Primary Care Provider Use Rates of a Clinical Decision Support Tool and Change in Diabetes Performance Measures

Authors

JoAnn Sperl-Hillen, GHI S&B for HealthPartners Institute for Education and Research, Bloomington, MN
Stephen Asche, GHI S&B for HealthPartners Institute for Education and Research, Bloomington, MNFollow
Darin Ruanpeng, GHI S&B for HealthPartners Institute for Education and Research, Bloomington, MN
Heidi Ekstrom, GHI S&B for HealthPartners Institute for Education and Research, Bloomington, MN
Patrick O'Connor, GHI S&B for HealthPartners Institute for Education and Research, Bloomington, MNFollow

Publication Date

8-10-2017

Keywords

primary care, diabetes

Abstract

Background: A previously published randomized controlled trial demonstrated that implementation of an electronic health record-linked personalized clinical decision support (PCDS) tool within primary care clinics improved mean A1c and blood pressure control. We subsequently implemented a modification of our CDS tool for expanded use with high cardiovascular risk adults (CV-PCDS) that also retained the decision support for glycemic control for patients with diabetes. Herein we analyze the association between primary care provider use rates of CV-PCDS with diabetes performance measures in patients with diabetes.

Methods: Using data from a cluster randomized trial in 2012–2014, we analyzed the association of CV-PCDS provider-specific use rates in March 2014 with diabetes performance measures 6 months later, using Pearson correlation coefficients. Performance measures included the proportion of a provider’s diabetes patients who (a) achieved A1c < 8%, and (b) achieved a composite measure of optimal diabetes care that required simultaneous achievement of A1c < 8%, systolic blood pressure < 140 mm Hg, low-density lipoprotein cholesterol < 100 mg/dl, non-tobacco user, and ASA (aspirin) use for secondary prevention.

Results: Providers (N = 43) used the CV-PCDS tool at a mean of 82.1% of targeted encounters of adults with high cardiovascular risk (range across providers: 36.0% to 100% of encounters). The mean percentage of the diabetes subgroup who achieved A1c < 8% was 73.7%, and the percentage of patients who achieved the optimal diabetes care goal was 46.8%. Pearson correlation coefficients between CV-PCDS provider use rates in March 2014 and A1c and optimal diabetes care performance measures in August 2014 were 0.16 (P = 0.31) and 0.24 (P = 0.12), respectively.

Conclusion: In this high-performing health care system with high CV-PCDS use rates, there was a positive but nonsignificant association of provider use of the CV-PCDS tool and provider-level quality of diabetes care 6 months later. The generalizability of this finding to lower-performing care systems, and to providers with lower baseline quality of diabetes care remains to be determined.

Recommended Citation

Sperl-Hillen J,Asche S, Ruanpeng D, Ekstrom H, O'Connor P. Primary care provider use rates of a clinical decision support tool and change in diabetes performance measures. J Patient Cent Res Rev. 2017;4:155.