Eligibility criteria

PICOS (population, intervention, comparator, outcome, study design) was used to establish inclusion criteria [28]:

• Population: CYP aged 2–16 years (or a mean age within that range) through direct or family-based interventions. A broad age range to include early years and adolescents was chosen to maximize the number of studies for the review, given that previously published reviews primarily reported studies using adults.
• Intervention: Targeting CYP’s physical activity behavior at home (inside or within the immediate vicinity of the home). No restrictions on intervention approach.
• Comparator: Control group (no treatment), wait-list control or an alternative intervention.
• Outcome: Device-based (e.g. accelerometer) or self-report (e.g. child or parent questionnaire) or change in minutes of CYP physical activity levels.
• Study design: Pre-post control design, including randomized and quasi control trials.

Where the home was the sole focus of the intervention, changes in physical activity were assumed to reflect home-based behavior change, as no other behavior was targeted. Interventions targeting the home within wider initiatives were excluded unless physical activity outcome data could distinguish between the home and other settings. Studies were excluded where interventions primarily targeted physical activity at locations outside the home, or lacked a comparator group. Exclusion criteria extended to review or discussion articles, not full articles published in a peer-reviewed journal (e.g. conference abstracts, commentaries), PhD thesis, grey literature, and non-English language articles.

Although the aim of the review was to identify behavior change techniques used in interventions, the use of behavior change techniques was not part of the eligibility criteria for inclusion.

Search strategy

Electronic databases searched included: AMED, PsychINFO, CINAHL, Cochrane, EMBASE, PubMed/Medline, Scopus, SPORTDiscus and Web of Science, from inception until June 2022. Dates of searches, coverage dates, and search criteria adapted for each database can be viewed in S1 Table. The search criteria included the following terms:

• Population = Child*, young person, adolescence, adolescents, teen*, girl*, boy*, infant*, family, toddler AND
• Setting = Home*, community, fam AND
• Target behavior = physical activity, exercise, fitness, play, move, dance, sport AND
• Outcome = step count, accelerometer, pedometer, GPS, global positioning system, moderate to vigorous physical activity, MVPA, minutes AND
• Intervention = intervention, toolkit OR resource OR campaign OR promotion OR trial or “randomi*ed controlled trial” OR “controlled trial” OR RCT OR “primary prevention” OR strategy OR program* OR experiment* OR quasi

Reference lists of included papers were searched for additional studies meeting the inclusion criteria.

Selection process

The lead author (AS) removed duplicate records and imported records retrieved into Microsoft Excel (Microsoft Corporation, 2010). Researchers (MW, IS, DM and EY) worked in pairs, and each pair independently reviewed article titles and abstracts. Titles were excluded if deemed irrelevant subject matter or where exclusion criteria were met. Abstracts were excluded if they were a conference abstract only, online thesis, or if at least 1 of the inclusion criteria was not met. Where agreement within pairs could not be reached, abstracts were reviewed by AS, ACh, JH or ZK. Researchers (MW, IS, DM and EY) worked in pairs to independently review all the full papers. All full papers were third reviewed by AS and ACr and excluded if at least 1 of the inclusion criteria were not met.

Data collection process

AS and ACh independently extracted data from included studies, then reviewed together to discuss discrepancies and reach a consensus on the level of information for inclusion. Protocol papers referred to within selected studies were also used to obtain data and provide clarification. Key study characteristics included participants, country, design, setting, outcome measure, effectiveness. Device-based measurement of physical activity outcome data was prioritized where more than 1 instrument was used (e.g. accelerometer and self-report). To identify how and why studies were/were not effective, process evaluation data within included studies (including those published separately) and intervention implementation characteristics (delivery mode [including the person/people delivering the intervention], duration and frequency, and engagement and compliance measures [including outcomes if reported sufficiently]) were extracted. Behavior change theory was also noted where mentioned by the authors, to identify the theory informing the intervention.

Behavior change technique coding

ACr and JH used the behavior change technique (BCT) taxonomy coding framework [26] to ascertain which BCTs were present within interventions. BCTs were coded as either “present beyond all reasonable doubt” (++) or “present in all probability” (+) as recommended [26]. BCTs were coded within source papers descriptions of intervention and control/comparator groups, and where they were applied to encourage parental support behavior; the target behavior being supporting and encouraging their children to be physically active. Reviewers met to discuss the coding, and any disagreements were resolved through discussion.

Quality of studies assessment

AS and ACh independently reviewed the selected articles for quality [29]. Discrepancies were discussed until a consensus was reached. The scoring criteria used a rating of ‘weak’, ‘moderate’ or ‘strong’ for the study components including: selection bias, design, confounders, blinding, data collection methods, and withdrawals and drop-outs. The overall paper rating was classed as ‘strong’, where no components were rated as ‘weak’; ‘moderate’ if 1 ‘weak’ component; and ‘weak’ if two or more ‘weak’ components.

Data synthesis

The key study variables and group descriptive statistics (including Cohen’s d effect size where feasible) were synthesized in a table, with studies themed according to the intervention mode. A narrative synthesis was used to compare and contrast each extracted element of the study and intervention characteristics. Recruitment rate was calculated as the proportion of participants randomized, out of those deemed eligible (where reported). Retention rate was calculated as the proportion of participants completing outcome measures, out of those randomized. Intervention characteristics were synthesized in a separate table. A narrative summary of process evaluation data was used to describe the barriers and facilitators to adherence to the interventions. Heterogeneity of studies made a meta-analysis impractical.

Study and intervention characteristics

Data are presented in Tables 1 and 2.

Download:

• PPT
  PowerPoint slide
• PNG
  larger image
• TIFF
  original image

Table 1. Study characteristics, outcomes and effects.

https://doi.org/10.1371/journal.pone.0289831.t001

Download:

• PPT
  PowerPoint slide
• PNG
  larger image
• TIFF
  original image

Table 2. Intervention characteristics.

https://doi.org/10.1371/journal.pone.0289831.t002

Engagement and compliance

Engagement and compliance data are presented in Table 2. All studies recorded engagement with the intervention, which was self-reported (n = 9), logged by the games console (n = 1), or a combination of both (n = 3).

One study reported significantly greater AVG engagement when supported by coaching, compared to AVGs alone [41]. Two studies reported a significant increase in AVG engagement between baseline and mid [36] or end [35] of the intervention period. Average duration of AVG play across the intervention period (where data were provided) ranged from ~90 minutes per week [42, 43] to ~280 minutes per week [23].

AVG engagement typically declined over the duration of the intervention period within several studies [35, 36, 40–43]. Where data were reported weekly, AVG engagement was greatest in the first week of the intervention [37, 40–42]. Where the earliest data collection was mid intervention, AVG engagement was greater than at the end [35, 36]. The provision of new AVGs during an intervention coincided with a ~75% increase in engagement [37] and the cessation of coaching coincided with a ~50% reduction in engagement [41]. One study observed a steady decline in the number of participants logging in to the web-based AVG over the intervention period [43].

Of the 9 studies that prescribed a frequency and/or duration of engagement with the intervention, 7 provided compliance data or engagement data (Table 2), with most achieving a mean intervention compliance of over 70% for part [40, 41] or all [38, 42, 45, 47] of the intervention duration.

Two studies highlighted issues with data quality. Errickson et al. [41] reported that some games console memory cards were not returned, and wide variation occurred in the completion of self-reported logs. Baranowski et al. [37] observed that some children in the intervention group had played inactive games, and that console data reported unlikely excessive engagement (e.g. 24 hours), could not distinguish between the study participant and other game players, and mismatched with self-reported engagement data.

Study quality

Two studies were rated ‘weak’, 10 studies rated ‘moderate’ and 1 study rated ‘strong’. Few papers clearly stated whether or not participants were single or double blinded, resulting in a low ‘blinding’ component score for 10 studies (Table 3). Individual scores for each component are shown in Table 3.

Download:

• PPT
  PowerPoint slide
• PNG
  larger image
• TIFF
  original image

Table 3. Global and component study quality assessment ratings.

https://doi.org/10.1371/journal.pone.0289831.t003

Effectiveness

The only non-AVG study included (Table 1) used home-based parent-led movement to music and was not effective [48]. Of the twelve studies incorporating AVGs, 3 showed a significant intervention effect for physical activity—1 for MVPA [47], and 2 for volume (counts per minute [23] and minutes of cycling [40]). AVGs within these included dance [23, 47] and cycling [40] activity that could be considered cardiovascular exercises. These interventions were ≥12 weeks. Two prescribed frequency, duration and intensity of game play [40, 47], while the other substituted typical video game play for AVG [23]. One study provided social support through telehealth coaching in addition to AVGs [47].

Staiano et al. [47] was the sole study reporting significant intervention effects post intervention (24 weeks), with a large effect size (d = 3.452). The significant 6-week intervention effect reported by Ni Mhurchu et al. [23] did not remain at 12 weeks (the end of the intervention). The significant intervention effect reported by Rhodes et al. [40] was only observed up to week 9 of the 13 week intervention, with the greatest group difference in week 1 (d = 0.52).

Behavior change theories and techniques

Self-determination theory [50] underpinned the components of 1 intervention and social cognitive theory [51] underpinned the components of 2 interventions (Table 4). No other interventions described a theoretical underpinning. Table 4 highlights the rationale provided, and how the theory was incorporated.

Download:

• PPT
  PowerPoint slide
• PNG
  larger image
• TIFF
  original image

Table 4. Behavior change theories described within interventions.

https://doi.org/10.1371/journal.pone.0289831.t004

The number of Behavior Change Techniques (BCTs) identified ranged from 1 [23, 37] to 15 [45], with a mean of 6 BCTs (Table 5). Among the 3 effective studies, there were 1 [23], 8 [40] and 13 [47] BCTs identified within the intervention groups.

Download:

• PPT
  PowerPoint slide
• PNG
  larger image
• TIFF
  original image

Table 5. Behavior change techniques used within intervention groups across all studies.

https://doi.org/10.1371/journal.pone.0289831.t005

Twenty-six out of 93 BCTs (28%) were identified (Table 5). The coded text for BCTs is in S1 Table. Adding objects to the environment was the only BCT common across all 3 effective interventions, with goal setting (behavior) and feedback on behavior, common within Rhodes et al. [40] and Staiano et al. [47] Among all interventions, adding objects to the environment—games consoles, peripherals, physical activity equipment and music DVDs—was the most prominent (12/13 studies). Practical social support was identified across nearly half (6/13) studies, usually as technical support and coaching to facilitate implementation and provide motivation to children and/or parents. Self-monitoring of behavior was identified in over a third (5/13) of studies through regular completion of logs/diaries/checklists to monitor engagement. Feedback on behavior was also identified in over a third (5/13) of studies, which ranged from grading on accuracy [41] to information on physical activity behavior [24, 40, 47]. Goal setting (behavior) was also used in over a third (5/13) of studies, through setting a goal or encouraging a set frequency and intensity of engagement. Instructions on how to perform the behavior and graded tasks were identified in a third (4/13). BCTs were also identified within the training of parents delivering an intervention for their child [45], in addition to the BCTs identified within the intervention protocol for the child participants. BCTs were also identified within alternative intervention group protocols, for example providing additional intervention elements [41], or adding non AVG objects to the environment for control groups [24, 40].

Process evaluations—Barriers and facilitators to engagement

Seven studies included a process evaluation, which included satisfaction surveys, [41, 42] acceptability surveys [47], interviews [37, 40, 52], and focus groups [24, 42] with children and/or parents. Children lacked motivation (reported by 32% of parents) [52], due to progressive difficulty of the intervention [52] or the AVGs being too challenging, causing frustration [24, 37, 41]. AVGs lacked sufficient variety of game choice [24], were unappealing [40], and some children found it difficult having no-one to play with [37]. Some children experienced issues with comfort of exergame bikes [24, 40] or problems with gaming equipment [40]. Participants stated competing priorities, either for the parent delivering the intervention (reported by 55% of parents) [52] or taking part in the intervention [24], or for the child directly [40, 47].

Many participants found AVGs fun and enjoyable [24, 40, 42, 47], distracted them from the physical work performed [24] and were easy to play [42, 47]. Participants reported monitoring success through using stickers and photographing high scores achieved, and had increased their game scores [42]. Having a schedule made the intervention easy to follow, and gave structure and routine [52]. Social support encouraging children to play AVGs with others led to 72% of children reporting to do so [47]. Social support from research staff gave parents a sense of belonging to the program and more parents valued this than support from other people in the home (32% vs 17% respectively) [52]. Self-efficacy was demonstrated through children’s self-belief that they were good at playing the AVGs, better than others, and satisfied with their performance [47]. Children valued having the opportunity to be more active and potentially improve their health/fitness [24, 40]. AVGs provided the opportunity for children to try activities they wouldn’t normally get to do (e.g. boxing or bowling), and meant that they “didn’t have to play outside” [37].

Active video games as the dominant intervention mode

In line with previous reviews, there was little variation in mode of delivery, with AVGs being the dominant mode [21, 22], While laboratory-based studies established adolescents found AVGs enjoyable, and game play elevated energy expenditure to moderate or vigorous intensity [53, 54], highly controlled efficacy studies may lack external validity. The lack of effectiveness among most interventions in this review suggests that observations during a controlled single dose of AVG play may not transfer to real world settings and sustained behavioral engagement. This is likely explained by uncontrollable factors within the physical and social environment. More recently, interventions targeting children’s physical activity at home have used a multi-component approach, supplementing AVG play with regular support from research staff or fitness coaches either in the home or via telephone. The most effective intervention in this review supported AVG play with regular coach-led telehealth sessions [47].

Mechanisms of effectiveness

The limited number of effective studies, wide variation in study design and intervention used, and dominance of low to moderate quality studies made it difficult to reach consensus on mechanisms of effectiveness. AVGs were considered fun and enjoyable [24, 40, 42, 47], suggesting that game play provided intrinsic motivation. However initial high engagement in AVGs was typically not sustained for the duration of interventions [23, 36, 37, 40–42], suggesting that additional mechanisms are important to overcome the novelty factor and support sustained behavior change. Focusing on the 1 high quality study [47], retention and compliance were high, and physical activity behavior change was sustained throughout the intervention. The focus on building social support through telehealth coaching and playing AVGs with others may have been important components contributing to the effectiveness. Social support through interaction with others may be important for sustaining children’s motivation for AVG, considering the greater dropout observed with single player vs. multiplayer AVGs games [55]. Social support from activity mentors and other participants joining home-based online physical activity sessions was reported as facilitators for engaging adolescent girls [56].

The intervention used by Staiano et al. [47] was underpinned by the social cognitive theory, including a focus on building self-efficacy within coaching and the AVG, and the study reported a significant increase in self-efficacy alongside increased physical activity. Self-efficacy has been highlighted as an important mechanism within children’s physical activity interventions [57, 58], and important for good AVG design [59]. The difficulties reported by children in relation to AVG play within the non-effective interventions [24, 37, 41, 52] may have led to low self-efficacy, limiting the effectiveness of the intervention.

Although the mechanisms are unclear, the use of regular video game players may have contributed to the effectiveness reported by other studies in the review [23, 40] given that a similar population showed sustained AVG engagement over a 24 week intervention [35].

Behavior change techniques

Within the 3 effective interventions, the most effective used the greatest number of BCTs [47]. However the highest number of BCTs across all studies was coded within a non-effective intervention [45]. This suggests that the type of BCTs, and/or implementation of BCTs, may be more important than the number used. Although 9 BCTs were common between these two interventions, many of these were coded within the telehealth component of the intervention which was implemented with parents in the non-effective study, suggesting that active involvement of children in telehealth coaching may be important to sustain engagement and enhance effectiveness.

Among the most common BCTs identified through this review, the only BCTs applied with some consistency across interventions (S3 Table) were ‘adding objects to the environment’, ‘self-monitoring of behavior’, and ‘goal setting (behavior)’. Other common BCTs showed wide variation in their implementation (S3 Table), for example, the extent of practical social support ranged from external technical support [38, 43], to at-home visits to setup gaming equipment [42, 47] and requiring children play AVGs with others [47]. The inconsistent implementation of BCTs limits the ability to identify active ingredients for effectiveness.

Enhancing intervention quality, acceptability and effectiveness

All 3 effective AVG interventions included lower body movements (dance or cycling), eliciting greater energy expenditure than AVGs involving primarily upper body movements [60]. AVGs involving lower body movements may maximize the potential of increasing MVPA, and should be tested within rigorously designed studies. Further research is also needed to understand how motivation for playing AVGs and enhancing PA can be sustained over time. Recent research on the use of animated narrative AVGs to enhance children’s MVPA [61] and the identification of principles for best practice in AVG design should inform choice of AVG in future interventions [62].

The use of social support and building self-efficacy using AVGs and telehealth coaching may be important mechanisms for effectiveness, and should be explored further. However, this may not be affordable or scalable to population level, suggesting other modes of intervention should be considered. In a multi-setting intervention with adolescent females completing live online workouts at home, other participants and scheduled messages from researchers provided social support, and motivated participants to complete physical activity sessions [56].

Only 28% of the 93 possible BCTs were identified within the 13 studies, highlighting the opportunity to develop novel interventions that incorporate other BCTs that may enhance effectiveness. For example, previously identified barriers of parental concerns around injury or damage to the home [63] and competing priorities [24, 40, 47, 52] could be addressed by restructuring the environment (BCT 12.1) and BCTs influencing reflective motivation such as providing information about health consequences (BCT ‘5.1), and feedback on the outcomes of behavior (BCT 2.7) [26].

Only 2 studies report using insight from children and parents to influence the design of the intervention [39, 46]. Co-producing interventions may enhance the feasibility, acceptability, quality, and impact of these interventions [64].

Study design and reporting of outcomes

Of the 13 studies included, 12 (92%) were rated low-moderate quality, with blinding of assessors and participants either not described, or not implemented (85%), and potential selection bias in the recruitment process (62%). Improving study design could reduce variability in study outcomes and help strengthen the evidence base. Few studies reported effect sizes, descriptive outcomes data and ethnicity of participants (Table 1). The use of the Consolidated Standards of Reporting Trials [65] would support methodological rigor and ensure outcome data is sufficiently reported, which may facilitate understanding effectiveness across different ethnic groups.

Most interventions measured physical activity using accelerometry, which will likely have included physical activity away from the home. Combining Global Positioning System (GPS) unit location data with heart rate [66] and/or accelerometry data [67] and indoor location and movement sensors [68] could be explored to improve the quality of objective measurement of home-based PA. Many studies did not measure short-term effectiveness (Tables 1 and 2) [38, 41–43, 45, 47] limiting the ability to identify potential initial engagement or effectiveness [47]. Weekly reporting of engagement and effectiveness would provide further insight into short- and medium-term effectiveness. Similarly, only 1 study measured outcomes following cessation of the intervention [42]. Follow-up measures of effectiveness could further knowledge around sustainability of behavior.

Although 3 studies referred to behavior change theory, none explicitly referred to using BCTs within their intervention design, limiting understanding of how theory was implemented within the intervention. Few studies provided a detailed description of the intervention and AVG play, or supplied copies of printed/online resources. Common BCTs within AVGs such as reinforcement and guided practice which can build self-efficacy may have been missed from coding [69]. Supplying copies of intervention materials/resources and use of the Template for Intervention Description and Replication [70] checklist would aid reviewers with the coding of BCTs.

Only 7 studies (58%) explored intervention barriers and facilitators. Future interventions should use a more mixed-method approach to help understand how an intervention has been effective [71, 72], especially within complex multi-component interventions [73].

Strengths and limitations of the study

The use of extensive search criteria and only including controlled, peer-reviewed studies ensured high-quality studies, with only two rated weak by the quality assessment tool for quantitative studies [29]. However, excluding uncontrolled studies may have omitted more varied home-based physical activity interventions. The review was also limited to English language publications and dominated by studies conducted in high income countries with majority White ethnic group populations, limiting the application of findings across low and middle income countries and among minority ethnic groups.

The review adhered to the updated PRISMA guidelines [27], which ensured transparent reporting and allows for future replication of the methods. A meta-analysis was not feasible due to heterogeneity in study designs and inconsistent reporting of effect estimates within studies.

The study also included an evaluation of BCTs used across interventions, and process evaluation data, helping synthesize intervention components which may be required for success. However, the low number of studies included in the review limits the ability to reach consensus on effective BCTs.