Poor sleep quality among adolescents is an important public health problem [1,2]. Research shows that 53% of adolescents get less than 8-hours of sleep per night and do not sleep for the recommended amount of 8-10 hours, especially on school nights [3-6]. Additionally, at least 36% of adolescents experience difficulty falling asleep, with 59% waking up feeling tired during the week [3-6]. Poor sleep quality in adolescents is associated with detrimental health outcomes, including unhealthy weight gain, reduced levels of physical activity, increased risk-taking behaviors, and increased levels of anxiety and depression, which could risk continuation into adulthood [7-13].

In recognition of the clear public health implications of poor adolescent sleep, there are several interventions aimed at improving sleep quality in adolescents [13]. These include school-based sleep education programs, which have been shown to successfully improve sleep knowledge in adolescents, but to be less useful for improving sleep behavior, as information is rarely enough to change behavior or guarantee positive sleep outcomes [14,15]. Interventions aimed at targeting specific behaviors, such as reducing screen time before sleep, have proven somewhat more successful for improving sleep duration and sleep-related behaviors among adolescents [16]. Likewise, interventions that include cognitive behavior therapy (CBT) have been found to be successful, producing significant improvements in both objective and subjective sleep outcomes among adolescents [17].

Determining the success of sleep interventions is difficult, as research in this area is often limited by a lack of control groups (CGs) and the application of a strong theoretical foundation, which makes it difficult to determine the mechanisms that make an intervention effective. Moreover, these interventions are often viewed as undesirable by adolescents due to being noninteractive and time-consuming [14,18-20]. Coupled with these shortcomings, sleep interventions targeted at adolescents can also be limited by small sample sizes, high dropout rates, little to no follow-up data, and being costly [13,17,21,22].

There is a pressing need for innovative solutions in this field of research. One clear recommendation is for intervention developers to utilize digital platforms [13,14]. There has already been some success in this field. For example, data suggests that the effects of digitalized CBT programs for insomnia are similar to those for face-to-face programs, indicating improvements in insomnia severity, sleep efficiency, subjective sleep quality, wake after sleep onset, sleep onset latency, total sleep time, and number of nocturnal awakenings [23]. These effects were comparable to those found with face-to-face CBT for insomnia (CBT-I) programs and were commonly maintained at 4- to 48-week follow-up [23]. However, these results were found in an adult population, and research is not as well established in the target population. Nonetheless, in a randomized-controlled trial, de Bruin et al [24] concluded that internet-based and face-to-face CBT-I result in long-term improvements in adolescent affective and anxiety problems by reducing insomnia symptoms, with adolescents preferring digital over face-to-face sleep treatment [25]. Therefore, this supports the use of digital methods of intervention delivery and aligns with adolescents’ preferences for sleep treatment. Evidence suggests that over 90% of adolescents in high-income countries, such as the United Kingdom, have access to the internet [26]. With digital platforms offering an attractive, feasible, and effective method to health management, including sleep behaviors among adolescents [20], there is an urgent demand for innovative, behavioral, and personalized approaches to treating adolescent sleep problems [13]—one that includes the adolescent voice in the development process—such that any intervention that is developed is acceptable to its intended end-users: the adolescents.

This study aims to overcome the limitations of previous sleep interventions targeted at adolescents, by undertaking a feasibility trial to assess the feasibility and preliminary impact of an evidence- and theory-based web-based sleep intervention called SleepWise. Specifically, the study adopted the person-based approach (PBA) [27] and the guidelines on the development of complex interventions [28]. The purpose of this feasibility study is to (1) explore recruitment rates, attrition, and uptake of the intervention, and its preliminary impact on sleep quality; (2) conduct a mixed methods process evaluation to determine participant engagement and experience with the intervention; and (3) understand the impact of incentivization on engagement, acceptability, and sleep quality, as the use of incentives needs careful consideration and understanding in research [29]. To investigate the third aim, 2 separate 2-arm randomized controlled feasibility trials were undertaken as part of this study; trial 1 was incentivized, and trial 2 was not.

SleepWise was uniquely developed using the PBA to intervention development [27] by integrating evidence-based behavioral techniques (eg, sleep education) and user feedback during the development process. The PBA combines participatory, theory-based, and evidence-based approaches to intervention development, with the aim of optimizing health interventions by grounding them within the perspectives of the people who use them [27]. While the PBA has been successfully used in previous research [30], it has not been used in this context. The development and optimization of SleepWise, which included addressing adolescent feedback based on a prototype version of the intervention, is reported in full elsewhere. In short, the optimized version of SleepWise reported in this paper consists of 4 weekly web-based sessions, completed over a 4-week period. Each session is tunneled (ie, participants work through a set number of compulsory or main pages), while “click-through” options provide optional and additional information. Session 1 provides information about the importance of sleep for young people and how eating habits can impact sleep. Session 2 contains information about the relationship between physical activity and sleep quality, and sessions 3 and 4 provide information about the importance and relationship between sleep and everyday habits among young people (eg, screen use, sleep environment, etc). Each session takes approximately 15-25 minutes to complete, depending on how much time participants want to spend on each session (eg, exploring optional pages). Participants in this study were encouraged to use SleepWise to set weekly goals and complete a daily sleep diary. At the start of each weekly session, participants reviewed their goals from the previous week and were given the option to set new goals or continue with the same goals. At the end of each weekly session, participants had the option to complete quizzes and take part in educational and interactive games (see Multimedia Appendices 1 and 2). On the SleepWise website, participants were also provided with optional sleep-aiding tools such as sleep routine charts and mindfulness recordings.

Two separate randomized controlled feasibility trials were undertaken as part of this study (OSF registration ID https://osf.io/yanb2). The only difference between the trials was that trial 1 was incentivized and trial 2 was not. This was to determine the impact of incentivization as part of the feasibility study. Trial 1 recruitment took place from November 2018 to May 2019 (study running in the summer term), with a rolling start time. Trial 2 recruitment took place from May 2019 to December 2019 (study running in the autumn term), also with a rolling start time.

Participants were recruited from 3 secondary schools and 2 further education colleges in Southeast England. The eligibility criteria for all parts of the study included being aged between 13 and 19 years and having access to a computer or laptop with internet access. Participants were not screened for any sleep, mental, or physical health conditions, as SleepWise is intended for use in a general population of adolescents, and not in an exclusive clinical or sub-clinical group.

Participant recruitment happened via the primary researcher reaching out to schools and colleges advertising the study and sending posters to those who responded as interested. An allocated member of staff (study facilitator) at each study site (schools and colleges) displayed or circulated the poster advertising the study to participants. Sites were offered a short talk advertising the study; if they chose this option, the researcher played a short video about adolescent health and talked about the aims of the study.

Ethical approval for the study was received from the university ethics committee (RKEEC190204), and the study was undertaken in accordance with the Helsinki Declaration. Adolescents who were interested in partaking in the study collected consent and information sheets from the study facilitator or by emailing the primary researcher. Signed consent forms were then collected by the primary researcher from each site or emailed directly to the researcher by participants. Parental consent was not required for participants aged 16 years and over, as per the British Psychological Society’s ethical guidelines [31].

This study maintained the confidentiality and anonymity of study participants. Study-related data were maintained on a password-protected computer, and any paper-based records were kept in a locked university cabinet. Participant data were anonymized. Intervention participants in trial 1 received a £20 (US $26.16) voucher once follow-up questionnaires were completed. This incentive was only offered to participants in the intervention group (IG) who undertook the intervention and completed the 5-week follow-up questionnaires in trial 1. All participants who took part in the qualitative interviews as part of the (poststudy) process evaluation received a £20 (US $26.16) voucher as a nominal payment for their time.

Participants in each trial were randomly allocated to an intervention or CG using an internet-based randomization generator (Randomisation.com). Participants were not blinded to allocation due to challenges associated with blinding participants in web-based trials [32,33]. The primary researcher was the intervention administrator, assessor, and data analyzer, and therefore it was not possible to blind them. Allocation was revealed to participants in an email containing information on how to access the intervention (IG), or to continue as normal (CG). Control participants continued to receive standard education (eg, at school) but were informed about the importance of their role in the study and were given the option to access the intervention (at the end of the trial) if they wished. After randomization, participants completed a baseline measure questionnaire about their sleep quality via an internet-based survey (Qualtrics). Participants in both groups were asked to complete this measure at baseline and at 5-week follow-up. Intervention participants were asked additional questions about the intervention’s acceptability at the 5-week stage. Text and email reminders were sent to participants if they had not responded to follow-up (5-week) questionnaires; however, participants had the option to opt out of these reminders if they wished. Participants in the IG were emailed weekly links to access the (free of charge) intervention and, with instructions, were asked to work through it for 4-5 weeks. Intervention participants were also sent weekly email and text reminders, reminding them about their goals and sleep logs. Participants also had the option to opt out of these reminders if they wished.

Regarding incentivization, intervention participants in trial 1 received a £20 (US $26.16) voucher once follow-up (5-week) questionnaires were complete. This incentive was only offered to participants in the IG who undertook the intervention and completed the 5-week follow-up questionnaires in trial 1. Participants in trial 2 did not receive incentives.

The aim of process evaluations is to explore implementation and contextual outcomes within a trial, which helps to better understand how best to improve interventions and their delivery [34]. The process evaluation in this study consisted of both quantitative and qualitative data collection and analysis. Specifically, to better understand participant engagement with the intervention, the quantitative component of the process evaluation evaluated participants’ engagement with each weekly session (date and time and pages viewed) and the number of goals and sleep logs completed across the intervention period, using SleepWise’s automatic data tracker. The qualitative component of the process evaluation explored participants’ experiences with SleepWise via qualitative follow-up interviews. The findings from these interviews can help inform the implementation of the intervention in a future trial of effectiveness. Participants from both trials in the intervention and CGs were contacted between 5 and 8 weeks post study via telephone interviews. Participants received a nominal £20 (US $26.16) voucher on completion of qualitative process evaluation interviews. This participant payment was separate from the incentivization evaluation, as the intervention period had ended.

In total, 19 participants (mean age 14.79, SD 0.98 years) were interviewed as part of the process evaluation interviews (see Table 1). Interviews lasted for approximately 20-30 minutes, depending on how much information participants provided. After conducting a thematic analysis, 3 main themes were identified: busy lifestyles, self-monitoring: recording sleep and setting goals, and changing behavior. These findings explore participants’ experiences with the intervention and can inform the implementation of the intervention in a future randomized controlled trial. Participant number (P, followed by number), gender, study group (IG and CG), are provided after each quote for informative purposes.

SleepWise shows promise as a potentially efficacious intervention for improving sleep quality among adolescents, as it was engaged with and deemed acceptable across both trials. The first session of the intervention (eating and sleep) was highly favored by adolescents, along with session 4 (understanding how everyday habits and environments impact sleep). Incentivization did not greatly influence intervention engagement, acceptability, or sleep quality, with participants in the nonincentivized group showing more success in achieving their set goals (ie, reported “meeting” their set goals more frequently). It was clear from participant feedback that adolescents lead busy lifestyles; however, the intervention delivery format was deemed time-efficient and suited participants’ busy schedules, albeit some further refinements to improve flexibility and personalization could be beneficial. Encouragingly, adolescents monitored their sleep behaviors using the sleep diary and goal-setting features of the intervention and attempted valuable changes to sleep-related behaviors (eg, reducing screen time before sleep).

The findings from this study suggest that incentivization does not greatly influence the acceptability of the intervention or sleep quality, with the nonincentivized trial (trial 2) showing a larger effect size, suggesting that participants in the nonincentivized trial showed a greater improvement in sleep quality. However, those in the incentivized trial viewed slightly more pages than those in the nonincentivized trial, suggesting that participants in the incentivized trial may have been more motivated to work through the intervention systematically. While this finding may be explained by individuals’ inclination to be more motivated by instant and guaranteed, versus delayed and lottery-based, incentives [44,45], those in the nonincentivized trial achieved their set goals more frequently, suggesting that incentivization may not necessarily encourage higher levels of behavior change. Clearly, using incentives in research requires careful consideration. The effectiveness and maintenance of behavior change when using incentives, especially when targeted at more challenging and complex behaviors and at different populations, is not well understood, and their feasibility is perhaps even less so [46-49]. While evidence suggests that health-promoting incentives have the potential to promote behavior change [50], controversy exists over the form and level of incentives, as there may be an expectation that incentives could be more attractive and effective for certain individuals, and therefore risk stigmatization or coercion [51,52]. Adding to this complexity, younger people may have a more favorable view toward incentives [49], yet the provision of incentives to adolescents remains unclear; therefore, careful consideration should be taken [53]. Nevertheless, the use of incentives may only be feasible to the extent to which they are deemed cost-effective to implement in practice when provided alongside health education and behavioral support [51,54]. Therefore, as the findings from this study suggest, incentivizing sleep interventions for adolescents may not impact acceptability, sleep quality and engagement with web-based sleep interventions, which may prove cost-effective in the context of a definitive trial.

Adolescents in this study viewed the intervention more favorably when the session content was deemed relevant to their needs (eg, age-appropriate), as well as having flexibility and easily accessible and personalized intervention features. In line with these findings, Werner-Seidler et al [55] show that adolescents deem an app-based sleep intervention more acceptable if the intervention content is tailored to their age group, its use is flexible, and the intervention can be personalized to each user. Personalized and flexible approaches to intervention delivery may also have the added benefit of reflecting users’ specific needs and allowing adaptive and targeted communication. With adolescents strongly favoring personalized and flexible approaches to intervention delivery, more innovative approaches to developing sleep interventions could replace one-off activity monitors and sleep logs [13], which may in turn decrease clinical time and increase availability and accessibility to treatment options for adolescents with sleep-related problems [13,56].

The findings from this study highlight the importance for sleep interventions to align their design goals with adolescents’ priorities and values. Adolescents in this study expressed leading busy lifestyles (eg, completing high volumes of homework), which may hinder their engagement with the intervention. As may be expected, high levels of homework [22], coupled with other adolescent-specific priorities such as maintaining social relationships [22,57], may make it more difficult to engage with sleep interventions. Given that adolescents face a unique set of age-related barriers, such as time constraints [58], and attending to early school start times [59], enabling bite-sized (eg, short weekly sessions) and interactive digital communications via platforms such as SleepWise could encourage adolescents to engage with interventions and subsequent sleep behavior change. Indeed, adolescents in this study reported attempts at behavior change when the intervention content was deemed important and relevant to them, suggesting that participants’ intention to change behavior may be influenced by how acceptable they deem an intervention (eg, interesting and time-efficient). With research indicating that adolescents prefer time-efficient and age-appropriate interventions [20,55], and adolescence being a potential marker for reduced application of intentional health-related self-regulatory behaviors [60], and characterized by increased impulsivity [61], it becomes critical to align interventions such as SleepWise with adolescents’ priorities and take into account the contextual factors (eg, academic commitments) that may facilitate or hinder engagement and subsequent behavior change [62].

Limitations of this study include the lack of an active CG. While adolescents in the CG continued to receive their usual education, it may have been preferential to include an active CG or psychological placebo. Additionally, a cluster-randomized trial was not undertaken. Implementing a cluster-randomized trial would reduce the likelihood of cross-contamination.

This study had several strengths. First, a novel approach that puts the voices of adolescent users at the center of intervention development was undertaken, ensuring the intervention was in line with participants’ priorities. Second, SleepWise was evaluated in the context of a feasibility trial, which helped answer key feasibility questions (eg, acceptability of the intervention) and contributed to reducing research waste and time. Lastly, and regarding the qualitative methodology used in this study, a strong commitment was made to prolonged engagement with the topic and undertaking a rigorous approach to qualitative data collection and analysis. Rigor in this sense refers to the elements of data that were collected and analyzed in relation to supplying all the information required for a comprehensive qualitative analysis [63].

This study addresses a core gap in the literature by testing the feasibility and potential efficacy of a person-based and web-based sleep intervention (SleepWise) aimed at adolescents. The SleepWise study has laid the foundation for developing an understanding of adolescents’ engagement and acceptability with web-based sleep interventions, with adolescents in this study perceiving SleepWise as an acceptable intervention (eg, time-efficient) with recommendations for future improvements (eg, personalizing the goal-setting feature). Interestingly, incentivization of the intervention did not show a notably large impact on engagement, acceptability, or sleep quality. Effect size analysis showed promise for improved sleep quality across both trials; thus, a main trial is warranted to test the effectiveness of SleepWise.