sional shadowing the nurse who administers medications, recording the medications prepared and administered, and comparing this information with what the prescriber ordered. Any discrepancy between what the patient received and what the prescriber ordered is an administration error. These data have been used to evaluate the accuracy of the output of the entire medication distribution system—whether the patient received the right drug, dose, form, and route—from the patient’s point of view (Barker et al., 2002). An important advantage of observation over voluntary reporting is that it does not rely on the heath care provider’s being aware of the error (providers typically do not realize they have made an error, and if they do, they may be reluctant to report it). Observation is best performed by nurses or pharmacists, although less costly resources, such as pharmacy residents, pharmacy or nursing students, or experienced pharmacy technicians, can also be used. The average cost of observation per dose was measured in one study as $6.65 when performed by a registered nurse and $4.56 by a licensed practical nurse (Flynn et al., 2002). Observation has been recommended for studying ADEs as well (Rothschild et al., 2005).
To derive the benefits of the lessons that can be learned from observation while keeping the process affordable, observation is best conducted for limited or periodic studies of settings of interest. The number of observations depends on the goal of the study, ranging from 100 per nursing unit over a day or two to see whether there is an error problem to over 1,000 for evaluations of technology effects. A hospital might conduct an observation on several nursing units selected as “typical” every few months to learn from the errors detected and determine whether there is a serious error problem. This process would be part of the hospitals routine quality monitoring program.
Detection of medication errors and ADEs in ambulatory care settings is a fairly recent development. For example, incident reports and review of patient records have been used to study ADEs among elderly ambulatory patients (Gurwitz et al., 2003). Medication dispensing errors on prescriptions filled in community pharmacies have been studied using a double-check by an independent observer pharmacist (Flynn et al., 2003); however, the standard in pharmacies is to rely on self-reports of errors detected by patients who notify the pharmacist. National databases of voluntary reports contain few error reports from the ambulatory setting, in part because of unawareness of such errors and in part because pharmacies do not want this information reported to external organizations. With the enactment of