gms | German Medical Science

Aim: Medication errors (ME) are a common and preventable cause of medical emergencies and frequently remain undetected. Clinical decision support systems (CDSS) have been introduced to aid physicians in preventing and detecting ME. However, in clinical practice CDSS often operate as a single source tool using medication data only for the detection of ME, whereas more information is needed to reliably detect ME. The aim of this study was to analyze which types of data are needed to correctly identify ME in patients presenting at an emergency department (ED).

Method: In 1510 patients presenting at an ED clinically relevant ME were identified by an interdisciplinary panel of specialists in emergency medicine, clinical pharmacology and pharmacists. In addition all data available to the emergency physician was retrospectively analyzed in order to explore the requirement and contribution of different clinical data sources to the detection of ME.

Results: Availability of medication data was an obvious requirement for the detection of all 663 ME. Only 68 (10.2%) ME could be identified based on medication data alone (detection of drug-drug interactions and overdosing) while 381 (57.5%) ME required at least one additional type of data (367 x underlying disease(s), 7 x laboratory values, 4 x ECG and 3 x acute clinical symptoms) for detection. Two additional types of data were required for the detection of 181 (27.3%) ME (two-way combinations of 178 x underlying disease(s), 109 x laboratory values, 43 x clinical symptoms and 32 x ECG) and three additional types of data for the detection of 33 (5.0%) ME (three-way combinations of 33 x underlying disease(s), 29 x clinical symptoms, 20 x laboratory values and 17 x ECG) respectively.

Conclusion: Only 10% of all ME observed in emergency patients could be identified on the basis of medication data alone. So far, the possible extent of this problem has not been quantified in medical emergencies. Focusing electronic decisions support on more easily available drug data alone may lead to an under-detection of clinically relevant ME.