Performance of referral strategies for spondyloarthritis: a population-based nationwide study

To evaluate the performance of the referral strategy (RS) for SpA of a nationwide epidemiological study (EpiReumaPt), as compared with previously proposed RSs.

EpiReumaPt was a three-stage national epidemiologic study. In phase one, 10 661 adult participants were randomly selected and screened for rheumatic and musculoskeletal diseases. In the second phase, positive screenings for ⩾1 rheumatic and musculoskeletal disease plus 20% negative screenings were assessed by a rheumatologist. Finally, three rheumatologists revised all the information and defined the final diagnosis. All participants from phase two were included. Thirteen RS were tested against the SpA diagnosis using several metrics, including sensitivity, specificity, the post-test probability of SpA given a positive RS (positive predictive value) and given a negative RS (1 - negative predictive value).

From the total 3877 participants, 92 received a SpA diagnosis [weighted national prevalence: 1.6% (95% confidence interval: 1.2, 2.1)]. Modified versions of the Assessment of SpondyloArthritis international Society-RS and EpiReumaPt-RS were the most sensitive (85% and 72%, respectively) and yielded the lowest post-test probabilities of SpA if negative (0.6% and 0.7%, respectively). Considering the national prevalence (pre-test probability) of SpA (1.6%), a negative screening by these two RSs decreased the probability of SpA substantially (Assessment of SpondyloArthritis international Society: -63%; EpiReumaPt: -56%). Other RSs performed less well in reducing disease probability (range: -6.3%; -37.5%). Overall, the probability of SpA given a positive RS was small (positive predictive value range: 2.2%; 7.6%) and the EpiReumaPt RS yielded the best balance between sensitivity and positive predictive value.

The proposed EpiReumaPt RS performed the best as a screening tool for SpA in patients from the general population when laboratory and imaging data were not available.