Identifying Crude Drugs in Kampo Medicines Associated with Drug-Induced Liver Injury Using the Japanese Adverse Drug Event Report Database: A Comprehensive Survey

Abstract

The current study aimed to identify the crude drugs associated with drug-induced liver injury (DILI) in 148 Kampo medicines prescribed throughout Japan using the Japanese Adverse Drug Event Report (JADER) database, a large-scale spontaneous reporting system in Japan. First, we tabulated the number of DILI reports from the report-based dataset and the background information from the patient-based dataset. Thereafter, we combined the 126 crude drugs into 104 crude drug groups to examine multicollinearity. Finally, the reporting odds ratios (RORs), 95% confidence intervals, p values for Fisher's exact test, and number of reports were calculated for each crude group to identify those associated with DILI. Notably, the number of adverse event reports for DILI (63,955) exceeded that for interstitial lung disease (51,347), the most common adverse event. In total, 78 crude drug groups (90 crude drugs) were reported to have an ROR > 1, a p < 0.05, and ≥10 reported cases. Our results highlight DILI as an essential issue, given that it was among the most frequently reported adverse drug reactions. We were able to clearly identify the crude drugs associated with DILI, which could help manage adverse drug reactions attributed to Kampo medicines and crude drugs.

Keywords: Japanese Adverse Drug Event Report (JADER) database; Kampo medicine; crude drug; drug-induced liver injury (DILI); pharmacovigilance; reporting odds ratio (ROR); spontaneous reporting system; volcano plot.

Figure 1

Flow diagram of dataset construction for analysis. The causes of drug-related adverse events (drug name, causal relationship, etc.) for each drug included in the DRUG table were classified into three categories: “suspected drug”, “concomitant drug”, and “interacting drug”. Only “suspected drug” information was extracted from the DRUG table. Duplicate data (adverse events, outcomes, etc.) in the DRUG and REAC tables were deleted. Data from the DEMO table (demographic information such as patient gender, age, weight, etc.) were combined into the DRUG and REAC tables using the patient identification number (ID number). This created a report-based dataset. The report-based dataset was assigned a flag to determine whether the adverse event was DILI, a flag to determine whether the suspected drug was a Kampo medicine, and a flag for the 126 crude drugs contained according to the composition of the Kampo medicines. The dataset was then integrated so that the ID number would be unique. In doing so, priority was given to “applicable” for each assigned flag. Finally, we created a patient-based dataset by excluding cases with missing information on gender and age.

Figure 2

Characteristics of the patient-based dataset (N = 638,876). (a) Bar graph for gender; (b) histogram for age group; (c) line graph for reporting year. The data collection period was from April 2004 to November 2021; thus, the number of reports for FY2021 seems to be low. (d) Horizontal bar chart for using Kampo medicines and adverse events related to DILI.

Figure 2

Characteristics of the patient-based dataset (N = 638,876). (a) Bar graph for gender; (b) histogram for age group; (c) line graph for reporting year. The data collection period was from April 2004 to November 2021; thus, the number of reports for FY2021 seems to be low. (d) Horizontal bar chart for using Kampo medicines and adverse events related to DILI.

Figure 3

Characteristics of patients who used Kampo medicines and experienced DILI (N = 1701). (a) Bar graph for gender; (b) histogram for age group; (c) line graph for the reporting year. The data collection period was from April 2004 to November 2021; thus, the number of reports for FY2021 seems to be low. (d) Ranking of the top 10 primary diseases. The data were obtained from 1055 patients whose primary disease information was available; in JADER, there were patients without primary disease information and patients who were assigned more than once. (e) Ranking of the top 20 most-used drugs (suspected and concomitant drugs). As JADER was the source of patient information, data regarding the drugs used were obtained from all 1701 patients; some patients used only one drug, whereas others used multiple drugs.

Figure 3

Characteristics of patients who used Kampo medicines and experienced DILI (N = 1701). (a) Bar graph for gender; (b) histogram for age group; (c) line graph for the reporting year. The data collection period was from April 2004 to November 2021; thus, the number of reports for FY2021 seems to be low. (d) Ranking of the top 10 primary diseases. The data were obtained from 1055 patients whose primary disease information was available; in JADER, there were patients without primary disease information and patients who were assigned more than once. (e) Ranking of the top 20 most-used drugs (suspected and concomitant drugs). As JADER was the source of patient information, data regarding the drugs used were obtained from all 1701 patients; some patients used only one drug, whereas others used multiple drugs.

Figure 4

Crude drug groups associated with DILI. The x-axis is the natural logarithm of the reporting odds ratio (Ln (ROR)), whereas the y-axis is the ordinary logarithm of the p value from Fisher’s exact test converted to negative (−Log10 (p value)). Reporting odds ratios (RORs) were calculated by cross tabulation. The dotted line on the y-axis represents p = 0.05. The color of the plot represents the number of adverse events reported. The red–green–blue dots are the ordinary logarithm of the total number of reports (range: 0–3.5); a positive ROR indicates a greater tendency for adverse events to occur, whereas a smaller p value indicates a greater statistical significance. Crude drugs having stronger associations with DILI are shown in the upper right-hand corner of the scatterplot.

Figure 5

Cross tabulation and formula used to calculate the reporting odds ratio for an adverse event.