Peer-Reviewed Journal Details
Mandatory Fields
Dalton, Kieran; O'Mahony, Denis; Cullinan, Shane; Byrne, Stephen
2020
July
Drugs & Aging
Factors affecting prescriber implementation of computer-generated medication recommendations in the SENATOR trial: A qualitative study
Published
Optional Fields
Prescriber implementation Medication recommendations
Background: The SENATOR trial intervention included the provision of computer-generated medication recommendations to physician prescribers caring for hospitalised older adults (≥ 65 years), with the aim of reducing in-hospital adverse drug reactions. Interim data analysis during the trial revealed that the prescriber implementation rates of the computer-generated STOPP/START recommendations were lower than expected across all six trial sites. Aim: The aim of this qualitative study was to identify the factors affecting prescriber implementation of the medication recommendations in the SENATOR trial. Methods: Semi-structured interviews were conducted with trial researchers and physician prescribers who were provided with SENATOR recommendations. Content analysis was used to identify the most relevant domains from the Theoretical Domains Framework (TDF) that affected recommendation uptake. Results: Ten trial researchers and fourteen prescribers were interviewed across the six trial sites. Eight TDF domains were found to be most relevant in affecting prescriber implementation: ‘environmental context and resources’, ‘goals’, ‘intentions’, ‘knowledge’, ‘beliefs about consequences’, ‘memory, attention and decision processes’, ‘social/professional role and identity’, and ‘social influences’. Interviewees felt that there was often a disconnect between the time prescribers were reviewing the patient and the point at which the recommendations were provided. However, when recommendations were reviewed, prescriber inertia was highly pervasive, with a particular reluctance to make pharmacotherapy changes outside their own specialty. Implementation was facilitated by recommendations reaching a ‘decision-maker’, but this was often not possible as the software could not evaluate the entire clinical context of patients, and thus frequently produced recommendations of low clinical relevance.
1170-229X
https://doi.org/10.1007/s40266-020-00787-6
10.1007/s40266-020-00787-6
Grant Details