This study takes a qualitative descriptive approach [21,22] to understand self-management practices among CMC. The qualitative descriptive approach focuses on generating a comprehensive summary of events with presentation of interpretive categories in everyday language [22]. Given that qualitative descriptive studies may begin with an overarching framework [21], data collection and analysis will be guided by the combined theoretical framework described above. However, both data collection and analysis will remain open to the data; in other words, it is possible that the concepts of interest will change over time [21,22]. Data collection will consist of semistructured interviews with children, parents, and social network members. Data analysis will consist of a combination of inductive and deductive methods of qualitative thematic analysis [23], which will use data generated by interviews and application of line-by-line codes; relevant categories to generate final themes will then be used describe the phenomena of interest. Relevant categories and themes will be analyzed at both individual and multiadic levels (ie, intragroup) [21,22] to understand similarities and differences within each case (ie, child) in terms of the various facets of the family and social networks and across cases regarding the same. Further, multiadic analysis means that relevant comparisons and contrasts will be made across similar thematic domains within each case. We expect there to be a diversity in disease representations but a shared experience in self-management burdens and difficulty in navigation between a multitude of settings.

Eligible families will initially be approached to participate in this study from an academic children’s hospital that provides specialist inpatient and outpatient services, including a pediatric complex care clinic. Each family will be comprised of a child between 0 and 21 years of age with diagnoses and/or a clinical presentation consistent with the definition of medical complexity, as well as the child’s parents. Operationalization of CMC will follow the Center of Excellence on Quality of Care Measures for Children with Complex Needs (COE4CCN) definition of the highest level of medical complexity of children with complex chronic disease [24]. This definition includes any of the following scenarios: significant chronic conditions in two or more body systems where the physical, mental, or developmental conditions have lasted at least a year; use of health care resources are above the level for a healthy child; and treatment is required for control of the condition. In addition, the conditions are expected to meet one of the three following descriptions: episodic or continuously debilitating, a progressive condition that is associated with deteriorating health with a decreased life expectancy in adulthood, child requires continuous dependence on technology for at least 6 months, or a progressive or metastatic malignancy that impacts life function. Maximum variance sampling [25] will be used to recruit participants; this will be based on condition as well as demographic characteristics shown to impact child self-management, including parent and child gender, age, race or ethnicity, geographic location, health literacy, educational background, and health status. This sampling strategy will enable us to gain insight into a wide range of established approaches to self-management in children with medical complexity.

If a child is unable to provide assent or consent, only the parents will be included. Furthermore, if only one parent is present or willing to participate, only this individual will be included. Thus, a family is eligible to participate in this study in any of the following combinations of individuals: a child and two parents, a child and one parent, two parents, or one parent. In this study, we refer to a social network set as the individual social networks of each family member. Thus, a social network set could be comprised of only one social network (ie, one parent’s network), two social networks (ie, both parents’ networks or one parental and one child network), or three social networks (ie, both parental networks and one child network). Together, a family and their social network set comprise a case. Social network members will be selected by participants based on prompts associated with the convoy model instrument [10] and may include individuals such as neighbors, friends, members of the community, and members of online social networks, in addition to family members [10]. During the consent process, it will be explained that the investigators are interested in contacting members of the participants’ social network for potential inclusion in the self-management study. Following completion of the parent interviews, participants will be asked for contact information for potential social network recruitment.

Semistructured interviews (see Table 1) will be conducted and will consist of four overarching domains. Part 1 of the interview will involve opening questions where both parents and children will answer open-ended questions about their daily routines. Part 2 of the interview process will elicit questions surrounding how self-management responsibilities are distributed and negotiated among the social network. This section of the interview will provide an exploration of participants’ social networks and the ways in which the network is engaged in self-management processes. Participants will be guided through a structured process of explicating their affective social networks [26]. They will be asked a series of open-ended questions to understand the characteristics of each social network member and their general relationship with the social network member. The interview will then transition to self-management. Participants will be asked about their own roles in specific self-management processes and the role of each social network member. Self-management processes include health-related domains of the following: therapeutics (ie, medications, technology dependence, and therapies), care coordination (ie, appointments, information exchange, logistics like transportation, and home activities of daily living), and management of unexpected health events (ie, side effects, unexpected illnesses, and unexpected hospitalizations).

Once each social network member has been discussed, participants will be asked if there is anyone else who helps with their own or their child’s self-management whom they did not place in their social network. Participants will be asked whether or not these individuals should be placed in the social network and asked questions about each person similar to those above. Finally, participants will be asked to reflect on their network as a whole and the ways in which they engaged this network in specific self-management processes.

Part 3 of the interview will provide an exploration of participants’ work systems and the ways in which work system elements shape self-management. Participants will be asked to think of times in the past when they were engaged in specific self-management processes. Participants will then be asked to reflect on how different aspects of their work system contributed to these experiences; these aspects include the following: personal attributes (eg, literacy, health literacy, education, socioeconomic status, and age), tasks (eg, complexity and frequency), technology (eg, access to and usability of personal health records, automated reminders, and health-related apps), physical environment (eg, clutter and noise), social environment (eg, support from social networks inclusive of individuals such as family members, friends, home care aides, and health care providers), community environment (eg, public transportation), and health policy environment (eg, insurance, characteristics of the health care delivery system such as financial assistance, and care coordination practices) [16]. A similar structure will be used to understand the challenges that participants have faced when engaging in the same self-management processes.

Interviews with social network members will also consist of four parts. In addition to opening and concluding questions that mirror those above and include demographic questions, the following questions related to the two aims will be asked. Questions will be oriented to the individuals that placed the social network member in their network (eg, child only or both parents and child). Social network members will be asked general questions about their relationship with the child and/or parents. They will also be asked where in their social network they would place the parent or child. Social network members will then be asked about their role in managing the child’s health, how they assumed that role, and how they feel about that role. Social network members will be asked about times in the past when they were involved in managing specific aspects of the child’s health and felt that things went well. They will then be asked to reflect on how different aspects of the parent or child’s work system (as described in Table 1) contributed to these experiences. Similar questions will be asked for when things were not perceived to go well. Follow-up questions will be skipped if a social network member does not engage in a specific aspect of managing the child’s health.

Analyses for both aims will begin with cross-sectional analysis, will move to within-case comparisons, and will end with across-case comparisons. These analyses will build on one another. That is, the cross-sectional analysis will be used to generate prevalent analytical categories and themes within the data. Within-case analysis will then be used to determine how these themes manifest and interact within each case. Finally, across-case analysis will be used as a higher order cross-sectional analysis that will allow for a meta-analysis of how these categories and themes manifest and interact across families and social networks.

Two types of cross-sectional views will be created: one drawn from all parent and child data and one drawn from all social network member data. In addition, analyses will be conducted both by self-management process and in aggregate across all self-management processes. Regarding the parent and child view, we will begin by describing parent and child social network structure and composition. Descriptive statistics (eg, demographics of social network members) and conventional qualitative content analysis [23], in which themes are derived from data (eg, relationship with social network members), will be used to develop cross-sectional views of these social networks. This latter method will also be used to analyze open-ended questions about the nature of self-management responsibilities and how they are distributed and negotiated among social network members. We will then use directed qualitative thematic analysis [23], in which operational definitions for initial categories and subcategories are determined using theory. The work system model and relevant empirical studies will be used to develop initial categories of barriers and facilitators. Regarding the social network view, analysis will parallel that described for the parent and child view, resulting in an aggregate view and views by self-management process. As described above, conventional qualitative content analysis will be used to analyze open-ended questions.

Methods of multiadic analysis will be used to create two types of within-case comparisons [27,28]. The first type of within-case comparison will occur at the level of the family. When two or three individuals participate from one family (ie, both parents, both parents and child, or one parent and child), multiadic methods will be used to compare the responses provided by each individual. In particular, points of overlap and difference between viewpoints will be explicitly coded using a combination of directed and conventional qualitative content analysis (ie, deductive and inductive methods). Drawing on Eisikovits and Koren’s framework [28], we will first code into four themes: (1) overlap on open and hidden reality, (2) contrast on open and hidden reality, (3) contrast on open reality and overlap on hidden reality, and (4) overlap on open reality and contrast on hidden reality. Open reality refers to what is described by individuals (eg, the event that occurred), whereas hidden reality refers to the ascribed meaning (eg, how the event was interpreted). This four-dimensional framework will be extended in the case of the participation of three family members. Within this framework, categories drawn from the data will be created to account for types of open and hidden realities. The second type of within-case comparison will occur at the level of the social network. A dyadic analysis will take place comparing the narratives of the family member with the narratives of each of their social network members (eg, What does the mother say about her own role and her sister’s role, and what does her sister say about her own role and the mother’s role?). This analysis will occur as described above. Another iteration of analysis will then be conducted to reduce the dyadic analyses into a synthesized view for each family member. For example, for each of 10 individuals in a child’s social network, a dyadic analysis will be conducted that will result in material under each of the four themes. The next round of analysis would consist of a qualitative content analysis of all of the material within each theme. Thus, the final analysis for each social network set would consist of an aggregate view of the four themes in Eisikovits and Koren’s framework across all social network members. One aggregate analysis will be created for each family member.

A final structural layer of analysis will be added with across-case comparisons. To create comparable analytic instruments that further reduce the data, we will create individual case reports [29,30]. These reports will be narrative in nature and will highlight key characteristics of each case (eg, number of family members involved, family demographics, and key themes from cross-sectional, dyadic, and multiadic analyses). Case reports will be comparatively analyzed using conventional qualitative content analyses. Guiding questions relevant to this analysis will include the following: (1) Which approaches to distributing and negotiating self-management responsibilities are common and distinct across families and their social networks? and (2) How do families and social networks with apparently similar characteristics differentially engage in distributing and negotiating self-management practices? Analyses guided by these questions will enable a nuanced understanding of mechanisms driving specific self-management experiences within and across various social networks. For all analysis steps, all investigators will individually analyze 20% of the data before reaching a consensus. Consensus guidelines will be documented for use in subsequent analyses. The remaining analyses will be divided between the principal investigators, who have in-depth experience with both forms of qualitative content analysis to be used in the study. Analysis will be discussed at regular full-team meetings. All analyses will be conducted using NVivo 11 (QSR International). Reflective journal entries will be maintained by the lead investigators during the research process. Methods to ensure rigor and trustworthiness [31] of the data analysis will include aspects of demonstrating (1) credibility (ie, direct observation of online communication, iterative questioning, and frequent debriefing), (2) transferability (ie, contextual review), (3) dependability (ie, maintaining an audit trail), and (4) confirmability (ie, bracketing, investigator triangulation, and member checking through the stakeholder advisory board). The key decision points for the qualitative research analysis will also be assessed and recorded in an online platform through the Center for Open Science to ensure rigor and reproducibility. Anticipated limitations of this study design include the following: lack of generalizability based on recruitment from a single-site study and potential for recall bias with self-reported data. Additionally, we lack the ability to triangulate self-management experiences within the school or other immersive environments.