The association between gaming disorder and impulsivity: A systematic review and meta-analysis

Abstract

Background:

Impulsivity, the tendency to act quickly without careful consideration, is a known risk factor and correlate of substance use and addictive disorders, including International Classification of Diseases (ICD)-11 gaming disorder (GD). The aim of this meta-analytic review was to critically evaluate associations between GD symptoms and trait impulsivity and its subtypes.

Methods:

Following the Preferred Reporting Items for Systematic Reviews and Meta-Analyses (PRISMA) guidelines, 33 studies (N = 24,818) met inclusion criteria, including being published between 2019 and 2024 (i.e. to focus on studies in ICD-11 era) and reporting data on problem gaming and trait impulsivity using psychometrically validated tools. Pooled effect sizes were calculated using reported correlations or means and standard deviations. Sources of heterogeneity, such as sample type, age, gender, region, assessment tool and impulsivity subtype, were examined using subgroup and moderator analyses.

Results:

The pooled association between GD and impulsivity was r = 0.29 (95% confidence interval [CI] = [0.24, 0.34]). Significant between-study heterogeneity was detected based on study region, with larger correlations in Asian studies compared to European and Western studies. Larger correlations were reported in studies employing the YIAT and IGDS9-SF and in studies employing the Barratt Impulsiveness Scale (BIS)-11 compared to other impulsivity tools.

Conclusion:

This meta-analytic review identified a consistent moderate association between trait impulsivity and GD. The result suggests that clinical evaluation and monitoring of GD should consider the influence of impulsivity on risk and recovery. Moreover, assessing specific subtype patterns of impulsivity may inform the implementation of tailored treatment. Future research should examine the relative influence of impulsivity subtypes in the initiation, maintenance and relapse of problematic gaming behaviour.

Introduction

Billions of people worldwide play video games (Statista, 2025). Many studies have highlighted the mass appeal and benefits of gaming, such as improvements in cognitive functioning, enhanced well-being and socializing opportunities (Adachi and Willoughby, 2017; Sampalo et al., 2023). However, for some individuals, gaming can be excessive and interfere with other life areas, resulting in significant negative psychological, physical, social and other consequences (Billieux et al., 2021). In its most severe form, problem gaming is recognized as an addictive disorder that shares many similarities to gambling disorder. In 2019, gaming disorder (GD) was officially included in the International Classification of Diseases (ICD-11; World Health Organization, 2024). GD is characterized by an inability to regulate gaming behaviour, prioritization of gaming to the exclusion of other activities and continued use despite harm (Brand et al., 2019). To advance understanding of this condition, the aim of the present meta-analytic review was to critically evaluate and synthesize the association between GD and impulsivity, given that impulsivity is a trans-diagnostic factor involved in a wide range of mental health problems and psychiatric conditions, including substance use and addictive disorders (Berg et al., 2015; Moeller et al., 2001; Vassileva and Conrod, 2019).

Current frameworks (e.g. the I-PACE model; Brand et al., 2016) conceptualize the development of GD as the product of biopsychosocial risk factors, including but not limited to difficult early life experiences, being male, escape motivations, comorbidities including depression, anxiety and Attention Deficit Hyperactivity Disorder (ADHD) and personality factors such as neuroticism (Ji et al., 2022; Király et al., 2023; Ostinelli et al., 2021; Stevens et al., 2021). An important and robust risk factor for addiction is trait impulsivity, which is central to explanations of psychological vulnerability to addictions (Lee et al., 2019; Vassileva and Conrod, 2019). Impulsivity as a personality trait is generally defined as the tendency to act quickly and without careful consideration, or with disregard, to potential negative consequences (Moeller et al., 2001). However, there is debate on the dimensionality and mechanisms in relation to impulsivity; while there is general consensus that impulsivity is a multi-faceted construct, there are different views on the constituent subtypes (Huang et al., 2024; Sharma et al., 2014; Strickland and Johnson, 2021).

The study of trait impulsivity has two widely used measurement tools: The Barratt Impulsiveness Scale (BIS; Patton et al., 1995) and the UPPS-P Impulsive Behaviour Scale (Cyders et al., 2007; Whiteside and Lynam, 2001). The BIS evaluates three dimensions of impulsivity: attentional, motor and non-planning and is generally considered a valid measure of impulsivity both as a stable, one-dimensional trait and according to each of its subscales (Meule et al., 2015). In comparison, the UPPS-P includes five factors: sensation seeking, positive urgency, negative urgency, lack of premeditation and lack of perseverance (Cyders et al., 2007; Whiteside and Lynam, 2001). Research suggests that the UPPS-P measures five distinct impulsivity constructs which cannot be combined into a single latent or one-dimensional model of impulsivity, although some studies still report aggregate scores (Billieux et al., 2012; Fournier et al., 2025; Whiteside and Lynam, 2001; Zsila et al., 2020). The literature therefore has been fragmented due to different approaches to conceptualizing impulsivity and its subtypes, which necessitates an approach to evaluating evidence that accounts for these differences.

Studies assessing the links between GD and impulsivity have assessed impulsivity as a relatively stable personality trait using self-report measures, but some studies have also used behavioural tasks to explore momentary state impulsivity. Behavioural tasks are designed to capture distinct components, such as response inhibition and delay of gratification, which are related to but are distinct from impulsivity (Cyders and Coskunpinar, 2011). Research indicates trait and behavioural measures of impulsivity often correlate poorly (Eben et al., 2023), and while trait measures strongly link to externalizing behaviours (e.g. drug, alcohol and cigarette use), state behavioural measures show minimal association, suggesting that they may tap into different underlying constructs (Creswell et al., 2019). As state and trait impulsivity are grounded in different theoretical frameworks, and methodological differences make them difficult to synthesize, treating them separately may help clarify conceptual distinctions.

Impulsivity is a well-established risk factor for GD, with high trait impulsivity shown to predict problematic gaming behaviour (Gentile et al., 2011; Kowalik et al., 2025; Raybould et al., 2022). However, there is to date less research exploring the relationship between specific impulsivity subtypes and problem gaming (see Müller et al., 2023, for a narrative review). Existing studies have suggested variance in trait impulsivity subtypes may account for differences in behavioural addiction symptom severity (López-Guerrero et al., 2023; Maxwell et al., 2020; Raybould et al., 2022) and treatment outcomes (Grall-Bronnec et al., 2012; Vinci et al., 2016). For example, higher levels of positive and negative urgency are associated with greater gambling addiction severity (López-Guerrero et al., 2023). In GD research, problematic gaming behaviours have been linked to lack of perseverance and attention subtypes (Cudo et al., 2020), positive and negative urgency subtypes (Raybould et al., 2022) or all five UPPS-P subfactors except positive urgency (Costes and Bonnaire, 2022). These mixed findings highlight the need for an integrated approach that synthesizes different lines of research to clarify which impulsivity subtypes tend to contribute to problematic gaming.

Several narrative and systematic reviews have assessed the role of impulsivity in contributing to the development and persistence of GD. A systematic review by Şalvarlı and Griffiths (2022) reported consistent associations between GD and overall impulsivity, potentially explained by altered neurobiological functioning in highly impulsive individuals but did not assess individual subfactor associations. Another review by Andrade et al. (2024) reported that GD was associated with overall impulsivity and high-risk subfactor trends including sensation seeking. This review also noted that four out of five studies evaluating gender differences reported greater impulsivity among males than females. However, no meta-analytic systematic review has quantified the magnitude of the relationship between GD and trait impulsivity, the effect of impulsivity subtypes or the magnitude by which the relationships vary in different subpopulations.

The present review

Previous research has identified positive associations across different measures of impulsivity and GD, but there is a need to synthesize evidence on the magnitude of the relationship. Moreover, it is not yet known how different subtypes of impulsivity relate to GD; how the relationship between GD and impulsivity may vary in different subpopulations. Therefore, the purpose of this review was to critically evaluate the literature on trait impulsivity and GD to: (1) determine the magnitude and direction of the relationship between impulsivity and GD through meta-analytic techniques; (2) investigate sources of potential heterogeneity in estimates according to various subgroups, including impulsivity subtype, assessment tool, study region and sample type and (3) assess potential underlying effects of age, gender and sample size using moderator analyses.

Methods

This systematic review was pre-registered on PROSPERO (registration number CRD42024591093). A preliminary search on MEDLINE and PsycINFO was performed to explore the literature and determine key terms and medical subject headings (MeSH) within the existing literature base. The search strategy was independently evaluated and verified by an academic librarian. Given impulsivity is a key symptom of ADHD, this condition was included in the search terms. Studies published in the last 5 years (2019–2024) were included to identify GD research in the ICD-11 era.

In this review, inclusion criteria were as follows: (1) peer-reviewed and published in English between 2019 and 2024, (2) investigated problem gaming symptoms and trait impulsivity using validated tools and (3) reported quantitative data on problem gaming and impulsivity. Regarding criterion (2), studies that utilized Internet addiction tools were included if they were used to assess gaming-related issues. Studies assessing trait impulsivity using a non-scalable, stand-alone tool (e.g. The BIS-11 [Patton et al., 1995]) were also included. Studies were excluded if they focused on psychometric validation, employed a sample with less than 50% gamers or focused on neural imaging (e.g. functional magnetic resonance imaging [fMRI], event-related potential [ERP], electroencephalography [EEG]).

Five academic databases (Ovid PsycINFO, Ovid MEDLINE, Ovid Emcare, CINAHL and Scopus) were searched using the following keywords and protocol: (ADHD OR [attention deficit hyperactivity disorder]) OR (inattent* OR hyperactiv* OR impulsiv* OR hyperkinesis) AND ([gaming disorder*] OR IGD OR [gaming addict*]). The first 10 pages of Google Scholar search results were reviewed (Paez, 2017). The final search was conducted on 28 November 2024.

Figure 1 displays a PRISMA (Preferred Reporting Items for Systematic Reviews and Meta-Analyses) flow diagram of the identification, screening and inclusion process. Initial database searches yielded 1,650 results (including duplicates). Titles and abstracts were independently screened for relevance by two reviewers (J.N. and L.N.), which led to the initial identification of 280 studies. Full-text review removed 247 studies. Pearling was undertaken to review the reference lists of the final selection of studies, but no additional studies were identified. All studies were checked for adequate effect size data for synthesis. The corresponding authors of studies not reporting adequate information (n = 13) were contacted. Six authors provided data, and the remaining authors (n = 7) could not be reached. Overall, there were 33 studies eligible for inclusion, encompassing 24,818 participants.

Figure 1.

PRISMA flow chart of study selection.

Data extraction

Effect sizes for each study were extracted, which either involved (where available) Pearson’s R correlation coefficients assessing the association between measures of trait impulsivity and GD. Where correlation coefficients were not available, means and standard deviations of impulsivity and impulsivity subscales were extracted for GD vs non-GD groups (where available). Where studies used longitudinal designs (i.e. intervention studies), only baseline scores were extracted. For the purposes of subgroup and moderator analyses, various demographic and methodological details were also extracted. These included study attributes, i.e. author(s), year of publication, country/region, study design, sample attributes (i.e. sample size, mean age, age range, female/male percentage, clinical or non-clinical population) and assessment attributes (i.e. impulsivity tool, GD tool, ADHD tool, diagnostic status i.e. clinical/imputed). Data were extracted and independently checked for agreement by two authors (J.N. and L.N.).

Risk of bias

The risk of bias of included studies was assessed using the Joanna Briggs Institute (JBI) Critical Appraisal Checklist for Analytical Cross-Sectional Studies (Moola et al., 2020). Each study was assessed for potential risk of bias on methodological domains covering inclusion criteria, participant description, measurement validity, identification and control of confounding, and appropriateness of statistical analysis. Two independent reviewers (J.N. and L.N.) individually rated each study, with disagreements settled by discussion between the two raters. JBI does not prescribe fixed thresholds for bias assessments but advises researchers to establish and justify their own criteria when using its appraisal tools. In this review, ‘yes’ responses were assigned a score of 1, while ‘no’ or ‘unclear’ responses were scored as 0. Based on the percentage of endorsed criteria, studies were classified as follows: low risk of bias if they achieved 80–100% of the appraisal criteria; moderate risk of bias if between 50% and 79% of the criteria were met and high risk of bias if less than half of the criteria were met. These percentage scores also served as indicators of overall study quality, with higher percentages reflecting superior methodological rigour and reporting standards. A table summarizing the risk of bias categories and reporting quality ratings of included studies is available in the Supplementary Materials.

Data analysis

The primary aim of the present meta-analytic review was to quantify the association between GD and impulsivity. Pooled effect sizes (Fisher’s Z) were calculated for each study using reported correlations or means and standard deviations. Where only impulsivity subscale scores were reported, an overall pooled effect size was calculated using the weighted average of scores (i.e. Wölfling et al., 2020; Yan et al., 2021) or the weighted average of correlations (i.e. Coelho et al., 2023; Fumero et al., 2020; Marrero et al., 2021; Tunney and Raybould, 2023). In studies where two groups were compared (e.g. GD vs non-GD), correlations were weighted and an overall pooled effect size calculated (i.e. Shin et al., 2019) or the mean difference between groups was extracted and used to calculate pooled effect sizes (i.e. Demetrovics et al., 2022; Marchica et al., 2020; Wang et al., 2022; Xiang et al., 2021). In studies where more than two groups were compared, groups were divided using scale cut-off scores to create GD and non-GD groups (i.e. Suárez-Soto et al., 2025). Outlier analysis detected no significant outliers.

Pooled correlations and 95% confidence intervals (CIs) were calculated using a random-effects model. A random-effects model was selected due to handling data from diverse populations (Borenstein et al., 2011). All analyses were conducted using R v.4.4.2, using the ‘dmetar’ package (Harrer et al., 2019). The degree of between-study heterogeneity was assessed using Cochrane’s Q, which measures the ratio of observed variation to within-study error (Borenstein et al., 2017) and the I² statistic, which indicates the proportion of total variation due to methodological differences rather than random sampling error. I² reflects the degree of inconsistency among study results, with of 25%, 50% and 75% indicating low, medium and high heterogeneity, respectively (Borenstein et al., 2017).

Sources of heterogeneity were explored using subgroup and moderator analyses. Studies were sorted into the following categories: sample type (i.e. in clinical treatment/non-clinical); age (i.e. mean age of participants); gender (i.e. proportion of male, female and other participants); region (i.e. geographical region of data collection); assessment (i.e. assessment tool); impulsivity subtype and study reporting quality (i.e. percentage of JBI criteria endorsed; note: the categorical risk of bias classifications were used to guide sensitivity analyses, while the continuous study reporting quality variable was employed as a moderator). BIS tools were categorized into either BIS-11 or brief versions (BIS-B) where sufficient numbers were available for subgroups. Short UPPS Impulsive Behaviour Scale (SUPPS) and UPPS scales were combined into a single subgroup (UPPS) (Pechorro et al., 2021). GD tools were categorized into IGDS9-SF, YIAT, Diagnostic and Statistical Manual of Mental Disorders (5th ed.; DSM-5)-based measures (IGD) and less commonly utilized DSM-5-based measures (Other). One study included data for a DSM-5-based tool and an ICD-11-based tool (Raybould and Tunney, 2024), so only the DSM-5 tool was retained for consistency across studies.

Gender rates were manually calculated if not reported and sufficient data were available. Pooled effect sizes were computed within each subgroup. ADHD symptom severity was initially considered as a potential moderator of the relationship between impulsivity and GD, but only two studies meeting inclusion criteria also reported ADHD outcome measures. This was deemed insufficient for robust statistical analysis and was therefore excluded from consideration in the current review (Higgins and Green, 2011).

Publication bias was assessed by visual inspection of funnel plots (which maps Fisher’s Z-transform for each study with its standard error), followed by evaluation of funnel plot asymmetry using Egger’s linear regression test (Egger et al., 1997). In both cases, a significant model intercept (β₀; p < .05) was used to indicate the presence of funnel plot asymmetry (see Supplementary Figure 1). Finally, a sensitivity analysis was conducted to assess the impact of influential studies on the overall results. This two-step approach involved first assessing the impact of removing individual studies using the leave-one-out method (Viechtbauer and Cheung, 2010) and second the impact of removing studies with a higher risk of bias (e.g. studies rated as Low or Moderate based on JBI ratings).

Results

Study characteristics

Table 1 presents a summary of study characteristics. The overall mean age of participants was 22.6 years (SD = 5.7). Most studies (k = 23) included samples with fewer than 50% females. Most studies were conducted in Asia (k = 10) and Europe (k = 9), and samples were primarily non-clinical (k = 30). The most widely used GD assessment tool was the IGDS9-SF (k = 7), and the most common trait impulsivity tool was the BIS-11 (k = 14). Almost half of the studies reported impulsivity subtype scores (k = 16).

Table 1.

Characteristics of included studies (N = 33).

Study	N	Location	Sample type	Mean age	Age range	Female (%)	GD tool	Impulsivity tool	Correlation	Fisher’s Z
Achab et al. (2022)	313	France	Non-clinical	27	18-54	17.9	DAS	BIS-10	0.32	0.33
Cerniglia et al. (2019)	656	Italy	Non-clinical	16	16-19	48.5	IGDS9-SF	BIS-11	0.29	0.3
Chen et al. (2024)	130	Malaysia	Non-clinical	21	NR	46.2	IGDS9-SF	BIS-11	0.56	0.63
Chew and Wong (2022)	123	Singapore	Non-clinical	25	18-59	57.0	IGDS9-SF	BIS-11	0.39	0.41
Chung et al. (2020)	158	Korea	Clinical	25	NR	15.8	YIAT	BIS-11	0.45	0.5
Coelho et al. (2023)	1697	Canada, USA	Non-clinical	23	NR	NR	SSBA	SUPPS-P	0.13 a	0.13
Cudo et al. (2020)	673	Poland	Non-clinical	21	17-38	58.1	IGDS9-SF	BIS-11	0.16	0.16
Delic et al. (2024)	560	Multiple	Non-clinical	28	18+	15.9	Petry IGD	UPPS-P	0.37	0.39
Demetrovics et al. (2022)	3600	Hungary	Non-clinical	40	18-64	51.0	IGDT-10	BIS-21	0.37 a	0.39
Efrati et al. (2021)	471	Israel	Non-clinical	16	14-18	60.0	IGDS9-SF	BIS-11	0.28	0.29
Entwistle et al. (2020)	1958	Multiple	Non-clinical	32	14+	12.0	IGDS	SUPPS	0.26	0.26
Fumero et al. (2020)	492	Spain	Non-clinical	14	11-18	48.5	PVP	BIS-15	0.15 a	0.15
Hammad and Al-Shahrani (2024)	350	Saudi Arabia	Non-clinical	21	18+	46.9	IGDS9-SF	BIS-15	0.68	0.83
Hing et al. (2023)	646	Australia	Non-clinical	15	12-17	42.1	Petry IGD	BBIS	0.29	0.29
King et al. (2020)	428	Multiple	Non-clinical	24	18-60	6.5	Petry IGD	BBIS	0.2	0.2
Li et al. (2020)	1127	China	Non-clinical	20	17-25	59.2	CGDS	BBIS	0.24	0.25
Li et al. (2021)	1288	China	Non-clinical	20	17-25	57.6	CGDS	BBIS	0.19	0.19
Maganuco et al. (2019)	364	Multiple	Non-clinical	25	18-48	33.3	YIAT	BIS-11	0.47	0.51
Marchica et al. (2020)	475	NR	Non-clinical	21	18-27	54.2	IGDS9-SF	BIS-5	0.26 a	0.27
Marrero et al. (2021)	550	Spain	Non-clinical	14	11-20	48.5	PVP	BIS-15	0.15 a	0.16
Martoncik et al. (2024)	1301	Multiple	Non-clinical	31	18+	23.0	IGDS9-SF	BIS-15	0.31	0.32
Raybould and Tunney (2024)	372	UK	Non-clinical	26	NR	49.5	IGD	BIS-11, UPPS-P	0.2	0.2
Shin et al. (2019)	150	Korea	Clinical	25	NR	0.0	YIAT	BIS-11	0.25 a	0.25
Su et al. (2019)	596	Hungary	Non-clinical	21	14-38	50.0	POGQ	BIS-21	0.33	0.34
Suárez-Soto et al. (2025)	1410	Spain	Non-clinical	21	18-34	33.6	GAS-7	UPPS-P	0.19	0.2
Tang et al. (2024)	1134	Multiple	Non-clinical	NR	18-25	46.2	IGDS9-SF	UPPS-P	0.17	0.17
Tunney and Raybould (2023)	500	UK	Non-clinical	30	18+	48.8	IGD	BIS-11	0.18 a	0.18
Wang et al. (2022)	109	China	Non-clinical	20	18-25	32.1	IGDQ, CIAS	BIS-11	0.26 a	0.27
Warburton et al. (2022)	866	Australia	Non-clinical	14	12-17	43.0	IGDT-10	BIS-15	0.21	0.21
Wölfling et al. (2020)	57	Germany	Clinical	30	NR	0.0	AICA-S	BIS-11	0.28 a	0.28
Xiang et al. (2021)	1525	China	Non-clinical	NR	12-26	42.4	YIAT	BIS-11	0.35 a	0.36
Yan et al. (2021)	115	China	Non-clinical	20	18+	39.7	YIAT	BIS-11	0.43 a	0.46
Zhu et al. (2023)	624	China	Non-clinical	20	18-24	58.8	IGD-20	BIS-11	0.26	0.27

Note. AICA: Scale for Assessment of Internet and Computer Gaming; CGDS: Chinese Gaming Disorder Scale; CIAS: Chen Internet Addiction Scale; DAS: Dependence Adapted Scale; GAS-7: Gaming Addiction Scale (7-item); IGD: DSM-5 Internet Gaming Disorder criteria; IGDT-10: Internet Gaming Disorder Test (10-item); IGD-20: Internet Gaming Disorder Scale (20-item); IGDS: Internet Gaming Disorder Scale; IGDQ: Internet Gaming Disorder Questionnaire; IGDS9-SF: Internet Gaming Disorder Scale–Short-Form (9-item); Petry IGD: Internet Gaming Disorder Scale (9-item); POGQ: Problematic Online Gaming Questionnaire; PVP: Problematic Videogame Playing Scale; SSBA: Brief Screener for Substance and Behavioral Addiction; YIAT: Young’s Internet Addiction Test; BBIS: Barratt Impulsiveness Scale–Brief; BIS-10: Barratt Impulsiveness Scale (34-item); BIS-11: Barratt Impulsiveness Scale (30-item); BIS-15: Barratt Impulsiveness Scale (15-item); BIS-21: Barratt Impulsiveness Scale (21-item); SUPPS: Short UPPS Impulsive Behavior Scale; SUPPS-P: Short UPPS-P Impulsive Behavior Scale (with Positive Urgency); UPPS-P: Impulsive Behavior Scale (with Positive Urgency); NR: data not reported.

^aCorrelation calculated from reported data.

Risk of bias

Risk of bias was low in 30 studies, with scores >80% on the JBI scale (see Supplementary Table 1). Three studies were of moderate risk of bias (scores 50–79%). No studies were classed as high risk of bias (i.e. <49%).

Data synthesis

Overall, the pooled association between GD and impulsivity across the 33 studies (N = 24,818) was r = 0.29, 95% CI = [0.24, 0.34], t(32) = 11.79, p < .001. However, significant heterogeneity between estimates was detected, Q(32) = 331.24, p < .001, τ² = 0.02 [0.01; 0.04], I² = 90.3%, with a prediction interval around r ranging from [0.02, 0.52]. Figure 2 presents a forest plot of pooled effect sizes across all 33 studies.

Figure 2.

Forest plot of pooled effect sizes across studies.

Subgroup and moderator analyses

Region

Table 2 presents a summary of the subgroup analyses for all variables. Significant between-study heterogeneity was detected based on study region, Q(3) = 9.58, p < .001. Significant correlations were found among Asian (k = 10, r = .35, CI = [0.25, 0.44], p < .001), European (k = 9, r = .26, CI = [0.18, 0.33], p < .001) and Western studies (k = 5, r = .20, CI = [0.12, 0.28], p = .002), with each group showing substantial I² estimates of heterogeneity (ranging from 70% to 90%). Larger correlations were found in Asian studies, compared with European and Western regions. The pooled effect size for Middle-Eastern studies was not significant (p = .288), due to the small number of studies (k = 2).

Table 2.

Subgroup analyses of the association between impulsivity and GD by region, sample type, impulsivity tool and GD tool.

			Combined estimates			Heterogeneity			Model
	Subgroup	k	r	95% CI	p-within	Tau2	Q	I 2	df	PI	p-between
Region	Overall	26	–	–	–	–	9.58	–	3	[-0.01, 0.61]	0
	Europe	9	0.26	[0.18, 0.33]	0	0.008	79.38	89.90%
	Asia	10	0.35	[0.25, 0.44]	0	0.013	50.3	82.10%
	Western	5	0.2	[0.12, 0.28]	0.002	0.003	13.66	70.70%
	Middle-East	2	0.56	[-2.88, 4.00]	0.288	0.144	58.41	98.30%
Population	Overall	33	–	–	–	–	0.41	–	1	[0.02, 0.58]	0.122
	Non-clinical	30	0.3	[0.24, 0.35]	0	0.019	323.97	91.00%
	Clinical	3	0.35	[0.01, 0.70]	0.048	0.012	4.88	59.00%
Impulsivity tool	Overall	28	–	–	–	–	4.17	–	2	[0.01, 0.61]	0
	BIS-11	14	0.34	[0.26, 0.42]	0	0.014	66.69	80.50%
	UPPS	5	0.23	[0.10, 0.35]	0.007	0.008	35.41	88.70%
	BIS-B	9	0.29	[0.13, 0.45]	0.003	0.04	138.87	94.20%
GD tool	Overall	33	–	–	–	–	16.36	–	3	[0.02, 0.58]	0
	Other	11	0.22	[0.17, 0.27]	0	0.004	34.3	70.80%
	IGDS9-SF	9	0.37	[0.20, 0.54]	0.001	0.045	145.24	94.50%
	YIAT	5	0.41	[0.29, 0.54]	0.001	0.006	11.99	66.60%
	Other_IGD	8	0.27	[0.20, 0.34]	0	0.006	55.55	87.40%

Note. BIS-11: Barratt Impulsiveness Scale; BIS-B: Brief versions of the Barratt Impulsiveness Scale (i.e. BIS-15, BBIS); UPPS: UPPS Impulsive Behavior Scales (i.e. UPPS-P, SUPPS-P); IGDS9-SF: Internet Gaming Disorder Scale–Short-Form (9-item); YIAT: Young’s Internet Addiction Test; Other_IGD: DSM-5-based measures (i.e. IGD, IGDT-10, IGDS); Other: less commonly utilized gaming disorder measures (i.e. AICA-S, CGDS, DAS, GAS-7, POGQ, PVP, SSBA); NA: Region data not provided.

Sample type

Sample type (clinical vs non-clinical) was not significantly associated with the overall effect, Q(1) = 0.41, p = .122. However, both clinical (k = 3, r = .35, CI = [0.01, 0.70], p = .048) and non-clinical (k = 30, r = .30, CI = [0.24, 0.35], p < .001) sample types yielded significant pooled effect sizes.

GD assessment

Significant between-study heterogeneity was detected among studies pooled by GD assessment. Significant correlations were found among studies using the IGDS9-SF (k = 9, r = .37, CI = [0.20, 0.54], p = .001), YIAT (k = 5, r = .41, CI = [0.29, 0.54], p = .001), IGD-variants (k = 8, r = .27, CI = [0.20, 0.34], p < .001) and other assessment tools (k = 11, r = .22, CI = [0.17, 0.27], p < .001), with each group showing substantial I² estimates of heterogeneity (ranging from 70% to 95%). Larger correlations were found among studies employing the YIAT, and IGDS9-SF, compared with other tools.

Impulsivity assessment

Significant between-study heterogeneity was detected among studies pooled by choice of impulsivity assessment. Significant correlations were found among studies using the BIS-11 (k = 14, r = .34, CI = [0.26, 0.42], p < .001), UPPS (k = 5, r = .23, CI = [0.10, 0.35], p = .007) and BIS-B assessment tools (k = 9, r = .29, CI = [0.13, 0.45], p = .003), with each group showing substantial I² estimates of heterogeneity (ranging from 80% to 95%). Larger correlations were found among studies employing the BIS-11, compared with other tools.

Subtypes

Individual analyses were performed for each impulsivity subtype to assess its association with GD. Table 3 presents a summary of separate meta-analyses for each subtype, which showed significant correlations between GD and motor impulsivity (k = 9, r = 0.25, CI = [0.16, 0.33], p = 0.001), attentional impulsivity (k = 9, r = .0.31, CI = [0.16, 0.44], p = .001), non-planning (k = 9, r = .0.23, CI = [0.08, 0.37], p = .008), negative urgency (k = 4, r = .33, CI = [0.18, 0.46], p = .001) and positive urgency (k = 4, r = .29, CI = [0.18, 0.40], p = .004). A multivariate regression model was then used to assess whether these subtypes differed significantly from one another. However, when accounted for in the overall model, coefficients became non-significant, suggesting an insufficient number of studies for each subtype to reliably assess subtype difference, F(10, 39) = 0.74, p = 0.690.

Table 3.

Impulsivity subfactor meta-analyses.

		Combined estimates			Heterogeneity			Model
Impulsivity subtype	k	R	95% CI	p-within	Tau2	Q	I 2	df	PI
BIS subfactors
Motor	9	0.25	[0.16, 0.33]	0.001	0.011	38.23	79.10%	8	[–0.01, 0.47]
Attentional	9	0.31	[0.16, 0.44]	0.001	0.034	75.42	89.40%	8	[–0.13, 0.65]
Non-planning	9	0.23	[0.08, 0.37]	0.008	0.039	184.71	95.70%	8	[–0.24, 0.62]
UPPS subfactors
Negative urgency	4	0.33	[0.18, 0.46]	0.001	0.009	33.04	90.90%	3	[0.00, 0.59]
Positive urgency	4	0.29	[0.18, 0.40]	0.004	0.004	15.52	80.70%	3	[0.06, 0.49]
Lack of premeditation	4	0.07	[–0.17, 0.30]	0.427	0.021	45.99	93.50%	3	[–0.42, 0.52]
Lack of perseverance	4	0.04	[–0.05, 0.13]	0.264	0.002	10.4	71.20%	3	[–0.13, 0.21]
Sensation seeking	4	0.07	[–0.10, 0.24]	0.287	0.012	48.56	93.80%	3	[–0.30, 0.43]

Moderator analyses

Table 4 presents a summary of moderator analyses performed to investigate potential underlying effects of participant age (k = 31), gender (k = 32), sample size (k = 33) and reporting quality (k = 33). All univariate predictors were not significant, suggesting heterogeneity in estimates was not explained by these variables. The absence of a moderating effect of reporting quality was likely a statistical artefact arising from the restricted range of quality scores (i.e. all studies scored between 57% and 100%).

Table 4.

Univariate regression analyses of the association between impulsivity and GD according to age, gender and sample size.

	Model fit					Model parameters
	k	F	df	R 2	I 2	B	95% CI	p
Age	31	0.94	[1, 29]	0.19	92.4%	0	[–0.00, 0.01]	0.34
Gender	32	0.16	[1, 30]	0	92.7%	0	[–0.00, 0.00]	0.69
Sample size	33	0.97	[1, 31]	0	92.6%	0	[–0.00, 0.00]	0.33
Reporting quality	33	0.51	[1, 31]	0	93.1%	0	[–0.01, 0.00]	0.48

Publication bias

A funnel plot assessing publication bias can be found in the Supplementary Materials (Supplementary Figure S1). Visual inspection of each plot identified no discernible asymmetry. Egger’s regression, t(31) = 0.81; β₀ = 1.11, 95% CI = [1.01, 1.21]; p = .421, confirmed the absence of significant funnel plot asymmetry and publication bias in the analysis.

Sensitivity analysis

Eleven studies were identified as potential influential cases (Chen et al., 2024; Coelho et al., 2023; Cudo et al., 2020; Demetrovics et al., 2022; Fumero et al., 2020; Hammad and Al-Shahrani, 2024; Li et al., 2021; Maganuco et al., 2019; Marrero et al., 2021; Suárez-Soto et al., 2025; Tang et al., 2024). Removal of these studies yielded an overall pooled association between GD and impulsivity across the remaining 22 studies (N = 13,183) was r = 0.28, 95% CI = [0.25, 0.32], t(21) = 18.80, p < .001, with moderate heterogeneity (I² = 58.5%), with a prediction interval around r ranging from [0.18, 0.38]. Removal of moderate risk of bias studies (Chen et al., 2024; Marchica et al., 2020; Xiang et al., 2021) yielded an overall pooled association between GD and impulsivity across the remaining 30 studies (N = 22,741) of r = 0.28, 95% CI = [0.23, 0.33], t(29) = 11.16, p < .001. A high degree of between-study heterogeneity was detected (I² = 90.5%), with a prediction interval around r ranging from [0.01, 0.51].

Discussion

The aim of this meta-analytic review was to quantify the magnitude of association between gaming disorder (GD) symptoms and trait impulsivity. Although recent systematic reviews have assessed the overall relationship between the two constructs, this was the first meta-analytic review aimed at quantifying differences in estimates based on demographic and sampling characteristics, as well as the first to examine the relationship based on impulsivity subtypes. Overall, small-to-moderate positive correlations between GD symptoms and impulsivity were identified. Subgroup analyses identified the strongest correlations among studies conducted in Asian regions and among studies employing the YIAT. When assessed individually, all impulsivity subtypes were correlated with GD, but differences according to each subtype could not be determined due to the small number of eligible studies.

Our primary finding of a small-to-moderate correlation between GD and impulsivity is consistent with recent systematic reviews. Andrade et al. (2024) identified a positive association between GD and overall impulsivity across 11 out of 16 studies in their review. Notably, their review focused primarily on studies comparing GD with non-GD samples. Similarly, Şalvarlı and Griffiths (2022) also reported a positive association between GD and impulsivity among 32 out of 33 studies included in their review and noted that impulsivity is a risk factor for GD. The present meta-analytic review was able to extend these reviews by exploring sources of heterogeneity in effect size estimates, such as stronger correlations in Asian countries compared to Western countries, and stronger associations depending on GD and impulsivity assessment tools. Furthermore, previous reviews had not quantified the magnitude of associations of impulsivity subtypes with GD symptoms.

The association between impulsivity and GD identified was consistent with the relationship reported in studies of other psychiatric conditions (Fields et al., 2021; Moeller et al., 2001). For comparison, the association between trait impulsivity and substance-based addictions has been reported in studies of the following conditions: alcohol use disorders (r = 0.28; Coskunpinar et al., 2013); cocaine dependence (r = 0.28; García-Marchena et al., 2018); gambling (r = 0.21; Dowling et al., 2017) and other behaviours proposed (but not formally recognized and often debated) to be behavioural addictions – including problematic smartphone use (r = 0.39; Li et al., 2020), Facebook use (r = 0.25; Rajesh and Rangaiah, 2022) and general problematic Internet use (r = 0.37; Li et al., 2021). These results support the notion that certain individuals exhibit a generalized predisposition towards impulsive, inappropriate and premature behaviours, which can lead to pathological outcomes and harm (Ioannidis et al., 2019). It is possible that the link between specific disorders and impulsivity is partly due to their conceptualization, because a lack of behavioural inhibition is a common diagnostic criterion (Moeller et al., 2001). Regardless, the highly consistent results highlight the potential role of impulsivity in the development, onset and maintenance of these conditions.

The present work identified variations in the magnitude of the GD-impulsivity association depending on demographic and methodological factors. Geographical region accounted for significant variability in estimates, with impulsivity most strongly related to GD in Asian regions compared to Western regions. Previous research has shown Asian regions are more widely studied and have higher GD prevalence compared with other regions (Stevens et al., 2021), possibly due to the popularity and cultural acceptance of gaming in these areas (Anh, 2021), but it bears noting the potential influence of GD tools that tend to overestimate harms. However, some research suggests that UPPS results may vary across regions, suggesting caution should be exercised when making cross-cultural comparisons (Fournier et al., 2025).

Another finding of this meta-analytic review was there was no moderating effect of gender, age, sample size or risk of bias. Previous literature indicates that younger males are significantly more at risk of GD (Stevens et al., 2021); however, our results indicate that these factors do not affect the association between GD and impulsivity. This implies that factors related to age and gender that were not measured in this review – such as social norms, peer influence, identity formation, cognitive emotional regulation difficulties and early exposure to gaming – play a role in the development of gaming disorder beyond impulsivity alone.

The choice of assessment tool for both GD and impulsivity was significantly associated with variability in outcomes. Although the differences associated with the choice of assessment for impulsivity were not statistically significant, the Barratt Impulsivity Scale (BIS-11) yielded stronger correlations (r = 0.34) compared with UPPS Impulsive Behaviour Scales (r = 0.23). The stronger effect for the BIS scale may be in part attributable to its validity in measuring impulsivity as a broad latent construct compared to the UPPS. Alternatively, there may again be an influence of conceptualization, with the stronger association found for the BIS potentially resulting from greater assessment of behavioural inhibition than with UPPS scales. In addition, studies assessing GD using Young’s Internet Addiction Test (YIAT) and the Internet Gaming Disorder Scale–Short-Form (IGDS9-SF) reported the strongest associations. It should be noted that studies conducted in Asian regions more frequently utilized the YIAT, which may contribute to the stronger predicted effect in these regions. Only one study included a GD tool based on ICD-11 criteria (Raybould and Tunney, 2024). Since ICD-11 criteria reflect a more robust conceptualization of GD compared with the Diagnostic and Statistical Manual of Mental Disorders (5th ed., text rev.; DSM-5-TR), it is likely that the reported subgroup effects will hold when employing stricter ICD-11 criteria. While variance in study results according to GD tool is not uncommon and tools are known to demonstrate varying psychometric properties, the present work supports generally consistent correlations across GD tools (King et al., 2020; Stevens et al., 2021).

Exploration of the relationship between impulsivity subtypes and GD revealed that three subtypes of the BIS-11 were all significantly independently correlated with GD. This may account for the overall stronger association of the BIS tools compared with the UPPS tools (where only two subtypes were significantly, independently correlated with GD). The consistent correlations across BIS subscales provide further evidence for the tool’s validity as a stable, unidimensional measure of impulsivity and its individual subscales (Meule et al., 2015). Positive and negative urgency subscales of the UPPS were independently associated with GD, whereas lack of premeditation, lack of perseverance and sensation seeking showed no significant associations. This result aligns with research on the importance of urgency subfactors in addiction severity for gambling, gaming and pornography use (López-Guerrero et al., 2023; Raybould and Tunney, 2024) and theoretical frameworks conceptualizing GD as a maladaptive emotional coping strategy (Brand et al., 2016). Some theoretical considerations may help to explain the absence of significant associations between non-urgency impulsivity subtypes and GD. For example, perseverance deficits typically impair performance on boring or effortful tasks but may be less evident in games that offer immediate and stimulating reinforcement (King et al., 2010). Similarly, sensation seeking may be a less prominent driver of sustained problematic play, as many problematic gamers engage in familiar, long-term games rather than constantly seeking novelty (Billieux et al., 2015) and engage in repetitive tasks (e.g. grinding) (Johnston and Dinc, 2024). With these considerations, future research should examine how individual subtypes relate to problematic gaming and how these relationships affect risk assessment and interventions (Raji et al., 2025).

The present review focused exclusively on trait measures of impulsivity and found that impulsive tendencies are positively associated with GD symptoms. This precludes us from commenting on state-level impulsivity, which has previously been assessed in the literature (e.g. Andrade et al., 2024; Ioannidis et al., 2019; Lee et al., 2019; Şalvarlı and Griffiths, 2022). In addition, we were able to include studies which reported impulsivity and GD data but which had an alternate research focus. This broadened the range and scope of studies in our analysis, which may have increased the generalizability of the evaluated associations.

The findings of this review have several practical implications. The consistent association between impulsivity and GD suggests that clinical evaluation of gamers should include measurement of impulsivity. Elevated trait impulsivity levels may offer clinical insight by indicating a higher likelihood of difficulty in regulating disordered gaming behaviours. Moreover, assessing specific subtypes of impulsivity may provide valuable insights and inform tailored treatment strategies. For example, individuals high in motor impulsivity may benefit from pharmacological approaches such as stimulant treatment (Chamberlain and Sahakian, 2007), while those with elevated negative urgency are more likely to benefit from interventions focused on distress tolerance and emotional regulation skills (Cyders and Smith, 2008; Nuske et al., 2025). In terms of research implications, future studies could explore whether particular impulsivity subtypes and interacting factors confer greater or lesser vulnerability to developing GD. In addition, exploring whether certain game features disproportionately attract impulsive players, thereby encouraging dysfunctional engagement, would offer significant benefits to understanding and addressing GD.

The main strength of this meta-analytic review is its novel focus on impulsivity subtypes, highlighting the potential relationship between factors such as urgency and attentional impulsivity. This review extended earlier reviews by providing further support for the relationship between overall trait impulsivity and GD. However, there are some limitations that warrant acknowledgement. First, some of the subgroups may have been underpowered due to sample size. The Cochrane Handbook advises that subgroup analyses include at least 10 studies for each continuous variable, but fewer can still be informative (Deeks et al., 2024). In addition, the analysis revealed large amounts of heterogeneity across the observed effects and subgroups. The absence of an observed association between study reporting quality and effect sizes may reflect limitations of the chosen quality assessment tool, which produced a restricted range of ratings (with most studies clustering at the higher end). This restricted variability likely reduced the ability to detect any moderating influence of study quality. Furthermore, some studies did not report complete data, and attempts to contact the authors were unsuccessful; as a result, correlations were calculated using only the available data, which may have impacted the robustness of the findings. Conclusions should therefore be interpreted cautiously.

Future research could include clinical samples assessing trait impulsivity subtypes, behavioural measures and possible impulsivity profiles. Use of validated tools and reporting of all outcome measures and correlations would increase the accuracy of pooled meta-analytic results. Studies should include GD tools based on both DSM-5 and ICD-11 criteria to test if there is variance in associations with impulsivity. Investigating comorbidities, family history, genetic contribution and other moderating environmental, personal and social factors will help us understand the specific contribution of impulsivity in GD. The current research also highlights the role of impulsivity as a target for intervention in longitudinal studies. Associations of impulsivity with gamer-specific variables could also be investigated, such as game genre preferences, gaming motivations, emotional regulation and comorbidity. Future research could also examine how impulsivity manifests in various gaming contexts, taking into account environmental influences and structural features of games.

Conclusion

This meta-analytic review identified a consistent, small-to-moderate association between trait impulsivity and GD. This finding aligns with other studies and meta-analytic reviews of other addictive and disordered behaviours. Although impulsivity subtypes were independently, positively correlated with GD, there was insufficient data to directly compare subtypes. This review supports psychological models which emphasize impulsivity in the development and maintenance of GD. Further research to delineate specific associations of impulsivity types and impulsive behaviours will help us understand how gamers can transition from engaged to dysfunctional. Understanding the specific role of impulsivity, whether as a causal or interacting factor, will ultimately inform whether it can be targeted through interventions and help decrease problem behaviours.