Can Practice Redesign Improve Diabetes Self-Management Education (DSME) Referrals?

Abstract

DSME is considered an essential component of diabetes care, but participation is abysmal. Referral processes and traditional delivery models may serve as barriers. The purpose of this study was to deploy a model that relied on elements of the patient-centered medical home (PCMH) and direct delivery of DE services in primary care (PC) and evaluate its impact on PC provider (PCP) referrals to DSME. Methods: Using a non-randomized controlled design, DEs and PC practices were assigned to the intervention group (IG; 3 DEs, 6 practices) or control (2 DEs, 6 practices). Practices in the IG were introduced to DSME Position Statement (PS) referral algorithm and PCMH elements and negotiated application for their practice, including frequency of direct delivery of DSME in PC. Traditional DSME delivery (no direct link between DE and PC) served as control. To examine DSME referrals, data were extracted from medical records for all patients with DM, age 18-75y, who presented to their PCP over 12 months (n=4,508; 60.7% IG) and compared between study groups. Results: Mean age was 59.8y, and primarily male 51.2%. The IG referred a higher percentage of patients compared to the control group (12.6% vs. 7.9%; p<0.0001). Using logistic regression modeling, it was found that after adjusting for age, gender, race, A1c, obesity, and micro/macrovascular comorbid conditions, the IG was 1.9 (Confidence Level 1.5-2.4) times more likely to refer patients to DSME. Other factors found to predict referrals were female (Odds Ratio=1.3, CL 1.1- 1.6), obesity (OR=2.7, CL 2.2-3.3) and higher A1c (OR=3.2, CL 2.6-3.9). Older patients were less likely to receive a referral (OR=0.99, CL 0.978-0.995). Conclusions: Findings suggest that the intervention positively influenced DSME referral rates within PC, a critical step towards improving patient participation in DSME and meaningful improvements in diabetes outcomes. Further study of DSME referral types, PS referral algorithm influence, and utilization patterns is warranted. Disclosure J. Krall: None. J. Kanter: None. V.C. Arena: None. K. Ruppert: None. F.X. Solano: None. L.M. Siminerio: None.