Introduction
US studies suggest that prescribing errors occur in 0.4 to 1.9% of all medication orders^1,2,3 and cause harm in about 1% of inpatients ⁴. However, there are no corresponding data for UK hospitals. A further problem is that the definition of a prescribing error used is often not given, or is inadequate or ambiguous. The objective of this study was to measure the incidence of prescribing errors in a UK hospital, using a practitioner-led definition of a prescribing error.

Methods
A definition of a prescribing error was developed using a two-stage Delphi process with a panel of 34 judges; these included prescribers, nurses, pharmacists and risk managers. Pharmacists in a large teaching hospital were then asked to record details of all prescribing errors identified in non-obstetric inpatients during a one month period, using the definition developed. The number of medication orders written during this time was estimated by examining the medical records of a 1 in 5 sample of inpatients. Prescribing errors were classified according to the stage of the drug use process and the stage of patient stay in which they occurred, as well as whether they originated in the prescribing decision or in medication order writing. Clinical significance was assessed by the investigator and a senior clinical pharmacologist.

Results
A general definition of a prescriber error was agreed upon, together with more detailed guidance concerning the types of events that should be considered errors. A total of 538 prescribing errors were identified during the study period; about 36,700 medication orders were written during this time. A prescribing error therefore occurred in 1.5% of all medication orders written. The majority (54%) occurred in the 'select dose regimen' stage of the drug use process. When analysed according to stage of patient stay, errors occurred in 1.3% of all medication orders written on admission, 1.8% written during the remainder of patient stay, 1.0% of re-written medication orders and 1.3% of those written at discharge (p < 0.001; chi-squared). Most (61%) originated in medication order writing, the remainder in the prescribing decision. Twenty-six percent were considered potentially serious. Of the potentially serious errors, 42% originated in medication order writing and 58% in the prescribing decision.

Conclusions
Prescribing errors occurred in 1.5% of medication orders written; serious prescribing errors in 0.4%. Errors were most likely to occur in new medication orders written during a patient's stay; error rates for medication orders written on admission, rewritten medication orders and discharge medication were also high considering that these mainly involve transcription. Although most prescribing errors originate in medication order writing, the majority of serious errors originate in the prescribing decision. These findings will be used together with qualitative work exploring prescribing errors' causes to suggest ways in which their incidence can be reduced.

References

1. Folli HL et al (1987). Medication error prevention by clinical pharmacists in two children's hospitals. Pediatrics 79, 718-722.
2. Blum KV et al (1988). Medication error prevention by pharmacists. AJHP 45, 1902-1903.
3. Lesar TS et al (1997). Factors related to errors in medication prescribing. JAMA 277, 312-317.
4. Bates DW et al (1995). Incidence of adverse drug events and potential adverse drug events: Implications for prevention. JAMA 274, 29-34.