Automated Medication-Dispensing Systems

Four studies evaluated automated medication-dispensing devices. In the only randomized trial, now more than 20 years old, a bedside automatic medication-dispensing machine was associated with a statistically significant reduction in medication error rate from 15.9 percent within the control group (a decentralized unit dose system) to 10.6 percent within the intervention group (Barker et al., 1984). In another study, implementation of an automated drug-dispensing system led to a reduction in medication errors, largely those related to time of administration, from 16.9 to 10.4 percent (p <0.001) (Borel and Rascati, 1995). This result is consistent with that of a later study demonstrating that the introduction of an automated medication-dispensing device led to an increase in the number of medications administered as scheduled from 59 to 77 percent of doses (p = 0.02) (Shirley, 1999). The remaining study reported on the introduction of automated medication-dispensing devices into a cardiovascular surgery unit and a cardiovascular ICU. Medication error rates decreased for patients in the surgical unit but increased for patients in the cardiothoracic unit; neither difference was statistically significant (Schwarz and Brodowy, 1995).

The AHRQ-funded review of patient safety practices concluded that the evidence provided by the above studies does not support the use of automated dispensing devices to reduce medication errors (Shojania et al., 2001).

Other Studies

National disease registries have become an important mechanism for correcting both under- and overprescribing of medications in hospitals for certain important groups of patients—those with ST elevation and non–ST elevation acute coronary syndrome and acute heart failure (Ferguson et al., 2003; Peterson et al., 2004; Jha et al., 2005).

A randomized control trial in a cardiac surgical ICU tested the efficacy of smart IV infusion pumps (incorporating an integrated decision-support system)—the intervention period. In the control period, the decision-support software was inactive. Although many errors were found that would not otherwise have been detected, the rates of serious medication errors in the control and intervention periods were not different (Rothschild et al., 2005).

A randomized controlled trial evaluated a continuous quality improvement initiative designed to increase the use of preoperative beta-blockers in patients undergoing coronary artery bypass graft surgery (Ferguson et al., 2003). The intervention included a call to action to a physician leader at the study site; educational products; and periodic longitudinal, site-specific feed-



The National Academies | 500 Fifth St. N.W. | Washington, D.C. 20001
Copyright © National Academy of Sciences. All rights reserved.
Terms of Use and Privacy Statement