Background Currently, in primary care no satisfactory referral model exists for referral to the rheumatologist of patients with chronic back pain (CBP) with possible axial spondyloarthritis (axSpA). Recently, a non-invasive, inexpensive and easily applicable two-step referral strategy was developed, based on a computer-generated model.1
Objectives The aim of our study was to validate this two-step referral strategy in the SPondyloArthritis Caught Early (SPACE)-cohort.
Methods In the first step of the referral strategy, the presence of psoriasis, alternating buttock pain (ABP) and improvement of back pain by exercise (IBPE) are registered. If ≥2 of these anamnestic features are present, the patient is referred to a rheumatologist. In the second step, if ≤1 anamnestic feature is present, HLA-B27 testing is performed; if the result is positive, the patient is also referred. The required data was obtained from the SPACE-cohort, in which patients with CBP (≥3 months, ≤2 years, onset <45 years) were included (n=192). The model was tested in this cohort; the performance of the model was evaluated by calculating sensitivity, specificity and likelihood ratios (LR) using classification by ASAS axSpA criteria as external standard.2 For the patients which were erroneously referred or erroneously not referred, post-test probabilities for axSpA were calculated based on LR products for the presence of SpA features3, with a LR product of ≥78 which equals a probability of ≥80% defined as cut-off value for probable axSpA.
Results A total of 74/192 patients had axSpA (ASAS), 61 of which would have been referred according to the strategy, yielding a sensitivity of 82.4%. The 13 false negatives are obligatory HLA-B27 negatives, have ≥1 SpA feature and sacroiliitis on imaging, 3/13 (23%) of which on X-ray. The referral strategy would have referred a total of 93/192 patients, yielding a specificity of 72.9%. Of the false positive patients, 8/32 (25%) would have been referred erroneously due to HLA-B27 positivity in absence of ≥2 SpA features. Half of these 32 patients (n=16; 50%) had a combination of ABP and IPBE, which are features of inflammatory back pain, which itself is not specific for axSpA. Although none of these false positives fulfil the ASAS axSpA criteria, 6/32 (19%) had probable axSpA (probability ≥80%) (figure).
Conclusions The two-step referral strategy performed well in the SPACE-cohort. These results show the potential of simple referral strategies for the early recognition of axSpA. However, all patients in this study had already been referred to the rheumatologist. Thus, the performance of this model in its target population (i.e. primary care) remains to be addressed in further studies.
Braun, Rheumatol 2013;52:1418-24.
Rudwaleit, ARD 2009;68:777-83.
Rudwaleit, ARD 2004;63:535-43
Disclosure of Interest None declared