Background and importance Automatic dispensing cabinets (ADCs) allow us to trace medications dispensed by patients. This also detects areas of improvement in the quality of care.

Aim and objectives To analyse discrepancies between prescription and dispensing of medications, investigate influencing factors and design areas for improvement.

Material and methods A cross sectional, descriptive, observational, retrospective study was conducted of prescriptions and dispensations through ADCs. 60 treatments for patients admitted on 8 July 2019 were randomised. Those who had surgery for that day or were discharged were excluded. The following variables were collected: number of prescribed medications, and number of parenteral and oral medications prescribed. Medications with conditional posology and multidose presentations were excluded from the analysis. The prescriptions and dispensations of each patient were reviewed. Discrepancy was defined when the number of units dispensed by medication were different from the number of units prescribed in 24 hours. Three variables were defined: total discrepancies, by default and by excess. A treatment complexity index (ICT) was calculated that took into account the number of prescribed medications (score 1 (0–4), 2 (≥5–9), 3 (≥10–14), 4 (≥ 15)), the dosage (1 point for each prescribed medication every 24 hours, 2 points every 12 hours, 3 points every 8 hours and 4 every 6 hours) and the administration route (1 point—only oral medication, 2 points—only parenteral medication and 3 points—both routes of administration). The index was the sum of the three sections. The ICT was related to the discrepancies detected by Pearson’s correlation. The data were extracted from the ATHOS prescription programme and the ADCs Dosys Software. Data were analysed with SSPS.V.20.

Results 40 treatments were reviewed. 68 discrepancies were found; 59 by default and 15 by excess. 30% of the treatments did not present discrepancies, 45% between 1 and 2, 20% between 3 and 4 and 5% ≥5. 35% of the treatments presented an ICT between 1 and 14 (low), 60% between 15 and 28 (medium) and 5%> 29 (high). Correlation between ICT and total discrepancies was statistically significant (r=0.614, p<0.01%).

Conclusion and relevance The discrepancy rate was high. The traceability of ADCs allowed us to identify areas for improvement. ICT can help identify those with the highest risk of discrepancies and establish measures to correct them.

Conflict of interest No conflict of interest