In recent years, social networks have garnered attention in both academic journals and the lay press because of strong associations demonstrated in retrospective studies between social networks and incidence of major health problems such as obesity and smoking. Financial incentives have also been demonstrated to improve health behaviors in obesity, smoking, and medication adherence. We propose to conduct a randomized controlled trial among a predominantly African American population with persistently poor diabetes mellitus (DM) control to examine whether two novel interventions, lottery based financial incentives and telephone based one-on-one peer mentoring (the 'buddy system'), can significantly ameliorate disparities in poor DM control.

The intervention is based on epidemiologic evidence, randomized controlled trials, and pilot studies demonstrating: 1) Lottery based financial incentives are a powerful motivator of behavior change; 2) One-on-one peer mentoring is a flexible, cost-efficient means of increasing DM specific social support and may be particularly salient in minority communities; and 3) Matching patients with poorly controlled DM with a similar individual from their community who has gained control of their DM draws on existing community assets in creating an inherently culturally competent intervention. DM patients with poor DM control will be randomized to 1 of 4 arms: usual care; telephone based one-on-one peer mentoring; lottery based financial incentives; and peer mentoring plus financial incentives.

The Specific Aims are to test: 1) The effectiveness of telephone based one-on-one peer mentoring in improving glycemic control relative to usual care; 2) The effectiveness of lottery based financial incentives in improving glycemic control relative to usual care; 3) The effectiveness of combining peer mentoring and financial incentives relative to control; and 4) The relative cost effectiveness of all four approaches. In exploratory analyses, we will examine whether African American patients enrolled in intervention arms have greater improvement in glycemic control than white patients, whether intervention group patients experience greater improvements in blood pressure (BP) and lipid control, and whether peer mentors experience improvements in their own health. We will pair mentors with mentees based on race, gender, age, and disease severity. The active intervention will be run for a 6-month time period, with participants followed for an additional 6 months to determine if effects persist post intervention.

The proposed interventions address multiple barriers to effective disease management common among patients with DM. If effective, these interventions could provide important models for improving glycemic control in general and, in particular, for addressing racial disparities in DM outcomes.

Inclusion Criteria:

• diagnosed with diabetes mellitus
• 30 to 70 years old
• persistently poor glucose control, defined as having the last two HbA1c values > 8%, with at least one measure within 3 months of enrollment
• receiving treatment for diabetes mellitus at one of five University of Pennsylvania outpatient clinics

Exclusion Criteria:

• does not speak English
• unable to provide informed consent