A comprehensive systematic search of four scientific medicine literature databases was conducted to access the most recent papers, covering the period from January 2013 to February 2024. The databases included Web of Science, Embase, Medline, and PubMed.

A total of 6233 articles were retrieved using search terms based on the PICO principle in the four databases (Appendix 1). The search terms were combined by adding “AND” for different concepts and adding “OR” for similar terms. Then all articles were subsequently imported into the Endnote X9 for reference management.

The review of inclusion criteria is based on the PICO framework, and give a further explanation in the study type and sample. (1) Participants: Children under 13 years of age, or studies with a mean age under 13, and parent-child dyads (where at least one parent and one child are directly involved in the physical activity together). There is no restriction on the number of individuals in the family (e.g., father/mother-son/daughter). (2) Interventions: Any family-centered physical activity interventions where both the parent and child engage together. There are no restrictions on the intervention setting (e.g., laboratory, classroom, home, hospital, gym) or duration. The FITTVP components of physical activity (i.e., frequency, intensity, time, type, volume, progression) are not limited. (3) Comparison: Studies must include a comparison condition (e.g., intervention group vs. control group, waitlist group, or usual care group). (4) Outcomes: The study must report on objective measures of: MVPA time (e.g., minutes per day). Sedentary behavior (e.g., sedentary time in minutes per day). Daily steps (e.g., steps per day). (5) Study Type: Only RCTs are included. (6) Study Samples: There are no restrictions on the sample size (e.g., pilot RCTs with limited sample sizes are included).

(1) Non-English Literature: Studies published in languages other than English. (2) Literature Reviews and Non-Original Research: literature reviews, patent literature, protocols, and studies with missing physical activity data. (3) Health Conditions: Studies where participants have physical diseases or exercise-related disorders. (4) Duplicate Samples: Articles using the same sample or dataset. (5) Non-Objective Measurements: Studies where children's physical activity levels are not objectively measured. (6) Unclear or non-family-centered Interventions: Studies with interventions that do not specifically involve family-centered physical activity.

Two authors (WQ and LH) and a qualified research assistant independently screened all articles and extracted the data. Any conflicts were addressed through discussion until unanimous agreement was achieved. Data were extracted according to the following aspects: (1) study characteristics (i.e., title, authors, publication year), (2) participant characteristics (i.e., age, sample size), (3) study design (i.e., two-arm RCT, three-arm RCT, cluster, stratified), (4) intervention duration, (5) intervention programs, and (6) the mean and standard deviation values of outcomes. If multiple studies used the same data, they were combined into one for analysis. In cases of missing data, the authors contacted the original researchers to obtain the necessary information. Data privacy and participant protection are critical ethical concerns in contemporary research. While advocating for data openness, it is essential to implement both technical and ethical measures to ensure the adequate protection of participants' privacy (46).

The risk of bias in the included studies was independently evaluated by two reviewers (YWX and LYW) utilizing the Cochrane Risk of Bias Tool (version 2.0). Discrepancies in the assessments were resolved through a deliberative process involving a third expert to reach a consensus. The tool assesses seven domains of bias risk. Each domain was classified as having a low, high, or unclear risk of bias. The evaluation of study quality was subsequently conducted based on the ratings within these domains. The resultant risk of bias chart was generated using RevMan 5.4 software to facilitate data presentation and analysis.

The review used a random-effects model to assess the degree of heterogeneity across the included studies. RevMan 5.4 was used to perform the primary analyses, including generating forest plots, conducting heterogeneity tests, performing sensitivity analysis, and carrying out subgroup analysis. For studies with different units of measurement, Statistical analysis was undertaken employing the standard mean difference (SMD) as the effect size measure, with a 95% confidence interval (95% CI) to provide an estimation of the precision and uncertainty associated with the calculated effect size (47). For studies with the same outcome variable and measurement unit, the weighted mean difference (WMD) was calculated. A p-value of < 0.05 was considered statistically significant for the differences between groups. Based on the characteristics of the selected studies, subgroup analysis was performed according to the primary outcome, intervention content, and duration (short-term, medium-term, or long-term) to evaluate whether the effect sizes differed across these subgroups.

The heterogeneity across the included studies was quantified using the I² statistic, with I² values classified as very low ( ≤ 25%), low (25% < I² ≤ 50%), moderate (50% < I² ≤ 75%), and high (>75%). To assess potential publication bias, Egger's test was applied. Additionally, a sensitivity analysis was performed to examine the robustness of the findings by systematically excluding each study in turn, thereby assessing whether the removal of any single study had a substantial impact on the pooled effect size.

A comprehensive search across four medical databases yielded a total of 6,233 studies by using predefined search terms. After applying the established inclusion and exclusion criteria, 6,222 studies were excluded. Ultimately, 10 studies were selected for inclusion in the systematic and quantitative analysis, as depicted in Figure 1.

This review includes 10 RCTs published between 2013 and 2024, comprising a total of 1,557 parent-child dyads involved in family-centered PA interventions (Table 1) (29, 30, 35, 36, 48–53). The participants' mean age ranged from 3 to 11 years, with the largest study including 826 parent-child dyads (53), and the smallest involving (50). All studies adhered to RCT methodology, all studies were the randomized controlled trial, with two being pilot RCTs (35, 50), three cluster RCTs (one of which was a cluster randomized clinical trial) (29, 30, 53), and one crossover RCT (48). Among the selected studies, three were published after 2020 (29, 30, 49). Three studies were from USA (49, 50, 52), and two studies were from Australia (35, 48). The left studies, each of one, came from China (30), Canada (36), Danish (29), UK (51), German (53), respectively. Of the 10 studies, nine involved both child-parent dyads, while one study focused only on child-father dyads. Five of the selected studies used the same devices, the Actigraph GT3X⁺. Two articles used the GT1M Actigraph accelerometer (50, 51). The rest of objective devices were Atciheart (53), Actical accelerometer (48), and non-commercial DeviceTracker apps (29). Theories used in the selected studies included socio-ecological theory (52), social cognitive theory (30, 49, 50), social learning theory (53), self-determination theory (51), theory of planned behavior (35, 48), family system theory (30), planning and self-regulation theory (36), and the behavioral model Social Cognitive Theory (29). The social cognitive theory was the most commonly applied framework in these family-centered interventions. The duration of interventions varied widely, ranging from 2 weeks to 2 years (29, 52). Each intervention program had a clear focus, such as teamplay (51) and mHealth intervention (49) and so on. The outcomes of the interventions were diverse, with many studies evaluating the MVPA of both children and parents separately, alongside other indicators such as sedentary behavior and daily step count.

Due to the RCT design, all studies involving randomization exhibited low selection bias (Figure 2). Additionally, most of the selected studies provided a clear flow of participant enrollment and demonstrated low attrition bias. The included studies adhered to methodological standards, including allocation concealment and blinding of participants and personnel. Importantly, since outcomes were measured using objective monitoring devices, detection bias was minimal. The researchers also focused on other potential biases, as shown below (Figure 3).

The aim of this review was to evaluate the effectiveness of family-centered interventions on objectively measured physical activity in children. One may conclude that the family-centered intervention programs were found to be effective in promoting children's MVPA, as evidenced by the 10 selected studies. Notably, significant improvements were observed during weekends and in the short term. However, no significant effect was found on children's sedentary behavior or parental MVPA. An unexpected finding was the lack of change in sedentary behavior despite the increase in MVPA. This result aligns with similar findings from other studies (26). Our findings extended the previous work based on the objective outcomes. Compared with other studies (26, 37), which utilized the mixed outcome measures, the results of this review, which exclusively used objective measurements, are more reliable. Given the lack of previous meta-analyses using a single objective measurement, this review may offer a novel contribution to the field.

Interestingly, social cognitive theory is the most frequently utilized theory in the interventions examined in the included articles. This highlights its potential as a valuable reference for selecting theories in future behavioral interventions. Family interventions should focus not only on the child but also on the parents. Only three articles collected data from the child and the parents. And fundamental movement skills (FMS) should be a novel outcome as early development of FMS is crucial for encouraging PA. Under conditions where equipment permits, gathering data from both parties would provide greater value. However, it is important to note that the included studies used different types of objective devices. Among the ten included studies, only half employed the same devices. The results of this review suggest that high-quality measurement devices were not widely used in earlier studies, although more recent research utilized higher-standard devices. Regarding sample size, only two trials included large-scale samples, with one study involving 534 parent-child dyads and the other 826. While all ten trials were RCTs, two pilot studies had a small sample size. The available data were insufficient to draw definitive conclusions.

The findings of this review offer important perspectives on the development of family-focused intervention programs designed to encourage physical activity, especially in children and young adolescents. Firstly, family-centered intervention programs were more effective in the short term. The duration of the intervention appears to be a key influencing factor. This may be because participants initially find the program engaging but gradually lose interest as time progresses. To maintain engagement, future interventions could periodically introduce new themes and incorporate fun, interactive activities to sustain participants' curiosity and enthusiasm. Given that children's interests and levels of engagement vary by age, program design should be developmentally appropriate. Except for a relevant factor of activity duration, additional elements—such as partnerships with external organizations—can further promote participation and extend the program's impact (54). For instance, collaboration with local sports associations, such as youth soccer leagues, can offer structured PA tailored to developmental needs, foster teamwork, and strengthen alignment between program objectives and community resources.

Additionally, the rest period emerged as another potential influencing factor. This review found that family-centered intervention programs had a significant impact on children's MVPA during weekends. During weekdays, children focus on their studies, and parents are often occupied with work, leaving limited time for physical activity. Family-centered interventions should consider maximizing the use of weekdays to encourage physical activity. Besides, leveraging existing resources such as school infrastructure can be an effective strategy for promoting PA (55). One potential approach involves incorporating family-oriented physical activity programs into the school day without increasing academic demands. For instance, schools could implement after-school programs or integrate brief activity breaks during the day that engage both children and their families.

Finally, the role of electronic devices is noteworthy. The review observed no significant effect on reducing sedentary behavior. With the increasing use of electronic devices, children are becoming more sedentary, spending long periods in front of screens. Even physically active children may still engage in excessive sedentary behavior. Family-centered interventions should encourage parents to actively manage screen time by removing or replacing traditional electronic devices at home and promoting outdoor, nature-based activities. Furthermore, it is crucial to acknowledge the increasing interest children have in technology. Future programs could explore integrating electronic devices with PA, such as through active video games or fitness apps that promote movement (56, 57). This approach could harness children's technological interests while fostering physical activity, offering a promising avenue for future interventions.

This review has several advantages. Firstly, it is the first one to quantify the effectiveness of family-centered intervention programs which focus on measuring physical activity objectively in children, offering a new perspective on family-centered interventions in this field. While some previous reviews have examined the impact of family-centered physical activity programs, they employed different measurement methods, resulting in significant variability in outcomes. Objective measurement helps reduce bias, further strengthening the findings. Additionally, all included studies were RCTs, which adhere to strict guidelines and ensure high-quality data compared to other experimental designs.

However, this review also has some shortcomings. The sample size was suboptimal, and the objective devices that was used to measure physical activity were very costly, with concerns about losing the devices during the intervention. Given the limited availability of such devices, the sample size should meet the minimum required value, and the latest devices, such as the Actigraph GT3X+, should be considered. Furthermore, most of the included studies were conducted in developed countries, with only one study from a developing country, which may limit the generalizability of the results. Despite these limitations, the review incorporates the most up-to-date and comprehensive research available. Lastly, this study's broad age range of participants may introduce variability in interests, abilities, and motor skills. Such developmental differences could affect the effectiveness of PA interventions, necessitating age-specific adaptations to better address the unique needs of each group. Future studies should consider narrowing the age range and accounting for developmental stages when designing interventions to ensure more tailored and effective outcomes.