Classifying idiopathic inflammatory myopathies: comparing the performance of six existing criteria

Various criteria have been proposed to classify the inflammatory myositides (IIMs) polymyositis (PM) and dermatomyositis (DM). However, none have received universal acceptance. Our aim was to assess the performance of the main criteria used to classify IIM. Specialist consultant diagnosis was considered the gold standard.

Patients attending King's College Hospital (KCH) or Reggio Emilia Hospital (REH) since 1990 with a diagnosis of IIM or non-inflammatory myopathy were identified, and their records and laboratory investigations retrospectively reviewed. Where the complete data required for the classification criteria or a final physician diagnosis was unavailable, patients were excluded. 52 patients with a specialist diagnosis of PM, DM, inclusion body myositis (IBM) or non-inflammatory myopathy were included. Agreement between specialist consultant diagnosis and classification criteria was measured using Cohen's kappa (κ) statistics. Sensitivity and specificity were also calculated.

The Dalakas (2003) criteria demonstrated substantial agreement with specialist diagnosis: κ=0.69, sensitivity 77%, specificity 99%. The European Neuromuscular Centre criteria (ENMC) demonstrated fair agreement: κ=0.49, sensitivity 71%, specificity 82%. Other criteria performed less well. In particular, the Bohan and Peter criteria demonstrated a specificity of only 29%.

The criteria of Dalakas (2003) agreed best with specialist consultant diagnosis. The criteria of Bohan and Peter demonstrated very poor specificity. Prospective studies are required to develop improved classification criteria.