Optimizing the identification of patients with axial spondyloarthritis in primary care-the case for a two-step strategy combining the most relevant clinical items with HLA B27

Methods. Consecutive patients < 45 years old (n = 950) with back pain (BP) > 2 months presenting to 143 orthopaedists were referred to 36 rheumatologists who made the diagnosis. The predictive value of HLA B27 (n = 298) alone and in combination, including modelling and a two-step strategy, was calculated. Results. Among all patients (mean age 36 years, 52% female, median duration of BP 32 months), 107 had axSpA (36%). Using a simple model, HLA B27 alone performed better than all combinations of clinical items and adding it did not improve likelihood ratios (LRs). Using modelling, two-phase strategies were analysed. Additional items were only relevant in the HLA B27-negative group: improvement by movement, buttock pain and psoriasis. Combining this information revealed the presumably best strategy to predict axSpA in primary care: more than one of these items or HLA B27 need to be present (sensitivity 80.4%, specificity 75.4%, LR+ 3.27 and LR- 0.26). Conclusion. This is the first study to show that patients with axSpA are more reliably identified in primary care by a strategy that includes HLA B27. Because of the two-step approach, the test needs to be performed in only about half of patients with chronic BP.