Medication records in the emergency department: agreement between paper-based charts and automated dispensing device.

BACKGROUND Research exploring the agreement between traditional medication records and electronic records generated by an automated dispensing device has been limited. OBJECTIVE To evaluate the extent of agreement between medication administration records written in paper-based emergency department charts and records generated by an automated dispensing device with regard to the presence or absence of a single, prespecified medication. METHODS Medication administration records in paper-based emergency department charts and medication dispensation records generated by an automated dispensing device were evaluated for concordance. The primary outcome measure was agreement between the 2 sources with regard to the presence or absence of a record for salbutamol by metered-dose inhaler (MDI) for randomly selected patients who presented to a pediatric emergency department with wheeze-related illness from January 1, 2008, to December 31, 2009. RESULTS In total, 1172 patient visits met the inclusion criteria. Of these, records for 1013 visits showed agreement between the paper-based emergency department chart and the dispensation record of the automated dispensing device (kappa = 0.71, 95% confidence interval 0.67-0.75). This value did not meet the target kappa of 0.80. Stratification by time of day, day of week, month, season, or year of presentation at triage or by triage level or disposition (whether or not the patient was admitted to the hospital ward) did not significantly affect the level of agreement between the 2 sources. CONCLUSIONS Agreement between records of salbutamol MDI administration in paper-based charts and dispensation records from an automated dispensing device was substantial, but discrepancies were present. There are significant quality management, legal, clinical, and research reasons to strive for concordance between multiple records with respect to medication use in the emergency department. Data generated by automated dispensing devices have potential value for research, but their strengths and limitations need to be understood.