Evaluation of Naranjo Adverse Drug Reactions Probability Scale in causality assessment of drug-induced liver injury

Background Causality assessment in hepatotoxicity is challenging. The current standard liver-specific Council for International Organizations of Medical Sciences/Roussel Uclaf Causality Assessment Method scale is complex and difficult to implement in daily practice. The Naranjo Adverse Drug Reactions Probability Scale is a simple and widely used nonspecific scale, which has not been specifically evaluated in drug-induced liver injury. Aim To compare the Naranjo method with the standard liver-specific Council for International Organizations of Medical Sciences/Roussel Uclaf Causality Assessment Method scale in evaluating the accuracy and reproducibility of Naranjo Adverse Drug Reactions Probability Scale in the diagnosis of hepatotoxicity. Methods Two hundred and twenty-five cases of suspected hepatotoxicity submitted to a national registry were evaluated by two independent observers and assessed for between-observer and between-scale differences using percentages of agreement and the weighted kappa (kappa(w)) test. Results A total of 249 ratings were generated. Between-observer agreement was 45% with a kappa(w) value of 0.17 for the Naranjo Adverse Drug Reactions Probability Scale, while there was a higher agreement when using the Council for International Organizations of Medical Sciences/Roussel Uclaf Causality Assessment Method scale (72%, kappa(w): 0.71). Concordance between the two scales was 24% (kappa(w): 0.15). The Naranjo Adverse Drug Reactions Probability Scale had low sensitivity (54%) and poor negative predictive value (29%) and showed a limited capability to distinguish between adjacent categories of probability. Conclusion The Naranjo scale lacks validity and reproducibility in the attribution of causality in hepatotoxicity.