LAUSR

Objective: To determine the number and type of errors found in the record of drug-related problems in the SINEA database, an electronic system for voluntary reporting of adverse events in healthcare, in order to quantify the differences between the raw and refined databases, suggest improvements, and determine the need for refining said databases. Methods: A Pharmacist reviewed the database and refined the adverse events reported from January to August, 2014, considering the “describe_what_happened” field as the gold standard. There was a comparison of the rates of medication errors, both potential and real, adverse reactions, impact on the patient, impact on healthcare, and medications more frequently involved in the raw and refined databases. Agreement was calculated through Cohen’s Kappa Coefficient. Results: 364 adverse events were reported: 66.7% were medication errors, 2.7% adverse reactions to the medication (2 were wrongly classified as both, showing a total percentage >100%) and 31% were other events. After refinement, the percentages were 69.5%, 5.8% and 24.7%, respectively (κ=0.85; CI95% [0.80-0.90]). Before refinement, 73.6% of medication errors were considered potential vs. 82.3% after refinement (κ=0.65; CI95% [0.54- 0.76]). The medication most frequently involved was trastuzumab (20.9%). The  “molecule” field was blank in 133 entries. A mean of 1.8±1.9 errors per entry were detected. Conclusions: Although agreement is good, the refinement process cannot be avoided, as it provides valuable information to improve pharmacotherapy. Data quality could be improved by reducing the number of type-in text fields, using drop-down lists, and by increasing the training of the reporters.