Page 42 - Recognizing axial spondyloarthritis - Janneke de Winter

CHAPTER THREE
according to predefined criteria. Second, we evaluated the degree of statistical heterogeneity and inconsistency by using the T , Chi2 and I2 statistics. Heterogeneity was considered significant at P < 0.10 (40). 2 I2 values of 25, 50, and 75% were considered low, moderate, and high inconsistency, respectively (40). For studies which we assessed sufficiently homogeneous to include in the meta-analysis, we used a random effects model since we anticipated not one true effect size for all studies, despite potential measured quantitative homogeneity (41).
Data synthesis
We pooled studies in a random-effects model by using the Mantel-Haenszel method, which estimates the between-study variation by comparing the result of each study with a Mantel-Haenszel random-effect meta-analysis result. Data were calculated as pooled prevalence with the corresponding 95% confidence interval (CI) and the difference in pooled prevalence between AS and nr-axSpA, with the corresponding 95% CI. We performed data analyses using MetaXL (42) in Microsoft Excel 2010 and Review Manager 5. Forest plots were produced for all analyses using Review Manager 5, and we adapted these to our specific needs (to show the difference in prevalence instead of the difference in risk).
RESULTS
Search
The electronic database search identified 447 articles. The use of citation pearl growing resulted in one additional article (Figure 1). After merging for duplicates and screening of the titles and abstracts, we completely reviewed 44 unique articles. One article reported retrospective data. One full-text article could not be retrieved. Five articles did not report symptoms of peripheral or extra-articular disease manifestations. In 25 articles, there was no clear differentiation between AS and nr-axSpA. Two articles did not use the ASAS and mNY criteria to classify axial SpA. Eight articles reported data on duplicate cohort studies. Finally, eight studies were included in the meta-analysis (6,9,11,23,43–46).
Characteristics of the included studies
Eight studies fulfilled the inclusion criteria. The characteristics of the enrolled studies are summarized in Table 1. Seven out of 8 studies were longitudinal SpA cohort studies. One study was a cross-sectional study of 100 consecutive patients visiting the outpatient clinics. In total, 3478 patients were included, of whom 2236 AS and 1242 nr-axSpA patients. In all studies patients with AS fulfilled the mNY- criteria. Patients with nr-axSpA fulfilled the ASAS criteria in 6 out of 8 studies. Both GESPIC and Swiss Clinical Quality Management (SCQM) used the ASAS criteria with minor modifications since patients were included before the development of the ASAS criteria.
40