Abstract
Diabetes self-management education programs are an essential strategy for improving health behaviors of adults with diabetes and, therefore, intermediate clinical outcomes. We conducted a retrospective observational study using a case:control design to estimate the impact of participation in a diabetes health education program on glycemic and lipid levels, accounting for nonrandom participation of adults with diabetes in the program ("regression to the mean"). Adults with diabetes in a group-model managed care organization who attended the diabetes health education program during the period January 1, 2003, through June 30, 2004 ("participants"), were randomly matched with 4 adults with diabetes who did not participate ("nonparticipants"). Participants (N = 1991) and nonparticipants (N = 7964) were matched on age group, gender, mean hemoglobin A1c (Hb A1c) (or low-density lipoprotein) in the 6 months prior to the class (or randomly selected index month for nonparticipants), and primary care practice where the patients received regular care. On average, participants had significantly (P < .05) worse glycemic and lipid levels in the 6 months prior to participation compared to nonparticipants. Participation in the diabetes education program significantly improved glycemic and lipid levels between baseline and follow-up periods above the improvement attributable to regression to the mean. For example, nonparticipants with baseline Hb A1c levels greater than 10.0% had improved Hb A1c levels of -1.7% (P < .01); however, among participants, mean Hb A1c levels improved an additional -1.6% (P < .01). Overall, the evidence suggests that participation in a multifactorial diabetes health education program significantly improved glycemic and lipid levels in the short-term, particularly among participants with extremely adverse Hb A1c or low-density lipoprotein levels prior to participation.