It is also essential that any decision support be practical. Overalerting is a frequent and important problem (Teich et al., 2005), especially for drug– drug interactions. To avoid this problem, it would be helpful if decision support rules were available in a publicly available location. AHRQ should consider approaches for developing a database to which organizations could contribute decision-support rules, expressed in a standard format, that could then be accessed by interested parties. This database would require periodic external vetting to ensure that it included only appropriate rules and to update the decision-support knowledge base. The need for such quality checks is illustrated by ISMP’s recent audit of pharmacy decision support, which found a very high rate of deficiencies and no improvement over a 6-year period (ISMP, 2005b).

An important adjunct to electronic prescribing is that all pharmacies should be able to receive prescriptions in coded form—a much lower-risk method than current paper or oral approaches (Bates, 2001), which are error-prone and require transcription and verification. A commonly used approach in the outpatient setting, for example, is to call prescriptions in to the pharmacy. These prescriptions are frequently left on voice mail. This approach, while efficient in some respects, has several limitations: there is no possibility of readback; if there is a problem, the pharmacist must contact the prescriber later; and the prescription cannot be checked at the time it is delivered to the pharmacy. Similarly, in the reverse direction, most communication by the pharmacist to the prescriber’s office must be left on voice mail, and sometimes the prescriber’s staff do not respond appropriately to queries requiring a clinical response. At the same time, a number of issues must be addressed for electronic transmission of prescriptions to be practical. Many of these issues are regulatory. For example, a number of states have laws that preclude the practice, particularly for narcotics, although there is no evidence that handwritten prescriptions are safer. In other situations, pharmacies can decide whether they will accept electronic transmission, a situation that creates substantial problems for providers attempting to implement safer prescribing practices.

It is also important to recognize that any technology can induce new errors as well as prevent them, and that computerization of prescribing thus does not represent a panacea (Koppel et al., 2005). When any intervention is introduced, it must be monitored, problems it creates must be identified, and appropriate changes must be made in the application and underlying databases to eliminate these problems. Typically, insufficient resources and energy are dedicated to this process, yet results of human factors research clearly demonstrate that it is more efficacious than training staff to work around difficulties (Gosbee, 2004).

Finally, to achieve the desired safety benefits, electronic prescribing must include basic clinical decision support (Bates et al., 1999; Gandhi et



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