This multisite RCT enrolled parents of young children with T1D from all regions of the United States through one of 3 recruitment sites located in the Southeast, Midwest, or Mountain regions of the United States. Our primary outcome was change in parental FH. Our secondary outcomes were change in parental diabetes distress and children’s HbA_1c. We measured outcomes pretreatment, immediately post treatment, and 3-month post treatment, with the last assessment point intended to explore any maintenance effects.

We recruited participants from 2019 to 2023; however, to comply with COVID-19 restrictions, we halted recruitment from March to July 2020. Additionally, due to slow recruitment, in August 2022, we expanded our efforts and launched a media campaign through the Type 1 Diabetes Exchange to advertise for the trial nationally. Our inclusion criteria were parents (including legal guardians) of young children with T1D, aged 2‐6.99 years old, ≥6 months post diagnosis, and using intensive insulin therapy. We excluded parents if their child had a comorbid chronic condition (eg, renal disease), their child was allergic to the adhesive or skin preparation used for continuous glucose monitors (CGMs), their child was on conventional insulin therapy, or parents did not identify as English speaking. We did not restrict families based on the child’s HbA_1c, though we required young children to be at least 6 months post diagnosis to allow for their HbA_1c to begin to stabilize following insulin initiation. We targeted parents who self-identified as primarily responsible for their child’s daily T1D management to complete the treatment sessions and study questionnaires. However, parents who identified as a secondary caregiver for daily T1D management were also welcome to attend the treatment sessions, and multiple caregivers from the same family were able to participate together if preferred, though only one caregiver was included in our analyses.

We used both in-clinic and remote recruitment procedures. We screened all families who expressed an interest in participating in the trial for eligibility. After obtaining parent-informed consent, we randomized families 1:1 to either our treatment (REDCHiP) or ATTN. Families then completed visit 1, which included REDCap (Research Electronic Data Capture) [25] questionnaires and a measurement of child HbA_1c via a validated home kit, which was analyzed in a central lab. Families completed visit 2 at the end of the treatment phase and visit 3 approximately 3 months later. Visits 2 and 3 repeated the same study procedures as visit 1.

We compensated families for completing the study visits (US $35 per study visit). Parents received this incentive via a reloadable debit card (Clincard). We also gifted children a toy, valued at up to US $15, at each study visit as a thank you for completing the home HbA_1c kit.

Families randomized to REDCHiP or ATTN completed the same online questionnaires, and young children completed a validated home HbA_1c kit. We previously published a description of our trial battery [13]. For this report, we focus on describing our sample characteristics and examining treatment effects based on changes in parent FH and diabetes distress and child HbA_1c.

We conservatively based our original power analysis on the assumption of completing 40 groups (20 REDCHiP and 20 ATTN) with an average size of 3.6 parents (40×3.6=144, or a final sample of n=144 parents). Based on this sample size, a standardized effect size of 0.4 SDs, and an intraclass correlation of 0.10, we projected an estimated power of 85%. We originally estimated an attrition rate of 20%, thereby setting our target sample size at 180 families. However, during the trial years affected by COVID-19, we experienced a higher rate of attrition. Thus, to account for this higher attrition rate, we over-recruited for the trial; our final sample comprised 197 families.

To compare families randomized to REDCHiP or ATTN based on sociodemographic characteristics collected at visit 1, we used independent 2-tailed t tests, chi-square tests, or 2-sample Wilcoxon rank-sum tests for nonnormally distributed variables. Because preliminary analyses indicated correlations between parent depressive symptoms and both FH and distress, we elected to include the presence of elevated parental depressive symptoms as a covariate in all our models to better isolate treatment effects. To examine reductions in parents’ FH (primary outcome), we ran a mixed effects model with participant as a random effect and fixed effects for randomization (REDCHiP or ATTN), visit (slope of the respective outcome over the study timeline), and the interaction of randomization by study timeline. For diabetes distress (secondary outcome), we repeated the mixed effects model when analyzing distress as a continuous variable. We conducted a logistic regression to analyze distress dichotomously (elevated vs not elevated) as a sensitivity analysis to examine whether parents were less likely to report clinically elevated distress over time based on an established threshold, using the same random effect structure and covariate. In exploratory analyses, we examined reductions in child HbA_1c by repeating the mixed effects model. We conducted all analyses in StataNow/SE 18.5 (StataCorp LLC). Because our hypotheses were directional and prespecified, we applied one-sided P values to test for changes in the expected direction. We did not adjust for multiplicity because we designed the trial as an ORBIT Phase 2b study exploring treatment effects rather than providing definitive efficacy estimates [23]. Consistent with CONSORT (Consolidated Standards of Reporting Trials; Checklist 1) reporting guidelines [34], we report estimates of effect sizes for observed treatment-slope trends based on the methods described by Feingold [35] to aid interpretation of the magnitude of change; they should not be interpreted as evidence of statistical significance.

We operated this multisite RCT under a single institutional review board (IRB) agreement, with Children’s Mercy Hospital-Kansas City as the IRB of record. We used IRB-approved procedures to obtain electronic documentation of parent informed consent before enrolling parents into the trial. All children were younger than 7 years; therefore, we did not require separate child assent. We registered the trial on ClinicalTrials.gov (NCT3914547) before proceeding with any trial activities. We collected and managed study data using secure, password-protected systems and deidentified all data prior to analysis. Participation involved minimal risk, and parents could skip questionnaire items or withdraw from the study at any time without penalty. We compensated participants for their time.

We enrolled 197 families of young children with T1D and randomized 183 families (n=93 to REDCHiP and n=90 to ATTN). At visit 1, we collected parent and child demographics. Young children had a mean age of 4.4 (SD 1.2) years, a mean HbA_1c of 7.54% (SD 1.15), and a mean time since diagnosis of 1.67 (SD 1.15) years. Children were 55% (100/183) male. Parents had a mean age of 36.3 (SD 5.9) years and 93% (170/183) self-identified as the child’s mother. Table 1 displays sample characteristics and comparisons by group. The CONSORT table details our recruitment and retention across the seven-and-a-half-month trial (Figure 1). Of the total enrolled, 77% (152/197) of families completed all study visits.

Overall, parents randomized to REDCHiP attended 95.2% of sessions, and parents randomized to ATTN attended 95.9% of sessions. While there was no difference in the mean length of group sessions for the REDCHiP and ATTN arms (P=.22), there was a difference in the mean length of individual sessions for the REDCHiP and ATTN arms (P=.003). Specifically, individual sessions in the REDCHiP arm were on average 7 minutes longer in duration than individual sessions in the ATTN arm (mean 39.6, SD 4.2 vs mean 32.7, SD .7 minutes for REDCHiP and ATTN, respectively). We coded treatment fidelity for 170 out of 517 (33%) of ATTN sessions and 199 out of 532 (37%) of REDCHiP sessions using structured, session-specific checklists. Overall, both arms achieved a high treatment fidelity rate (85/100, 85% for ATTN and 89/100, 89% for REDCHiP).

Table 2 describes outcomes for parent FH, diabetes distress, and the percentage of parents with clinically elevated depressive symptoms for the entire sample and within each treatment arm. Across both arms, parent FH, diabetes distress, and the percentage of parents with clinically elevated depressive symptoms decreased from visits 1 to 3. However, at all study visits, parents with clinically elevated depressive symptoms reported higher FH and distress than parents without clinically elevated depressive symptoms (P=.001; Table 3).

In a series of mixed effects models, we examined treatment effects for parents’ FH and diabetes distress, with randomization (REDCHiP or ATTN), visit (slope of the respective outcome over study timeline), interaction of randomization by study timeline, and presence of elevated depressive symptoms as covariates. In our first model, results revealed that parents’ FH decreased over time for both treatment arms (coefficient=−2.57; 95% CI −3.81 to −1.33; P=.001), with a treatment-slope effect that slightly favored the REDCHiP group, though it was not significant (coefficient=−1.16; 95% CI −2.88 to 0.57; P=.09). Similarly, in our second model, results revealed a decrease in parents’ diabetes distress over time for both treatment arms (coefficient=−3.93; 95% CI −5.63 to −2.23; P=.001), with a nonsignificant treatment slope effect (coefficient=−1.87; 95% CI −4.23 to 0.50; P=.06) that slightly favored the REDCHiP group. Although the interaction terms were not statistically significant, we calculated effect sizes to describe the magnitude of observed trends in accordance with CONSORT recommendations [34]. For parents’ FH, the effect size for REDCHiP was Feingold d=0.18, and for diabetes distress, the effect size for REDCHiP was Feingold d=0.21.

We also examined treatment effects for parents dichotomized as having either clinically elevated or not elevated diabetes distress, with randomization (REDCHiP or ATTN), visit (slope of the respective outcome over the study timeline), interaction of randomization by study timeline, and presence of elevated depressive symptoms as covariates. There was no time effect (coefficient=−0.25; 95% CI −0.61 to −0.11; P=.08). However, we observed a significant treatment-slope effect (coefficient=−0.70; 95% CI −1.35 to −0.05; P=.02), suggesting that, over time, REDCHiP parents were less likely to endorse clinically elevated diabetes distress.

In mixed effects models, we explored treatment effects for children’s HbA_1c, with randomization (REDCHiP or ATTN), visit (slope of the respective outcome over study timeline), interaction of randomization by study timeline, and presence of elevated depressive symptoms as covariates. Results revealed a small but nonsignificant reduction in child HbA_1c over time for both treatment arms (coefficient=−0.05; 95% CI −0.12 to 0.03; P=.11), with a nonsignificant treatment-slope effect for children of parents in the REDCHiP group (coefficient=0.04; 95% CI −0.06 to 0.15; P=.21).

Our aim was to examine the source of treatment effects for REDCHiP, a telehealth-delivered intervention to reduce FH in parents of young children with T1D, by conducting an RCT. We recruited a large sample of parents from across the United States and randomized them to receive either REDCHiP or an attention control treatment (ATTN). We hypothesized parents receiving REDCHiP would report greater reductions in FH and diabetes distress compared to parents who were randomized to ATTN. We also explored whether children experienced an HbA_1c reduction based on parents’ exposure to REDCHiP or ATTN.

Contrary to our primary hypothesis, our results suggest that all participating parents experienced significant reductions in FH with treatment-slope effects that slightly favored REDCHiP parents and were nonsignificant (P=.09). When we examined parents’ diabetes distress as a secondary outcome, our model with diabetes distress as a continuous variable showed a similar time and treatment slope effect (P=.06). However, the model analyzing diabetes distress as a dichotomous variable identified a significant treatment-slope effect for REDCHiP. Specifically, we found that parents receiving REDCHiP were less likely to report clinically elevated diabetes distress over time than parents receiving ATTN (P=.02). Our results also suggest a small and nonsignificant decrease in children’s HbA_1c over time, regardless of intervention group.

Although both REDCHiP and ATTN were associated with reductions in FH and diabetes distress, the reasons for improvement in the ATTN group remain unclear. ATTN sessions did not include content on emotional regulation or stress management, and there is limited evidence that general support groups reduce anxiety or depressive symptoms in adults [36]. One possibility is that nonspecific factors, such as social connection during sessions (especially surrounding the COVID-19 pandemic), contributed to improvements, though this is only speculative and would require further dismantling studies to confirm. Importantly, we designed ATTN as a comparator versus its own scalable intervention. Furthermore, it is unknown whether a longer follow-up post treatment would reveal greater divergence between groups. In a context where virtually no interventions exist to address FH in parents of young children with T1D, we believe the observed reductions, even if nonsignificantly different by condition, represent an important step forward in the field.

While there is an emerging trend in testing new behavioral interventions uniquely tailored to parents of young children with T1D, the literature remains small compared to parents of older children with T1D [37]. As such, we believe this trial adds valuable insight into how to support parents of young children with T1D in managing symptoms of FH and diabetes distress. Moreover, because this is the first behavioral intervention trial among parents of young children with T1D to use an attention control arm, we assert that it marks a notable improvement in trial design rigor directed at families of young children with T1D [38,39].

Our trial recorded a very high attendance rate (~95% of sessions attended) for parents randomized to REDCHiP and ATTN. We think the telehealth format of our treatments was likely the main contributor to our high session attendance. Parents could participate from home, thereby eliminating typical barriers to traditional, in-person behavioral health sessions (eg, transportation, childcare, and time off from work) [19,21,22]. However, it is possible another contributor to our high attendance rate may be parents’ keen interest in receiving support for the emotional and behavioral challenges of managing T1D in their young child. Indeed, our telehealth format allowed us to recruit nationally, potentially reaching families who may not have access to behavioral health care through their local diabetes center [40]. Thus, it appears REDCHiP’s telehealth format helped enhance feasibility and engagement in this trial and may provide preliminary evidence to support its broader dissemination and integration into pediatric diabetes care through existing telehealth infrastructure.

Though we did not find support for our main hypothesis, our trial builds on previous research that suggests that parents of young children with T1D may be particularly vulnerable to FH [3,6]. Parents in this trial reported baseline levels of FH that were comparable to levels reported in published studies [6]. We think this is notable in the context of our sample’s high rate of personal CGM (151/183, 83%) and automated insulin delivery (AID; 93/183, 51%) use. Within the larger literature, the results are mixed as to whether CGM and AID use are associated with lower FH in parents of young children [11,41-43]. Thus, our data may further support the notion that some parents require treatment specifically focused on their fear to experience any improvements [36]. We also think it is notable that REDCHiP parents experienced an absolute mean reduction in FH (−8.22) that was comparable to our previous REDCHiP pilot trial (−9.02) [12]. This may suggest that REDCHiP is still an effective treatment in helping parents to learn the skills to manage FH, even though the reduction was only slightly and non-significantly greater than among ATTN parents.

Our trial examined reductions in parents’ diabetes distress using 2 approaches, as a continuous variable and as a dichotomous variable based on clinically elevated versus nonelevated diabetes distress. We examined distress as a continuous variable to provide a precise estimate of treatment effects and calculate effect sizes. We conducted a sensitivity analysis using the dichotomous measure to assess clinical significance, whether parents moved from elevated to nonelevated distress, a threshold clinics would likely use to guide referral and treatment decisions. Results suggested that REDCHiP had a small effect on parents’ diabetes distress when analyzed continuously, with an absolute mean reduction (−12.88) comparable to our pilot trial (−11.11) [12]. This suggests REDCHiP helped parents manage diabetes distress, even if the treatment-specific effect was attenuated by the ATTN group. However, when analyzed dichotomously, exposure to REDCHiP was associated with a lower occurrence of elevated distress over time compared to ATTN, suggesting REDCHiP may help parents achieve clinically meaningful reductions in diabetes distress. This finding may be particularly relevant for clinical practice.

It is important to recognize the impact of a parent report of elevated depressive symptoms on FH and diabetes distress. At every assessment point and regardless of treatment group, parents with elevated depressive symptoms reported significantly higher FH and diabetes distress than parents without depressive symptoms. Unfortunately, we conducted this trial during the COVID-19 pandemic, a time when depressive symptoms were on the rise and parents of children vulnerable to COVID-19 were more likely to report depressive symptoms than adults prepandemic [44]. While not formally recorded, many parents in our trial described limiting social behaviors or not enrolling children in childcare or preschool. Many parents reported challenges in balancing working from home with parenting secondary to the COVID-19 pandemic. Moreover, some parents described additional concerns specific to COVID-19 and its potential added risk in youth with chronic illness. In our earlier REDCHiP pilot, we did not observe a change in parents’ depressive symptoms after exposure to the intervention [12]. We believe this may be because REDCHiP focuses specifically on changing parents’ unhelpful thoughts and behaviors related to fear. Nevertheless, the data from this trial suggest that parents with elevated depressive symptoms may require a different treatment approach. Therefore, a future direction could be to adapt REDCHiP to incorporate strategies specifically targeting depressive symptoms, such as behavioral activation [45], to better support parents experiencing depressive symptoms and FH.

While our trial demonstrated a small but nonsignificant reduction in child HbA_1c related to parent exposure to either REDCHiP or ATTN, this outcome may be expected. In our REDCHiP pilot, we only observed a small change in HbA_1c among young children who started the trial with an HbA_1c that was above the clinical target [12]. In the current sample, nearly one-third of young children had a baseline HbA_1c <7%. This suggests the possibility that we may have encountered a floor effect; there was simply not enough variability in young children’s HbA_1c to detect a meaningful change. In the future, it may be helpful to use young children’s CGM data to explore any glycemic changes related to their parents’ exposure to REDCHiP or stratify enrollment by baseline child HbA_1c to better detect glycemic effects.

Our use of a large national sample may enhance the generalizability of our results, though as a limitation we acknowledge that our sample was majority non-Hispanic White and that future research should explore the efficacy of REDCHiP in a more racially and ethnically diverse sample. Related, we acknowledge a limitation in that our trial primarily recruited mothers versus fathers and other caregivers. Because less is known about FH in fathers and other caregivers of young children with T1D [4], it is important that future research explore these other caregivers’ experience with FH and response to treatment. We acknowledge a limitation in that we did not restrict our sample to only include families of children with an above-target HbA_1c. We made this decision based on past data showing moderately high levels of FH even among parents of young children with an HbA_1c at or below their target [6]. Nonetheless, to see any child glycemic effects related to parents’ exposure to REDCHiP, it may be important for future trials to limit enrollment to families of young children with an above-target HbA_1c. Because this trial included multiple outcomes and 1-sided P values without adjustment for multiplicity, we acknowledge the risk of type 1 error. We made these analytic decisions to prioritize clinical relevance and to explore treatment effects in this Phase 2b trial. Nevertheless, future studies should confirm these findings using more conservative approaches. As important strengths of the trial, we highlight the high rate of CGM and AID use among young children, our use of a central laboratory to analyze children’s HbA_1c, our acceptable retention rate, our high attendance rate (~95% of sessions), our relatively low rate of data missingness (<10%) [46], and our rigorous trial design that included an attention control condition [38] and an assessment of treatment fidelity.

In summary, even with uptake of CGM and AID in young children with T1D, some parents continue to experience FH and diabetes distress and may require specialized treatment. REDCHiP is a telehealth-delivered intervention to reduce FH in parents of young children with T1D. Though our trial shows treatment-slope effects that only slightly and nonsignificantly favored REDCHiP parents with respect to FH and diabetes distress, we remain optimistic with respect to REDCHiP’s potential treatment impact. We found that parents’ exposure to REDCHiP was associated with a lower occurrence of elevated diabetes distress over time than ATTN. Also, we observed that parents exposed to REDCHiP reported high treatment satisfaction. REDCHiP aligns with broader trends in pediatric digital behavioral health with its remote delivery, personalization, and integration of evidence-based strategies from cognitive behavioral therapy and behavioral parent training into parents’ daily T1D care [19]. Moreover, because of its telehealth format, it is possible REDCHiP could be integrated into pediatric diabetes care as a first-line treatment, leaving more intensive in-person treatment for families with more complex needs [40]. Future refinements and adaptations of REDCHiP will aim to increase its treatment effect, especially for parents experiencing comorbid depressive symptoms and young children with elevated HbA_1c. We also intend to explore ways to tailor REDCHiP for other caregiver populations, including fathers, grandparents, and teachers. Finally, future adaptations of REDCHiP will explore how to incorporate other digital health delivery methods, such as asynchronous content for self-paced learning and mobile app integration to track and reinforce skill learning to further promote its scalability and implementation.