INTRODUCTION: Collecting information about drugs in clinical practice is essential for ongoing riskbenefit analysis of the drug use. Medical literature is an important source of new information on drug safety, in particular for the signal assessment. A signal is an information about a new potentially causal association, or a new aspect of a known association (e.g. change in frequency or severity of the reaction) between a drug and an adverse event (AE). AIM OF THE STUDY: To verify the effectiveness of the identification of adverse drug reaction (ADR) reports published in the local medical literature using MEDLINE and Embase, versus manual full text review of journals. MATERIAL AND METHODS: The study was performed for 20 randomly selected drugs and 84 Polish medical journals and covers a review of 1,576 individual journal issues with 20,146 articles. Retrospective analysis of literature reports collected during manual full text review was performed and compared to the outcome of database search. RESULTS: ADRs for analyzed drugs were identified only in 17 out of 84 journals, as a result of which 66 reports were analyzed. The majority of reports (55%) were found in local non-indexed journals. Three reports originated from journals indexed in MEDLINE and 9 reports from journals indexed in Embase were not found in these databases because databases do not fully cover conference abstracts and journal supplements. Moreover, while using databases for ADR report search there is a risk of missing up to 30% of ADR reports. The average gap between article publication date and database entry was 119 days. CONCLUSIONS: We verified that the effectiveness of the identification of ADR reports published in the local medical literature is more accurate based on manual full text review than by searching in bibliographic databases.

Department of Gastroenterology and Internal Diseases Medical University of Warsaw

MS Pharm s r o Prague Czech Republic

Citace poskytuje Crossref.org