(1982) demonstrated relatively high rates of medication administration errors in nursing homes and the potential of bar coding to decrease these rates, although controlled trials of this technology have not been conducted in this setting.

In the outpatient setting, electronic prescribing will be important (Gandhi et al., 2003), although evidence to date for its effectiveness in this setting is limited, and electronic prescribing without associated decision support is unlikely to yield the potential safety benefits (Gandhi et al., 2005). Indeed, it may be more important to improve communication between patients and providers; the available evidence suggests that many ADEs might have been prevented or ameliorated had communication occurred earlier in the medication-use process (Gandhi et al., 2003). In this regard, personal health records that are linked to provider EHRs represent one attractive approach that deserves further evaluation (Katz et al., 2004). Other technologies, such as on-line communications, also warrant further investigation. In addition, automation may be useful in pharmacies to improve the likelihood that prescriptions will be filled accurately, and to free pharmacists to do more counseling with patients, which too often does not occur today.

The adoption of CPOE with computerized decision support has been slow (Kaushal et al., 2005). High upfront capital costs and the difficulty of demonstrating the financial benefits have been major barriers to the adoption of CPOE. A recent study demonstrated that the investment in CPOE with decision support at the Brigham and Women’s Hospital, Boston, Massachusetts, has resulted in substantial operating budget savings (Kaushal et al., 2006). Over the period 1993–2002, the hospital invested $11.8 million to develop, implement, and operate a CPOE system and achieved net operating budget savings of $9.5 million. The majority of the savings were derived from a relatively small number of interventions. The annual savings generated in 2002 dollars were from renal dosing guidance ($2.24 million), ADE ($1.05 million), improved nursing time utilization ($0.96 million), and specific/expensive drug guidance ($0.88 million).

A key lesson from the implementation of CPOE with computerized decision support at Brigham and Women’s Hospital is that hospitals should focus initially on a small number of high-impact interventions (for example, renal dosing guidance, ADE prevention, and specific/expensive drug guidance). There are other high-impact interventions not implemented at Brig-