Summary: In four primary care clinics serving adults with type 2 diabetes, this proof-of-concept cluster randomised pragmatic trial tested whether the INTEGRATE-D implementation support intervention was feasible and acceptable. Clinics judged it acceptable, appropriate, and feasible, but effects on process-of-care metrics and clinical outcomes (PHQ-9 and A1C) were mixed, and the authors call for further pilot testing rather than claiming efficacy.
PICO Summary
| Element | Detail |
|---|---|
| Population | Adults with type 2 diabetes attending primary care clinics in the United States; cluster randomised pragmatic, mixed-methods, proof-of-concept design across 4 clinics. |
| Intervention | INTEGRATE-D implementation support (tailored training on ADA-recommended psychosocial care plus 15 months of facilitation; delivered virtually owing to COVID-19). 2 intervention clinics. |
| Comparison | Usual care with no intervention. 2 control clinics. |
| Outcome | Primary implementation outcomes (acceptability, appropriateness, feasibility) were favourable: clinics found INTEGRATE-D acceptable and well-aligned with their needs, though COVID-19 stressors tempered feasibility. Secondary process-of-care metrics (e.g. depression screening, diabetes management) and clinical measures (PHQ-9, A1C) showed mixed effects. As a 4-clinic proof-of-concept study, no formal between-arm effect sizes, 95% CIs, or p values for efficacy were reported, and none should be inferred. |
Expert Commentary
This study is best read as a feasibility and acceptability signal, not as evidence that structured psychosocial support improves diabetes outcomes. The verdict is that INTEGRATE-D was found acceptable, appropriate, and feasible by participating clinics, while its effect on process-of-care and clinical measures such as PHQ-9 and A1C was explicitly mixed. That distinction matters, because implementation outcomes and patient outcomes are not interchangeable, and an intervention can be welcomed by clinicians yet still fail to move glycaemic or depression metrics. The dominant limitation is scale: with only two intervention and two control clinics, the trial was never powered to detect clinical effects, and baseline differences between clinics not balanced by randomisation could plausibly explain the mixed signal. The design was pragmatic and unblinded, and the protocol was diluted mid-study when COVID-19 forced facilitation and training to move online, which the authors note tempered the intended intensity. There was no commercial or manufacturer sponsorship, and the effect sizes reported are modest rather than implausibly strong, which is reassuring for credibility. Can I use this with my patients? Not yet; this is hypothesis-generating work that supports a larger, in-person pilot rather than a change in how psychosocial care is delivered today. Future trials should enrol enough clinics to test patient-level outcomes with adequate power and fidelity.
References
Cohen DJ, Sweeney SM, Springer R, Balasubramanian BA, Michaels L, Marino M, et al. Intervention to Improve Psychosocial Care for People with Type 2 Diabetes. J Am Board Fam Med. 2025;38(2):253-274. doi:10.3122/jabfm.2024.240265R1
