The performance of the EULAR/ACR Idiopathic Inflammatory Myopathies classification criteria in an expert-defined ten-year incident cohort

Objectives - To assess the performance of the EULAR/ACR idiopathic inflammatory myopathies (IIM) classification criteria in a cohort of incident IIM cases and examine how criteria-assigned IIM subtype correlates with expert opinion.
Methods - Adults with newly diagnosed IIM attending Salford Royal NHS Foundation Trust were identified over a ten-year period. A retrospective review of all putative cases was performed, and those fulfilling a consensus expert-opinion diagnosis of IIM included. Clinical, serological and histological data were collected and each case assigned a single IIM subtype. The EULAR/ACR classification criteria were then applied and sensitivity, specificity, positive and negative predictive values calculated, presented with 95% confidence intervals (CI).
Results – 1637 cases were screened with 255 consensus expert-opinion IIM ultimately identified. Then applying the EULAR/ACR classification criteria, the sensitivity to diagnose an IIM was 99.6% (97.2-100) and 80.9% (76.0-85.8) for the criteria cut-points of ‘probable’ and ‘definite’ diagnoses respectively. In 94/255 cases the IIM subtype differed between consensus expert opinion and classification criteria, most strikingly in the group subtyped as ‘polymyositis’ by the EULAR/ACR criteria, where there was discrepancy in the majority (i.e. in 87/161).
Conclusion - The EULAR/ACR criteria performed with high sensitivity in identifying IIM in this external cohort of incident IIM. However, substantial disagreements arose between consensus expert opinion and the criteria regarding IIM subtype assignments, resulting in a large proportion of criteria assigned cases of ‘polymyositis’ having heterogeneous features. These results may have important implications for future use of these criteria in subsequent research.