Four controlled studies (including two time-and-motion studies), one prospective study with a historical control, five before-and-after studies (4 prospective and 1 retrospective) and one retrospective time series evaluated CPOE. One controlled study and six before-and-after studies (5 prospective and 1 retrospective) evaluated ADEs. Three controlled studies and four before-and-after studies (all prospective) evaluated bar coding. The total number of participants was unclear.
Only results for the major outcomes of medication errors and ADEs and the appropriateness of use are reported in this abstract; other outcomes were reported in the review.
CPOE.
Medication errors and ADEs (3 studies): few controlled studies evaluated CPOEs. One prospective before-and-after study found that CPOE reduced preventable and potential ADEs; non-intercepted errors decreased by 55% (P=0.01) and preventable ADEs were reduced by 17% (P=0.37). One retrospective time series found that CPOE reduced 'all major categories' of medication errors: non missed dose errors by 81% (P<0.0001), non-intercepted serious errors by 86% (P<0.0003), and all medication errors by 83%. The ADE rate fell from 14.7 to 9.6 per 1,000 patient days towards the end of the study period, (P<0.09). One prospective before-and-after study (number of patients was not reported) found that CPOE reduced ADEs due to antibiotics from 28 to 4 (P=0.018).
Appropriateness of use (1 prospective before-and-after study): the study found that physicians overrode the CPOE system about 50% of the time. Overriding the system increased the number of drugs prescribed (from 1.5 to 2.7), the duration of therapy (from 103 to 330 hours) and the number of doses (from 11.4 to 27.6).
ADMs.
Medication errors and ADEs (5 studies): all five studies found that ADMs reduced medication errors. One prospective before-and-after study found that AMDs reduced the number of medication errors (10.4% versus 16.9%, P<0.001). Most errors were wrong-time errors. Other studies found that ADMs reduced dispensing errors compared with manually filled prescriptions (0.65% versus 0.84%); reduced error rates compared with the use of traditional cassettes (0.61% versus 0.89%); reduced errors on the cardiovascular surgery unit (from 0.0075 to 0.0058 errors per day, P>0.05), although they increased errors on the cardiovascular intensive care unit (from 0.0051 to 0.0090, P>0.05); and reduced errors compared with a decentralised unit dosing system (10.6% versus 15.9%, P<0.05).
No studies assessed appropriateness of use.
Bar coding.
Medication errors and ADEs (5 studies): the studies found that the error rate was reduced following implementation of bar coding (1 study, 0.2% versus 1.0%); the accuracy of the medical supplies inventory was improved following implementation of bar coding (1 study, P<0.001); and the error rate in data entry was reduced compared with manual entry (3 studies: 0.79% versus 1.53%, P=0.0167; 1.7% versus 5.8%; and 2.63 errors versus 4.48 errors, P<0.0001).
No studies assessed appropriateness of use.
CMARs.
One study was identified but the authors were unable to obtain the publication.