The Final Frontier in Diabetes Care: Implementing Research in Real-World Practice

Despite decades of evidence related to type 2 diabetes (T2D) and its complications, most people at risk for and people who have diabetes still do not receive recommended care (1–5). In this article, we discuss the promise and challenges of implementing effective approaches across the continuum of diabetes prevention and care.

In this context, implementation refers to the translation of research findings into clinical practice, with the goal of improving the health of real-world individuals and populations with diabetes (6). Some evidence-based therapies and approaches are easier to implement than others; unfortunately, when efficacy trials are designed, intervention translatability is not always thoroughly considered. However, implementation is of the utmost importance because without it, even the most impressive research findings will remain of purely academic interest.

Much can be learned from rigorous examination of the processes of mplementation, including predictors of implementation fidelity, which intervention approaches are acceptable to patients, and which outcomes patients value most (6). Herein we provide examples of implementation research in three key areas of diabetes care—prevention, glycemic control, and prevention of complications—in order to improve the prevention and control of T2D in the community.

The foundational evidence demonstrating that structured lifestyle interventions can prevent the onset of T2D emerged from five efficacy trials, commonly referred to as the “landmark trials,” conducted in China, Finland, the U.S., Japan, and India (7–11). These trials, conducted between 1982 and 2005, specifically targeted individuals with impaired glucose tolerance with or without overweight (BMI ≥25 kg/m²). The interventions in these trials varied in duration from 3 to 6 years and resulted in relative risk reductions for diabetes incidence ranging from 28% to 67%. Moreover, for three of these trials the follow-up periods were extended, lasting from 7 to 30 years (12–24). In these long-term follow-up studies, for the intervention participants there was consistent maintenance of significantly lower weights, reduced diabetes incidence, and fewer occurrences of micro- and macrovascular complications, as well as lower rates of all-cause mortality, in comparison with the control participants. In addition to their effectiveness, lifestyle interventions in these landmark trials were found to be cost-effective in the short- to medium-term (3–10 years) (25–28) and cost saving over a 30-year or lifetime horizon from the health system and societal perspectives (29).

Building on the success of the landmark efficacy trials and subsequent efficacy trials (30), researchers and program implementers worldwide have made attempts to translate the findings into real-world settings (31–33). In these real-world implementation trials adjustments were made to efficacy trials, including the use of hybrid implementation-effectiveness designs and broadening inclusion criteria for participants and engaging laypeople and community health and allied health professionals for program delivery to increase the reach of programs in order to achieve public health impact. Investigators also evaluated the implementation of programs in a range of community settings and with different populations, including those who have been traditionally underserved, disadvantaged, and vulnerable. Adapted from the curricula of the U.S. Diabetes Prevention Program (DPP) (9) and the Finnish Diabetes Prevention Study (8), these real-world studies have generally shown significant positive health outcomes but not to the same extent as in the more intensive programs delivered in landmark efficacy trials. For instance, a meta-analysis of 28 U.S.-based studies with use of the curriculum of the U.S. DPP demonstrated an average weight loss of 4% relative to baseline after 12 months of intervention (31). Notably, this weight loss remained consistent regardless of whether the intervention was delivered by clinically trained professionals or lay educators (31). Furthermore, a global systematic review and network meta-analysis of real-world lifestyle-based intervention studies demonstrated a clinically meaningful 29% reduction in diabetes incidence (32). Additionally, mean weight loss and fasting glucose reduction were 1.5 kg and 0.09 mmol/L more, respectively, for intervention participants than for control participants (32). However, despite the promising clinical and behavioral effectiveness of these more real-world studies, participation rates have still generally been low, and the reporting of implementation outcomes has been inadequate, particularly concerning program reach and penetration and implementation fidelity (33). Moreover, the vast majority (86%) of real-world translation studies have been conducted in North America and Europe, with only a small fraction (3%) in Asia, and few have been undertaken in the Middle East and Africa (32).

Only a few countries, primarily in North America and Europe, have attempted to implement national diabetes prevention programs (NDPP) at scale in their populations with interventions adapted from those used in the landmark efficacy trials (). These initiatives have shown positive effects on clinical and behavioral outcomes, including weight loss, reduction in diabetes incidence, and improvements in cardiometabolic risk factors (37–46). However, their reach and adoption have only been very modest. An analysis of U.S. nationally representative, population-based data from the 2016 and 2017 National Health Interview Surveys, involving 50,912 adults, showed that merely 4.9% reported receiving a referral to a diabetes prevention lifestyle modification program (44,45). Among them, just under 40% reported actual participation in diabetes prevention programs. In the U.K. National Health Service Diabetes Prevention Programme, of 324,699 individuals referred to the program between June 2016 and December 2018 fewer than one-third (29.7%) attended more than one intervention session (36). Furthermore, individuals referred to the program were more likely to have higher BMI levels and be ever smokers than their counterparts, despite matching for age, sex, and time to diagnosis of prediabetes (42). Intervention attendance varied significantly by sex, age, ethnicity, and deprivation: lower among men (<0.001), higher among older age-groups (<0.001), and higher for individuals of Asian, Afro-Caribbean, mixed, and other ethnic groups compared with individuals of White European groups (<0.001) (37). Additionally, attendance rates were higher among those in the most deprived quantile than rates of those in the least deprived quantile.

Despite the progress made in the research on preventing T2D through lifestyle interventions, moving from efficacy trials to real-world studies and national prevention programs, the effective translation of these initiatives into clinical practice remains a persistent challenge, even after two decades of research and program development (32,45,46). From a translation perspective, few of the real-world studies have reported sufficient information for estimation of the overall impact of such programs (32). This reduces the utility of many diabetes prevention trials to inform resource allocation and real-world implementation. More rigorous evaluation methods are required for better understanding of the factors that influence the likely success of such interventions (32).

Evidence for the effective implementation and translation of diabetes prevention programs and their population impact needs to be established in ways that are different from just measuring program effectiveness (32). In this context, implementation science provides the tools and methods for evaluating program penetration, implementation, adoption, and maintenance (47). While high-frequency, intensive interventions can play an important role in achieving outcomes of high reductions in weight loss, programs with “low”-intensity interventions have also shown high reductions in the incidence of T2D (32). This means that even when the effectiveness of an intervention is moderate in terms of weight loss, it can still have a profound impact on the development of diabetes at the population level—provided enough effort is put into maximizing high penetration and participation as well as sustainability at both the individual and program levels (32).

Findings from trials undertaken in low- and middle-income countries and with vulnerable and disadvantaged people in high-income countries over the past 15 years have identified potential strategies and ways of achieving high penetration and participation to achieve population benefit (31,32). For example, community-based approaches with peer support and community health workers can be very powerful. Key considerations for the translation of proven diabetes prevention efforts into real-world clinical care can be found in Table 1. Importantly, such approaches have also been shown to be cost-effective (48) and to be supported by policy-makers and program implementers, which is very important for the future scalability and sustainability of such programs (49). There are learnings for high-income countries from these trials.

The Kerala Diabetes Prevention Program (K-DPP) stands as a notable phase II effectiveness-implementation hybrid trial (50) conducted within a low- and middle-income setting. The K-DPP was designed for implementation of an effective structured lifestyle intervention within rural communities in the Indian state of Kerala, as a modest, sustainable, and scalable program primarily facilitated by lay peer leaders (50). Kerala, in the advanced stage of epidemiological transition in India (51), with the highest prevalence of T2D in the country (52), is an ideal context for the implementation of a community-based diabetes prevention program in India.

The intervention design of K-DPP was methodically formulated on the theoretical underpinnings of the “Health Action Process Approach” (53), a sociobehavioral framework. This model facilitated the systematic identification of determinants influencing behavior change, which were subsequently translated into learning objectives for program participants. The development of the K-DPP intervention drew on various sources of knowledge and best practices. Firstly, it integrated the fundamental principles of peer support outlined in the U.S. Peers for Progress program (54). Additionally, behavior change strategies were identified through a needs assessment study (55). Moreover, elements from established programs in high-income countries, including the U.S. DPP (9), the Good Ageing in Lahti Region (GOAL) program in Finland (56), and the Greater Green Triangle (GGT) project in Australia (57), were culturally adapted to suit the local context of Kerala (58). Prior to full-scale implementation, pilot testing was conducted in a randomly selected community within the study area (59). This allowed for further refinement of both program objectives and intervention components based on real-world feedback and insights, ensuring that the intervention was effectively tailored to address the specific needs of the local population.

The K-DPP enrolled a diverse population of 1,007 individuals at high risk, with 500 in the intervention group and 507 in the control group. This cohort included individuals across the entire glycemic spectrum, excluding only those with diabetes, with a high prevalence of several cardiometabolic risk factors (59). Moreover, the sociodemographic profile of K-DPP participants was similar to that of adults in rural Kerala (59), enhancing the generalizability of the findings to a broader population at high risk. Peer group sessions were conducted by trained lay peer leaders, with one male and one female leader per group, selected from within the group based on their willingness, social credibility, and acceptability according to their peers (60). These sessions took place in local community facilities, such as schools or community halls, scheduled over weekends at times convenient for participants, and typically lasted 60–90 min. Of note, besides receiving training and ongoing support from the K-DPP staff throughout the intervention period, the peer leaders did not receive any incentives.

The lifestyle change strategies focused on increasing physical activity through identification of enjoyable activities for both individuals and groups and incorporating them into daily routines (60). Dietary guidance included increasing fruit and vegetable consumption, reducing portion sizes of rice, and minimizing intake of fried foods and refined sugars. Participants used goal setting, action planning, and self-recording of activities as key behavioral strategies for enhancing physical activity and adopting healthy eating habits. Outside of formal peer group sessions, peer leaders maintained regular contact with their group members to offer ongoing support for lifestyle changes. Surprisingly, participants also engaged in “spin-off” activities of K-DPP, such as community-based initiatives like establishing kitchen gardens, yoga training, and joining walking groups. The groups were encouraged to continue meeting after the formal 12-month program period.

The K-DPP trial replicated some of the major findings of landmark efficacy trials in the significant reduction of diabetes incidence, by 34% (relative risk 0.66 [95% CI 0.45, 0.98]), among individuals with impaired glucose tolerance after 24 months (60), at a considerably lower relative cost (48,60). Furthermore, the K-DPP intervention significantly reduced predicted cardiovascular risk (61) and its risk factors (60) and was deemed cost-effective from both the health system and societal perspectives (48). The implementation evaluation of K-DPP, guided by the frameworks of Reach, Effectiveness, Adoption, Implementation, and Maintenance (RE-AIM) of Glasgow et al. (62) and Penetration, Implementation, Participation, and Effectiveness (PIPE) of Pronk (63), was conducted from the perspectives of providers, participants, and the community (64). From the provider’s standpoint, intervention fidelity and adoption were notably high, with 29 of 30 peer groups conducting all 12 sessions as outlined in the protocol. Furthermore, the intervention cost per participant was US$22.50 (cost in 2013) (48)—significantly lower than reported costs in efficacy trials conducted in India (25) and high-income countries like the U.S. DPP (26). Participants exhibited significant levels of engagement and satisfaction with the program. They attended a median of eight peer group sessions, with a retention rate of 95% after 24 months. Goal attainment was also impressive, with 96% achieving their physical activity goals and 99% successfully reaching their dietary objectives. Furthermore, 92% of intervention participants expressed satisfaction with the peer group sessions, citing the convenience of session locations and timings as primary facilitators for attendance. From the community perspective, nearly 60% of participants reported participating in one or more community-based activities such as walking groups, kitchen gardening, and yoga sessions. In some communities, support from local political leaders emerged as a major facilitator for the high uptake of the program. Conversely, a lack of community support led to the dropout of one peer group to run sessions.

Building on the success of K-DPP, the program underwent a scale-up initiative in three other districts of Kerala, selected randomly from the southern, middle, and northern regions (49). This expansion was achieved through collaboration with stakeholders across different tiers of the health system and in partnership with the Kudumbashree State Mission (KSM) (65), a prominent women’s organization. Refinements to the K-DPP were made based on expert insights gathered both on the health system level and from the women associated with KSM. A total of 15,222 women affiliated with KSM were trained as peer leaders to facilitating monthly sessions on diabetes prevention within their local communities, with 20–25 women and their family members in attendance. This outreach resulted in an impressive engagement of 375,000 adults over the span of 12 months. Despite the fact that KSM is a women’s organization, we extended invitations to men for group sessions due to lower male participation rates in the K-DPP. Beyond the program’s extensive reach, notable outcomes were observed, including a reduction in tobacco use from 30% to 25% (P = 0.02) and alcohol consumption from 40% to 32% (P = 0.001) among men after 12 months of active intervention. Furthermore, there was a significant reduction in mean waist circumference, by 2 cm (P < 0.001), for the entire population (44).

The insights gained from the original K-DPP trial and subsequent scale-up project have informed the development of another large-scale implementation initiative, called Community Control of Hypertension and Diabetes (COCO-HD). Its overarching objective is to extend the reach of K-DPP across the entire Kerala state and the neighboring state of Tamil Nadu in partnership with the two state governments. In close collaboration with the respective state governments, this initiative aims to improve control rates for hypertension and diabetes among adults in these states.

Glycemic control is central to high-quality care of T2D. Glycemic control yields meaningful reductions in microvascular complications and may reduce macrovascular complications (2,66–69). The greatest absolute benefits of glycemic control occur among those with marked elevations in hemoglobin A_1c (HbA_1c) (70,71).

Achieving recommended glycemic targets in T2D demands holistic management, with pharmacologic and nonpharmacologic approaches having equal roles. Glucose-lowering medications are central to glycemic control; >75% of people with T2D take at least one medication for T2D, and use of most T2D medication classes has increased in recent years (72). Behavioral interventions, such as nutrition therapy and physical activity support, likewise provide valuable glycemic benefits. Structured diabetes self-management education and support (DSMES) programs have a strong evidence base supporting effective delivery of behavior-focused content.

Despite the existence of effective treatments for T2D, many people do not meet recommended glycemic targets. Even in high-performing systems, 10%–15% of people with T2D maintain an HbA_1c >8.5% despite receiving available diabetes care (73,74). Such persistent HbA_1c elevations result from multiple factors. For example, therapeutic inertia and medication nonadherence are known to blunt the real-world effects of glucose-lowering medications. Additionally, DSMES programs remain underused, in part because many patients have geographic and financial barriers to access (75–77). Meanwhile, racial and ethnic disparities in glycemic control persist (78), highlighting the need for T2D care to be delivered not only effectively but also equitably, with recognition of social drivers of health (79).

Beyond the content of approaches to glucose lowering, the manner in which they are delivered can influence glycemic outcomes. Traditionally, T2D care has been delivered in person, in brick-and-mortar clinic settings. Due to busy schedules and travel requirements, the patient-provider contact frequency achievable in clinic settings is typically no more than every 3–6 months (80–82). Such infrequent contact limits opportunities for treatment intensification (exacerbating therapeutic inertia), suboptimally supports medication adherence, and perpetuates underdelivery of formal and informal DSMES. When the demands of T2D self-management exceed a patient’s capacity for self-management—and when clinic-based care cannot offer sufficient support to balance demands and capacity—persistently elevated HbA_1c values and downstream complications result (70,73,83).

Telehealth augments in-person T2D management via remote delivery of evidence-based care, feasibly enabling frequent patient-provider contact (84,85). Broadly, telehealth programs leverage technology-enabled approaches to gather patient-generated health data (PGHD), which is then used to support delivery of medical, behavioral, and educational services by providers, nurses, or other team members (86,87). Telehealth is a particularly valuable strategy for people with limited access to in-person care, including residents of rural areas, because it enables immediate access to specialized medical and behavioral management that might otherwise require extensive travel (88,89). There is also evidence that telehealth fosters equity by expanding access to evidence-based care among people from minoritized groups (90). Satisfaction with telehealth is generally high due to perceived effectiveness, ease of use, improved communication quality, and reduced travel (91–93).

In part due to the heterogeneity of studied telehealth programs, reported HbA_1c reductions vary from 0.2% to 0.7% versus usual care (88,94–98). However, comprehensive telehealth interventions carry greater potential for HbA_1c lowering. Comprehensive telehealth programs organize diabetes care professionals from across disciplines into teams capable of targeting multiple drivers of hyperglycemia (78,88). For example, comprehensive telehealth often includes active medication management to facilitate medication intensification and reduce therapeutic inertia (88). Simultaneously, comprehensive telehealth programs deliver DSMES-related education and behavior-focused teaching designed to boost adherence to T2D self-management and medication taking. As a result, comprehensive telehealth often lowers HbA_1c by ≥1.0% versus usual care (89,99,100). The American Diabetes Association recommends structured telehealth programs that leverage multidisciplinary teams as part of care for T2D (78,101).

Critically, although many health systems have embraced telehealth, few use comprehensive telehealth (102,103) Currently, “telehealth” as used in real-world practice connotes intermittent video or phone appointments with neither interim review of PGHD nor frequent patient-provider contact (102,104–107). Just as clinic-based care provides insufficient self-management support for many people with T2D, “clinic-like” telehealth is similarly inadequate (73). Using Veterans Affairs data from 2020–2022 (N = 3,778), Zupa et al. (102) recently demonstrated the limitations of telehealth as typically used for T2D. During the observation period, those receiving only telehealth appointments (annual mean 2.1) experienced no significant improvement in HbA_1c, while average improvement in HbA_1c of those receiving in-person appointments was −0.37%.

Thus, when used in a clinic-like manner—as is currently the norm—telehealth does not fulfill its potential to improve glycemic control in T2D. However, even before the coronavirus disease 19 pandemic, there was a strong argument for using thoughtfully designed telehealth approaches to improving glycemic control among individuals with T2D responding insufficiently to clinic-based care. Now that telehealth has gained increased acceptance, systems have the opportunity to implement comprehensive telehealth for people with T2D not responding to available care.

The complex nature of comprehensive telehealth programs introduces challenges for translation to clinical practice settings (Table 1). Barriers to implementing evidence-based telehealth include 1) intervention designs reliant on research-funded resources, unavailable in real-world settings; 2) variable staffing and infrastructure across implementation contexts; 3) insufficient clinical or electronic health record (EHR) integration of PGHD; 4) uncertain reimbursement models, which may threaten sustainability; and 5) attenuated program impact in real-world settings (108–111). These challenges may intensify in resource-limited settings.

Although implementing comprehensive telehealth comes with challenges, appropriate planning can overcome many barriers (Table 1). First, program reliance on research-funded resources can be avoided through consideration of the program’s eventual implementation from the outset of development and authentic engagement of end users, system leaders, and community partners in the design process (112). Second, issues related to variable staffing across contexts can be overcome through explicit design of programs for delivery by commonly available staff, as well as definition of core and adaptable program components to facilitate context-specific tailoring. Third, while inadequate integration of PGHD can impede EHR-integrated delivery of comprehensive telehealth, it can be mitigated through creative use of existing EHR infrastructure and third-party vendor websites. Fourth, although reimbursement-related sustainability challenges are problematic, system-specific solutions do exist. Some payers reimburse for chronic care management and telehealth services (113,114), and comprehensive telehealth can be designed to align with these opportunities; furthermore, systems with affiliated Accountable Care Organizations can seek to integrate comprehensive telehealth services with care management infrastructure (115). Fifth, for tracking and addressing problems with implementation or fidelity that might attenuate real-world program effectiveness, a rigorous, theory-informed implementation evaluation should be maintained. Of note, future research will continue to smooth the path for telehealth implementation across systems, including an ongoing trial examining viable reimbursement models for comprehensive telehealth and integration of PGHD into the EHR (116).

Advanced Comprehensive Diabetes Care (ACDC) is a comprehensive telehealth program that was developed and implemented within the U.S. Veterans Health Administration (VHA) for people whose HbA_1c remained above 8.5%–9.0% despite receiving available T2D care. ACDC was previously described in detail (89,99,100).

ACDC was designed to leverage existing VHA Home Telehealth (HT) program staffing and infrastructure. The VHA HT program provides remote patient monitoring across the U.S., but prior to ACDC, was not used to deliver comprehensive telehealth for diabetes (117). Because HT infrastructure is ubiquitous across VHA, every site has the necessary resources to deliver ACDC; accordingly, reimbursement for ACDC services occurs through existing HT channels. At each site, clinical HT nurses deliver ACDC using standard-issue HT equipment, with collaboration from medication managers (typically VHA clinical pharmacists), who have no direct contact with patients. As delivered in VHA practice, ACDC includes three components: telemonitoring, self-management support, and medication management (Table 2). HT nurses deliver ACDC content during telephone-based intervention encounters occurring every 2 weeks for 6 months. Clinical documentation and intrateam communication occur within the EHR.

ACDC was effective in a randomized pilot (n = 50), which demonstrated a statistically significant HbA_1c reduction, −1.0%, versus usual care alone, and in a randomized effectiveness trial (n = 200), which showed a statistically significant HbA_1c reduction, −0.6%, versus an active telehealth comparator (telemonitoring and care coordination) (99,100). ACDC is now being implemented in VHA clinical practice via two related efforts. The first is a VHA Office of Rural Health–supported project, through which ACDC has been implemented at 20 VHA centers. Implementation analyses are guided by the RE-AIM framework; the program was delivered with strong fidelity to N = 755 participants between 2017 and 2023, with a mean HbA_1c improvement of −1.6% at 6 months (sustained at 36 months) (62,89). The second effort is a Quality Enhancement Research Initiative–funded, cluster-randomized, hybrid type 3 effectiveness-implementation trial, through which ACDC is being delivered at 10 additional sites with the aim of comparing two evidence-based implementation strategies (clinical trial reg. no. ISRCTN91461910, www.isrctn.org). Consolidated Framework for Implementation Research (CFIR)-guided analyses are ongoing (118).

In conclusion, comprehensive telehealth can improve glycemic control in T2D, even in cases where clinic-based care has proven insufficient. For successful, sustainable implementation, interventions must be designed and implemented in a system-specific manner that emphasizes feasibility and leverages available clinical resources. Although many challenges remain, the effectiveness and patient centeredness of comprehensive telehealth make these problems worth solving.

Micro- and macrovascular complications contribute substantially to the excess morbidity and mortality associated with diabetes. For example, diabetes is a leading cause of vision loss, with an estimated 10.1% of U.S. adults with diabetes having severe visual compromise or blindness (119). Of U.S. adults with diabetes >8% have major cardiovascular disease and nearly 40% have chronic kidney disease based on estimated glomerular filtration rate values alone. Individuals with diabetes are also hospitalized for lower-extremity amputations ∼160,000 times per year in the U.S. (119). Fortunately, many interventions designed to screen for, prevent, and treat diabetes-related complications are effective in modifying the risks posed by these conditions (4,120–123).

Recommended strategies to reduce the risk and progression of cardiovascular and kidney diseases in diabetes have evolved dramatically over the past decade. Contemporary guidelines issued by cardiovascular, kidney, and diabetes care societies are now unified in recommending the use of glucagon-like peptide 1 receptor agonist (GLP-1RA), sodium–glucose cotransporter 2 inhibitor (SGLT2i), and/or nonsteroidal mineralocorticoid receptor antagonist therapy for people with diabetes who have or are at high risk for cardiorenal complications (120–125). Importantly, these newer recommendations are in addition to, rather than in place of, traditionally used cardiovascular risk reduction strategies such as smoking cessation and blood pressure and lipid management (121). Unfortunately, even these long-standing effective and relatively inexpensive interventions are inconsistently applied. This is evident from nationally representative U.S. data collected in the National Health and Nutrition Examination Survey between 2015 and 2018. During this time, 66.8% of U.S. adults with diabetes achieved individualized HbA_1c targets, 48.2% had blood pressure <130/80 mmHg, 59.7% had LDL cholesterol level <100 mg/dL, and only 21.2% achieved all three (126). These targets were less likely to be met by younger adults and by non-Hispanic Black and Mexican American adults compared with older and non-Hispanic White adults (126).

Adherence to the current, expanded cardiovascular risk reduction guidelines for people with diabetes at high risk may be even more limited. In a multicenter cohort study performed in 2018, using data from the U.S. National Patient-Centered Clinical Research Network (PCORnet), investigators found that <1 in 20 patients with diabetes and established atherosclerotic cardiovascular disease (ASCVD) were prescribed all three guideline-indicated therapies including high-intensity statin, an ACE inhibitor (ACEi) or angiotensin receptor blocker (ARB), and an SGLT2i and/or a GLP-1RA (127). Surprisingly, more than one-third of these patients at very high risk were on none of these indicated therapies. Use of newer drugs with proven cardiovascular benefit was particularly low, with only 3.9% of patients prescribed a GLP-1RA and 2.8% prescribed an SGLT2i (127). Although limited access to newer medications is often considered the most significant obstacle to delivery of care, this would not be the explanation for under- or non-use of the older classes of medications. In the U.S. Veterans Affairs system, where the newer drugs are readily available for patients at higher risk, SGTL2i or GLP-1RA use in patients hospitalized with coronary artery disease and T2D increased to 30% in 2021 but remains clearly suboptimal (128). Obviously, there are other patient-, system-, and provider-related barriers to the delivery of guideline-based care (129).

The Coordinating Cardiology clinics randomized trial of interventions to improve outcomes (COORDINATE) – Diabetes (COORDINATE-Diabetes) was designed and conducted based on these and other, similar data demonstrating inadequate delivery of guideline-based care to patients with T2D at high risk (130). This pragmatic, prospective, two-arm, cluster-randomized, controlled trial was conducted at cardiology clinics across the U.S. where participants were recruited between July 2019 and May 2022 and followed up through December 2022 (131). Twenty-five sites were randomized to the trial intervention, and 24 continued their usual-care practices. Trial participants were adults with T2D and established ASCVD who were not already taking all three of high-intensity statin, an ACEi or ARB, and an SGLT2i and/or a GLP-1RA and who did not have an absolute contraindication to use of those medications. The primary objective of the trial was to test the effectiveness of a clinic-level multifaceted intervention aimed at addressing locally important barriers to the adoption of guideline-recommended therapy in patients with T2D and ASCVD, with the primary outcome being the proportion of enrolled participants prescribed all three of the indicated therapies at the final visit at between 6 and 12 months of follow-up (131).

The intervention consisted of strategies provided to the clinics by a coordinating center trio of a cardiologist, an endocrinologist, and an implementation specialist who also had clinical nursing experience (131). The coordinating center trio initially traveled to each site to begin the intervention, but these meetings were conducted virtually after the onset of the coronavirus disease 2019 pandemic. At the intervention sites, there was an analysis of the local barriers to evidence-based care and development of local interdisciplinary teams and plans to address those issues.

Coordination of care among clinicians, particularly among the site cardiologists, local endocrinologists, and primary care clinicians, was facilitated at the sites. Education on current guidelines was delivered to the site clinicians and to the trial participants to emphasize the importance of multifactorial cardiovascular risk reduction, and with auditing feedback on quality metrics was provided to the intervention sites during each month of the trial (131).

The primary outcome analysis included data for 457 participants (99.6%) at the intervention sites and 588 participants (99.7%) at the usual-care sites who had medication data available at final follow-up (1,021 participants [97.3%] had available 12-month data) (131). Participants at intervention sites were significantly more likely than those at usual-care sites to be prescribed all three evidence-based therapies (high-intensity statins, ACEi or ARB, and SGLT2i and/or GLP-1RA) (173 of 457 [37.9%] vs. 85 of 588 [14.5%], respectively, a difference of 23.4%; odds ratio 4.46 [95% CI 2.55–7.80], P < 0.001). Importantly, intervention site participants were more likely than participants at usual-care sites to be prescribed medications from each of the three recommended medication groups, and the effect of the intervention did not differ by medication use at baseline. There was also 89.2% agreement between medication prescriptions and participant-reported medication use. In addition, consistent benefit of the trial intervention was seen in Hispanic and Black populations for the primary outcome and for the individual components of the primary outcome (132). The trial was not powered to assess the effect of the intervention on clinical events; however, a nonsignificant reduction was found for a composite clinical outcome of all-cause death or hospitalization for myocardial infarction, stroke, decompensated heart failure, or urgent revascularization. An outcome event occurred in 23 of 457 participants (5%) at the intervention sites compared with 40 of 588 participants (6.8%) at the usual-care sites (adjusted hazard ratio 0.79 [95% CI 0.46–1.33]) (131).

Although the effectiveness of the trial intervention was demonstrated in COORDINATE-Diabetes, the clinic sites and participants involved with the trial were selected and may not be representative of the broader U.S. clinical landscape. In addition, a substantial proportion of the enrolled participants were not receiving all three components of care at the end of follow-up. Initiatives that are focused on care delivery in other settings, such as the hospital or diabetes care clinic, may be needed to complement the types of interventions studied in COORDINATE-Diabetes. However, the trial was quite successful in increasing the delivery of guideline-based care to patients with T2D and ASCVD.

The trial intervention itself incorporated many of the key principles of implementation research but was also quite straightforward and likely can be readily adopted to a variety of clinical care settings. As outlined in Table 1, key stakeholders were engaged at each intervention site to identify and prioritize local barriers to delivery of guideline-based care, and coordination between cardiology and diabetes care specialists was highly encouraged. These individuals developed strategies to best address the identified barriers, with assistance from the trial team as necessary. Personnel at the intervention sites were also regularly provided data regarding prescription of indicated therapies to their enrolled participants, including progress over time and how their performance compared with that of the other sites. Full trial details and information on the materials used are available free of charge from the COORDINATE-Diabetes website (133).

Of course, a broader array of interventions are needed to reduce the risks of multiple diabetes-related complications. Examples of successful approaches include the adoption of artificial intelligence screening for diabetic retinopathy via retinal photography (134). Analysis of data from >30,000 individuals has shown that an artificial intelligence–enabled algorithm provided adequate sensitivity and specificity for high-risk retinopathy in a real-world setting, with the potential to reduce the human workload by half (135). A new public awareness initiative by the U.S. Department of Health and Human Services, the National Kidney Foundation, and the American Society of Nephrology, which provides education about the risks of chronic kidney disease and the need for its early detection with albuminuria screening, is intended to improve the outcomes of affected patients (136,137). Other analyses indicate that the implementation of guidelines established by the International Working Group on the Diabetic Foot improves outcomes including for amputations and is likely cost-effective (138–140). Given this accumulation of evidence, it is clear that there must be local, institutional, and systemic prioritization of initiatives to promote the delivery of guideline-based care in diabetes.

The gap between the existence of evidence-based interventions that improve diabetes outcomes and receipt of optimal care by people with or at risk for diabetes persists, but in this article we show how implementation science approaches to translating evidence into practice can help us to close that gap (2,5,141). Such systematic approaches may also be effective in identifying, addressing, and minimizing disparities in care. The studies highlighted in this article demonstrate the importance of conducting research in real-world settings that incorporates local context and includes consideration of long-term scalability and sustainability (47).

The K-DPP was designed explicitly to determine how the existing evidence on diabetes prevention interventions could be translated in lower-resourced areas (50). Important elements to note include its community-based approach from participant engagement to intervention implementation, acceptance of the tradeoff between intervention intensity and reach, and the multilevel nature of the intervention. The reach of the K-DPP—more than three quarters of eligible individuals actually received the intervention—was particularly remarkable given that in the U.S., the most recent published data suggest that <5% of eligible adults have participated in a NDDP lifestyle change program (32,142).

The ACDC intervention uses existing, funded clinical resources in the VHA integrated health system to address persistently elevated HbA_1c in individuals with diabetes already engaged in care (99). This multilevel intervention leverages telemedicine, which addresses transportation and logistical barriers, but importantly, ACDC provides multicomponent care outside of the traditional clinic visit paradigm. During telemedicine visits, along with addressing medication management, participants receive additional relevant services such as diabetes self-management education, which is generally underused in diabetes care (143). The ACDC intervention has also been adapted for local settings in which it is implemented, with acknowledgment of heterogeneity even across sites within the same integrated health system (89). The example of the effective ACDC intervention also highlights the importance of operational relationships in translating evidence into practice in the clinical setting.

For COORDINATE-Diabetes a multilevel, multicomponent intervention was implemented in cardiology clinics with tailoring to local clinics (130). A distinguishing feature of the COORDINATE-Diabetes implementation approach is the focus on receipt of multiple evidence-based interventions among individuals with T2D and ASCVD: prescription of a high-intensity statin, prescription of an ACEi or ARB, and prescription of an SGLT-2i or a GLP-1RA.

These and multitudes of other examples of successful implementation of evidence-based interventions demonstrate that we can indeed translate evidence into practice to stem the tide of the epidemic of T2D and its complications. A true transformation of how we approach clinical medicine is needed, though, and this requires a substantial reallocation of resources. Our system of care needs to expand beyond its focus on visit-based care, which relies mainly on the one-to-one patient-provider interaction; evidence-based care has to occur in multiple settings and at multiple levels tailored to local context in acknowledgment of the social determinants of health. Team-based care has to become the standard. Each time we implement a quality improvement intervention, we must consider local resources and what is truly scalable beyond the confines of a given project. In diabetes, we can no longer focus on a single outcome such as HbA_1c as there are multiple targets to improve across the spectrum of screening and monitoring; receipt of clinical services, including diabetes self-management training and appropriate technology; weight management; receipt of indicated medications; and cardiometabolic risk factors. We have the evidence in our hands, and each of us involved in the diabetes ecosystem must consider how to change the system at our level, informed by the approaches described herein, so that someone else is not writing this commentary again in 10 years.

Funding. This publication was supported by the American Diabetes Association. M.J.C. acknowledges support for the ACDC program from the Veterans Health Administration Office of Rural Health (Veterans Rural Health Resource Center, Iowa City), the Veterans Affairs Quality Enhancement Research Initiative (VA QUE 20-012), and the Durham Center of Innovation to Accelerate Discovery and Practice Transformation within the Durham VA Health Care System (VA CIN 13-410). S.T. was supported by the Woodruff Health Sciences Center Synergy Awards and the Georgia Clinical & Translational Science Alliance (CTSA) funded by the National Center for Advancing Translational Sciences (NCATS) of the National Institutes of Health (NIH) under award no. UL1TR002378. S.T. was also partially supported by grant no. 75D30120P0742 from the Centers for Disease Control and Prevention (CDC), Atlanta, GA.

Duality of Interest. J.B.G. reports research support from Boehringer Ingelheim, Lilly, Merck, Roche, and Bluedrop and serving as an advisor or a consultant for Boehringer Ingelheim, Lilly, Bayer, AstraZeneca, Valo, Anji Pharmaceuticals, Vertex Pharmaceuticals, and Novo Nordisk. N.M.M. is a co-inventor of an online diabetes prevention program. Under a license agreement between Johns Hopkins HealthCare Solutions and Johns Hopkins University, N.M.M. and the university were entitled to royalty distributions. This technology is not discussed in this publication. No other potential conflicts of interest relevant to this article were reported.

Prior Presentation. Parts of this work were presented at the 84th Scientific Sessions of the American Diabetes Association, Orlando, FL, 21–24 June 2024. A video presentation can be found in the online version of the article at https://doi.org/10.2337/dci24-0001.

Handling Editors. The journal editor responsible for overseeing the review of the manuscript was Mark A. Atkinson.