Despite the introduction of multiple novel therapeutics over the past two decades, rates of control in type 2 diabetes (T2D) and its comorbidities remain suboptimal and may have even declined (1,2). An often-underappreciated strategy for improving outcomes in T2D has been the development of disease management programs, which leverage multidisciplinary approaches to redesign the care of T2D and other chronic diseases to extend beyond the traditional clinic setting. The American Diabetes Association and other organizations recommend the use of such programs based on frameworks such as the Chronic Care Model and the Patient-Centered Medical Home (3–6). A host of studies and systematic reviews have examined the effects of these programs over the years, with varying results (7–13).
As an example, in the landmark Steno-2 trial, Gaede et al. (7) used patient-level randomization to evaluate the impact of an intensified, multifactorial intervention versus conventional care on cardiovascular and microvascular events in individuals with T2D and microalbuminuria. They demonstrated a significant reduction in cardiovascular events (hazard ratio [HR] 0.47, 95% CI 0.24–0.73) and microvascular events (nephropathy, retinopathy, and autonomic neuropathy) with the intensive intervention compared with conventional care over a mean follow-up of 7.8 years. In contrast, the Anglo-Danish-Dutch Study of Intensive Treatment in People With Screen Detected Diabetes in Primary Care (ADDITION)-Europe utilized a cluster-randomized design to evaluate the effect of intensive, multifactorial treatment of individuals with T2D versus routine care on 5- and 10-year cardiovascular outcomes. While the intensive management group demonstrated greater improvements in cardiovascular risk factors (blood pressure, cholesterol, and HbA1c) compared with routine care, between-group reductions of cardiovascular events (HR 0.83, 95% CI 0.65–1.05) and mortality (HR 0.91, 95% CI 0.69–1.21) were not statistically significant at 5 years (13). After 10 years, differences in improvements in cardiovascular risk factors were sustained within the intensive management group but were similar to improvements seen in the routine care group. The difference in cardiovascular events and mortality remained nonsignificant (8).
While Steno-2 and ADDITION-Europe utilized randomization to allocate patients into treatment arms, randomization has not always been used for T2D disease management program evaluations. In one meta-analysis assessing the effectiveness of diabetes disease management programs, Pimouguet et al. (14) excluded 60% of published articles due to lack of randomization. While randomized studies remain the gold standard for evaluating therapeutic interventions, there is value in also considering observational studies in the overall evaluation of T2D management programs. Information from these studies may provide real-world insights on innovative approaches to care in diverse practice settings over extended periods of time. In addition, study design and analytic techniques have evolved over time to address confounding and bias to strengthen causal inferences from observational data (15–19).
In this issue of Diabetes Care, Tang et al. (20) present 10-year follow-up data on reductions in macrovascular complications, microvascular complications, and all-cause mortality in association with the Risk Assessment and Management Programme–Diabetes Mellitus (RAMP-DM). This article complements prior reports of outcomes associated with RAMP-DM over shorter follow-up periods (21–23). RAMP-DM was implemented by the Hong Kong Hospital Authority to enhance the care of patients with T2D. The intervention utilized nurse case managers to facilitate systematic risk assessment, triage, and care management activities. In the present analyses, a prospective cohort design with 1:1 propensity score matching was used to assess the outcomes between the RAMP-DM and usual care groups (Fig. 1). A total of 36,746 patients were included, with a median follow-up of 9.5 years. Individuals receiving RAMP-DM were observed to have lower risks of macrovascular complications (HR 0.52, 95% CI 0.50–0.54), microvascular complications (HR 0.68, 95% CI 0.64–0.72), and all-cause mortality (HR 0.45, 95% CI 0.43–0.47) compared with the usual care group. Of note, the impact of the program on microvascular and macrovascular complications appeared to diminish in the eighth year and ninth year of follow-up, respectively. Tang et al. observed impressive between-group absolute risk reductions and corresponding numbers needed to treat for the long-term outcomes (e.g., absolute risk reductions of 16.2% and numbers needed to treat of 6 for all-cause mortality).
The incorporation of a pragmatic design, extended follow-up period, and propensity score matching to mitigate underlying between-group differences makes the study a valuable addition to the body of evidence supporting T2D management programs. The team’s impressive success in implementing the RAMP-DM intervention across a broad population within a real-world primary care framework provides a potential roadmap for efforts to redesign T2D care in other large populations and health systems. The study also reported on a broad spectrum of outcomes of interest, from intermediate clinical markers to long-term T2D complications and death. The authors’ robust subgroup and sensitivity analyses may further enhance the actionability of these results.
This study suggests that involvement of nurse case managers translates to substantial benefits for patients. However, while Tang et al. (20) found that large reductions in microvascular complications, macrovascular complications, and all-cause mortality were associated with receiving RAMP-DM, it is important to consider factors that may diminish the reliability of the study’s effect estimates, such as unmeasured confounding. Although propensity score matching was appropriately utilized to account for known between-group differences at baseline, the reductions in T2D complications and mortality with RAMP-DM exceeded what might be expected based on the relatively modest between-group differences in measures of diabetes control and cardiovascular risk factors. While RAMP-DM may have benefitted patients in ways not fully reflected by changes in cardiovascular risk factors, the apparent disconnect between risk factor control and long-term outcomes also raises the possibility that unmeasured confounding may not have been fully eliminated and that residual selection bias may have contributed to the differences in long-term outcomes. As unmeasured confounding is a major threat to the internal validity of an observational study, analytic techniques to detect the presence and potential magnitude of the issue should be utilized. Reassuringly, the current study incorporated both an evaluation of a negative control and E-value calculation, supporting the robustness of the association between RAMP-DM and long-term outcomes (15–17).
Continued efforts to evaluate innovative disease management programs for T2D in real-world settings remain critical as large health systems and individual clinicians alike work to improve the lives of individuals with T2D. Long periods of follow-up are necessary to understanding the true return on investment of T2D disease management programs from a health system and payer perspective. Furthermore, due to broad differences in patient populations, available resources, and a host of other attributes, continued dissemination of outcomes associated with a wide range of interventions provides the best opportunity for health systems to find interventions that best fit their specific context. While randomized trials remain appropriate for earlier-phase program assessments and longer-term studies whenever feasible, well-designed, methodologically sound studies such as that of Tang et al. (20) contribute to these overarching goals.
Innovations in T2D care delivery should be explored, evaluated, and discussed with the same energy as those related to pharmacotherapy and other new treatment modalities. Accordingly, future evaluations of T2D management programs should seek to provide the highest possible levels of evidence to evaluate causal inference, whether they be randomized trials or well-designed observational studies. To facilitate implementation and dissemination, authors and journals should be encouraged to publish detailed information about intervention design and delivery within supplementary material. Scaling programs with optimally generated supportive evidence across the health care ecosystem is one approach to maximizing efforts to improve the outcomes of individuals with T2D.
See accompanying article, p. 2871.
Funding. M.J.C. reports funding from the National Institutes of Health (1R01NR019594-01), the Veterans Affairs Quality Enhancement Research Initiative (VA QUE 20-012), and the Veterans Affairs Office of Rural Health.
Duality of Interest. No potential conflicts of interest relevant to this article were reported.