Background: Although clinical trials have shown that proper management of diabetes can improve outcomes, and treatment guidelines are widespread, glycated hemoglobin (HbA1c) levels in the United States are rising. Since process measures are improving, poor glycemic control may reflect the failure of health care providers to intensify diabetes therapy when indicated--clinical inertia. We asked whether interventions aimed at health care provider behavior could overcome this barrier and improve glycemic control.
Methods: In a 3-year trial, 345 internal medicine residents were randomized to be controls or to receive computerized reminders providing patient-specific recommendations at each visit and/or feedback on performance every 2 weeks. When glucose levels exceeded 150 mg/dL (8.33 mmol/L) during visits of 4038 patients, health care provider behavior was characterized as did nothing, did anything (any intensification of therapy), or did enough (if intensification met recommendations).
Results: At baseline, residents did anything for 35% of visits and did enough for 21% of visits when changes in therapy were indicated, and there were no differences among intervention groups. During the trial, intensification increased most during the first year and then declined. However, intensification increased more in the feedback alone and feedback plus reminders groups than for reminders alone and control groups (P<.001). After 3 years, health care provider behavior in the reminders alone and control groups returned to baseline, whereas improvement with feedback alone and feedback plus reminders groups was sustained: 52% did anything, and 30% did enough (P<.001 for both vs the reminders alone and control groups). Multivariable analysis showed that feedback on performance contributed independently to intensification and that intensification contributed independently to fall in HbA1c (P<.001 for both).
Conclusions: Feedback on performance given to medical resident primary care providers improved provider behavior and lowered HbA1c levels. Similar approaches may aid health care provider behavior and improve diabetes outcomes in other primary care settings.