1005 — The Impact of a Quality Improvement Project on Statin Prescribing in Primary Care: Quality Improvement and Personalization for Statins (QUIPS)

Lead/Presenter: Jeremy Sussman, COIN - Ann Arbor
All Authors: Sussman JB (VA Ann Arbor Healthcare System) Holleman R () Youles B () Lowery J () Hofer T () Kerr EA ()

Objectives:
Recent clinical practice guidelines, including those by a joint VA/DoD Committee, changed recommendations for the use of the cholesterol-lowering statin drugs to be based on cardiovascular risk, rather than cholesterol level. This is a substantial change, in part because providers do not have automated information about cardiovascular risk. Therefore, we developed an intervention to improve statin use that included individualized clinical decision support. We examined its effect in the VA Ann Arbor Healthcare System.

Methods:
We performed a parallel-design, pragmatic, cluster randomized, controlled trial with clustering by primary care team and provider. The population was patients who did not meet the 2014 VA/DoD statin guidelines. The intervention was a quality improvement intervention with personalized clinical decision support, provider education, and audit and feedback. Our analysis used hierarchical logistic regression with patient visits clustered within 47 providers clustered within 5 clinical teams, adjusting for baseline treatment rate. The primary outcome measure was initiation of moderate or high potency statins. Secondary analyses included prespecified clinical subgroups, the intensity of statin use, and reasons for non-use.

Results:
The final analysis included 7,980 visits before the intervention and 8,974 afterwards among 45 providers. At baseline, intervention providers were less likely to provide statins to eligible patients than providers in the control group (0.058 vs. 0.089, = 0.04). On adjusted analysis, the intervention increased guideline-concordant prescribing (odds ratio of receiving change 2.2 +/- 0.39, p < 0.001) The clinical reminder influenced care more when the patient was calculated to be at high risk of cardiovascular disease (and hence recommended for statin use) and less when s/he was intermediate risk (and recommended neither for nor against statin use), compared to the impact of patients with diabetes and a history of heart attack or stroke, for which calculations were unnecessary.

Implications:
A relatively simple and scalable intervention improved statin prescribing. It was most influential for patients where the intervention helped providers meet guidelines.

Impacts:
Simple implementation programs with decision support can help implement risk-based prescribing. This will gain in importance as big data and risk prediction become more central to medical practice.