Evaluation of drug-induced liver injury developed during hospitalization using Electronic Health Record (EHR)-based algorithm

Purpose: The incidence of drug-induced liver injury (DILI) has been increasing; however, few algorithms are available to identify DILI in electronic health records (EHRs). We aimed to identify and evaluate DILI with an appropriate screening algorithm.Methods: We collected data from 3 university hospitals between June 2015 and May 2016 using our newly developed algorithm for identifying DILI. Among patients with alanine transferase (ALT) ≤ 120 IU/L and total bilirubin (TB) ≤ 2.4 mg/dL in blood test results within 48 hours of admission, those who either had 1) ALT > 120 IU/L and TB > 2.4 mg/dL or 2) ALT > 200 IU/L at least once during hospitalization were identified. After excluding patients with liver disease-related diagnosis at discharge, medical records were retrospectively reviewed to evaluate epidemiological characteristics of DILI. Results: The total number of inpatients was 256,598, of whom 1,100 (0.43%) were selected by the algorithm as suspected DILI. Subsequently, 365 cases (0.14% of total inpatients, 95% confidence interval, 0.13-0.16) were identified as DILI, yielding a positive predictive value of 33.1%. Antibiotics (n = 214, 47.2%) were the major class of causative drug followed by chemotherapeutic agents (n = 87, 19.2%). The most common causative drug was piperacillin-tazobactam (n = 38, 8.4%); the incidence of DILI by individual agent was highest for methotrexate (19.4 cases/1,000 patients administered the drug). Common reasons for excluding suspected DILI cases were ischemic hepatitis and postoperative liver dysfunction. Conclusions: Using our EHR-based algorithm, we identified that approximately 0.14% of patients developed DILI during hospitalization. Further studies are needed to modify criteria for more accurate identification of DILI.