One year following an empowerment-based DSMS intervention designed for African Americans, participants sustained the postintervention self-care and quality-of-life improvements and demonstrated further improvements in metabolic and cardiovascular outcomes. Most notable among our findings was the significant improvement in glycemic control associated with the 1-year follow-up period (8.0% vs 7.1%). This −0.93% change in HbA1c is comparable to that achieved from oral agents (23
). Consistent with our results, other studies have found significant reductions in glycemic control to emerge at long-term follow-up (18 months and beyond) rather than immediately postintervention (9
). For example, in a study of a diabetes education group visit intervention, HbA1c was significantly lower at 2-year follow-up but not at 1-year postintervention (9
). Findings from these studies provide some support for a delayed intervention effect.
That clinical outcomes such as glycemic and lipid control would improve 1-year following intervention withdrawal rather than immediately following the 2-year DSMS intervention is curious. There was no significant change in the percentage of participants who were using insulin and taking cholesterol medication during this time. Although these improvements may be attributed to treatment intensification, it is unlikely that this level of intensification occurred only during the final year of the study. Another viable explanation could be that improvements in lifestyle and behavioral changes achieved in the 2-year DSMS intervention (eg, making good dietary choices, spacing out carbohydrates, using insulin as prescribed) require more time to produce clinical benefits. However, after further statistical examination, we found no relationship between these behavioral and clinical improvements. Future studies should investigate other factors that could contribute to a delayed intervention effect.
Although DBP (74.8 mm Hg vs 78.7 mm Hg) and HDL cholesterol (49.9 mg/dL vs 43.4 mg/dL) appeared to worsen at 1-year follow-up, a mean DBP of 78.7 mm Hg still falls within the recommended target range of at or below 80 mm Hg. Similarly, with regard to cardiovascular disease risk, the ratio of total cholesterol to HDL cholesterol is more clinically meaningful than HDL cholesterol alone. At 1-year follow-up, the mean ratio of total cholesterol to HDL cholesterol was 3.14, which is within the target range (≤4).
Some studies evaluating DSME interventions have incorporated reinforcement during or immediately following the intervention with the objective of promoting continued health improvements. Although some research has found reinforcement to be a critical component for sustained behavior change (12
), other research has not produced compelling evidence for its value (14
). An example favoring reinforcement is a 4-year study of an intensive, structured hospital-based diabetes education program with a built-in yearly “booster” education session conducted in a group-based setting, where previous self-management topics were discussed and “up-to-date” diabetes care information was presented; in this study, more health-related improvements were evidenced in the intervention group compared with the control group (12
). Alternatively, another study used a reinforcement component for a 6-week Internet-based diabetes self-management intervention via a discussion board (over 18 months) where participants could post messages asynchronously, with the goal of fostering reciprocal peer support; in this study, there were almost no intervention-control differences (14
). Perhaps the critical question is not whether a reinforcement mechanism is present or absent but rather what type of reinforcement mechanism is employed and how frequently and how long the reinforcement mechanism is used.
Rather than incorporating a reinforcement component into a larger DSME intervention, our DSMS intervention could be conceptualized as a participant-driven reinforcement program in and of itself. Specifically, we invited participants to access support (ie, weekly sessions) as frequently as they needed or as was feasible, given their schedule and life circumstances. In other words, participants had complete control over the frequency and extent of support they sought. Essentially, participants functioned as active agents of their self-management and lifestyle change, and the weekly DSMS sessions served as a resource they could use in their self-management efforts. Considering this interpretation, participants may not have been adversely affected when the intervention was withdrawn. Instead, they were already directing their own self-management practices and decisions and could continue to do so even when the resource was no longer available. Clearly, a goal of DSMS interventions is to foster this type of self-sufficiency and self-efficacy.
Although our attrition rate fell within the expected range, having 33% of participants drop out of the study was not ideal, and we did not follow up with participants regarding reasons for study discontinuation. Subsequent investigations should include a formal protocol to inquire about dropout. This study had other limitations. Because of restricted funding, we were not able to conduct a randomized controlled trial. Without a rigorous methodological design, making definitive conclusions about the efficacy of this intervention was not possible. However, considering the extended length of the intervention and subsequent follow-up period, it is unlikely that improvements could be attributed to factors such as the initial enthusiasm at the start of a study or the attention one receives simply by enrolling in a study. Similarly, any changes associated with secular trends would have more likely reflected health deterioration rather than health improvement, particularly for this sample of older participants. Regardless, future investigations should use a randomized controlled trial or comparative effectiveness design to answer these questions.
Effective diabetes self-management requires initial DSME followed by ongoing DSMS (16
). Given the overwhelming evidence for the short-term effect of diabetes self-management interventions, greater attention is needed for developing, implementing, and evaluating interventions that yield positive outcomes that are sustained throughout life. Interest in examining peer support models as an effective and viable approach to long-term DSMS is growing, and understanding which aspects or underlying mechanisms of self-management interventions lead to long-lasting diabetes-related health benefits is needed.