medications. These estimates were derived from health plan administrative data (Solberg et al., 2004). A retrospective analysis of data from the 1995– 2000 National Ambulatory Medical Care Survey and National Hospital Ambulatory Medical Care Survey found inappropriate drug–drug combinations in 0.74 percent of visits involving two or more prescriptions and inappropriate drug–disease combinations in 2.58 percent of visits involving at least one prescription (Zhan et al., 2005).

Concerns about labeling have been researched for sample medications dispensed from the ambulatory care setting. In one study (Dill and Generali, 2000), involving 35 frequently used sample medications from 16 different manufacturers with nine drug classifications, information on the usual dosage was not present on 12 percent of the labels evaluated; 17 percent gave dosage and frequency; and 9 percent gave dosage, route, frequency, and duration. The remaining 62 percent referred the user to enclosed prescribing information, which in 27 percent of cases was not in fact enclosed.

The committee identified only one study of medication monitoring in an ambulatory care setting. In a retrospective chart review of 400 outpatients being treated with levothyroxine at a large North American tertiary care hospital, only 56 percent of patients were found to have received minimum monitoring based on criteria derived from the literature and established through expert consensus (Stelfox et al., 2004). Those patients who received the recommended monitoring had fewer levothyroxine-related ADEs than those who did not (1 percent versus 6 percent).

Three studies have examined the rate of medication discrepancies in the outpatient medical record. A study in an outpatient geriatric center found that 0.37 of current medications per patient (43 medications/117 patients) were missing from the patient record, and 0.38 of medications per patient (44 medications/117 patients) were included in the record but were not currently being taken by the patient (Wagner and Hogan, 1996). About a third of these errors were judged to have been caused by patients who misreported a medication at a previous visit or changed (stopped, started, or dose-adjusted) a medication between visits. A study carried out in a private practice affiliated with an academic center, involving 312 patients from the practices of five cardiologists and two internists, found that 0.89