1064-P: Order Set Design and "Nudges" to Change Care in Diabetes-Towards a Novel Pragmatic Clinical Trial Design

Introduction and Objective: When there are multiple reasonable treatment approaches, electronic health record (EHR) design can 'nudge' providers to choose one: for example, a default drug selected on an order set may be more likely to be used. Our group has proposed use of such nudges in a pragmatic trial design called a Decision Architecture Randomization Trial (DART). Methods: Retrospective analysis of how 3 modifications to an inpatient electronic health record 'nudged' prescribing practices: 1) introduction of an order set to facilitate use of nutritional + correction insulin (as opposed to sliding scale); 2) phased rollout of an order set change to encourage long-acting insulin use by adding a radio button to facilitate ordering; and 3) introduction of default diabetes orders pre-checked in order sets for patients undergoing colorectal surgery. Results: The first nudge increased use of nutrition-correction insulin; the second nudge actually decreased rates of long-acting insulin use; and the third nudge increased glucose monitoring and insulin use for surgery patients (Figure). Conclusion: EHR design offers opportunities to improve care and power innovative randomized trials. These findings support the feasibility of the DART clinical trial design, but research plans should take into account the potential for unpredictable nudge effects.