Randomised trial of a diabetes clinical decision support system: use rates, provider satisfaction, and impact on clinical outcomes [abstract]
- View All
Background and aims: CV Wizard is a web-based EHR-integrated point-of-care clinical decision support (CDS) system that presents personalized cardiovascular (CV) risk information to primary care providers (PCP) and patients in both a low numeracy and high numeracy format. Here we report use rates, PCP satisfaction, and impact of the CDS system on clinical outcomes of eligible diabetes patients. Materials and methods: Nineteen primary care clinics with 7035 eligible diabetes patients with high CV risk were randomized to either usual care (UC) or use of the CDS system. This CDS system identifies target high risk patients and provides prioritized and personalized EHR-linked web-based CDS for management of glucose, BP, lipids, tobacco, weight, and aspirin to both the PCP and the patient at the point of care. Consented providers (n=102) were surveyed at baseline and 18 months after implementation, with survey response rates of 92% at baseline and 80% at follow-up. Electronic health record (EHR) data and multilevel regression models tested for differential trends in CV risk among patients at UC and CDS clinics. Results: The CDS system was used at 70-80% of targeted visits made by diabetes patients. Compared to UC, PCPs in the CDS group reported increased use of CV risk calculation while seeing patients (73% vs. 28%, p=.006), being more prepared to discuss CVrisk reduction priorities with patients (98% vs. 78%, p=.03), being more able to provide accurate advice on aspirin use for primary prevention (75% vs. 48%, p=.02), and more frequent discussion of CV risk reduction with patients (60% vs. 30%, p=.06). PCPs at CDS clinics reported that the CDS system improved CV risk factor control (98%), saved time talking to patients about CV risk reduction (93%), efficiently elicited patient treatment preferences (90%), was useful for shared decision making (95%), influenced treatment recommendations (89%), and helped initiate CV risk discussions (94%); 85% of PCPs reported that their patients liked CV Wizard. During the 14-month period when the final CDS intervention was in place, CV risk declined 0.2% per visit for diabetes patients at UC clinics, and 0.6% per visit for diabetes patients at CDS intervention clinics. Conclusion: In a randomized trial, the CV Wizard CDS system was successfully integrated into the workflow of primary care visits with sustained use rates of 70-80% at targeted clinic visits, high PCP satisfaction, perceived positive impact on shared decision making and patient-centered care, and a favorable effect on CV risk factor control.