Abstract
Although research has shown that proper management of diabetes can improve outcomes, glucose control is worsening. This partly reflects the failure of providers to intensify diabetes therapy when indicated, termed clinical inertia. Our intervention used (a) decision support reminders which provided patient specific recommendations for management at each visit, and (b) computer generated provider specific feedback on performance. This intervention improved the frequency with which providers intensified the therapy and improved glycemic control.
Problem
Over 20 million Americans have diabetes which is the 6th leading cause of death and the leading cause of kidney failure and new blindness. Although better control can improve outcomes, recent analyses indicate that diabetes control (A1c) levels are getting worse rather than better.
To the extent that clinical inertia is a limiting problem, interventions that improve provider behavior should lead to better diabetes control.
Methods
Data for visits of 4038 patients with type 2 diabetes to 345 physicians over 3 years were recorded with the incoming and outgoing medications in an Oracle database. Their physician providers were randomized either to be controls (usual care), or to receive interventions thought to be capable of changing provider behavior – computer generated reminders, performance feedback, or both.
The decision support reminders included a flowsheet section to show prior clinical data and a recommendations section. The decision support consisted of detailed management algorithms which used each patient’s prescribed medications, and the most recent laboratory and clinical values to recommend specific changes in dose and/or types of medications. Feedback sessions were one-on-one between an endocrinologist investigator and a primary care provider, ~5 minutes long every 2 weeks. The sessions used performance graphs generated by a Visual Basic application which selected provider specific data, passed it to an Excel procedure and printed out the 4 or 5 “report cards” for the current 2 week session. There were more than 40 such reports that reviewed individual provider action or outcomes specific to the patients seen by that provider.
Results
Provider intensification of therapy increased most the first year. It increased more in Feedback groups than in the Reminders and Controls (p<0.001). After 3 years, provider behavior in the non-Feedback groups decayed, whereas improvement with Feedback groups was sustained, at 52% from ~35% initially. In multivariate analysis performance feedback contributed independently to intensification, and intensification independently improved diabetes control (both p<0.001).
Conclusions
This informatics infrastructure made possible compelling, credible performance feedback to primary care providers. This resulted in improved diabetes control that ultimately should decrease complications. Similar approaches can improve provider behavior and diabetes outcomes in other primary care settings.