Testing CDS in OSCAR EMR Using STOPP Criteria

Overview

This mixed methods study seeks to assess and understand the impact of using an advanced CDS tool to present the STOPP criteria to support prescribers at the point of decision-making. We ask the overarching question: \*\*How can clinical decision support tools effectively translate a complex set of evidence-based rules into primary care clinical practice?\*\* The work is divided into 2 phases. PHASE 1: Randomized control trial. We will measure change in potentially inappropriate prescribing by primary care prescribers (e.g. family doctors) who are using the Clinical Decision Support (CDS) tool in their Electronic Medical Record (EMR), as compared to those who are not. Potentially Inappropriate Prescribing is defined by the STOPP criteria, the same criteria that will be used in the clinical decision support tool in the EMR. We will use the OSCAR EMR for this study. EMR based queries will be developed to assess the number of potentially inappropriate prescriptions are current in each of 12 physician practices at baseline and at 16 weeks. PHASE 2: We will then complete a qualitative follow up with participants to aid in understanding how primary care prescribers perceived the CDS. This will help explain Phase 1 findings and provide additional information to enhance future decision support tools.

The STOPP (Screening Tool of Older People's potentially inappropriate Prescriptions) criteria are a previously clinically validated set of suggested medications to avoid in patients aged 65+ with certain characteristics (e.g. benzodiazepines should be avoided in those with a fall history). These have previously been translated into point of care reminders that can be used with the OSCAR Electronic Medical Record (EMR) so that physicians are notified of potentially inappropriate prescriptions when a patient's chart is opened. This two phase explanatory study will examine if those STOPP based point of care prescribing alerts as presented to physicians in the OSCAR EMR can change potentially inappropriate prescribing patterns in primary care physicians. PHASE 1: A cluster randomized control trial in private primary care physician offices. The intervention is at the level of the physician and consists of activating the set of STOPP based alerts for the physician so that they see them during their workflow. These alerts will be triggered and displayed to physicians at the point of care. The alerts will suggest the physician consider Potentially Inappropriate Prescriptions (PIPs), as defined by STOPP. The control group will not have the alerts presented. (STOPP criteria are described in the full proposal, attached). The primary outcome measure is a composite indicator: * Δ in Number of PIPs in patients \>64yo with a visit during the intervention Number of patients \>64yo with a visit during the intervention Phase 1 Data Analysis: The outcomes will be measured using the University of British Columbia (UBC) Department of Family Practice's research network infrastructure. The network has been designed and successfully tested in a previous study (UBC ETHICS H13-02531). This network connects to OSCAR EMRs and, with permission from the physician, is able to answer practice level questions WITHOUT collecting patient level data. The practice level queries for potentially inappropriate prescriptions will be distributed to the practices through the network and only the aggregate answers (not patient level data) will be returned for analysis. For example, one of the STOPP recommendations is to not use a non-cardioselective beta blocker in patients with Chronic Obstructive Pulmonary Disease (COPD). The research network would be able to query the physician practices before and after the intervention and receive counts back on the number of patients with Chronic Obstructive Pulmonary Disease (COPD) who are also on a non-cardioselective beta-blocker. PHASE 2: Will seek to qualitatively explain the findings from phase 1 with the physician participants. This will consist of one to two, two-hour, semi-structured discussion group(s) with family physicians from the intervention group. The discussion group will seek to understand specific issues with the Phase 1 tool. The discussion group will move through the following steps: (a) consent; (b) participants provide overall impressions of the tool; (c) review preliminary results from Phase 1; (d) specific feedback on how to improve the tool. The focus group will be audio recorded and qualitatively analyzed through the lens of pragmatic content analysis to better understand perceptions of the Clinical Decision Support (CDS) tool and its integration into workflow. There may be 2 discussion groups to allow for broader participation if scheduling is an issue and it may be a remote discussion group, depending on location of physician participants.