Background Medicines reconciliation at discharge is a key strategy to ensure proper drug prescription and the effectiveness and safety of any treatment.

Purpose To analyse the effectiveness of an information technology-based medicines reconciliation intervention to reduce reconciliation errors at discharge.

Material and methods A quasi-experimental interrupted time series study carried out in the cardio-pneumology unit of a general hospital from February to April 2013. The study consisted of three phases: pre-intervention, intervention and post-intervention, each involving 23 days of observations. The intervention consisted of incorporating a pharmacist in the medical team, who included the patient’s pre-admission medicines in an information technology-based application integrated into the electronic clinical history of the patient. The effectiveness was evaluated by a segmented regression analysis of the mean daily proportion of reconciliation errors per patient in the discharge report using the Prais–Winsten method. The types of error identified and their potential seriousness were then analysed, as was the effectiveness of the intervention to reduce the errors considered to be of clinical importance.

Results 321 patients (119, 105 and 97 in each phase, respectively) were included in the study. For the 3966 medicines recorded, 1087 reconciliation errors were identified in 77.9% of the patients. Pharmaceutical intervention led to a gradual reduction in these errors (β₃ = −0.42; p = 0.553), especially in the case of those of clinical importance (β₃= −0.54; p = 0.029). When pharmaceutical intervention was withdrawn, the number of errors increased again, both overall (β₄ = 29.06; p = 0.003) and in the case of clinically important errors (β₄ = 10.8; p = 0.002). Most errors involved omission of medicines (46.7%) or incomplete prescription (43.8%), 35% being considered of clinical importance.

Conclusion The proposed intervention highlighted the high incidence of reconciliation errors at discharge and was effective in reducing both the overall percentage of errors and those considered to be of clinical importance.

References and/or Acknowledgements We would like to thank the participating patients and physicians who supported our study.

No conflict of interest.