Page 43 - SPONDYLOARTHRITIS IN COLOMBIA -

There is an important problem regarding the interpretation of the performance of criteria in observational studies like ours. In some observational studies, such as SPACE and DESIR, MRI and HLA-B27 testing are performed in all patients ‘by protocol’, regardless of the likelihood of a diagnosis of SpA. In other observational SpA cohorts, including the Colombian cohort, many of these tests are performed on clinical indication only, namely in those patients with the highest probability of having SpA, or –alternatively- in those in whom the presence of other SpA-features does not convince the diagnosing physician. Here the danger of bias by indication arises, because the presence of a certain test result may reflect the perceived severity of the disease, and consequently information about these tests in patients with milder symptoms may be lacking. Additionally, the lack of information about HLA-B27 and MRI in a proportion of patients could have influenced the performance of the classification criteria, since these parameters (HLA-B27 and MRI) have a rather prominent place in the classification criteria (HLA-B27 in the Amor criteria and HLA-B27 and MRI in the ASAS classification criteria). It is unclear how this lack of data may have impacted the results.
There are several limitations in this study, including the retrospective design and missing data, which reflects the daily clinical practice. First, there was not a fixed protocol requesting for additional clinical testing for the totality of patients; in contrast, these test were performed based on the opinion of the physician. Second, we selected the rheumatologist diagnosis of SpA as the external standard. A concern with our choice of external standard is that the rheumatologist´s diagnosis might be biased and influenced by current knowledge, by the experience and the existing criteria classification sets, by new diagnostics developments (MRI) or by discussion with experts. Third, data were only collected at the initial visit to the clinic (cross-sectional approach) and there was no information of follow-up visits available, not information about (response to) treatment.
41