Accuracy and reproducibility of two scales in causality assessment of unexpected hepatotoxicity

What is known and Objective: There is almost no published information about reliability of scales for causality assessment in hepatotoxicity at pharmacovigilance centres. The aim of this study was to compare two commonly used scales in cases of unexpected hepatotoxicity, in evaluating their accuracy and reproducibility at pharmacovigilance centres (in signal detection). Methods: Two scales [Council for International Organizations of Medical Sciences or Rousel Uclaf Causality Assessment Method (CIOMS/RUCAM) and NARANJO] were compared in 19 cases of unexpected hepatotoxicity reported during 2004-2009. Data of the cases (with initial clinical assessments) were collected by a network of medical specialists using a structured reporting form. Later, two independent observers assessed each case using both scales. The accuracy and reproducibility of the scales were analysed by Kappa weighted (Kw) test. Results: Both scales (CIOMS/RUCAM vs. NARANJO) showed moderate agreement with the initial clinical assessments (accuracy) for observer A (Kw: 0 56 vs. 0 60) and substantial agreement for observer B (Kw: 0 72 vs. 0 70), with high agreement between observers (Kw: 0 84 vs. 0 67). Both observers (A vs. B) found low agreement between scales (Kw: 0 21 vs. 0 50), with lower scores for the CIOMS/RUCAM scale in 11 and nine cases, respectively. For an early perception of unexpected serious reactions, the scale is more useful if it is not asked for 'previous knowledge' and if it gives higher causality score. What is new and Conclusion: The CIOMS/RUCAM scale showed similar accuracy, but better reproducibility (agreement between observers) than the NARANJO scale, and therefore is recommended for use at pharmacovigilance centres. Fine-tuning of the CIOMS/RUCAM method could contribute to better detection of unexpected hepatotoxicity.