Non-intercepted dose errors in prescribing anti-neoplastic treatment: a prospective, comparative cohort study.

BACKGROUND The incidence of non-intercepted prescription errors and the risk factors involved, including the impact of computerised order entry (CPOE) systems on such errors, are unknown. Our objective was to determine the incidence, type, severity, and related risk factors of non-intercepted prescription dose errors. PATIENTS AND METHODS A prospective, comparative cohort study in two clinical oncology units. One institution used a CPOE system with no connection to the electronic patient record system, while the other used paper-based prescription forms. All standard prescriptions were included and reviewed. Doses were recalculated according to the guidelines of each institution, using the patient data as documented in the patient record, the paper-based prescription form, or the CPOE system. A non-intercepted prescription dose error was defined as ≥10% difference between the administered and the recalculated dose. RESULTS Data were collected from 1 November 2012 to 15 January 2013. A total of 5767 prescriptions were evaluated, 2677 from the institution using CPOE and 3090 from the institution with paper-based prescription. Crude analysis showed an overall risk of a prescription dose error of 1.73 per 100 prescriptions. CPOE resulted in 1.60 and paper-based prescription forms in 1.84 errors per 100 prescriptions, i.e. odds ratio (OR) = 0.87 [95% confidence interval (CI) 0.59-1.29, P = 0.49]. Fifteen different types of errors and four potential risk factors were identified. None of the dose errors resulted in the death of the patient. CONCLUSIONS Non-intercepted prescribing dose errors occurred in <2% of the prescriptions. The parallel CPOE system did not significantly reduce the overall risk of dose errors, and although it reduced the risk of calculation errors, it introduced other errors. Strategies to prevent future prescription errors could usefully focus on integrated computerised systems that can aid dose calculations and reduce transcription errors between databases.