ACT1VATE: Addressing Emotional Distress to Improve Outcomes Among Diverse Adults With Type 1 Diabetes

HbA1c (%) reflects average glucose over the past 2-3 months, with higher values indicating greater risk for developing diabetes-related complications. HbA1c for up to 5 data points (0, 3, 6, 9, 12 months) will be analyzed. Multilevel models using full information maximum likelihood estimation will be conducted to examine HbA1c changes. Analyses will include the between-subjects factor of group and the within-subjects factor of time. Month 0 will be the referent time-point with post-intervention and follow-up time-points as comparison time-points in dummy-coded predictors. The group by time interaction is of primary interest. If an interaction is found significant, follow-up analyses will determine the nature of differential change between treatment conditions.

The Type 1 Diabetes Distress Scale (T1-DDS; 28 items averaged to obtain a total score ranging 1-6, with higher scores indicating greater diabetes-related emotional stress) will be analyzed. Multilevel models using full information maximum likelihood estimation will be conducted to examine change in diabetes distress over time. Analyses will include the between-subjects factor of group and the within-subjects factor of time. Month 0 will be the referent time-point with post-intervention and follow-up time-points as comparison time-points in dummy-coded predictors. The group by time interaction is of primary interest. If an interaction is found significant, follow-up analyses will determine the nature of differential change between treatment conditions.

Seven items regarding dietary and exercise behaviors from the Summary of Diabetes Self-Care Activities scale (SDSCA; items averaged to obtain a total score for each behavioral domain, ranging from 0-7, with higher scores indicating better outcomes/greater adherence to diabetes self-management behaviors) will be analyzed. Multilevel models using full information maximum likelihood estimation will be conducted to examine behavioral changes. Analyses will include the between-subjects factor of group and the within-subjects factor of time. Month 0 will be the referent time-point with post-intervention and follow-up time-points as comparison time-points in dummy-coded predictors. The group by time interaction is of primary interest. If an interaction is found significant, follow-up analyses will determine the nature of differential change between treatment conditions.

The Generalized Anxiety Disorder Assessment (GAD-7; 7 items are summed to obtain a total score, ranging from 0 to 21, with higher scores indicating worse outcomes/greater anxiety symptoms) will be analyzed. Multilevel models using full information maximum likelihood estimation will be conducted to examine anxiety symptom changes over time. Analyses will include the between-subjects factor of group and the within-subjects factor of time. Month 0 will be the referent time-point with post-intervention and follow-up time-points as comparison time-points in dummy-coded predictors. The group by time interaction is of primary interest. If an interaction is found significant, follow-up analyses will determine the nature of differential change between treatment conditions.

Patient Health Questionnaire (PHQ-8; 8 items averaged to obtain a total score, with higher scores indicating worse outcomes/greater depressive symptomatology) will be analyzed. Multilevel models using full information maximum likelihood estimation will be conducted to examine depressive symptom changes over time. Analyses will include the between-subjects factor of group and the within-subjects factor of time. Month 0 will be the referent time-point with post-intervention and follow-up time-points as comparison time-points in dummy-coded predictors. The group by time interaction is of primary interest. If an interaction is found significant, follow-up analyses will determine the nature of differential change between treatment conditions.

Perceived Stress Scale (PSS-4; 4 items are summed to obtain a total score, ranging from 0-16, higher scores indicating worse outcomes/greater levels of perceived stress) will be analyzed. Multilevel models using full information maximum likelihood estimation will be conducted to examine stress level changes over time. Analyses will include the between-subjects factor of group and the within-subjects factor of time. Month 0 will be the referent time-point with post-intervention and follow-up time-points as comparison time-points in dummy-coded predictors. The group by time interaction is of primary interest. If an interaction is found significant, follow-up analyses will determine the nature of differential change between treatment conditions.

The WHO Well-Being Index (WHO-5; 5 items are summed and multiplied by 4 to obtain a total score ranging from 0-100, with higher scores indicating better outcomes/greater well-being/overall quality of life) will be analyzed. Multilevel models using full information maximum likelihood estimation will be conducted to examine quality of life changes. Analyses will include the between-subjects factor of group and the within-subjects factor of time. Month 0 will be the referent time-point with post-intervention and follow-up time-points as comparison time-points in dummy-coded predictors. The group by time interaction is of primary interest. If an interaction is found significant, follow-up analyses will determine the nature of differential change between treatment conditions.

Hypoglycemic Attitudes and Behaviors Scale [HABS; a 14-item self-report scale that highlights three dimensions of hypoglycemia-related concerns (anxiety, avoidance, confidence), with higher scores indicating higher levels of patient worry and concerns related to hypoglycemia] will be analyzed. Multilevel models using full information maximum likelihood estimation will be conducted to examine changes in hypoglycemic attitudes and behaviors over time. Analyses will include the between-subjects factor of group and the within-subjects factor of time. Month 0 will be the referent time-point with post-intervention and follow-up time-points as comparison time-points in dummy-coded predictors. The group by time interaction is of primary interest. If an interaction is found significant, follow-up analyses will determine the nature of differential change between treatment conditions.

The Diabetes Knowledge Test (DKT; a 23-item multiple-choice questionnaire developed by the University of Michigan, with higher scores reflecting greater diabetes knowledge) will be analyzed. Multilevel models using full information maximum likelihood estimation will be conducted to examine knowledge changes over time. Analyses will include the between-subjects factor of group and the within-subjects factor of time. Month 0 will be the referent time-point with post-intervention and follow-up time-points as comparison time-points in dummy-coded predictors. The group by time interaction is of primary interest. If an interaction is found significant, follow-up analyses will determine the nature of differential change between treatment conditions.

The Diabetes Support and Isolation Questionnaire (an 11-item, multiple choice questionnaire developed by University of California, San Francisco to assess available support resources for managing one's diabetes) will be analyzed. Multilevel models using full information maximum likelihood estimation will be conducted to examine changes in support resources over time. Analyses will include the between-subjects factor of group and the within-subjects factor of time. Month 0 will be the referent time-point with post-intervention and follow-up time-points as comparison time-points in dummy-coded predictors. The group by time interaction is of primary interest. If an interaction is found significant, follow-up analyses will determine the nature of differential change between treatment conditions.

Long-term cost-effectiveness of ACT1VATE will use a simulation model. The UKPDS Outcomes Model Risk Engine (UKPDS-OM) developed the predictive equations for diabetes-related complications, mortality, and assigned utilities conditional on disease. Estimated clinical effects + costs of each arm will be inputs into the UKPDS-OM model. Estimated clinical effects (HbA1c change) come from Aim 1. Intervention costs will be estimated using standard accounting methods, process mapping + time-based activity costing. Base case assumes a health system perspective, 40-year time horizon, and 3% discount rate for both QALY and costs. Sensitivity analyses will investigate the influence of the estimated treatment effects and intervention costs, the influence of time horizon and discount rate, and second order uncertainty. UKPDS-OM provides equation parameters derived from bootstrap samples of the original UKPDS population. Estimates will be used to calculate an incremental cost-effectiveness plan.

Diabetes-Specific Self-Compassion Scale (D-SCS; range 0-5 with higher scores indicating higher levels of diabetes-specific self-compassion) will be analyzed to evaluate theoretical fidelity. Multilevel models using full information maximum likelihood estimation will be conducted to examine change in the construct over time. Analyses will include the between-subjects factor of group and the within-subjects factor of time. Month 0 will be the referent time-point with post-intervention and follow-up time-points as comparison time-points in dummy-coded predictors. The group by time interaction is of primary interest. If an interaction is found significant, follow-up analyses will determine the nature of differential change between treatment conditions.

The Acceptance and Action Diabetes Questionnaire (AADQ; a 6-item scale measuring cognitive avoidance related to diabetes with higher scores indicating higher levels of cognitive avoidance) will be analyzed to evaluate theoretical fidelity. Multilevel models using full information maximum likelihood estimation will be conducted to examine change in the construct over time. Analyses will include the between-subjects factor of group and the within-subjects factor of time. Month 0 will be the referent time-point with post-intervention and follow-up time-points as comparison time-points in dummy-coded predictors. The group by time interaction is of primary interest. If an interaction is found significant, follow-up analyses will determine the nature of differential change between treatment conditions.

Body Mass Index (BMI) - will be measured as Weight in kg/Height squared in meters. BMI for up to 5 data points (0, 3, 6, 9, 12 months) will be analyzed. Multilevel models using full information maximum likelihood estimation will be conducted to examine BMI changes. Analyses will include the between-subjects factor of group and the within-subjects factor of time. Month 0 will be the referent time-point with post-intervention and follow-up time-points as comparison time-points in dummy-coded predictors. The group by time interaction is of primary interest. If an interaction is found significant, follow-up analyses will determine the nature of differential change between treatment conditions.