Peer-Delivered Cognitive Behavioral Training to Improve Functioning in Patients With Diabetes: A Cluster-Randomized Trial

The purpose of the study was to examine the effectiveness of a cognitive behavioral therapy (CBT) based training program for patients with diabetes in rural towns in Alabama with high burden of chronic disease. Outcomes included improved functional status, increased physical activity, and physiologic measures. The study used community members/lay health workers to do the CBT intervention and attention control.

Because the effects of diabetes are known to improve with lifestyle modification it is a good choice for a CBT-based intervention. Goals of CBT in this study were to identify barriers and challenges that an individual faces, and to explore what they need to do to overcome them. The group discussed that patients with diabetes can greatly benefit from the 1:1 support and repeated reinforcement provided by the intervention. We discussed that people with diabetes often also have co-morbid conditions treatable with CBT such as depression and chronic pain, so the benefits could be even greater than simply treating blood glucose levels. There are several advantages of using lay healthcare workers: medical professionals may not be available, they can develop closer relationships and identify with participants, and economically are more feasible than paying medical professionals. Using peer coaches who have worked in the community previously or are key leaders in the community is very important for establishing trust and for the effectiveness of the intervention. There is a history of mistrust in underserved communities when it comes to researchers using their population to collect data and leaving without implementing long term solutions. Allowing community members to lead the intervention alleviates these concerns. Peer coaches have greater understanding of culture and experiences in a community. The factor of cost comes into greater play when examining the great application of interventions of this nature. The group discussed that several states are in the process of enacting policies which would ensure reimbursement for the services of lay health workers from insurance companies. These changes would increase the availability of services to many people especially those in rural areas with low access to healthcare and high burden of chronic diseases such as diabetes. The success of lay healthcare workers depends greatly on the quality of training. In addition, cross training coaches in a variety of topics is difficult. The group discussed that this approach raises questions about the generalizability of the effect of the intervention.

This study uses a cluster-randomized control trial (Living Health as intervention vs General Health as control). The clusters randomized were rural towns in Alabama. The group discussed that an individually randomized design is very difficult in a rural community because they tend to be so tight-knit and it is difficult to prevent communication between groups. Participants and coaches were not blinded due to the nature of the study, but the analysts and researchers were blinded. Our opinion is that this is not as ideal as a double-blind study, but necessary. Recruitment was accomplished by distributing posters/flyers at community events, health fairs, and by participants referring other people in the community. Advantages of this approach are the low cost and increased visibility within social circles, but it also makes the study vulnerable to selection bias. The peer coaches in the study were from the same communities as the participants, and were of similar demographics, and either had diabetes or a family member with diabetes. They went through a 10 week CBT training program for another intervention/study done by the same researchers The intervention group coach training - The Living Health intervention discussed healthy eating, physical activity, and pain control; it was more action-oriented. The control group training - The General Health had high level education on various health topics, which were independent of the study outcomes. Our opinion is that the structure of the control group was acceptable, but without further elaboration on the specific material covered in each intervention it is possible that the results could be confounded by content differences in the programs rather than attributable to the CBT delivery method. In addition, we also believe that the results of the study would have been strengthened greatly by a crossover design rather than the longitudinal one that was selected. We were impressed by the extent of oversampling required to produce the study. Baseline characteristics of the control and intervention group were similar. The study population included mostly middle aged African-American females with incomes <$20,000. The group had questions about why adjusted data was not available; even if baseline characteristics were similar they were not exactly the same and adjusted data would have accounted for those differences. There was also confusion regarding whether there were differences in the mean days per week walked vs. mean days per week of intense exercise based on differences in tables and narrative of the paper. There were statistically significant improvements in functional status, stiffness, pain, number of days walked, and the mental health component of quality of life between baseline and follow-up for both groups, but difference in days participated in intense physical activity, physical component scores of quality of life, and clinical indicators (HbA1c, weight) were not significant. The group thought that the subjective measures were appropriate for this 3 month intervention, but the chosen objective measures were not. Using HbA1c as a clinical indicator for this study was problematic because HbA1c is an average over 3 months and the study was only 3 months. It would be a more appropriate measure at a later follow-up time. In general, changing clinical indicators takes a long time (longer than 3 months) with only lifestyle modification. There also was no mention of the participants being on glucose control medications or not being on medications, which would also play a significant role in their HbA1c.The tool used to measure physical activity was not a validated instrument and therefore less reliable. One question we thought was particularly problematic was one that used sweating as a marker for intense activity, but sweating is variable between people. We appreciated that the tools used to examine quality of life and functional status were validated instruments. Mental health component scores of the quality of life scale improved, but not physical component scores. The group considered that this may be from the effect of physical activity and interpersonal support on mood. Intervention group participants had increased ability to exercise in spite of or without being hindered by pain, which may indicate that it is easier to manage and work through pain with a support system. However, there was no data available on whether participants had any other pain management techniques in place. It would be useful to know if patients were taking pain medications or making other efforts to lessen pain during the intervention.

Future studies may benefit from looking at longer-term follow-up to examine the effects of the intervention. We discussed examining a more diverse or more specific population and found that either direction has merit. A more specific group might provide suggestions for tailored intervention, but a more diverse group would require a much bigger study. The goal of a future study would be to examine if there was a persisting lifestyle change and establish long term follow-up. CBT is not sustainable for a lifetime and requires a manner of transition after the intensive intervention.

The results of the study support a prevention care model and the use of lay healthcare workers. Ideally, results can be used to convince health care systems that the cost of a similar intervention or any provision for CBT would be worth the investment.

Reference Andreae, Susan J., et al. "Peer-delivered cognitive behavioral training to improve functioning in patients with diabetes: a cluster-randomized trial." The Annals of Family Medicine 18.1 (2020): 15-23.

Competing interests: None declared