The University of Galway Research Ethics Committee granted ethical approval for this study on January 16, 2023 (Ref: 2023.02.014).

In this paper, we use the term PPI to refer specifically to the advisory groups that informed the study’s design and conduct. Eight months before launching The Kid’s Trial, we established a Children’s Research Advisory Group (CRAG) and a Parents’ Research Advisory Group (PRAG), comprising 7 family units (8 children aged 7-11 years and 7 parents). Child literacy levels and parent education levels were not collected for PPI group members. These groups were recruited using social media and email campaigns. Parent-child dyads were recruited to support participation in a fully online setting, particularly for younger children. Meetings were held online to accommodate the geographic diversity of participating families (Estonia, Ireland, Kenya, Norway, Tanzania, and the United States).

Over approximately 18 months, members of the CRAG and PRAG collaborated with the research team to refine the visual identity and tone of The Kid’s Trial website, codevelop age-appropriate consent and assent materials, shape the content and language of the animated explainer videos, and review survey wording for clarity and accessibility. Engagement was maintained through regular online meetings, structured feedback surveys, and anonymous polls. A detailed account of the activities, challenges, facilitators, and experiences of both PPI group members and researchers is provided in a separate report [20].

At each step, children and parents were directed to the relevant page on the website, viewed the corresponding animation, and then invited to join that step by clicking a link to the associated QuestionPro [21] survey. Demographic data collected in these surveys included a parent’s email address, the child’s age, gender, country of residence, and ethnicity. The parents’ email address and the child’s age were required fields, while gender, country of residence, and ethnicity were optional and were collected to assess the study’s demographic and geographical reach, and, specifically, whether the decentralized design facilitated participation across different countries, age groups, and genders. Demographic data were collected for each survey except for the final 2, “Collecting the results!” and “Sharing the results!” as only previously participating children were invited to join those steps.

This section summarizes the purpose and methods of each step of The Kid’s Trial. Figure 2 illustrates the sequence and timeline of these steps. The duration of each step varied depending on operational requirements: “Running the trial!” and “Collecting the results!” remained open for the longest period to facilitate recruitment for the embedded RCT, whereas “Sharing the results!” was necessarily concise due to project deadlines.

Between September 4, 2024, and November 18, 2024, children were invited to join the second step of The Kid’s Trial, in which children were asked to plan the trial that would address their selected question. They determined both the primary and secondary outcomes for the REST trial and the treatment arm procedures for participation in the “Try-it-Out” and “Wait-and-See” groups. To identify the outcomes, children were asked how sleep quality should be measured after the trial. They could select more than one of the following options:

The 2 most popular options helped identify the primary and secondary outcomes for the REST trial. To identify the primary outcome, we asked which outcome best showed how children slept. Children selected daytime sleepiness as the primary outcome and overall sleep quality as the secondary outcome.

To define treatment and control arm procedures, the children were asked:

Children’s “usual beds” were defined as any bed where they regularly sleep, for example at home, at a relative’s house, or in another regular residence. The “Planning the trial!” survey is available in Multimedia Appendix 3

In “Running the trial!” children were invited to take part in the REST trial. By clicking the “Join the trial here!” button on the website, children were redirected to a QuestionPro survey [21], in which baseline data on daytime sleepiness and overall sleep quality were collected before randomization. This step opened on January 16, 2025, and closed September 15, 2025.

Daytime sleepiness was measured using the Patient-Reported Outcomes Measurement Information System (PROMIS) Pediatric Short Form v1.0 Sleep-Related Impairment 4a [26], which yields a standardized T-score, with lower scores indicating less impairment [27,28]. Overall sleep quality was measured with the single-item Sleep Quality Scale (SQS) [29]. The SQS instrument consists of one question that asks participants to rate their overall sleep quality over the previous 7 days on a categorical scale coded 1‐5, with higher scores indicating better sleep quality [29].

The REST trial was a 2-arm, parallel-group, pragmatic superiority RCT. The intervention period lasted one week, and participants were randomized using a 1:1 ratio to either the “Try-it-Out” or “Wait-and-See” group using computer-generated randomization using QuestionPro [21]. Due to resource constraints, it was not possible to engage an independent researcher to clean and manage the data; consequently, the researchers became aware of group allocation after participant randomization. Because of the nature of the intervention, it was impossible to blind participants to their group. However, no party could influence or predict a participants’ group allocation. This survey is available in Multimedia Appendix 4.

In the fourth step, “Collecting the results!” parents of children who took part in the REST trial received an email with a survey link specific to their child’s allocated group 8 days after randomization. These surveys collected primary (sleep-related impairment [SRI]) and secondary (SQS) outcome data, along with additional group-specific questions on intervention adherence. These adherence data are reported in the REST trial report [30]. If the survey was not completed within 10 days postrandomization, a reminder email was sent, followed by a final reminder on day 13 postrandomization. These surveys are available in Multimedia Appendix 5 (S5A and S5B).

The REST trial protocol [31], statistical analysis plan (SAP) [32], and full trial report [30] detail the methods, analyses performed, and full results. Analyses followed an intention-to-treat (ITT) approach using a mixed effects model with treatment group as the main predictor, adjusting for baseline outcome, baseline comfort item use, age, and gender as fixed effects and country of residence as a random effect. Additionally, exploratory subgroup and sensitivity analyses were conducted as outlined in the SAP to assess differences by baseline characteristics.

In “Sharing the results!” all parents whose children had previously participated in any step of The Kid’s Trial received an email inviting their children to rank different methods and locations in which to share the study results. This step opened on September 16, 2025, and closed on October 9, 2025. To determine which options children would rank, the research team first consulted the PPI groups to ensure the choices were appropriate and gather additional suggestions.

Additionally, this survey repeated the 3 questions from the Pre-Trial Quiz to assess children’s understanding of RCTs and learning gained through participation. Finally, children were asked about their experience of taking part in The Kid’s Trial and their views on the learning materials. This survey is available in Multimedia Appendix 6.

Participation across the steps of The Kid’s Trial is summarized in Figure 4. In total, 224 children from 15 countries took part in at least one survey (Figure 4A). The number of participants per survey ranged from 20 to 166. The optional Pre-Trial Quiz had the highest participation (n=166), followed by “Running the trial!” (n=139) and “Collecting the results!” (n=101). The earlier surveys (“Sending in your questions!” “Sorting the questions!” “Choosing the question!” and “Planning the trial!”) had lower participation counts (24, 63, 64, and 65 participants, respectively).

Figure 4B and C illustrate how participation overlapped across steps. Figure 4B shows the number of children who completed each unique combination of surveys, ordered from the most common (left) to the least common (right). Each bar represents a specific participation pattern. Figure 4C visualizes these combinations: filled circles indicate the surveys completed for each pattern, while empty circles show surveys not completed. Together, these plots reveal that 37% (n=82) completed only one survey, 19% (n=42) completed two, 15% (n=33) completed three, 6% (n=14) completed four, 0.9% (n=2) completed five surveys, and 21% (n=48) completed six surveys. Only 0.4% (n=1) completed seven surveys, and 0.9% (n=2) completed all eight. The most common participation patterns involved combinations of the trial-conducting steps (“Running the trial!” and “Collecting the results!) with or without the Pre-Trial Quiz.

The mean age of participants was 9.3 (SD 1.6) years. Most parents reported their child’s gender; 55% (n=122) listed “boy,” 45% (n=101) listed “girl,” 0.4% (n=1) selected “prefer not to answer,” and no parents selected “something not listed here.” Most parents provided their child’s country of residence (99%, n=222) and ethnicity (98%, n=219). Of those, approximately 10% (n=20) chose to specify ethnicity in a free-text response. Most participants lived in Ireland (44%, n=98) or Kenya (39%, n=87). Most parents listed “White” (53%, n=116) as their child’s ethnicity, and 34% (n=75) listed “Black/Black Irish.” Detailed demographic information of participating children is presented in Table 1.

The Kid’s Trial demonstrated that a global, child-led citizen science RCT can be conducted using a fully decentralized design. Children engaged across multiple stages of the research process, including choosing a research question, defining trial procedures, participating in the trial, and choosing how the findings would be shared. While recruitment for the REST trial fell short of its target and attrition was higher than expected, the successful completion of all study steps and overall engagement of 224 children and families demonstrates that children can meaningfully contribute to core trial activities within a fully online RCT. These findings highlight key methodological and practical considerations for designing and conducting participatory, decentralized trials with children.

The embedded REST trial found no meaningful effect of comfort items on sleep-related impairment, with a small improvement in perceived sleep quality that was not robust to sensitivity analyses. These findings should be interpreted in the context of an underpowered study. As such, the REST trial is best understood as a proof-of-concept that demonstrates the implementation of a child-led, decentralized RCT, rather than providing definitive evidence on the effect of comfort items on the quality of children’s sleep. While the trial findings were inconclusive, the successful conduct of a participatory, child-led trial, demonstrated by children’s participation across all steps, illustrates that children can meaningfully engage with the process of generating, refining, and testing research questions. It also highlights that participation in these processes can provide opportunities for children to learn an important principle: not all interventions show effects, and learning to interpret null or uncertain findings is an important aspect of critical appraisal.

By adapting the principles of public co-production for a younger audience, The Kid’s Trial extends the approach pioneered by The People’s Trial [9]. Like its predecessor, it used experiential participation to engage children with RCT processes, aligning with educational frameworks that emphasize active participation as central to learning [34]. Children’s engagement throughout the study supports previous research showing that inquiry-based participation fosters curiosity and critical thinking about science and health [12,35,36]. Importantly, the study also provided researchers with practical insight into how children can be active partners in health literacy and research engagement initiatives, rather than passive recipients of information. Involving children in the co-creation of the trial meant that they chose the research question, selected and prioritized outcomes, contributed to trial procedures, and selected dissemination routes according to their own priorities and perspectives. As a result, The Kid’s Trial reflects children’s interests and decision-making processes in ways that may have differed substantially from an adult-designed study.

One of the aims of The Kid’s Trial was to explore whether participation in a co-created RCT could improve children’s engagement with and understanding of trials. The limited posttrial results are consistent with respondents’ perceived improvements in RCT understanding. These findings should be interpreted cautiously, given the small number of posttrial responses and potential self-selection bias, but they are consistent with the broader educational rationale underpinning participatory approaches.

It is important, however, to distinguish between the dual objectives of The Kid’s Trial. While the study successfully engaged children in an experiential citizen-science initiative, the embedded REST trial fell short of its recruitment target, was underpowered for its primary analysis, and yielded null or uncertain findings. These dual outcomes underscore the tension in citizen science between rigorous scientific methods and broad public engagement. While this tension is not unique to our study, it highlights the importance of clearly defining success criteria at the outset: for educational citizen-science initiatives, engagement and learning outcomes may be more relevant benchmarks than statistical power or definitive scientific conclusions.

Participation patterns showed considerable variability but demonstrated that sustained engagement in a multistep online study is achievable for children in this age group. Because the final survey sample was small, we did not attempt demographic comparisons between early- and late-stage completers. While many participants engaged briefly, a meaningful subset remained active across several stages, particularly during the trial conduct steps. Children and parents appeared to be most motivated by concrete, action-oriented activities such as running the trial and reporting results, with fewer joining the earlier planning steps. This likely reflects a combination of factors, including the intensified recruitment efforts during that period and increased awareness of the study after several months of ongoing advertising and dissemination of materials. This pattern suggests future studies should consider a “core-periphery” design that distinguishes between essential trial-conduct steps (which attract higher engagement) and optional enrichment activities (for sustained participants). Alternatively, condensing all steps into a shorter, more intensive timeframe might preserve momentum, though this would need to be balanced against the cognitive and time demands on child participants.

Recruitment for The Kid’s Trial proved challenging despite extensive and varied outreach. The intentionally wide target population, chosen to maximize inclusivity, may have reduced the personal relevance or urgency that drives participation in more narrowly focused studies, particularly given the many competing demands on families’ schedules. Additionally, the lack of established school or community partner networks limited the reach of the recruitment effort. The REST trial’s extended recruitment period (9 vs 6 months) may have contributed to attrition, as delays between steps reduced continuity and momentum. The final step, “Sharing the results!” was open for only 2 weeks due to study deadlines, which may have further contributed to the low participation at that stage. The decline to only 9% completion in the final step raises important questions about study design. While the abbreviated timeframe was a contributing factor, this attrition likely also reflects cumulative fatigue, the absence of personal relevance of disseminating the outcome, and the substantial time elapsed since randomization (up to 9 months for early participants). Future studies might maintain engagement by integrating dissemination planning earlier, offering multiple dissemination formats throughout the study, or creating milestone celebrations to sustain momentum. The low response rate also prevented robust assessment of learning outcomes, limiting our ability to evaluate whether participation meaningfully improved trial literacy, a core objective of the initiative.

Notably, paid digital advertising increased website traffic but failed to convert visitors to participants, suggesting that trust and personal connection, rather than visibility alone, drive enrollment in child-focused citizen science. This finding has practical implications for future resource allocation, suggesting that investment in community partnerships and institutional networks may yield better returns than paid advertising.

Despite these challenges, the study demonstrates the feasibility of this approach. Early and sustained involvement of PPI groups ensured that study materials were age-appropriate and engaging. The decentralized model allowed participation across cultural and geographic contexts at relatively low cost, showing that such designs can enhance inclusion and accessibility in child-focused research.

The Kid’s Trial had several notable strengths. It is, to our knowledge, the first global citizen-science RCT co-created and conducted by children, integrating authentic public involvement throughout the trial. The decentralized design allowed participation from 15 countries without prespecified geographic limitations. However, participation was heavily concentrated in 2 countries (Ireland and Kenya, representing 83% of participants), reflecting the research team’s networks and the successful partnership with Toto Centre Initiative in Kenya. While the study achieved international reach, future work should distinguish between “global access” and “global engagement” when describing decentralized studies. The extensive PPI input enhanced the study’s relevance and child-centered design, and the trial materials, including the animations, simplified complex trial concepts, making them accessible and engaging for children.

However, the limitations should also be acknowledged. The online format excludes children without internet access, estimated at up to one-third of the global population [37]. At the same time, the decentralized format may offer opportunities to engage children who are geographically dispersed or less connected to traditional research settings, particularly when combined with community-based partnerships. The reliance on parental support and the use of English-only materials also limited inclusivity. Although our work with the CRAG members enhanced the accessibility and child-friendly tone of study materials, the study did not address the needs of children with additional communication or learning requirements. In addition, we did not collect children’s literacy or parents’ education levels among PPI group members, which limits how fully we can describe the advisory groups’ educational and literacy context or assess whether further adaptations were needed. Future work should explore how participatory trials can be adapted to support a wider range of access needs, for example, through easy-read or multilingual materials, additional visual supports, hybrid delivery options, and accessibility testing with children at the younger end of the age range.

The REST trial was underpowered, and its outcomes and children’s understanding of RCTs was self-reported. Because participants were aware of their allocated group and outcomes were self-reported, the REST trial was also vulnerable to expectancy and social desirability biases. This may have influenced the outcome measures, such as perceived sleep quality, particularly given that the small between-group difference in SQS was not robust to sensitivity analyses. Posttrial data were limited to 20 responses (9% of participants), and although all respondents reported understanding what RCTs were and how they were conducted, compared with 70% and 60% at baseline, respectively, these findings require cautious interpretation, given the small sample size and potential selection bias (those who completed the final survey may have been more engaged learners). Rather than providing definitive evidence of learning, these results are consistent with the possibility that experiential participation may support children’s perceived understanding of trial concepts and their importance. Future studies should prioritize learning assessment by embedding brief knowledge checks throughout the study rather than relying solely on final surveys, and by using validated instruments where available.

The Kid’s Trial provides proof-of-concept that children can meaningfully contribute to the design and conduct of an RCT through a fully decentralized participatory approach. It highlights both the potential and the practical challenges of involving children as research partners. By introducing trial literacy through experiential participation, the study suggests that children can begin to engage critically with evidence from an early age through active participation in generating and testing research questions.

Future child-led citizen-science studies could improve this model by prioritizing early partnerships with schools and community organizations, which proved the most effective recruitment channel in the present study (44% of participants), rather than relying on social media or paid advertising, which generated traffic but failed to convert visitors to participants. Condensing study phases to maintain momentum between steps (limiting intervals to 4‐6 weeks rather than the months-long gaps in our timeline) and integrating brief learning assessments throughout rather than only at the end point would address the substantial attrition observed in later stages. Researchers should also clearly distinguish between educational outcomes (engagement, interest, and satisfaction) and scientific outcomes (statistical power and definitive findings) when defining success metrics, as these dual objectives may require different evaluation frameworks and can be in tension. Finally, incorporating gamification elements, milestone celebrations, and regular progress updates may help sustain participant engagement across multiple study phases, while developing multilingual materials and hybrid delivery models would enhance true global inclusivity beyond simply enabling international access.