Sage Journals: Discover world-class research

Abstract

Background:

Gaming disorder was included in the latest revision of the International Classification of Diseases (11th ed.). Worldwide, prevalence estimates of gaming disorder are considerably heterogeneous and often appear to be exceedingly high. However, few studies have examined the methodological, cultural and/or demographic factors that might explain this phenomenon. This review employed meta-analytic techniques to compute the worldwide-pooled prevalence of gaming disorder and evaluate the potential contributing factors for varied prevalence estimates.

Method:

Prevalence estimates were extracted from 53 studies conducted between 2009 and 2019, which included 226,247 participants across 17 different countries. Study findings were meta-analyzed using a random-effects model. Subgroup and moderator analyses examined potential sources of heterogeneity, including assessment tool and cut-off, participant age and gender, sample size and type, study region, and year of data collection.

Results:

The worldwide prevalence of gaming disorder was 3.05% (confidence interval: [2.38, 3.91]); this figure was adjusted to 1.96% [0.19, 17.12] when considering only studies that met more stringent sampling criteria (e.g. stratified random sampling). However, these estimates were associated with significant variability. The choice of screening tool accounted for 77% of the variance, with the Lemmens Internet gaming disorder-9, Gaming Addiction Identification Test and Problematic Videogame Playing scales associated with the highest estimates. Adolescent samples, lower cut-off scores and smaller sample size were significant predictors of higher prevalence. Gaming disorder rates were approximately 2.5:1 in favor of males compared to females.

Conclusion:

The worldwide prevalence of gaming disorder appears to be comparable to obsessive-compulsive disorder and some substance-related addictions, but lower than compulsive buying and higher than problem gambling. Gaming disorder prevalence rates appear to be inflated by methodological characteristics, particularly measurement and sampling issues.

Keywords

Gaming disorder prevalence meta-analysis addiction video game ICD-11

Introduction

Following a provisional status for ‘Internet gaming disorder’ (IGD) in the Diagnostic and Statistical Manual of Mental Disorders (5th ed.; DSM-5), gaming disorder (GD) was officially adopted at the World Health Assembly in May 2019 as a diagnosis in the 11th edition of the International Classification of Diseases (ICD-11; https://icd.who.int/en). GD is characterized by persistent gaming behavior, impaired control over gaming and functional impairment due to gaming usually for a period of at least 12 months. Individuals with GD play games to the exclusion of other activities, resulting in missed life opportunities and interference with normal routine and basic self-care (i.e. sleep, eating, personal hygiene); real-world social interaction (i.e. meeting friends, visiting family); and important responsibilities (i.e. school, work, care of children) (Brand et al., 2019; Humphreys, 2019; Wölfling et al., 2019). Individuals with GD often feel unable to regulate or cease their gaming behavior and experience intense negative mood states (e.g. irritability, sadness and boredom) when unable to play (Kaptsis et al., 2016; King et al., 2018; Rumpf et al., 2018; Saunders et al., 2017). Personal distress may relate to a fear of missing out on the online game world, where the user feels a strong sense of personal identity and self-efficacy (King and Delfabbro, 2014; Leménager et al., 2013; Marino and Spada, 2017; Stevens et al., 2019; Wegmann et al., 2017). With the inclusion of GD as a diagnostic category in the ICD-11, it was considered timely to evaluate the worldwide prevalence of GD to assist in estimating the potential burden of harm related to excessive gaming behaviors.

The conceptualization of GD shares features and symptoms (e.g. loss of control and continuation despite awareness of harm) with other addictive behaviors (e.g. gambling disorder), substance use disorders (SUDs), impulse control disorders (e.g. compulsive buying disorder [CBD]) and obsessive-compulsive disorder (OCD). However, the extent to which the prevalence of GD may be comparable to these disorders is unclear. For context, Calado and Griffiths (2016) reported that the worldwide prevalence of problem gambling was between 0.12% and 5.80%, with most studies reporting figures less than 1%. Degenhardt et al.’s (2017) review of 28 nationally representative studies reported the prevalence of various SUDs was 2.6%. Epidemiological studies of OCD have estimated that its prevalence falls between 0.6% and 3.5% (Crino et al., 2005; Kessler et al., 2005; Ruscio et al., 2010). Higher prevalence rates have generally been reported for compulsive buying, with a meta-analysis of 40 studies by Maraz et al. (2016) reporting the prevalence of CBD was between 3.4% and 6.9%.

Understanding the epidemiological aspects of GD provides valuable insight into its etiology and risk factors (Mihara and Higuchi, 2017). Prevalence data are also important for guiding decisions on public health policy interventions and the allocation of research funding and health resources. In the past few decades, a number of research teams, particularly those in Europe and Asia, have made substantial efforts to define the prevalence of GD and/or problematic gaming. Some studies have reported figures between 1% and 2% (Haagsma et al., 2012; Müller et al., 2015; Rehbein et al., 2015), whereas other estimates have been as high as 14% (Wang et al., 2015; Wartberg et al., 2019) and even in excess of 20% (Vadlin et al., 2018). Given that such figures greatly exceed those of related disorders, it is necessary to evaluate whether there may be sources of bias that affect the validity of these prevalence estimates (King et al., 2020a).

Variability across GD prevalence rates is proposed to be attributable to the influence of demographic and cultural factors, as well as the methodological characteristics of studies. Research has consistently reported, for example, that adolescents have higher rates of gaming-related problems than older age groups (Mak et al., 2014; Mentzoni et al., 2011; Müller et al., 2015; Van Rooij et al., 2011). Studies also report that males engage in longer gaming sessions (Desai et al., 2010) and experience more gaming-related problems than females (Rehbein et al., 2010; Turner et al., 2012; Vadlin et al., 2015; Wang et al., 2014). Age and gender differences have been reported in studies of impulse control disorders, but not necessarily in the same direction, for example, compulsive buying appears to be more prevalent among females (Maraz et al., 2016).

In their systematic review of GD prevalence, Feng et al. (2017) examined 27 studies and reported prevalence estimates ranging from 0.7% to 15.6% among school-aged children. The authors concluded that the average prevalence from 1998 to 2016 ‘[did] not appear to have increased in prevalence to the extent that internet usage has increased’ (p. 22). However, the authors did not conduct subgroup analyses to identify potential sources of bias. Another review by Fam (2018), which involved a meta-analysis of 16 studies of adolescents, reported that the pooled prevalence estimate was 4.6%, including 6.8% among males. Fam’s (2018) subgroup analyses indicated that prevalence estimates tended to be higher among older (pre-2000) studies with smaller (N < 1000) samples conducted in Asia and for studies which employed measurement adapted from DSM pathological gambling criteria. Another systematic review by Mihara and Higuchi (2017) of 50 GD prevalence studies—the largest to date—found prevalence rates ranged between 0.7% and 27.5%. The authors noted that methodological differences, particularly the choice of instrument and cut-off score, made it difficult to compare prevalence estimates across studies. These studies highlight the complexities that arise when attempting to obtain an accurate pooled prevalence rate of GD but have not yet determined clearly the most significant factors contributing to variability in rates.

Another important reason for undertaking this meta-analysis of GD prevalence was to account for differences that may arise due to cultural factors (King et al., 2018; Long et al., 2018). A review of 36 studies published in Chinese by Long et al. (2018) reported that the prevalence of GD ranged from 3.5% to 17%. Gaming is a popular activity in some East Asian countries, including Japan and China and particularly in South Korea, where gaming is highly accessible and available ‘24/7’ (e.g. ‘PC bang’ or Internet café culture) and supported by high-speed Internet services and related infrastructure propelled by booming esports industries. These countries have also been active in promoting public health and regulatory initiatives to reduce excessive use of gaming and other screen-based activities (Humphreys, 2019). One explanation for higher GD rates in these regions has been the greater familial and societal pressures on young people to succeed in education, which may be pronounced in single-child families (Lim, 2012). However, higher prevalence rates in East Asian regions (Fam, 2018) should be considered cautiously given that these studies have often employed instruments (e.g. the Internet Addiction Test) that were developed outside of Asia and without adequate attempts to ensure that these measures are valid and culturally appropriate (King et al., 2020; Wang et al., 2014).

Methodological characteristics of studies, including the instrument and its scoring, are an important influence on GD prevalence (Lortie and Guitton, 2013; Petry et al., 2015); however, these issues have received less attention than demographic and regional differences. The tools used to screen for GD have changed greatly over time, and many tools have been based on criteria transposed from other disorders, such as gambling and SUDs. This may affect the validity and applicability of certain items (e.g. tolerance and withdrawal, deception and chasing losses). Variability in GD measurement, including different profiles of strengths and weaknesses across instruments, has been highlighted by King et al. (2020) in a systematic review that identified 32 tools used across 320 studies of GD in Europe and Asia. Tools have varied in their scope of coverage of DSM-5 and ICD-11 criteria for GD, which may potentially affect their sensitivity and specificity. For example, the criterion referring to continued use despite harm (i.e. an important feature of addictive disorders, including in the ICD-11 conceptualization of GD) has appeared in only 9 out of 32 tools. Currently, the extent to which specific tools may contribute to variability in prevalence estimates is unclear.

The present review

Previous work has suggested that demographic and/or regional differences as well as methodological characteristics are likely to contribute to variability in GD prevalence estimates. However, these factors have not yet been examined systematically using meta-analytic techniques. The purpose of this review was to review the literature on the prevalence of GD to: (1) calculate the global pooled prevalence of GD and (2) determine which factors may be implicated in the variability of estimates by conducting subgroup analyses based on the demographic, regional and methodological characteristics of studies. Given limited previous reviews to guide hypotheses, this review was designed to be exploratory but expected to identify significant differences in prevalence rates according to all these characteristics.

Method

Data sources

The present review sought to collect the findings of all published studies reporting prevalence rates of GD. Although numerous studies refer to ‘Internet gaming disorder’ or IGD (i.e. the DSM-5 terminology), this review uses the abbreviation GD to encompass both the DSM-5 and ICD-11 classifications, including when referring to studies that predate the DSM-5. This review employed the same methodology used in King et al.’s (2020) review, which involved the identification of all empirical literature (N = 320 studies) that employed a named instrument to assess GD. King et al.’s (2020) review involved a database search using the following keywords and protocol: ([or tool or test or validation or psychometric or screening or diagnostic or item or instrument] AND gaming). Databases included PsycINFO, PsycArticles, PubMed, ScienceDirect, Scopus, Web of Science and Google Scholar. A PRISMA (Preferred Reporting Items for Systematic Reviews and Meta-Analyses) chart and supplementary file containing all 320 references is available in this open-access paper. There were 5828 results (including duplicate results). The titles and abstracts of all results generated by each database were screened for relevance, which led to the initial identification of 75 studies. Three duplicate studies (i.e. same data re-analyzed) were removed. Sixteen studies did not provide a prevalence estimate or sufficient data for manual calculation. Note that the reference lists of included studies were not checked for potential additional studies; however, this would likely not have identified further studies due to the requirement of studies to have used a named GD tool (which would likely be cited and therefore identified by the protocol). Thus, there were 53 studies for inclusion in the present review, which encompassed a total of 226,247 participants across 17 different countries.

In the present review, studies from the list of 320 studies met the following inclusion criteria: (1) peer-reviewed and published in English, (2) assessed problematic gaming or GD using a named IGD/GD tool and (3) reported a prevalence rate of IGD/GD. In relation to criterion (3), a further check of each study was performed to assess each prevalence rate against the sample characteristics of the study (i.e. identifying level of rigor and variability in study procedures and reporting). This subsequent check led to studies being categorized into one of four categories based on aspects of their sampling (see Data Analysis or Table 1 legend for a complete description of each ‘sample type’). Regarding criterion (2), studies that employed Internet addiction tools were included when they were used to screen for gaming-related problems, for example, the Young Internet Addiction Test (YIAT) and Compulsive Internet Use Scale (CIUS).

Table 1.

Characteristics of included studies (N = 53) in order of year of data collection.

Study	Data collected^a	N	Location	Mean age	Age range	Female (%)	Prevalence (%)			Sample type^b	Tool	Cut-off
Study	Data collected^a	N	Location	Mean age	Age range	Female (%)	Total	Male	Female	Sample type^b	Tool	Cut-off
Turner et al. (2012)	2007	2826	CA	15.00	12–19	52.70	8.53	15.1	3.1	1	PVP	5/8
Rehbein et al. (2010)	2008	15,168	DE	15.30	15–16	51.30	1.7	3.0	0.3	1	CSAS-II	42/56
Van Rooij et al. (2010)	2009	8299	NL	14.35	13–16	48.50	1.04	NR	NR	4	CIUS	40/56
Mentzoni et al. (2011)	2009	816	NO	NR	15–40	NR	0.61	7.2	1.3	1	GAS-7	3/7
Brunborg et al. (2013)	2009	1320	NO	13.60	12–14	52.12	4.24	6.5	2.2	4	GAS-7	28/35
Jeong and Kim (2011)	2010	600	SK	NR	12–18	53.20	2.17	NR	NR	2	YIAT-20	80/100
Shiue (2015)	2010	5003	JP	NR	20–89	42.26	5.50	NR	NR	1	JGSS	NR
Seok and Da Costa (2012)	2011	1332	SK	NR	18–29	14.00	2.70	NR	NR	2	A-EQ	6/11
Festl et al. (2013)	2011	4382	DE	37.80	14–90	NR	0.16	4.1	3.2	4	GAS-7	7/7
Király et al. (2014)	2011	4875	HU	16.40	15–18	68.40	4.33	NR	NR	4	POGQ-SF	32/60
Pápay et al. (2013)	2011	5045	HU	16.40	15–18	51.00	4.56	3.8	0.6	4	POGQ-SF	32/60
Kim et al. (2017)	2011	1401	SK	33.13	18–74	69.90	7.71	5.2	2.5	4	PVP	4/9
Faulkner et al. (2015)	2011	3338	CA	15.90	11–20	51.10	1.89	1.6	0.3	1	PVP	5/9
Van Rooij et al. (2014)	2011	8478	NL	14.20	12–16	49.00	2.41	4.8	1.3	1	VAT	4/5
Khazaal et al. (2016)	2011	5983	MN	20.00	19–22	100.00	2.3	NR	NR	3	GAS-7	21/35
Müller et al. (2015)	2012	12,938	MN	15.80	14–17	47.10	1.62	3.6	0.3	1	AICA-S	14/14
Colder Carras and Kardefelt-Winther (2018)	2012	13,460	MN	15.80	NR	62.90	2.20	NR	NR	1	AICA-S	13/14
Vadlin et al. (2015)	2012	1783	SE	NR	13–15	45.20	1.29	1.3	0.0	1	GAIT	5/9
Colder Carras et al. (2017)	2012	9733	NL	NR	NR	48.83	1.30	NR	NR	1	VAT	4/5
Ustinavičienė et al. (2016)	2012	1730	LT	15.84	13–18	50.20	9.13	10.6	7.7	1	YDQ	5/8
Kuss et al. (2013)	2012	3105	NL	14.20	11–19	48.30	3.64	1.7	2.0	1	CIUS	28/56
Wartberg et al. (2015)	2012	1744	DE	15.50	14–17	50.00	3.21	1.9	1.4	1	CIUS	28/56
Rehbein et al. (2015)	2013	11,003	DE	14.88	13–18	51.09	1.16	.25	0.07	1	CSAS	5/9
Wang et al. (2014)	2013	503	HK	14.58	13–16	49.50	14.71	22.7	8.7	2	GAS-7	3/9
Brunborg et al. (2015)	2013	3037	NO	35.00	16–74	58.00	0.72	NR	NR	1	GAS-7	21/35
Wang et al. (2015)	2013	920	HK	15.03	NR	36.30	13.04	10.21	2.82	2	GAS-7	21/35
Wittek et al. (2016)	2013	10,081	NO	32.60	16–74	62.70	0.53	.23	0.0	1	GAS-7	28/35
Henchoz et al. (2016)	2013	5223	SW	21.25	NR	100.00	1.6	NR	NR	3	GAS-7	21/35
Müller et al. (2014)	2013	2512	DE	NR	14–94	48.50	2.11	5.2	2.5	1	AICA-S	7/14
Strittmatter et al. (2015)	2014	8807	MN	15.00	NR	44.50	3.62	2.64	0.97	1	YDQ	5/8
Männikkö et al. (2017)	2014	271	FI	18.70	13–24	52.70	8.86	NR	NR	4	GAS-7	4/7
Vink et al. (2016)	2014	5278	NL	15.60	13–18	39.80	4.00	NR	NR	1	CIUS	28/56
Strittmatter et al. (2016)	2014	515	DE	14.50	13–17	37.90	3.11	NR	NR	1	YDQ	5/8
Gmel et al. (2019)	2015	2921	SW	NR	15–35 +	50.30	3.36	2.5	3.8	1	CIUS	13/32
Vadlin et al. (2018)	2015	1576	SE	NR	13–18	42.01	21.76	16.03	3.66	1	GAIT	44/52
Pontes et al. (2016)	2015	1071	SI	13.44	12–16	50.20	2.43	NR	NR	1	IGDS9-SF	25/45
Durkee et al. (2016)	2015	11,931	MN	14.89	NR	43.40	4.43	5.2	3.9	1	YDQ	5/8
Li et al. (2016)	2015	654	CN	20.29	18–22	45.60	4.74	3.7	1.1	2	OGCAS	32/80
Przybylski et al. (2017)	2015	1247	US	NR	18–24	42.26	2.80	NR	NR	2	IGD	5/9
Przybylski et al. (2017)	2015	1899	UK	NR	18–24	49.55	2.70	NR	NR	2	IGD	5/9
Przybylski et al. (2017)	2015	10,009	MN	NR	18–65	49.9	2.60	NR	NR	2	IGD	5/9
Przybylski et al. (2017)	2015	5777	US	45.49	NR	57.6	1.20	NR	NR	2	IGD	5/9
Yu and Cho (2016)	2015	2024	SK	14.5	14–15	50.6	5.88	10.4	1.2	1	IGD-9	5/9
Müller et al. (2018)	2016	1489	DE	13.10	10–17	45.90	2.48	1.9	3.0	1	AICA-S	13/14
Wartberg et al. (2017)	2016	1531	DE	18.86	12–25	51.40	5.75	8.4	2.9	2	IGD-9	5/9
Kim et al. (2018)	2016	402	SK	13.00	12–14	55.50	8.96	NR	NR	2	IGUESS	10/27
Dreier et al. (2017)	2016	3967	DE	15.55	12–18	54.50	1.94	5.7	4.6	1	AICA-S	13/14
Vadlin et al. (2016)	2016	2110	SE	14.07	12–16	43.44	9.48	17.3	3.7	1	GAIT	19/52
Wartberg et al. (2019)	2017	985	DE	13.40	12–14	50.80	14.42	NR	NR	1	IGD-9	5/9
Van den Eijnden et al. (2018)	2017	538	NL	12.90	12–15	48.90	9.76	10.23	9.32	2	IGD-9	5/9
Chiu et al. (2018)	2017	8110	TW	15.66	15–16	53.50	3.14	2.6	0.4	4	IGDT-10	15/27
Myrseth and Notelaers (2018)	2017	2055	NO	17.50	NR	47.10	1.22	NR	NR	1	GASA	28/35
Lin et al. (2019)	2018	4442	IR	15.30	13–18	50.30	2.39	NR	NR	1	IGDS9-SF	NR

A-EQ: Addiction Engagement Questionnaire; AICA-S: Scale for Assessment of Internet and Computer Gaming; CIUS: Compulsive Internet Use Scale; CSAS-II: Videogame Dependency Scale; GAIT: Gaming Addiction Identification Test; GAS-7: Gaming Addiction Scale (7-item); GASA: Gaming Addiction Scale–Adolescent; IGD: Internet gaming disorder; IGD-9: Lemmens IGD Scale (9-item); IGDT-10: Internet Gaming Disorder Test (10-item); IGUESS: Internet Game Use-Elicited Symptom Screen; JGSS: Japanese Gaming Social Survey; OGCAS: Online Gaming Attitudes and Cognitions Scale; POGQ-SF: Problematic Online Gaming Questionnaire–Short Form; PVP: Problematic Videogame Playing Scale; VAT: Videogame Addiction Test; YIAT-20: Young Internet Addiction Test (20-item); YDQ: Young Diagnostic Questionnaire; CA: Canada; CN: China; DE: Germany; FI: Finland; HK: Hong Kong; HU: Hungary; IR: Iran; JP: Japan; LT: Lithuania; MN: multi-national; NL: Netherlands; NO: Norway; SE: Sweden; SI: Slovenia; SK: South Korea; SW: Switzerland; TW: Taiwan; UK: United Kingdom; US: United States.

Year of data collection.

Sample type: Type 1: Representative [age, gender, education, socio-economic status, urbanicity] sampling, stratified random or randomized cluster sampling; population or cohort registry; Type 2: Convenience or purposive sampling; online (social media) marketing or advertising; non-representative sample via crowd-sourcing platform; Type 3: Selective sample based on male gender; Type 4: Selective sample based on gaming participation.

Data extraction

Data of interest were extracted from each study independently and checked for agreement by two authors (M.W.R.S. and D.L.K.). The variables of interest were participants (i.e. total number of participants, and total number with GD status); age (i.e. mean age, SD and age range); gender (i.e. total number of male and female participants); prevalence (i.e. proportion of individuals with GD among the total sample; and separate prevalence rates for males and females); methodology (i.e. location and year of data collection); and assessment (i.e. screening tool and cut-off score).

Data analysis

The first aim of the review was to determine a pooled estimate for prevalence of GD based on all 53 included studies. Reported prevalence estimates were extracted from each study and checked for accuracy by dividing the number of GD cases (n) by the number of total participants (N). Where possible, calculated rates were chosen in favor of reported rates to yield more precise estimates and to reduce error caused by rounding of figures.

When prevalence estimates are small (i.e. <0.10), heterogeneity tends to be artificially inflated (Higgins and Green, 2011). To counteract the influence of heterogeneity and outliers, all prevalence estimates were transformed into logits (or log odds). By transforming prevalence estimates onto a logit scale, the data become normally distributed, centered around a mean of zero, making statistical analysis more straightforward. The formula for the logit transformation and calculating the standard error were as follows

E S_{k} = \ln (\frac{p_{k}}{1 - p_{k}})

and

S E_{k} = \sqrt{\frac{1}{n_{k} p_{k}} + \frac{1}{n (1 - p_{k})}}

In the above case, ES is the log odds of study k, where ln is the natural log, and p is the raw GD prevalence estimate; and SE is the standard error for study k, where n is the total number of participants. Pooled prevalence estimates (point estimates) and 95% confidence intervals (CIs) were calculated using a random-effects model. All analyses were conducted using the logits and then converted back to the original proportion metric. A random-effects model was selected due to handling data from diverse populations (Borenstein et al., 2011). The formula for converting back to the original proportion metric was

p = (\frac{e^{E S_{k}}}{1 + e^{E S_{k}}})

The degree of between-study heterogeneity, or the variance in prevalence estimate, was examined using the following two indices: Cochrane’s Q, which determines the ratio of observed variation to within-study error (Borenstein et al., 2017) and the I² statistic, which estimates the true effect variance over sampling error observed. I² reflects the level of inconsistency of findings across studies and is expressed as a percentage, with values of 25%, 50% and 75% indicating low, medium and high heterogeneity, respectively (Borenstein et al., 2017). All analyses were conducted using Meta-Essentials v.1.4.

Subgroup and moderator analyses were conducted to identify possible sources of heterogeneity among prevalence estimates. Studies were sorted into the following categories: sample type (Type 1: Representative [age, gender, education, socio-economic status, urbanicity] sampling, stratified random or randomized cluster sampling; population or cohort registry; Type 2: Convenience or purposive sampling; online (social media) marketing or advertising; non-representative sample via crowd-sourcing platform; Type 3: Selective sampling based on male gender; Type 4: Selective sampling based on gaming participation); assessment (i.e. assessment tool used to identify potential cases of GD); region (i.e. geographical region of data collection); and year (i.e. year of data collection). In three cases where data collection dates were not provided, the year prior to publication was substituted. Point estimates and 95% CIs were then compared for each group using a random-effects model.

Three other potential sources of variance were identified. These were age (i.e. mean age of participants); cut-off (i.e. cut-off value used to determine cases of potential GD as a percentage of total score, or as percentage of endorsed criteria); and sample size (i.e. size of the total sample). Total sample size was reported in all included studies. In studies where GD cut-off scores were not reported, the cut-off recommended by the tool’s authors was used. The number of endorsed criteria (rather than a continuous score) was used in some cases. Cut-off scores were converted to a percentage. Similarly, gender rates were manually calculated if not reported and sufficient data were available. All three variables (i.e. age, cut-off and sample-size) were first analyzed individually using univariate regression to determine the ability of each variable to predict variance in GD prevalence scores. Significant univariates were then included in a multivariate regression analysis.

Potential publication bias was also checked in this analysis. Fail-safe N’s were calculated using Rosenthal’s (1979) method. A conservative Fail-safe N was adopted for the present review. Specifically, publication bias was deemed to not be an issue if the Fail-safe N value was greater than the number of studies (k) that contributed to each analysis.

Results

Study characteristics

Table 1 presents a summary of the study characteristics. Of studies that provided demographic information, the majority included adolescent-only samples (n = 25, 47.2%) with an overall mean age of 17.55 years (SD = 6.1). Of the 53 studies, 58.5% of studies (k = 31) belonged to the Type 1 (‘representative’) sample type. There were 12 studies (22.6%) in Type 2 (‘convenience/purposive’), 8 studies (15.1%) in Type 3 (‘male only’) and 2 studies (3.8%) in Type 4 (‘gamer only’) category. The majority of studies were conducted in Europe (k = 37, 69.9%), with the Gaming Addiction Scale (7-item; GAS-7) being the most used assessment tool (k = 11, 20.8%). In terms of cut-off, the mean adjusted score was 59.8% (SD = 17.0%), which equated to endorsement of 5.3 out of 9 DSM-5 GD criteria. The lowest cut-off value was 33% (i.e. 3/9 IGD criteria) (Wang et al., 2014), with the strictest tool requiring all criteria are met or a total score of 100% (Muller et al., 2015). Most studies (81.1%) measured GD specifically (k = 43) whereas other studies measured GD either by using a tool designed for problematic internet use (PIU) (k = 4, 7.5%) or a tool for both GD and PIU (k = 6, 11.3%). In terms of ICD-11 qualifiers for GD (i.e. online/offline distinction), the most common type of GD was online-only (k = 18, 33.4%) and both online and offline forms (k = 9, 17.0%), with the other studies (49.0%) not providing sufficient information about predominantly online/offline distinctions (k = 26).

Data synthesis

Overall, the pooled prevalence of GD across the 53 studies was 3.05% (k = 53 [2.38, 3.91]). However, significant heterogeneity between estimates was detected. Subgroup and moderator analyses were conducted to explore possible sources of heterogeneity. Tables 2 and 3 present the results of subgroup and moderator analyses, respectively. Figure 1 presents a forest plot of subgroups containing at least five studies. Figure 2 provides a visual summary of the GD prevalence rate across all 53 studies, in chronological order of data collection.

Table 2.

Subgroup analyses of prevalence according to assessment, region and year of data collection (N = 53 studies).

		Combined estimates				Heterogeneity				Tau			Model
	Subgroup	k	Prevalence (%)	95% CI	p	Q	df	p	I ²	Tau²	Tau	SE	p	Pseudo R² (%)
By sample type	−	−	−	−	−	−	−	−	−	−	−	−	0.061	9.93
	Type 1	31	1.96	[0.19, 17.12]	<0.01	2.10	30	0.147	52.34	0.04	0.19	0.18
	Type 2	12	2.91	[2.10, 4.01]	<0.01	2124.85	11	<0.01	98.59	0.70	0.84	0.16
	Type 3	2	4.56	[2.71, 7.56]	<0.01	1551.57	1	<0.01	99.29	0.91	0.95	0.24
	Type 4	8	2.97	[1.05, 8.07]	<0.01	473.23	7	<0.01	98.52	0.44	0.67	0.45
By assessment	−	−	−	−	−	−	−	−	−	−	−	−	<0.01	77.97
	CSAS^a	2	1.41	[0.12, 14.17]	<0.01	12.50	1	<0.01	92.00	0.06	0.26	0.19
	GAS-7^b	11	1.75	[0.71, 4.24]	<0.01	545.36	9	<0.01	98.17	1.27	1.12	0.41
	VAT	2	1.77	[0.03, 49.23]	<0.01	30.17	1	<0.01	96.69	0.19	0.43	0.31
	AICA-S	5	2.01	[1.64, 2.45]	<0.01	14.38	4	<0.01	72.18	0.02	0.15	0.07
	IGD Survey	4	2.20	[1.15, 4.17]	<0.01	36.73	3	<0.01	91.83	0.14	0.38	0.21
	IGDS-9-SF	2	2.39	[2.20, 2.61]	<0.01	0.01	2	0.937	0	0.00	0.00	0.01
	CIUS	5	2.78	[1.39, 5.47]	<0.01	123.67	4	<0.01	96.77	0.28	0.53	0.25
	POGQ-SF	2	4.45	[3.19, 6.17]	<0.01	0.31	1	0.577	0	0.00	0.00	0.03
	YDQ	4	4.76	[2.19, 10.02]	<0.01	99.95	3	<0.01	97.00	0.16	0.40	0.25
	PVP	3	5.07	[0.60, 32.16]	<0.01	124.34	2	<0.01	98.39	0.55	0.74	0.51
	IGD-9	5	6.89	[3.28, 13.89]	<0.01	239.24	2	<0.01	98.33	0.52	0.72	0.28
	GAIT	3	6.90	−	0.242	256.64	2	<0.01	99.22	1.02	1.01	0.90
By region	−	−	−	−	−	−	−	−	−	−	−	−	<0.05	12.33
	Europe	37	2.72	[1.96, 3.75]	<0.01	3219.21	36	<0.01	98.88	0.74	0.86	0.16
	USA	4	2.74	[0.70, 10.21]	<0.01	285.82	10	<0.01	98.95	1.21	1.10	0.44
	Asia	10	5.08	[3.29, 7.78]	<0.01	217.41	9	<0.01	95.86	0.28	0.53	0.20
By year	−	−	−	−	−	−	−	−	−	−	−	−	<0.05	16.53
	2009	3	1.46	[0.12, 15.16]	<0.01	74.12	2	<0.01	97.30	0.99	0.99	0.59
	2010	2	3.60	[0.01, 94.43]	<0.01	11.32	1	<0.01	91.17	0.65	0.43	0.48
	2011	8	2.54	[1.06, 5.96]	<0.01	245.76	7	<0.01	97.15	0.55	0.30	0.38
	2012	7	2.55	[1.31, 4.88]	<0.01	398.78	6	<0.01	98.50	0.72	0.52	0.28
	2013	8	2.03	[0.82, 4.90]	<0.01	536.18	7	<0.01	98.69	0.97	0.95	0.39
	2014	3	4.81	[1.23, 17.06]	<0.01	15.88	2	<0.01	87.40	0.46	0.21	0.33
	2015	6	5.37	[2.30, 12.03]	<0.01	683.00	5	<0.01	99.27	0.81	0.90	0.34
	2016	10	3.59	[2.19, 5.84]	<0.01	427.97	9	<0.01	99.40	0.53	0.73	0.23
	2017	4	4.97	[0.79, 25.55]	<0.01	297.88	3	<0.01	98.99	1.06	1.12	0.59

GAS-7: Gaming Addiction Scale (7-item); VAT: Videogame Addiction Test; AICA-S: Scale for Assessment of Internet and Computer Gaming; IGD: Internet gaming disorder; IGD-9: Lemmens IGD Scale (9-item); CIUS: Compulsive Internet Use Scale; POGQ-SF: Problematic Online Gaming Questionnaire–Short Form; YDQ: Young Diagnostic Questionnaire; PVP: Problematic Videogame Playing Scale; GAIT: Gaming Addiction Identification Test; CI: confidence interval; SE: standard error; GASA: Gaming Addiction Scale–Adolescent.

Indicates both CSAS and CSAS-II combined.

Indicates GAS-7 and GASA scales.

Table 3.

Univariate and multivariate regression analyses of prevalence according to age, tool cut-off and sample size.

	Model fit					Model parameters
	n	F	df	p	R ²	B	95% CI	p
Univariate
Age	41	14.39	1, 40	<0.01	26.9	–0.52	[–0.10, –0.03]	<0.01
IGD tool cut-off	51	9.82	1, 50	<0.01	16.7	–0.41	[–0.04, –0.01]	<0.01
Sample size	53	8.28	1, 52	<0.01	14.0	–0.37	[–0.00, –0.00]	<0.01
Multivariate
Age × sample size × cut-off	38	14.25	3, 35	<0.01	55.0	−	−	−

IGD: Internet gaming disorder; CI: confidence interval.

Figure 1.

Forest plot of pooled prevalence estimates across all groups containing >5 studies.

Figure 2.

Prevalence rates in order of data collection date (publication date in parentheses).

Subgroup analyses

Sampling

Statistical heterogeneity is inevitable in meta-analysis due to the diverse range of participants and study designs (Higgins et al., 2003). A subgroup that contains significant heterogeneity indicates that the underlying populations are too diverse to be considered homogeneous. Thus, non-significant results are not necessarily more ideal than significant results; the difference lies in the interpretation. For GD prevalence according to the sample type (i.e. Types 1, 2, 3 and 4; see Data Analysis for descriptions), the point estimate for Type 1 (k = 31) was significant, with I² estimates of heterogeneity indicating moderate heterogeneity. Types 2, 3 and 4 also had significant point estimates, with each group displaying large I² estimates of heterogeneity. Overall, sample type explained 9.93% of the variance in prevalence rates but was only considered significant at the 0.05 level (p = 0.023). Of the three subgroups including five or more studies, Type 1 samples (k = 31) provided the lowest estimates (1.96 [0.19, 17.2]), and Type 3 (k = 8, male only) samples reported the highest pooled prevalence (4.56 [2.71, 7.56]).

Assessment

Considering GD prevalence according to assessment tool, the point estimates for all assessment tools were significant except for the Gaming Addiction Identification Test (GAIT; k = 3, p = 0.242). I² estimates indicate heterogeneity was not detected for the Problematic Online Gaming Questionnaire–Short Form (POGQ-SF) and Internet Gaming Disorder Scale-9-Short Form (IGDS-9-SF); moderate for the Scale for Assessment of Internet and Computer Gaming (AICA-S) and high for the remaining tools. Overall, the choice of assessment tool was a significant predictor (p < 0.01) that accounted for 77.97% of the variance across prevalence estimates. Of the four subgroups including five or more studies, the GAS-7 reported the lowest prevalence estimates (1.75 [0.71, 4.24]) and the Lemmens IGD-9 reported the highest prevalence estimates (6.89 [3.28, 13.89]).

Region

For GD prevalence according to geographical region, the point estimates for Europe (k = 37), Asia (k = 10) and USA (k = 4) were significant, with each group displaying large I² estimates of heterogeneity. Overall, study region accounted for 12.33% of the variance in prevalence rates but was only considered significant at the 0.05 level (p = 0.039). Only two subgroups contained five or more studies, of which Asia reported higher prevalence estimates (5.08 [3.29, 7.78]) than Europe (2.72 [1.96, 3.75]).

Year

For GD prevalence according to year of data collection, the point estimates for all values were significant. In all cases, I² estimates of heterogeneity were large. Overall, year of data collection explained 16.53% of the variance in prevalence rates but was only considered significant at the 0.05 level (p = 0.012). Of the five subgroups including five or more studies, 2013 (k = 8) provided the lowest estimates (2.03 [0.82, 4.89]) and 2015 was the highest (5.37 [2.30, 12.03]).

Moderator analyses

Moderator analyses were conducted to examine possible effects of underlying univariates of participant age (k = 41), cut-off (k = 51) and sample size (k = 53). Univariate regression analysis determined that age, F(1,40) = 14.39, p = 0.001, R² = 26.9%, cut-off, F(1,50) = 9.82, p = 0.003, R² = 16.7%, and sample size, F(1,52) = 8.28, p = 0.006, R² = 14.0%, were all significant predictors of GD prevalence rate. A stepwise multivariate model was then generated using age, cut-off and sample size. The model was significant, F(3,35) = 14.25, p < 0.001, R² = 55.0%, with standardized coefficients: age (B = −0.05 [−0.08, −0.02], p = 0.001, B = −0.41); cut-off (B = −0.02 [−0.04, −0.01], p = 0.003, B = −0.41); and sample size (B = −0.00 [−0.00, −0.00], p = 0.022, B = −0.31), indicating that younger participants, smaller sample sizes and lower cut-off values predicted higher prevalence rates. Table 3 provides a summary of the meta-regression analyses.

Gender differences in GD prevalence

There were 31 studies that stratified prevalence by gender or provided sufficient information to calculate distinct prevalence rates according to gender. Gender-specific prevalence rates were included in a separate analysis. GD rates were approximately 2.5:1 in favor of males compared to females. Only 3 out of 31 studies (9.6%) reported a higher GD prevalence rate among females. Results from a one-way analysis of variance (ANOVA) were statistically significant, F(1,60) = 13.04, p = 0.001, indicating that the average GD rate was higher for male (6.31% [4.30, 8.31]) than female samples (2.54 [1.65, 3.43]).

Publication bias

No evidence of publication bias was found (Fail-safe N = 12,350), indicating 12,350 studies with null findings would need to be included in the analysis to render the true prevalence 0 (i.e. no effect).

Discussion

The purpose of this review was to determine the global prevalence of GD. To our knowledge, this work represents the largest systematic review and meta-analysis of GD prevalence rates, including a total of 226,247 participants across 17 different countries. The main findings were as follows: (1) the global pooled prevalence of GD was 3.05% (95% CI = [2.38, 3.91]), and this figure was adjusted to 1.96% (95% CI = [0.19, 17.12]) when considering only those studies which met more stringent sampling criteria (e.g. stratified random sampling); (2) prevalence estimates varied largely as a function of the GD screening tool; (3) studies with smaller sample sizes, adolescent samples and/or studies employing lower cut-off scores reported significantly higher prevalence rates; and (4) GD rates were approximately 2.5:1 in favor of males compared to females. There was also some evidence that higher prevalence rates were reported in more recent (post-2013) studies and in studies conducted in Asia as compared to Europe.

The results suggest that the global prevalence of GD (3.05%) may be comparable to some SUDs (2.6%) (Degenhardt et al., 2017) and OCD (0.6–3.5%) (Crino et al., 2005), but appears to be lower than compulsive buying (3–6%) (Maraz et al., 2016) and higher than problem gambling (close to 1%) (Calado and Griffiths, 2016). This GD prevalence rate is lower than the rate reported in Fam’s (2018) review, which appears attributable to its focus on adolescents. However, interpretation of these findings should take several factors into consideration. First, the results from subgroup analyses indicated GD prevalence estimates varied significantly according to the GD screening tool. In addition, as expected, moderator analyses identified lower cut-off scores as a significant predictor of higher prevalence rates. In some cases, cut-off scores for identifying potential cases of GD were as low as 33% of the total score. Studies employing cut-off scores lower than 50% should be treated cautiously, because these studies may inflate prevalence estimates above thresholds recommended by the DSM-5 (i.e. at least 5/9 criteria) and may be more likely to count cases without functional impairment.

Prevalence rates varied according to several study characteristics. Regression analysis identified that adolescent samples tended to report higher prevalence estimates. Higher GD prevalence among adolescents is a common finding in the literature (King et al., 2013; Mak et al., 2014; Müller et al., 2015). Given that the overall mean age of the total sample was 17.5 years (SD = 6.1), the total pooled prevalence figure may represent the peak of the distribution of GD scores. Gender was another demographic factor that affected prevalence rates in the expected direction (i.e. male bias). Consistent with previous reviews (Feng et al., 2017; Mihara and Higuchi, 2017), GD prevalence rates were found to be 2.5 times higher for males than for females. Although female prevalence rates were lower, the results suggested that female prevalence rates increased over the last decade. The average female GD prevalence rate was 1.75% in 2009 compared to 4.47% in 2017, which may be partly due to the increasing popularity of games among females, particularly for games that intersect with social media and related online activities (King and Potenza, 2020).

This review contributes to critical discussion of GD screening approaches by highlighting that GD prevalence is strongly influenced by the choice of instrument. The field has an abundance of screening tools (i.e. more than 30); many appear to be quite similar on face value but demonstrate varying psychometric properties (King et al., 2020). The relative performance of different tools according to sampling approaches warrants closer attention. This review highlights, for example, that the CSAS assessment tool has been administered exclusively in non-convenience samples and yielded the lowest prevalence estimate (1.41%). However, the GAIT was also administered in non-convenience samples, and the Lemmens IGD-9 was primarily administered (65%) in non-convenience samples, and these two tools yielded some of the highest prevalence estimates (6.90% and 6.89%, respectively). Thus, when considering prevalence rates according to tools, the effect of sampling, while relevant, does not necessarily tell the whole story.

As often the case in psychological research, it is the interaction between several factors that becomes important in understanding the bigger picture. Overall, GD prevalence was affected by the combination of small-sized and adolescent-only samples. For example, three of the five smallest studies in the present review had quite low mean ages and reported some of the highest prevalence rates (i.e. Kim et al. (2018) [8.96%, N = 402, age = 13.0 years]; Wang et al. (2014) [14.71%, N = 503, age =14.58 years]; and Van den Eijnden et al. (2018) [9.76%, N = 538, age = 12.9 years]). Such studies can contribute to the perception of GD prevalence as being unreasonably high which can then draw strong criticism of the validity of the condition. Although some studies have been described as ‘prevalence’ studies, a closer inspection of some studies reveals that their methods were not likely to yield genuinely representative data. The present review differentiated 31 studies, which we termed ‘Type 1’ studies, based on clearly identifiable representative sampling (e.g. according to age, gender, education, socio-economic status, urbanicity), stratified random or randomized cluster sampling, and population or cohort registry sampling. The pooled prevalence rate for Type 1 studies was 1.96%. However, it bears noting that, like the total sample of 53 studies, this figure had a large CI, partly due to lower cut-off scores applied in some of these studies which increased heterogeneity.

Prevalence rates were found to be higher for Asian countries than for other regions. These results support previous findings in which East Asian prevalence rates were among the highest worldwide (Fam, 2018; Long et al., 2018). The reasons for these regional differences remain unclear, but it is well-documented that gaming is very popular and considered a national pastime in regions including South Korea. The greater popularity of gaming in Asian regions is often understood with reference to the rise in popularity of esports, more abundant gaming infrastructure (i.e. 24/7 gaming cafes), and more accepting cultural attitudes toward gaming in general despite strong regulatory measures. However, methodological factors including a higher proportion of high-school survey studies conducted in those regions should be considered. There is a need for cross-cultural research on GD (i.e. comparing Eastern and Western countries) employing comparable methodological approaches, including recruitment, sampling and measurement (Rumpf et al., 2019).

Results suggest that GD prevalence has increased slightly over the last decade. In 2009, the average prevalence estimate was 1.97%, whereas estimates in 2017 averaged 4.70%. One reason for this apparent increase may be related to the adoption of the DSM-5 criteria into screening measures. The DSM-5 refers to nine criteria that include symptoms of preoccupation, tolerance and withdrawal, which have been shown to converge with highly engaged but non-problematic gaming (Charlton and Danforth, 2007). Although these criteria are listed in other addictive disorders (including gambling disorder), their inclusion in GD screening tools may lead to tests being overly sensitive. Some authors have argued that applying DSM-5 criteria to activities, including gaming and other everyday activities, may result in over-pathologization (Billieux et al., 2015). The DSM-5 approach (i.e. any five criteria, with harm not required) employed in some tools (e.g. the Lemmens IGD-9) may lead to artificially high prevalence rates that do not reflect genuine gaming-related problems. In support of this, the present review showed that the pre-2013 (i.e. pre-DSM-5) pooled prevalence estimate of GD was 2.35% compared to 4.45% post-DSM-5. As an alternative to the DSM-5, the ICD-11 guidelines refer to three GD criteria (over-prioritization of gaming, loss of control and continued use despite harm) and requires functional impairment. Therefore, future studies adopting the ICD-11 approach may be expected to report lower prevalence figures which may not necessarily reflect an actual decrease in prevalence rate. This potential decrease (due to the ICD-11 approach) should be anticipated by policymakers and public health authorities in evaluating the GD-related burden in populations over time, or in evaluating effects of potential GD preventive countermeasures (King and Delfabbro, 2017). On the contrary, there may be a general increase in gaming and associated problem gaming in the context of the COVID-19 pandemic (i.e. due to stay-at-home mandates and lockdowns) which may counteract this potential decrease (King et al., 2020c; Király et al., 2020).

The main strength of the present review was its scope. This meta-analysis was the largest to date, including 53 studies and 226,247 participants across 17 different countries. However, some limitations warrant acknowledgment. First, some of the subgroup analyses may have been underpowered. The Cochrane Handbook recommends the inclusion of at least 10 studies for each continuous variable in subgroup analyses (Higgins and Green, 2011). However, Fu et al. (2011) contend that between 6 and 10 studies is sufficient. Differences in sampling approaches were another potential limitation to the present review. Some studies did not provide detailed information about their sampling methods, which affected the designation of some studies to the four types of sampling. Another limitation is that only English-language studies were included, which affected the representativeness of cross-cultural comparisons. This review only included peer-reviewed studies; gray literature was not included, which may have excluded some important studies, such as those published in government reports. The relative number of adolescent samples also affects the overall representativeness of the data. About half of the studies did not report whether gaming occurred predominantly online versus offline or both (i.e. ICD-11 category distinctions).

As the GD evidence base continues to grow, it would be helpful for future reviews that consider prevalence rates to consider some additional variables. Specifically, this work could examine the potential effects of sampling methodologies in more detail (e.g. school-based samples, online convenience sampling and crowd-sourced data, such as Amazon MTurk). Another issue for consideration is the influence of comorbidity, for example, axis I disorders, other addictive disorders. A stronger focus on gaming-related harm (see Delfabbro et al., 2020; King and Delfabbro, 2018) to gain insights into genuine harm and functional impairment would strengthen future studies and meta-analyses.

Conclusion

This review found that the worldwide prevalence of GD was 3.05% (CI: [2.38, 3.91]) and this rate reduced to 1.96% [0.19, 17.12] when considering only those studies that met more stringent sampling criteria (e.g. stratified random sampling). However, given the larger proportion of adolescent samples in this evidence base, these prevalence rates may be substantially lower among other demographic groups. The results indicated that the choice of assessment tool accounted for a large proportion of the variance in GD prevalence. Furthermore, higher GD rates were found among adolescents and males, as well as in studies employing smaller sample sizes or lower cut-offs. This review affirms previous observations of the varied methodological approaches employed in GD research (King et al., 2020b; Zajac et al., 2017) and was consistent with Mihara and Higuchi’s (2017) observation that variability in GD screening greatly affects comparability of studies of prevalence. Future epidemiological research should apply standardized approaches to GD screening and report more detailed information on sampling. Given that much of the variability in prevalence is related to screening, a potentially helpful next step for the global GD research agenda would be to conduct studies employing cognitive interviews to gain insights into how people are interpreting screening questions or whether classifications based on screening instruments stand up to independent clinical review of the same cases. Future research should also consider the potential effect of other factors including the role of comorbidity and different types of games (Green et al., 2020; King et al., 2019a, 2019b).

Footnotes

Author Contributions

M.W.R.S. and D.L.K. designed the review and conducted literature searches and provided summaries of the identified content. M.W.R.S. and D.D. conducted the meta-analysis. M.W.R.S. wrote the first draft of the manuscript, and all authors contributed to and approved the final manuscript.

Declaration of Conflicting Interests

The author(s) declared no potential conflicts of interest with respect to the research, authorship and/or publication of this article.

Funding

The author(s) disclosed receipt of the following financial support for the research, authorship and/or publication of this article: This work received financial support from a Discovery Early Career Researcher Award (DECRA) DE170101198 funded by the Australian Research Council (ARC).

ORCID iDs

Matthew WR Stevens

Diana Dorstyn

Daniel L King

References

Billieux

Schimmenti

Khazaal

, et al. (2015) Are we overpathologizing everyday life? A tenable blueprint for behavioral addiction research. Journal of Behavioral Addictions 4: 119–123.

Borenstein

Hedges

Higgins

, et al. (2011) Introduction to Meta-Analysis. Chichester: John Wiley & Sons.

Borenstein

Higgins

Hedges

, et al. (2017) Basics of meta-analysis: I2 is not an absolute measure of heterogeneity. Research Synthesis Methods 8: 5–18.

Brand

Rumpf

H-J

Demetrovics

, et al. (2019) Gaming disorder is a disorder due to addictive behaviors: Evidence from behavioral and neuroscientific studies addressing cue-reactivity and craving, executive functions, and decision making. Current Addiction Reports 6: 296–302.

*Brunborg

Hanss

Mentzoni

, et al. (2015) Core and peripheral criteria of video game addiction in the game addiction scale for adolescents. Cyberpsychology, Behavior, and Social Networking 18: 280–285.

*Brunborg

Mentzoni

Melkevik

, et al. (2013) Gaming addiction, gaming engagement, and psychological health complaints among Norwegian adolescents. Media Psychology 16: 115–128.

Calado

Griffiths

(2016) Problem gambling worldwide: An update and systematic review of empirical research (2000–2015). Journal of Behavioral Addictions 5: 592–613.

*Colder Carras

Kardefelt-Winther

(2018) When addiction symptoms and life problems diverge: A latent class analysis of problematic gaming in a representative multinational sample of European adolescents. European Child & Adolescent Psychiatry 27: 513–525.

*Colder Carras

Van Rooij

Van de Mheen

, et al. (2017) Video gaming in a hyperconnected world: A cross-sectional study of heavy gaming, problematic gaming symptoms, and online socializing in adolescents. Computers in Human Behavior 68: 472–479.

10.

Charlton

Danforth

(2007) Distinguishing addiction and high engagement in the context of online game playing. Computers in Human Behavior 23: 1531–1548.

11.

*Chiu

Pan

Lin

(2018) Chinese adaptation of the Ten-Item Internet Gaming Disorder Test and prevalence estimate of Internet gaming disorder among adolescents in Taiwan. Journal of Behavioral Addictions 7: 719–726.

12.

Crino

Slade

Andrews

(2005) The changing prevalence and severity of obsessive-compulsive disorder criteria from DSM-III to DSM-IV. American Journal of Psychiatry 162: 876–882.

13.

Degenhardt

Glantz

Evans-Lacko

, et al. (2017) Estimating treatment coverage for people with substance use disorders: An analysis of data from the World Mental Health Surveys. World Psychiatry 16: 299–307.

14.

Delfabbro

Georgiou

King

(2020) Opportunity cost and gambling: Distinguishing between competing activities and harm. Journal of Gambling Issues 44. DOI: 10.4309/jgi.2020.44.9.

15.

Desai

Krishnan-Sarin

Cavallo

, et al. (2010) Video-gaming among high school students: Health correlates, gender differences, and problematic gaming. Pediatrics 126: e1414–e1424.

16.

*Dreier

Wölfling

Duven

, et al. (2017) Free-to-play: About addicted Whales, at risk Dolphins and healthy Minnows. Monetarization design and internet gaming disorder. Addictive Behaviors 64: 328–333.

17.

*Durkee

Carli

Floderus

, et al. (2016) Pathological internet use and risk-behaviors among European adolescents. International Journal of Environmental Research and Public Health 13: 294.

18.

Fam

(2018) Prevalence of internet gaming disorder in adolescents: A meta-analysis across three decades. Scandinavian Journal of Psychology 59: 524–531.

19.

*Faulkner

Irving

Adlaf

, et al. (2015) Subtypes of adolescent video gamers: A latent class analysis. International Journal of Mental Health and Addiction 13: 1–18.

20.

Feng

Ramo

Chan

, et al. (2017) Internet gaming disorder: Trends in prevalence 1998–2016. Addictive Behaviors 75: 17–24.

21.

*Festl

Scharkow

Quandt

(2013) Problematic computer game use among adolescents, younger and older adults. Addiction 108: 592–599.

22.

Gartlehner

Grant

, et al. (2011) Conducting quantitative synthesis when comparing medical interventions: AHRQ and the Effective Health Care Program. Journal of Clinical Epidemiology 64: 1187–1197.

23.

*Gmel

Khazaal

Studer

, et al. (2019) Development of a short form of the compulsive internet use scale in Switzerland. International Journal of Methods in Psychiatric Research 28: e1765.

24.

Green

Delfabbro

King

(2020) Avatar- and self-related processes and gaming disorder: A systematic review. Addictive Behaviors 108: 106461.

25.

Haagsma

Pieterse

Peters

(2012) The prevalence of problematic video gamers in the Netherlands. Cyberpsychology, Behavior, and Social Networking 15: 162–168.

26.

*Henchoz

Studer

Deline

, et al. (2016) Video gaming disorder and sport and exercise in emerging adulthood: A longitudinal study. Behavioral Medicine 42: 105–111.

27.

Higgins

Green

(eds) (2011) Cochrane handbook for systematic reviews of interventions. Version 5.1.0 (updated March 2011). The Cochrane Collaboration. Available at: http://www.cochrane-handbook.org

28.

Higgins

Thompson

Deeks

, et al. (2003) Measuring inconsistency in meta-analyses. British Medical Journal 327: 557–560.

29.

Humphreys

(2019) Sharpening the focus on gaming disorder. Bulletin of the World Health Organization 97: 382–383.

30.

*Jeong

Kim

(2011) Social activities, self-efficacy, game attitudes, and game addiction. Cyberpsychology, Behavior, and Social Networking 14: 213–221.

31.

Kaptsis

King

Delfabbro

, et al. (2016) Withdrawal symptoms in Internet gaming disorder: A systematic review. Clinical Psychology Review 43: 58–66.

32.

Kessler

Chiu

Demler

, et al. (2005) Prevalence, severity, and comorbidity of 12-month DSM-IV disorders in the National Comorbidity Survey Replication. Archives of General Psychiatry 62: 617–627.

33.

*Khazaal

Chatton

Rothen

, et al. (2016) Psychometric properties of the 7-item game addiction scale among French and German speaking adults. BMC Psychiatry 16: 132.

34.

*Kim

Kim

Lee

, et al. (2017) Internet game addiction, depression, and escape from negative emotions in adulthood: A nationwide community sample of Korea. The Journal of Nervous and Mental Disease 205: 568–573.

35.

*Kim

Yim

Jeong

, et al. (2018) The association between aggression and risk of Internet gaming disorder in Korean adolescents: The mediation effect of father-adolescent communication style. Epidemiology and Health 40: e2018039.

36.

King

Adair

Saunders

, et al. (2018) Clinical predictors of gaming abstinence in help-seeking adult problematic gamers. Psychiatry Research 261: 581–588.

37.

King

Delfabbro

(2014) The cognitive psychology of Internet gaming disorder. Clinical Psychology Review 34: 298–308.

38.

King

Delfabbro

(2017) Prevention and policy related to Internet gaming disorder. Current Addiction Reports 4: 284–292.

39.

King

Delfabbro

(2018) The concept of ‘harm’ in Internet gaming disorder. Journal of Behavioral Addictions 7: 562–564.

40.

King

Billieux

Carragher

, et al. (2020a) Face validity evaluation of screening tools for gaming disorder: Scope, language, and overpathologizing issues. Journal of Behavioral Addictions 9: 1–13.

41.

King

Chamberlain

Carragher

, et al. (2020b) Screening and assessment tools for gaming disorder: A comprehensive systematic review. Clinical Psychology Review 77: 101831.

42.

King

Delfabbro

Billieux

, et al. (2020c) Problematic online gaming and the COVID-19 pandemic. Journal of Behavioral Addictions 29: 184–186.

43.

King

Delfabbro

Deleuze

, et al. (2019a) Maladaptive player-game relationships in problematic gaming and gaming disorder: A systematic review. Clinical Psychology Review 73: 101777.

44.

King

Delfabbro

Doh

, et al. (2018) Policy and prevention approaches for disordered and hazardous gaming and Internet use: An international perspective. Prevention Science 19: 233–249.

45.

King

Delfabbro

Zwaans

, et al. (2013) Clinical features and axis I comorbidity of Australian adolescent pathological Internet and video game users. Australian & New Zealand Journal of Psychiatry 47: 1058–1067.

46.

King

Koster

Billieux

(2019b) Study what makes games addictive. Nature 573: 346.

47.

King

Potenza

(2020) Gaming disorder and female adolescents: A hidden problem? Journal of Adolescent Health 66: 650–652.

48.

*Király

Griffiths

Urbán

, et al. (2014) Problematic Internet use and problematic online gaming are not the same: Findings from a large nationally representative adolescent sample. Cyberpsychology, Behavior, and Social Networking 17: 749–754.

49.

Király

Potenza

Stein

, et al. (2020) Preventing problematic internet use during the COVID-19 pandemic: A consensus guideline. Comprehensive Psychiatry 100: 152180.

50.

*Kuss

van Rooij

Shorter

, et al. (2013) Internet addiction in adolescents: Prevalence and risk factors. Computers in Human Behavior 29: 1987–1996.

51.

Leménager

Gwodz

Richter

, et al. (2013) Self-concept deficits in massively multiplayer online role-playing games addiction. European Addiction Research 19: 227–234.

52.

*Li

Zou

Wang

, et al. (2016) Role of stressful life events, avoidant coping styles, and neuroticism in online game addiction among college students: A moderated mediation model. Frontiers in Psychology 7: 1794.

53.

Lim

(2012) Regulatory initiatives for managing online risks and opportunities for youths–the East Asian experience. In: Walrave

(ed.) e-Youth: Balancing between Opportunities and Risks? Brussels: Peter Lang, pp. 271–290.

54.

*Lin

Imani

Broström

, et al. (2019) Evaluating the psychometric properties of the 7-item Persian Game Addiction Scale for Iranian adolescents. Frontiers in Psychology 10: 149.

55.

Long

Liu

, et al. (2018) Prevalence and correlates of problematic online gaming: A systematic review of the evidence published in Chinese. Current Addiction Reports 5: 359–371.

56.

Lortie

Guitton

(2013) Internet addiction assessment tools: Dimensional structure and methodological status. Addiction 108: 1207–1216.

57.

Mak

Lai

, et al. (2014) Psychometric properties of the revised Chen internet addiction scale (CIAS-R) in Chinese adolescents. Journal of Abnormal Child Psychology 42: 1237–1245.

58.

*Männikkö

Billieux

Nordström

, et al. (2017) Problematic gaming behaviour in Finnish adolescents and young adults: Relation to game genres, gaming motives and self-awareness of problematic use. International Journal of Mental Health and Addiction 15: 324–338.

59.

Maraz

Griffiths

Demetrovics

(2016) The prevalence of compulsive buying: A meta-analysis. Addiction 111: 408–419.

60.

Marino

Spada

(2017) Dysfunctional cognitions in online gaming and internet gaming disorder: A narrative review and new classification. Current Addiction Reports 4: 308–316.

61.

*Mentzoni

Brunborg

Molde

, et al. (2011) Problematic video game use: Estimated prevalence and associations with mental and physical health. Cyberpsychology, Behavior, and Social Networking 14: 591–596.

62.

Mihara

Higuchi

(2017) Cross-sectional and longitudinal epidemiological studies of Internet gaming disorder: A systematic review of the literature. Psychiatry and Clinical Neurosciences 71: 425–444.

63.

*Müller

Glaesmer

Brähler

, et al. (2014) Prevalence of internet addiction in the general population: Results from a German population-based survey. Behaviour & Information Technology 33: 757–766.

64.

*Müller

Janikian

Dreier

, et al. (2015) Regular gaming behavior and internet gaming disorder in European adolescents: Results from a cross-national representative survey of prevalence, predictors, and psychopathological correlates. European Child & Adolescent Psychiatry 24: 565–574.

65.

*Müller

Wölfling

Beutel

, et al. (2018) Insights into aspects behind Internet-related disorders in adolescents: The interplay of personality and symptoms of adjustment disorders. Journal of Adolescent Health 62: 234–240.

66.

*Myrseth

Notelaers

(2018) A latent class approach for classifying the problem and disordered gamers in a group of adolescence. Frontiers in Psychology 9: 2273.

67.

*Pápay

Urbán

Griffiths

, et al. (2013) Psychometric properties of the problematic online gaming questionnaire short-form and prevalence of problematic online gaming in a national sample of adolescents. Cyberpsychology, Behavior, and Social Networking 16: 340–348.

68.

Petry

Rehbein

, et al. (2015) Internet gaming disorder in the DSM-5. Current Psychiatry Reports 17: 72.

69.

*Pontes

Macur

Griffiths

(2016) Internet Gaming Disorder among Slovenian primary schoolchildren: Findings from a nationally representative sample of adolescents. Journal of Behavioral Addictions 5: 304–310.

70.

*Przybylski

Weinstein

Murayama

(2017) Internet gaming disorder: Investigating the clinical relevance of a new phenomenon. American Journal of Psychiatry 174: 230–236.

71.

*Rehbein

Kliem

Baier

, et al. (2015) Prevalence of internet gaming disorder in German adolescents: Diagnostic contribution of the nine DSM-5 criteria in a state-wide representative sample. Addiction 110: 842–851.

72.

*Rehbein

Psych

Kleimann

, et al. (2010) Prevalence and risk factors of video game dependency in adolescence: Results of a German nationwide survey. Cyberpsychology, Behavior, and Social Networking 13: 269–277.

73.

Rosenthal

(1979) The file drawer problem and tolerance for null results. Psychological bulletin 86: 638.

74.

Rumpf

H-J

Achab

Billieux

, et al. (2018) Including gaming disorder in ICD-11: The need to do so from a clinical and public health perspective. Journal of Behavioral Addictions 7: 556–561.

75.

Rumpf

H-J

Brandt

Demetrovics

, et al. (2019) Epidemiological challenges in the study of behavioral addictions: A call for high standard methodologies. Current Addiction Reports 6: 331–337.

76.

Ruscio

Stein

Chiu

, et al. (2010) The epidemiology of obsessive-compulsive disorder in the National Comorbidity Survey Replication. Molecular Psychiatry 15: 53–63.

77.

Saunders

Hao

Long

, et al. (2017) Gaming disorder: Its delineation as an important condition for diagnosis, management, and prevention. Journal of Behavioral Addictions 6: 271–279.

78.

*Seok

Da Costa

(2012) The world’s most intense online gaming culture: Addiction and high-engagement prevalence rates among South Korean adolescents and young adults. Computers in Human Behavior 28: 2143–2151.

79.

*Shiue

(2015) Self and environmental exposures to drinking, smoking, gambling or video game addiction are associated with adult hypertension, heart and cerebrovascular diseases, allergy, self-rated health and happiness: Japanese General Social Survey, 2010. International Journal of Cardiology 181: 403–412.

80.

Stevens

MWR

King

Dorstyn

, et al. (2019) Cognitive-behavioral therapy for Internet gaming disorder: A systematic review and meta-analysis. Clinical Psychology and Psychotherapy 26: 191–203.

81.

*Strittmatter

Kaess

Parzer

, et al. (2015) Pathological Internet use among adolescents: Comparing gamers and non-gamers. Psychiatry Research 228: 128–135.

82.

*Strittmatter

Parzer

Brunner

, et al. (2016) A 2-year longitudinal study of prospective predictors of pathological Internet use in adolescents. European Child & Adolescent Psychiatry 25: 725–734.

83.

*Turner

Paglia-Boak

Ballon

, et al. (2012) Prevalence of problematic video gaming among Ontario adolescents. International Journal of Mental Health and Addiction 10: 877–889.

84.

*Ustinavičienė

Škėmienė

Lukšienė

, et al. (2016) Problematic computer game use as expression of Internet addiction and its association with self-rated health in the Lithuanian adolescent population. Medicina 52: 199–204.

85.

*Vadlin

Åslund

Nilsson

(2018) A longitudinal study of the individual- and group-level problematic gaming and associations with problem gambling among Swedish adolescents. Brain and Behavior 8: e00949.

86.

*Vadlin

Åslund

Hellström

, et al. (2016) Associations between problematic gaming and psychiatric symptoms among adolescents in two samples. Addictive Behaviors 61: 8–15.

87.

*Vadlin

Åslund

Rehn

, et al. (2015) Psychometric evaluation of the adolescent and parent versions of the Gaming Addiction Identification Test (GAIT). Scandinavian Journal of Psychology 56: 726–735.

88.

*Van den Eijnden

Koning

Doornwaard

, et al. (2018) The impact of heavy and disordered use of games and social media on adolescents’ psychological, social, and school functioning. Journal of Behavioral Addictions 7: 697–706.

89.

*Van Rooij

Kuss

Griffiths

, et al. (2014) The (co-) occurrence of problematic video gaming, substance use, and psychosocial problems in adolescents. Journal of Behavioral Addictions 3: 157–165.

90.

Van Rooij

Schoenmakers

Van

Eijnden

, et al. (2010) Compulsive internet use: The role of online gaming and other internet applications. Journal of Adolescent Health 47: 51–57.

91.

*Van Rooij

Schoenmakers

Vermulst

, et al. (2011) Online video game addiction: Identification of addicted adolescent gamers. Addiction 106: 205–212.

92.

*Vink

Van Beijsterveldt

Huppertz

, et al. (2016) Heritability of compulsive Internet use in adolescents. Addiction Biology 21: 460–468.

93.

*Wang

Chan

Mak

, et al. (2014) Prevalence and correlates of video and Internet gaming addiction among Hong Kong adolescents: A pilot study. The Scientific World Journal 2014: 874648.

94.

*Wang

Chan

, et al. (2015) Exploring personality characteristics of Chinese adolescents with internet-related addictive behaviors: Trait differences for gaming addiction and social networking addiction. Addictive Behaviors 42: 32–35.

95.

*Wartberg

Kriston

Thomasius

(2017) The prevalence and psychosocial correlates of internet gaming disorder: Analysis in a nationally representative sample of 12-to 25-year-olds. Deutsches Ärzteblatt International 114: 419.

96.

*Wartberg

Kriston

Kammerl

, et al. (2015) Prevalence of pathological internet use in a representative German sample of adolescents: Results of a latent profile analysis. Psychopathology 48: 25–30.

97.

*Wartberg

Kriston

Zieglmeier

, et al. (2019) A longitudinal study on psychosocial causes and consequences of Internet gaming disorder in adolescence. Psychological Medicine 49: 287–294.

98.

Wegmann

Oberst

Stodt

, et al. (2017) Online-specific fear of missing out and Internet-use expectancies contribute to symptoms of Internet-communication disorder. Addictive Behaviors Reports 5: 33–42.

99.

*Wittek

Finserås

Pallesen

, et al. (2016) Prevalence and predictors of video game addiction: A study based on a national representative sample of gamers. International Journal of Mental Health and Addiction 14: 672–686.

100.

Wölfling

Müller

Dreier

, et al. (2019) Efficacy of short-term treatment of internet and computer game addiction: A randomized clinical trial. JAMA Psychiatry 76: 1018–1025.

101.

*Yu

Cho

(2016) Prevalence of internet gaming disorder among Korean adolescents and associations with non-psychotic psychological symptoms, and physical aggression. American Journal of Health Behavior 40: 705–716.

102.

Zajac

Ginley

Chang

, et al. (2017) Treatments for Internet gaming disorder and Internet addiction: A systematic review. Psychology of Addictive Behaviors 31: 979–994.

Global prevalence of gaming disorder: A systematic review and meta-analysis

Abstract

Background:

Method:

Results:

Conclusion:

Keywords

Introduction

The present review

Method

Data sources

Data extraction

Data analysis

Results

Study characteristics

Data synthesis

Subgroup analyses

Sampling

Assessment

Region

Year

Moderator analyses

Gender differences in GD prevalence

Publication bias

Discussion

Conclusion

Footnotes

Author Contributions

Declaration of Conflicting Interests

Funding

ORCID iDs

References