Abstract
Background/Aims The Roussel Uclaf Causality Assessment Method (RUCAM) is the most commonly used algorithm to evaluate drug-induced liver injury (DILI). It uses 7 criteria to assess causality and provides a total score ranging from −9 to 15, which is stratified into 5 categories describing the likelihood of DILI ranging from ‘Highly Probable’ to ‘Excluded’. An electronic RUCAM (eRUCAM) was previously developed at Kaiser Permanente Southern California (KPSC) with the Observational Medical Outcomes Partnership and GlaxoSmithKline to detect DILI in an electronic medical record. The study aim was to identify areas of improvement for future iterations of the eRUCAM based on results from chart review.
Methods The eRUCAM was used to score potential cases of DILI identified in patients taking at least one of 14 drugs commonly associated with DILI. Patients were required to be at least 18 years of age and have 12 months of continuous membership plus drug benefit prior to drug initiation. A random sample of 20 patients without pre-existing liver disease was selected for chart review to perform quality assurance on the programming specifications and identify areas for improvement in the algorithm. A hepatologist manually scored each patient using the paper-based RUCAM. Scores from the hepatologist were compared to those from the eRUCAM using the Wilcoxon signed rank test. Qualitative findings from chart review regarding future improvements to the algorithm were summarized.
Results The total score from eRUCAM was identical to those of the hepatologist for 6 cases (30%). The absolute difference between the hepatologist and eRUCAM scores was greatest for Criteria 5, which assesses non-drug causes of liver injury (P =0.001). Differences in scores were not statistically significant for other remaining criteria. Issues that were identified with Criteria 5 included: inadequate time windows for identifying non-drug causes, failure to consider laboratory results (e.g., hepatitis serology), incomplete lists of ICD-9 codes for non-drug causes, and inability to extract information from physician progress notes.
Conclusions Future work related to the development of an electronic causality assessment tool based on the RUCAM should focus on improving Criteria 5 by using wider time windows to identify non-drug cases, laboratory results, and natural language processing.
