Canadian development cohort with 125 of 609 patients (21%). The external validation of the Canadian model resulted in C-indices of 0.85 (interquartile range [IQR], 0.83-0.87) and 0.66 (IQR, 0.63-0.68) for prediction of severe disease course and nonachievement of remission, respectively. After fine-tuning in training sets, the median C-indices of the Canadian model were 0.85 (IQR, 0.80-0.89) and 0.69 (IQR, 0.65-0.73), respectively. The internal validation of the Nordic model yielded a comparable performance for predicting severe disease course, with a median C-index of 0.90 (IQR, 0.86-0.92).

Study limitations included the large number of clinical variables used to construct the original models and missing data for predictor and outcome variables.

Investigators suggested that both the Canadian and Nordic prediction models demonstrated an excellent ability to predict severe disease course in children with JIA. They added that a clinically useful prediction model for JIA outcomes should be further tested across diverse cohorts, countries, and ethnicities.


1. Rypdal V, Guzman J, Henrey A, et al. Validation of prediction models of severe disease course and non-achievement of remission in juvenile idiopathic arthritis; part 1 – results of the Canadian model in the Nordic cohort [published online December 5, 2019]. Arthritis Res Ther. doi:10.1186/s13075-019-2060-2

2. Guzman J, Henrey A, Loughin T, et al. Predicting which children with juvenile idiopathic arthritis will have a severe disease course: results from the ReACCh-Out cohort. J Rheumatol. 2017;44(2):230-240.

3. Rypdal V, Arnstad ED, Aalto K, et al. Predicting unfavorable long-term outcome in juvenile idiopathic arthritis: results from the Nordic cohort study. Arthritis Res Ther. 2018;20(1):91.