Impact of a computerized physician order entry system on compliance with prescription accuracy requirements.

OBJECTIVE: To assess the change in non-compliant items in prescription orders following the implementation of a computerized physician order entry (CPOE) system named PreDiMed. SETTING: The department of internal medicine (39 and 38 beds) in two regional hospitals in Canton Vaud, Switzerland. METHOD: The prescription lines in 100 pre- and 100 post-implementation patients' files were classified according to three modes of administration (medicines for oral or other non-parenteral uses; medicines administered parenterally or via nasogastric tube; pro re nata (PRN), as needed) and analyzed for a number of relevant variables constitutive of medical prescriptions. MAIN OUTCOME MEASURE: The monitored variables depended on the pharmaceutical category and included mainly name of medicine, pharmaceutical form, posology and route of administration, diluting solution, flow rate and identification of prescriber. RESULTS: In 2,099 prescription lines, the total number of non-compliant items was 2,265 before CPOE implementation, or 1.079 non-compliant items per line. Two-thirds of these were due to missing information, and the remaining third to incomplete information. In 2,074 prescription lines post-CPOE implementation, the number of non-compliant items had decreased to 221, or 0.107 non-compliant item per line, a dramatic 10-fold decrease (chi(2) = 4615; P < 10(-6)). Limitations of the computerized system were the risk for erroneous items in some non-prefilled fields and ambiguity due to a field with doses shown on commercial products. CONCLUSION: The deployment of PreDiMed in two departments of internal medicine has led to a major improvement in formal aspects of physicians' prescriptions. Some limitations of the first version of PreDiMed were unveiled and are being corrected.

A qualitative systemic analysis of drug dispensing in Swiss hospital wards.

OBJECTIVE: The traditional floor stock drug distribution system entails that, in most Swiss hospitals, drugs are prepared in a pillbox for each patient by nurses before administration. The aim of this study was to analyse and evaluate the process of inpatient pillbox dispensing in Switzerland. SETTING: Internal medicine wards in three regional hospitals (32-39 beds). METHOD: A qualitative exploratory study with 4 days of participant observation and three semi-structured interviews on each of the three wards. MAIN OUTCOME MEASURE: Thematic content analysis of observations, casual conversations and semi-structured interviews. RESULTS: Pillboxes are prepared in a somewhat tumultuous context. The space for preparation is not isolated from health carers' other activities which makes it prone to numerous interruptions. Drug blisters and tablets often lack the necessary inscriptions to identify them as unit doses. Storage places are small and unstructured. Nurses have to translate medical orders into the name and dosage of the hospital formulary drugs. Use of patients' own drugs hampers both storage and preparation. Nurses take easily a task over one from another, which affects continuity. Practices vary and few standard rules have been set. The position of nurses in the medication process makes them routinely compensate for failures and gaps from healthcare workers involved in the upper stream of their tasks, without any opportunity of giving some feedback. CONCLUSION: The preparation of a pillbox is error prone not only because single doses of oral drugs are not identifiable in a safe way, but also because of various context-related factors. The latter can be grouped into four categories that are linked in a dynamic relationship: tools, practices, organization and information transmission. Improvement strategies should consider all four categories as well as the specific position of nurses in the medication process which makes them fundamental guarantors of the system resilience.