Clinician preferences for computerised clinical decision support for medications in primary care: a focus group study

Katy E Trinkley, Weston W Blakeslee, Daniel D Matlock, David P Kao, Amanda G Van Matre, Robert Harrison, Cynthia L Larson, Nic Kostman, Jennifer A Nelson, Chen-Tan Lin, Daniel C Malone

To improve user-centred design efforts and efficiency; there is a need to disseminate information on modern day clinician preferences for technologies such as computerised clinical decision support (CDS). To describe clinician perceptions regarding beneficial features of CDS for chronic medications in primary care. This study included focus groups and clinicians individually describing their ideal CDS. Three focus groups were conducted including prescribing clinicians from a variety of disciplines. Outcome measures included identification of favourable features and unintended consequences of CDS for chronic medication management in primary care. We transcribed recordings, performed thematic qualitative analysis and generated counts when possible. There were 21 participants who identified four categories of beneficial CDS features during the group discussion: non-interruptive alerts, clinically relevant and customisable support, presentation of pertinent clinical information and optimises workflow. Non-interruptive alerts were broadly defined as passive alerts that a user chooses to review, whereas interruptive were active or disruptive alerts that interrupted workflow and one is forced to review before completing a task. The CDS features identified in the individual descriptions were consistent with the focus group discussion, with the exception of non-interruptive alerts. In the individual descriptions, 12 clinicians preferred interruptive CDS compared with seven clinicians describing non-interruptive CDS. Clinicians identified CDS for chronic medications beneficial when they are clinically relevant and customisable, present pertinent clinical information (eg, labs, vitals) and improve their workflow. Although clinicians preferred passive, non-interruptive alerts, most acknowledged that these may not be widely seen and may be less effective. These features align with literature describing best practices in CDS design and emphasise those features clinicians prioritise, which should be considered when designing CDS for medication management in primary care. These findings highlight the disparity between the current state of CDS design and clinician-stated design features associated with beneficial CDS.