US
Coaching, peer health coaching, and peer support use health care providers or volunteers, collectively referred to as coaches or peer supporters, to provide self-management support for persons who may be considered peers or who have the same health condition [ 27 , 28 ]. These coaches and peers can include patients, community health workers, lay educators, family members, and health care professionals. Peer health coaching is based on the idea that a patient will connect to others who have similar experiences [ 29 ]. Regardless as to the type of coaching or peer support, the goal is to engage and motivate patients in self-management.
Coaching and peer support interventions have been well documented in diabetes education. In the context of diabetes support, coaches and peers can have multiple roles, including educator, advocate, cultural translator, mentor, case manager, and group facilitator [ 27 ]. Peer coaching and support is most commonly delivered by a trained peer, and focuses on self-management interventions that are time limited and based on a scripted standardized curricula [ 30 ]. In terms of effectiveness, peer health coaching and support have been successful in improving self-management and in lowering HbA1c [ 31 ]. Because of these favorable results, peer health coaching and support has received increased interest as a model for more long-term diabetes self-management support interventions.
For this integrative review, ten randomized controlled trials (RCT) were reviewed (see Table 2 ). Locations of research included seven studies in the United States, and one in the Netherlands, Thailand, and Australia. Six studies compared peer-led interventions, two compared health professional-led interventions, one compared a CHW intervention, and one used a family-oriented approach to self-management, with all intervention groups being compared to usual care. The duration of the interventions ranged from 4 weeks to 18 months. Studies showed improvements in self-efficacy and knowledge of self-management [ [32] , [33] , [34] , [35] , [36] ]. Results for reduction in HbA1c were mixed. Four studies described reductions in HbA1c levels in peer-led intervention groups and CHWs [ 1 , 28 , 31 , 37 ]; three studies showed no significant reduction in HbA1c levels [ 32 , 33 , 38 ].
Peer health coaching/peer.
Study & Location | Design | Sample | Outcome Measures | Intervention (I) and Control (C) Groups | Results |
---|---|---|---|---|---|
Nishita et al. (2012) [ ] US | RCT | = 190 Female: 62.6% Mean age: 48.5 Hawaiian or Asian: 71% | Height, weight, and HbA1c, & self-reported self-efficacy, & QOL, collected at baseline, 6 and 12 months. | (I): ( = 128) Over 12 months, individualized, self-directed support from life coach and a pharmacist. Appointments made by individual participants (LTA = 45); Intervention delivered by pharmacist and “trained” life coach (bachelor's degree in social sciences); Intervention fidelity addressed. (C): ( = 62) No treatment; (LTA = 10). | No significant difference between groups on HbA1c or BMI. Self-efficacy and QOL improved in those subjects who had 10 or more sessions. |
Ruggiero et al. (2014) [ ] US | RCT | = 270 Female: 68.8% Mean age: 53.2 African American: 52.6%; Hispanic/Latino: 47.4% | A1c, BMI, & self-reported self-care, depressive symptoms, & confidence, collected at baseline, 6 and 12 months. | (I): ( = 136) Medical Assistant coaching intervention delivered by trained MA's over a 12-month period with in-person contacts at regular clinic visits (30 min sessions), and monthly follow-up phone calls in between visits. The focus was on providing information and skills to make informed self-care choices and changes; (LTA = 43); Intervention delivered by medical assistants; Intervention fidelity addressed. (C): ( = 134) Usual care; (LTA = 51). | All groups reported improvements in self-care across time, but no intervention effect was found. No differences were found in HbA1c between groups or across time. |
Wichit et al. (2017) [ ] Thailand | RCT | = 140 Female: 72.8% Mean age: 58.4 | Self-management activities, QOL, self-efficacy, and HbA1c, collected at baseline, 5 weeks and 13 weeks. | (I): ( = 70) Family intervention consisting of three 2-h group session delivered at baseline, 5 weeks and 9 weeks. Groups of 8–12 dyads (patient and family member); (LTA = 3); Intervention delivered by nurse; Intervention fidelity not addressed. (C): ( = 70) Usual care; (LTA = 3). | Improvements seen in self-efficacy, self-management, and QOL in the intervention group. No between group differences in HbA1c. |
Wu et al. (2010) [ ] Australia | RCT | = 30 Female: 28.6% Mean age range: 62.7–71.5 | Self-reported self-efficacy, self-management behavior, & knowledge, collected at baseline & 4-week follow up. | (I): ( = 15) Usual care plus peer support (Peer CDSMP). The program is 3 face to face sessions with research nurse (week 1), and follow up weeks 2–4 by peers who used weekly one telephone call and two text messages after each phone call; Intervention delivered by nurses and “trained” peers; Intervention fidelity not addressed. (C): ( = 13) Usual care. | Significant differences in knowledge were found for the intervention group, but no differences between the two groups over time for self-efficacy and self-management. |
Van der Wulp et al. (2012) [ ] Netherlands | RCT | = 133 Female: 45.4% Mean age: 54 | Self-reported self-efficacy, coping, diet, physical activity, well-being, depressive symptoms, & distress, collected at baseline, 3 and 6 months. | (I): ( = 68) Three monthly home visits by a peer (expert patient) with a follow up phone call or email within two weeks after each visit. Visit 1 explored areas of lifestyle change. Visit 2 had participants assign importance and feasibility to proposed lifestyle changes, and set goals related to those changes. Visit 3 evaluated goals; (LTA = 9); Intervention delivered by “trained” expert patient peers; Intervention fidelity addressed. (C): ( = 65) Usual care; (LTA = 5). | The peer-lead coaching intervention improved self-efficacy in patients experiencing low self-efficacy. No significant differences were found in remaining outcome variables. |
Carrasquillo et al. (2017) [ ] US | RCT | = 300 Female: 55% Mean age: 55.2 Latino | BP, lipids, HbA1c, BMI, & self-reported diet, physical activity, and medication adherence, collected at baseline and 12 months. | (I): ( = 150) CHW intervention for 12 months that included 4 home visits and 12 phone calls, and additional monthly CHW led educational groups; (LTA = 39). Intervention delivered by a “trained” CHW; Intervention fidelity addressed. (C): ( = 150) Enhanced usual care that included additional mailed educational materials; (LTA = 46). | The intervention group had lower HbA1c (reduction of 0.51), compared to control. No difference in any other outcome variables. |
Moskowitz et al. (2013) [ ] US | RCT | = 299 Female: 51.4–53% Mean age: 54.1–56.3 | A1c, and self-reported depression, social support, literacy, & self-management, collected at baseline and at 6 months. | (I): ( = 148) Coaching intervention with peer coaches interacting with patients in person – telephone contact 2 times/month, and an in-person contact 2 or more times over 6 months. Intervention delivered by “trained” peers; Intervention fidelity not addressed. (C): ( = 151) Usual care. Study attrition not addressed | Peer health coaching was more effective in lowering HbA1c for patients with low medication adherence and self-management than for patients with higher levels of adherence and self-management. |
Sinclair et al. (2013) [ ] US | RCT | = 82 Female: 63% Mean age: 53-55 Hawaii | A1c, height, weight, BP, & lipids, collected at baseline and 6 months. | (I): ( = 48) Diabetes self-management program (Partners in Care), led by peer educators. Focus on knowledge and skills related to blood glucose monitoring, adherence to medications, healthy eating, physical activity, and stress reduction; (LTA = 14); Intervention delivered by “trained” peer educators; Intervention fidelity addressed. (C): ( = 34) Wait listed for intervention; (LTA = 3). | Significant reduction in HbA1c (reduction of 1.6) and distress in intervention group at 6 months. |
Thom et al. (2013) [ ] US | RCT | = 299 Female: >50% Mean age: 56.1 African American: 30.7–37.5% | A1c, lipids, height, weight, BMI, & BP, collected at baseline and 6 months. | (I): ( = 148) Coaching intervention with peer coaches interacting with patients in person - telephone contact 2 times/month, and an in-person contact 2 or more times over 6 months; (LTA = 8); Intervention delivered by “trained” peer coaches; Intervention fidelity not addressed. (C): ( = 151) Usual care; (LTA = 16). | At 6 months, significant differences in HbA1c levels, with reduction of 1.07% in intervention group, and only 0.3% in the usual care group. |
Tang et al. (2015) [ ] US | RCT | = 106 Female: 67% Mean age: 56.3 African American | HbA1c, lipids, BP, BMI, waist circumference, & self-reported distress, & social support, collected at baseline, 3, 9, and 15 months. | (I): ( = 54) 3 months DSME plus 12 months peer support; (LTA = 20); Intervention delivered by nurses and peer leaders; Intervention fidelity not addressed. (C): ( = 52) 3 months DSME; (LTA = 20). | No significant changes in HbA1c between groups. |
Problem solving therapy (PST) is an intervention approach for behavior change that entails a series of cognitive operations used to figure out what to do when the way to reach a goal is not apparent [ 39 ]. The goal of PST is to facilitate behavior change, aiming to facilitate positive emotional reactions and reduce negative emotional reactions [ 40 ]. PST involves teaching the patient a step-by-step process to solving life problems, generally broken down into two major parts: applying a problem solving orientation to life, and using problem solving skills [ 41 ]. PST is based on teaching the following skills: (1) identifying a problem, (2) defining the problem, (3) understanding the problem, (4) setting goals related to the problem, (5) identifying alternative solutions, (6) evaluating and choosing best alternatives, (7) implementing alternatives, and (8) evaluating the effort at problem solving [ 42 ].
PST has a long history in clinical and counseling psychology to address multiple mental health disorders, family and relational distress, stress management and coping skills, and substance abuse [ 39 ]. PST has been a frequently used component of interventions within diabetes education and care, usually one component of a larger diabetes self-management intervention. PST has been recognized as an important process, intervention, and skill in diabetes self-management [ 43 ].
For this integrative review, three studies were reviewed: a RCT, a systematic review, and a meta-analysis (see Table 3 ). Locations of research included the United States, with systematic reviews including studies from English and Chinese electronic databases. The RCT compared an intensive program including eight PST session to a condensed program including just one PST session. Results showed a significant difference in HbA1c (0.71%) in the intensive PST group [ 44 ]. The systematic review assessed 56 papers exploring the association of PST to diabetes self-management and control. Six studies used PST as an intervention for adults. Results of the review suggest that evidence for the effectiveness of PST on HbA1c is weak [ 45 ]. The meta-analysis assessed 16 RCTs of interactive self-management interventions, with seven being specific to PST. The studies specific to PST showed a mean difference of −0.39% when comparing intervention to control groups, demonstrating a significant reduction in HbA1c [ 46 ].
Problem-solving therapy.
Study & Location | Design | Sample | Outcome Measures | Intervention (I) and Control (C) Groups | Results |
---|---|---|---|---|---|
Hill-Briggs et al. (2011) [ ] US | RCT | = 56 Female: 58.9% Mean age: 61.3 African American | A1c, lipids, BP, literacy, & self-reported depression, knowledge, health problems, barriers, self-management, & satisfaction, collected at baseline, 1-week post-intervention, & 3 months. | (I - Intensive group): ( = 29) 1 session (diabetes and CVD education session + 8 PST session - delivered bi-weekly, 8–10 participants/group); (LTA = 3); Intervention delivered by “trained interventionist”; Intervention fidelity addressed. (I - Condensed group): ( = 27) CVD education session + one PST session; (LTA = 1). | Intensive group had significant improvement in SBP, DBP, LDL, and cholesterol, improved HbA1c (reduction of 0.71%), problem solving skills, self-management behavior of diet, and knowledge. |
Hill-Briggs et al. (2007) [ ] | Systematic review | = 52 Qualitative, quantitative, cross-sectional prospective, RCTs, and quasi-experimental designs Type 1 and type 2 diabetes | Problem solving, self-management behaviors, physiological, psychosocial, and process outcomes. | Six studies of adults (out of 52 studies) used problem solving as an intervention. | Ineffective problem solving was associated with poor glycemic control; more studies are needed to make conclusions about the impact of problem solving on self-management; evidence for problem solving effectiveness on HbA1c is inconsistent and weak. |
Cheng et al. (2017) [ ] | Meta-analysis | = 16 RCTs Adults type 2 diabetes English and Chinese | A1c | Seven studies of adults (out of 16 studies) used problem solving as an intervention. | Problem solving studies showed a mean difference in HbA1c of −0.39% (95% CI: −.73% to −.05%; p = .03). |
Technology based interventions involve the use of equipment, devices, or tools to augment care through improved communication and increased ability to process information. Often referred to as telehealth, these various modalities include telephone, teleconferencing by video, computer, and internet/web-based technology [ 47 ]. Technology based interventions incorporate various technological modalities to monitor outcomes, provide self-management education, and deliver self-management strategies.
In general, technology based interventions have been used to provide support for patients with multiple health conditions including heart disease, chronic lung disease, and diabetes [ 48 ]. These telehealth interventions have been developed in response to access to care issues in various rural and regional communities [ 47 ]. The telephone is a customary technology that is commonly available for communication with patients [ 49 ]. More advanced telephone management includes mobile phone-based applications, referred to commonly as apps, which allow smart-phone applications and texting in addition to basic telephone components. Videoconferencing requires even more complex technology, such as webcams and software to communicate by video. Computer-assisted modules (CAM) typically include computer hardware and software that provide programs for education and/or support. CAMs can be further stratified to include web-based interventions. A final category includes mixed modalities of these various components. Systematic reviews of technology interventions for mixed populations of type 1 and type 2 diabetes have shown limited to no impact on hemoglobin HbA1c [ 47 , 50 , 51 ]. For this integrative review, 30 studies were reviewed (see Table 4 ). Research on telehealth yielded articles on telephone/mobile phone (16), computer-assisted modules (2), web-based interventions (7) and mixed modalities (5). Locations of research included eleven studies in the United States, and thirteen studies done in eight different countries.
Technology based interventions.
Study & Location | Design | Sample | Outcome Measures | Intervention (I) and Control (C) Groups | Results |
---|---|---|---|---|---|
Wu et al. (2010) [ ] | Systematic review & meta-analysis | = 7 RCTs, ≥16 years old Type 2 diabetes White: 81% | A1c | Telephone follow up interventions directed at improving self-management in comparison with a control group in which the telephone was the only difference in the intervention being provided. (I): = 1020. (C): = 744. | Standardized effect of the telephone follow up showed a mean weighted difference in HbA1c of −0.44% in favor of the intervention. |
Graziano et al. (2009) [ ] US | RCT | = 120 Female: 45% Mean age: 62 White: 77% | A1c, medication changes, SMBG, & self-reported perceived severity, perceived susceptibility, perceived benefits, barriers, & attitudes, collect at baseline and 90 days. | (I): ( = 62) Usual care plus a daily automated prerecorded voice message relaying a short (less than 1 min) message focused on self-care behaviors to influence attitudes and beliefs, and reduce barriers for self-care behaviors; (LTA = 1); Intervention delivered by “investigator”; Intervention fidelity not addressed. . (C): ( = 58) Usual care; (LTA = 4). | No significant change in HbA1c or secondary outcomes between groups, except for SMBG. The telephone group had significant increase in frequency of SMBG. |
Williams et al.(2012) [ ] Australia | RCT | = 120 Female: 37% Mean age: 57.4 Australian born: 70%. | A1c, & self-reported health-related QOL, collected at baseline and 6 months. | (I): ( = 60) Telephone Linked Care (TLC) with automated interactive telephone response, where users had to call in weekly. Calls lasted 5–20 min, and system gave feedback and encouragement based on participant responses; (LTA = 9); Intervention delivered by “coordinator”; Intervention fidelity not addressed. (C): ( = 60) Usual care; (LTA = 5). | The intervention group had a significant reduction in HbA1c (0.8%) compared to the control group (0.2%), and in mental health related QOL. |
Lim et al.(2011) [ ] Korea | RCT | = 154 Females: 55.8% Mean age: 67.5 Korean | A1c, weight, BMI, glucose levels, lipids, & SMBG, collected at baseline, 3 and 6 months. | Three groups: (I-1): ( = 51) SMBG group; (LTA = 4). (I-2): ( = 51) U-healthcare group that received a glucometer that transmitted SMBG readings to the Clinical Decision Support Server, with subsequent participant feedback message on their mobile phone; (LTA = 2). Intervention delivered by diabetologists, nurses, dieticians, and exercise trainers; Intervention fidelity not addressed. (C): ( = 52) Usual care; (LTA = 4). | U-healthcare group had significant improvement in HbA1c and SMBG, but did not meet study goal of less than 7% for HbA1c. No other significant findings. |
Walker et al. (2011) [ ] US | RCT | = 526 Female: 67.1% Mean age: 55.5 Black: 62%; Hispanic: 23%; 77% foreign born | A1c, medication adherence (pill counts), & self-reported self-management behaviors, collected at baseline and 12 months. | (I): ( = 262) Telephone intervention involving 10 calls at 4–6 week intervals from a health educator over a 12-month period. Focus was on medication and life style changes (no face-to-face interaction); (LTA = 34); Intervention delivered by “health educators” supervised by nurses; Intervention fidelity addressed. (C): ( = 264) Print materials only (no face-to-face interaction). Outcome variables of HbA1c, medication adherence (pill counts), and self-reported self-management behaviors collect at baseline and 12 months; (LTA = 48). | Telephone group had greater reduction in HbA1c (0.23% ± 1.1%) over 1 year, and improved medication adherence among those not taking insulin. No significant changes in self-management behaviors were related to HbA1c changes. |
Trief et al. (2016) [ ] US | RCT | = 280 Female: 38.4% Mean age: 56.8 30% self-described minority | A1c, BMI, BP, distress, self-efficacy, depressive symptoms, & satisfaction collected at baseline, 4, 8, and 12 months. | Three arms: (IC): ( = 94) Individual call group, with 2 phone sessions, plus 10 additional calls (50–55 min) addressing self-management; (LTA = 1); Intervention delivered by dieticians; Intervention fidelity addressed. (CC): ( = 104) Collaborative couple call group, with 2 phone sessions, plus 10 additional calls (50–55 min) addressing self-management; (LTA = 7). (DE): ( = 82) Diabetes education with 2 phone sessions and no additional contact; (LTA = 4). | Significant reduction in HbA1c in all groups with no difference between groups. The Collaborative Couples intervention resulted in lasting improvements in HbA1c, obesity, and psychosocial variables. |
Goode et al. (2015) [ ] Australia | RCT | = 302 Female: 72% Mean age: 57.8 Caucasian: 43.7% | Weight, PA, HbA1c, & diet collected at baseline, 6, 18, and 24 months. | (I): ( = 135) 18-month intervention with 27 phone calls, weekly for first 4 weeks, then every 2 weeks for 5 months, then monthly for the remaining 12 months. Counseling to increase PA, diet, and weight loss provided. Given pedometer and digital scales; (LTA = 33); Intervention delivered by counselors with bachelor's-level training in nutrition and dietetics; Intervention fidelity addressed. (C): ( = 144) Usual care plus educational brochures; (LTA = 13). | Increased dose of intervention was associated in greater weight loss. |
Sacco et al. (2009) [ ] US | RCT | = 62 Female: 58% Mean age: 52 Caucasian: 77% African American: 14.5% Hispanic: 8.1% | A1c, BMI, and self-report of symptoms, depression, knowledge, self-efficacy, awareness of goals, and adherence to diet, SMBG, foot care, & medications, collected at baseline and 6 months. | (I): ( = 31) Telephone coaching call weekly for 3 months, then bi-weekly for additional 3 months. Telephone sessions averaged 17.4 min. Telephone sessions were guided by a Weekly Coaching Checklist addressing self-care, and reviewed weekly blood glucose readings; (LTA = 10); Interventions delivered by undergraduate psychology students; Intervention fidelity addressed. (C): ( = 31) Usual care; (LTA = 4). | Significant treatment effects on adherence, diabetes-related medical symptoms, and depression Symptoms. No significant effects on BMI or HbA1c. |
Anderson et al. (2010) [ ] US | RCT | = 295 Female: 58% Age: > 18 White: 26–27% Other: 62–65% with majority being African American or Hispanic | Weight, BMI, HbA1c, lipids, & BP, and self-reported overall health, depressive symptoms, diet and physical activity, collected at baseline, 6 and 12 months. | (I): ( = 146) One-year of telephonic disease management with phone calls including a brief clinical assessment, self-management discussion. Patients were called weekly, bi-weekly or monthly depending on a risk-stratification, or if the patient requested a change in call frequency; (LTA = 52); Intervention delivered by nurses; Intervention fidelity not addressed. (C): ( = 149) Usual care; (LTA = 32). | No significant difference in HbA1c or other secondary outcome measures after 12 months. |
Frosch et al. (2011) [ ] US | RCT | = 201 Female: 50% Mean age: 55 Latino: 55%; African American:16%; White: 20% | A1c, lipids, BP, BMI, & prescribed medications, and self-reported knowledge of self-management behaviors, collected at baseline, 1 and 6 months. | (I): ( = 100). A 24-min video behavior support intervention with a workbook and 5 sessions of telephone coaching by a trained diabetes nurse. The telephone sessions varied in length from 15 to 60 min with a cap of 150 min total. Time intervals between calls determined collaboratively; (LTA = 17); Intervention delivered by nurses; Intervention fidelity not addressed. (C): ( = 101) Usual care; (LTA = 14). | No significant overall reduction in HbA1c between groups. Secondary outcome measures were nonsignificant. |
Nesari et al. (2010) [ ] Iran | RCT | = 61 Female: 71.7% Mean age: 51 Iranian | A1c, and self-reported disease characteristics, diet, exercise, medications, foot care, and SMBG, collected at baseline and after 12 weeks. | (I): ( = 30) Telephone follow up 12 weeks, twice weekly for the first month and then weekly for second and third months. Each session averaged 20 min and each person received 16 phone calls. Calls included self-management education, and medication adjustments coordinated by the nurse and consulting endocrinologist; (LTA = 0); Intervention delivered by nursing student; Intervention fidelity not addressed. (C): ( = 30) Usual care; (LTA = 1). | No significant HbA1c change between groups; Significant changes in adherence for diet, exercise, foot care, medication taking and SMBG. |
Wayne et al. (2015) [ ] Canada | RCT | = 131 Female: 72% Mean age: 53.2 Black: 45%; Caucasian: 27% | A1c, weight, BMI, & waist circumference collected at baseline, 3 and 6 months. | (I): ( = 67) 6-month intervention using a health coach and smart phone, with 24/7 access to coach; (LTA = 19); Intervention delivered by behavior-change counseling specialist; Intervention fidelity not addressed. (C): ( = 64) Using health coach, but no smart phone; (LTA = 15). | No difference between groups in HbA1c reduction. Both groups reduced HbA1c (−0.84 intervention; −0.81 control). |
Cui et al. (2106) [ ] | Systematic review | = 13 Adults with type 2 diabetes from 7 countries: Finland, Norway, US, Korea, Spain, Canada, Netherlands | A1c Baseline and at study completion | Thirteen RCTs compared mHealth smart phone applications to control groups receiving usual care only. Studies included a primary outcome variable of HbA1c, and measured change in HbA1c. | Significant reduction in HbA1c by 0.40% (p < .01) mean difference, when compared to control group. |
Wu et al.(2018) [ ]. | Systematic review & meta-analysis | = 17 Adults with type 2 diabetes | A1c Baseline and at study completion | Seventeen RCTs of smartphone technology that used apps or internet access via the smartphone or personal digital assistants, compared to a control group receiving usual care only. Outcome variable of HbA1c, and measured change in HbA1c. | Meta-analysis showed a pooled HbA1c reduction of −0.51% when comparing smartphone technology to usual care. |
Aikens et al.(2015) [ ] US | Descriptive comparative study | = 301 Male: 92.8% Mean age 66.7 Caucasian: 92.8% from Veterans Affairs clinics | Self-reported self-management behaviors, physical & mental functioning, depressive symptoms, & distress, collected at baseline, 3 and 6 months. | Two intervention groups: a 3 month group ( = 108), and a 6 month group ( = 193). The intervention was an Interactive voice response (IVR) mobile health service with questions via a tree-structured algorithm and verbal reinforcement for self-management. Calls were 5–10 min, and performed weekly for 3 or 6 months. A pattern of abnormal blood glucose or BP triggered a clinician notification for follow up. Attrition for total sample 23%, more likely in the 6-month group; Intervention delivered by research team; Intervention fidelity addressed. | Significant improvements in all health outcomes (except psychological functioning), and in self-management behaviors of medications, SMBG, and foot care. Duration of study had no significant effects on IVR outcomes. |
Hou et al. (2016) [ ] | Systematic review | = 14 RCTs Adults with type 1 or type 2 diabetes | A1c (baseline and follow up, and not self-reported) | Ten RCTs (out of 14) were of type 2 diabetes, and using a total 9 different apps for type 2 diabetes. Apps were designed to improve self-management by providing personalize feedback on self-monitoring of blood glucose, diet, and physical activity | All studies of type 2 diabetes reported a mean reduction in HbA1c of 0.49% compared to controls. |
Pal et al. (2014) [ ] | Systematic review | = 16 RCTs(UK) Adults with type 2 diabetes | A1c, BP, lipids, weight, death, health-related QOL, changes in cognition, behaviors, social support, emotional outcomes, adverse effect, complications, & economic data. | Interventions included those that were computer-based and interactive with users to generate tailored content aimed at improving self-management. | Computer-based interventions had a small effect on HbA1c, with a pooled effect of −0.2%, with the sub-group of mobile phone-based interventions having a larger effect (−0.50%) on HbA1c. No evidence of benefit for other biological, cognitive, behavioral or emotional outcomes. |
Jaipakdee et al. (2015) [ ] Thailand | RCT | = 403 Females: 76.7% Mean age: 61.3 | A1c, glucose, weight, BMI, BP, waist circumference, and self-reported depressive symptoms, self-management behaviors, & QOL, collected at baseline, 3 and 6 months. | (I): ( = 203) DSMS over 6 months with computer-assisted instruction (CAI) that included educational sessions by computer plus a monthly 3 h educational session; (LTA = 9); Intervention delivered by nurses; Intervention fidelity addressed. (C): ( = 200) Usual care; (LTA = 16). | Significant improvements in HbA1c (reduction of 0.34), fasting blood glucose, health behaviors, and QOL in intervention group. |
Pacaud et al. (2012) [ ] Canada | RCT | = 79 Female: 52.9% Mean age: 54.2 | A1c, diabetes knowledge, self-efficacy, self-care behaviors, satisfaction, QOL, collected at baseline, 3, 6, 9, & 12 months. | Two intervention conditions: (I-1): ( = 18) Web static group; (I-2): ( = 29) Web interactive group. (C): ( = 21) Standard face-to-face care. All groups received 60–90 min assessment with trained clinician and research assistant. Follow up during study was done by same clinicians for each group.(LTA: of the 79 enrolled, LTA 25% web static group, 16% face-to-face group, 2.6% web interactive group. . | Significant findings when comparing website use, such that higher website use was associated with higher knowledge, self-efficacy, and lower HbA1c. |
Hansel et al. (2017) [ ] France | RCT | = 120 Female: 66.7% Mean age: 57 | Weight, waist circumference, BMI, lipids, HbA1c, aerobic fitness, & self-reported diet, physical activity, & satisfaction collected at baseline and 4 months. | (I): ( = 60) Web-based support tool designed to improve lifestyle habits, including diet and PA. Participants progress through modules as they answer questions. Human contact is limited to technical support. Program runs on a personal computer; (LTA = 11); Intervention delivered by study team; Intervention fidelity not addressed. (C): ( = 60) Usual care; (LTA = 5). | Significant improvements in HbA1c, weight and waist circumference in intervention group at 4 months. |
Avdal et al. (2011) [ ] RCT Turkey | = 122 Female: 50.8% Mean age: 51.5 | A1c & rate of attendance at health check visits were collected at baseline and 6 months. | (I): ( = 61) Web site intervention that provided information, education, and feedback; (LTA = 9); Intervention delivered by nurses; Intervention fidelity not addressed. . (C): ( = 61) Usual care; (LTA = 8). | The intervention group had a mean reduction (0.13) in HbA1c, and increased health check visits. No significant changes seen in the control group. | |
Glasko et al. (2012) [ ] US | RCT | = 463 Female: 50% Mean age: 58 White: 72%, African American: 15%, Latino 21% | A1c, BMI, lipids, BP, health literacy, and self-reported diet, physical activity, medication adherence, self-efficacy, problem solving, supportive sources, health status, distress, collected at baseline, 4 and 12 months. | 3 arm trial using CASM, an internet-based computer assisted self-management intervention. (Group 1): ( = 169) CASM (LTA = 49). (Group 2): ( = 162) CASM+, with added human support; (LTA = 38). (Group 3): ( = 132) Enhanced usual care group that included a computer-based health risk appraisal feedback and recommended preventive care behaviors but did not include the key intervention procedures; (LTA = 18). Intervention delivered by research team; Intervention fidelity not addressed. | Internet based programs significantly improved health care behaviors compared to usual care. All conditions improved moderately on biological and psychosocial outcomes, but between group differences not significant. |
Lorig et al. (2010) [ ] US | RCT | = 761 Female: 76% Mean age: 54.3 White: 76% | A1c, and self-reported health status, health care utilization, patient activation, self-efficacy, distress, & physical activity, collected at baseline, 6, and 18 months. | 3 arm trial: (Group 1): ( = 259) Internet-based Diabetes Self-Management Program (IDSMP) that included a 6-week asynchronous training program with 6 weekly sessions and a reference book; (LTA = 50). (Group 2): ( = 232) IDSMP plus e-mail reinforcement; (LTA = 46). Intervention delivered by “trained” peer facilitators; Intervention fidelity not addressed. (C): ( = 270) Usual care; (LTA = 32). | Significant improvements in HbA1c, patient activation, and self-efficacy at 6 months, and self-efficacy and patient activation at 18 month, for the intervention groups. No changes in other health or behavioral indicators. |
Heinrich et al. (2012) [ ] Netherlands | RCT | = 99 Female: Mean age: | Diabetes self-management knowledge, and use of website intervention, collected at baseline and two weeks. | (I): ( = 43) Web-based Diabetes Interactive Education Programme (DIEP) that provides an overview of type 2 diabetes in seven chapters; (LTA = 7); Interventions delivered by research team; intervention fidelity not addressed. (C): ( = 56) Usual care; (LTA = 2). | Significant improvement in knowledge scores in the experimental group at post-test. The total time spent on the website averaged 58 min, and was not correlated to increased knowledge. |
Tang et al. (2013) [ ] US | Cohort study | = 415 Female: 40% Mean age: 54 White: 59% Asian: 21% Hispanic: 10% | A1c, BP, lipids, cardiovascular risk, and self-reported knowledge, distress, depression & treatment satisfaction, collected at 6 and 12 months. | (I): ( = 202): An online, disease management support system that included wirelessly uploaded home glucometer readings with graphical feedback, comprehensive patient-specific diabetes summary status report, nutrition and exercise logs, insulin record, online messaging with the health team; (LTA = 9); Intervention delivered by nurses and dieticians; Intervention fidelity not addressed. (C): ( = 213) Usual care; (LTA = 24). | Compared to usual care, the intervention group had significant HbA1c reduction at 6 months (reduction of 1.32), but no significant differences between groups on HbA1c at 12 months. |
Jackson et al. (2006) [ ] | Systematic review | = 26 RCTs & observational studies Type 1 & type 2 diabetes | A1c, weight, BP, micro-albumin, lipids, creatinine, depression, hematocrit, & health care utilization, self-care behaviors, satisfaction, & cost. | 14 out of 26 studies were RCTs. Studies used various technologies including internet (3 RCTs), telephone (4 RCTs), and computer-assisted integration of clinical information (7 RCTs). | Six out of 14 RCTs showed significant declines in HbA1c (>1%) when compared with controls. Overall increases in patient satisfaction with the interventions, personal health care, perceived support, QOL, and knowledge. |
Fisher et al. (2013) [ ] US | Cohort study | = 392 Female: 53.8% Mean age: 56 White: 40% Asian: 19% African American: 16% Hispanic: 11% | A1c, & self-reported diabetes distress, & self-reported physical activity, diet, & medication adherence, collected at baseline, 4 and 12 months. | 3 Intervention groups: All groups received live phone calls at weeks 2, 4, 7, 12, 24, 28, 34 & 48 to check progress. (Group 1): ( = 150) Computer-Assisted Self-Management (CASM) is a 40 min web-based diabetes program with interactive self-management feedback, and a booster program at month 5; (LTA = 29). (Group 2): ( = 146) CASM plus PST (CAPS) included a 60-min in-person intervention which introduced PST in addition to the CASM and a live booster session at month 5; (LTA = 29). (Group 3): ( = 96) Leap Ahead (LEAP) is a minimal intervention with a 20-min computer-delivered health risk appraisal along with diabetes information regarding healthy living, and a repeat risk appraisal at month 5; (LTA = 15). Intervention delivered by nonprofessional college graduate interventionists; Intervention fidelity not addressed. | No significant time or group main effects were found for HbA1c. Significant reductions in distress across all three groups without significant between group differences. |
Noh et al. (2010) [ ] Korea | Cohort study | = 44 Female: 22.5% Mean age: 42 Koreans | A1c, fasting and post-prandial blood glucose levels, collected at baseline and 6 months. | (I): ( = 24), eMOD intervention is a web-based system providing diabetes education that participants can log into when convenient by either cell phone or computer; (LTA = 4); Intervention delivered by research team; Intervention fidelity not addressed. (C): ( = 20) Received education books with content similar to eMOD website; (LTA = 0). | A1c reduction (1.53%) and post-prandial blood glucose decreased significantly over time in the eMOD group, with significant relationship between change in HbA1c and frequency of access to eMOD. |
Greenwood et al. (2015) [ ] US | RCT | = 90 Female: 23% Mean age: 58 Caucasian 64% | A1c, diabetes knowledge, self-management activities, & self-efficacy collected at baseline and 6 months. | (I): ( = 45) A telehealth remote monitoring system using a tablet connected to a modem and a glucometer that has a touch screen to answer daily health questions. Data is sent to a certified diabetes educator. 84 daily sessions delivered; (LTA = 4); Intervention delivered by certified diabetes educators; Intervention fidelity not addressed. (C): ( = 45) Usual care; (LTA = 5). | Both groups lowered HbA1c with a significant difference (−.41%) at 6 months, with greater reduction in the intervention group. |
Wild et al. (2016) [ ] UK | RCT | = 321 Female: 33.3% Mean age: 61 Ethnicity/race not reported | A1c, BP, weight, lipids, self-reported self-management, & QOL collected at baseline and 9 months. | (I): ( = 160) 9-month telehealth intervention using remote monitoring equipment for weight, BP, and blood glucose, with the information being delivered via modem to nurses. Advice then given to participant based on data at weekly intervals, and as needed; (LTA = 14); Intervention delivered by research team; Intervention fidelity not addressed. (C): ( = 161) Usual care; (LTA = 22). | Intervention group showed reduction in HbA1c (0.51%), and blood pressure. No differences between groups in weight, self-management behaviors, or QOL. |
For telephone interventions, a systematic review with meta-analysis of seven RCTs examining the impact of telephone follow up interventions on glucose control found little impact on glycemic control, with a mean weighted difference in HbA1c of −0.44% in favor of the intervention [ 35 ]. Eleven RCTs studied the impact of telephone interventions on glycemic control, symptoms, and self-management behaviors. Two RCTs that explored the impact of automated response systems showed no improvement in HbA1c [ 49 , 52 ]. Nine RCTs examined live telephonic interactive interventions that involved a consultation, counseling, or coaching interaction(s), demonstrating mixed results on self-management behaviors. Five studies demonstrated improvements in HbA1c [ [53] , [54] , [55] , [56] ], weight loss [ 56 ], and symptoms [ 57 ]. However, four studies found no significant change impact on HbA1c level [ [58] , [59] , [60] , [61] ]. In a descriptive study by Aikens et al. [ 62 ], improvements in self-management behaviors were noted (medication adherence self-monitoring blood glucose, foot care) of varying significance.
More specifically, mobile phone technology and access to these devices is increasing the use of this technology in self-management of diabetes. Two systematic reviews of mobile phone applications designed to improve glycemic control by supporting type 2 diabetes self-management report an overall mean reduction in HbA1c of 0.40% and 0.49% when compared to controls [ 63 , 64 ]. One systematic review with meta-analysis also demonstrated a reduction in HbA1c of 0.51% when comparing smart phone to standard care [ 65 ].
Two studies examined CAMs and the impact on physiologic and psychosocial outcomes. A systematic review of 16 RCTs examined the impact of computer-based diabetes intervention, showing only a small benefit on reduction of HbA1c level, with no other evidence of benefit noted on cardiovascular risk factors, QOL and health status [ 66 ]. A RCT assessed the effectiveness of a computer-assisted diabetes self-management intervention, finding no significant HbA1c improvement and only small improvements in fasting plasma glucose and body weight [ 67 ].
Seven studies examined web-based interventions and the impact on physiologic and psychosocial outcomes, including six RCTs and a cohort study. Of the six RCTs, one showed significant improvement in HbA1c, weight, and waist circumference at four months [ 68 ] three showed improved HbA1c at six months [ [69] , [70] , [71] ], and two showed improvement in knowledge scores, healthcare behaviors, and HbA1c [ 72 , 73 ]. The cohort study showed a significant HbA1c reduction at six months but not at twelve month [ 74 ].
The mixed modalities studies were all CAMs & telephone and included one systematic review, two cohort studies, and two RCTs. The systematic review included six studies that found significant declines in HbA1c and an overall increase in satisfaction, personal health care, knowledge and quality of life [ 75 ]. Cohort studies showed mixed results with significant changes to HbA1c; however one found significant reductions in distress [ 76 , 77 ]. In the RCTs, intervention groups using telehealth with provider feedback showed significant decreases in HbA1c [ 78 ] and blood pressure, but no differences in weight, self-management adherence behaviors, or QOL [ 79 ].
Lifestyle modification program (LMP) is a general description given to an intervention designed to promote health through lifestyle and behavior change. LMPs can include a wide range of topics, including diet, exercise, medications, and stress; can occur in a wide range of settings, including healthcare organizations, workplaces, and the community; and can be delivered through a variety of mediums ranging from face-to-face, to telephonic, to online technologies. LMP's have a long history in diabetes care, and typically combine interventions targeting diet, exercise, medication management, and behavior modification. Individualizing LMPs has been identified as a key to their success.
Seven RCTs were reviewed with various LMP interventions (see Table 5 ). Locations of research included three in the United States, and one in the United Kingdom, Canada, and the Netherlands. Programs ranged from twelve months to two years in length. Of these seven, only one study had a short-term statistically significant impact on HbA1c but this did not persist over the duration of the study [ 80 ]. Of the multiple LMP outcome variables, the most positive impacts were noted in diet (6/7 studies) [ 10 , [80] , [81] , [82] , [83] , [84] ]; physical activity (4/7 studies) [ 10 , 81 , 83 , 84 ]; self-efficacy (2/6 studies) [ 80 , 85 ]; and stress (2/6 studies) [ 82 , 83 ].
Lifestyle modification programs.
Study & Location | Design | Sample | Outcome Measures | Intervention (I) and Control (C) Groups | Results |
---|---|---|---|---|---|
Rosal et al. (2011) [ ] US | RCT | = 252 Female: 76.5% Age: > 18 Latino | Fasting glucose, HbA1c, BP, weight, BMI, waist circumference, medication intensity, physical activity, BGSM, diet, knowledge, & self-efficacy, collected at baseline, 4 and 12 months post-intervention. | (I): ( = 124) A 1-year long program with 12 weekly sessions with follow up phase of 8 monthly sessions. Focus of program: DM knowledge, attitudes, self-management, cultural tailoring; (LTA = 19); Intervention delivered by a nutritionist or health educator and “trained” and lay individuals; Intervention fidelity not addressed. (C): ( = 128) Usual care; (LTA = 16). | Significant difference in HbA1c at 4 months (reduction 0.88), but not at 12 months. Significant changes at 12 months for diabetes knowledge, self-efficacy, BGSM, and diet self-management. |
Clark et al. (2004) [ ] UK | RCT | = 166 Female: 42% Mean age: 59.5 United Kingdom | Self-management activities, diet behaviors, physical activity, weight, BMI, waist circumference, lipids, HbA1c, stages of change, barriers, & self-efficacy, collected at baseline, 12, 24, and 52 weeks. | (I): ( = 50) Tailored LMP with meetings with interventionist at baseline, and weeks 12, 24, and 52, for goal setting and MI techniques for behavior change. Follow up phone calls by interventionist at weeks 1, 3, and 7; (LTA = 2); Intervention delivered by an “interventionist”; Intervention fidelity not addressed. (C): ( = 50) Usual care; (LTA = 4). | Fat intake reduced and physical activity increased in intervention group. No other significant differences between groups. |
Thoolen et al. (2009) [ ] Netherlands | RCT | = 227 Female: 45% Mean age: 62 | BMI, & self-reported intentions, self-efficacy, proactive coping, self-care behaviors, physical activity, diet, & medications, collected at baseline, 3 and 12 months. | (I): ( = 89) Proactive coping intervention lead by RN, two individual and 4 group sessions (each session 2 h), over 12 weeks. Taught a 5-step proactive coping plan, involving goal setting and planning processes; (LTA = 11); Intervention delivered by nurses; Intervention fidelity not addressed. (C): ( = 108) Usual care; (LTA = 4). | Diet and physical activity behavior improved, resulting in significant weight loss at 12 months; proactive coping was a better predictor of long-term self-management than intentions or self-efficacy. |
Toobert et al. (2007) [ ] US | RCT | = 289 Female: 100% Mean age: 61 Post-menopausal women | Self-reported lifestyle behaviors (diet, physical activity, smoking, stress management), social support, problem solving, self-efficacy, depression, QOL, & cost analysis, collected at baseline, 6, 12, and 24 months. | (I): ( = 163) Mediterranean Lifestyle Program (MLP), a 2 and a half days retreat, followed by 4-h weekly meetings for the first 6 months addressing diet, PA, stress management, and support groups. After 6 months, participants randomized to either (a) faded schedule of weekly meeting led by lay leader, or (b) 4 meetings over 18 months led by project staff to complete a personalized computer assisted program; Intervention delivered by a dietician, exercise physiologist, stress-management instructor, and professional and lay support group leaders; Intervention fidelity not addressed. (C): ( = 116) Usual care. LTA = 15% of total randomized sample. | Significant improvements at all time points for diet, stress management, & problem solving ability. Improvements noted in physical activity, social resources, and self-efficacy. |
Toobert et al. (2011) [ ] US | RCT | = 280 Female: 100% Mean age: 55.6–58.7 Latina | Problem solving (coded by interviewers), and self-reported self-efficacy, social support, diet, stress management, & physical activity, collected at baseline, 6 and 12 months. | (I): ( = 142) Usual care plus Viva Bien program - a 12-month lifestyle modification program addressing diet, stress management techniques, exercise, smoking cessation, problem solving. Involves a 2 and a half days retreat followed by weekly 4-h meetings for 6 months, then twice monthly for 6 months. Intervention delivered by “study staff”; Intervention fidelity not addressed. (C): ( = 138) Usual care. LTA: 23.2% intervention group; 21.7% control group. | Significant improvements in behavior change (diet, practice of stress management, exercise, and engagement in social support), and HbA1c; however, these changes were not maintained at 12 months. Improvements in psychosocial outcomes (problem solving, self-efficacy, and perceived support). |
McGowan (2015) [ ] Canada | RCT | = 361 Male: 54–64% Mean age range: 63.8–64.6 | HbA1c, lipids, weight, BMI, BP, waist circumference, self-reported self-efficacy, attitudes, behaviors, health status, & QOL, collected at baseline, 6 and 12 months. | Three groups: (I-1): ( = 130) DSMP program - lead by pairs of trained lay leaders, groups of 10–16 meet once a week for 2.5 h over a 6 wee time period; (LTA = 44) (I-2): ( = 109) CDSMP (same as DSMP, but not specific to diabetes); (LTA = 46). Both intervention groups led by ““trained program leaders”; Intervention fidelity not addressed. (C): ( = 122) Usual care; (LTA = 33). | Significant improvements in 5 of 30 outcome measures: fatigue, cognitive symptom management, self-efficacy, communication with physician, and diabetes empowerment. Marginal differences in HbA1c between both groups. Both programs effective in bringing about positive changes, but little difference between the programs. |
Markle-Reid et al. (2018) [ ] Canada | RCT | = 159 Female: 55.9% Age: 30% aged 65 to 69, 40% aged 70 to 74, and 30% aged 75 and older. | HRQOL, mental health, & self-efficacy, collected at baseline and 6 months after intervention | (I): ( = 80) Participated in a community-based lifestyle modification program focused on self-efficacy, self-management, holistic care, and individual and caregiver engagement. The program, delivered by trained nurses, dietitians, program coordinator, and peer volunteers, involved 3 in-home visits, monthly group sessions, monthly case conferences, and on-going nurse-led care coordination. (LTA = 5). Fidelity addressed. (C): ( = 79) Usual care. (LTA = 13). | Intervention group showed improved quality of life and self-management and reduced depressive symptoms. |
Identified as a critical element in the care for all people with diabetes, diabetes self-management education (DSME) has been a long-standing intervention in the care of persons with diabetes [ 86 ]. DSME has evolved over time to include behavioral and affective strategies [ 87 ], and biopsychosocial treatment models addressing both medical and psychosocial needs of persons with diabetes [ 88 ]. Educational interventions can be administered by peers or professionals, to individual or groups, in short term or extended sessions, and by different modalities. Current thought on optimal diabetes self-management is that DSME needs to be followed by diabetes self-management support (DSMS) [ 89 ]. DSMS involves several essential components that must be maintained long-term to prevent diabetes-related complications: adherence to diet, physical activity, treatments, and monitoring checks [ 90 ].
The National Standards for Diabetes Self-Management Education and Support are reviewed and revised approximately every five years by a Task Force jointly convened by the American Association of Diabetes Educators (AADE) and American Diabetes Association (ADA) [ 86 ]. While there are many models of DSME, the standards do not endorse any one approach, but rather, specifies what constitutes effective self-management strategies [ 86 ]. Many studies have explored the impact of DSME on self-management with outcomes measures covering a range of physiological, behavioral, and psychosocial variables. Research suggests that DSME is associated with changes in diabetes knowledge, clinical outcomes, self-efficacy, and quality of life [ 91 ].
For this review, eleven sources were reviewed: one systematic review, two meta-analysis, and eight RCTs. Locations of research included the United States, Sweden, Australia, Saudi Arabia, Japan, and Norway. Looking specifically at HbA1c as the outcome, a systematic review of 118 DSME interventions found that DSME resulted in a significant decrease in HbA1c [ 91 ]. Two meta-analyses analyzing RCTs specific to persons with type 2 diabetes show that the benefits of DSME are modest [ 92 ] and that the positive effects tend to gradually decline over time [ 93 ]. Eight RCTs conducted in six countries were reviewed with various educational interventions (see Table 6 ). Sample sizes ranged from 75 to 670 participants with the intervention groups ranging from 36 to 335 individuals in the RCTs. Statistically significant improvements in select biophysical, psychosocial, and self-management measures, including knowledge [ 94 , 95 ], distress and quality of life [ 96 , 97 ], and physiologic outcomes [ 98 , 100 ]. One study found no differences in biophysical or self-management behaviors [ 101 ].
Educational interventions.
Study & Location | Design | Sample | Outcome Measures | Intervention (I) and Control (C) Groups | Results |
---|---|---|---|---|---|
Chrvala, Sherr, & Lipman (2016) [ ] | Systematic review | = 118 RCTs >18 years, type 1 and type 2 diabetes. | Must have HbA1c as outcome variable. | Out of 118 RCTs, most reported on a single discrete DSME intervention with follow up HbA1c level at 3 months or greater. Several RCTs compared 2 or 3 methods of DSME to a control condition. | 61.9% of studies reported significant change in HbA1c, with an average reduction of 0.57. Education hours <10 were associated with a greater proportion of interventions with significant reductions in HbA1c. |
Klein et al. (2013) [ ] | Meta-analysis | = 52 RCT Type 2 diabetes, age > 18. | HbA1c values at baseline and post-intervention. | Of the 52 RCTs, 17 had 13 weeks or less for length of intervention, 17 had 14–16 weeks of intervention, and 19 had 27 or more weeks of intervention. | DSME resulted in significant reductions in HbA1c compared to control conditions. However, most participants did not achieve recommended HbA1c level. |
Adolfsson et al. (2007) [ ] Sweden | RCT | = 101 Female: 41% Mean age: 63 Sweden | HbA1c, BMI, and self-reported confidence in diabetes knowledge, self-efficacy, & satisfaction, collected at baseline and 1-year follow up. | (I): ( = 42) A group of 5–8 participants had 4-5 empowerment group education sessions, and a follow up session within 7 months; (LTA = 8); Intervention delivered by “trained” doctors and nurses; Intervention fidelity addressed. (C): ( = 46) Usual care; (LTA = 5). | Higher confidence in diabetic knowledge only statistically significant difference in intervention group. No significant change in HbA1c. |
Campbell et al. (2013) [ ] Australia | RCT | = 670 Female: 46.3% Mean age: 55.7 Australia | Self-reported self-efficacy, and self-management behaviors, collected at baseline, 4 weeks, and 6 months. | (I): ( = 335) Received Fact Sheets and DVD comprising patient narratives of type II diabetes management during a 3-week intervention; (LTA = 49); Intervention delivery personnel and intervention fidelity not addressed. (C): ( = 335) Received diabetic Fact Sheets only; (LTA = 23). | Mean difference in self-efficacy was 7.2 better in intervention group. Change in self-care behaviors during previous 7 days significantly greater in intervention group. |
Beverly et al. (2013) [ ] US | RCT | = 135 Female: 51% Mean age: 59 Caucasian: 75% | HbA1c, weight, BMI, waist circumference, BP, pedometer readings, fitness assessment, blood glucose, and self-reported self-care, symptoms, coping, distress, QOL, confidence, and health literacy, collected at baseline, 3, 6, and 12 months. | (I): ( = 68) Four 1-h group education sessions each with a different topic (diabetes overview, healthy eating, BGL monitoring, natural course of diabetes); (LTA = 10); Intervention delivered by RNs and dieticians; Intervention fidelity not addressed. (C): ( = 67) Two classes 2 h in length focused on BP & cholesterol; (LTA = 4). | Intervention group had modest improvement in HbA1c at 3 months (reduction of 0.4%), with no maintenance of improvement at 6 and 12 months. Control group had no improvement of HbA1c at any time. Both groups improved frequency of self-care, QOL, distress and frustration over time. |
Sugiyama et al. (2015) [ ] US | RCT | = 516 Female: 70% Mean age: 63 Latino: 61%; African American: 39% | HbA1c, and self-reported mental and physical health-related QOL, & social support, collected at baseline and 6 months. | All given 2 h training on SMBG. (I): ( = 258) Six weekly small group self-care sessions based on empowerment model. Sessions were for 2 h, with 8–10 persons per group; (LTA = 55); Intervention delivered by trained “health educators”; Intervention fidelity addressed. (C): ( = 258) Six lectures on geriatric topics unrelated to diabetes; (LTA = 62). | Education increased health-related QOL, and significant reduction in HbAlc (0.4%) compared to control. |
Mohamed et al. (2013) [ ] Saudi Arabia | RCT | = 430 Female: majority Mean age: 53.5 Saudi Arabia | HbA1c, fasting glucose, lipids, BMI, BP, albumin/creatinine ratio, and self-reported knowledge, attitudes, & practice, collected at baseline, 6 and 12 months. | (I): ( = 215) CSSEP (culturally sensitive structured education program), consisting of 4 educational sessions following the ADA standards of care clinical and behavioral goals, 3–4 h each, in groups 10–20 patients; (LTA = 106); Intervention delivered by “educators”; Intervention fidelity not addressed. (C): ( = 215) Usual care; (LTA = 34). | Significant improvements in intervention group in HbA1c reduction (0.55%), fasting blood sugar, BMI, albumin/creatinine ratio, knowledge, attitude & practice. |
Moriyama et al. (2009) [ ] Japan | RCT | = 75 Female: 54% Mean age: 65.8 Japan | Weight, abdominal circumference, BP, fasting blood glucose, HbA1c, lipids, and self-reported QOL, stage of change, goal attainment, & self-check, collected at baseline and 3, 6, 9, and 12 months. | (I): ( = 50) Monthly face-to-face individual sessions, 30 min each, after clinical exam. Required patient setting behavioral goals on exercise and diet and contact every 2 weeks to check if practicing goal setting behaviors over the next 12 months; (LTA = 8); Intervention delivered by “educator”; Intervention fidelity addressed. (C): ( = 25) Usual care; (LTA = 2). | Intervention group had significant improvements in weight, HbA1c reduction (0.55%), self-efficacy, dieting and exercise stages, QOL, diastolic BP, total cholesterol and complication prevention behaviors. |
Sperl-Hillen et al. (2011) [ ] US | RCT | = 623 Female: 49.4% Mean age: 61.8 Caucasian: 65.2%; Hispanic: 22.1% | HbA1c, weight, waist circumference, BP, and self-reported depression, general health, support, attitudes, caring ability, distress, understanding, empowerment, diet, & physical activity, collected at baseline, 1, and 4 months. | 3 groups: (I-1): ( = 243) Group education using Conversation Maps in four 2-h sessions with groups scheduled at 1 week intervals, 8–10 people per group; (LTA = 29). (I-2): ( = 246) Individual education; 3 sessions, 1 h each, one month intervals; (LTA = 37). Intervention delivered by “trained” nurses and dieticians; Intervention fidelity addressed. (C): ( = 134) Usual care (LTA = 13). | HbA1c reduction in all groups (0.27, 0.51, 0.24) but significantly more with individual education, compared to group education or usual care. Individual education improved physical health, but not mental health scores. |
Rygg et al. (2012) [ ] Norway | RCT | = 146 Female: 45% Mean age: 66 White Norwegians: 100% | BP, BMI, HbA1c, lipids, creatinine, and self-reported patient activation, QOL, distress, global health, diabetes knowledge, & self-management skills, collected at 6 and 12 months. | (I): ( = 73) DSME group of 8–10 patients, 15 h of education over 3 sessions, one week between each session; (LTA = 9); Intervention delivered by “trained” nurses; Intervention fidelity not addressed. (C): ( = 73) Usual care; (LTA = 4). | No difference in primary outcomes between groups at 12 months. Diabetes knowledge and some self-management skills improved significantly in the intervention group. |
Ferguson et al. (2015) [ ] | Systematic review and meta-analysis | = 13, with 11 included in meta-analysis. Hispanic or Latino majority. | A1c Baseline and at follow up. Follow up periods ranged from 6 months to 5 years. | Studies included a DSME intervention in combination with primary care. Seven RCTs included culturally tailored DSME; 9 reported the level of involvement of the primary care provider. Five of 13 studies reported statistically significant changes in HbA1c in the intervention group; Six found no significant changes in HbA1c between groups. | The pooled effect across studies was and HbA1c reduction of −0.25 (95% CI, −0.42 to −0.07, P = .01), indicating a greater improvement in glycemic control for the intervention group at 6 months–12 months. |
Mindfulness is a type of meditation practice that has been described as being attentive to the present moment in an open and non-judgmental way [ 102 ]. Described as both a trait that can vary between persons, and a skill that can be learned, the concept of mindfulness has measureable aspects including: non-reacting, observing, acting with awareness, describing, and non-judging [ 103 ]. Mindfulness as an intervention engages and strengthens an individual's internal resources for optimization of health through self-awareness and taking responsibility for one's life choices [ 104 ]. Mindfulness interventions emphasize different practices, depending on the philosophy of mediation practice used, and can incorporate components of stress reduction therapy, cognitive behavior therapy, and spiritual components. However, while mindfulness interventions take on a variety of forms, most follow a systematic procedure for developing self-awareness, and have clear learning objectives based on theory and science [ 105 ].
Mindfulness interventions have be used in chronic disease care to address symptom management and the emotional distress caused by disease and its management. Research suggests that mindfulness has a negative association with both anxiety and depression symptoms in a sample of 666 persons with type 1 and 2 diabetes [ 106 ], and was negatively correlated with depression and positively correlated with health-related quality of life in a sample of 75 adults with type 2 diabetes [ 107 ]. A mindfulness-based cognitive therapy intervention has been shown to reduce emotional distress and increase quality of life in persons with type 1 and 2 diabetes [ 108 ]. In a systematic review of 45 studies using meditation interventions for chronic disease, Chan and Larson [ 105 ] conclude that meditation improved symptoms of anxiety, depression, and chronic disease; but conclude that the lack of consistency across diseases and types of meditation interventions warrants further research.
For this integrative review, mindfulness was studied as an intervention in three studies. Locations of research included the United States and Germany. The frequency and length of mindfulness interventions included a one-day workshop, to two 90-min session two months apart, and weekly meetings for 8 weeks. In a RCT of 81 persons from the community with type 2 diabetes, providing education and teaching mindfulness and acceptance of diabetes, compared to providing education alone resulted in improvements in HbA1c at three months post-intervention [ 109 ]. However, two other RCTs did not find improvements in physical measures of diabetes. An 8-week mindfulness-based intervention compared to a control group demonstrated lower levels of self-reported depression and improved health status at a one-year follow up, but no differences in albuminuria [ 110 ]. In a cohort study, a mindfulness-based eating intervention was compared to an educational intervention over a six-month period, resulting in no significant differences between groups for change in weight or diet intake [ 111 ]. See Table 7 .
Mindfulness.
Study & Location | Design | Sample | Outcome Measures | Intervention (I) and Control (C) Groups | Results |
---|---|---|---|---|---|
Gregg et al. (2007) [ ] US | RCT | = 81 Female: 46.9% Mean age: 50.9 Hispanic: 28.4% | A1c, and self-reported self-management (diet, exercise, and blood glucose monitoring), knowledge, treatment satisfaction, & acceptance, collected at baseline and 3 months. | (I): ( = 43) The ACT condition, involving a one-day workshop with education, acceptance, and mindfulness training; (LTA = 10); Intervention delivered by “author of manual”; Intervention fidelity not addressed. (C): ( = 38) Education alone; (LTA = 3). | ACT condition more likely to use the coping strategies, to report better diabetes self-care, and to have HbA1c values in the target range. ACT had no significant effect on HbA1c. |
Hartmann et al. (2012) [ ] Germany | RCT | = 110 Males: 78.1% Mean age: 59 European | Albuminuria, and self-reported psychiatric comorbidity, levels of Depression, & stress, collected at baseline & 12 months. | (I): ( = 53) Mindfulness based stress reduction (MBSR) intervention, groups of 6–10 participants, meeting weekly for 8 weeks, with a booster session after 6 months; (LTA = 10); Intervention delivered by psychologist and resident physician; Intervention fidelity not addressed. (C): ( = 57) Usual care; (LTA = 6). | The MBSR group showed significant reduction in psychosocial distress, but not on albuminuria. o significant reduction in HbA1c. |
Miller et al. (2014) [ ] US | Cohort | = 68 Female: 63.5% Mean age: 54 Caucasian: 76.5% | Weight, & self-reported diet, knowledge, outcome, expectancy, self-efficacy, anxiety, depression, & mindfulness, collected at baseline, post-intervention, then again 1 month and 3 months after the second data collection. | Group 1: ( = 32) Mindful based eating awareness training (MB-EAT), 2 CDs to guide mindfulness meditation, encouraged to meditate 6 days/week and to practice mini-meditations at other times, basic information on self-management; (LTA = 5). Group 2: ( = 36) Smart Choice (SC) intervention which is behavioral DSME and in-depth nutrient information. All groups had 90 min 1 and 3 month follow up session reviewing key principals of interventions; (LTA = 11);. For both groups, intervention delivered by dietician and social worker; Intervention fidelity addressed. | Both groups with significant improvements in depressive symptoms, expectations, self-efficacy, and cognitive control regarding eating behaviors. |
Cognitive behavioral therapy (CBT) is a form of psychotherapy focused on problem-solving through improving negative thinking and behavior [ 112 ]. In CBT, the therapist focuses on the impact that dysfunctional thoughts have on current behavior and future functioning. CBT is aimed at evaluating, challenging, and modifying a patient's dysfunctional beliefs (cognitive restructuring) [ 113 ]. CBT is used as an intervention for multiple disorders including but not limited to, anxiety, depression, panic disorder, phobias, obsessive compulsive disorder, post-traumatic stress, schizophrenia, anger, eating disorders, somatic disorders, and chronic pain syndromes [ 114 ].
In relation to the study of diabetes, CBT has been used as an intervention to treat depression due to its association with glycemic control and self-management. The incidence of major depression as a comorbid condition in both type 1 and type 2 diabetes is well documented, estimated to affect 15–20% of persons with diabetes [ 115 ]. Furthermore, depression, even at low levels, has been associated with suboptimal adherence, worse diabetes control, and risk of complications [ 116 ]. In an early study done by Lustman et al. [ 117 ], the use of CBT with supportive diabetes education demonstrated effectiveness in treatment for major depression and potential improvement in glycemic control in persons with type 2 diabetes. Following this work, other studies have explored the use of CBT for treatment of depression and the impact on glycemic control. CBT has been explored in studies of both type 1 and type 2 diabetes mellitus, demonstrating positive effects of CBT on depressive symptoms, but with mixed findings on the impact on glycemic control. In a study of 94 outpatients with diabetes and depressive symptoms, improvements in depressive symptoms and HbA1c, and in self-reported depressive symptoms, anxiety, well-being, and diabetes-related distress were found [ 118 ]. Additional studies of both type 1 and type 2 diabetes using CBT interventions show improvements in depression, but are inconclusive regarding the impact on improving self-management and physical health outcomes [ 119 , 120 ].
For this integrative review, five studies examined the use of CBT for depression and the relationship to type 2 diabetes self-management: Three RCTs, one systematic review, and one meta-analyses (see Table 8 ). Two of the RCTs were done in the United States, and one in Germany. One study included five weekly 90-min CBT sessions, and two studies included eight to twelve one-hour weekly CBT sessions. All RCTs compared the intervention group to usual care that included diabetes self-management education. The RCTs show improvements in depression and distress [ 121 , 123 ], but only one study showed improvements in glycemic monitoring and control [ 123 ]. In a systematic review and meta-analysis of RCTs of psychological interventions to improve glycemic control in persons with type 2 diabetes, 23 out of 25 RCTs examined CBT as the intervention. Results suggest that there are improvements in long-term glycemic control and psychological distress, but not in weight control and blood glucose level [ 124 ]. In a meta-analyses of 45 RCTs assessing efficacy of psychological interventions for self-management of type 2 diabetes in adults from mainland China, 33 studies focused on CBT as the intervention. Analysis suggest that CBT was more effective than the control condition in reducing HbA1c, depression, and anxiety [ 125 ].
Cognitive behavioral therapy.
Study & Location | Design | Sample | Outcome Measures | Intervention (I) and Control (C) Groups | Results |
---|---|---|---|---|---|
Hermanns et al.(2015) [ ] Germany | RCT | = 214 Female: 56.5% Mean age: 43.3 German-speaking | Self-reported depressive symptoms, distress, self-care activities, well-being, QOL, diabetes acceptance, & treatment satisfaction, collected at baseline, immediately after intervention, then 6 and 12 months. | (I): ( = 106) DIAMOS program, delivered by psychologist using CBT, comprised of five 90-min lessons; (LTA = 13); Intervention delivered by “certified” psychologist; Intervention fidelity not addressed. (C): ( = 108) Usual care, consisting of a group-based diabetes education program; (LTA = 20). | 12-month follow up showed significant reduction in depressive symptoms, and diabetes related distress in the intervention group. |
Penckofer et al. (2012) [ ] US | RCT | = 74 Females Mean age 54.8 White 63%; Black 29%; Hispanic: 8% | Fasting glucose, HbA1c, & self-reported depression, anxiety, anger, health related QOL, & knowledge, collected at baseline, 3 and 6 months. | (I): ( = 38) One hour CBT intervention done in group sessions, delivered by a nurse, weekly for 8 weeks; (LTA = 12); Interventions delivered by nurses; Intervention fidelity addressed. (C): ( = 36) Usual care; (LTA = 2). | CBT significantly reduced depression, anxiety, and anger symptoms compared to usual care, but there were no significant differences between groups on HbA1c levels. |
Safren et al. (2014) [ ] US | RCT | = 87 Female: 50% Mean age: 55-58 Majority Caucasian | A1c, medication adherence, SMBG, distress & self-reported depression, collected at baseline, 4, 8, and 12 months. | (I): ( = 45) CBT for adherence & depression + ETAU (enhanced treatment as usual). Received 9–12 CBT sessions over 4 months; (LTA = 5); Intervention delivered by a “therapist”; Intervention fidelity not addressed. (C): ( = 42) ETAU with series of diabetes support and adherence interventions (included one meeting with nurse educator, two with dietician, one with adherence counselor); (LTA = 4). | Intervention group at 4 months: statistically significant improvement in medication adherence, SMBG, reduction in HbA1c (0.63%), & improvement in depression score. At 8 & 12 months medication adherence, HbA1c and SMBG adherence maintained in CBT group. |
Ismail et al. (2004) [ ] | Systematic review | = 25 RCTs Psychological interventions for Type 2 diabetes control | A1c, blood glucose, weight, BMI, & psychological distress. | 23 studies of adults (out of 25 studies) used CBT as an intervention in relation to diabetes control in type 2 diabetes. | In persons receiving psychological therapies, there are improvements in long term glycemic control (mean HbA1c reduction of 0.32%), and psychological distress, but not in weight or blood glucose level. |
Chapman et al. (2015) [ ] | Meta-analysis | = 45 RCTs(US and China) Psychological interventions for type 2 diabetes control | HbA1c, blood glucose, anxiety, depression, & QOL. | 33 studies of adults (out of 45 studies) used CBT as an intervention in relation to diabetes control in type 2 diabetes. | CBT was more effective than the control condition in reducing HbA1c (SMD = −0.97), depression, and anxiety. |
4.1. impact of interventions.
This integrative review examined 70 studies (8 systematic reviews, 3 meta-analyses, 53 RCTs, 4 cohort, and 2 descriptive), summarizing eight categories of interventions targeting physiologic, behavioral, and psychological outcomes in patients with type 2 diabetes. Studies were examined from seventeen countries including a broad range of cultures and ethnicities within the research, including Caucasian American, African American, Native American, Hispanic/Latino, European, Canadian, Australian, Middle-Eastern, and Asian populations.
While interventions have shown mixed results in all interventions categories, many studies do support small to modest improvements in physiologic, behavioral, and psychological outcome measures. Interventions have shown small to modest improvements for HbA1c. Often the significant HbA1c change was only within the intervention group, but not significant when compared between groups. Levels of improvement ranged from 0.13% to 1.6% reductions, with the highest reductions seen in peer support/coaching and technology-based interventions. Small to modest improvements were also seen in physiologic outcomes of weight loss, behavioral outcomes of self-reported diet and physical activity, and psychological outcomes of self-reported improvement in self-efficacy and reduction in distress.
In addition to a wide variety of interventions being tested for self-management of type 2 diabetes, considerable heterogeneity of interventions exist within similar types of interventions. Areas of heterogeneity included length, duration, and number of sessions, content, method of delivery (i.e., in-person and technology-based, individual or group-based), and facilitation (i.e., self-directed, health care professional, peer). For example, motivational interviewing interventions ranged in length from one 60-min session to five 45-min sessions over one year, could be either individual or group based sessions, including face-to-face and self-directed internet based sessions. Considerable variation was found in all intervention categories in this review. This heterogeneity makes it difficult to aggregate findings on specific interventions.
A wide range of professionals and non-professionals were used for intervention delivery. Out of 59 studies, 18 (30%) had nurses facilitating the interventions, with most being education or technology interventions. Twenty-three studies used non-specified personnel to deliver the intervention, including health educators, trained personnel, and peer and/or lay persons. Most of these studies included a peer or coaching intervention, or a lifestyle modification program. Ten studies indicated that a research team delivered the intervention, mostly of which were technology-based interventions. Other types of professionals delivering interventions included certified diabetes educators, psychologist/counselors, pharmacists, dieticians and nutritionists, exercise physiologists and trainers, and social workers, medical assistants, physicians, and students. While only 30% of the studies had nurses as interventionists, they are well positioned to contribute to all intervention types. As it was noted with the exception of mindfulness, nurses were the only professionals used as interventionists across all types of interventions in this review. However, it is also to be noted that components of mindfulness have been embedded within some larger multi-modality and education interventions that have been led by nurses.
In addition to heterogeneity, many intervention approaches are multi-modal, and include components of different categories of intervention in one intervention program. For example, a life style management program may include components of education, motivational interviewing, and technology. Technology interventions, while focusing on the use of the specific technology, may include education, problem solving, and peer coaching. And while the use of multi-modal approach may be beneficial to helping to improve self-management, this overlap makes it challenging to separate out impact of specific interventions. And lastly, fidelity of interventions is another area of consideration. Out of 59 studies, only 21 (35.5%) addressed procedures for intervention fidelity.
The studies in this review examined the impact of a self-management intervention on the major outcomes of physiologic measures of disease control, self-management behaviors, and psychological outcomes. The most commonly reported physiologic measure of disease control was HbA1c level. Other commonly used physiologic measures included weight, BMI, waist circumference, and blood lipid levels. The most commonly reported behavioral outcomes were for diet and physical activity. Other behavioral outcomes included SMBG and medication adherence. In addition, behavioral outcomes were mainly self-report. The most commonly reported psychological outcomes were self-efficacy and distress, and as in behavioral outcomes, these outcomes were also mainly self-reported.
Outcome measures were collected mostly at 6 months (19 studies) and 12 months (22 studies) follow up. Twelve studies collected outcome data at three to four months, two at 18 months, and two studies at 24 months. Overall, duration of most research was limited to one year.
In terms of attrition rates, the majority of studies (64.4%) had less than 20% attrition at final data collection time. Approximately 25% had 1.2–10% attrition, and 39% had 10.0–20% attrition. Three studies had attrition rates between 32.6 and 37.7%, with study duration lasting between 12 and 15 months. The majority of studies report attrition as a number or percentage, with limited information about participant characteristics and attrition. Five studies did not report attrition.
5.1. impact of interventions.
The results of the integrative review support prior reports from the literature on diabetes self-management. A vast amount of literature exists describing intervention research for diabetes self-management. Interventions in general have demonstrated short-term improvements in glycemic control [ 126 , 127 ], and in promoting knowledge, self-efficacy, and in distress reduction [ 46 ]. However, results of intervention effectiveness are inconsistent [ 45 ], with many studies producing mixed results in relation to physiological, behavioral, and psychosocial outcomes.
The levels of improvement of HbA1c in this integrative review ranged from 0.13% to 1.6%. To elaborate on those findings, most studies that showed improvements in HbA1c had reductions of approximately 0.50%. Of the four studies that had showed HbA1c reductions of greater that 1.00%, three of them collected outcome data at 6 months, and two had sample sizes less than 65 subject. These finding bring consideration to the question of statistical versus clinical significance. It has been suggested that 0.5% HbA1c is a clinically significant change [ 128 ]. This reference to this reduction in HbA1c is drawn from the earlier work of the Diabetes Control and Complications Trial Research Group [ 129 ], and the UK Prospective Diabetes Study [ 7 ]. A difference in HbA1c of only approximately 2% between intensive and standard treatment groups demonstrated significant differences in outcome risks [ 129 ], and even lower differences in HbA1c (7.0% intensive vs 7.9% conventional treatment) demonstrated significant reduction of microvascular complications in persons with type 2 diabetes [ 7 ].
The results of this integrative review demonstrated that in addition to a wide variety of interventions being tested for self-management of type 2 diabetes, there is considerable heterogeneity of interventions that exists within similar types of interventions. This result is also reported in systematic reviews on interventions for self-management of type 2 diabetes [ 23 , 124 , 125 ] describing considerable variability in studies with respect to methods of intervention delivery, duration, and intensity, and in measurement of outcome variables and follow-up interval [ 91 ]. In addition, many intervention approaches are multi-modal and include components of different categories of intervention in one intervention program. This overlap makes it challenging to separate out impact of specific interventions, and makes it challenging to aggregate findings and draw solid conclusions on the impact on outcomes of physiologic, behavioral, and psychological outcomes [ 35 , 91 ].
Fidelity of interventions is another area of consideration. In this integrative review, out of 59 studies, only 21 (35.5%) addressed procedures for intervention fidelity. Intervention fidelity has been identified as a limitation in diabetes self-management research, with issues concerning inconsistency in intervention delivery, quality in training to assure fidelity, and lack of fidelity assessment [21, 44, [ 125 ]. A systematic review specific to intervention fidelity in diabetes self-management interventions reported that intervention fidelity of interventions remains under-investigated [ 130 ], with most fidelity assessment done through direct observation, and with intervention dose being assessed by self-reported measures [ 130 ].
The most commonly reported physiologic measure of disease control was HbA1c level. This is consistent with the diabetes literature on treatment and research [ 7 , 131 ], with HbA1c being considered the gold standard for glycemic control. HbA1c reflects average glycemia over approximately 3 months and has strong predictive value for diabetes complications [ 132 ], and provides the most objective and reliable information about glucose control of patients with type 2 diabetes. Most studies in this review reported HbA1c value changes between groups from points in time, as opposed to identifying target HbA1c reduction value. While a specific number or percentage considered to be the target value for reduction has not been identified or consistently used in reference for HbA1c reduction, the common approach has been consistency in lowering HbA1c. Consistent with the literature, studies in this review referenced the American Diabetes Association [ 132 ] goal for HbA1c for most adults to be 7%, and presented HbA1c results in terms of reductions towards that goal.
The most commonly reported self-management behavioral outcomes were for diet and physical activity. Diet and physical activity are two of the four major cornerstones of care for self-management of diabetes [ 133 ]. Poor diet and physical inactivity are major contributors to disabilities that result from diabetes. The importance of proper nutrition and physical activity in reducing rates of disease and death from chronic diseases has been well-established [ 8 , 134 ]. The balancing of diet and physical activity are well-established keys to managing diabetes [ 132 ], and in many cases, the most challenging of the self-management behaviors to manage due to being complicated and difficult to integrate into daily life [ 135 ]. In addition, they can be challenging to measure, with most measures in research studies being self-report. Self-report measures may present certain limitations in capturing aspects of dietary and physical activity behavior, with over-reporting being a known problem [ 68 , 136 ].
The most commonly reported psychosocial outcomes were self-efficacy and distress. Self-efficacy and distress have received considerable attention in the chronic disease and diabetes literature. Self-efficacy has been defined as the judgment of capabilities to organize and execute courses of action required to attain desired types of performance and expected outcomes [ 137 ]. Diabetes distress has been described as unique emotional issues directly related to the burdens and worries of living with a chronic disease [ 11 ]. Both self-efficacy and distress have been associated with diabetes self-management and HbA1c levels [ 138 , 139 ]. In general, a broad range of interventions have favorable impact on both self-efficacy and distress, however, sustaining impact on glycemic control and self-management behaviors remains a challenge. Successful treatment and management of emotional needs of patients is needed so that people can be successful with diabetes self-management [ 122 ]. And as in the measurement of diet and physical activity, measures of self-efficacy and distress are self-reported, thus the risk of over-reporting on these variables exists.
Outcome measures were collected mostly at 6 months (19 studies) and 12 months (22 studies) follow up. For studying the impact of interventions on physiologic, behavioral, and psychological outcomes, this timeline presents limitations. Research suggests that results of interventions begin to diminish over twelve months [ 46 ], and that longer follow up periods extending beyond twelve months are needed [ 75 ]. However, the challenges of longitudinal studies are well documented. Challenges such as incomplete and interrupted follow-up with study participants, attrition with loss to follow-up over time, and the generally increased time and financial demands associated with longitudinal research are implicit in study designs [ 140 ].
Because this was an integrative review we chose to include systematic reviews, meta-analyses, RCTs, and descriptive work. This was done in order to not miss nuances found within individual studies that can sometimes occur with larger review studies. However, because some of the RCTs may have also been in larger review studies, there may be some duplication of findings and enhanced or diminished intervention impact. Because of the exhaustive nature of the literature on this topic, it is challenging to stay informed of the entirety of the body of work in this area. Thus, not every piece of evidence, nor every aspect of intervention success/failure maybe completely accounted. And lastly, because of the multi-modal aspects of interventions, it was difficult to initially categorize the broad array of interventions. In each selected category, there may be other interventions. Thus the true impact of a singular category (i.e.: coaching, technology-based, etc.) is difficult to separate out and report outcomes.
Based on the synthesis of findings from this review, the following recommendations for future research are offered. To address the concerns of multi-modal interventions, research that includes a theoretical basis/model of investigation would be beneficial to explicitly describe and provide rationale for the foundation of the intervention. Complex interventions need to be developed based on theoretical frameworks, which is important because a simple explanation or model applied to a complex intervention risks overstating the causal contribution of the intervention [ 141 ]. All elements of a complex intervention need to be identified and described, giving the intervention its theoretical and pragmatic basis that is thought to account for the effectiveness of the intervention [ 142 , 143 ]. Using a theoretical framework provides a guide to appropriately implement, and analyze the intervention [ 144 ].
Research studies need to include full protocols/descriptions of the interventions to provide researchers with the details for comparison and reproduction of the intervention. Many intervention descriptions are too brief or ambiguous, making it difficult to identify specific actions taken, and in turn, making replication challenging. Often words such as self-management, education program, or healthy lifestyle are used with little clarity into what exactly constitutes the intervention. Mixed modality approaches make it difficult to sort out contributions of components of intervention, or to examine the association of components with each other and the impact on outcomes. For example, tech-based interventions may be enhanced by adding coaching components.
Long-term studies and analysis are needed to assist in evaluating the ways in which study variables impact self-management behavior. Longer follow up may provide participants more opportunity to implement strategies targeting behavior change. Prolonged follow up is needed to monitor maintenance of skills gained, many of which may improve over time (i.e. problem solving skills, CBT). In addition, research incorporating more objective measures of self-management is needed. Much of the self-management behaviors are self-report. More objective measures, in addition to HbA1c, for self-management are needed. Objective measures of physical activity and diet are needed.
While efforts have been made to expand the diversity of research participants, many groups continue to be under-represented in diabetes research. More strategies for recruiting representative numbers of ethnic minorities and underserved populations, and research seeking to determine whether interventions are equally effective in these groups is needed. There is a need for new strategies to control the growing diabetes epidemic in the underserved and marginalized population, to better understand diabetes self-management patterns and correlates, and to identify and overcome barriers to self-care in an effort to identify effective culturally tailored self-management interventions [ 33 ].
And lastly, care delivery models that incorporate what is known about effective interventions in the management of diabetes is an area of nursing research wide open for investigation. Specifically, the role of the registered nurse in the management of diabetes care. An interesting point to consider about the issues with intervention heterogeneity, fidelity, and duration focuses on the role of the nurse in the primary care setting. All of the interventions included in this review fall within the scope of practice of general practice nurses in the primary care setting. The RN can be uniquely positioned as part of an inter-professional team to take on expanded primary care functions in managing the complex care of patients with diabetes, leading complex care management teams, and comprehensive care coordination between the primary care home and providers of care services [ 145 ].
Diabetes is a global health problem, as evidenced by the findings of this integrative review. The vast amount of research exploring the impact of interventions for self-management has made major contributions to the care of persons with type 2 diabetes, from offering suggestions for improving care, to stimulating new questions for research. However, implications for clinical practice remain inconclusive [ 126 ], and there remain limitations in the existing body of research, suggesting caution in interpreting results of studies. Moving research forward with attention to intervention development, study design features, and exploring innovative care delivery models offers potential to move this body of research forward to achieving impactful and sustainable physiologic, behavioral, and psychosocial outcomes, and improve the health of those with type 2 diabetes.
This research did not receive any specific grant from funding agencies in the public, commercial, or not-for-profit sectors.
Peer review under responsibility of Chinese Nursing Association.
Appendix A Supplementary data to this article can be found online at https://doi.org/10.1016/j.ijnss.2018.12.002 .
The following is the Supplementary data to this article:
Introduction Gestational diabetes mellitus (GDM) is one of the most common medical complications of pregnancy. Glycaemic control decreases the risk of adverse pregnancy outcomes for the affected pregnant individual and the infant exposed in utero. One in four individuals with GDM will require pharmacotherapy to achieve glycaemic control. Injectable insulin has been the mainstay of pharmacotherapy. Oral metformin is an alternative option increasingly used in clinical practice. Both insulin and metformin reduce the risk of adverse pregnancy outcomes, but comparative effectiveness data from a well-characterised, adequately powered study of a diverse US population remain lacking. Because metformin crosses the placenta, long-term safety data, in particular, the risk of childhood obesity, from exposed children are also needed. In addition, the patient-reported experiences of individuals with GDM requiring pharmacotherapy remain to be characterised, including barriers to and facilitators of metformin versus insulin use.
Methods and analysis In a two-arm open-label, pragmatic comparative effectiveness randomised controlled trial, we will determine if metformin is not inferior to insulin in reducing adverse pregnancy outcomes, is comparably safe for exposed individuals and children, and if patient-reported factors, including facilitators of and barriers to use, differ between metformin and insulin. We plan to recruit 1572 pregnant individuals with GDM who need pharmacotherapy at 20 US sites using consistent diagnostic and treatment criteria for oral metformin versus injectable insulin and follow them and their children through delivery to 2 years post partum. More information is available at www.decidestudy.org .
Ethics and dissemination The Institutional Review Board at The Ohio State University approved this study (IRB: 2024H0193; date: 7 December 2024). We plan to submit manuscripts describing the results of each study aim, including the pregnancy outcomes, the 2-year follow-up outcomes, and mixed-methods assessment of patient experiences for publication in peer-reviewed journals and presentations at international scientific meetings.
Trial registration number NCT06445946 .
This is an open access article distributed in accordance with the Creative Commons Attribution Non Commercial (CC BY-NC 4.0) license, which permits others to distribute, remix, adapt, build upon this work non-commercially, and license their derivative works on different terms, provided the original work is properly cited, appropriate credit is given, any changes made indicated, and the use is non-commercial. See: http://creativecommons.org/licenses/by-nc/4.0/ .
https://doi.org/10.1136/bmjopen-2024-091176
Request permissions.
If you wish to reuse any or all of this article please use the link below which will take you to the Copyright Clearance Center’s RightsLink service. You will be able to get a quick price and instant permission to reuse the content in many different ways.
DECIDE ( www.decidestudy.org ) is a patient-centred and pragmatic comparative-effectiveness randomised control trial that will compare oral metformin versus injectable insulin for the prevention of adverse pregnancy outcomes and the safety of postpartum outcomes among pregnant individuals with gestational diabetes mellitus who require pharmacotherapy and for their exposed children.
Strengths of the DECIDE trial include a non-inferiority clinical trial design, assessment of postpartum outcomes to confirm safety, integration of patient-reported outcomes and inclusion of a racially, ethnically and geographically diverse population.
Limitations of the DECIDE trial include no follow-up beyond 2 years post partum and assessment of participant and infant anthropometry and adiposity by physical exam instead of imaging.
Challenges of this trial will include recruitment across 20 US sites and postpartum retention.
Gestational diabetes mellitus (GDM) is one of the most common medical complications of pregnancy and affects nearly 400 000 or ~1 in 10 pregnant individuals in the USA every year. 1–3 The incidence of GDM has more than doubled in the past decade in an environment of rising prevalence of advanced reproductive age and obesity. 4 5 Moreover, GDM has risen inequitably among racially and ethnically minoritised and lower-income individuals. 2 6 More than one in four infants born to individuals with GDM will experience an adverse neonatal outcome, such as large-for-gestational age (LGA) birth weight, hypoglycaemia or hyperbilirubinaemia. 7–9 After delivery, individuals with prior GDM are at >10 fold increased risk of diabetes, and infants exposed to GDM are at 2-fold increased risk of obesity. 10 11
The goal of GDM treatment is to achieve optimal glycaemic control and prevent adverse pregnancy outcomes. 9 12 The initial therapeutic approach is dietary modification and regular exercise, 12 13 but >1 in 4 individuals will not achieve glucose control with these interventions. 13–15 When pharmacological treatment is needed, guidelines from the American Diabetes Association (ADA) and American College of Obstetricians and Gynecologists (ACOG) recommend insulin as the first-line medication, 14 16 while the Society for Maternal-Fetal Medicine states both insulin and metformin are reasonable medication options. 17
In the past, insulin has been the first-line option because it provides glycaemic control, improves pregnancy outcomes and does not cross the placenta. 7 18 19 An alternative to insulin is metformin, which also provides glycaemic control and improves pregnancy outcomes ( table 1 ). 18–20
Advantages versus disadvantages of metformin versus insulin
Patients and providers may prefer metformin to insulin because it is convenient to take as an oral pill, well-tolerated, cheaper and practical when medication is needed for a brief time period. Additionally, metformin does not cause hypoglycaemia and reduces gestational weight gain. Metformin use is increasing in clinical practice—while insulin remains the most common medication for GDM, one in three individuals with GDM in the USA were prescribed metformin by late 2018. 21 22 Yet metformin has limitations, including that more than one in four individuals will ultimately need supplemental insulin to achieve glucose control, and there is known placental transfer. Historically, another oral agent, glyburide, had been used, but guidelines have since advised against its use following trials that showed it did not appear to be efficacious. 18 23
Follow-up data on metformin from individuals with prior GDM and their exposed children are limited. 24 Extant data suggest that children exposed to metformin had similar body fat composition but slightly higher body mass index (BMI) compared with those exposed to insulin 25 26 ; but recent population-based data show no difference in BMI. 27 However, these studies were limited due to inadequate randomised controlled trial (RCT) follow-up and observed heterogeneity in the effect across different study sites. Also, whether participant metabolic health postpartum varies based on prior metformin versus insulin exposure in pregnancy requires further study. 28
Deciding between metformin and insulin can be challenging for patients and providers given variation in treatment guidelines, provider recommendations and lack of conclusive comparative efficacy and safety data. 29 Understanding whether patients take medications as directed, how satisfied they are with their medication decision, and how their medication decision impacts their pregnancy experience may help to explain observed heterogeneity of treatment effects (HTE). 30 Patient perspectives on barriers to and facilitators of metformin versus insulin use may identify opportunities to improve outcomes. 31
DECIDE: A Comparative Effectiveness Trial of Oral Metformin vs Injectable Insulin for the Treatment of Gestational Diabetes is a randomised, patient-centred, open-label and pragmatic comparative effectiveness trial in pregnancy with postpartum follow-up. This protocol is written in accordance with the Standard Protocol Items: Recommendations for Interventional Trials 2013 statement. 32
Primary aims.
Aim 1: To evaluate whether pregnant individuals randomised to metformin are not inferior to pregnant individuals randomised to insulin for the composite adverse neonatal outcome (LGA birth weight, hypoglycaemia, hyperbilirubinemia or death).
Aim 2: To evaluate whether mean BMI at 2 years of age is higher in the offspring of pregnant individuals randomised to metformin.
Aim 3: To understand facilitators and barriers associated with metformin versus insulin use and HTE to facilitate evidence-based pharmacotherapy.
Aim 1: We hypothesise that metformin is not inferior or worse than insulin by an absolute margin or difference of more than 8% in the composite adverse neonatal outcome.
Aim 2: We hypothesise that metformin does not result in increased child BMI at 2 years (not inferior by an absolute margin of 0.31 kg/m 2 ) compared with insulin.
Aim 3: We hypothesise that patient-reported factors associated with metformin compared with insulin use will be different, which is important to identify to enable clinical implementation of study findings.
We will compare outcomes at delivery between pregnant individuals randomised to metformin versus insulin (hypertensive disorder of pregnancy, gestational weight gain, mode of delivery and obstetric anal sphincter injuries) and their infants (preterm birth, mechanical ventilation, neonatal intensive care unit (NICU) admission, oxygen support, respiratory distress syndrome and small-for-gestational-age at birth); as well as the frequency of treatment supplementation with insulin among pregnant individuals randomised to metformin.
We will compare outcomes at 2 years post partum between individuals randomised to metformin versus insulin (obesity, anthropometry, adiposity, diabetes, cholesterol and hypertension) and their children (obesity, anthropometry and adiposity).
We will compare patient-reported outcomes (PROs) at randomisation (mental and physical health; Diabetes Knowledge Questionnaire (DKQ), Diabetes Distress Scale (DDS) and Diabetes Management Self-Efficacy Scale (DMSES); lifestyle; health behaviours and diet), and at 6 weeks and 2 years post partum for the individual (pregnancy and childbirth experiences; treatment adherence and satisfaction; Maternal-Infant Bonding Scale (MIBS); lactation practices; lifestyle; health behaviours and diet) and child (lifestyle; health behaviours and diet).
DECIDE is a randomised, controlled, open-label, patient-centred and pragmatic multicentre comparative effectiveness trial that is designed to determine whether metformin is not inferior to insulin in reducing adverse pregnancy outcomes and is comparably safe for exposed pregnant individuals and children and to identify patient-reported factors associated with metformin versus insulin that facilitate and enable implementation of study findings ( online supplemental file 1 ).
The DECIDE Study Consortium includes 20 clinical sites under a clinical coordinating centre (CCC) and an independent data coordinating centre (DCC). The consortium is governed by a steering committee and guided by a patient advisory board and stakeholder engagement group. Data management, coordination and analysis will be completed by the DCC, led by the study statisticians (CM and RGC). Participant data will be collected, stored and maintained in OpenClinica, a browser-independent electronic data capture system. Enrolled individuals will be randomised in a 1:1 ratio of metformin to insulin within the web-based data management system according to a computer-generated permuted block design with variable block sizes. Randomisation will be stratified by study site.
Individuals will be recruited across 20 US clinical sites with diabetes and prenatal care programmes ( figure 1 ). These sites have been selected with the goal of achieving racial and ethnic, urban and rural, and geographical diversity at both academic and community-based medical centres. Individuals who continue to receive routine prenatal care in their local community, and then receive high-risk prenatal and diabetes care from the clinical site will also be eligible for study participation. After delivery, individuals and their infants will be followed up with data ascertainment at 6 weeks and 2 years post partum.
Geographic distribution of DECIDE sites across the USA. CCC, clinical coordinating centre; DCC, data coordinating centre.
Inclusion criteria are age >18 years, singleton pregnancy, gestational age between 20 0/7 and 31 6/7 weeks, GDM diagnosis between 20 0/7 and 31 6/7 weeks, requiring medication for glycaemic control, and willingness and ability to attend 2-year follow-up visit ( table 2 ). The decision to initiate medication will be consistent with current US recommendations, defined as ≥30% elevation of either fasting or 1-hour or 2-hour postprandial glucose values in the prior week.
Inclusion and exclusion criteria
We will exclude individuals who have known underlying chronic kidney disease; a fetus with a chromosomal, genetic or major structural malformation; contraindication to metformin or insulin; pregestational diabetes (either type 1 or 2); early-onset GDM <20 weeks; prior haemoglobin A1c >6.5%; concurrent enrollment in a trial with a primary aim that influences the primary study outcome; planned delivery at an outside clinical site where access to medical records cannot be obtained for outcome data abstraction; language barrier (appropriate translation resources unavailable at the site); participation in this trial in a previous pregnancy and fasting hyperglycaemia defined as >115 mg/dL for ≥50% glucose values in the past week ( table 2 ). We include fasting hyperglycaemia as an exclusion criterion as prior data suggest that individuals with this finding are likely to require insulin to achieve glycaemic control. 20
The start date for recruitment is 1 August 2024 and is anticipated to end by 1 May 2026, with final data collection at the 2-year follow-up ending on 1 May 2028. All individuals who present for prenatal care at sites in the DECIDE Study Consortium will be screened for eligibility. Individuals who meet study criteria will be approached for participation by study staff, which will include a study pamphlet with a weblink and QR code ( www.decidestudy.org ) and a 3 min video about GDM medication management ( https://youtu.be/CGmYCmF4vDo ). After eligibility is confirmed, individuals will be asked to participate after study information is given. Individuals who agree will complete the written informed consent process (see online supplemental file 2 for sample consent document). Reasons for ineligibility and rates of declining to participate will be collected. The patient advisory board will review recruitment and retention materials to create participant-friendly information and to assist with provider trainings. 33
Baseline visit.
A research team member or healthcare provider will ask patients if they are interested in joining the study either in person or virtually. Those who are interested will be given an orientation to the study by a research team member ( figure 2 ). Written informed consent will be obtained in English or Spanish. Enrolled individuals will be randomised in a 1:1 ratio to metformin or insulin. Study staff will inform the provider about randomisation arm via telephone, email and the electronic medical record. Because this is an open label, non-blinded pragmatic trial comparing two treatments that are standard of care, individuals allocated to either arm will obtain their medication from their preferred pharmacy with a prescription from their provider, which will account for brand of insulin on formularies of their insurance plans.
Flow diagram of DECIDE study events.
Using defined data fields, we will record participant demographics, medical history and obstetric characteristics. Participants will complete standardised surveys at randomisation to assess lifestyle, health behaviours, diet, mental and physical health during pregnancy, DKQ, DDS and DMSES.
Consistent with a pragmatic trial, all dosing changes will be performed by the participant’s provider. The frequency of participant clinical encounters will be about every 2 weeks, which is standard clinical practice for GDM management, 34 with virtual or in-person clinic visits at the discretion of the provider. Study staff will visit with participants monthly (preferably in person and otherwise virtually), ask them about adherence to assigned treatment, review side effects related to their medication including nausea and symptoms of hypoglycaemia and assess medication adherence. Additionally, study staff will review and abstract the following information from the participant’s medical record: capillary blood glucose log values or continuous glucose monitoring logs for the past 1 week period, current metformin and insulin dosing, type(s) of insulin, and gestational weight gain. Finally, study staff will assess for adverse events (AEs) and serious AEs (SAEs) at each study visit.
The assigned treatment (insulin vs metformin) will be discontinued at delivery. The provider will have the responsibility for intrapartum and postpartum management. Data on intrapartum and postpartum glycaemic management will be abstracted by the study team. 35 Additionally, participant and infant data will be collected from the EHR until hospital discharge. We will collect comprehensive antepartum and intrapartum data, such as labour and delivery details; glycaemic control and neonatal outcomes. To address concerns about placental transfer of medication and fetal safety, cord blood and placental samples will be collected when possible for further analyses.
~6 weeks post partum.
At ~6–8 weeks post partum, participants will complete standardised surveys with the study team in person, by a virtual platform, or by mail, per participant and site preference. This visit will include standardised measures to assess treatment adherence and satisfaction, lactation, lifestyle, health behaviours, diet, pregnancy and childbirth experience, and MIBS. Additional participant and child postpartum data through the ~6 weeks postpartum visit will be collected from the EHR. Testing for diabetes at the postpartum visit is standard of care. We will collect these results and emphasise best practices to increase uptake of diabetes screening. 36 Research staff will actively maintain contact with participants every 6 months after delivery by telephone, email or post.
At or after 2 years post partum, participants will be invited to return for an in-person assessment and physical exam of both the participant and child. Participants will complete standardised surveys to assess lifestyle, health behaviours and diet in the mother and child. Physical exam of the child will include growth (weight and height), anthropometry (arm and abdominal circumference) and adiposity (skinfolds) measurements. For the participant, blood pressure, height, weight, anthropometry (waist and hip circumferences) and adiposity (skinfolds) will be obtained. The assessor at the visit will be masked to study arm assignment. A calibrated scale with stadiometer will be used for weight and height, a tape measure for anthropometry and callipers for skinfold measurements. As has been done in prior GDM cohorts (co-I PC), all sites will undergo central training and assessment in these techniques to promote standardisation of measurements and adherence to the study protocol. 37 38 Ongoing training and strict monitoring of the study team measurement techniques will be performed and regular assessment of interobserver variation will be conducted via training videos every 6 months. For the postpartum individual, blood will be obtained using standard venipuncture techniques, including for haemoglobin A1c, cholesterol panel and 2-hour 75 g oral glucose tolerance test using standardised collection procedures. 39
In years 2–4 of the study, 150 individuals across all sites will be invited to complete a 30–45 min interview approximately 6 weeks after delivery. Individuals will be purposefully recruited from each study site and across each year of the study to ensure diversity with respect to race and ethnicity, age, insurance status and study arm. Participants will be given the option to be interviewed by phone or via video (eg, zoom).
A semistructured interview guide has been developed and refined based on feedback from patients and experts on qualitative research methods ( online supplemental file 2 ). These interviews will be performed centrally at OSU under the supervision of co-I ASM. The guide will include open-ended questions in the following domains: DECIDE trial participation, GDM and pregnancy, medications to treat GDM, experiences taking medication to treat GDM, and GDM and postpartum health. 31 During the interviews, participants will be asked to describe their experiences using metformin and/or insulin with question probes to address specific aspects of their experiences, including barriers to and facilitators of metformin and insulin use, and the factors that might improve adherence and pregnancy experience. The draft guide will be pilot tested and finalised prior to use in the study. All interviews will be audio recorded and transcribed verbatim to allow rigorous qualitative analysis.
Patient and stakeholder interactions have helped develop the research question, comparators, study participants characteristics and relevant outcomes. 31 The DECIDE investigator team will engage key opinion leaders, patients with a history of GDM or living with diabetes and stakeholders, including public and private-sector insurers, advocacy organisations and professional societies, to elicit feedback through the Patient Advisory Board (co-I SC) and Stakeholder Engagement Board (co-I AT), both of which were established for this study and DiabetesSisters (co-I DR), a national patient advocacy organisation dedicated to women living with diabetes. The team will draw on participatory learning approaches, such as adaptive management, rapid assessments, data-driven decision-making and human-centred design to codevelop recommendations to inform project improvement.
The outcomes in this study include both clinical outcomes and PROs that cumulatively measure obstetric and perinatal morbidity and mortality that impact quality of life, well-being and pregnancy experience. Clinical outcomes relevant to contemporary practice are based on prior GDM pharmacotherapy trials, 20 23 40 meta-analyses comparing treatment strategies for GDM to prevent adverse outcomes, 18 19 26 engagement with stakeholder organisations and providers, 41 and core outcome sets for GDM that used a Delphi methodology. 42 43 PROs 44 are based on validated and standardised instruments that address birth experiences, living with diabetes, and treatment experiences and adherence; systematic reviews of prior qualitative studies of patient experiences with GDM; and testimonials from affected individuals and their families. 29 30 45–47
The primary pregnancy outcome (aim 1) is a neonatal composite adverse outcome of LGA birth weight, hypoglycaemia, hyperbilirubinaemia and/or death ( table 3 ). 48 This measure is based on neonatal outcomes causally related to glycaemic control and consistent with that used in recent trials 23 and meta-analyses. 19
Primary and secondary outcomes
In the follow-up at 2 years, the primary child outcome is BMI as a continuous measure, which is consistent with prior GDM RCT follow-up studies and meta-analyses. 18 19 40 An updated sex-adjusted US reference will be used for standardisation of height and weight for age. 49 Anthropometry will be measured with standardised protocols that have been successfully implemented in prior GDM cohorts at birth (co-I PC). 37 38 50
Secondary pregnancy outcomes ( table 3 ) include neonatal outcomes (preterm birth, small-for-gestational-age, NICU admission, mechanical ventilation by duration, oxygen support by type and duration, and respiratory distress syndrome by clinical features and oxygen or respiratory support for any time during the first 72 hours after birth and participant outcomes (hypertensive disorder of pregnancy, mode of delivery, total gestational weight gain and obstetric anal sphincter injuries).
In the follow-up at 2 years, secondary child outcomes include obesity and measures of adiposity and anthropometry. Secondary participant outcomes at 2 years will include type 2 diabetes, obesity, pre-diabetes, hypertension, metabolic profile, and measures of anthropometry and adiposity.
At randomisation, baseline assessments will include mental and physical health (PROMIS Global Short Form 51 52 ; DKQ, 53 DDS, 54 DMSES 55 and social determinants of health (Accountable Health Communities Health-Related Social Needs Screening Tool 56 and Williams Everyday Discrimination Scale 57 ( table 3 ).
At ~6 weeks postpartum follow-up visit, PROs will include treatment adherence and satisfaction Treatment Satisfaction Questionnaire for Medication 58 and Acceptability of Treatment 20 ; infant feeding practices (CDC Infant Feeding Practices, selected questions) 59 ; pregnancy and childbirth experience (Birth Satisfaction Scale-Revised Indicator) 60 ; MIBS 61 ; International Physical Activity Questionnaire (IPAQ), short-form 62 ; Mini-EAT (Eating Assessment Tool) 63 and the Brief Infant Sleep Questionnaire-Revised Short Form (selected questions). 64
At 2 years, PROs of the postpartum individual will include IPAQ, long-form 62 ; Mini-EAT 63 ; social determinants of health 56 ; mental and physical health 51 and of the child will include CDC Child Health and Diet Survey (selected questions) 65 ; Movement Behaviour Questionnaire (selected questions) 66 ; Brief Infant Sleep Questionnaire-Revised Short Form (selected questions) 64 and the Child Eating Behaviour Questionnaire. 67
Study guidelines for metformin and insulin management including initiation, dosing, titration and monitoring have been developed based on current clinical guidelines. 68 Pharmacotherapy will be initiated when ≥30% of fasting and/or postprandial glucose values are elevated in the past week. Given this is a pragmatic trial, clinical practice may vary slightly across sites based on local standard-of-care and individualised provider–patient decision-making.
Metformin (either extended or immediate release) will be started at 500 mg two times per day and titrated to a maximum daily dose of 2500 mg. Participants randomised to metformin will be given uniform advice by study personnel on how to minimise gastrointestinal distress, such as taking study tablets prior to meals and using antiemetics.
Providers will be encouraged to use trimester-specific and weight-based insulin dosing criteria for both basal and prandial insulins for up to a total of 4 daily injections. Consistent with clinical practice, some participants may be managed with a single dose of intermediate-acting or long-acting insulin at night to treat isolated fasting hyperglycaemia, while others may require additional treatment of postprandial hyperglycaemia with shorter-acting insulin. The sites’ insulin formularies may include rapid (Novolog, Humalog and Aspart), intermediate (Humulin N, Novolin N and NPH) and long-acting insulins (detemir, Lantus and degludec) with comparable efficacy in pregnancy. Participant insurance coverage will be considered when selecting insulin type.
We will standardise the incorporation of best practices regarding metformin and insulin titration per ADA and ACOG guidelines. Providers at each site will be instructed on the study protocol and trained on study procedures, including glycaemic monitoring ~1–2 weeks and uptitration. Glucose assessment by those participants electing self-monitored blood glucose monitoring will be performed at fasting and three times postprandial; those who elect continuous glucose monitoring will be asked to similarly document their fasting and postprandial values. Adherence to these goals will be monitored by research staff monthly from participant interviews and medical record review. Concerns regarding protocol adherence will be discussed with site PIs. Weekly participant glucose logs and total metformin and insulin doses and type of insulin will be recorded and considered in data analysis.
Participants receiving metformin will have insulin supplemented (ie, addition of insulin to base regimen of metformin) only if they have not achieved euglycaemia for at least 30% of glucose values after approaching the maximum daily dose of metformin (>2000 mg or the maximum tolerated dose). Participants will be asked to continue taking metformin after treatment supplementation with insulin, which is generally the current clinical practice. In rare circumstances (0%–2%) in which severe gastrointestinal distress or intolerable side effects are present with metformin, participants may be prescribed insulin before reaching the maximum daily dose of metformin (2500 mg) or switched to insulin entirely. 69 Reasons that patients and providers decide on treatment supplementation will be collected.
An independent data safety monitoring board (DSMB) has been created to provide oversight of trial accrual and of privacy and safety of study participants. DSMB members have appropriate expertise (obstetrics and gynaecology, maternal–fetal medicine, endocrinology, neonatology, bioethics and biostatistics). The DSMB will meet to review the protocol prior to study initiation and then yearly to review study progress. The DCC will provide reports to the DSMB that include recruitment, protocol adherence and safety outcomes.
Detailed information about AEs and SAEs will be collected and evaluated throughout the trial. If a patient develops an SAE, the primary clinician in collaboration with the site PI will ascertain the safety of continuing the intervention. All unanticipated and possibly study-related AEs and SAEs will be reported to the IRB per regulatory reporting guidelines. Metformin may be temporarily stopped in the setting of acute kidney injury or intravenous contrast administration. Metformin has been reported to be very rarely associated with lactic acidosis (<10 cases per 100 000 patient-years), although the validity of this association has been challenged. 70 We will include lactic acidosis on metformin as a safety stopping rule.
Sample size and power.
Published data suggest that upwards of 30% of individuals with GDM have an associated adverse neonatal outcome. Using data from recent meta-analyses that compared the two treatment regimens, 18 19 71 and the most recent RCT (although comparing glyburide to insulin) that assessed the same primary composite outcome as in our study, 23 we estimate the frequency of the primary composite perinatal outcome to be 28% with insulin. To be conservative, we have used an estimate of 25%.
We have chosen a non-inferiority trial design because metformin’s advantages in terms of cost and ease (eg, oral, no refrigeration needed, less costly) suggest that metformin may be the preferred first-line treatment for GDM if it were found to be non-inferior to insulin in terms of efficacy and safety. 41 A non-inferiority margin of 8% was selected for the primary outcome based on a survey and interviews we conducted in January to June 2021 with each of our 20 site PIs, all of whom are maternal-fetal medicine specialists, as well as interviews with 144 patients. This conservative margin is also consistent with recent non-inferiority RCTs for GDM. 23 Additionally, we estimate that 20% of individuals who are randomised to metformin will require supplemental insulin, 21 which is lower than prior trials because we will exclude those with fasting hyperglycaemia (>115 mg/dL for >50% in the prior week) who are at the highest risk of failing metformin. 20
Based on the above assumptions, we plan to enrol 1572 individuals to determine if metformin is non-inferior to insulin for the composite primary outcome, with 90% power, one-sided significance level of 0.025, a loss to follow-up at delivery of 2% and 20% supplementation with insulin in addition to metformin.
For the 2-year follow-up, if outcomes are obtained on 1415 participants (ie, a loss to follow-up rate of 10%), there will be 90% power to rule out an effect size of at least 0.172 SD. This translates to a 0.31 unit difference in BMI or a 0.29 kg mean difference in child weight. 25 There will be 80% power to rule out an effect size of at least 0.149 SD, or a 0.27 unit difference in BMI or a 0.25 kg mean difference.
We will use descriptive statistics to characterise participants to determine comparability of treatment groups at baseline. As an intention-to-treat analysis, the comparison is between individuals randomised to start metformin regardless of whether they later required supplemental insulin or stopped metformin due to side effects and switched to insulin versus individuals randomised to start on insulin. Analyses of the primary outcome will consist of summarising the proportions of trial participants with the primary endpoint for each group and calculating the corresponding between-group risk difference (insulin minus metformin) with 95% CIs.
Data analyses will adhere to the CONSORT (Consolidated Standards of Reporting Trials) guidelines and follow the intention-to-treat principle in which patients are analysed in the group to which they were randomised, regardless of whether they received the assigned intervention or altered their assigned medication prior to delivery. Metformin will be determined as non-inferior if the lower 95% confidence limit for the risk difference is −8 percentage points or greater (ie, closer to 0). If treatment groups differ on a pretreatment factor known to be a risk factor for the outcome, the analysis will adjust for these differences and an adjusted risk difference will be reported. If metformin is determined to be non-inferior to insulin, a superiority test will be conducted without adjusting the type I error, with metformin considered superior if the lower 95% confidence limit for the risk difference is more than 0.
Since the sample size estimate is based on the assumption that the primary endpoint rate will be 25% in the insulin group, it is important to evaluate this proportion in the study after 20% of the participants (N=315) have delivered. In addition, the proportion of patients in the metformin group who require supplemental insulin will be reported. Once 50% of the participants have delivered (N=786), a formal interim analysis will be performed to determine whether metformin is inferior to insulin, with an upper boundary for the stopping rule for harm based on a one-sided type I error of 0.025 and the Lan-DeMets generalisation of the O’Brien-Fleming boundary. If the upper confidence bound for the risk difference is less than 0, the DSMB will evaluate this in the context of the other safety outcomes. We also plan to calculate conditional power given the observed data and conditional on future data showing no difference between treatment strategies. If the conditional power is high (>90%) that the neonatal composite rate will be more that 8% higher in the metformin arm, the DSMB will consider termination for futility, although any decision to terminate the study would not be reached solely on statistical grounds but on a number of clinical and statistical considerations.
Child BMI is the primary outcome at 2 years of age. Analyses will consist of summarising the mean BMI standardised for age and sex for each group and calculating the corresponding between-group mean difference with 95% CIs using generalised linear models. Metformin will be determined as non-inferior to insulin if the lower 95% confidence limit for the mean difference is 0.31 units or greater (ie, closer to 0). Additional analyses as detailed above for the primary neonatal composite in the RCT will be performed, including for measures of child adiposity and anthropometry. Fetal sex will be evaluated for predefined interaction analyses with treatment group, and anthropometry will be standardised by sex-specific standards. 48
We will use the constant comparative method and a grounded theory approach to analyse interview data. 72 This iterative approach to analysis will include reading interview transcripts and discussing findings among investigators as the study progresses. Our approach will enable exploration of emergent themes and ensure saturation in data collection. Analysis will prioritise the elucidation of key concepts from individuals’ interview statements (extraction), conceptual development based on constant comparative analysis, and classification of data through code development. 72 73 The coding team (co-I ASM) will create a preliminary coding dictionary based on the interview guide, defining broad categories of findings to enable coding of responses to interview questions. Frequent discussions among coding team members will allow the characterisation of emergent codes and ensure agreement about identified themes and subthemes. ATLAS.ti software will be used to support the analysis process.
Treatment effectiveness for subgroups may differ due to barriers related to social determinants of health (eg, race/ethnicity), bioavailability of medication, physiologic insulin resistance (eg, BMI) or factors related to GDM and its severity (eg, maternal age, gestational age at medication initiation) ( online supplemental file 3 ). We will employ existing rigorous checklist for addressing the design, analysis and context of subgroup analyses. 74 These risk factors were selected based on differences in the frequency of GDM and adverse pregnancy and postpartum outcomes, and hence, at least a theoretical possibility as to why HTE may exist. We will formally assess for effect modification (interaction effect). Should we note significant heterogeneity of treatment effect across these prespecified groups (p<0.05), we will then systematically examine two-way effect modification. Should there be evidence of HTE, the proposed exploratory subgroup analyses will employ a non-inferiority approach consistent with the overall trial design and analysis plan.
We will investigate the robustness of the observed differences between the two groups with respect to any missing data. First, an inverse probability weighting (IPW) analysis will be conducted with each case weighted by the inverse probability of being a complete case. Under a missing-at-random mechanism, the IPW approach would result in an unbiased estimate of the difference between groups assuming a correctly specified model for the missing data. Second, a tipping-point analysis will describe the additional number of events in the insulin group versus the metformin group among the participants with missing data that would change the conclusion related to non-inferiority. In addition, a sensitivity analysis will be performed among participants in the metformin group who did not require supplemental insulin versus participants randomised to insulin only.
Participants will be asked to provide contact information (eg, phone, email and address) for themselves and two relatives who would know how to contact them. Research staff will actively maintain contact with participants throughout their pregnancies and by telephone, email or post, every 6 months after delivery. Participants will be asked to verify or update information at each contact. We will also maintain contact with participants and their families through flyers, cards and electronic communications in order to provide study updates.
Participant reimbursement will be provided for completing assessments at multiple time points: randomisation (US$100), 6 weeks post partum (in person, virtual and/or telephone) (US$50) and 2-year follow-up visits for the participant and child (in person) (US$125). Participants selected for qualitative interviews will receive additional compensation (US$100).
The OSU Institutional Review Board (IRB), which will serve as the single IRB of record for all sites, has approved this protocol. All protocol amendments will be communicated for approval to the OSU IRB. Before a site may start the trial, it must be certified, which involves certification of research staff and an IRB reliance agreement with the single IRB.
We will submit study results for publication in peer-reviewed journals. The DCC and CCC will maintain access to the final trial dataset, and a limited deidentified dataset will be released via the online portal of the primary funder. A key component of our dissemination plan will be increasing patient and provider awareness about the comparative effectiveness results. Our partnership with DiabetesSisters and the Stakeholder Engagement Group will be leveraged for dissemination of results, including appropriate forums (eg, meetings, newsletters, social media communities, online videos). We will share accessible evidence-based factsheets and provide our primary publications for free download, including to study participants.
In this two-arm, open-label, pragmatic, comparative effectiveness RCT, we will examine whether metformin is not inferior to insulin in reducing adverse pregnancy outcomes and is comparably safe for exposed mothers and children, and whether patient-reported factors including facilitators and barriers of medication use differ between metformin versus insulin use. The DECIDE trial will randomise 1572 pregnant individuals with GDM who need pharmacotherapy at 20 US sites—with uniform diagnostic and treatment criteria—to oral metformin versus injectable insulin and follow them and their children through delivery and then to 2 years post partum.
The proposed comparative effectiveness study is designed to inform one of the most frequent medication decisions in pregnancy. The clinical equipoise that currently exists in use of these medications for GDM underscores that a trial with pregnancy and postpartum follow-up in a diverse, representative and contemporary US population is necessary and will fill a key knowledge gap affecting everyday practice, patient experience and clinical outcomes. 41 These themes, listed in bold below, have been identified as critical by stakeholders including patients, providers, researchers and professional societies.
Among the major limitations of the RCTs to date are (1) using varying GDM diagnostic criteria, (2) unclear criteria or guidelines for supplemental insulin, (3) lack of sufficient power for important outcomes, (4) insufficient long-term assessment of outcomes in exposed children, (5) unreported patterns of hyperglycaemia potentially influencing treatment effectiveness and (6) results from populations that do not reflect a contemporary US population. DECIDE will address each of these limitations with uniform diagnostic and treatment criteria and inclusion of 20 academic and community centres representative of major US geographical regions with diverse population characteristics.
Experts have cautioned that a GDM treatment trial without a plan for robust postnatal follow-up will not meaningfully fill the evidence gap and allow best practices to be determined. 24 71 DECIDE embeds a seamless, preplanned and rigorous follow-up of all randomised mother–child dyads.
An in-depth understanding of patient and other key stakeholder perspectives on barriers to and facilitators of metformin versus insulin use is necessary to identify opportunities to improve outcomes. DECIDE includes PROs and outcomes that focus on the same constructs to bolster patient and stakeholder confidence. 75 DECIDE also assesses patient experiences, such as medication side effects, whether patients take medicines as directed, how satisfied they are with their medication choice, and how their medication choice impacts their pregnancy and postpartum experience, which may explain observed HTEs.
The proposed study is designed with the goal of informing healthcare decisions, both by filling an important evidence gap and by ensuring that the evidence provided is aligned with and informed by patients and other healthcare partners. While conducting the study, we will engage with the patient advisory board and stakeholder engagement group, which includes patients, patient advocates, clinicians, researchers, purchasers, payors, industry, health systems and policy-makers. We will discuss the study protocol and startup in a cooperative learning environment, and these stakeholders will be invited to participate in data analysis to add their perspectives to promote authenticity.
Limitations.
First, while randomisation to pharmacotherapy minimises selection bias, lack of patient and provider blinding to treatment can introduce bias. Second, because this is a pragmatic RCT, variations in insulin formulary and differences in medication titration may result in heterogeneity in outcomes. To minimise the impact of variation of treatment effects across study sites, we have instituted uniform criteria for treatment initiation, defined as ≥30% elevated glucose values in the prior week. Also, the DECIDE manual of operations will contain guidelines for insulin and metformin management and standardised glycaemic targets for medication titration. We will stratify randomisation by site, and we will consider adjustment for site in analyses via both stratification and interaction effects. Finally, we include follow-up through 2 years postpartum, although longer follow-up may be necessary to assess the long-term impact of pharmacotherapy on outcomes.
We have powered our study to a conservative non-inferiority margin, which is consistent with recent non-inferiority RCTs for GDM 23 and allows for substitution of supplemental insulin for those on metformin. Second, we examine postpartum safety following exposure to metformin versus insulin on child and maternal/paternal health. Third, we integrate rigorous assessment of patient preferences and values through PROs, standardised measures and qualitative interviews as part of the RCT and follow-up. Finally, DECIDE includes a racially, ethnically and geographically diverse patient population with broad inclusion criteria reflective of obstetric practice to maximise relevance, impact and generalisability.
Patient consent for publication.
Not applicable.
Contributors KKV, CM, RGC, GS and ML designed the study. KKV, CM, RGC, AB, DG and ML wrote the methods manuscript. CP, ASM, LF, PC, AT and DR provided oversight for study design and implementation. CM and RGC provided statistical support and oversight. Under the clinical oversight of KKV, ML, MC, ANB, KB, KE, TE, MNF, LH, AK, MK-W, HM-F, MM, AS, NS, DS, SW, and CAZ assisted with the clinical trial development and execution. All authors revised the manuscript for relevant scientific content and approved the final version of the manuscript.
Funding This work was supported through a Patient-Centered Outcomes Research Institute (PCORI) BPS-2022C3-30268.
Disclaimer All statements in this report, including its findings and conclusions, are solely those of the authors and do not necessarily represent the views of the Patient-Centered Outcomes Research Institute (PCORI), its Board of Governors or Methodology Committee.
Map disclaimer The depiction of boundaries on this map does not imply the expression of any opinion whatsoever on the part of BMJ (or any member of its group) concerning the legal status of any country, territory, jurisdiction or area or of its authorities. This map is provided without any warranty of any kind, either express or implied.
Competing interests None declared.
Patient and public involvement Patients and/or the public were involved in the design, or conduct, or reporting, or dissemination plans of this research. Refer to the Methods section for further details.
Provenance and peer review Not commissioned; peer reviewed for ethical and funding approval prior to submission.
Supplemental material This content has been supplied by the author(s). It has not been vetted by BMJ Publishing Group Limited (BMJ) and may not have been peer-reviewed. Any opinions or recommendations discussed are solely those of the author(s) and are not endorsed by BMJ. BMJ disclaims all liability and responsibility arising from any reliance placed on the content. Where the content includes any translated material, BMJ does not warrant the accuracy and reliability of the translations (including but not limited to local regulations, clinical guidelines, terminology, drug names and drug dosages), and is not responsible for any error and/or omissions arising from translation and adaptation or otherwise.
Research design and methods, conclusions, article information, diabetes, prediabetes, and brain aging: the role of healthy lifestyle.
Abigail Dove , Jiao Wang , Huijie Huang , Michelle M. Dunk , Sakura Sakakibara , Marc Guitart-Masip , Goran Papenberg , Weili Xu; Diabetes, Prediabetes, and Brain Aging: The Role of Healthy Lifestyle. Diabetes Care 20 September 2024; 47 (10): 1794–1802. https://doi.org/10.2337/dc24-0860
Download citation file:
Diabetes is a well-known risk factor for dementia. We investigated the association between (pre)diabetes and older brain age and whether this can be attenuated by modifiable lifestyle behaviors.
The study included 31,229 dementia-free adults from the UK Biobank between the ages of 40 and 70 years. Glycemic status (normoglycemia, prediabetes, or diabetes) was ascertained based on medical history, medication use, and HbA 1c measured at baseline. Information on cardiometabolic risk factors (obesity, hypertension, low HDL, and high triglycerides) and lifestyle behaviors (smoking, drinking, and physical activity) was also collected at baseline. Participants underwent up to two brain MRI scans over 11 years of follow-up. Brain age was estimated using a machine learning model based on 1,079 brain MRI phenotypes and used to calculate brain age gap (BAG; i.e., brain age minus chronological age).
At baseline, 13,518 participants (43.3%) had prediabetes and 1,149 (3.7%) had diabetes. Prediabetes (β = 0.22 [95% CI 0.10, 0.34]) and diabetes (2.01 [1.70, 2.32]) were both associated with significantly higher BAG, and diabetes was further associated with significant increase in BAG over time (0.27 [0.01, 0.53]). The association between (pre)diabetes and higher BAG was more pronounced in men and in people with two or more cardiometabolic risk factors. In joint exposure analysis, having a healthy lifestyle (i.e., no smoking, no heavy drinking, and high physical activity) significantly attenuated the diabetes-BAG association.
Diabetes and even prediabetes are associated with accelerated brain aging, especially among men and people with poor cardiometabolic health. However, a healthy lifestyle may counteract this.
Type 2 diabetes (hereafter, diabetes) is a well-established risk factor for cognitive impairment and has been associated with approximately double the risk of dementia ( 1–3 ). In brain MRI studies, diabetes has been related to global brain atrophy, increased burden of small-vessel disease, and microstructural lesions before the onset of cognitive symptoms ( 4 ). While prediabetes has been related to more modest levels of many of the cerebrovascular and neurodegenerative abnormalities associated with overt diabetes in some MRI studies ( 5 , 6 ), the association of prediabetes with cognitive decline and dementia remains controversial, with previous studies reporting conflicting results ( 7–10 ).
Recently, modeling methods have been introduced to estimate brain age based on MRI features such as volume loss, cortical thinning, white matter degradation, loss of gyrification, and ventricle enlargement ( 11 ). Brain age gap (BAG) reflects the difference between brain age and chronological age. Having an older-appearing brain for one’s chronological age—that is, a high BAG—can indicate deviation from the normal aging process and has been linked to mortality and increased risk of cognitive decline and dementia ( 11 ). Early detection of accelerated brain aging could support timely identification and intervention for people who are most at risk for developing dementia.
A growing body of cross-sectional studies has linked diabetes to brain age that is between 0.85 and 4.6 years older than chronological age ( 12–18 ), but longitudinal evidence on the association between diabetes and changes in brain age is lacking, and the relationship between prediabetes and brain age has not been explored. Given the heterogeneity of the diabetes population, another important consideration is how clinically relevant factors, such as sex, comorbidities, and lifestyle behaviors, might influence the association between (pre)diabetes and brain age. A variety of lifestyle behaviors, including physical activity and smoking/alcohol avoidance, have been related to decelerated brain aging ( 12 , 16 , 19 , 20 ), but whether a healthy lifestyle can counteract the detrimental influence of (pre)diabetes is unknown.
To address these questions, we comprehensively investigated the relationship between hyperglycemia and brain aging, leveraging detailed neuroimaging data from the UK Biobank covering six different MRI modalities in >30,000 middle-aged and older adults. Specifically, we aimed to 1 ) examine the cross-sectional and longitudinal relationship between (pre)diabetes and BAG; 2 ) explore the role of sex and cardiometabolic risk factors in these associations; and 3 ) investigate whether a healthy lifestyle, characterized by high physical activity and abstention from smoking and heavy drinking, can attenuate the influence of (pre)diabetes on BAG.
The UK Biobank is an ongoing longitudinal study including >500,000 adults between the ages of 40 and 70 from across the United Kingdom ( 21 ). Between 2006 and 2010, participants took part in a baseline examination at 1 of 22 assessment centers across the country consisting of physical and medical assessments and a series of questionnaires about sociodemographic information and lifestyle behaviors. Approximately 9 years later, between 2014 and 2020, >40,000 participants additionally underwent a brain MRI scan. Beginning in 2019, participants were invited to return for a follow-up brain MRI scan.
Selection of the study population is illustrated in Supplementary Fig. 1 . The analysis was restricted to 34,296 participants who underwent brain MRI scans and had complete information on all available imaging-derived phenotypes (IDPs). We then excluded 630 participants with chronic neurological disorders (including dementia) at the time of the MRI scan (see Supplementary Table 1 for details), 15 with type 1 diabetes, and 2,422 with missing information on baseline HbA 1c , leaving a sample of 31,229, including 2,414 who underwent two MRI scans.
All data collection procedures have been approved by the UK National Research Ethics Service (Ref 11/NW/0382) and the use of the data for the present analyses were additionally approved by the Regional Ethical Review Board in Stockholm, Sweden (Ref 2024-00520-01). All participants provided informed consent at baseline.
Baseline diabetes and prediabetes were defined according to the American Diabetes Association standard diagnostic criteria ( 22 ). Participants were classified as having diabetes if they had any one of the following: medical record of diabetes, use of glucose-lowering medications, self-reported history of diabetes, or HbA 1c ≥6.5% (see Supplementary Table 2 for field codes). Among diabetes-free participants, prediabetes was defined as HbA 1c 5.7% to 6.4%, and normoglycemia was defined as HbA 1c <5.7%. Diabetes was further categorized according to level of glycemic control: <7.0% (well-controlled), ≥7.0 to <8.0% (moderately controlled), or ≥8.0% (poorly controlled) ( 23 ).
Brain MRI scans were conducted using a Siemens Skyra 3T scanner. Detailed descriptions of the UK Biobank brain MRI image acquisition and processing protocols have been previously published ( 24 , 25 ) and are summarized in Supplementary Table 3 .
A total of 1,079 IDPs were extracted across six MRI modalities: 165 from T1-weighted MRI, 1 from T2-fluid attenuated inversion recovery (FLAIR), 14 from T2*, 675 from diffusion MRI, 210 from resting-state functional MRI (fMRI), and 14 from task fMRI. Briefly, T1-weighted imaging provides information on the volume and thickness of different brain regions, T2-FLAIR imaging detects white matter hyperintensities (reflecting vascular brain damage), T2* detects brain microbleeds, diffusion MRI assesses white matter microstructural integrity, resting-state fMRI measures brain activity at rest for assessment of intrinsic functional connectivity of neural networks, and task fMRI does so when the participant is performing a task or experiencing a sensory stimulus (in this case, a face/shapes matching task) ( 24 ). A full list of all 1,079 IDPs is provided in Supplementary Material .
The procedure for brain age estimation has been described in previous studies ( 26 , 27 ). A detailed description is available in the Supplementary Material , and the workflow is illustrated in Supplementary Fig. 2 .
Briefly, from the entire sample of participants with complete brain MRI data ( N = 34,296), we first identified 4,355 healthy individuals between the ages of 40 and 70 with no ICD-10 diagnoses and who were free from self-reported long-term illness, disability, or frailty (Field ID: 2188) and self-reported fair or poor health status (Field ID: 2178) ( Supplementary Table 4 ). These participants were randomly allocated in a 4:1 ratio to a training set ( n = 3,484) and a validation set ( n = 871). Next, all 1,079 IDPs were Z standardized and nine machine learning models were trained for modeling brain age in the training set. These included least absolute shrinkage and selection operator regression (LASSO), eXtreme gradient boosting, and support vector regression, which were combined with three possible feature selection strategies (no feature selection, FeatureWiz, or recursive feature elimination with cross validation). Bayesian optimization was performed to optimize the hyperparameters of all nine models through 100 epochs ( Supplementary Tables 5 and 6 ). Once optimized, all nine models were applied to the validation set so that their performance could be compared. Ultimately, the LASSO model without feature selection achieved the lowest mean absolute error ( Supplementary Table 7 ) and was therefore chosen to predict brain age for the entire sample. Of the 1,079 IDPs, 285 contributed significantly to the brain age estimate and are listed in Supplementary Table 8 .
Next, because brain age tends to be overpredicted in younger individuals and underpredicted in older individuals, we corrected brain age estimates for age bias as follows ( 28 , 29 ): brain age corrected = [brain age original – β/α] , where coefficients α and β are the slope and intercept of brain age training set = α × chronological age training set + β ( Supplementary Fig. 3 ).
Finally, BAG, which represents the difference between an individual’s brain age and their chronological age, was calculated as BAG = brain age – age time of MRI . Positive values for BAG indicate a brain that is older (i.e., less healthy) and negative values for BAG indicate a brain that is younger (i.e., more healthy) than expected based on the individual’s chronological age.
Sociodemographic factors.
Education (college/university vs. not) was dichotomized based on the highest level of formal education attained. Socioeconomic status (SES) was assessed using the Townsend deprivation index, a measure of neighborhood-level socioeconomic deprivation based on the prevalence of unemployment, household overcrowding, car nonownership, and home nonownership in a given postcode of residence.
Cardiometabolic risk factor burden was operationalized in terms of the components of the metabolic syndrome (MetS) ( 30 ). BMI was calculated using height and weight measurements from the baseline examination and classified as underweight (<20 kg/m 2 ), normal weight (≥20 to <25 kg/m 2 ), overweight (≥25 to <30 kg/m 2 ), or obese (≥30 kg/m 2 ). Hypertension was defined based on self-report, blood pressure measurement (systolic ≥140 mmHg, diastolic ≥90 mmHg), or antihypertensive medication use. HDL cholesterol and triglycerides were measured from blood samples collected at baseline. A score reflecting cardiometabolic risk factor burden (ranging from 0 to 4) was generated according to the total number of MetS components present, including obesity, hypertension, low HDL (<40 mg/dL [1.03 mmol/L] for men and <50 mg/dL [1.29 mmol/L] for women), and high triglycerides (≥150 mg/dL [1.7 mmol/L]). (Notably, the fifth MetS component, hyperglycemia, was not included because it was already considered as the exposure in all analyses.)
Information was collected on three readily modifiable lifestyle behaviors: smoking, alcohol drinking, and physical activity. Smoking status was categorized as nonsmoker, former smoker, or current smoker according to self-report. Intake of various alcoholic beverages was self-reported and converted into U.K. alcohol units (1 unit = 8 g ethanol) ( 31 ). Alcohol consumption was categorized as nondrinker, light/moderate drinking (≤14 units/week), or heavy drinking (>14 units/week) according to current U.K. guidelines on alcohol consumption for both men and women ( 32 ). Physical activity was measured using the International Physical Activity Questionnaire. Participants were classified as inactive (<600 MET-min/week), moderate (600 to <3,000 MET-min/week), or active (≥3,000 MET-min/week); 600 MET-min/week is equivalent to the World Health Organization recommendation of 150 min of moderate-intensity or 75 min of vigorous physical activity per week ( 33 ). An optimal lifestyle was defined as never smoking, no or light/moderate alcohol consumption, and high physical activity.
Alzheimer disease (AD)-related polygenic risk score (PRS AD ) was obtained from the UK Biobank’s Standard PRS Set ( 34 ). Briefly, PRS AD represents the Z-standardized sum of each participant’s number of AD-related alleles (including the well-known APOE ε4 polymorphism) weighted by the strength of each allele’s association with AD ( 34 ).
Baseline characteristics of the study participants by glycemic status were assessed using χ 2 tests for categorical variables and one-way ANOVA for continuous variables.
Linear regression models were used to estimate β-coefficients and 95% CIs for the association between glycemic status at baseline and BAG at the time of brain MRI. Least-squares means of BAG in the normoglycemia, prediabetes, and diabetes groups were additionally estimated from the margins of the linear regression models. Similar analyses were conducted using HbA 1c as a continuous variable. Restricted cubic splines with three knots at fixed percentiles of the HbA 1c distribution (10th, 50th, and 90th) were used to model the possible nonlinear association between HbA 1c and BAG. Among participants who underwent two brain MRI scans, linear mixed-effects models were used to estimate β-coefficients and 95% CIs for the association between glycemic status and changes in BAG between the first and second scans. The fixed effect included baseline glycemic status, follow-up time (in years), and their interaction. The random effect included random intercept and slope, allowing individual differences in BAG to be reflected at baseline and across follow-up.
Next, stratified linear regression models were used to explore the role of sex (women vs. men) and cardiometabolic health (0–1 vs. ≥2 risk factors) in the association between glycemic status and BAG. Finally, we performed joint exposure analysis by incorporating a six-category indicator variable that combined glycemic status (normoglycemia, prediabetes, or diabetes) and lifestyle (optimal or nonoptimal) into the linear regression model. Interactions between glycemic status and sex, cardiometabolic risk factor level, and lifestyle were assessed by incorporating the cross-product term into the models.
All models were first basic adjusted for sociodemographic factors (i.e., age, sex, education, and SES), followed by further adjustment for number of cardiometabolic risk factors, lifestyle behaviors (i.e., smoking, alcohol consumption, and physical activity), and PRS AD . Missing values for covariates were imputed using fully conditional specification, with estimates pooled from five iterations.
In sensitivity analysis, we repeated the main analyses 1 ) using BAG calculated based on brain age estimates from other candidate machine learning models; 2 ) using nonimputed data; 3 ) after adding an additional covariate for brain MRI assessment center; 4 ) after excluding participants with possible prodromal/undiagnosed dementia (i.e., incident dementia during follow-up; n = 42) or possible cognitive impairment (i.e., baseline cognitive test scores <25th percentile; n = 7,806) to minimize the possibility of reverse causality; and 5 ) using diabetes status defined at the time of brain MRI scan to address the possibility of changes in glycemic status since baseline. All analyses were performed using Stata SE 16.0 software (StataCorp, College Station, TX). P values <0.05 were considered statistically significant.
Requests for access to the UK Biobank data can be made here: https://www.ukbiobank.ac.uk/enable-your-research/apply-for-access .
Baseline characteristics of the 31,229 study participants (mean age 54.8 ± 7.5; 53.0% female) are summarized in Table 1 . At baseline, 13,518 participants (43.3%) had prediabetes and 1,149 (3.7%) had diabetes. Compared with participants with normoglycemia, those with (pre)diabetes were more likely to be older, male, have a lower education level and SES, be physically inactive, and have cardiometabolic risk factors. The study sample was comparatively younger and had greater educational attainment, higher SES, and a more favorable cardiometabolic risk profile compared with the UK Biobank population as a whole ( Supplementary Table 9 ).
Baseline characteristics of the 31,229 study participants by glycemic status
. | . | By glycemic status . | . | ||
---|---|---|---|---|---|
Characteristics . | Full sample ( = 31,229) . | Normoglycemia ( = 16,562) . | Prediabetes ( = 13,518) . | Diabetes ( = 1,149) . | value . |
Age, years | |||||
At baseline | 54.8 ± 7.5 | 53.1 ± 7.5 | 56.7 ± 7.1 | 57.5 ± 7.0 | <0.001 |
At time of brain MRI | 63.7 ± 7.6 | 62.0 ± 7.6 | 65.6 ± 7.2 | 66.4 ± 7.2 | <0.001 |
Female sex | 16,556 (53.0) | 9,015 (54.4) | 7,105 (52.6) | 436 (38.0) | <0.001 |
College/university-educated | 14,503 (46.6) | 8,006 (48.5) | 6,038 (44.8) | 459 (40.2) | <0.001 |
Townsend deprivation index | −1.9 ± 2.7 | −1.9 ± 2.7 | −2.0 ± 2.7 | −1.5 ± 2.9 | <0.001 |
BMI, kg/m | 26.5 ± 4.2 | 25.9 ± 3.8 | 26.9 ± 4.3 | 30.2 ± 5.5 | <0.001 |
Underweight (<20) | 799 (2.6) | 490 (3.0) | 302 (2.2) | 7 (0.6) | <0.001 |
Normal (20–25) | 11,651 (37.3) | 6,870 (41.5) | 4,613 (34.2) | 168 (14.6) | |
Overweight (25–30) | 13,382 (42.9) | 7,020 (42.4) | 5,910 (43.8) | 452 (39.4) | |
Obese (≥30) | 5,367 (17.2) | 2,163 (13.1) | 2,683 (19.9) | 521 (45.4) | |
Hypertension | 6,594 (21.1) | 2,832 (17.1) | 3,197 (23.7) | 565 (49.2) | <0.001 |
HDL, mg/dL | 57.2 ± 14.6 | 58.1 ± 14.5 | 57.0 ± 14.5 | 48.6 ± 13.2 | <0.001 |
Triglycerides, mg/dL | 144.5 ± 84.0 | 135.0 ± 79.3 | 152.5 ± 85.3 | 186.4 ± 108.8 | <0.001 |
HbA , % | 5.7 ± 0.5 | 5.4 ± 0.2 | 5.9 ± 0.2 | 7.1 ± 1.0 | <0.001 |
Smoking | <0.001 | ||||
Nonsmoker | 19,036 (61.1) | 10,446 (63.2) | 8,007 (59.4) | 583 (50.8) | |
Former smoker | 10,258 (32.9) | 5,241 (31.7) | 4,537 (33.6) | 480 (41.8) | |
Current smoker | 1,873 (6.0) | 847 (5.1) | 941 (7.0) | 85 (7.4) | |
Alcohol consumption | <0.001 | ||||
Nondrinker | 1,970 (7.2) | 886 (6.0) | 983 (8.4) | 101 (10.4) | |
Low/moderate drinking | 11,785 (42.9) | 6,237 (42.4) | 5,134 (43.7) | 414 (42.4) | |
Heavy drinking | 13,705 (49.9) | 7,599 (51.6) | 5,645 (48.0) | 461 (47.2) | |
Physical activity | <0.001 | ||||
Low | 4,880 (18.1) | 2,575 (17.8) | 2,049 (17.7) | 256 (26.0) | |
Moderate | 11,320 (41.9) | 6,022 (41.7) | 4,896 (42.3) | 402 (40.8) | |
High | 10,801 (40.0) | 5,835 (40.4) | 4,638 (40.0) | 328 (33.3) | |
ε4 carrier | 7,322 (27.5) | 3,992 (28.1) | 3,105 (27.1) | 225 (23.8) | 0.008 |
PRS | 0.04 ± 0.99 | 0.05 ± 0.99 | 0.04 ± 0.98 | −0.04 ± 0.96 | 0.026 |
. | . | By glycemic status . | . | ||
---|---|---|---|---|---|
Characteristics . | Full sample ( = 31,229) . | Normoglycemia ( = 16,562) . | Prediabetes ( = 13,518) . | Diabetes ( = 1,149) . | value . |
Age, years | |||||
At baseline | 54.8 ± 7.5 | 53.1 ± 7.5 | 56.7 ± 7.1 | 57.5 ± 7.0 | <0.001 |
At time of brain MRI | 63.7 ± 7.6 | 62.0 ± 7.6 | 65.6 ± 7.2 | 66.4 ± 7.2 | <0.001 |
Female sex | 16,556 (53.0) | 9,015 (54.4) | 7,105 (52.6) | 436 (38.0) | <0.001 |
College/university-educated | 14,503 (46.6) | 8,006 (48.5) | 6,038 (44.8) | 459 (40.2) | <0.001 |
Townsend deprivation index | −1.9 ± 2.7 | −1.9 ± 2.7 | −2.0 ± 2.7 | −1.5 ± 2.9 | <0.001 |
BMI, kg/m | 26.5 ± 4.2 | 25.9 ± 3.8 | 26.9 ± 4.3 | 30.2 ± 5.5 | <0.001 |
Underweight (<20) | 799 (2.6) | 490 (3.0) | 302 (2.2) | 7 (0.6) | <0.001 |
Normal (20–25) | 11,651 (37.3) | 6,870 (41.5) | 4,613 (34.2) | 168 (14.6) | |
Overweight (25–30) | 13,382 (42.9) | 7,020 (42.4) | 5,910 (43.8) | 452 (39.4) | |
Obese (≥30) | 5,367 (17.2) | 2,163 (13.1) | 2,683 (19.9) | 521 (45.4) | |
Hypertension | 6,594 (21.1) | 2,832 (17.1) | 3,197 (23.7) | 565 (49.2) | <0.001 |
HDL, mg/dL | 57.2 ± 14.6 | 58.1 ± 14.5 | 57.0 ± 14.5 | 48.6 ± 13.2 | <0.001 |
Triglycerides, mg/dL | 144.5 ± 84.0 | 135.0 ± 79.3 | 152.5 ± 85.3 | 186.4 ± 108.8 | <0.001 |
HbA , % | 5.7 ± 0.5 | 5.4 ± 0.2 | 5.9 ± 0.2 | 7.1 ± 1.0 | <0.001 |
Smoking | <0.001 | ||||
Nonsmoker | 19,036 (61.1) | 10,446 (63.2) | 8,007 (59.4) | 583 (50.8) | |
Former smoker | 10,258 (32.9) | 5,241 (31.7) | 4,537 (33.6) | 480 (41.8) | |
Current smoker | 1,873 (6.0) | 847 (5.1) | 941 (7.0) | 85 (7.4) | |
Alcohol consumption | <0.001 | ||||
Nondrinker | 1,970 (7.2) | 886 (6.0) | 983 (8.4) | 101 (10.4) | |
Low/moderate drinking | 11,785 (42.9) | 6,237 (42.4) | 5,134 (43.7) | 414 (42.4) | |
Heavy drinking | 13,705 (49.9) | 7,599 (51.6) | 5,645 (48.0) | 461 (47.2) | |
Physical activity | <0.001 | ||||
Low | 4,880 (18.1) | 2,575 (17.8) | 2,049 (17.7) | 256 (26.0) | |
Moderate | 11,320 (41.9) | 6,022 (41.7) | 4,896 (42.3) | 402 (40.8) | |
High | 10,801 (40.0) | 5,835 (40.4) | 4,638 (40.0) | 328 (33.3) | |
ε4 carrier | 7,322 (27.5) | 3,992 (28.1) | 3,105 (27.1) | 225 (23.8) | 0.008 |
PRS | 0.04 ± 0.99 | 0.05 ± 0.99 | 0.04 ± 0.98 | −0.04 ± 0.96 | 0.026 |
Data are presented as means ± SD or n (%). Missing data: 92 for education level; 28 for Townsend deprivation index; 30 for BMI; 15 for hypertension; 4,030 for HDL; 1,406 for triglycerides; 62 for smoking status; 3,769 for alcohol consumption; 4,228 for physical activity level; 4,638 for APOE ε4 status; and 242 for PRS AD .
Indicates significant ( P value <0.05) pairwise comparison (reference group = normoglycemia).
Compared with normoglycemia, prediabetes (β = 0.22 [95% CI 0.10, 0.34]) and diabetes (β = 2.01 [1.70, 2.32]) were associated with significantly higher BAG ( Table 2 ). Specifically, brain age was on average 0.50 years older than chronological age among people with prediabetes and 2.29 years older than chronological age among people with diabetes ( Fig. 1A ). BAG rose as high as 4.18 years among people with poorly controlled diabetes (HbA 1c ≥8.0%). Consistent with this, HbA 1c as a continuous variable was associated with significantly higher BAG (β = 0.77 [0.65, 0.90]), and the restricted cubic spline analysis showed a strong increase in BAG with higher levels of HbA 1c ( Fig. 1B ).
Cross-sectional and longitudinal associations between glycemic status and BAG: results from linear regression and linear mixed-effects models
. | . | BAG . | |||
---|---|---|---|---|---|
. | . | Basic adjusted . | Multiadjusted . | ||
Glycemic status . | Participants ( ) . | β (95% CI) . | value . | β (95% CI) . | value . |
Cross-sectional | |||||
Normoglycemia | 16,562 | Reference | Reference | ||
Prediabetes | 13,518 | 0.32 (0.20, 0.44) | <0.001 | 0.22 (0.10, 0.34) | <0.001 |
Diabetes | 1,149 | 2.40 (2.10, 2.71) | <0.001 | 2.01 (1.70, 2.32) | <0.001 |
HbA <7.0% | 671 | 1.81 (1.42, 2.21) | <0.001 | 1.43 (1.04, 1.83) | <0.001 |
HbA ≥7.0% to <8.0% | 303 | 2.62 (2.04, 3.20) | <0.001 | 2.19 (1.61, 2.77) | <0.001 |
HbA ≥8.0% | 175 | 4.29 (3.53, 5.05) | <0.001 | 3.90 (3.15, 4.66) | <0.001 |
HbA (continuous) | 0.95 (0.82, 1.08) | <0.001 | 0.77 (0.65, 0.90) | <0.001 | |
Longitudinal | |||||
Normoglycemia × time | 1,354 | Reference | Reference | ||
Prediabetes × time | 982 | −0.03 (−0.12, 0.07) | 0.597 | −0.03 (−0.12, 0.07) | 0.596 |
Diabetes × time | 78 | 0.27 (0.01, 0.53) | 0.045 | 0.27 (0.01, 0.53) | 0.045 |
HbA (continuous) × time | 0.13 (0.03, 0.23) | 0.012 | 0.13 (0.03, 0.23) | 0.012 |
. | . | BAG . | |||
---|---|---|---|---|---|
. | . | Basic adjusted . | Multiadjusted . | ||
Glycemic status . | Participants ( ) . | β (95% CI) . | value . | β (95% CI) . | value . |
Cross-sectional | |||||
Normoglycemia | 16,562 | Reference | Reference | ||
Prediabetes | 13,518 | 0.32 (0.20, 0.44) | <0.001 | 0.22 (0.10, 0.34) | <0.001 |
Diabetes | 1,149 | 2.40 (2.10, 2.71) | <0.001 | 2.01 (1.70, 2.32) | <0.001 |
HbA <7.0% | 671 | 1.81 (1.42, 2.21) | <0.001 | 1.43 (1.04, 1.83) | <0.001 |
HbA ≥7.0% to <8.0% | 303 | 2.62 (2.04, 3.20) | <0.001 | 2.19 (1.61, 2.77) | <0.001 |
HbA ≥8.0% | 175 | 4.29 (3.53, 5.05) | <0.001 | 3.90 (3.15, 4.66) | <0.001 |
HbA (continuous) | 0.95 (0.82, 1.08) | <0.001 | 0.77 (0.65, 0.90) | <0.001 | |
Longitudinal | |||||
Normoglycemia × time | 1,354 | Reference | Reference | ||
Prediabetes × time | 982 | −0.03 (−0.12, 0.07) | 0.597 | −0.03 (−0.12, 0.07) | 0.596 |
Diabetes × time | 78 | 0.27 (0.01, 0.53) | 0.045 | 0.27 (0.01, 0.53) | 0.045 |
HbA (continuous) × time | 0.13 (0.03, 0.23) | 0.012 | 0.13 (0.03, 0.23) | 0.012 |
Basic-adjusted models included age, sex, education, and socioeconomic status. Multiadjusted models additionally included cardiometabolic risk factor burden, smoking status, alcohol drinking, physical activity, and PRS AD .
Relationship between glycemic status and BAG. A : Least-squares means and SDs of BAG in participants with normoglycemia, prediabetes, and diabetes. B : The relationship between HbA 1c (as a continuous variable) and BAG is modeled using restricted cubic splines. The red line and red shaded area represent the least-squares means and 95% CIs of BAG as a function of baseline HbA 1c . Gray bars represent the distribution of HbA 1c in the study population. C : The relationship between glycemic status and changes in BAG is modeled using linear mixed-effects models. All models were adjusted for age, sex, education, SES, cardiometabolic risk factor burden, smoking status, alcohol drinking, physical activity, and PRS AD .
In an exploratory longitudinal analysis among the 2,414 participants (7.7%) who underwent two brain MRI scans, diabetes was associated with a 0.27-year annual increase in BAG ( Table 2 and Fig. 1C ). No significant relationship was detected between prediabetes and changes in BAG, although HbA 1c as a continuous variable was associated with a significant increase in BAG (β = 0.13 [95% CI 0.03, 0.23]).
In stratified analyses ( Fig. 2A and Supplementary Tables 10 and 11 ), the association between diabetes and higher BAG was more pronounced in men compared with women (β = 2.32 [95% CI 1.90, 2.74] vs. 1.51 [1.04, 1.99]) and people with a higher burden of cardiometabolic risk factors (0–1 risk factors: 1.91 [1.45, 2.36]; ≥2 risk factors: 2.20 [1.74, 2.66]). The same was true for prediabetes. Specifically, brain age was on average 0.75 years older than chronological age among men with prediabetes, compared with only 0.27 years older for women. Moreover, BAG rose to 2.63 years for men with diabetes compared with 1.76 years for women. Similarly, among individuals with two or more cardiometabolic risk factors, prediabetes and diabetes were associated with an average BAG of 1.32 and 3.08 years compared with 0.24 and 1.96 years, respectively, among their counterparts with a lower cardiometabolic risk factor burden.
Role of sex, cardiometabolic risk factor burden, and healthy lifestyle in the association between glycemic status and BAG. A : Least-squares means and SDs of BAG among participants with normoglycemia, prediabetes, and diabetes, stratified by sex and cardiometabolic burden. Significant interactions were detected between glycemic status and sex ( P < 0.001) and between glycemic status and cardiometabolic burden ( P < 0.001). Models were adjusted for age, education, SES, cardiometabolic risk factor burden, smoking status, alcohol drinking, physical activity, and PRS AD as well as sex or cardiometabolic risk factor burden, depending on the stratification factor. B : β-Coefficients for the joint effect on glycemic status and lifestyle on BAG. A significant interaction was detected between glycemic status and healthy lifestyle ( P = 0.04). Models were adjusted for age, sex, education, SES, cardiometabolic risk factor burden, and PRS AD . Note: The reference group was changed to (pre)diabetes and optimal lifestyle when assessing whether lifestyle significantly modified the (pre)diabetes-BAG association.
Significant interactions were detected between glycemic status and both sex and cardiometabolic burden with respect to BAG ( P < 0.001 for all).
In joint exposure analysis, an optimal healthy lifestyle (i.e., nonsmoking, no or light/moderate drinking, and high physical activity) significantly attenuated the association between diabetes and BAG ( Fig. 2B and Supplementary Table 12 ). Brain age was on average only 0.78 years older than chronological age among people with diabetes and an optimal lifestyle compared with 2.46 years older with a nonoptimal lifestyle. Therefore, healthy lifestyle was related to a 1.68-year reduction in BAG. More modest reductions in BAG were seen between individuals with normoglycemia and prediabetes and an optimal vs. nonoptimal lifestyle, respectively, although the difference for individuals with prediabetes was not statistically significant. A significant interaction was detected between glycemic status and lifestyle ( P = 0.04).
Sensitivity analyses are described in detail in the Supplementary Material . Overall, similar results were obtained when we repeated the analyses using BAG calculated based on brain age estimates from other candidate machine learning models ( Supplementary Table 13 ), using nonimputed data ( Supplementary Table 14 ), after additionally adjusting for brain MRI assessment center ( Supplementary Table 15 ), after excluding 42 participants with possible prodromal/undiagnosed dementia ( Supplementary Table 16 ), and after excluding 7,806 participants with possible cognitive impairment ( Supplementary Table 16 ). Moreover, 558 people with normoglycemia or prediabetes transitioned to diabetes during the ∼9-year period between baseline and the first MRI scan ( Supplementary Fig. 4 ), but results remained consistent using diabetes status defined at the time of this scan ( Supplementary Table 17 ).
In this large-scale neuroimaging study, diabetes and even prediabetes were related to significantly older brain age in relation to chronological age, and diabetes was further associated with significant widening of the gap between brain and chronological age over time. These associations were more pronounced in men and people with poorer cardiometabolic health but may be counteracted with a healthy lifestyle characterized by physical activity and abstention from smoking and heavy drinking.
Diabetes was associated with a BAG of 2.29 years in the current study, consistent with previous reports in which diabetes has been related to a BAG between 0.85 and 4.6 years ( 12–16 ). Drawing on the >2,000 participants in our study who underwent two brain MRI scans, we further determined that diabetes was associated with a 0.27-year annual increase in BAG over time, a compelling signal that diabetes is related not only to older brain age but also to an accelerated pace of brain aging. In line with this, a small study ( n = 25) exploring the longitudinal relationship between diabetes and brain aging reported that BAG widened by an estimated 0.2 years annually among people with diabetes ( 12 ).
Notably, whereas most previous studies estimated brain age used only T1-weighed imaging ( 12–15 , 17 , 18 ), ours leveraged information across six brain MRI modalities (T1-weighted imaging plus T2-FLAIR, T2*, diffusion MRI, resting-state fMRI, and task fMRI). A recent study also conducted using UK Biobank data concluded that whereas T1-weighted imaging is the MRI modality with the highest independent accuracy for brain age estimation, the best performance is achieved when multiple MRI modalities are combined ( 16 ).
Owing to our use of multimodal brain MRI data to estimate brain age, combined with the large sample size, we were able to detect a modest but highly statistically significant association between prediabetes ( P < 0.001) and higher BAG. In light of conflicting findings on the relationship between prediabetes and cognitive impairment and dementia ( 7–10 ), our results provide compelling evidence that prediabetes may accelerate brain aging during the very earliest stages of dementia development. Given the substantial and growing prevalence of prediabetes—estimated at ∼9% of the global population ( 35 )—even a modest effect of prediabetes on brain health could make a substantial difference at the population level. Encouragingly, prediabetes is a reversible state, and population-based studies have demonstrated that it is more common for people with prediabetes to regress to normoglycemia than progress to overt diabetes ( 36 , 37 ). Potential benefits for brain health could be yet another motivation to tighten glycemic control during this critical window.
Considering the heterogeneity of the diabetes population, we additionally investigated the role of a variety of other biological factors in the relationship between (pre)diabetes and brain age. In stratified analyses, the association between diabetes and higher BAG was more pronounced in men compared with women (2.63 vs. 1.76 years) and people with two or more as opposed to zero or one cardiometabolic risk factors (3.08 vs. 1.96 years). The prediabetes-BAG association was also stronger in men (0.75 vs. 0.27 years) and people with a higher cardiometabolic risk factor burden (1.32 vs. 0.24 years). Two previous studies have also reported a stronger relationship between brain age and diabetes among men ( 13 , 15 ), and the stronger diabetes-BAG association in in the context of a poorer cardiometabolic health is generally consistent with what has been observed for the diabetes-dementia association ( 2 , 3 ). These results highlight the complex interplay between hyperglycemia, sex, and cardiometabolic factors on brain health and underscore the importance of identifying populations that may benefit most from preventative interventions.
Although lifestyle behaviors such as a healthy diet, smoking/alcohol avoidance, physical activity, and social engagement have been associated with younger brain age ( 12 , 16 , 19 , 20 ), a relevant and so-far unexplored question is whether a healthy lifestyle can counteract the damaging influence of existing risk factors, such as diabetes, on brain aging. In our study, a lifestyle characterized by high physical activity and avoidance of smoking and heavy drinking significantly attenuated the association between diabetes and higher BAG. These results provide the encouraging suggestion that adoption of these healthy lifestyle behaviors could improve brain health among people with diabetes, although interventional studies are warranted to verify this hypothesis. Our findings are consistent with previous studies highlighting the mitigating role of lifestyle behaviors in the association between diabetes and dementia ( 38 , 39 ) and emphasize the significance of a healthy lifestyle for not only cardiometabolic health but also the brain.
There are several potential biological pathways through which (pre)diabetes may impact brain health. Hyperglycemia, the defining pathophysiological feature of diabetes, can promote endothelial dysfunction, oxidative stress, systemic inflammation, and the accumulation of advanced glycation end products ( 1 ). Together these contribute to disruption of blood-brain barrier permeability (exposing the brain to potentially toxic substances, leading to abnormal neuronal activity), demyelination and loss of axons (leading to brain atrophy and disruptions in neurotransmitter signaling), and alterations in Ca 2+ signaling (leading to excitotoxicity and disruptions in gene expression) ( 1 ). Additionally, the micro- and macrovascular complications of diabetes can contribute to brain atherosclerosis and cerebrovascular pathologies that may lower the threshold for neurodegeneration ( 1 ). Finally, the insulin resistance that characterizes diabetes has been linked to AD-related processes, including amyloid-β generation, τ-hyperphosphorylation, and impaired amyloid-β clearance ( 1 ). A healthy lifestyle may enhance cardiovascular and metabolic health, thereby minimizing the impact of hyperglycemia, insulin resistance, and vascular damage.
Strengths of this study include the large sample size and the use of multimodal brain MRI data to estimate brain age. However, some limitations should be acknowledged. First, healthy volunteer bias in the UK Biobank could limit the generalizability of our findings and may have contributed to an underestimation of the observed associations. Selection bias may be stronger in our sample because it was restricted to participants who underwent a brain MRI scan, a comparatively younger and more cardiometabolically healthy subgroup ( Supplementary Table 9 ).
Second, diet could not be considered in the healthy lifestyle construct due to a high proportion of missing data (35%); additional analyses integrating diet into the optimal lifestyle measure are presented in Supplementary Table 18 .
Third, there is the possibility of reverse causality insofar as having an older brain may contribute to the development of (pre)diabetes by making it more difficult to manage medical conditions and adhere to a healthy lifestyle. However, results remained consistent in sensitivity analyses excluding participants with possible cognitive impairment or prodromal dementia ( Supplementary Table 16 ), suggesting that reverse causality is unlikely to have a major impact on our findings.
Additionally, misclassification of baseline glycemic status may have occurred because HbA 1c is less sensitive than alternative measures such as fasting plasma glucose or the oral glucose tolerance test ( 40 ). Moreover, because HbA 1c was measured only at baseline, we could not assess changes in glycemic control or progression/reversion of prediabetes in relation to BAG.
Finally, longitudinal data were available for only 2,414 participants (7.7%). Repeat collection of brain MRI scans is still ongoing, presenting an opportunity for future studies to explore the longitudinal relationship between (pre)diabetes and brain aging in greater detail.
In conclusion, the current study provides evidence that hyperglycemia—including diabetes and even prediabetes—may contribute to accelerated brain aging. These associations were more pronounced in men and people with poorer cardiometabolic health but were attenuated with a healthy lifestyle characterized by physical activity and abstention from smoking and heavy drinking. Our findings highlight diabetes and prediabetes as ideal targets for lifestyle-based interventions to promote brain health.
This article contains supplementary material online at https://doi.org/10.2337/figshare.26417971 .
G.P. and W.X. are co-last authors.
Acknowledgments. The authors would like to express their gratitude to the UK Biobank study participants and the staff involved in the UK Biobank data collection and management.
Funding. A.D. received funding from Alzheimerfonden (AF-993470) and Demensfonden. W.X. received grants from the Swedish Research Council (No. 2021-01647), the Swedish Council for Health, Working Life and Welfare (No. 2021-01826), Alzheimerfonden, and the Karolinska Institutet Board of Research. G.P. received funding from the Riksbankens Jubileumsfond (No. P20-0779) and Swedish Research Council (No. 2019-02804). M.G.M. received funding from the Marianne and Markus Wallenberg Foundation (No. 2020.0013) and Swedish Research Council (No. 2021-02046).
Duality of Interest. No potential conflicts of interest relevant to this article were reported.
Author Contributions. A.D. conducted the statistical analyses, performed the literature search, and drafted the first version of the manuscript. A.D. and W.X. contributed to the conception and design of the study. J.W. and H.H. created the brain age variable. J.W., H.H., M.M.D., S.S., M.G.-M., G.P., and W.X. interpreted the data and provided critical revisions to the manuscript. All authors made a significant contribution to finalizing the manuscript and approved the final version for publication. A.D. is the guarantor of this work and, as such, had full access to all the data in the study and takes responsibility for the integrity of the data and the accuracy of the data analysis.
Handling Editors. The journal editors responsible for overseeing the review of the manuscript were Elizabeth Selvin and Alka M. Kanaya.
Sign In or Create an Account
IMAGES
VIDEO
COMMENTS
It is a crucial component in the care of people with type 1 diabetes, and it becomes increasingly important in the care of patients with type 2 diabetes who have a constellation of comorbidities, all of which must be managed for successful disease outcomes.
Abstract. Unhealthy behaviours, including diet and physical activity, coupled with genetic predisposition, drive type 2 diabetes (T2D) occurrence and severity; the present review aims to summarise the most recent nutritional approaches in T2D, outlining unmet needs. Guidelines consistently suggest reducing energy intake to counteract the ...
Type 2 diabetes (T2D) is a rapidly growing epidemic, which leads to increased mortality rates and health care costs. Nutrients (namely, carbohydrates, fat, protein, mineral substances, and vitamin), sensing, and management are central to metabolic homeostasis, ...
Introduction Prevalence of type 2 diabetes (T2D) is increasing rapidly and lifestyle interventions to reverse diabetes are seen as a possible solution to stop this trend. New practice-based evidence is needed to gain more insight in the actual, and above all scientific, basis for these claims. Methods This observational study with a pretest post-test design aimed to pilot a 6-month ...
The nutrition Consensus Report (1) and four featured papers (2 - 5) in the special section on nutrition in this issue of Diabetes Care focus on nutrition therapy and medical nutrition therapy (MNT) in the management and prevention of diabetes.
Backgrounds Although the significance of diet in preventing or managing diabetes complications is highlighted in current literature, there is insufficient evidence regarding the correlation between nutrient patterns and these complications. The objective of this case-control study is to investigate this relationship by analyzing the dietary intake of nutrients in participants with and without ...
This randomized clinical trial compared the effects of an intensive lifestyle intervention vs standard care on glycemic control and medication reduction among participants with non-insulin-dependent type 2 diabetes.
Background: Although the body of evidence indicates clear benefits of dietary modifications for prevention of type-2 diabetes mellitus (T2DM), it may be difficult for healthcare providers to recommend which diet interventions or dietary factors are appropriate ...
Abstract Objective: The objective of this Expert Consensus Statement is to assist clinicians in achieving remission of type 2 diabetes (T2D) in adults using diet as a primary intervention. Evidence-informed statements agreed upon by a multi-disciplinary panel of expert healthcare professionals were used.
Nutrition therapy provided by dietitians was associated with better clinical parameters of type 2 diabetes, including clinically significant improved glycaemic control, across diverse multiethnic patient groups from all six inhabited continents. This conclusion should be reflected in clinical guidel …
This Review highlights the evidence from clinical trials that diet-induced weight loss interventions can be used to treat type 2 diabetes mellitus. The composition of the diet and the importance ...
Abstract Purpose: Type 2 diabetes mellitus represents a significant health problem. Many studies have reported that intensive nutritional intervention by itself or in addition to medications is the best method to improve glycaemic control in type 2 diabetes mellitus.
Introduction A wealth of evidence supports short-term efficacy of lifestyle interventions in type 2 diabetes (T2D). However, little is known about long-term effects of lifestyle interventions in real-life settings. Methods This observational, single-arm study evaluated long-term impact of 'Voeding Leeft: Reverse-Diabetes2-Now', a 6-month multicomponent lifestyle programme, on glycaemic ...
Common ground on dietary approaches for the prevention, management, and potential remission of type 2 diabetes can be found, argue Nita G Forouhi and colleagues Dietary factors are of paramount importance in the management and prevention of type 2 diabetes. Despite progress in formulating evidence based dietary guidance, controversy and confusion remain. In this article, we examine the ...
Questions What type of lifestyle modification should be prescribed for L.H.? What was the outcome after 2 months of treatment? Can risk factors for diabetes be reversed in a short period of time with intense exercise and diet?
To investigate the effects of combined diet-and-exercise interventions in patients with type 2 diabetes mellitus (T2DM). ... Further studies are necessary to compare the effectiveness of other diet-and-exercise combinations than those analyzed in this review and to clarify the role of other interacting factors, e.g. patients' medication. ...
Comprehensive lifestyle interventions are effective strategies to prevent type 2 diabetes among at-risk populations in LMICs. The heterogeneity identified in our results should be considered when using these interventions to address the onset of type 2 diabetes.
Roy Taylor and colleagues explain how type 2 diabetes can be reversed by weight loss and avoidance of weight regain Type 2 diabetes mellitus was once thought to be irreversible and progressive, but a series of clinical studies over the past 12 years have clarified the mechanisms that cause the disease. We now know that the processes that cause type 2 diabetes can be returned to normal ...
Aims/hypothesis We conducted the largest and longest clinical trial comparing a whole-food, plant-based intervention with standard medical care (SMC) in individuals with type 2 diabetes. Methods We randomised (parallel-arm; computerised 1:1 randomisation ratio) 169 adults aged 18-75 years with type 2 diabetes in the Marshall Islands to an intensive whole-food, plant-based intervention with ...
Abstract Type 2 diabetes mellitus (T2DM) accounts for >90% of the cases of diabetes in adults. Resistance to insulin action is the major cause that leads to chronic hyperglycemia in diabetic patients. T2DM is the consequence of activation of multiple pathways and factors involved in insulin resistance and β-cell dysfunction.
The comparative effectiveness of glucose-lowering medications for use with metformin to maintain target glycated hemoglobin levels in persons with type 2 diabetes is uncertain. In this trial involv...
Biotechnol., 15 September 2020. The Role of Nutrition in the Prevention and Intervention of Type 2 Diabetes. Type 2 diabetes (T2D) is a rapidly growing epidemic, which leads to increased mortality rates and health care costs. Nutrients (namely, carbohydrates, fat, protein, mineral substances, and vitamin), sensing, and management are central to ...
Studies such as PLATEDIAN (Telemedicine on Metabolic Control in Type 1 Diabetes Mellitus Andalusian Patients), provide positive results and conclude that a telemedicine intervention has a similar impact to the traditional office-based intervention in terms of glycemic control, acute complications, and perceived QOL (Ruiz de Adana et al., 2020).
Study with Quizlet and memorize flashcards containing terms like Mandana is given a prescription for metformin, a biguanide medication prescribed for management of type 2 diabetes. What best describes how metformin works to control blood glucose? a. It delays carbohydrate digestion. b. It increases urinary glucose excretion. c. It delays stomach emptying. d. It improves insulin secretion. e ...
Aims and objectives To assess the effect of an instructional WhatsApp group on self-care and HbA1c levels among female patients with type 2 diabetes mellitus (T2DM). Background T2DM is a chronic disease that requires effective self-care. WhatsApp is a free application that can be effectively used for patient education. Design This study used a quasi-experimental design. Methods A convenience ...
Background Type 2 diabetes mellitus has been identified as one of the most challenging chronic illnesses to manage. Since the management of diabetes is mainly accomplished by patients and families, self-management has become the mainstay of diabetes care. However, a significant proportion of patients fail to engage in adequate self-management.
Introduction Gestational diabetes mellitus (GDM) is one of the most common medical complications of pregnancy. Glycaemic control decreases the risk of adverse pregnancy outcomes for the affected pregnant individual and the infant exposed in utero. One in four individuals with GDM will require pharmacotherapy to achieve glycaemic control. Injectable insulin has been the mainstay of ...
Type 2 diabetes (hereafter, diabetes) is a well-established risk factor for cognitive impairment and has been associated with approximately double the risk of dementia ().In brain MRI studies, diabetes has been related to global brain atrophy, increased burden of small-vessel disease, and microstructural lesions before the onset of cognitive symptoms ().