Major Finding: None of 97 analyzed case reports or series reports of drug-induced liver injury included all 42 elements, which the investigators deemed necessary for evaluating causality of such events. A median of 48% of the elements were missing in the reports.
Data Source: Retrospective analysis.
Disclosures: Funding for the Drug-Induced Liver Injury Network is provided by the National Institute of Diabetes and Digestive and Kidney Diseases; the study was conducted on behalf of this network. One author, Dr. John G. McHutchison, reported receiving research support from, and acting as a scientific adviser for, GlaxoSmithKline and Merck & Co.
Key clinical information is often lacking in published reports of drug-induced liver injury, Dr. Vijay K. Agarwal and his colleagues reported.
The finding is concerning, because accurate reporting of drug-induced liver injury—the single leading cause of acute liver failure in the United States—is essential for the development of reliable, interpretable data to help promote early detection and awareness of drug-induced hepatotoxicity, said Dr. Agarwal of Duke University, Durham, N.C., and colleagues.
For the study, which was conducted on behalf of the Drug-Induced Liver Injury Network, the investigators developed a list of 42 elements necessary for evaluating causality of drug-induced liver injury, and they analyzed 97 published case reports or series of such injuries for the presence of these elements.
Basic disease, drug, and demographic information was present in the vast majority of reports, but numerous important elements for determining causality and eliminating alternative causes of liver injury were lacking. These elements included bilirubin level (missing in 12% of reports), initial alkaline phosphatase level (missing in 58% of reports), and competing viral etiologies (missing in more than 50% of reports).
None of the reports included all 42 elements, which the investigators considered to be minimally necessary for evaluating the causes of the adverse effects. A median of 48% of the elements were missing in the reports.
The reports evaluated included 23 single case reports, 7 brief communications, 46 small case series, and 21 letters to the editor, and they focused on six drugs from three drug classes: amoxicillin/clavulanic acid (35 reports), troglitazone (32), rosiglitazone (10), pioglitazone (8), zafirlukast (8), and montelukast (4). The first two drugs are known to cause clinically apparent drug-induced liver disease, while the other four rarely cause liver injury, but case reports have been important in documenting the medications' potential for hepatotoxicity, the investigators noted.
Some studies included only vague descriptions of how certain diagnoses were excluded, and data on abnormal results from serial liver tests often were not included. Single case reports had significantly fewer missing elements than letters to the editor and small case series (a median of 33% vs. 50% and 48% of the elements were missing, respectively).
No significant differences were observed on the basis of journal type: The median percentage of missing elements was 50% for major internal medicine journals, 48% for gastroenterology and liver subspecialty journals, and 45% for other types of journals.
The authors noted that unless the essential details for interpreting the findings are included in such reports, it will be impossible to determine if episodes of hepatotoxicity can be causally assigned to a specific drug or combination of drugs. In addition, opportunities to identify rare events that might not be apparent in clinical trials, and to increase awareness of issues possibly associated with a drug early in its development and use, will be missed, the investigators said. They argued that a more standardized approach to the reporting of drug-induced liver injury is needed.
A checklist of minimal elements for diagnosing drug-related liver injury and for assessing causality should be developed, and a secondary list of elements that are helpful in many situations—such as results of assays for anti–hepatitis E virus antibodies to exclude hepatitis E, or magnetic resonance cholangiopancreatography to fully exclude biliary obstruction—would be useful, they said.
Such standards, which have been suggested in the past but not widely adopted, could be posted on a publicly funded Web site, with the goal that they would ultimately be adopted by journal editors, the authors suggested.