What is the state of children’s participation in qualitative research on health interventions?: a scoping study

The question that guided our study was: What is the state of participation of children ages 6–11 years in health intervention research? To formulate our question, we drew on the PCC (Population, Concept, and Context) elements of designing a scoping study question [37]. Our population of interest was children ages 6 to 11 years. The concept we employed was health intervention research, defined broadly to include empirical research on interventions with children, including research on programs, processes, and guideline and including all points on Proctor’s conceptual model of implementation research from implementation strategies to outcomes [38]. In selecting search terms we recognized that the language of participatory research rarely enters the lexicon of intervention science where studies may nonetheless include children as stakeholders. We therefore elected to use broader search terms while focusing on OVID Medline in order to balance a higher tolerance for terminological variation with the time constraints of manually assessing a higher number of articles by the inclusion criteria.

Research with children was defined as research in which children are “actively participating and expressing their views and opinions” [39]. We focused only on qualitative studies for this review because of the increasing emphasis on the value of qualitative data for intervention development, adaptation, and evaluation [2, 40, 41]. The context was open, including all geographical, healthcare, and sociocultural contexts.

We worked with a medical librarian and conducted a comprehensive search of Ovid Medline for literature on child participation in qualitative health intervention research. The search (completed July 3, 2018) encompassed a decade (January 1, 2007 to July 2, 2018) for breadth. Medline was selected as it is the prominent bibliographic database of life sciences and biomedical information and the most appropriate database to search for health intervention research. The decision to focus solely on Medline was made in conjunction with our medical school’s librarian who has expertise in scoping and systematic reviews. She identified that, because our goal was to map the health intervention research, a search of Medline was most strategic. It would capture the state of the research and also how researchers in the medical and health sciences typically search scholarly work to inform their interventions. The search included the following keywords: “(children) AND (health intervention) AND (participation) AND (qualitative research),” as well as the corresponding Medical Subject Headings terms, restricted to middle childhood (6–11 years) (see Supplementary 1). Grey literature was not searched as we focused on intervention research that is published in peer-reviewed journals to understand the state of the scholarly field.

To determine study eligibility, the following criteria were applied: children 6–11 years old; published from 2007 to the present (date of search – July 2, 2018); included a health intervention; study designs were qualitative; and studies were published in English. We included studies that had children below age 6 and above age 11 if those studies also included children in our target category (6–11 years). We did this because study participant samples often do not map directly on to specific age ranges. We focused only on qualitative studies for this review because there is increasing emphasis on the value of qualitative data for intervention development, adaptation, and evaluation [1, 2]. Studies were excluded if they were unpublished studies, review articles, or not focused on health.

Search results were screened in EndNote X8. The selection of sources of evidence was undertaken by two authors (JH, CM) for both title and abstract screening and full text review. To ensure that we did not miss articles for relevant inclusion criteria, such as a qualitative components as part of a range of methods, we did full text reviews of all articles in which authors did not definitively detail participants or methods in their abstracts. All full texts were read to identify if they fit our definition of intervention research. We defined pre-intervention and/or pre-implementation as any article that made it clear that intervention creation and/or implementation was the next step (eg. this included studies reporting on a needs assessment). We made this decision to limit the ambiguity and author guessing about what types of study could be useful for intervention development or implementation. JH and CM independently reviewed and met to iteratively ensure agreement, and JN resolved any differences in agreement until consensus was reached on all articles.

Relevant data from the included articles were charted by the research team (JH, CM, JN, and 3 research interns) and recorded in a Microsoft Excel spreadsheet. JH and JN created categories for the chart and refined the categories in an iterative process of testing the data chart. The finalized chart included: author, title, year, research country, authors’ countries, age of youngest child, age of oldest child, adults included in research, family included in research, rationale for including children, aim of paper, when in the implementation process children participated in research (i.e., methods, methodological adaptations, participatory research, meaningful participation, arts-based research, ethical issues discussed, power dynamics discussed), and a range of categories for conditions and social issues covered in the research (e.g., disability, cancer). All team members extracted data and 90 (78%) articles were double extracted with JH and JN resolving disagreement through discussion at weekly meetings. A critical appraisal of individual sources of evidence was not conducted for this review.

Following Arksey and O’Malley [27], we used descriptive and content analysis as our framework for examining our findings. First, descriptive analysis mapped the studies, focusing on the dissemination of the studies, geographical distribution of studies and researchers, and the types of interventions. Second, we used a combination of descriptive and content analysis [42] to examine age and research methods used. For example, we coded not only for the protocol related to age and method, but also examined how authors discussed age and how/if they reported on findings by age and how they discussed any challenges or benefits of methods used with children.

We coded articles’ rationales for children’s participation to understand why children were included in the research. To minimize subjective inferences, we only coded authors’ explicit statements of why children were included. We documented what methods were used with children and the descriptions and justifications for these methods to understand what constituted research with children for the researchers.

To examine meaningful participation, we used Roger Hart’s distinction between participation and non-participation [14]. Hart developed a widely-used Ladder of Participation, which has 8 rungs. Rungs 1–3 cover what Hart refers to as non-participation (e.g., token participation, participation only as decoration, and participation manipulated by adults). We use the term “limited participation” to describe studies that fell into rungs 1–3 or for which there was not enough information available to assess participation. We defined meaningful participation as any study that fell in Hart’s category of participation, rungs 4–8. Meaningful participation covers many ways of including children from having them assigned roles by adults to children participating in decisions about the project. We included this wide range of participation within the category of “meaningful” because we recognize that although the ladder metaphor suggests a hierarchy of value, childhood studies scholars have critiqued assumptions that higher rungs of the ladder necessarily produce higher quality or more ethical research [43]. Through discussion of a sub-set of articles and drawing from childhood studies theory, we developed a basic schema for evaluating each study based on Hart’s ladder. Our evaluation schema incorporated considerations of children’s meaningful inclusion at various stages of the research process, including sampling, study design, analysis, reporting, and application of findings. We used only the information available to us from the articles for our assessment and did not reach out to authors for additional information. Because studies were heterogenous and there are many dimensions to children’s meaningful inclusion, we applied the schema flexibly and with consideration for the research goals. Three co-authors trained in childhood studies (JH, JN, JS) coded studies according to the best-matched ladder rung and provided brief justification for the assigned code. A sub-set of articles, including all articles assessed as below level 3 (limited participation), were double coded with high levels of agreement between coders. While coding the articles, each coder kept memos on the article related to the coding schema, and we used these memos, alongside our other findings, to develop a summarizing reflective guide. We note that this guide is not a validated tool but rather a summary of the key issues we identified.

For an overview of children’s participation in intervention research, we documented what types of health issues were targeted, where the research took place, when children were involved, who participated in the research and how they participated. Most studies targeted disease prevention (n = 77, 67.5%), which included physical activity and nutrition interventions, tobacco and other substance use prevention, HPV vaccination, and HIV prevention. Physical activity, nutrition, and obesity prevention interventions were the most prominent, making up 39.5% of the total sample of articles (n = 45). Only a quarter of the reviewed studies (n = 30, 26.3%) covered illnesses that children actively dealt with in the present (e.g., cancer, cystic fibrosis, diabetes). Even fewer focused on children living with and/or managing the care and health of family members (n = 7, 6.1%). Perhaps related to the topical foci, most studies took place in high-income countries, such as the United States (n = 41, 36.0%) or the United Kingdom (n = 23, 20.2%). Few studies (n = 18, 15.8%) were carried out in low- or middle-income countries (e.g., Brazil, India, Zambia).

When engaging children in the 6–11 year age group, the most common qualitative methods used were focus groups (n = 74; 64.9%) and interviews (n = 47, 41.2%), with lesser-used approaches including observational (n = 22, 19.3%) and arts-based methods (n = 16, 14.0%). Forty-two (36.8%) used more than one qualitative approach (e.g., both focus groups and interviews). We discuss trends in meaningful participation related to age and method in detail below. For more details on study characteristics, see Table 1.

Seventy-four articles (64.9%) included the authors’ rationales for including children. We grouped rationales into two inductively developed categories to classify the authors’ intents. The first category linked their rationales to children as inhabiting a distinct social category and to children’s “unique worldviews and perspective” that had been “dismissed,” “missing,” or “excluded” from previous intervention development and implementation [44‐46]. Forty-six articles (40.4% of the total sample; 62.2% of the articles stating a rationale) fit this child-focused category. The second category focused on child participants as “stakeholders,” “target populations,” “consumers,” “patients,” “community members,” and “intervention users,” and included 28 articles (24.6% of the total sample; 37.8% of the articles with rationales). In contrast to the first child-focused category, this stakeholder category made no mention of children’s age category or of childhood as a social category as part of the rationale for including them.

When compared, these two categories had similarities and differences. Both sets of rationales focused on concepts of voice and empowerment as a central reason for including children as participants. This focus on voice and empowerment, however, seemed to draw from different literatures. The latter category that focused on stakeholders and target groups was aligned more with a disease and health focus that places emphasis on learning from those who experience a particular health issue. However, the emphasis in the former child-focused category (i.e., childhood as a social category) was on childhood status as disempowering, and those researchers were more explicit about addressing age-related power dynamics during the research process. That is, the researchers sought to critically consider children’s position in their society in relation not just to the topical focus, but also in relation to the researchers. Most explicitly, this set of articles focused on methodological adaptations to facilitate children’s inclusion by disrupting common ways of interacting between children and adults that might discourage children’s participation or feelings of inclusion. Some adaptations included inviting children to lead group discussions [47], sit at eye level [48], or address adults by their first names [49]. Additionally, this set of articles was more likely to incorporate participant observation [50] and use task- and arts-based methods (e.g., drawing, photography, performance) [51].

Using Hart’s distinction between meaningful child participation and non-participation [14], we assessed children’s participation as meaningful in 83 studies (72.8%). We identified two primary ways that researchers in the meaningful participation category engaged children in intervention research. The first approach was more structured, with researchers asking children directed questions aimed at the adaptation and evaluation of intervention materials. Children participated in content design, assessed the content of messaging, and offered their opinions on what they liked or disliked and why [46, 52‐56]. In the second approach, researchers focused more broadly on identifying children’s knowledge and experiences of particular health topics and interventions. Such work was often based on interpretivist paradigms in qualitative research, often with a child-focused rationale and using open-ended interviews, task- and arts-based methods, participant observation, and multiple methods (e.g., see [44, 51, 57‐61]). For example, Meininger et al. [62] examined how children of different ages experienced and perceived food and physical activities, and then they drew on their perceptions to tailor an intervention. Bond et al. [51] identified that children in Zambia were wary of an intervention that invited children to promote tuberculosis testing in their households, because of the stigma related to tuberculosis and children’s subordinate positions in relation to adults, but that the children also wanted to be involved. They tailored their intervention to take into account TB stigma, children’s positions in households, and children’s desires to participate. Thus, research with children could directly shape delivery of interventions to be more sensitive to children’s experiences or to understand the uptake of interventions during and after completion. Supplementary Table 2 offers examples of the range in ways that authors meaningfully engaged children.

The 31 (27.2%) remaining articles categorized as limited participation were informative of when and how children’s participation may get tokenized. Many did not include enough information to assess the meaningfulness of children’s participation, and thus we could not classify these articles in the category of meaningful participation. Some common reporting approaches that made it difficult to assess children’s meaningful participation included a lack of information about the methods used with children and the exclusive use of adult and youth (> 11 years) quotes and responses in results and discussion sections, without any mention of the younger children in those sections. Additionally, many of the articles in this category were studies with participants of different ages and roles (e.g., child and parent, student and teacher). Articles in this limited participation category also conflated child and adult responses, using statements such as, “adults and children said,” making it difficult to ascertain how much children informed findings or conclusions.

Assessing the content of the articles in the limited participation category brought to light limitations we were also identifying in articles in the meaningful participation category. Notably, we also identified this “adults and children” conflation in reporting results in articles in the meaningful participation category, albeit to a lesser extent. To offer a further example, we identified a strong trend within the meaningful participation category to not report or discuss younger children’s quotes or other data (e.g., drawings and their explanations). As reported above, numerous studies included participants above the age of 6–11. The perspectives of older children (i.e., close to and above 11 years) and adults were more often reported than those of the children in our scoping review’s age range. Some researchers reported why they focused more on older age ranges, often identifying methodological challenges in working with younger children in their samples. For example, Laroche et al. [63] began over-sampling children above age 12 during recruitment because, in their perspective, they “express [ed] themselves more clearly than younger children” (p. 428).

The omissions of children’s quotes and other data from results and discussion sections of studies in our review, even those with meaningful participation, might suggest that many researchers also found difficulties including and interpreting the perspectives of children in middle childhood. Hieftje et al. [44] noted that children in their study produced unanticipated and difficult to interpret responses. However, they argued against disregarding such responses: “Although researchers might intend for an activity to open discussion about a particular topic, young adolescents may use the activity in a different way … The data that subsequently emerge … provide poignant insights into young adolescents’ beliefs, attitudes, and behaviors” (p. 720). This suggests the need to further efforts and upskill researchers in how to not just carry out methods with children, but also analyze the data produced by children (e.g., quotes, stories, drawings, performances, and other artwork). Finally, we note in Table 1 an increase in annual number of publications over the review time period (2007-July 2, 2018). However, we did not identify an increase in meaningful participation through this time period, which further supports the need for upskilling.

The review revealed clear patterns in the time point at which children were engaged in the research on interventions as well as in topical and geographical areas. First, most articles did not seek children’s input early on in the research/intervention development process; at best they engaged children during or immediately prior to implementation. As we showed above, some researchers who engaged children in pre-implementation research gained valuable input from children as to how to adapt their interventions to children’s lives and to children’s strengths and needs [43]. Without this initial involvement, researchers miss an opportunity to improve interventions and their implementation. They may in fact introduce interventions that do not address children’s needs or that could cause harm to children.

Second, there was a strong trend in the focus of interventions, with most focusing on disease prevention, especially on physical activity and nutrition interventions. While this focus is important given the range of health issues throughout the lifespan and the link between childhood obesity and obesity in adulthood [64, 65], a critical perspective from childhood studies offers a different insight on this topical trend. Childhood researchers in the social sciences have identified a tendency for researchers to view children in terms of their futures as adults, often focusing on future needs, health, and productivity [66]. They have shown that viewing children as “future adults” can limit researchers’ understandings of children in the present and can miss out on issues, including health issues, that children find meaningful and that actively affect children now [67, 68]. Our review raises the question: Are such interventions more common because the adults funding and designing studies are viewing children in future terms? What effects might such views have on the interventions? What types of health issues might children focus on and what are we missing out on related to children’s illness experience during childhood?

Third and relatedly, we note that most studies reported on research with children living in high-income countries. As such, we are missing important perspectives from children on the topics covered in the review that affect children in both high- and low-income countries and on specific health issues that children may face in low-income countries (e.g., tuberculosis, waterborne illnesses). This is likely indicative of global inequities in academia, where researchers working in high-income countries have greater access to research funding to carry out work, and to the institutional and social resources to publish such research in the journals indexed by Ovid Medline. However, given that childhood studies is based on the premise that childhood is not singular, but experienced in context, developing an understanding of children’s involvement in intervention research will be at best partial if it is only done with certain populations living in certain contexts [68, 69]. Further, as global health research and interventions increase, we would warn researchers, specifically those from high-income countries who work in low-income countries, that mapping understandings of childhood in one place onto those in another could lead to ineffective or detrimental interventions [70].

Several gaps we identified relate to a need to improve methods and the interpretation of findings within qualitative approaches to intervention research. First, many articles lacked sufficient clarity about methodological choices and adaptation, the process of engaging children in the qualitative research, and analytic approaches, as Larsson et al.’s review of young people’s (ages ≤25 years) participation in intervention also identified [18]. This was especially the case in the studies that included adults. Reporting details about research with children may be a challenge for authors who face publisher word and image restrictions, or who need to describe multiple methods and participants. Researchers may also face constraints related to publishing in more traditionally quantitative journals because of much lower word counts or the need to revise articles based on comments from peer reviewers unfamiliar with qualitative research, who may de-emphasize the qualitative methods components or need standard qualitative terms defined in greater depth. Nevertheless, the lack of clarity and detail made it difficult to assess the rigor of the qualitative work and to understand how children participated in projects or if their participation was meaningful.

Second, our review identified a reliance on two types of qualitative methods in published research on intervention work— focus groups and interviews. While both provide useful insights and are standard in qualitative research, they are not the only options available, and they may not always be the best approaches for answering certain research questions, or for working with children in middle childhood [71]. Childhood studies offers a wealth of techniques and approaches for working with children, from ethnography to more visual and arts-based approaches, that can yield valuable insights for health interventions [3, 7, 15, 16, 72]. In sum, investment in qualitative approaches is likely to enhance implementation of interventions, as extending the range and improving the rigor of qualitative techniques in health intervention research will enable meaningful interpretations of children’s data, secure trust in those interpretations, and promote the value of research with children.

Deepening qualitative approaches and qualitative reporting is not enough to facilitate children’s meaningful participation without an understanding of both children’s and researchers’ positions and voice in research. Our analysis of rationales for inclusion of certain populations can offer insight into researchers’ theoretical framings, assumptions, and ethical approaches, especially in empirical articles on interventions, where word counts and conventions may constrain authors from detailed discussion. Childhood studies researchers have demonstrated the importance of attending to children’s social positions (e.g., as structurally dependent, bound by different institutional and social rules) in order to promote their inclusion [73]. Without a consideration of power relations that lead to children’s exclusions in research, researchers might use methods and promote interpretations that reinforce relations of power (e.g., teacher, student) and dampen children’s voices (e.g., conducting a study as part of a class activity where children may not be able to decline participation). Asking children to participate without attending to the power dynamics that shape interventions may in fact be disempowering to young people (e.g., when researchers tell children that they will listen to their input but do not reveal their own limits to make the children’s suggested changes). The studies in our review in which authors gave explicit attention to power inequities between children and adults seemed to find points of meaningful inclusion through methodological adaptations (e.g., using drawing or photography) and in tailoring of interventions to children’s needs.

Relatedly, our review identified substantial gaps in the representation of children’s participation in research reporting. Some researchers expressed legitimate uncertainties around how to interpret and represent children’s responses in order to apply these responses to interventions. And many articles exhibited biases in how they weighed and reported on younger children’s versus older children’s, youth’s, or adults’ responses. These gaps in representation in our scoping study do not speak to the impossibility of including younger children; childhood studies researchers have amassed substantial evidence demonstrating the inclusion of younger children. It speaks, instead, to the need to broaden the concept of what child participation looks like and how to represent children’s perspectives and experiences [6, 10, 74]. The challenge, as we see it, is in how to represent children’s responses as some among many, without diminishing or tokenizing them or over-selling children's agency in relation to those in more powerful positions [23].