Dematerialized drug-related problems (DRPs) interventions might be provided by clinical decision support systems (CDSS) to allow clinical pharmacists (CP) to intervene on the most complex situations. Moreover, this would help prioritize pertinent information to display to the physician to avoid alert fatigue.

To determine which DRPs could be dematerialized and integrated into CDSS and to assess the feasibility of dematerialization.

Data were collected over a six months prospective study, conducted in internal medicine wards. Identified DRPs were confronted to a beforehand developed classification of CDSS in order to allocate them in different CDSS classes. A score of implementation complexity was developed to determine the feasibility of dematerialization in the CDSS. This score was divided in 3 parts each scored from 1 to 4: technical, axes of CDSS; accessibility, data type; organizational, impact on population affected in institutions. It was stratified in 3 groups of complexity of implementation: low level: [3–6 points[; medium level: [6–9[; high level: [9–12].

Three hundred and eighty-three DRPs were highlighted by the CP whose 32 were excluded from this study (main reason: data loss). Dematerialization of 20/351 (5.7%) DRPs were classified as impossible (e.g. drug-drug interaction (DDI) for more than 2 drugs). The remaining 331 could technically be integrated into CDSS and among them, 89/331 (26.9%) were related to the co-medications (e.g. 78 related to DDI), and 196/331 (59.2%) to the patient (e.g. 44 targeting untreated indication).

According to the score of implementation complexity, DRPs were classified from the lowest to the highest level of implementation complexity as follows: 40% low (e.g. co-medication), 24% medium (e.g. drug-lab interaction) and 35% high complexity of dematerialization (e.g. drug-lifestyle interaction).

The CP realized 146 interventions on the 331 DRPs, with a rate of 50% for low, 62% for medium and 44% for high levels of complexities. The rate of physician's acceptance was comparable for the medium (84%) and low (85%) levels and the best rate was for the highest complexity level (94%).

Most of CP's interventions could theoretically be replaced by implementation of CDSS. However, even when a dematerialization is possible, the real pertinence of the alerts is not guarantee. Indeed, in many situations, information related to the decision-making is still insufficient to generate alerts considering weighing of interests. This point merits to be evaluated in further research.