A behavioral randomized control trial (RCT) for high diabetes distress. In Phase 2 we propose a practical two-arm, nine month RCT to reduce high DD and improve glycemic control and disease management. We will compare two active interventions: (1) a Type 1 diabetes education protocol ('KnowIt') that brings together new advances in diabetes education and behavioral management, and (2) A DD-reduction protocol ('OnTrack') that helps identify and address the personal stresses and strains of having diabetes.
Patients in both study arms will receive three months of intervention and six months of follow-up. Each arm includes a one-day in-person group workshop for 8-10 participants; four real-time, 65-minute web-based group meetings; and six individual telephone calls from the group leader. The web-based meetings will occur at two, four, eight and 12 weeks after the workshop. Four of the personal phone calls will occur one week prior to each web-based group meeting to provide support and encourage participation in the group meetings; two personal phone calls will occur at weeks 20 and 28 (between months thee and nine) to maintain support and participation. Assessment will take place at baseline, 12 and 36 weeks from the initial workshop.
Hypotheses:
H1. The OnTrack arm will be superior to KnowIt on the primary outcomes (HbA1c and DD) and the secondary outcomes (diabetes management) at 12 and 36 weeks.
H2. (mediator hypothesis). Differential between-group change in primary outcomes will be mediated by changes in self-efficacy and fear of hypoglycemia across time.
H3. (moderator hypothesis). The differences between the two study arms and changes in the primary outcomes will be qualified by: (a) patient characteristics, with stronger positive effects for patients with younger age, shorter diabetes duration, higher baseline distress or HbA1c, BMI, and more frequent hypoglycemic episodes; and (b) intervention engagement, with stronger positive effects for patients who attend more calls or have higher use of intervention media. We will test our findings against the null hypothesis of no difference between the two groups. All statistical tests will be two tailed.
Statistical Analyses:
We will perform analyses to determine whether differential attrition occurred by patient characteristics. Likelihood based approaches (e.g. GLMMs) and multiple imputation will be used to handle missing data. Background demographics will be examined to describe the sample and will be included in multivariate analysis if they are related to the outcome at p\<.2, differ between treatment arms, or associated with dropout. Correlations among variables will be examined and redundant measures will be combined or eliminated to avoid multicollinearity. We will select potential moderators by retaining those at p\<.2 and then reassessing models to assess for significant interactions.
Assessment of Primary \& Secondary Outcomes: We will fit regression models for cross-sectional and longitudinal outcomes. Demographics will be compared for the two arms at baseline. Comparisons of OnTrack and KnowIt on changes in DD and HbA1c will be estimated with linear mixed models and generalized linear models with generalized estimating equations (e.g., SAS PROC MIXED ROC GENMOD, SAS Institute Inc., 1999). Generally, we seek to model longitudinally the trajectory of these outcomes as a function of time, intervention group assignment, and group-by-time interaction. We begin by fitting base models, those with minimum covariates and, for longitudinal models, the most restricted residual covariance structures. Additional covariates then will be considered. Similar models will be used to examine the secondary outcomes of diet, physical activity, and self-management, e.g., BG variability). We will examine change from baseline for each outcome as the dependent variable. In addition to testing the groups-by-time interaction term, custom contrasts will assess group differences at each time point.
Mediating \& Moderating Analyses. These analyses will further elaborate any observed treatment effects. A first set will determine whether demographic strata, use of media, and dose effects moderate any intervention effects. Each outcome will be regressed onto measures of intervention, baseline covariates and their interactions with intervention group assignment. A second set of analyses will test specific mediation hypotheses.
Cost Estimates. To calculate the per-participant costs of each intervention, we will first adjust all of the costs obtained from project records for inflation to constant US dollars for the final year of the study using the Bureau of Labor Statistics' Consumer Price Index. Capital costs will be annualized and discounted at the investment interest rate in the final year of the program. Fixed costs will be allocated among all participants in each arm, and we will divide group leader costs by the number of respective participants. After these adjustments and allocations are completed, the newly calculated per-participant costs will be combined with any additional costs to compute the total per-participant cost for each intervention. These cost estimates can be used to project the cost of future dissemination and implementation efforts.
Power Analysis. Sample size and power estimates for comparison of OnTrack and KnowIt in Phase 2 are based on alpha = 0.05 and 2-sided t-tests on change from baseline to 12, 24 and 36 weeks. Estimates of means (SDs) at follow-up from similar studies in type 1 adults (48, 51)and previous experience with REDEEM suggest effect sizes that are small to moderate, with 0.35 to .40 SD unit differences. A sample of 116 participants assigned to OnTrack and to KnowIt will provide power to exceed 0.80 for tests on the primary outcomes of DD and HbA1c. This allows detectable mean differences of changes from baseline of magnitude 0.48 for HbA1c. At the projected sample size for each group, power to discriminate between OnTrack and KnowIt will be at least 80%. This sample size also will allow us to determine a medium effect size for the mediation effect (H2) of fear of hypoglycemia with a difference of 15%, and of self-care behavior with a difference of 17% between the arms. Sample size estimates for moderator effects (H3) are based upon a group-by-moderator strata interaction test in a regression model. Findings (48)showed between-group differences on age ranging from of -1.2 to 0.55. Based upon these estimates, the proposed study will be powered to detect the moderator effect, conditional on the main effects, equal to 0.93. Our preliminary findings for group by baseline HbA1c interactions and those from the literature suggest power of at least 0.80.
Procedures:
Following screening to assess initial eligibility and interest, a research assistant will email a baseline survey using Qualtrics with a unique study ID number assigned by the researcher. The survey begins with the consent form that will be completed electronically. The research assistant will also email the patient a PDF copy of the consent form and brochure to assist the patient in fully considering participation, and to retain a copy of the study consent form to keep for reference.
Baseline assessment (assessment 1): Following informed consent in Qualtrics, participants will navigate to complete the baseline survey. The survey includes the Type 1 Diabetes Distress Scale (T1-DDS), plus scales to assess self-management, eating and weight patterns, medication adherence, mindfulness, problem solving, depression and quality of life. Once participants complete the survey, their T1-DDS score will be calculated. If this meets our inclusion threshold (mean score 2.0 and above) participants will be sent a slip for a baseline blood draw to measure HbA1c and cholesterol levels at a community laboratory (Quest) of their choosing. Participants whose mean diabetes distress score (measured using the T1-DDS scale) is below 2.0 will be informed by letter and phone call that they are no longer eligible. Participants who meet the threshold for severe symptoms of depression (PHQ score 15 and above) will also be informed by letter and phone call from a research assistant, who will administer a script to encourage the participant to seek help for their symptoms and to screen for suicide risk.
Laboratory results will be reviewed to determine whether the HbA1c results meet inclusion criteria (7.5% and above). Participants who meet the HbA1c criteria will be randomized by the research assistants. Upon completion of baseline assessment participants will be compensated via gift card (pro-rated for partial baseline completion) for survey completion and for blood draw.
Group workshop (week 0): Participants will be randomly allocated to either OnTrack or KnowIT. Both programs are the same length and contain the same frequency and types of events, however the content of each differs and each will be delivered by separate, trained interventionists. We plan to deliver the workshop to 8-10 participants.
Each workshop will be delivered by a trained interventionist for that workshop. Participants will have the opportunity to participate in group discussions and share experiences throughout the workshop to reflect on the material being delivered. During the workshops the heights, weights and blood pressures of participants (required for baseline characteristic assessment and the outcome analyses) will be recorded in a private location by one of the attending members of the study team. Participants in both groups will receive a copy of their baseline HbA1c result, cholesterol results and blood pressure results. In addition, participants in the OnTrack group will receive a copy of their diabetes distress score profile (since diabetes distress is specifically targeted in this intervention arm). At the end of the workshops, participants in both groups will write a brief Action Plan that allows participants to develop a series of personal goals to enhance their diabetes management. Participants in both groups will be given a copy of the workshop content at the end of the workshops for reference.
Group web meetings and individual phone calls (weeks 1- 12): Following the workshop, participants will receive four 10-15 minute individual phone call from their group facilitator at one, three, seven and 11 weeks following the initial workshop, at mutually agreed times. The purpose of the call is to review participant progress with their Action Plans and to enable participants to discuss any challenges one-on-one with the facilitator. Participants will also take part in four 65 minute online web group meetings using WebEx, at two, four, eight and 12 weeks after the workshop. WebEx allows individuals to connect in an online meeting using video and audio. The purpose of the meetings is to reinforce and build upon workshop content, and to allow participants to support each other and to raise any challenges they are experiencing with reaching their Action Plan goals.
Assessment 2 (week 12 / 3 months): Participants will be asked to complete the same survey and blood draws as they completed at baseline. A research assistant will email a personalized and confidential link to the survey via Qualtrics, and will mail a lab slip for a blood draw from a local community laboratory of the participants' preference.
MAINTENTANCE PHASE (WEEKS 13 - 39): Following the workshop, participants will receive two 10-15 minute individual phone call from their group facilitator at 20 and 28 weeks following the workshop to review their progress with their Action Plans.
Assessment 3 (week 39 / 9 months): Participants will be asked to complete the same survey and blood draws as they completed at baseline and assessment 2.
