Efficacy of Gamified Digital Mental Health Interventions for Pediatric Mental Health Conditions

Key Points

Question

Given the challenges confronting the accessibility of pediatric mental health care, are gamified digital mental health interventions (DMHIs) effective for treating pediatric anxiety, depression, and attention-deficit/hyperactivity disorder (ADHD)?

Findings

This systematic review and meta-analysis found that gamified DMHIs were effective in improving the severity of ADHD and depressive symptoms in randomized clinical trials (RCTs) but not for reducing the severity anxiety symptoms. Moderator analyses revealed that intervention delivery modality and sex differences influenced treatment effects of DMHIs for ADHD, whereas preset time limits improved treatment effects of DMHIs for depressive disorders.

Meaning

The findings suggest that gamified DMHIs may constitute a key piece of a comprehensive pediatric mental health treatment plan for youth with ADHD or depression, and future research should consider identified moderators to optimize treatment outcomes.

This systematic review and meta-analysis evaluates the efficacy of gamified digital mental health interventions in treating anxiety, depression, and attention-deficit/hyperactivity disorder among children and adolescents.

Abstract

Importance

Anxiety, depression, and attention-deficit/hyperactivity disorder (ADHD) affect up to 20% of children and adolescents. Despite demonstrated efficacy, evidence-based treatments for these conditions are often inaccessible; innovative solutions are essential to meet the demand for pediatric mental health care.

Objective

To examine the efficacy and moderators of gamified DMHIs for anxiety, depression, and ADHD in randomized clinical trials (RCTs) for children and adolescents.

Data Sources

A systematic search of PubMed, PsycInfo, and Web of Science was conducted for RCTs published before March 20, 2024.

Study Selection

RCTs that evaluated the efficacy of gamified DMHIs for treating pediatric ADHD, depression, or anxiety were included. Studies were excluded if they did not use a gamified DMHI, provide sufficient data for effect sizes, or were unavailable in English.

Data Extraction and Synthesis

Efficacy data were extracted from rating scales for ADHD, depression, and anxiety. Extracted moderator variables included participant characteristics (eg, age and sex), intervention characteristics (eg, delivery modality and time limit), and trial design characteristics (eg, outcome measure and risk of bias).

Main Outcomes and Measures

The primary outcome was change in ADHD, depression, or anxiety severity in the treatment group compared to the control group. Hedges g quantified treatment effects.

Results

The search strategy identified 27 RCTs that included 2911 participants across ADHD, depression, and anxiety disorders. There were modest significant effects of gamified DMHIs on ADHD (g, 0.28; 95% CI, 0.09 to 0.48) and depression (g, 0.28; 95% CI, 0.08 to 0.47) but small, nonsignificant effects for anxiety disorders (g, 0.07; 95% CI, −0.02 to 0.17). Moderator analyses revealed that DMHIs for ADHD delivered on a computer and those RCTs that had a greater preponderance of male participants produced larger treatment effects. DMHIs for depressive disorders that used preset time limits for gamified DMHIs also exhibited larger treatment effects.

Conclusions and Relevance

The findings suggest a benefit of gamified DMHIs for youth with ADHD or depressive disorder. Pediatricians and other health care professionals have new information about novel, accessible, and efficacious options for pediatric mental health care.

Introduction

Anxiety, depression, and attention-deficit/hyperactivity disorder (ADHD) are common mental health conditions that affect up to 20% of children and adolescents.¹ These conditions often co-occur² and are characterized by symptoms, including low mood, excessive worry, trouble with attention, hyperactivity, and impulsivity—which can cause severe impairment across life domains.³ Left untreated, these conditions place youth at risk of the development of severe psychopathology and health-related sequalae in adulthood.⁴ Consequently, the effective treatment of anxiety, depression, and ADHD is a pressing public health concern that has implications across the life span.⁵

Fortunately, evidence-based treatments exist. While behavioral therapies such as parent training are efficacious for the management of ADHD, cognitive behavioral therapy has shown efficacy for treating anxiety and depression in youth.^6,7,8 Stimulant and nonstimulant medications have shown therapeutic benefit for ADHD,^7,9 and selective serotonin reuptake inhibitors have demonstrated efficacy for anxiety and depression.^6,8 However, evidence-based psychotherapies are often limited to specialty clinics and restricted by the number of trained health care professionals, which results in long wait times to access care.^10,11,12 Similarly for pharmacotherapy, there is limited accessibility to trained child and adolescent psychiatrists. Moreover, some evidence-based pharmacotherapies are controlled substances (eg, stimulant medications) that require regular in-person visits that may not be feasible for patients and families (although this requirement has been suspended through December 31, 2024, due to the COVID-19 pandemic¹³). Indeed, experts suggest that doubling the mental health care workforce would yield only a minor public health impact.¹⁴ While pediatric primary care settings have served as intermediaries to provide mental health care services,^15,16 innovative solutions are needed to address the increasing demand.

Digital mental health interventions (DMHIs) comprise activities accessed via technology platforms—computers, tablets, video game consoles, and smartphones—that aim to improve the user’s mental health.¹⁷ DMHIs have demonstrated some efficacy for treating mental health conditions.^18,19 However, engagement with therapeutic content is a well-known challenge for DMHIs.²⁰ For example, a teenager may download an efficacious DMHI for anxiety or depression, but the patient still needs to open the app, engage with therapeutic content, and implement therapeutic skills—all of which can prove challenging for youth who have low motivation or struggle with inattention. One approach to improve engagement is gamification, which refers to “the process of adding games or gamelike elements to something (such as a task) so as to encourage participation.”²¹ Initial evidence suggests that gamified DMHIs can increase engagement among young people relative to traditional therapies.²² Thus, gamified DMHIs represent a novel way to deliver evidence-based mental health care in an accessible, scalable, and engaging manner.

In response, we conducted a meta-analysis to examine the therapeutic effects of gamified DMHIs for anxiety, depression, and ADHD. Furthermore, we evaluated moderators of treatment effects from DMHIs investigating participant-level characteristics, intervention-level characteristics, and trial-design characteristics. Findings can help frontline clinicians (eg, primary health care professionals) make informed recommendations about using gamified DMHIs as part of a comprehensive mental health treatment plan.

Methods

Design, Search Strategy, and Eligibility Criteria

Following the Preferred Reporting Items for Systematic Reviews and Meta-Analyses (PRISMA) guidelines, PubMed, Web of Science, and PsycInfo were searched using the following key terms: “ADHD” or “depression” or “anxiety” and “youth” or “children” or “adolescents” focusing on a “videogame” or “computer game” or “game-based” intervention (see the eMethods in Supplement 1 for exact search terms). The initial search was conducted in April 2023 and updated to incorporate any recent published reports through March 20, 2024 (Figure 1²³). Identified titles and abstracts were reviewed for appropriateness independently by 2 raters. The references of pertinent trials and review articles were also searched. Identified abstracts and citations were evaluated for the following inclusion criteria: (1) a randomized clinical trial (RCT) that used a gamified DMHI to reduce anxiety, depression, or ADHD severity; (2) evaluated the efficacy of the gamified DMHI relative to a control condition; (3) included only participants younger than 18 years; (4) available in English; and (5) provided sufficient data to allow for treatment effect size calculation. When insufficient data were available, attempts were made to contact study authors to obtain data.

Figure 1. PRISMA Flow Diagram of Selection of Randomized Controlled Trials (RCTs).

ADHD indicates attention-deficit/hyperactivity disorder.

^aOne article²³ included 2 gamified digital mental health intervention (DMHI) conditions, yielding an analysis of 27 RCTs in total.

Data Extraction

Data extraction was completed independently by 2 raters (B.R.B. and M.R.S.). When raters disagreed, consensus was achieved through consultation with the senior investigator (J.F.M.). Efficacy data were extracted from severity rating scales for ADHD, depression, and anxiety. A hierarchy of preferred outcomes was established a priori to limit possible reporting bias. If multiple outcome measures were present (eg, 2 anxiety rating scales), priority was given to parent, clinician, and then child-reported rating scales. Extracted moderators included participant characteristics (age, sex, and sample type [clinically significant vs community recruited]), intervention characteristics (specific DMHIs, treatment duration, time limit, aesthetic sophistication, public accessibility, and delivery modality), and trial design characteristics (sample size, attrition, publication year, control condition, outcome measure, and risk of bias). Sample type was categorized as either clinically significant or community recruited. If participants had to meet symptomatic requirements for enrollment (eg, clinician-diagnosed), the sample was classified as clinically significant. Meanwhile, if a study recruited participants without requiring a diagnosis but participants experienced some symptoms or were at risk of developing a disorder (eg, recruitment of youth with elevated anxiety levels from a school), this was classified as community recruited. Treatment duration referred to the number of weeks over which participants used the DMHI to allow for comparability across studies. Time limits were coded in a binary manner based on whether the RCT imposed an upper time limit for DMHI use by restricting access to the DMHI. Aesthetic sophistication was coded in a binary manner based on 3 DMHI characteristics: graphic complexity, concordance with game theme, and consistency across game elements. Public accessibility was determined based on whether the raters could access the DMHI. Risk of bias was calculated for each study using the Cochrane Risk Bias Calculator version 2.²⁴

Effect Size Calculation and Statistical Analyses

Hedges g was selected to quantify treatment effects and was calculated in Comprehensive Meta-Analysis version 3 (Biostat).²⁵ Effect sizes were calculated using change scores to increase the precision of effect size estimators by controlling for pretreatment group differences in symptom severity. Preatment and posttreatment means and standard deviations were entered into Comprehensive Meta-Analysis, and were divided by the pooled posttreatment standard deviation. Effect sizes were standardized so that a positive result indicated that gamified DMHIs performed better than control conditions. Separate random-effects models using inverse variance weights examined the effect sizes of gamified DMHIs for anxiety, depression, and ADHD severity. A random-effects model was selected because the true effect size was expected to vary across RCTs due to different study characteristics.²⁶ Heterogeneity of effect size was assessed using forest plots, the Q statistic, and the I² statistic. Publication bias was assessed by visual inspection of the funnel plot and the Egger test for bias. Moderators were analyzed using either method-of-moments meta-regression using inverse variance weights, or an analog to the analysis of variance. For studies with multiple active intervention arms, the same control condition was used as the comparator to calculate effect size and affected only 1 RCT for ADHD.²³

Results

Search Results and Included RCTs

Figure 1 details the literature search that produced 27 RCTs^{23,27,28,29,30,31,32,33,34,35,36,37,38,39,40,41,42,43,44,45,46,47,48,49,50,51} that met all inclusion criteria. One included ADHD trial had 2 gamified DMHI conditions²³ and 5 included RCTs provided both depression and anxiety outcomes.^{38,39,40,41,42} This provided 32 treatment comparisons for inclusion in this meta-analysis (ADHD: 11 RCTs; depression: 9 RCTs; anxiety: 12 RCTs) with 2911 unique participants (ADHD: 771 participants; depression: 1332 participants; anxiety: 1719 participants). Table 1 displays participant characteristics for the RCTs. Table 2 presents the intervention and trial design characteristics. Table 3 shows the moderator analyses of treatment effects for each condition.

Table 1. Participant Characteristics.

Trial type	Participant characteristics
	Source	Mean age, y	% Female	Sample type
ADHD	Bikic et al,27 2018	9.96	16	Clinically significant
	Dovis et al,23 2015, intervention arm 1	10.55	20	Clinically significant
	Dovis et al,23 2015, intervention arm 2	10.4	21	Clinically significant
	García-Redondo et al,28 2019	11.56	39	Clinically significant
	Kim et al,29 2022	9.1	23	Clinically significant
	Kollins et al,30 2020	9.65	29	Clinically significant
	McDermott et al,31 2020	9.57	30	Clinically significant
	Rajabi et al,32 2020	10.13	0	Clinically significant
	van der Oord et al,33 2014	9.75	18	Clinically significant
	Wang et al,34 2023	8.15	25	Clinically significant
	Yerys et al,35 2019	11.25	11	Clinically significant
Depression	Bohr et al,36 2023	NR	NR	Clinically significant
	David et al,37 2019	12.89	64	Clinically significant
	Fleming et al,38 2012	14.9	44	Community recruited
	Knox et al,39 2011	12.88	0.38	Clinically significant
	McCashin et al,40 2021	9.94	42	Clinically significant
	Merry et al,41 2012	15.56	66	Clinically significant
	Perry et al,42 2017	16.76	63	Community recruited
	Poppelaars et al,43 2016	13.24	100	Clinically significant
	Poppelaars et al,44 2021	17.11	66	Clinically significant
Anxiety	Fleming et al,38 2012a	14.9	44	Community recruited
	Knox et al,39 2011a	12.88	38	Clinically significant
	McCashin et al,40 2021a	9.94	42	Clinically significant
	Merry et al,41 2012a	15.56	66	Clinically significant
	Perry et al,42 2017a	16.76	63	Community recruited
	Sanchez et al,45 2017	8.9	40	Community recruited
	Scholten et al,46 2016	13.27	65	Clinically significant
	Schoneveld et al,47 2016	9.95	58	Clinically significant
	Schoneveld et al,48 2018	9.97	59	Clinically significant
	Schuurmans et al,49 2018	13.97	16	Clinically significant
	Tsui et al,50 2021	13.5	63.80	Clinically significant
	Wijnhoven et al,51 2020	11.1	23	Clinically significant

Abbreviation: NR, not reported.

^a Data also included in depression subset.

Table 2. Trial Design and Intervention Characteristics.

Trial design and intervention characteristics
Source	Sample size	Attrition, %	Comparison condition	Outcome measure	Risk of bias	Specific DHMI intervention	Intervention duration, wk	Time limit (Y/N)	Aesthetically sophisticated (Y/N)	Delivery modality
ADHD trials
Bikic et al,27 2018	70	7	TAU /waitlist	ADHD-RS	Low	ACTIVATE	8	N	N	Computer
Dovis et al,23 2015, intervention arm 1	61	10	Placebo digital intervention	DBRS	Low	Braingame Brian	5	N	N	Computer
Dovis et al,23 2015, intervention arm 2	58	7	Placebo digital intervention	DBRS	Low	Braingame Brian	5	N	N	Computer
García-Redondo et al,28 2019	44	NR	TAU/waitlist	EDAH-ADHD	Low	Boogies Academy and Cuibrain	14	Y	N	Tablet
Kim et al,29 2022	30	0	TAU/waitlist	K-ARS	Low	NeuroWorld DTx	4	Y	Y	Tablet
Kollins et al,30 2020	348	5	Placebo digital intervention	ADHD-RS	Low	AKL-T01/Project EVO	4	N	Y	Tablet
McDermott et al,31 2020	46	13	TAU/waitlist	ADHD-RS	Some concerns	CogoLand	8	Y	N	Computer
Rajabi et al,32 2020	32	NR	TAU/Waitlist	CPRS-R	Some concerns	SmartMind	10	Y	N	Computer
van der Oord et al,33 2014	43	7	TAU/waitlist	DBRS	Some concerns	Braingame Brian	5	Y	N	Computer
Wang et al,34 2023	20	0	Placebo digital intervention	ADHD-RS	Low	DTFAD001	4	N	Y	Tablet
Yerys et al,35 2019	19	5	Educational intervention	ADHD-RS	Low	AKL-T01/Project EVO	4	Y	Y	Tablet
Depression trials
Bohr et al,36 2023	48	33	TAU/waitlist	CESD-R	Some concerns	SPARX	7	N	Y	Computer
David et al,37 2019	109	12	Rational emotive behavior education	EATQ-R depressive mood	Some concerns	REThink	4	Y	N	Computer
Fleming et al,38 2012	32	6	TAU/waitlist	CDRS-R	Low	SPARX	5	Y	Y	Computer
Knox et al,39 2011	30	20	TAU/waitlist	CDI	High	Freeze-Framer 2.0 and Journey to the Wild Divine: The Passage	8	Y	Y	Computer
McCashin et al,40 2021	122	34	TAU/waitlist	DSM-problem subscales: depressive	Low	Pesky gNATs	7	N	N	Computer
Merry et al,41 2012	187	9	TAU/waitlist	CDRS-R	Low	SPARX	7	N	Y	Computer
Perry et al,42 2017	540	75	Placebo digital intervention	MDI	High	SPARX-R	7	Y	Y	Computer
Poppelaars et al,43 2016	102	5	Active monitoring of depressive symptoms	RADS-2	Low	Journey	7	N	Y	Videogame console
Poppelaars et al,44 2021	162	4	Placebo digital intervention	CDI	Low	Journey	4	N	Y	Videogame console
Anxiety trials
Fleming et al,38 2012	32	6	Waitlist/TAU	SCAS	Low	SPARX	5	Y	Y	Computer
Knox et al,39 2011	30	20	Waitlist/TAU	MASC	High	Freeze-Framer 2.0 and Journey to the Wild Divine: The Passage	8	Y	Y	Computer
McCashin et al,40 2021	122	34	Waitlist/TAU	CBCL DSM-problem subscales: anxiety	Low	Pesky gNATs	7	N	N	Computer
Merry et al,41 2012	187	9	Waitlist/TAU	SCAS	Low	SPARX	7	N	Y	Computer
Perry et al,42 2017	540	75	Digital placebo	SCAS GAD and SOC	High	SPARX-R	7	Y	Y	Computer
Sanchez et al,45 2017	93	26	Waitlist/TAU	Social Anxiety Scale for Children Revised	Some concerns	Adventures aboard the S.S. GRIN	9	N	Y	Computer
Scholten et al,46 2016	138	7	Digital placebo	SCAS	Low	Dojo	3	Y	Y	Computer
Schoneveld et al,47 2016	136	9	Digital placebo	SCAS	Low	Mindlight	2.5	Y	N	Computer
Schoneveld et al,48 2018	174	17	CBT	SCAS	Low	Mindlight	6	Y	N	Computer
Schuurmans et al,49 2018	41	10	Waitlist/TAU	SCAS	Low	Dojo	4	Y	Y	Computer
Tsui et al,50 2021	117	11	CBT	SCAS	Some concerns	Mindlight	3	Y	N	Computer
Wijnhoven et al,51 2020	109	37	Digital placebo	SCAS	Some concerns	Mindlight	6	Y	N	Computer

Abbreviations: ADHD-RS, attention-deficit hyperactivity disorder rating scale; CBCL, Child Behavior Checklist; CBT, cognitive behavioral therapy; CDI, Children’s Depression Inventory; CDRS-R, Children’s Depression Rating Scale–Revised; CESD-R, Center for Epidemiologic Studies Depression Scale–Revised; CRPS-R, Connor’s Parent Rating Scale–Revised; DBRS, disruptive behavior rating scale; DSM, Diagnostic and Statistical Manual; EATQ-R, Early Adolescent Temperament Questionnaire–Revised; EDAH-ADHD, assessment of ADHD scale; GAD, generalized anxiety disorder; K-ARS, Korean ADHD rating scale; MASC, multidimensional anxiety scale for children; MDI, Major Depression Inventory; RADS-2, Reynold’s Adolescent Depression Scale, 2nd Edition; SCAS, Spence Children’s Anxiety Scale; TAU, treatment as usual.

Table 3. Regression Analyses and Analog to Analysis of Variance Examining Moderators of Gamified Digital Interventions Treatment Effects.

Characteristic	β (95% CI)	No.	Q	df	P value	No.
ADHD
Mean participant age	−0.13 (−0.39 to 0.14)	11	NA	NA	NA	NA
Sex (% female)	−2.11 (−3.88 to −0.34)	11	NA	NA	NA	NA
Clinically significant/community recruited	NA	NA	NA	NA	NA	NA
Intervention characteristics
Treatment duration, wk	−0.02 (−0.09 to 0.05)	11	NA	NA	NA	NA
App name	NA	NA	7.35	7	.39	11
Time limit	NA	NA	3.24	1	.07	11
Aesthetically sophisticated	NA	NA	1.53	1	.22	11
Accessible to public	NA	NA	0.01	1	.99	11
Intervention delivery modality	NA	NA	4.45	1	.04	11
Trial design characteristics
Sample size	<−0.01 (<−0.01 to <0.01)	11	NA	NA	NA	NA
Participant attrition	2.43 (−3.65 to 8.81)	9	NA	NA	NA	NA
Publication year	−0.01 (−0.01 to 0.07)	11	NA	NA	NA	NA
Control condition	NA	NA	4.61	2	.10	11
Outcome measure	NA	NA	4.41	4	.35	11
Risk of bias	NA	NA	8.61	1	.003	11
Depression
Mean participant age	−0.03 (−0.12 to 0.06)	8	NA	NA	NA	NA
Sex (% female)	−0.01 (−0.02 to 0.01)	8	NA	NA	NA	NA
Clinically significant/community recruited	NA	9	0.72	1	.40	9
Intervention characteristics
Treatment duration, wk	−0.06 (−0.21 to 0.09)	9
App name	NA	NA	7.07	5	.22	9
Time limit	NA	NA	4.43	1	.04	9
Aesthetically sophisticated	NA	NA	1.49	1	.22	9
Accessible to public	NA	NA	<0.01	1	.96	9
Intervention delivery modality	NA	NA	3.28	1	.07	9
Trial design characteristics
Sample size	<−0.01 (−0.01 to 0.01)	9	NA	NA	NA	NA
Participant attrition	<−0.01 (−0.01 to 0.01)	9	NA	NA	NA	NA
Publication year	−0.04 (−0.09 to 0.02)	9	NA	NA	NA	NA
Control condition	NA	NA	5.42	3	.14	9
Outcome measure	NA	NA	7.08	6	.31	9
Risk of bias	NA	NA	0.00	2	.99	9
Anxiety
Mean participant age	<0.01 (−0.03 to 0.04)	12	NA	NA	NA	NA
Sex (% female)	<0.01 (−0.01 to 0.01)	12	NA	NA	NA	NA
Clinically significant/community recruited	NA	NA	0.50	1	.48	12
Intervention characteristics
Treatment duration, wk	<−0.01 (−0.06 to 0.05)	12	NA	NA	NA	NA
App name	NA	NA	0.81	6	.99	12
Time limit	NA	NA	1.37	1	.24	12
Aesthetically sophisticated	NA	NA	0.12	1	.73	12
Accessible to public	NA	NA	0.04	1	.84	12
Intervention delivery modality	NA	NA	NA	NA	NA	NA
Trial design characteristics
Sample size	<0.01 (<−0.01 to <0.01)	12	NA	NA	NA	NA
Participant attrition	<0.01 (<−0.01 to <0.01)	12	NA	NA	NA	NA
Publication year	−0.02 (−0.05 to 0.02)	12	NA	NA	NA	NA
Control condition	NA	NA	1.58	2	.45	12
Outcome measure	NA	NA	0.59	4	.97	12
Risk of bias	NA	NA	0.81	2	.67	12

Abbreviation: NA, not applicable.

ADHD Efficacy Outcomes

The random-effects meta-analysis found a modest therapeutic effect of identified gamified DMHIs on ADHD outcomes compared to control conditions (g, 0.28; 95% CI, 0.09 to 0.48; z, 2.83; P < .001) (Figure 2A). Visual inspection of the forest plot, Q statistic, and I² statistic identified the presence of some heterogeneity (Q₁₀ = 14.28; P = .16; I², 29.95%). Visual inspection of the funnel plot and Egger test for bias (t₉ = 2.06, P = .07) did not suggest publication bias to be significant. As a precautionary step, the Duval and Tweedie trim-and-fill method was applied and filled 3 trials to the left of the mean. This produced a smaller adjusted effect size (g, 0.17).

Figure 2. Overall Summary Effects and by Condition.

ADHD indicates attention-deficit/hyperactivity disorder.

ADHD Moderator Outcomes

Participant Characteristics

There was a significant association between biological sex and treatment effects, such that RCTs that had a greater percentage of boys exhibited larger effects. However, no significant associations between average participant age and treatment effects were found (Table 3).

Intervention Characteristics

DMHIs delivered on a computer (k, 6; g, 0.41) produced larger effects compared tablet-delivered DMHIs (k, 5; g, 0.09). Additionally, DMHIs delivered with imposed time limits (k, 6; g, 0.49) tended to exhibit greater effects compared to those without time limits (k, 5; g, 0.12, P = .07) (Table 3). However, no significant association was found between treatment duration and therapeutic effects (Table 3). Additionally, there was no significant difference in treatment effects across gamified DMHIs (Table 3; eResults in Supplement 1). Furthermore, no difference was found between DMHIs that were not aesthetically sophisticated (k, 7; g, 0.34) compared to those that were (k, 4; g, 0.13, Table 3), nor between DMHIs that were accessible (k, 2; g, 0.24) vs inaccessible to the public (k, 9; g, 0.25).

Trial Design Characteristics

Risk of bias was found to influence treatment effects such that RCTs with a low risk of bias (k, 8; g, 0.13) had smaller effects compared to those with some concerns about risk of bias (k, 3; g, 0.74). However, there were no significant associations between treatment effects and either sample size, participant attrition, publication year, comparison conditions, or outcome measures (Table 3; eResults in Supplement 1).

Depression Efficacy Outcomes

The random-effects meta-analysis found a modest therapeutic effect of gamified DMHIs on depression outcomes compared to control conditions (g, 0.28; 95% CI, 0.08 to 0.47; z, 2.80; P < .005) (Figure 2B). Visual inspection of the forest plot, Q statistic, and I² statistic identified significant heterogeneity in effects across RCTs (Q₈ = 17.28; P = .03, I² = 53.69%). Visual inspection of the funnel plot and Egger test for bias (t₇ = 1.11; P = .30) did not suggest publication bias was present.

Depression Moderator Outcomes

Participant Characteristics

There were no significant associations between participant characteristics (eg, age and biological sex) and treatment effects on depression outcomes (Table 3). Although descriptively larger, there was no difference between participants with clinically significant symptoms (k, 7; g, 0.23) and those without (k, 2; g, 0.76).

Intervention Characteristics

Gamified DMHIs that had imposed time limits exhibited larger effects (k, 4; g, 0.63) compared to DMHIs that did not (k, 5; g, 0.12) (Table 3). However, there was no significant association between treatment duration and therapeutic effects (Table 3). There was also no significant difference in treatment effects across gamified DMHIs (Table 3; eResults in Supplement 1). Additionally, there were no significant differences between DMHIs that were not aesthetically sophisticated (k, 2; g, 0.47) compared to DMHIs that were (k, 7; g, 0.23), between DMHIs delivered on a computer (k, 7; g, 0.37) compared to a videogame console (k, 2; g, 0.05), or between those that were accessible (k, 8; g, 0.29) vs inaccessible (k, 1; g, 0.30) to the public (Table 3).

Trial Design Characteristics

There were no significant associations between treatment effects and sample size, participant attrition, publication year, comparison conditions, or outcome measures (Table 3; eResults in Supplement 1). Finally, risk of bias did not significantly influence treatment effects, with little difference found between RCTs with high concerns (k, 2; g, 0.26), some concerns (k, 2 g, 0.24), and low concerns (k, 5; g, 0.26).

Anxiety Efficacy Outcomes

The random-effects meta-analysis found a small non-significant effect of gamified DMHIs on anxiety outcomes compared to control conditions (g, 0.07; 95% CI, −0.02 to 0.17; z, 1.48; P < .14) (Figure 2C). Visual inspection of the forest plot, Q statistic, and I² statistic identified little heterogeneity (Q₁₁ = 6.25; P = .86; I² = 0%). Visual inspection of the funnel plot and Egger test for bias (t₁₀ = 0.35; P = .73) did not suggest publication bias was present.

Anxiety Moderator Outcomes

Participant Characteristics

There were no significant associations between participant characteristics (eg, age and biological sex) and treatment effects on anxiety outcomes (Table 3). Similarly, there were no differences between participants with clinically significant symptoms (k, 9; g, 0.10) and those without (k, 3; g, 0.03) (Table 3).

Intervention Characteristics

There was no significant association between treatment duration and therapeutic effects (Table 3). Additionally, there was no significant difference between specific DMHIs for anxiety (Table 3; eResults in Supplement 1). Furthermore, there was no significant difference between DMHIs that imposed a time limit (k, 9; g, 0.04) compared to DMHIs that did not (k, 3; g, 0.19). Finally, there were no differences between DMHIs that were aesthetically sophisticated (k, 7; g, 0.09) or accessible to the public (k, 4; g, 0.08), compared to those DMHIS that were not aesthetically sophisticated (k, 5; g, 0.05) or inaccessible to the public (k, 8; g, 0.06).

Trial Design Characteristics

There were no significant associations between treatment effects and either sample size, attrition, publication year, comparison conditions, or outcome measures (Table 3; eResults in Supplement 1). Finally, risk of bias did not significantly influence treatment effects, with little descriptive difference between RCTs with high concerns (k, 2; g, 0.06), some concerns (k, 3; g, 0.01), and low concerns (k, 7; g, 0.11).

Discussion

This systematic review and meta-analysis examined the therapeutic effects of gamified DMHIs for pediatric anxiety, depression, and ADHD and explored moderators of treatment effects. A modest therapeutic effect of gamified DMHIs was identified for both ADHD and depression outcomes (g, 0.28) but not for anxiety outcomes (g, 0.07). Several DMHI moderators were identified for ADHD and depression outcomes.

First, sex differences in treatment effects were observed for ADHD, with RCTs that had more male participants having larger effects. On one hand, this is consistent with literature identifying sex differences in the phenomenology, clinical presentation, and neurobiology of female individuals with ADHD.^52,53 On the other hand, this finding may be less influenced by underlying neurobiological sex differences but rather the specific gamification approaches used in DMHIs to promote engagement. For instance, gamified DMHIs for ADHD used action and driving games that may appeal more to some patients who have an interest in engaging with those types of games. Further research is needed to elucidate the cause of the discrepancy in treatment effects to optimize therapeutic outcomes for female individuals with ADHD.

Second, computer-based delivery—compared to other modalities such as tablets or videogame consoles—produced larger treatment effects for ADHD. Given the proliferation of mobile Health applications for mental health, it is important to consider possible reasons for this finding. For instance, the context in which gamified DMHIs are used may influence outcomes. In these RCTs, computers were stationary and often placed in environments with limited distractions (eg, a desktop or laptop in a quiet room with a research assistant nearby). Meanwhile, tablets can be used in a variety of settings (eg, home, in the backseat of a car, and waiting rooms of physicians’ offices) that have a higher likelihood of distraction. Similarly, the other intervention characteristics, such as imposing time limits on the use of gamified DMHIs, were associated with larger treatment effects for depression—which may also lead to its delivery in contexts that used a structured approach and increased focus on DMHI content. Thus, it may be important to balance the accessibility of gamified DMHIs alongside contextual factors (eg, setting and time limits) to optimize clinical outcomes.

Current gamified DMHIs did not produce significant benefit for anxiety severity. Given that DMHIs show modest benefit for adults with anxiety,¹⁸ it is worth considering whether gamified DMHIs use the optimal approach to delivering therapeutic components or whether therapists and other expert health care professionals are needed to deliver these therapeutic components (eg, exposure therapy). Intervention (eg, DMHI game design) or trial design characteristics may have affected findings (eg, using the Spence Children’s Anxiety Scale as the outcome measure and robust comparison conditions). As many DMHIs were not publicly accessible, further investigations will necessitate collaborative academic-industry partnerships. which can leverage collaborative expertise (eg, game design, computer programming, and evidence-based treatments) to develop, design, and test empirically informed DMHIs for youth.

Limitations

This report is not without limitations. First, most RCTs used parent- or child-reported outcome measures. While psychometrically valid, a multimodal assessment that incorporates independent evaluator ratings would reduce risk of bias. Additionally, there were limited characteristics available for inclusion across RCTs. Thus, there may be unexamined factors that could also influence treatment effects (eg, participants’ engagement, digital literacy, skill learning assessment, and social determinants).⁵⁴

Conclusions

Our work identified several areas for future research and highlights the need for establishing reporting guidelines for RCTs on DMHIs in youth.⁵⁴ First, adverse effects from gamified DMHIs were largely unreported across RCTs, which is a concern due to emerging research suggesting some negative effects from screen time for youth.^55,56 Furthermore, minimal information was available about data safety and privacy policies for gamified DMHIs (eg, who has access to users’ data and for what purposes). It will be important for future RCTs to report adverse effects and consider data safety (or data privacy policies) before incorporating gamified DMHIs into clinical practice.

Second, there was inconsistent reporting of concurrent pharmacotherapy and psychotherapy across RCTs. As gamified DMHIs can play an important role in the continuum of pediatric mental health care, this is important information. Future RCTs should report concurrent pharmacological and psychotherapeutic interventions received by participants during study participation, which can provide insight into the effects of DMHIs as stand-alone or adjunctive therapies. Third, although most RCTs reported the intervention duration, participants’ level of engagement with DMHIs was absent (eg, average or total time interacting with DMHI over the intervention period). This can provide essential information on the dose of DMHIs, which can provide insights into whether the total use (or screen time) relates to treatment outcomes or adverse effects. Furthermore, when reporting information on DMHI dose (eg, engagement and duration), it will be important to consider the context and settings in which DMHIs are used as well—which can moderate DMHI treatment effects. Finally, it will be important to monitor the efficacy, safety profile, and dose of DMHIs over the short term and long term. This should provide greater insight into the lasting effects of DMHIs, which is important for adoption in clinical practice.

In summary, our findings demonstrated that gamified DHMIs are modestly efficacious for children and adolescents with ADHD or depression. While moderators of treatment effects provide initial insights to optimize clinical outcomes, several questions remain that warrant future investigation. Addressing these questions is particularly important, as most commercially available DMHIs remain untested in RCTs.⁵⁷

Supplement 1.

eMethods

eResults

eFigure 1. Funnel Plot for Publication Bias For ADHD

eFigure 2. Funnel Plot for Publication Bias For Depression

eFigure 3. Funnel Plot for Publication Bias For Anxiety

Supplement 2.

Data sharing statement