Background: Inflammatory arthritis (IA) is an immune-related condition defined by the presence of clinical synovitis. Its most common form is rheumatoid arthritis.
Objective: To develop scores for predicting IA in at-risk persons using multidimensional biomarkers.
Design: Prospective observational cohort study.
Setting: Single-center, Leeds, United Kingdom.
Participants: Persons with new musculoskeletal symptoms, a positive test result for anticitrullinated protein antibodies, and no clinical synovitis and followed for 48 weeks or more or until IA occurred.
Measurements: A simple score was developed using logistic regression, and a comprehensive score was developed using the least absolute shrinkage and selection operator Cox proportional hazards regression. Internal validation with bootstrapping was estimated, and a decision curve analysis was done.
Results: Of 455 participants, 32.5% (148 of 455) developed IA, and 15.4% (70 of 455) developed it within 1 year. The simple score identified 249 low-risk participants with a false negative rate of 5% (and 206 high-risk participants with a false-positive rate of 72%). The comprehensive score identified 119 high-risk participants with a false-positive rate of 29% (and 336 low-risk participants with a false-negative rate of 19%); 40% of high-risk participants developed IA within 1 year and 71% within 5 years.
Limitations: External validation is required. Recruitment occurred over 13 years, with lower rates of IA in later years. There was geographic variation in laboratory testing and recruitment availability.
Conclusion: The simple score identified persons at low risk for IA who were less likely to need secondary care. The comprehensive score identified high-risk persons who could benefit from risk stratification and preventive measures. Both scores may be useful in clinical care and should also be useful in clinical trials.
Primary funding source: National Institute for Health and Care Research Leeds Biomedical Research Centre.