Background In order to secure the drug circuit, health institutions are encouraged to deploy computerised prescriptions. The literature shows that computerised prescriptions generate some risks. To limit these risks, a certification of such software by the Haute Autorité de Santé (HAS) was set up in 2015. However, an important part of the set up of this software remains under the health care institution’s responsibility, especially hospital pharmacists.

Purpose To identify computerised physician order entry (CPOE) system configurations placed under the responsibility of the hospital pharmacist, and to quantify the risks.

Material and methods In our establishment, a multidisciplinary team identified different setting actions of the CPOE system implemented by hospital pharmacists. For each of these actions, the higher risk modalities of failure were identified by estimating the risk priority number (RPN). To do this, on the basis of failure mode and effects analysis (FMEA), failure severity and the possibility of its occurrence and detection were estimated (scoring from 1 to 10). Preventive actions were suggested for those modes of failure with an RPN value >100.

Results Three configuration groups managed by hospital pharmacists were identified: product sheets, alerts and the requirements filled setting. Product sheets setting include the information belonging to the drug formulary, prescription units, administration routes, breakable, procedures for reconstitution/dilution, synonyms and common unit of dispensation (UCD) code. This code allows an interface with external database software, which permits calculation of interactions and contraindication alerts associated with the field. Alerts configuration is to define their perimeter that will be visible for prescribers. The pre-requirements filled in to facilitate the lines capture of complex prescriptions. The FMEA highlighted a criticality high for the following settings: prescriptions pre-filled, the alerts filter definition, the UCD code sheet, prescriptions unit and the reconstruction/dilution terms. The criticality is intermediate for administration route, breakable and drug formulary inscription. It is weak for synonyms.

Conclusion This analysis has led to management measures setting up of a priori risk (validation circuit of configurations, e-learning implementation, risk mapping) and a posteriori (adverse drug events analysis reported in connexion with computerised prescribing, followed by pharmaceutical interventions related to CPOE errors).

No conflict of interest.