Medication errors: comparison of electronic and hand-written prescribing in the ICU

The most common stage of medication error was reported in one study to be the prescribing stage, which accounted for 56% of errors detected [1]. Electronic prescribing (EP) with decision support has been shown to reduce the medication error rate [2], but no studies could be found in critical care. This study compares the impact of this change on the medication error rate before and after the implementation of the GE Systems QS 5.6 clinical information system (CIS), which does not have decision support.

During the study periods, all medication errors identified by the ICU pharmacist were recorded. Errors were identified using a published definition [3], except abbreviations of drug names were not regarded as errors. ICU staff were unaware that the study was taking place. The location was a 22-bed general ICU/HDU at a teaching hospital. Data were collected in 2002 for two periods of 9 days in total before introduction and for four periods (17 days in total) on weeks 2, 10, 25 and 37 after CIS introduction. The total number of drugs prescribed was recorded.

There was a statistically significant reduction in the medication error rate following the introduction of CIS. The error rate before CIS was 6.7% (69 errors from 1036 prescriptions) and after CIS introduction was 4.7% (115 errors out of 2429 prescriptions) (χ² = 5.34, one degree of freedom [df], P < 0.03). There was variation of the error rate with EP over time (χ² = 21.7, three df, P < 0.001) as the staff got used to the new system and prescribing systems were improved. There is also strong evidence of a linear trend (χ² = 11.9, one df, P < 0.001). Thus the error rate appeared to reduce with time with EP.

In conclusion, introduction of the CIS coincided with a reduction in the overall medication error rate, with some suggestion of a 'learning curve'.

Figure 1