By 2010, all prescribers should write and all pharmacies should be able to receive prescriptions electronically. By 2008, all prescribers should have plans in place to implement electronic prescribing. Providers should subject prescriptions to evidence-based, current clinical decision support.

Increasing evidence demonstrates that paper-based prescribing is associated with high error rates (Bates et al., 1995a; Kaushal et al., 2003) and that electronic prescribing is safer (Bates et al., 1998). Electronic prescribing has a number of advantages (Bates et al., 1999; Teich et al., 2005): it eliminates handwriting, helps ensure that the key fields (for example, drug name, dose, route, and frequency) contain meaningful data, and makes it possible to suggest a default dose. More important, however, computerized prescribing enables a range of clinical decision support (Teich et al., 2005), including checks for allergies, drug–drug interactions, overly high doses, clinical conditions, drug–laboratory issues, and pregnancy-related issues, as well as suggestions about dose given the patient’s level of renal function and age.

While all decision-support checks contribute to an effective medication-use system, results of recent studies suggest that dose adjustment may be especially important. It is clear that 10-fold dosage calculation errors in particular are a major clinical issue, especially in pediatrics (Rowe et al., 1998). One inpatient study found that suggesting a dose appropriate for a patient’s level of renal function substantially improved the likelihood that the patient would receive the appropriate dose of medication (Chertow et al., 2001); another demonstrated that alerts decreased the likelihood that patients would receive too high a dose of medication given their level of renal function (Galanter et al., 2005). Another study (Peterson et al., 2005) demonstrated that suggesting an appropriate starting dose of medication for geriatric inpatients improved the likelihood that the recommended daily dose would be prescribed, reduced the likelihood of 10-fold overdose, and was associated with a lower rate of falling.

It is not easy to implement decision supports, however, and problems can arise with all of them. For example, many issues remain to be addressed with regard to allergy checking, although some best practices have been suggested (Hsieh et al., 2004). Drug–drug interactions are especially complex since so many have been identified, but the number that are clinically important is more modest (Hansten et al., 2001; Peterson and Bates, 2001; Glintborg et al., 2005). Fewer data are available regarding the frequency of problems in such areas as diagnoses contraindicating drugs, generic checks for overly high doses, and drug–laboratory decision support.

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