cation errors in the selection and procurement of drugs by the hospital pharmacy.

Medication errors occur in all stages of the medication-use process, most frequently at the prescribing and administration stages. Several U.S. studies using differing definitions of error and methods of error identification found that rates of prescribing errors for adults in hospitals varied considerably (see Table 3-1). Prescribing errors occurred at rates of 12.3– 1,400.00 per 1,000 patient admissions (Bates et al., 1995a; Lesar et al., 1997; Lesar, 2002; LaPointe and Jollis, 2003; Winterstein et al., 2004). Such errors occurred at rates of 0.6–53.0 per 1,000 orders (Lesar et al., 1990; Bates et al., 1995a; Lesar et al., 1997; Lesar, 2002). And in studies that evaluated prescribing errors per opportunity for error, rates of 1.5–9.9 per 100 opportunities were found (Dean et al., 2002; van den Bemt et al., 2002; Bobb et al., 2004; Lisby et al., 2005).

Errors rates depend on the thoroughness of the error detection methods that are used (Gandhi et al., 2000). Most of the above studies used less comprehensive error detection methods, such as spontaneous reports by pharmacists after review of written orders (Lesar et al., 1997; Lesar, 2002), prompted reporting (Winterstein et al., 2004), and reporting by a clinical pharmacist participating in patient care (LaPointe and Jollis, 2003). The study that found by far the highest rate (Bates et al., 1995a) used much more comprehensive detection methods—chart review, including review of written medication orders by a dedicated trained reviewer, in addition to prompted reporting from nurses and pharmacists. This study found a rate of 1,400 prescribing errors per 1,000 patient admissions or 0.3 prescribing errors per patient per day. Of the errors identified, 7.5 percent were adjudged serious—preventable or potential ADEs. By comparison, a study (Kaushal et al., 2001) using similar error detection methods in pediatric units identified 405 prescribing errors per 1,000 patient admissions or 0.1