Engaging Caregivers and Providers of Children With Sickle Cell Anemia in Shared Decision Making for Hydroxyurea: Protocol for a Multicenter Randomized Controlled Trial

Background Sickle cell anemia (SCA) is a genetic blood disorder that puts children at a risk of serious medical complications, early morbidity and mortality, and high health care utilization. Until recently, hydroxyurea was the only disease-modifying treatment for this life-threatening disease and has remained the only option for children younger than 5 years. Evidence-based guidelines recommend using a shared decision-making (SDM) approach for offering hydroxyurea to children with SCA (HbSS or HbS/β0 thalassemia) aged as early as 9 months. However, the uptake remains suboptimal, likely because caregivers lack information about hydroxyurea and have concerns about its safety and potential long-term side effects. Moreover, clinicians do not routinely receive training or tools, especially those that provide medical evidence and consider caregivers’ preferences and values, to facilitate a shared discussion with caregivers. Objective The aim of this study is to understand how best to help parents of young children with sickle cell disease and their clinicians have a shared discussion about hydroxyurea (one that considers medical evidence and parent values and preferences). Methods We designed our study to compare the effectiveness of two methods for disseminating hydroxyurea guidelines to facilitate SDM: a clinician pocket guide (ie, usual care) and a clinician hydroxyurea SDM toolkit (H-SDM toolkit). Our primary outcomes are caregiver reports of decisional uncertainty and knowledge of hydroxyurea. The study also assesses the number of children (aged 0-5 years) who were offered and prescribed hydroxyurea and the resultant health outcomes. Results The Ethics Committee of the Cincinnati Children’s Hospital Medical Center approved this study in November 2017. As of February 2021, we have enrolled 120 caregiver participants. Conclusions The long-term objective of this study is to improve the quality of care for children with SCA. Using multicomponent dissemination methods developed in partnership with key stakeholders and designed to address barriers to high-quality care, caregivers of patients with SCA can make informed and shared decisions about their health. Trial Registration ClinicalTrials.gov NCT03442114; https://clinicaltrials.gov/ct2/show/NCT03442114 International Registered Report Identifier (IRRID) DERR1-10.2196/27650

shared decision making, hydroxyurea offered) and improvement in decisional outcomes (decisional uncertainty, HU knowledge, satisfaction) (see Figure 1 Conceptual Model). If parents feel more confident, less uncertain, more knowledgeable about hydroxyurea, and involved in the decision to initiate hydroxyurea, then they are more likely to initiate hydroxyurea and ensure their child is adherent with the medication (see distal outcomes in Figure 1). Furthermore, if more children with SCD are offered hydroxyurea, hydroxyurea uptake should increase. Increased HU uptake means that more children with SCD would benefit from this disease-modifying treatment, and subsequently, experience less neurological impairment and better quality of life (distal outcomes - Figure 1). Ultimately, increased hydroxyurea uptake would result in fewer ill visits, emergency room visits and hospitalizations (distal outcomes - Figure 1). We have included this information on page 7 and revised the entire research plan to better reflect our conceptual model. It is unclear how many sites will be participating in the study. On page 9 of the application under population/settings it states "one of 9 clinics" participating. However, in the following paragraph for study design it states that there will be 4 clusters with each cluster consisting of 2 sites for a total of 8 sites. Then again on page 16, 9 sites are listed in Table 3. Participating Sites. On page 15 of the application it states data from the 9 th site would be used if another site cannot meet recruitment goals. Please provide clarification on how the 8 sites for initial data collection will be selected, when data collection will begin at the 9 th site, and what the criteria will be for including the 9 th site if needed.
Our initial plan was to have 8

Use of some of the RE-AIM constructs are incorrect throughout the application.
We have revised the evaluation of study using the RE-AIM framework below so that the constructs are aligned with the correct data being proposed for collection/assessment.
• On page 11 of the application it states that "we chose an Adoption-related primary dissemination outcome: parent report of shared decision-making." Participants reporting of whether shared decision-making occurred would be considered efficacy/effectiveness of the intervention. Adoption is the proportion and representativeness of the settings (in this case clinics) and the staff who are willing to initiate the intervention(s).
This section now reads: "Guided by the RE-AIM model and with input from our clinician and advocacy stakeholders, we chose an Efficacy/Effectiveness-related primary dissemination outcome: parent report of shared decision-making."

Secondary Outcomes
Parent reported Satisfaction with decision making 3 item survey -E* adapted from the empirical research related to the concept of procedural justice (PC-3). 5 If the Cronbach's alpha for these items is acceptable (≥ .70), ratings will be summed to obtain a total score; otherwise, items will be analyzed separately.  6 If the Cronbach's alpha for these items is acceptable (≥ .70), items will be summed to obtain a total score; otherwise, items will be analyzed separately. ( • On page 16 of the application under D.1. Subgroup analyses #2 -characteristics of drop-outs versus completers. This is categorized as effectiveness. This is should be included in understanding reach.
• On page 17 -#4 -It states that characteristics of sites who adopt the full H-SDM toolkit verse the core components is considered Implementation. This is adoption. Implementation should focus on consistency of delivery of the intervention protocols.
• For item #5 on page 17 -which is classified as maintenance, there is no indication of when this assessment would be conducted. This is typically done 6 months or greater after the intervention has been completed. Please provided greater detail as to how maintenance will be evaluated.

At CCHMC, we used counseling in cases of influenza vaccine hesitancy as a proof of concept to assess the impact of virtual reality training on communication skills related to motivational interviewing. This training resulted in a statistically significant decrease in rates of influenza vaccine refusal among
providers that underwent this virtual reality curriculum when compared to those that did not. 13 Additionally, learners described the training as realistic, immersive, and fun.  15 We have included this information on page 8 of the application.
We have a summer research fellow who will be working on developing cases and mining qualitative data for phrases from existing parent interviews, provider interviews and clinic visit observations. This work will be completed using institutional funds. The research and VR team will be meeting in August to finalize the cases. Our STORM engagement group will be invited to these meetings, but if not convenient for engagement group members, we will review the cases and language during existing STORM engagement group meetings. Please see the Finalize the cases and language for virtual reality training August 2017 Pilot the virtual reality cases for content and messaging with key stakeholders including physicians, nurses, health educators and families and amend accordingly September 2017 Create the shared decision making cases in the virtual reality platform, modified from the virtual reality created for the prior influenza study 13 October 2017 Pilot the actual virtual reality scenarios to determine issues prior to clinical implementation November 2017 Update the virtual reality scenarios based on pilot information to have the system ready for implementation December 2018 6. A few concerns were raised related to the sample size calculations. Please address.
• The application needs to provide justification for the sample size calculations. Did the calculations use the pilot data? Why is 0.6 a meaningful difference in this context? An ICC of 0.00001 is not very realistic and too low. For the ICC please provide preliminary data that supports that this estimation is correct or provide a power analysis table which should include a range of effect sizes, the variance of the effect, the level of significance, and minimum sample size given these expectations.

Sample size calculation (CI-2; CI-4; GM-5; CI-5):
We based our sample size calculation on minimal effect sizes basedon studies of the DCS (effect sizes range from 0.4 to 1.2), 1 our primary outcome, and a stepped wedge design (Hussey and Hughes approach) 16,17 . Specifically, power analyses were calculated using Optimal Design 18 Table). • There is no discussion of or accounting for attrition. 7. There were a few concerns related to the analytics for this application. Please address.

Participant attrition is an inherent problem in clinical trials
• The only model considered is a linear mixed model, which is not suitable for categorical data such as hydroxyurea offered.

Data Analysis for Aim 1 (GM-3; CI-4; CI-5; MD-5):
The data will be analyzed based on the intent-to-treat principle. All patients will remain in the arm of the study to which they were randomized, regardless of whether or not they receive the assigned dissemination method.
We will report our findings using the Consolidated Standards of Reporting Trials (CONSORT) statement extension to cluster randomized trials 20  • It is not clear why and how certain sites are clustered together.
We will randomly assign sites to the 3 clusters (3 sites per cluster). The randomization will guarantee that each cluster has a site considered large, medium and small (<270 patients = small; >270 -500 = medium; >501 = large). This information is now included on page 17 and throughout the application.
• The effect size characterization for the toolkit testing is incorrect for two of the three outcomes in the toolkit pilot testing.
In our pilot study, we tested the H-SDM toolkit dissemination method with clinicians of parents newly facing the decision to initiate hydroxyurea locally and at a second Midwest SCA clinic (intervention group, n = 27). Parents completed the 16-item decisional conflict scale (DCS) to assess decisional uncertainty 25 and a 9-item survey to assess hydroxyurea knowledge specific to content of the decision aids (highest score 9). We evaluated acceptability using a 10-item survey used in previous studies 26 . Concurrently, parents from a third pediatric SCA clinic in the West whose clinicians used usual care completed the DCS and hydroxyurea knowledge measures (control group, n = 20).

We examined changes in DCS scores and hydroxyurea knowledge with repeated measures
ANOVAs for the intervention group (pre and post). All parents using the decision aids reported that they were useful in decision-making (100%). Hydroxyurea knowledge increased and decisional conflict decreased pre to post using the decision aids (Table 3).
We used an independent samples ANOVA to determine effect sizes (Cohen'