Participant Characteristics
Table 1 shows the characteristics of the clinicians interviewed and parents surveyed. For the interviews, participants were predominately physicians including pediatricians, family medicine, and medicine-pediatric (84.6%), but other roles included medical assistants (7.7%), nurse practitioners (3.8%), and physician assistants (3.8%); over 75% were females. Approximately 40% of clinician interviewees were from the Boston area, 40% were from the Denver area, and approximately 20% were from the Greenville area. Of the parents who completed a survey, 37.5% were from the Boston area, 37.5% were from the Denver area, and 25% were from the Greenville area. Approximately half of parents reported their primary language spoken at home was Spanish (49.2%) and over 75% of parents reported an annual income of < $50,000.
Clinician Interview Emergent Themes
Despite differences between clinical roles, patient sociodemographics, geographic locations (i.e., urban v. rural), and workflow across the sites, we found thematic saturation after 52 clinician interviews. The findings from the engagement we completed with other key stakeholders (e.g., Quality Improvement Committee, Clinical Unit Chief meetings, Medical Assistant Council) were similar to the clinician interview findings in regards to EHR tools, workflows, and aligning with healthcare organizations’ priorities. The findings from the interviews represented the CFIR domains and resulted in six emergent themes. Table 2 shows the CFIR domains and constructs, emergent themes, and representative quotes.
We asked questions about the CFIR domain of intervention characteristics regarding preferences and needs of EHR flagging and clinical decision support tools and found clinicians wanted evidence-based clinical decision support tools for screening and management that are actionable, integrate into their workflow, and do not detract from patient care or extend visit time. This theme provided important information regarding the CFIR constructs of intervention adaptability and complexity and allowed us to make modifications to the clinical-facing tools to be responsive to clinicians’ needs and preferences while increasing best evidence-based practice for screening and management of childhood obesity.
Two themes emerged when exploring the CFIR domain of outer setting by asking questions regarding needs and preferences for family-facing tools. The themes included (1) family resources should be responsive to the needs of families by being concrete, culturally sensitive, available in multiple languages, and include local resources; and (2) the delivery of the resources should be multimodal to suit the needs of families, clinicians, and staff. These themes highlighted patient needs and resources across the four healthcare organizations and the need to ensure the family-facing tools would be responsive to their needs.
We asked clinicians questions about implementation readiness and their organization’s culture representing the CFIR domain of inner setting and two themes emerged (1) childhood obesity is an important issue and clinicians are open to implementing new programs, but they are aware that competing priorities may detract from this program; and (2) for successful adoption, the program should highlight the importance of improving patient care rather than documentation, and sustainability plans should be addressed early as clinicians have seen other programs fade out. The first theme represented the CFIR constructs of implementation climate and organizational readiness, while the second theme represented the CFIR constructs of implementation climate and culture.
We explored the CFIR domain of process by asking questions about training and implementation strategies and we found clinicians prefer a combination of in-person, individual, and online trainings that are concise, interactive, and case-based that are offered throughout the program duration and provide feedback to clinicians and practice. This theme illustrated the CFIR constructs of engaging, champions, and reflecting and evaluating, and helped elucidate ways to engage clinicians in program adoption, identify key implementation leaders, and understand the role clinician champions should play in the implementation process.
Parent Survey Findings
We completed 400 parent surveys (220 in English and 180 in Spanish) in the Boston, Denver, and Greenville area (see Figure 2 and Table 3). Parents reported their perceptions of pediatric weight management programs, resources, and referrals which aligned with the CFIR construct of patient needs and resources (outer setting) and felt it was extremely or very important that their child’s primary care clinician discuss, make a plan, and provide referrals and resources related to weight management. Referrals to dieticians (30.5%) and resources about food assistance (36.8%) and structured activity programs (29.8%) were reported to be the most helpful. Parents reported their preferences for delivery of resources and likelihood of accessing resources which aligned with the CFIR construct of adaptability (intervention characteristics). We found approximately half of parents preferred after-visit summaries to be printed while at their child’s visit and approximately half of families wanted other resources, such as educational materials and other content, to be texted to them. Parents reported being willing to download an App (59.5%) or visit websites (62.3%) to find additional information about behavioral changes. The questions about preferences for the text messaging program and mobile phone usage and behaviors aligned with the CFIR constructs of adaptability (intervention characteristics) and patient needs and resources (outer setting). We found nearly three-quarters of families thought it would be helpful to receive text messages with behavioral change tips and reported being willing to follow links to access additional content. Approximately 30% of families reported running out of data on their mobile phone. The findings from engagement with parent stakeholders through family and community advisory councils were similar to the findings of the surveys. For example, parents expressed interest in the educational topics covered in the program and preferred multi-modal delivery of materials and resources. Findings across sites were similar and differences did not drive any modifications.
Connect for Health Program Adaptations
Following the mixed methods assessments, we made several adaptations to the core program components and to the implementation strategies. We found the results of the engagement across all stakeholders converged which strengthened our modification decisions. As a research team across all four organizations, we collectively reviewed results of the clinician interviews, parent surveys, and other stakeholder engagement activities. Based on the findings, we made modifications at the program level and then each healthcare organization identified additional changes for their setting. During the pre-implementation phase, the research team met monthly via video conferencing, as well as an in-person meeting to review and discuss program adaptations. Across the core components, we identified cultural and geographic, mode of delivery, and target audience adaptations that were necessary due to differences in geographic location and patient populations as compared to the original trial, differences in organizational culture and clinical workflow, and in consideration of future program scalability (14). Table 4 shows the nature of the content modifications, level of delivery of the modifications, and by whom made the modifications using Stirman and colleagues’ adaptation classification system (15,16). This adaptation classification system was used to assist with future interpretation of patient, service, and implementation outcomes, and sustainability and scalability planning.
Based on the interviews, we learned each of the healthcare organizations had their own unique practice workflow and had customized EHR instances, therefore, each site modified the flagging of children with an elevated BMI to fit their needs. For example, changes included creating a non-interruptive best practice alert and having medical assistants receive alerts rather than the physician. Changes to the clinical decision support tools included housing family-facing tools within the EHR and creating an order to enroll patients in the text messaging program that would support future program sustainability. We aligned changes to the best practice alert and clinical decision support tools with performance metrics at the healthcare organizations to incentivize the usage of the EHR tools and for sustainability purposes.
Modifications to the family-facing tools (i.e., patient educational materials, community resource guide, and text messaging program) included translation of materials into languages spoken by the healthcare organizations’ patient populations, revision to the materials to meet the needs of patients in urban and rural settings and geographic locations across the United States, and consolidation of educational materials for ease of delivery. Additional changes included a handout focused on creating a balanced nutrition plan and modifying the text messages to be unidirectional rather than bidirectional. Several of these modifications (i.e., consolidation of materials, unidirectional text messages, elimination of health coach) were made at the program-level in consideration of program sustainability and scalability.
Modifications to the implementation strategies included selecting a clinician who was embedded within the clinical practices to engage other clinicians and champion the program, adding a practice coach to provide technical assistance, and inviting all staff to in-person trainings and extending them throughout the implementation phase. Based on the interest in on-demand trainings, we also added a virtual learning community for capacity building of childhood obesity screening and management. For the in-person trainings and virtual learning community, we intend to offer continuing educational units and quality improvement bonuses (in collaboration with other hospital departments as available) to incentive participation and increase the likelihood that clinicians will be familiar with the program. Modifications to the implementation strategies were made at the program level, but each of the healthcare organizations were encouraged to customize as needed.