Enviar rexistro por email: Adjustment for misclassification bias is necessary in survey-based arthritis surveillance to produce credible estimates of prevalence