Background and importance We developed a clinical decision support system (CDSS) to monitor high-risk situations related to drugs from electronic health records. It involves clinical rules (CR) that trigger alerts to clinical pharmacists based on drug prescriptions, laboratory values, vital signs, and medical problems (eg, vitamin K antagonists + international normalised ratio (INR) ≥4).

Aim and objectives We describe a methodology to select CR to extend our approach to the paediatric department.

Material and methods CR were identified with a literature review and scored by 14 senior physicians (expert) divided in two groups (A: general/specialised paediatrics; B: neonatology/intensive care) for two criteria: criticality (low, moderate, high, extreme—risk); relevance (no, need to be adapted to be, highly, very—relevant). The pharmacist in charge of CDSS scored CR technical feasibility (no, hardly, easily, very easily—feasible). ‘Very relevant’ and ‘easily feasible’ CR were retained if average criticality score was ‘high’ when applicable for different specialties (assessed by numerous experts) or ‘extreme’ when applicable for a specific specialty (assessed by only one expert).

Results Fifty-six CR potentially relevant for children were selected from the literature and divided into five risk classes: drug contraindicated (34%), medication and abnormal laboratory value (27%), drug–drug interaction (19%), inadequate administration mode (11%) and prescription omission (9%). Twenty-four CR were retained after expert assessment, 8 (33.3%) concerned both groups, 14 (58.3%) were specific for group A and 2 (8.3%) for group B. The three most critical CR involved prescribing potassium and hyperkalaemia, glucose-lowering drugs and hypoglycaemia, and vancomycin not adjusted to renal function. Development in CDSS was assessed as ‘very easily’ feasible for 5 CR (21%) including 3 CR (12.5%) concerning both groups.

Conclusion and relevance We identified 24 CR in five risk classes that could be monitored using our CDSS. Assessment based on expert opinion according to risk (criticality), clinical practice (relevance) and technical consideration (feasibility) allowed CR prioritisation to be developed. One-fifth of CR would be immediately implementable with some likely to cover the entire paediatric department since they are common to both groups. A pilot study using these CR will assess the workload associated with this new practice.

References and/or acknowledgements 1. Fox et al. BJCP 2016.

Conflict of interest No conflict of interest