Effect of intervention on screen time in preschoolers: a systematic review and meta-analysis of randomized controlled trials

Abstract

Background

Excessive screen time (ST) is linked to adverse physical and mental development in children, with habits forming in preschool and solidifying over time. Previous studies on interventions to reduce ST in preschoolers aged 0–7 years have shown inconsistent results. We conducted a systematic review and meta-analysis of randomized controlled trials (RCTs) to evaluate intervention effectiveness and identify potential moderators. 

Methods

We searched CNKI (Chinese), Wanfang Data (Chinese), SinoMed, Ebscohost, Web of Science, EMBASE, PubMed, and Cochrane Library for RCTs (January 1, 2000–December 31, 2024) targeting ST reduction in preschool children. Study quality was assessed using the Cochrane Risk of Bias Assessment tool and Jadad scale. Effect size were reported as standardized mean difference (SMD) with 95% confidence intervals (CI). Subgroup analyses (e.g., sample size, child age, gender, intervention duration) were conducted, with statistical tests (e.g., the chi-square test) comparing subgroup differences.

Results

A total of 41 studies with 14,514 participants were included in the meta-analysis. The overall summary of random effects showed a small, beneficial impact of ST interventions compared to controls (SMD = 0.26, 95% CI: 0.12 to 0.39). Subgroup analysis showed that the intervention effect was moderated by child age (χ² = 6.25, P = 0.04), intervention duration (χ² = 7.38, P = 0.007), target behavior of the intervention (χ² = 6.72, P = 0.01) and intervention recipient (χ² = 10.5, P = 0.01).

Conclusion

Despite heterogeneity in study methods and results, overall interventions to reduce ST in early childhood show significant reductions, suggesting that this may be an opportune time to intervene. Future research should explore strategies for implementing large-scale interventions and sustaining long-term intervention effects, investigate the potential impact of different theoretical frameworks on intervention outcomes, and examine the role of caregivers in supporting behavior change. These findings provide a robust foundation for developing evidence-based strategies to promote healthy ST habits in young children.

Supplementary Information

The online version contains supplementary material available at 10.1186/s12889-025-23700-5.

Background

Screen time (ST) refers to the duration of screen-based activities conducted on visual devices such as televisions, smartphones, computers, video games, tablets, iPads, and other handheld devices [1]. Excessive ST is associated with poor physical and mental development in children [2–4], leading to issues such as overweight/obesity, myopia, depression, and others [5, 6]. ST patterns are formed in preschool years and tend to become ingrained over time [7]. Therefore, it is crucial to manage ST during early childhood.

The World Health Organization recommends that the average daily ST for preschool children should not exceed 1 h per day (h/d) [8]. Many countries have also issued recommendations to limit ST for young children. Canada’s 24-Hour Movement Guidelines suggest that infants aged 0–1 year and toddlers aged under 2 years should not engage in prolonged sedentary ST. Preschoolers (3–4 years old) should not spend more than 1 h per day in sedentary ST [9]. The Australian government advises against prolonged sedentary ST for children under 5 years old, with no single session exceeding 1 h (e.g., in strollers, car seats, or high chairs) [10]. The American Academy of Pediatrics (AAP) recommends: (1) no screen use for children under 18 months; (2) limiting ST to less than 30 min per day for children under 2 years old; and (3) no more than 1 h per day of ST for children aged 2 to 5 years old [11]. The Chinese Ministry of Education and the National Health Commission recommend that parents consciously control the use of electronic devices by children, especially preschoolers (0–6 years old), with a daily cumulative usage time of no more than one hour [12]. However, currently, the ST of preschool children in most countries exceeds these recommendations. In Finland, 22.7% of 18-month-old children use screen media devices for more than 60 min per day [13], and two-thirds of 5-year-olds spend more than 60 min per day in front of screens [14]. In 2020, the average daily total ST for children aged 2 to 4 in the United States was 2.5 h [15]. Data from the United States Centers for Disease Control and Prevention in 2022 showed that more than half of preschool children exceed 1 h of ST per day [16]. In China, more than one-third of preschool children have excessive ST (> 1 h per day) [17, 18]. This highlights the need for interventions to reduce the prevalence of these behaviors.

To effectively reduce ST among preschool children, researchers have developed and implemented various intervention strategies, including parent education [19, 20], school curricula [21, 22], community activities [23, 24], technological tools (such as ST management apps) [25], and comprehensive intervention approaches [26, 27]. However, current intervention studies on excessive ST in preschool children have yielded inconsistent results. For example, interventions based on clinical personalized counseling effectively reduced ST in children aged 2 to 6 years [28], and the use of a TV lockout device intervention resulted in a reduction of 17.5 h per week of ST after six months [25]. Yet, other interventions such as online, face-to-face measures or parent-centered 24-hour movement behavior interventions showed no significant differences in changing ST compared to control groups [27, 29, 30]. Furthermore, despite the effectiveness of some intervention strategies, there is a need for systematic evaluation and summarization regarding the specific efficacy of these interventions, the existence of best practices, and the factors that may influence intervention effectiveness.

Existing meta-analyses on interventions for ST primarily focus on children and adolescents [31–33], with fewer meta-analyses specifically addressing ST interventions for young children. Due to the fact that young children have much less independence than school-aged children and adolescents, their behavior is more strongly influenced by parents and the family environment [34]. Therefore, strategies proven effective in older children may not translate well to this younger population. The purpose of this study is to evaluate the impact and effectiveness of different ST interventions on preschoolers (0–7 years old) by conducting a systematic review and meta-analysis of existing randomized controlled trials (RCTs) targeting this age group. It aims to explore the heterogeneity of intervention effects and their potential influencing factors, thereby providing a scientific basis for the development of future intervention strategies.

Methods

This review is registered with the PROSPERO International Prospective Register of Systematic Reviews (CRD420251009910). The PRISMA Statement [35] guidelines were followed in reporting.

Data source and retrieval strategy

A comprehensive literature search was conducted between January and February 2025. The search encompassed the following databases: China National Knowledge Infrastructure (CNKI), Wanfang Data (in Chinese), SinoMed, Ebscohost, Web of Science, EMBASE, PubMed, and Cochrane Library, targeting publications from January 1, 2000, to December 31, 2024. Database-specific terms were utilized in this study, including keywords such as “intervention OR RCT OR trial” AND “children OR preschool children OR toddler OR infant” AND “television OR computer OR media use OR screen time OR video games OR entertainment media OR sedentary behavior”. Detailed information on the search strategy is provided in Supplementary Tables. Additionally, the reference lists of the included studies were reviewed to identify any further relevant research. One author (YW) screened the titles identified in the initial search. Two authors (CZ and JJ) independently reviewed the abstracts of the included studies; abstracts were excluded when both authors deemed they did not to meet the inclusion criteria for the review. Subsequently, two authors (CZ and YW) examined the full text of the remaining articles to determine final inclusion. Disagreements were resolved through discussions between the aforementioned two authors, and if consensus could not be reached, discussions with all other authors were conducted. Citations of the included studies were uploaded to the EndNote (version X9.2) library for reference management.

Inclusion and exclusion criteria

The inclusion criteria are as follows: (1) participants aged ≤ 7 years (population); (2) behavioral interventions aimed at reducing screen/sedentary time (i.e., computer use, video games, television, etc.), or studies reporting screen/sedentary time reduction as an outcome; (3) published in peer-reviewed Chinese or English journals; (4) randomized controlled trials (study design). In this review, “screen/sedentary time” is defined as screen-based activities (i.e., watching television, playing video games, using computers, etc.). The combination of ST and sedentary time is included, because despite being different behavioral constructs, they are often conflated in the literature. However, for the purposes of this review, interventions specifically targeting the reduction of sitting time (without screen exposure) are excluded. There are no geographical restrictions.

Exclusion criteria: (1) studies in the form of conference reports, review articles, book chapters, policy documents, research proposals, etc.; (2) interventions targeting children/adolescents/adults (> 7 years old) who are not parents or caregivers of children aged 0–7 years; (3) clinically disordered populations, including those with clinical diagnoses (e.g., cancer), psychological disorders (e.g., depression, mood disorders, eating disorders), and developmental disorders (e.g., intellectual developmental disorder); (4) studies involving non-human specimens (e.g., animals, viruses/bacteria, computer algorithms, mathematical simulation models); (5) interventions that do not target a reduction in screen/sedentary time or do not report screen/sedentary time as an outcome; (6) interventions specifically targeting the reduction of sitting or seated time; (7) non-randomized controlled trials (study design); (8) The data type of ST is a non-continuous variable.

Study screening and data extraction

Two researchers (YW and CZ) independently conducted information extraction using standardized forms. In case of disagreement, all other researchers were consulted and reached an agreement. The extracted data included the year of publication, sample size, intervention measures (refer to the specific methods taken to reduce preschool children’s ST, such as psychological counseling, curriculum design, goal setting, etc.), site, period, outcomes, etc. If a study collected data at multiple time points, only the baseline level and the level closest to the end of the intervention were used. If multiple intervention groups were compared with the same control group, according to the method provided in the Cochrane Handbook [36], these intervention groups can be combined into a single composite intervention group and then compared with the control group.

Quality assessment

The quality evaluation was carried out according to the Cochrane bias risk assessment tool [37] and Jadad scale [38], which was completed by 2 authors (YW and CZ). The Cochrane Risk of Bias tool assesses literature quality and risk of bias from seven aspects: (1) random sequence generation (selection bias); (2) allocation concealment (selection bias); (3) blinding of participants and personnel (performance bias); (4) blinding of outcome assessment (detection bias); (5) incomplete outcome data (attrition bias); (6) selective reporting (reporting bias); and (7) other bias. Based on the instructions accompanying the tool, each component was assigned a quality score of low, unclear and high risk of bias. Components that were not reported were given an unclear rating. The Jadad scale was scored by evaluating the generation of random sequences, randomization hiding, blindness (“appropriate” = 2, “unclear” = 1, “inappropriate” = 0), and withdrawal (“described” = 1, “not described” = 0). A score of 0–3 was low quality, and a score of 4–7 is regarded as high quality. Disagreements between the two investigators in assessing the risk of bias were resolved by consensus or consultation with other investigators.

Statistical analysis

Meta-analyses were conducted utilizing Review Manager V.5.3 (RevMan). ST reported in each study was extracted. There were variations in the reported ST across different studies. Some studies reported the total daily ST, while others provided ST for specific periods (e.g., weekdays vs. weekends). ST was reported either as the cumulative total across all screen-based devices or as estimates for “specific devices,” such as minutes spent watching TV or playing video games per day. Information on how ST was measured (i.e., time of day, device-specific time, weekdays vs. weekends) was extracted from each study, and standardized it into average daily ST (min/d). When the average daily ST on weekdays and the average daily ST on weekends are known, the average daily ST for children can be calculated using the following formula:

The baseline and post-intervention mean, standard deviations (SD), and sample sizes (N) for both the intervention and control groups were extracted and included in RevMan. When an interquartile range (IQR) was reported, the SD was calculated using the following formula, . When a standard error (SE) was reported, the SD was calculated using the following formula, . In the reports, adjusted mean differences were utilized. The meta-analysis employed a random-effects model to assess heterogeneity through the χ² and I² statistics. Effect size were reported as standardized mean difference (SMD) with 95% confidence intervals, where SMD represents the difference between groups in standard deviation units (Cohen’s interpretation: 0.2=small, 0.5=medium, 0.8=large) [36]. A χ²-value p ≤ 0.05 was considered indicative of significant heterogeneity, with I² values interpreted as low (25%), moderate (50%), and high (75%) heterogeneity. We predetermined that, in the presence of high heterogeneity, subgroup analyses would be conducted based on sample size, child age, child gender, intervention duration, intervention setting, measurement tools for ST, target behavior (whether the intervention directly aimed to reduce ST or focused on increasing physical activity, weight control, or improving healthy lifestyle behaviors while reporting ST outcomes), intervention recipient, intervention approach, intervention theory and regions of intervention.

The presence of publication bias was evaluated using the Begg and Egger tests and funnel plots. To evaluate the reliability of the study results, sensitivity analysis was performed by sequentially excluding one study at a time to examine whether each individual study had a significant influence on the pooled effect size. Statistical analyses were performed using R software (Version 4.4.2). The results were considered statistically significant at P values < 0.05.

Results

Study identification and selection

A detailed flow diagram of the study selection process for the meta-analysis is presented in Fig. 1. A total of 18,677 articles were retrieved (PubMed: 3,092, EBSCO: 2,681, Web of Science: 12,251, Embase: 633, and 20 from the reference lists of relevant journal articles). Among these, 8,378 studies were excluded due to duplication, and 10,228 records were excluded after reviewing titles and abstracts. The full texts of the remaining 71 articles were assessed for eligibility. Reasons for exclusion included: 4 studies were duplicate reports of already included studies, 10 studies lacked data on the outcome of interest (sedentary time was not screen-based behavior time or the outcome data were not continuous variables), 6 studies involved populations outside the preschool age range, 6 studies were not randomized controlled trials, and 4 studies lacked a control group. Ultimately, 41 studies were included in the meta-analysis.

Fig. 1.

Flow diagram of trials included in the meta-analysis

Study characteristics

The main characteristics of the included studies are presented in Table 1. These trials were published between 2000 and 2024. Among the included studies, 20 were conducted in North America [21, 23, 25, 28, 39–54], 9 in Europe [20, 26, 55–61], 8 in Oceania [29, 30, 62–67], and 4 in Asia [19, 27, 68, 69]. The mean age of participants ranged from 0 to 6.5 years. The sample sizes varied from 16 to 2,434, with a total of 14,514 participants. The duration of the interventions ranged from 4 weeks to 156 weeks. The interventions or behavior change strategies covered in the studies included counseling, curriculum (e.g., intervention on ST, dietary trial), environmental change (e.g., posters of the intervention were put up in the classroom), goal setting, material support (e.g., intervention curriculum paper materials, children’s books), screen rules, physical activity, rewards, and social support. Regarding the intervention settings, 14 studies were conducted at home, 10 at school, 10 in the community, and 9 in primary-care settings.

Table 1.

The characteristics of included studies

Author	Years	Participant description			Country	Interventions				Outcomes#	Jadad scores
		N	Mean age, years	Males(%)		Intervention*	Intervention duration, weeks	Intervention sites	Control
Beck et al. [39]	2023	96	0.1	50	USA	IC	65	Primary-care setting	None	HF, ST, SD	7
Birken et al. [53]	2012	132	3.1	46.2	Canada	C, G, I	52	Primary-care setting	C, I	ST, BMI	7
Byrd-Bredbenner et al. [40]	2018	172	2.7	52	USA	C, I, PA	32	Home	C, I	NT	6
Campbell et al. [62]	2013	542	0.3	52.6	Australia	C, IC, I, P, PA, S	60	Primary-care setting	C, UC	BMI, ST	7
Cespedes et al. [41]	2014	445	4.9	52	USA	C, IC, I, EMD	52	Primary-care setting	None	ST	5
De Coen et al. [55]	2012	1806	5	49.9	Belgium	C, IC, PA, I, S	104	Community	None	BMI, PA, ST	3
De Creamer et al. [56]	2016	859	4.4	54.4	Belgium	IC, I, PA	24	School	None	ST	5
Dennison et al. [21]	2004	77	3.9	49.35	USA	IC, E, P, I	6	School	GC	ST, BMI	7
Downing et al. [63]	2018	57	3.1	45.6	Australia	C, G, I, R	6	Home	None	ST	7
Epstein et al. [25]	2008	70	6	52.85	USA	G, R, E, I	106	Home	None	ZBMI, ST, PA	5
Feng et al. [27]	2024	147	4.8	56.46	China	IC, G, I, PA	12	Home	None	PA, ST	7
Fitzgibbon et al. [42]	2006	324	4.3	50.6	USA	IC, PA, I, R	14	School	I, R	BMI, PA, ST	3
Fitzgibbon et al. [43]	2011	543	4.3	47.8	USA	IC, PA, S, I	14	School	GC, I	BMI, PA, ST	3
Fitzgibbon et al. [44]	2013	116	4.5	46.9	USA	IC, C, PA, E	14	Home	GC	BMI, PA, ST	5
French et al. [45]	2018	530	3.4	50	USA	C, IC, E	156	Primary-care setting	UC	BMI, DI, PA, ST	7
Haines et al. [46]	2013	111	4.1	52.3	USA	C, G, R, I	24	Community	R, I	SD, ST	5
Hammersley et al. [64]	2019	78	3.5	50	Australia	C, IC, I, S	11	Community	I	BMI, ST	5
Hammersley et al. [29]	2021	240	3.4	52.4	Australia	C, S, I, G	13	Community	I	BMI, ST	5
Handel et al. [57]	2017	307	4	57.5	Denmark	C, IC	60	Community	None	PA, BMI, ST	7
Hinkley et al. [67]	2015	16	2.9	63.6	Australia	C, G, R	5	Home	None	ST	7
Kaur et al. [68]	2024	340	2–5	50.9	Indian	C, E, R, S	8	Home	UC	ST	7
Knowlden et al. [47]	2016	44	4–6	43.9	USA	IC, I, S	4	Home	GC, I, S	PA, NT, ST	3
Kong et al. [48]	2016	553	1	46.8	USA	C, IC	14	School	GC	ZBMI, ST, HEI	7
Latomme et al. [58]	2017	2434	4.7	52.2	European	C, IC, E, I	48	School	None	ST	4
Lerner-Geva et al. [69]	2015	137	4–6	50.98	Israel	IC, PA, I	16	School	None	BMI, PA, ST	3
Lin et al. [19]	2021	129	5.6	47.3	China	IC, I	8	School	I	ST, SD	6
Marsh et al. [65]	2020	54	2.6	Not provided	New Zealand	C, I, P	6	Home	C	ST, SD	4
Martínez-Andrade et al. [23]	2014	306	3.4	52.6	Mexico	IC, S, PA, P	6	Primary-care setting	None	BM, ST, PA	6
Mendoza et al. [49]	2016	160	4.5	52.5	USA	IC, I	8	Community	None	ST	3
Morgan et al. [30]	2022	125	3.9	60.8	Australia	S, I	8	Community	None	PA, ST	5
Ostbye et al. [50]	2012	308	3.1	55.8	USA	C, IC, E,I	32	Home	I	DI, MVPA, ST, ZBMI	6
Puder et al. [26]	2011	652	5.2	50	Switzerland	PA, IC, E, I	36	School	S	BMI, PA, ST	3
Ray et al. [20]	2020	586	5.2	53.2	Finland	IC, E	23	School	None	ST, PA	3
Skouteris et al. [68]	2016	151	2.8	49.7	Australia	PA, IC, I	10	Community	None	NT, ST	5
Taveras et al. [28]	2011	446	4.9	52	USA	C, I	104	Primary-care setting	None	BMI, ST, PA	4
Tomayko et al. [51]	2018	450	3.8	49.7	USA	S, I	104	Home	I	ZBMI, ST, PA	4
Tuominen et al. [59]	2017	164	6.5	50.25	Finland	PA	7	Home	None	PA, ST	5
Van Grieken et al. [60]	2014	298	5.8	38.1	Netherlands	C, I, IC, P, G	104	Primary-care setting	C, UC	PA, ST	6
Verbestel et al. [61]	2013	98	1.3	54.3	Belgium	C, I, G	52	School	None	ZBMI, NT, PA, ST,	3
Walton et al. [54]	2016	48	3	41.75	Canada	IC, I, G, R	9	Community	GC	SE, ST	4
Yilmaz et al. [52]	2014	363	3.5	65.56	USA	I, G, PO, C	8	Home	None	ST, BMI	5

^*C Counseling, IC Curriculum (e.g., intervention on ST, dietary trial), GC General curriculum, E Environmental change, G Goal setting, I Material support, P Screen rules, PA Physical activity, R Rewards, S Social support, UC Usual care, EMD Electronic monitoring device

^#BMI Body mass index, ZBMI z-scored BMI, ST Screen time, SE Self-efficacy, SD Sleep duration, NT Nutrition trial, HF Healthy Feeding, DI Dietary Intake, HEI Healthy Eating Index, MVPA Moderate to vigorous physical activity

Quality assessment

Figure 2 summarizes the details of the Cochrane risk of bias assessment. Thirty-eight trials provided detailed descriptions of random sequence generation, and twenty-six trials reported appropriate allocation concealment. Fifteen trials reported that both participants and personnel were blinded to the nature of the intervention. Twenty-four trials reported blinding of outcome assessment. Twenty-nine studies explained the reasons for attrition and conducted appropriate data handling in their analyses. All included trials were judged to have a low risk of reporting bias and other biases. According to the Jadad scale assessment, 32 studies with a score of 4 or higher, 78.05% were of high quality, as shown in Table 1.

Fig. 2.

Forest plot of mean differences in screen time (minutes/day) between intervention and control groups

Meta-analysis of intervention effects

The results of the random-effects model analysis showed that the overall mean difference in ST was SMD = 0.26, 95% CI (0.12, 0.39), with a significant overall effect (Z = 3.81, p = 0.0001). Heterogeneity was high, with an I² value of 93%, indicating significant heterogeneity among studies. Figure 2 shows the forest plot of the mean difference in daily ST (in minutes).

Subgroup analysis for moderators of the intervention effects

The meta-analysis indicated the presence of heterogeneity. As pre-specified, we conducted subgroup analyses based on sample size, child age, child gender, intervention duration, intervention setting, measurement tools for ST, target behavior (whether the intervention directly aimed to reduce ST or focused on increasing physical activity, weight control, or improving healthy lifestyle behaviors while reporting ST outcomes), intervention recipient, intervention approach, intervention theory and regions of intervention. The results of the subgroup analyses are presented in Table 2.

Table 2.

Subgroup analysis of intervention on ST in preschoolers

Subgroup	Number of studies	Effect size		Test		Heterogeneity within subgroups			Test for subgroup differences
		SMD (min/day)	95% CI	Z	p value	χ2	I2 (%)	p value	χ2	p value
Sample size*
≤ 111	11	−0.07	(−0.48, 0.33)	0.36	0.72	71.52	86	< 0.001	4.18	0.24
111–172	10	−0.43	(−0.84, −0.02)	2.07	0.04	128.45	93	< 0.001
172–446	10	−0.42	(−0.79, −0.05)	2.24	0.03	270.00	97	< 0.001
≥ 446	10	−0.13	(−0.21, −0.06)	3.74	< 0.001	22.27	60	0.008
Child age
≤ 3 years [39, 40, 48, 54, 61, 62, 65–68]	10	−0.14	(−0.25, −0.03)	2.58	0.01	11.81	24	0.22	6.25	0.04
3–5 years [21, 23, 27–30, 41–47, 49–53, 56–58, 63, 64, 69]	24	−0.38	(−0.58, −0.17)	3.64	< 0.001	441.97	95	< 0.001
≥ 5 years [19, 20, 25, 26, 55, 59, 60]	7	0.03	(−0.25, 0.30)	0.18	0.85	76.82	92	< 0.001
Sex of child								< 0.001
Boys > 50%	21	−0.3	(−0.54, −0.06)	2.49	0.01	497.11	96	< 0.001	1.00	0.32
Girls > 50%	19	−0.17	(−0.26, −0.09)	3.86	< 0.001	39.34	54	0.003
Duration of intervention
<6 months [19–21, 23, 27, 30, 42–44, 46–49, 52, 54, 56, 59, 63–69]	24	−0.44	(−0.69, −0.20)	3.57	< 0.001	412.85	94	< 0.001	7.38	0.007
≥ 6 months [25, 26, 28, 29, 39–41, 45, 50, 51, 53, 55, 57, 58, 60–62]	17	−0.07	(−0.19, 0.05)	1.19	0.23	96.25	83	< 0.001
Setting								< 0.001
Home	13	−0.26	(−0.73, 0.22)	1.06	0.29	331.43	96	< 0.001	1.13	0.77
School	11	−0.18	(−0.29, −0.06)	3.06	0.002	39.57	75	< 0.001
Community	9	−0.38	(−0.75, −0.02)	2.04	0.04	127.22	94	< 0.001
Primary-care setting	8	−0.19	(−0.32, −0.07)	3.03	0.002	17.63	60	0.01
Measurement tool
Parent-Report	31	−0.25	(−0.38, −0.11)	3.61	< 0.001	350.7	91	< 0.001	0.02	0.88
Objective Tool#	10	−0.28	(−0.71, 0.15)	1.28	0.20	187.9	95	< 0.001
Behavior targeted								< 0.001
Targeted ST alone [21, 41, 45, 49, 52, 53, 56, 58, 63, 67, 68]	11	−0.69	(−1.10, −0.27)	3.24	0.001	415.26	98	< 0.001	6.72	0.01
Targeted ST and others [19, 20, 23, 25–30, 39, 40, 42–44, 46–48, 50, 51, 54, 55, 57, 59–62, 64–66, 69]	30	−0.13	(−0.22, −0.04)	2.72	0.006	112.16	74	< 0.001
Intervention recipient
Child Only [21, 26, 41, 43, 55–57, 69]	8	−0.23	(−0.34, −0.11)	3.93	< 0.001	21.72	68	0.003	10.5	0.01
Parent Only [29, 40, 52, 54, 60–65, 67]	11	−0.44	(−0.95, 0.08)	1.67	0.09	262.79	96	< 0.001
Child and Parent [19, 23, 25, 27, 28, 30, 39, 42, 44–47, 49–51, 53, 59, 66, 68]	19	−0.2	(−0.42, 0.02)	1.79	0.07	207.59	91	< 0.001
Teacher [20, 48, 58]	3	−0.03	(−0.11, 0.05)	0.72	0.47	2.33	14	0.31
Methods of intervention
Face-to-face Intervention	21	−0.17	(−0.34, −0.00)	1.99	0.05	221.62	91	< 0.001	1.19	0.55
Online Intervention	11	−0.41	(−0.88, 0.05)	1.74	0.08	262.29	96	< 0.001
Hybrid Intervention	9	−0.25	(−0.35, −0.15)	4.80	< 0.001	18.43	57	0.02
Theoretical frameworks
Yes	22	−0.38	(−0.63, −0.14)	3.08	0.002	405.04	95	< 0.001	2.99	0.08
No	19	−0.14	(−0.27, −0.02)	2.25	0.02	116.97	85	< 0.001

^*<25th percentile = sample size ≤ 111, 26-50th percentile = 111–172, 51-75th percentile = 172–446, 76-95th percentile = 446–2434

^#Objective Tool = Accelerometer, TV Time Monitor and Environmental Audit

The subgroup analysis of mean differences in ST revealed significant variations across subgroups stratified by child age (P for interaction = 0.04), intervention duration (P for interaction = 0.007), target behavior of the intervention (P for interaction = 0.01), and intervention recipient (P for interaction = 0.01). Significant reductions in ST were observed in studies involving children younger than 3 years (SMD = 0.14, 95% CI: 0.03 to 0.25, P = 0.01) or aged 3 to 5 years (SMD = 0.38, 95% CI: 0.17 to 0.58, P < 0.001), interventions lasting no more than 6 months (SMD = 0.44, 95% CI: 0.20 to 0.69, P < 0.001), interventions specifically aimed at reducing ST (SMD = 0.69, 95% CI: 0.27 to 1.10, P = 0.001), and interventions directly targeting children (SMD = 0.23, 95% CI: 0.11 to 0.34, P < 0.001). In contrast, no significant effects were found in studies involving children aged 5 years or older, interventions lasting 6 months or longer, or interventions directly targeting parents or teachers. Even after excluding studies that contributed to heterogeneity (based on quality assessment results, low-quality studies were removed), the moderating effects of sample size, child gender, intervention setting, measurement tools for ST, intervention approach, theory-based interventions and regions of intervention were not significant (P > 0.05).

ST interventions demonstrated the greatest effect size in studies with sample sizes ranging from 111 to 172 (SMD: 0.43, 95% CI: 0.02 to 0.84, P = 0.04), with a downward trend in intervention effectiveness as sample sizes increased (from 172 to 446 to ≥ 446) (SMD: 0.42 vs. 0.13, P < 0.05). Significant impacts were noted for both boys and girls, with a greater effect size observed in boys compared to girls (SMD: 0.30 vs. 0.17, P < 0.05). In terms of intervention settings, school-based, community-based, and primary-care setting-based interventions all showed significant effects, with community-based interventions yielding the largest effect size compared to school-based and primary-care setting-based interventions (SMD: 0.38 vs. 0.18 vs. 0.19, P < 0.05). Compared to studies where children’s ST was measured using objective tools (e.g., accelerometer, TV time monitor, and environmental audit), interventions relying on parent-reported ST showed a greater effect size (SMD: 0.25, 95% CI: 0.11 to 0.38, P < 0.001). Among the methods of intervention, blended interventions combining face-to-face and online approaches showed significant effects (SMD: 0.25, 95% CI: 0.15 to 0.35, P < 0.001), followed by face-to-face interventions, which demonstrated marginally significant effects (SMD: 0.17, 95% CI: 0.00 to 0.34, P = 0.05). Interventions supported by theoretical frameworks exhibited greater effect sizes (SMD: 0.38, 95% CI: 0.14 to 0.63, P = 0.002) compared to those without theoretical support (SMD: 0.14, 95% CI: 0.02 to 0.27, P = 0.02).

Publication bias

To assess potential publication bias, we performed Egger’s regression test and Begg’s rank correlation test. Egger’s test indicated significant publication bias (intercept = −0.123, p = 0.036), suggesting that smaller studies with larger effect sizes may have been over represented. Begg’s test further supported the presence of publication bias (Kendall’s tau = 0.234, p = 0.045). In addition, a funnel plot was drawn to test publication bias, as shown in Fig. 3. The funnel plot exhibited asymmetry, indicating potential publication bias. The Trim and Fill method identified 3 potentially missing studies. After adjustment, the overall effect size decreased from 0.26 (95% CI: 0.12, 0.38) to 0.21 (95% CI: 0.11, 0.31), but remained statistically significant (p < 0.05).

Fig. 3.

Funnel plot of publication bias of included studies

Sensitivity analysis

To assess the influence of individual studies on the overall results, we conducted sensitivity analyses by sequentially excluding each study. The pooled mean difference (95% CI) for ST ranged from 0.19 (0.10, 0.29) (excluding the trial by Yilmaz et al.) to 0.30 (0.17, 0.43) (excluding the trial by Epstein et al.). After removing individual studies, substantial heterogeneity remained across the studies, and the pooled effect sizes did not significantly differ from the overall effect size obtained from the included studies. The sensitivity analysis indicated that the pooled effect size for the impact of interventions on ST among preschool children was robust and stable.

Discussion

This study systematically reviews interventions aimed at altering ST among preschool children. The review included 41 randomized controlled trials. Meta-analysis results indicate that, compared to control groups, ST interventions had a significant positive effect on reducing ST among preschool children. The effect size and direction in this meta-analysis are consistent with previous meta-analyses on behavioral interventions to reduce children’s ST [31, 32, 70]. Additionally, this study found that the age of children, intervention duration, intervention goals, and direct intervention recipients may be important factors influencing the effectiveness of ST interventions, which holds significant implications for designing interventions aimed at reducing ST in preschool children.

To optimize intervention effectiveness and maximize resource utilization, it is necessary to identify the “active ingredients” in behavioral interventions aimed at reducing children’s ST. The study found that characteristics such as younger children age, shorter intervention duration, direct ST reduction as the intervention goal, and children as the direct intervention target were associated with larger effect sizes. Subgroup analysis revealed that interventions were more effective for children under 3 years old and those aged 3–5, compared to children aged 5 or older, with the largest effect observed in the 3–5 age group. A review study targeting children and adolescents aged 0–18 also found that interventions were more effective for children under 6 years old [71]. This suggests that early interventions may have a greater impact, likely due to the greater plasticity of habits and behaviors in younger children [72]. Furthermore, parents may exert stronger control over ST for younger children, whereas increasing autonomy with age may weaken intervention effectiveness. Therefore, interventions targeting younger children should be prioritized, as they may yield more significant and lasting effects.

This study found that interventions lasting less than 6 months were more effective in reducing ST among preschool children compared to those lasting longer than 6 months. Previous reviews on children and adolescents also found that shorter-duration interventions were more effective in reducing ST than longer ones [31, 73]. This may be related to the ease of maintaining participants’ adherence and motivation in short-term interventions [31]. In contrast, long-term interventions may suffer from participant fatigue or declining interest, leading to reduced effectiveness. However, the optimal duration for reducing ST remains controversial [71, 74]. For example, Dobbins et al. (2009) reported that intervention duration was unrelated to reductions in television viewing time [75], while Biddle et al. (2011) found that interventions lasting less than 4 months or more than 1 year appeared effective for reducing ST in children and adolescents [76]. Jones et al. (2021) observed significant reductions in ST in interventions lasting less than 52 weeks [32]. These findings may reflect the challenges of maintaining engagement and adherence in long-term interventions. Given that most interventions are short-term, future interventions should consider incorporating strategies to sustain motivation and participation over longer periods to identify optimal strategies for reducing ST in preschool children.

Interventions directly targeting ST reduction showed significant effects, whereas those aimed at increasing physical activity or improving healthy lifestyles had weaker effects. Similar findings were reported by Buchanan et al. (2016) [77] and Martin et al. (2021) [78], who found that interventions specifically targeting ST were more strongly associated with reductions in ST compared to those addressing multiple lifestyle behaviors. This highlights the importance of designing focused interventions with clear, specific goals. Interventions targeting multiple health behaviors may dilute the effect of reducing ST due to goal dispersion [31]. Additionally, multi-behavior interventions often have longer duration, and behavior changes achieved through one or two sessions on ST may not be sustained when additional topics are introduced [78]. However, if ST remains a primary focus, combining ST reduction with other health-promoting behaviors may still be beneficial [19, 46].

Interventions directly targeting children were more effective than those targeting parents or teachers. This suggests that involving children directly in behavior change may have a greater impact than relying solely on indirect methods through caregivers or educators. However, three direct child-targeted interventions showed non-significant reductions in ST [55, 57, 69], which may be related to the heterogeneity of interventions. Future research should further explore designs targeting different intervention recipients. Additionally, some studies found that a combined approach involving both children and their caregivers may enhance effectiveness [19, 27, 28], as parents play a critical role in shaping children’s habits [79].

Among studies with sample sizes of 111–172, ST interventions showed the largest effects, with intervention effectiveness declining as sample sizes increased. This pattern suggests that, within a specific range, larger sample sizes may dilute intervention effects. This may be because smaller studies are better able to control for confounding variables, whereas larger studies may include more heterogeneous populations, thereby reducing intervention effectiveness [80]. This study found that interventions were more effective for boys than girls, consistent with previous research [74] indicating that boys benefit more from interventions. This may be related to boys’ greater interest in electronic media and longer ST compared to girls [81, 82].

In terms of intervention settings, school-, community-, and primary care-based interventions all showed significant effects, with community-based interventions having the largest effect size, consistent with previous findings [70, 76]. However, this study identified high heterogeneity among studies, necessitating further exploration of designs for different intervention settings. Previous research suggests that intervention settings have a limited impact on ST interventions [78, 83, 84]. Hybrid interventions (combining face-to-face and online approaches) showed larger effects, likely due to their flexibility and ability to engage participants through multiple channels, whereas purely face-to-face interventions showed only marginal significance. Previous research found that face-to-face interventions were more effective than online or indirect interventions [78]. This suggests that combining multiple intervention modalities may better align with the behavioral habits and needs of modern children. Additionally, subgroup analysis revealed that interventions supported by theory (e.g., social cognitive theory, self-efficacy theory, self-determination theory, chronic care model) were significantly more effective than those without theoretical support. This finding underscores the importance of theoretical frameworks in intervention design, as they provide scientific rationale and logical support [52, 62], thereby enhancing intervention effectiveness. Future research should place greater emphasis on theory-based intervention design and explore the potential impact of different theoretical frameworks on intervention outcomes.

Compared to studies using objective measurement tools (e.g., accelerometers), interventions relying on parent-reported child ST showed larger effect sizes. This finding raises critical questions about the mechanisms underlying this difference. One possible explanation is social desirability bias [85], where parents may under report their children’s ST to align with intervention goals, thereby exaggerating perceived effectiveness. Another explanation is that self-reporting may heighten parental vigilance, promoting behavior change as parents actively monitor and limit their children’s screen use, a process not directly captured by objective measures [86]. This aligns with previous evidence that subjective reports often reflect intentional behavior change, whereas objective measures capture more nuanced and less malleable behavioral habits [87]. However, parent-reported data have limitations, including recall bias and cultural differences in perceptions of ST [88, 89], necessitating cautious interpretation. Future research should consider mixed methods, combining objective monitoring with parent engagement strategies to balance feasibility and rigor. These findings highlight the need for standardized, context-sensitive measurement frameworks in ST research.

The consistency of the results before and after the Trim and Fill adjustment suggests that the intervention effect is robust to publication bias. However, the presence of publication bias highlights the need for caution in interpreting the magnitude of the effect size, as it may have been overestimated. Future studies should consider including unpublished data or gray literature to minimize this bias.

This meta-analysis has several strengths. Firstly, in terms of study design, it exclusively includes RCTs, ensuring high-quality research design and reliable causal inference. The study population (preschool children aged 0–7), intervention types (ST/screen-based sedentary behavior), and outcome measures are clearly defined, ensuring homogeneity and comparability. Secondly, in data collection, the latest research data are incorporated, reflecting the most recent advancements in ST interventions for preschool children. Major Chinese and English databases are covered to reduce publication bias. Lastly, in methodology, random-effects models, trim-and-fill methods, and other techniques are employed to comprehensively assess effect sizes and heterogeneity. Subgroup analyses explore sources of heterogeneity (e.g., study subjects, intervention duration, intervention methods, sample size), providing deeper insights. Sensitivity analyses and publication bias tests verify the robustness of the results, ensuring conclusions are not influenced by individual studies. The study quantifies the impact of ST interventions on preschool children and identifies the most effective intervention characteristics (e.g., age group, duration) through subgroup analyses, offering evidence for policy-making and parental practices.

There are some limitations of this review. Firstly, some studies reported non-continuous measures of children’s ST, which could not be included in the meta-analysis. The inclusion of these studies might have altered the observed outcomes. Secondly, the limitations of the individual studies included in this review must also be considered. Some pilot studies with relatively small sample sizes were included. These studies may have lacked sufficient statistical power to detect minor changes in ST, which could influence the results of the meta-analysis. Furthermore, all included trials in this study were assessed to have a low risk of reporting bias and other biases. However, it was still identified that some studies exhibited a high risk of bias in terms of blinding and allocation concealment. It must be acknowledged that these represent sources of heterogeneity, and the generalizability of the findings of this study requires careful consideration, although sensitivity analysis did not reveal any impact on the stability of the study outcomes. Thirdly, the interventions in the included studies were primarily conducted in the United States or other developed countries (e.g., Canada, Australia, Finland) and were predominantly located in urban areas, which may limit the generalizability of the findings to rural or less-developed settings. Therefore, caution should be exercised when generalizing our findings to other populations, such as those in developing countries and rural regions. Fourthly, the studies included in this review exhibited significant heterogeneity in intervention objectives, settings, methodologies, and modalities, complicating the comparability of the results. The high degree of heterogeneity observed in most of the meta-analyses underscores this issue. Fifthly, in this review, two of the included studies featured two intervention groups each. Given the similarity in their intervention methods, the data from these intervention groups were combined following the approach recommended by the Cochrane Handbook. However, it should be noted that this merging process may introduce some degree of heterogeneity. Lastly, while our analysis encompassed children aged 0–7 years to accommodate global variability in preschool age definitions, subgroup analyses revealed significantly stronger intervention effects among 3-5-year-olds. This indicates that although including a broader age range maintains developmental continuity in ST behaviors, it may mask optimal intervention windows. Future studies should focus on 3-5-year cohorts to optimize intervention design while retaining broader age ranges in policy-oriented studies to ensure global applicability.

Conclusions

This meta-analysis demonstrates that ST interventions are effective in reducing ST among preschool children, particularly for younger age groups (under 5 years), shorter intervention duration (less than 6 months), interventions directly targeting ST reduction, and interventions where children are the primary recipients. Community-based interventions and hybrid interventions appear particularly promising due to their adaptability and potential for scalability. Future research should explore strategies for implementing large-scale interventions and sustaining long-term intervention effects, investigate the potential impact of different theoretical frameworks on intervention outcomes, and examine the role of caregivers in supporting behavior change. These findings provide a robust foundation for developing evidence-based strategies to promote healthy ST habits in young children.

Authors’ contributions

SY, YW and XX conceptualized and designed the study, drafted the initial manuscript, and critically reviewed and revised the manuscript. YW, CZ and JJ were in charge of the collected data. All authors have read and approved the final manuscript.

Funding

This study was funded by the Major Humanities and Social Sciences Research Projects in Zhejiang higher education institutions (2023QN019), Zhejiang Provincial Philosophy and Social Sciences Planning Research Project (25NDJC162YBMS).

Data availability

The datasets used and/or analyzed during the current study available from the corresponding author on reasonable request.

Declarations

Ethics approval and consent to participate

Ethics approval was obtained from the Ethics Committee of The First Hospital of Jiaxing (2023-LY-024). All methods were performed in accordance with the relevant guidelines and regulations (Declaration of Helsinki).

Consent for publication

Not applicable.

Competing interests

The authors declare no competing interests.