Methods: Data come from a two-year cluster-randomized trial of 9,166 adult diabetic patients cared for by 78 primary care physicians in 10 practices in Northeast Ohio, all using the Epic© EMR. Five practices (51 PCPs and 5,854 diabetic patients) were assigned to an Epic-facilitated EDS (diabetes related Alerts and other forms of decision support). A self-administered questionnaire was distributed to all experimental physicians after the trial.

Results: To date, 27 of the 51 (53%) physicians have completed the questionnaire. Of these, 89% rated the five ADA Guideline-based Best Practice Alerts (BPA) as at least somewhat helpful. Ninety-six percent felt that the BPAs should remain in the EMR after the Trial. The proportion of physicians who rated each EDS component as at least “somewhat helpful” was 50% (Measures and Medication Tool), 54% (Diabetes Patient List), 64% (Diabetes Practice Profile), and 76% (Measures Over Time Tool). Physicians felt the Practice Profile should remain after the end (88%); 75% felt the Diabetes Patient List should remain; 75% indicated that the Patient List should remain; and 81% felt the Medication Tool should remain. Overall, 85% of physicians felt the Alerts improved their ability to provide quality care to their patients and 75% indicated that Alerts improved their efficiency to care for their diabetic patients. Comments made on the questionnaire about decision support that required the physician to retrieve the support indicated that these would have been more useful if they were automated.

Discussion: These results indicate that physicians find decision support that is automatically triggered within a patient encounter is useful in caring for diabetic patients. A decision support that has to be retrieved is not seen as useful. Designers of decision supports should take into consideration the amount of time and effort physicians need to expend to effectively utilize decision support.