Background:
Unintentional medication discrepancies during hospitalization can contribute to patient harm. Discrepancies can be reduced by performing medication reconciliation; however, effective implementation is challenging. The goals of the Multi-Center Medication Reconciliation Quality Improvement Study (MARQUIS) were to operationalize best practices for inpatient medication reconciliation and test their effect on potentially harmful medication discrepancies.
Methods:
Five U.S. hospitals participated in a two-year quality improvement (QI) study between September 2011 and July 2014. With the guidance of trained mentors and using standard QI principles, each site implemented at least one of 11 intervention components on some or all medical-surgical inpatient units. A toolkit and supplementary materials such as instructional videos and slide presentations described the design and implementation of these intervention components in detail. Mentors conducted monthly phone calls and two site visits during the intervention period.
The primary outcome was the number of potentially harmful unintentional medication discrepancies per patient. This was determined in approximately 22 randomly selected patients per month at each site during a 6-month baseline period and throughout the intervention. Trained on-site pharmacists took “gold standard” medication histories on these patients, compared these histories to admission and discharge medication orders, and identified and categorized all discrepancies. Trained physician adjudicators at each site determined the potential for harm of all unintentional discrepancies.
To analyze the effect of the entire intervention on the primary outcome, we conducted an interrupted time series analysis using multivariable Poisson regression to detect both sudden improvement with initiation of the intervention and change in the temporal trend after initiation, adjusted for baseline temporal trends and baseline differences between intervention and any control units.
Results:
Across the five participating sites, 1479 patients were enrolled, including 548 patients during the baseline period and 931 patients during the intervention period. Implementation of the intervention as a whole was associated with a reduction in the number of potentially harmful discrepancies over time, beyond any baseline temporal trends: incidence rate ratio 0.89 per month (95% CI 0.80 to 0.99), p=0.03. Of the 4 sites that implemented anywhere from 4 to 7 different intervention components during the study period, 3 sites saw reductions in their potentially harmful discrepancy rate. The site that saw an increase in their discrepancy rate implemented a new electronic medical record shortly after beginning the intervention.
Conclusions:
Adoption of a multi-faceted medication reconciliation quality improvement initiative using a mentored implementation model was associated with a reduction in potentially harmful medication discrepancies over time.