Estimating the incidence of rheumatoid arthritis: trying to hit a moving target?

To examine the effect of delay between symptom onset and notification to an arthritis register and the effect of application of the American College of Rheumatology (ACR; formerly, the American Rheumatism Association) 1987 criteria in a cumulative manner on estimates of the incidence of rheumatoid arthritis (RA).

General practitioners and/or hospital consultants in the Norwich Health Authority, Norfolk, UK, notified the Norfolk Arthritis Register (NOAR) of all patients who had onset of inflammatory polyarthritis (swelling of > or =2 joints) during 1990. The patients were assessed within 2 weeks of notification and annually thereafter. The ACR 1987 criteria for RA were applied at each assessment. Age- and sex-specific incidence rates were calculated.

If up to 12 months elapsed from symptom onset to notification to NOAR and the ACR criteria were applied at the baseline assessment, RA incidence estimates, age-adjusted to the population of England and Wales, were 30.8/100,000 for women and 12.7/100,000 for men. If up to 5 years elapsed from symptom onset to notification, these estimates rose by 45% for women and 36% for men. If up to 5 years elapsed between symptom onset and notification and the criteria were applied cumulatively, the estimates rose by 75% and 93% for women and men, respectively, compared with the 1-year data, reaching 54.0/100,000 for women and 24.5 per 100,000 for men.

Accurate estimation of the incidence of RA requires long-term followup of patients who present with undifferentiated inflammatory polyarthritis. The highest age-adjusted estimates from this study are probably the best that are available.