across the 35 homes). Using observation of 300 doses, Cooper and colleagues (1994) found a 6 percent administration error rate in one 300-bed nursing home during the baseline evaluation, using observation of 300 doses prior to implementation of an automated system.
Wrong-time error is a significant problem in residential care settings. An error rate of 27 percent in assisted-living settings was reduced to 15 percent when a 4-hour interval (as opposed to a 2-hour interval) around the scheduled time was used to designate on-time administration (Young et al., 2005). Four types of error were observed: wrong time (43 percent of errors), wrong dose (30 percent), omitted dose (10 percent), and unauthorized drug (10 percent). Wrong-time errors were even more prevalent in nursing homes, where the error rate decreased from 35.6 to 6.7 percent when wrong-time errors were excluded (Scott-Cawiezell et al., 2005).
Errors identified by observation of drug administration, which detects errors in transcription as well as administration, are described in the section below on administration. A study of discrepancies in medication orders on documents (medication administration record, hospital summary, and discharge orders) accompanying 20 newly admitted residents at the time of transfer from a Veterans Administration hospital to a Veterans Administration nursing home revealed at least one medication discrepancy for every subject (Siple and Joseph, 1992). Discrepancies were found in medication name, medication dose, omitted or added medications, and instructions for use. Investigators attributed 75 percent of the discrepancies to error and 25 percent to intentional changes.
Although the committee could identify no studies focused specifically on monitoring errors, Gurwitz and colleagues (2005) pointed out that the high rate of preventable ADEs (4.1 per 100 patient months) identified in their cohort study of long-term residents of two academic nursing homes argued for a special focus on ordering and monitoring. Errors occurred at the monitoring stage in 80 percent of the preventable ADEs. The most common monitoring errors were inadequate monitoring and failure to act on monitoring.
Few studies directly compare error rates across the stages of the medication-use process. In such a comparison, the method of error detection will