Type 2 Diabetes
Conditions
Keywords
Mobile health, Mhealth, T2.coach, Smartphone app
Brief summary
In this project, the investigators will evaluate the efficacy of a novel approach to personalizing behavioral interventions for self-management of type 2 diabetes (T2DM) to individuals' behavioral and glycemic profiles discovered using computational learning and self-monitoring data. This study is a two-arm randomized controlled trial with n=280 participants recruited from the participating Federally Qualified Health Centers (FQHCs). The participants will be randomly assigned to the intervention group and the usual care (control) group with 1-1 allocation ratio. Half of the participants (n=140) will be randomly assigned to a usual care (control) group. Both groups will receive standard diabetes education at their respective FQHC site. In addition, the experimental group will receive instructions to use T2.coach for a minimum of 6 months.
Detailed description
One of the main difficulties in managing diabetes is that each affected individual requires personally tailored combination of diet, exercise, and medication to effectively control their blood sugar. Rather than strictly following a doctor's prescription, individuals need to carefully examine their lifestyle choices and their impact on their health. Independent learning, experimentation and problem solving become of great importance. However, they can be challenging for individuals with diabetes. In this project, the investigators will refine and evaluate a novel intervention for diabetes self-management that uses computational analysis of self-monitoring data to help individuals with type 2 diabetes identify what daily activities, including consumption of meals, physical activity, and sleep, have impact on blood glucose levels, and suggest modifications to these daily activities to improve blood glucose levels. Growing evidence highlights significant differences in glycemic function and cultural, social, and economical circumstances of individuals with type 2 diabetes (T2DM) that impact their self-management. Precision medicine strives to personalize medical treatment to an individual's genetic makeup, computationally discovered clinical phenotypes and lifestyle. Studies showed the benefits of tailoring not only medical treatment, but also behavioral interventions. Yet, currently, personalization of self-management in T2DM requires each individual to engage in discovery, reflection, and problem-solving-critical but cognitively demanding activities-or to rely on their healthcare providers. Both of these may present considerable barriers to individuals from medically under-served low income communities. Mobile health (mHealth) solutions in T2DM bring promise of reaching wider populations in need of self-management; however, few such solutions provide assistance with personalizing self-management behaviors. Ongoing efforts on personalizing behavioral interventions outside of T2DM focus on tailoring behavior modification techniques to individuals' psycho-social characteristics, such as self-efficacy ), and tailoring delivery of intervention to individuals' context rather than on personalizing self-management strategies. The ongoing focus of this research is on developing informatics interventions for diabetes self-management, with a specific focus on discovery with self-monitoring data and on problem-solving for improving glycemic control. In the proposed research the investigators introduce T2.coach, an mHealth intervention that uses computational analysis of self-monitoring data to identify behavioral patterns associated with poor glycemic control and formulate personalized behavioral goals for changing problematic behaviors. This study will evaluate T2.coach's efficacy in a two-arm RCT with stratified randomization conducted with Clinical Directors Network (CDN), a well-recognized primary care practice-based research network (PBRN) of Federally Qualified Health Centers (FQHCs), and Agency for Healthcare Research and Quality (AHRQ)-designated Center of Excellence (P30) for Practice-based Research and Learning.
Interventions
BEHAVIORALT2.coach
T2.coach is a smartphone app for low-burden capture of diet and blood glucose (BG) levels and for reviewing past records, integrated with FitBit for captured of physical activity and sleep. All captured data are sent to the computational inference engine that uses machine learning methods and expert system to formulate personalized behavioral goals. Examples of behavioral goals include the following: "For high carbohydrate breakfasts, reduce your carbs to be about 1 carb choice. Examples of 1 carb choice are 1 slice of whole wheat toast, 1 cup of oatmeal, or 1 apple." The T2.coach chatbot companion uses text messages to help individuals set goals that are consistent with evidence based guidelines for diabetes self-management, inferences on data captured with T2.coach, and their own preferences, as well as send individuals goal reminders and prompts for reflection on goal achievement.
Sponsors
Columbia University
Clinical Directors Network
University of Colorado, Denver
National Institute of Diabetes and Digestive and Kidney Diseases (NIDDK)
Study design
Allocation
RANDOMIZED
Intervention model
PARALLEL
Primary purpose
OTHER
Masking
NONE
Masking description
Because of the nature of the intervention (smartphone app), masking is not possible.
Intervention model description
Two-arm RCT with 1:1 randomization at participant level, with stratified randomization to balance by clinical site, sex, and language, evaluate the efficacy of the T2.coach intervention
Eligibility
Sex/Gender
ALL
Age
18 Years to 65 Years
Healthy volunteers
No
Inclusion criteria
* Patient of the health center for ≥ 6 months and a diagnosis of T2DM * HbA1c ≥ 8.0, * Aged 18 to 65 years * Attends diabetes education program at the health center * Owns a basic mobile phone * Proficient in either English or Spanish
Exclusion criteria
* Pregnant * Presence of severe cognitive impairment (recorded in patient chart), * Existence of other serious illnesses (e.g. cancer diagnosis with active treatment, advanced stage heart failure, dialysis, multiple sclerosis, advanced retinopathy, recorded in patient chart), * Plans for leaving the FQHC in the next 12 months, * Participation in the previous trial of diabetes self-management technologies
Design outcomes
Primary
| Measure | Time frame | Description |
|---|---|---|
| Mean HbA1c Value | Baseline, 6 months, 12 months | The main outcome is the mean Hemoglobin A1c at 12 months. In the statistical analysis, we examine difference in mean HbA1c between the study arms at baseline, 6 months, and 12 months. |
Secondary
| Measure | Time frame | Description |
|---|---|---|
| SCA-I Score | Baseline, 6 months, 12 months | Diabetes Self-Care Inventory (SCA-I) is a 15-item 5-point Likert scale (1-never engage; 5-always engage) for measuring different aspects of diabetes self-care. The final score ranges from 1 (lowest) to 5 (highest) with a higher score indicating better self-care (better outcome). To account for missing values, the final score was normalized to a 1-100 scale with a higher score indicating better self-care. All analysis was conducted with normalized scores. |
| DSES Score | Baseline, 6 months, 12 months | Diabetes Self-Efficacy Scale (DSES) is a 15-item 10-point Likert scale (1-not at all confident; 10-totally confident) that measures the belief that one can self-manage one's own health, adapted to diabetes. Final scores are averaged, and the total score ranges from 1 (lowest) to 10 (highest) with a lower score indicating poor self-efficacy (worse outcome). |
| PAID Score | Baseline, 6 months, 12 months | Problem Areas in Diabetes (PAID) is a 20-item 5-point Likert scale (0=not a problem; 4=very serious problem) that measures the emotional aspect of living with diabetes. The final score ranges from 0 (lowest) to 80 (highest), with a higher score indicating greater emotional discomfort (worse outcome). |
Countries
United States
Contacts
PRINCIPAL_INVESTIGATOROlena Mamykina, PhD
Columbia University
Participant flow
Recruitment details
Recruitment occurred from 2020-2023. Initially, participants were recruited in person at two FQHCs via waiting-room outreach and staff referrals. Recruitment paused in March 2020 due to COVID-19. By August 2020, the study transitioned to fully virtual procedures, expanding to four additional sites. The team used secure EHR access to identify eligible participants and conducted outreach by phone.
Pre-assignment details
7,616 participants were assessed for eligibility. Of these, 4,552 did not meet inclusion criteria, 775 declined to participate, and 1,989 were excluded for other reasons. The remaining 300 met the inclusion criteria, were consented and randomized into study arms. Participants were randomized on an individual level. No units other than participants were randomized.
Baseline characteristics
| Characteristic | — |
|---|---|
| Age, Continuous | 48.7 years STANDARD_DEVIATION 9.1 |
| Annual Income <10,000 USD | 57 Participants |
| Annual Income >=10,000 USD | 90 Participants |
| Annual Income Unreported | 1 Participants |
| Body Mass Index (BMI) | 33.3 kg/m2 STANDARD_DEVIATION 7.9 |
| Education Did not complete high school | 39 Participants |
| Education High school and above | 100 Participants |
| Employment Employed | 88 Participants |
| Employment Unemployed | 56 Participants |
| Ethnicity (NIH/OMB) Hispanic or Latino | 92 Participants |
| Ethnicity (NIH/OMB) Not Hispanic or Latino | 117 Participants |
| Ethnicity (NIH/OMB) Unknown or Not Reported | 0 Participants |
| Insurance Medicaid/Medicare/Dual Enrollment/none | 21 Participants |
| Insurance Other | 156 Participants |
| Language English only | 122 Participants |
| Language Other (including bilingual) | 37 Participants |
| Language Spanish only | 50 Participants |
| Marital status Married | 133 Participants |
| Marital status Other | 73 Participants |
| Race (NIH/OMB) American Indian or Alaska Native | 4 Participants |
| Race (NIH/OMB) Asian | 4 Participants |
| Race (NIH/OMB) Black or African American | 50 Participants |
| Race (NIH/OMB) More than one race | 5 Participants |
| Race (NIH/OMB) Native Hawaiian or Other Pacific Islander | 0 Participants |
| Race (NIH/OMB) Unknown or Not Reported | 65 Participants |
| Race (NIH/OMB) White | 37 Participants |
| Sex: Female, Male Female | 88 Participants |
| Sex: Female, Male Male | 61 Participants |
Adverse events
| Event type | EG000 affected / at risk | EG001 affected / at risk |
|---|---|---|
| deaths Total, all-cause mortality | 3 / 148 | 3 / 152 |
| other Total, other adverse events | 1 / 148 | 0 / 152 |
| serious Total, serious adverse events | 4 / 148 | 4 / 152 |
Outcome results
None listed