Evaluation of adverse drug reactions in medical intensive care units.

Purpose: Patterns of adverse drug reactions (ADRs) in the medical intensive care unit (MICU) were analysed, and signals for detecting ADRs were developed from the analysis.

Method: A retrospective study was conducted in MICU wards at a tertiary care teaching hospital in Seoul, Korea. The areas included one general MICU and one cancer centre MICU. Two pharmacists evaluated ADRs in terms of length of stay, causality, severity, preventability, types, related organs, and incidence. Differences in ADR perception rates between physicians and pharmacists were also evaluated. ADR cases detected through the evaluation were reviewed to develop specific alerting signals for ICU ADRs.

Results: The study group included 346 patients admitted to the ICU over 4 months. The overall incidence of ADRs was 32%. ICU length of stay is closely related to ADRs (p = 0.014). Most ADR cases were mild, temporary, and harmful to the patient. Twenty percent of ADRs were preventable, and 74% were type A. Of the ADRs, 70% were noted by physicians; 80% required intervention. The most commonly implicated drug was amphotericin B, and the clinical presentation was a haematologic reaction. Data on the time required for pharmacists to identify ADRs indicated that they were not slower than physicians. Six signals for early detection of the ADRs were developed.

Conclusions: The overall ADR incidence in the MICU was about one-third, and the length of stay of the ADR group was longer than that of those without this experience. Automated signal generation was developed. It seemed to be a valuable tool for faster and more efficient patient management, and possibly prevention of ADRs. A future study should scientifically evaluate the clinical relevance of this tool.