Eligible for inclusion in the review were all studies that assessed the accuracy of the original or a modified version of the Leiden clinical prediction rule in identifying patients (older than 16 years) who are at high risk of developing rheumatoid arthritis. Eligible patients had to have recent-onset undifferentiated arthritis. Symptoms had to have been present for between six weeks and 12 months, and include synovitis in at least one joint and/or two or more swollen joints. The comparator (reference) test was progression to rheumatoid arthritis as defined by the American College of Rheumatology criteria (ACR 1987).
Included studies were conducted in The Netherlands, Canada, UK, Germany and Japan. Most studies were conducted in early arthritis clinics, with one based in a hospital setting. Some patients were receiving or had received disease-modifying antirheumatic drugs, and a small proportion of patients were also receiving glucocorticoids. The mean age of participants ranged between 45 and 52.5 years, and most participants were female. The proportion of participants developing rheumatoid arthritis ranged from 31.14% to 76.19%.
Study inclusion criteria varied across studies, as did definitions for undifferentiated arthritis, where this was reported. Four cohorts used the modified version of the Leiden clinical prediction rule.
Two reviewers independently screened studies for inclusion; discrepancies were resolved by a third reviewer.