This study was approved by the University of Michigan Institutional Review Board (HUM00196546). The protocol was designed according to the Standard Protocol Items: Recommendations for Interventional Trials 2013 (SPIRIT) [20]. SPIRIT subheadings are used throughout the manuscript and noted in curly brackets. Trial registration: clinicaltrials.gov, NCT05235425. Registered February 11, 2022, https://clinicaltrials.gov/ct2/show/NCT05235425.
Study setting {9}
The intervention will be delivered remotely by Zoom™, a HIPAA-compliant video conferencing software. All data will be collected in the United States.
Eligibility criteria {10}
Inclusion criteria include: (1) aged 21-75 years; (2) overweight, defined as BMI ≥ 25 kg/m2 or ≥ 23 kg/m2 if of Asian descent; (3) HbA1c of 5.7% to 6.4% (as measured on baseline blood draw); (4) willingness and ability to participate in group-based, online sessions using audio and video; (5) ability to engage in at least light physical activities, such as walking; (6) willingness to follow either prescribed diet through random assignment; (7) willingness to self-monitor weight, dietary intake, and physical activity minutes; and (8) physician approval to participate.
Exclusion criteria include: (1) inability to read, write, or speak English; (2) inability to provide written informed consent; (3) history of type 1 diabetes or type 2 diabetes; (4) pregnant or planning to become pregnant during the intervention period; (5) breastfeeding; (6) use of anti-obesity medications; (7) participation in another weight loss program or intervention; (8) previous bariatric surgery or plan to have bariatric surgery during the study period; (9) use of glucose-lowering medications other than metformin; (10) blood disorders that require transfusion or phlebotomy, including anemia, hemoglobinopathies, or polycythemia; (11) adherence to a vegan or vegetarian diet; (12) adherence to a very low-carbohydrate diet; (13) difficulty chewing or swallowing; (14) inability to control foods that are purchased, prepared, or served; (15) untreated eating disorder or unstable serious mental illness (such as depression with suicidal ideation, bipolar or schizophrenia with psychosis); (16) abnormal baseline labs, including triglycerides ≥ 600 milligrams per deciliter (mg/dL), thyroid stimulating hormone (TSH) < 0.4 milli-international units per liter (mIU/L) or > 5.0 mIU/L, serum potassium < 3.6 millimoles per liter (mmol/L) or > 5.2 mmol/L; (17) chronic kidney disease ≥ stage 4; (18) use of loop diuretic equivalent to furosemide 20 mg per day or greater; (19) warfarin use; (20) chronic oral corticosteroid use; (21) any condition for which the study team deems participation to be unsafe or inappropriate.
Recruitment {15}
Based on study inclusion criteria, we will use DataDirect [21], a self-serve tool that enables access to clinical data on more than 4 million unique patients within the University of Michigan Health System, to identify potentially eligible individuals. Potentially eligible individuals will be sent a letter with information about the study by postal mail or e-mail. The letter will include a web address to a website with trial information, study team contact information, and a link to an online screening survey. The screening survey will be used to obtain sociodemographic characteristics, HbA1c, weight, height, and medication use. If we are not able to meet our recruitment target with this initial strategy, we may recruit by: (1) direct telephone outreach; (2) referral by primary care providers; and/or (3) posted flyers and advertisements on UMHealthResearch.org [22] or social media platforms, such as Facebook™.
Who will take informed consent? {26a}
Consent for baseline blood draw: Potentially eligible participants who complete the online screening survey will receive an email that includes a link to a pdf blood draw consent form and a link to an orientation video explaining the study goals and procedures. Individuals will be asked to watch the video and then answer a brief questionnaire to assess their understanding of the study and to provide consent for a blood draw, which will include HbA1c. Consent will provided via REDCap [23] and potential participants are encouraged to contact the study team with any questions. Individuals that provide baseline blood draw consent will receive information by email regarding participating laboratory locations and hours.
Consent for full study participation: Individuals who complete the blood draw and have an HbA1c of 5.7% to 6.4%, triglycerides ≤ 600 mg/dL, TSH within normal range, and serum potassium within normal range will be informed of their study eligibility by email and invited to schedule a virtual visit with a trained study team member to complete study enrollment and consent processes. During this virtual visit, a study team member will review intervention procedures (e.g., randomization process), discuss expectation of participants (e.g., class attendance, completion of assessments), and answer any questions. Individuals that remain interested in study participation will be asked to provide written informed consent and complete a baseline survey; these will be administered via REDCap [23].
Additional consent provisions for collection and use of participant data and biological specimens {26b}
Participants are required to consent to be recorded to participate in this trial.
Explanation for choice of comparators {6b}
The NDPP is a low-fat, calorie restricted dietary intervention; it is the standard lifestyle change approach for T2DM prevention. Dietary carbohydrate restriction is effective for T2DM management, but less is known about its role in T2DM prevention. We will test the effectiveness of the NDPP versus a very low-carbohydrate adaption of the program.
Intervention description {11a}
All participants will be encouraged to engage in lifestyle changes to support weight loss and T2DM prevention. Both NDPP and VLC-NDPP groups will include 16 core sessions over 4 months followed by 8 monthly sessions. All sessions will last an hour and will be held on Zoom, in groups of approximately 15-18 participants. Sessions will be delivered online by coaches from the National Kidney Foundation of Michigan (NKFM), our local leader in the delivery of CDC-recognized DPPs [24] and our community partner in prior pilot studies [17,25]. Session topics relate to dietary change, meal planning, grocery shopping, increasing physical activity, managing stress, and supporting self-efficacy for initiating and sustaining lifestyle change. The VLC-NDPP differs from the NDPP primarily in terms of its dietary advice; the content of the non-dietary sessions is minimally altered. Participants in both groups will be asked to self-weigh at least once weekly, maintain food logs, and track minutes of physical activity each week; these data will be collected via online survey prior to each session and shared with the coach. The coach will discuss any issues during group sessions and may contact individual participants, as needed.
National Diabetes Prevention Program (NDPP): Participants will receive lifestyle change recommendations in accordance with the CDC’s NDPP, which teaches individuals to follow a low-fat, calorie-restricted diet and to engage in at least 150 minutes of moderate-intensity physical activity per week. With regards to dietary advice, participants are taught that fat contains more calories per gram than protein or carbohydrate and a reduction in dietary fat thus potentiates weight loss. Participants are encouraged to eat whole grains, vegetables, fruit, dairy, and lean protein throughout the day with the following explanation: “Spread your calories out through the day. Doing so helps keep you from getting too hungry and losing control. Eat 3 meals each day and 1 or 2 healthy snacks.” The NDPP includes handouts for participants and a detailed guide for coaches [26].
Very low-carbohydrate adaptation of the NDPP (VLC-NDPP): Participants will receive lifestyle change recommendations in accordance with our adapted curriculum, which replaces low-fat dietary recommendations with very low-carbohydrate dietary recommendations. Program topics related to physical activity or behavioral change techniques were minimally altered. Table 1 summarizes the VLC-NDPP session topics and the degree to which components, such as diet, physical activity, and behavior change strategies were modified from the NDPP.
The VLC-NDPP teaches participants to reduce dietary carbohydrates to 20-35 grams of non-fiber carbohydrates per day. Participants will be advised to reduce their carbohydrate intake back to their prior tolerated level if their weight increases. Participants will be encouraged to eat when they are hungry and stop when they are full, to keep their protein levels similar to baseline, presuming they are meeting the recommended dietary requirement [27], and to derive their remaining calories from fat. In general, participants are taught to avoid foods such as potatoes, rice, pasta, bread, donuts, and sugar-sweetened beverages; they are instructed to consume foods such as meat, fish, tofu, tempeh, full-fat dairy, eggs, fats, nuts, seeds, berries, and leafy or other low-carbohydrate vegetables. We provide information about low-carbohydrate versions of foods, such as eggplant-based lasagna, spiralized zucchini pasta, and cauliflower rice.
Table 1. Very Low-Carbohydrate-National Diabetes Prevention Program topics and summary of changes made to the Center for Disease Control and Prevention’s National Diabetes Prevention Program.
General topics and timing
|
Changes from standard NDPP
|
Session titles
|
Introduction
(month 1)
|
Adds low-carbohydrate meal plan goals
|
· Welcome to the DPP
|
Nutrition (months 1-12)
|
Completely altered; now teaches how to follow a very low-carbohydrate meal plan
|
· Be a Carbohydrate Detective
· What to Eat
· Cooking and Shopping
· Four Keys to Eating Out
· Welcome to the Post-Core Phase
· Fat: Saturated, Unsaturated, and Trans Fats
· Handling Holidays, Vacations, and Special Events
· Grocery Store Information
· Revisiting Recipes and Cooking
|
Physical Activity
(months 1-12)
|
Adds strategies for being physically active when following a very low-carbohydrate meal plan
|
· Move Those Muscles
· Being Active: A Way of Life
· Jump Start Your Activity Plan
· Staying on Top of Physical Activity
· Being Physically Active Together
|
Behavior Change Strategies
(months 1-12)
|
Slightly adapted to include very low-carbohydrate food recommendations
|
· Challenges and Support
· Take Charge of What’s Around You
· Problem Solving
· Talk Back to Negative Thoughts
· The Slippery Slope of Lifestyle Change
· Make Social Cues Work for You
· You Can Manage Stress
· Ways to Stay Motivated
· Stress and Time Management
· Long-Term Maintenance and Looking Forward
|
After the program’s core phase, if participants have reached their weight loss goal and desire to liberalize their carbohydrate intake, they are taught to gradually increase their carbohydrate consumption by adding no more than 5 net grams of carbohydrates to their daily goal per week. For example, an individual consuming 35 grams of non-fiber carbohydrates per day may increase to 40 grams of non-fiber carbohydrates per day and continue that goal for a minimum of one week before making additional changes.
Training coaches
All coaches will receive standard Lifestyle Coach Training according CDC guidelines [28] to ensure consistency in the skills necessary to run group sessions. Both the NDPP and VLC-NDPP curricula include handouts for participants, as well as a detailed guide for coaches. VLC-NDPP coaches will receive additional training by study team members to ensure adequate content knowledge regarding dietary carbohydrate restriction. Following the training period, VLC-NDPP coaches will complete a knowledge assessment quiz. Individuals who score low on the quiz will be required to repeat training.
Ensuring intervention fidelity
All sessions will be recorded. We will randomly select 20% of sessions, stratified by intervention phase (either core phase or maintenance phase), to assess curriculum fidelity. Study staff will review recordings of the group sessions and assess whether each class meets content objectives, including the extent to which the coaches: (a) covered curriculum content, (b) maintained appropriate control over the pacing of the session, (c) troubleshot individual participant challenges effectively, (d) conveyed enthusiasm for the topics, (e) described the topics clearly, and (f) described the topics accurately. Coaches will receive feedback for classes with < 90% adherence to the protocol or low scores in any of these areas; additional training will be provided, as needed.
Coaches will report on every group session by completing an online survey immediately following the session. This survey will ask coaches to report any issues with participant adherence, adverse events, or side effects. Coaches will also be asked to self-evaluate their adherence to the curriculum content.
Criteria for discontinuing or modifying allocated interventions {11b}
Study physicians will use their clinical judgement to determine if participants should be removed from the trial due to the development of exclusionary diagnoses or addition of exclusionary medications (e.g., glucose-lowering medication other than metformin). We do not anticipate modifying or discontinuing modifying allocated interventions.
Strategies to improve adherence to interventions {11c}
The books "Calorie King"[29] and "Dana Carpender’s Keto Fat Gram Counter"[30] will be sent to NDPP and VLC-NDPP participants, respectively. All participants without access to a home scale will receive one by mail at the start of the trial.
Relevant concomitant care permitted or prohibited during the trial {11d}
Participants are encouraged to continue with their health care as normal throughout the trial. All blood test results will be sent to participants and their PCPs. Enrollment in other nutrition, weight, or diabetes related trials or programs is prohibited. It is expected that some participants from either group may advance from prediabetes to T2DM during the study period. If PCPs decide to start their patient on glucose-lowering medications, those participants will be allowed to remain in the trial, except for sodium-glucose cotransporter-2 (SGLT-2) inhibitors. Participants who start an SGLT-2 inhibitor will be removed from the trial due to concerns for increased risk of euglycemic diabetic ketoacidosis when used with very low-carbohydrate diets.
Provisions for post-trial care {30}
We do not plan for any post-trial care. Any adverse events, reactions, or symptoms reported by participants during the trial will be addressed immediately. We do not anticipate the need for post-trial care or follow-up.
Plans for assessment and collection of outcomes {18a}
Pre-enrollment
The following laboratory testing will be obtained when individuals present for their screening (baseline) HbA1c: (1) lipids; (2) serum insulin; (3) fasting plasma glucose; (4) high-sensitivity C-reactive protein (hsCRP); (5) fasting β-hydroxybutyrate (ketones); (6) a comprehensive metabolic panel; and (7) TSH.
Baseline
Consented individuals will be asked to schedule an in-person appointment at Michigan Diabetes Research Center’s Clinical Research Unit to complete the following procedures: (1) height, hip, and waist measurements; (2) body weight measurement; and (3) continuous glucose monitor (CGM) placement. Participants will wear their CGM (which will not provide feedback to participants) for two weeks before mailing it back to the study team. Following consent and prior to the start of the intervention, participants will be contacted by phone for one unannounced 24-hour dietary recall. Within the 2 weeks before classes begin, participants will be asked to complete an online survey; survey items will assess self-reported symptoms (using measures adapted from the literature [39]), quality of life [40], and medication use.
4 months
The following measures will be collected at 4 months: (1) HbA1c; (2) lipids; (3) serum insulin; (4) fasting plasma glucose; (5) hsCRP; (6) fasting β-hydroxybutyrate; (7) comprehensive metabolic panel; and (8) body weight as measured on home scale. Participants will be asked to complete an online survey (the same one they completed at baseline with added items to assess program satisfaction) and will complete one unannounced 24-hour dietary recall.
12 months
The following laboratory measures will be collected at 12 months: (1) HbA1c; (2) lipids; (3) serum insulin; (4) fasting plasma glucose; (5) hsCRP; (6) fasting β-hydroxybutyrate; and (7) comprehensive metabolic panel. Participants will also complete an in-person appointment at Michigan Diabetes Research Center’s Clinical Research Unit for the following procedures: (1) height, hip, and waist measurements; (2) body weight measurement; and (3) CGM placement. Participants will wear their CGM (which will not provide feedback to participants) for two weeks before mailing it back to the study team. Participants will be asked to complete an online survey (the same one they completed at baseline with added items to assess program satisfaction) and will complete one unannounced 24-hour dietary recall.
Outcomes {12}
Primary outcome measure
Change in glycemic control: HbA1c is the most widely accepted measure of overall glycemic control in clinical care and predicts the risk of microvascular diabetes complications in people with type 2 diabetes [31]. HbA1c levels will be measured at 0, 4, and 12 months, and the change in HbA1c from baseline to 12 months will be the primary study outcome.
Secondary outcome measures
Mean change in body weight: Participants’ body weight will be measured using a calibrated scale at 0 and 12 months. Participants will also be advised to self-weigh at least once weekly using a home scale (provided by the study team, if necessary). Participants will report home weight data to their coach through a weekly online survey prior to each session. We will calculate average body weight change at 4 and 12 months compared to baseline.
Mean percent body weight loss: We will calculate percent body weight loss (100-(weight at baseline/weight at 4 or 12 months multiplied by 100)).
Percentage of participants who achieve ≥5% and ≥10% body weight loss: We will calculate the percentage of participants who achieve ≥5% and ≥10% body weight loss at 4 and 12 months by dividing the number of participants per treatment arm who achieve these weight targets by the total number of participants in treatment arm and multiplying by 100.
Change in glycemic variability: Glycemic variability contributes to vascular damage [32], and intraday blood glucose variability is greater in people with prediabetes compared to people with normal glycemic levels [33]. We will place an Abbott Libre Pro continuous glucose monitoring (CGM) device on a participant’s upper arm at 0 and 12 months. Participants will wear each sensor for 14 days. This type of CGM records participants’ glucose levels in the interstitial fluid by a glucose oxidase method every 15 minutes; the sensor is blinded, and participants will not receive feedback. The research team will then download sensor data at the end of the measurement period. Following previous research, we will assess the glucose variability and the proportion of time spent in the euglycemic (3-7.8 mmol/l) and hyperglycemic (≥ 11.1 mmol/l for at least 15 minutes) states, following previous standards for interstitial glucose concentrations [34,35].
Change in serum insulin and insulin resistance: Homeostatic Model Assessment-Insulin Resistance (HOMA-IR) is a widely used method of estimating insulin resistance based on fasting insulin and glucose levels [36]. HOMA-IR will be calculated at 0, 4, and 12 months and mean change from baseline to 4 and 12 months will be determined.
Change in serum lipids: Understanding how a carbohydrate-reduced diet affects lipids for those with prediabetes may help assess the likely impact of the diet on macrovascular complications. We will assess lipids using NMR Lipoprofile® by LabCorp [37].
Change in inflammatory markers: High-sensitivity CRP will be measured with nephelometric methods utilizing latex particles coated with CRP monoclonal antibodies and standardized against a CRP reference preparation at 0, 4, and 12 months.
Exploratory outcomes
Class attendance: We will report mean session attendance. Additionally, in accordance with CDC Diabetes Prevention Program Recognition Standards, we will report the number of participants that completed at least 8 sessions during the program’s core phase and who remained engaged in the program for at least 9 months [38].
Dietary adherence: We will assess dietary adherence with one unannounced 24-hour dietary recall at 0, 4, and 12 months, which allows us to measure absolute and percent of calories of each macronutrient. For participants in the very low-carbohydrate group, blood ketone levels are a biomarker to help assess whether target levels of carbohydrate restriction have been achieved. We will assess fasting β-hydroxybutyrate at 0, 4, and 12 months for all participants.
Rate of conversion to T2DM: We will calculate the percentage of participants that progress from prediabetes to T2DM at 4 and 12 months, as determined by HbA1c level of ≥ 6.5%.
Initiation of anti-hyperglycemic agents: We will calculate the percentage of participants in each arm that initiate use of any anti-hyperglycemic medication during the study period.
Change in survey measures: We will evaluate change in survey measures, like self-reported symptoms and quality of life, from baseline to 4 and 12 months.
Qualitative data collection
Participants’ experiences in the intervention: We will conduct semi-structured interviews with a total of 68 participants; participants will have the opportunity to complete interviews at either 4 or 12 months after baseline. During these interviews, we will explore participants’ experiences during the intervention, including perceived barriers to and facilitators of dietary change and their ability to maintain these changes over time. We will purposively sample individuals with varied weight change and HbA1c outcomes. We will maximize sample variation by inviting participants with differences in gender, age, and race/ethnicity to complete interviews.
Participant timeline {13}
Assignment of interventions: allocation
Sequence generation {16a}
Individuals who meet study inclusion criteria, provide written informed consent, and complete baseline assessments will be randomized to one of two groups, standard or very low-carbohydrate version of the DPP, with a 1:1 ratio. The order will be created using block randomization procedures, with blocks randomly allocated to size 2, 4, or 8 and two strata: baseline HbA1c of 5.7%-6.0% or 6.1%-6.4% and sex of male or female. We will use the R package blockrand() with a random seed of 2345.
Concealment mechanism {16b}
A web-based tool will be used for blinded treatment allocation.
Implementation {16c}
A trained study team member will notify participants of their treatment assignment.
Assignment of interventions: blinding
Who will be blinded {17a}
Although treatment condition will be apparent to participants and researchers, outcome assessment and data analyses will be blinded. Participants will be made aware of their treatment group assignment via email 1 week prior to the first session.
Procedure for unblinding if needed {17b}
Not applicable.
Plans to promote participant retention and complete follow-up {18b}
To encourage participant retention, participants will be paid $55 at 4 months and $75 at 12 months for completing study-related assessments at each time point. Participants will receive an additional $20 at baseline for returning their CGM on time. Additionally, individuals who participate in optional semi-structured interviews will receive $25.
Data management {19}
Study data will be collected and stored in REDCap [23] or University of Michigan’s HIPAA-compliant DropBox account [41]. All computers used by study staff will be encrypted. All blood drawn from participants will be analyzed and then destroyed by LabCorp. Laboratory test results will be shared with participants using secure email. Participants’ physicians may be contacted through the Electronic Health Record, HIPAA-compliant fax, or by phone. Participant data will be retained by the study team until analyses are complete. Data related to any identifying information will be destroyed once analyses are complete (approximately 3-6 years post-study). All data will be kept in encrypted storage.
Confidentiality {27}
We will use a crosswalk file to link participants to a unique study specific identifier. The identifier will be used, whenever possible, in place of identifiable participant information.
Plans for collection, laboratory evaluation and storage of biological specimens for genetic or molecular analysis in this trial/future use {33}
Not applicable; no biological specimens will be stored in this trial.
Statistical methods
Statistical methods for primary and secondary outcomes {20a}
Primary intent-to-treat analysis: The principal analysis will be intent-to-treat, with all observations included for all individuals based on initial group assignment, regardless of adherence to the treatment protocol. Our primary outcome is a continuous longitudinal outcome, so we will use hierarchical linear mixed effects models to assess differences over time. The final model will include fixed effects for linear and quadratic time-by-arm interaction terms, and randomization and stratification variables. Covariance terms will be included for repeated measures over time. Because we are studying a group intervention, there may be effects that vary due to differences in group dynamics or teachers. To address this, we will include a cluster variable as a random effect to account for the lack of full independence of observations within the same group. Our previous trials’ intraclass correlation coefficient (ICC) for the clustering effect of coaches has been <0.001, suggesting that the effect of individual dynamics of groups is minimal. We will investigate model diagnostics for the linear model. If the assumption of normal errors is not met, we will consider transformations, such as the log transform, to reduce skewness.
For secondary outcomes that are continuous, we will use linear mixed models that are analogous to the model described above. For secondary outcomes that are binary (e.g., percentage of participants who achieve weight loss thresholds), we will use an analogous general linear mixed effects model. All analyses will be conducted using either SAS or R statistical software.
Qualitative analysis
Interviews will be recorded and transcribed verbatim by a professional transcriptionist. At least two members of the study team will participate in inductive, thematic data analysis, which will involve: 1) reviewing transcripts; 2) developing codes; 3) applying codes to transcripts; 4) resolving coding differences through consensus conferences; 5) developing themes that interrelate codes; 6) reviewing and refining themes; and 7) describing themes, evidenced by participant quotes [42].
Integrated analysis
Mixed methods integration occurs when quantitative and qualitative data are combined to further clarify the research question [43]. We will integrate the data in several ways. Following thematic analysis, we will use the resulting codes and themes to create categorical variables of barriers and facilitators of dietary changes. This approach has been used in health services research to develop variables that can be tested in subsequent quantitative analyses [44,45]. Using this approach, we will explore the association between variables derived from qualitative data and quantitative intervention outcomes. Second, we will create a visual joint display to co-present qualitative and quantitative outcomes and enhance our understanding of the study’s findings [46].
Interim analyses {21b}
None.
Methods for additional analyses (i.e., subgroup analyses) {20b}
Secondary per-protocol analyses: As a secondary approach, we will also perform a per protocol analysis based on session attendance. The CDC Diabetes Prevention Recognition Program defines a program completer as “an eligible participant enrolled in an evaluation cohort who attended at least 8 sessions in months 1-6 and whose time from first session held by the cohort to last session attended by the participant is at least 9 months.” [47] We will analyze our results based on CDC completer status and also by using session attendance as a continuous measure.
Methods in analysis to handle protocol non-adherence and any statistical methods to handle missing data {20c}
To investigate the impact of missing data on the analysis, we will examine the pattern of missingness in the data. We will compare the means among each observed pattern of missingness. If the means are similar, suggesting that the data are missing completely at random, then we will exclude missing values from the analysis. If the pattern of missingness suggests that the data are missing not at random, we will use multiple imputation to estimate the treatment effect while correctly modeling the variability in the data.
Plans to give access to the full protocol, participant level-data and statistical code {31c}
There are no plans to grant public access to the full protocol, participant-level dataset, or statistical code.
Sample size {14}
To estimate the number of subjects to recruit, we note that the DPP showed an approximate 0.1% decrease in HbA1c with a standard deviation of the difference approximately equal to 0.1% [3]. This decrease in HbA1c was deemed clinically significant because it led to a clear reduction in type 2 diabetes incidence compared to control. Two studies of VLCDs in adults with prediabetes had a standard deviation for the difference approximately equal to 0.1%, similar to the DPP [48,49]. Thus, we used 0.1% as the standard deviation in our sample size calculation. We used an alpha level of 0.05. Finally, although this is a superiority trial, the potential of non-inferiority is of clinical value, thus we used a power of 95%. To be conservative, and because the low ICC (cited above) causes minimal increase in variability due to clustering, we consider a test of the difference between the two groups at 12 months. Our proposed hierarchical linear model that pools the repeated measurements will enable us to maintain high power while adjusting for important covariates like sex and examining the effect of treatment phase. The sample size needed to detect a 0.1% decrease in HbA1c with a standard deviation of 0.1% with alpha of 0.05 and power of 0.95 using a two-sample t-test is 105 participants per group. Our ongoing and published trials of diet and lifestyle programs with adults with prediabetes have had a retention rate of 70% or better [17]. Assuming a similar rate of 70% retention, the number of participants we need to enroll is 150 per group.