Needles in haystacks.

Drug-induced liver disease remains a significant problem for physicians caring for patients with this therapeutic complication and for those developing and licensing new drugs. There has been renewed interest recently in drug-induced liver injury (DILI) investigation, with respect to the assessment of causality to permit prompt and accurate diagnosis and the determination of risk factors and the mechanisms by which drugs injure the liver, both in a predictable dose-dependent manner and idiosyncratically. Current investigations focus on determining the true incidence of such adverse drug hepatotoxic reactions, both for drug development and clinical practice, as well as on elucidating the mechanisms of disease and on finding diagnostic biomarkers. The Drug Induced Liver Injury Network (DILIN), supported by the National Institute of Diabetes and Digestive and Kidney Diseases (NIDDK), includes in its charge the study of four drugs (phenytoin, isoniazid, amoxicillin/clavulanic acid, and valproic acid) that are relatively frequent causes of DILI. In the current study, the authors have attempted to search an institutional diagnostic database of the International Classification of Diseases-9th Edition-Clinical Modification (ICD-9-CM) codes, using three different strategies, to identify patients with DILI due to the four targeted drugs. Unfortunately, the search strategy that yielded the greatest number of specific DILI cases was too insensitive and labor-intensive to be useful for DILI research, whereas strategies that were more specific were also far less sensitive and yielded an inadequate number of cases to study. The findings in this report emphasize the inadequacy of administrative patient diagnosis databases for specific research in DILI and lend support to the DILIN concept that is funded by NIDDK as the most suitable means to further understanding in this burgeoning field.