Effects of a pharmacy-driven medication history program on patient outcomes
John Marshall, Bryan D. Hayes, Jennifer L. Koehl, William Hillmann, Marjory A. Bravard, Susan Jacob, Rosy Gil, Elisabeth Mitchell, Francesco Ferrante, Jennifer M. Giulietti, Andrea Tull, Xiu Liu, David Lucier
Abstract
PURPOSE: Obtaining an accurate medication history is a vital component of medication reconciliation upon admission to the hospital. Despite the importance of this task, medication histories are often inaccurate and/or incomplete. We evaluated the association of a pharmacy-driven medication history initiative on clinical outcomes of patients admitted to the general medicine service of an academic medical center. METHODS: Comparing patients who received a pharmacy-driven medication history to those who did not, a retrospective stabilized inverse probability treatment weighting propensity score analysis was used to estimate the average treatment effect of the intervention on general medical patients. Fifty-two patient baseline characteristics including demographic, operational, and clinical variables were controlled in the propensity score model. Hospital length of stay, 7-day and 30-day unplanned readmissions, and in-hospital mortality were evaluated. RESULTS: Among 11,576 eligible general medical patients, 2,234 (19.30%) received a pharmacy-driven medication history and 9,342 (80.70%) patients did not. The estimated average treatment effect of receiving a pharmacy-driven medication history was a shorter length of stay (mean, 5.88 days vs 6.53 days; P = 0.0002) and a lower in-hospital mortality rate (2.34% vs 3.72%, P = 0.001), after adjustment for differences in patient baseline characteristics. No significant difference was found for 7-day or 30-day all-cause readmission rates. CONCLUSION: Pharmacy-driven medication histories reduced length of stay and in-hospital mortality in patients admitted to the general medical service at an academic medical center but did not change 7-day and 30-day all-cause readmission rates. Further research via a large, multisite randomized controlled trial is needed to confirm our findings.