Longterm health outcomes and quality of life in American and Italian inception cohorts of patients with juvenile rheumatoid arthritis. II. Early predictors of outcome

Objectives. To determine whether demographic, and immunogenetic variables measurable during the first 6 months of illness predict longterm health outcomes and quality of life in patients with juvenile rheumatoid arthritis (JRA). Methods. Patient eligibility criteria; (1) first examined in our units between 1958 and 1990 within 6 months of onset of symptoms; (2) diagnosis of JRA by American College of Rheumatology criteria; (3) disease duration of at least 5 years at the time of assessment of outcome. Instruments used: (1) the Health Assessment Questionnaire (HAQ, short form), or HAQ (CHAQ) to measure disability (0-3 scale), (2) pain, and (3) parental assessment of overall well being, each scored on a 15 cm visual analog scale; (4) the Quality of Life Scales (QOLS) (adults only). Independent variables that showed significant results using univariate tests underwent multiple logistic regression analysis. Results. 227 patients were available for analysis. Mean duration of disease at time of assessment of outcome was 15 years (range 5.3-36.1). Univariate tests allowed 11 variables for disability, 9 for pain, 7 for overall well being, and 4 for QOL into the multivariate analysis. The best predictor of higher disability was the articular severity score (odds ratio, OR, 5.69) while antinuclear antibody positivity foretold less disability (OR 0.29), HLA-DR5 positivity conferred the greatest risk for pain (OR 3.34), while HLA-B5, DR3, and C3 were protective (OR 0.25, 0.28, 0.33, respectively). Early hand involvement was the strongest predictor of poorer overall well being (OR 8.75). On the erythrocyte sedimentation rate was predictive of future QOL, but the model yielded a low C statistic (< 70%) and the OR 95% confidence limits were extreme (OR 9.77; 95% confidence interval, 1.22-77.8). Conclusion. Clinical and immunogenetic variable measurable within 6 months of onset of JRA can be used to predict future disability, pain, and well being. QOL appears more difficult to forecast, perhaps due to the multiple domains that make up this outcome. Further study is needed to identify other genetic and laboratory factors that predict outcome in JRA with greater precision.