Treatments for Internet Gaming Disorder and Internet Addiction: A Systematic Review

Abstract

Problems related to excessive use of the Internet and video games have recently captured the interests of both researchers and clinicians. The goals of this review are to summarize the literature on treatment effectiveness for these problems and to determine whether any treatments meet the minimum requirement of an evidence-based treatment as defined by (Chambless et al. 1998). Studies of treatments for Internet gaming disorder (IGD) and Internet addiction were examined separately, as past studies have linked IGD to more severe outcomes. The systematic review identified 26 studies meeting predefined criteria; 13 focused on treatments for IGD and 13 on Internet addiction. The results highlighted a paucity of well-designed treatment outcome studies and limited evidence for the effectiveness of any treatment modality. Studies were limited by methodological flaws, including small sample sizes, lack of control groups, and little information on treatment adherence, among other problems. In addition, the field is beset by a lack of consistent definitions of and established instruments to measure IGD and Internet addiction. The results of this review highlight the need for additional work in the area of treatment development and evaluation for IGD and Internet addiction. Attention to methodological concerns identified within this review should improve subsequent research related to treating these conditions, and ultimately outcomes of patients suffering from them.

Researchers and the public have been increasingly interested in problems arising from excessive use of technology, including Internet use more generally and video gaming particularly. Currently, debate exists over whether these problems should be categorized as mental disorders. In the case of excessive video game use, Internet gaming disorder (IGD) was included in the most recent version of the Diagnostic and Statistical Manual of Mental Disorders, 5^th Edition (DSM-5; American Psychiatric Association [APA], 2013) as a condition warranting further study. The nine proposed symptoms for IGD are similar to criteria for substance use and gambling disorders and include preoccupation, withdrawal, tolerance, inability to reduce or stop despite a desire to do so, giving up other activities in favor of gaming, continuing to game despite significant problems, deception or covering up of amount of gaming, using gaming to escape adverse moods, and risking or losing relationships or opportunities (e.g., educational, career) due to gaming.

The APA’s DSM-5 Substance Use Disorders workgroup found less evidence to support the inclusion of Internet addiction more generally and cautions that, until more systematic data are collected, gaming and other excessive uses of the Internet should be considered separately (Petry et al., 2014). To best understand a specific condition, limiting heterogeneity is expedient, especially in the early stages. The Internet can be used for many purposes, but an increasing number of studies demonstrate that video gaming is a distinct form of excessive Internet use and associated with unique harms (e.g., Siomos, Dafouli, Braimiotis, Mouzas, & Angelopoulos, 2008; van Rooij, Schoenmakers, van de Eijnden, & van de Mheen, 2010). Although many studies confound multiple forms of Internet use and problems (Ko, Yen, Yen, Lin, & Yang, 2007; Li, Zhang, Lu, Zhang, & Wang, 2014), some excessive uses of the Internet appear to represent different etiologies and expressions. For example, problematic gambling using the Internet is likely better represented as a gambling disorder than an Internet addiction, which may include uses ranging from excessive social networking to viewing pornography. Further, IGD and more globally defined Internet addiction are associated with different risk factors; most notably, male gender is a much stronger risk factor for IGD than it is for Internet addiction more generally (Király et al., 2014). Importantly, in studies that compare different types of online behaviors, gaming is consistently the activity most strongly associated with compulsive or problematic use (Kuss, Griffiths, Karila, & Billieux, 2014; van Rooij et al., 2010; Whang, Lee, & Chang, 2003; Siomos et al., 2008; Yen, Ko, Yen, Wu, & Yang, 2007). Thus, the literature suggests that IGD and Internet addiction may represent distinct problems and, in turn, may require different treatment approaches.

One major controversy surrounding IGD and Internet addiction is whether these problems are severe enough to warrant classification as mental disorders. However, there is increasing evidence that excessive gaming can be associated with substantial impairment. For example, a systematic review linked IGD to a wide array of negative outcomes, including poor school achievement, oppositional behavior, suicidality, and sleep abnormalities, among others (Kuss & Griffiths, 2012). On the extreme end, there have been descriptions of game-induced seizures (Chuang, 2006) and media reports of game-induced deaths (Spragg, 2015). IGD has also been linked to depression (e.g., Desai, Krishnan-Sarin, Cavallo, & Potenza, 2010), social difficulties (e.g., Lo, Wang, & Fang, 2005), attention deficit hyperactivity disorder (ADHD; e.g., Swing, Gentile, Anderson, & Walsh, 2010), and substance abuse (e.g., van Rooij et al., 2014).

Similar to IGD, Internet addiction has been associated with multiple comorbidities, including depression, anxiety, obsessive-compulsive symptoms, and aggression (see Carli et al., 2013 for a review) as well as negative outcomes, including problems at school or work, sleep deprivation, family conflict, and social withdrawal (see Kuss et al., 2014 for a review). However, many of the studies included in these reviews confound IGD with Internet addiction. Given the tendency for gaming problems to be related to the highest risk for adverse effects, it is quite possible that problems with gaming account for much of the relation between Internet addiction and problematic outcomes.

Unfortunately, the scientific literature on IGD and Internet addiction has been hindered by methodological problems. Because neither are established conditions, studies applied inconsistent definitions and measurement, which affected results from epidemiological as well as treatment studies. For example, several large population studies reported that the prevalence of IGD is less than 2% of the population (e.g., Rehbein, Kliem, Baier, Mößle, & Petry, 2015; Rehbein, Kleimann, & Mössle, 2010; van Rooij, Schoenmakers, Vermulst, van den Eijnden, & van de Mheen, 2011), whereas others have been as high as 8–10% (e.g., Choo et al., 2010; Gentile, 2009). The studies that used definitions and assessments of IGD that most closely align with the DSM-5 proposed symptoms tended to find prevalence rates at the lower end of this spectrum and also point to similar risk and protective factors.

Estimated rates of Internet addiction range from 0.8 to 11.8% in Western samples (e.g., Durkee et al., 2012; Johansson & Gotestam, 2004; Morahan-Martin & Schumacher, 2000; Poli & Agrimi, 2012) and well over 10% in some Asian samples (e.g., Li et al. 2014; Mak et al., 2014; Shek & Yu, 2012). Some reports even indicate rates of “smartphone addiction” or social media addiction to exceed 30% (e.g., Andreassen, 2015; Gutiérrez, de Fonseca, & Rubio, 2016).

Regardless of the existence of a recognized mental disorder or its specific manifestations, individuals are seeking professional help for these problems, and in some countries, entire psychiatric centers are dedicated to treating gaming and Internet addictions. In addition, researchers have evaluated various treatments for IGD and Internet addiction. Reviews and theoretical papers have been published (Griffiths & Meredith, 2009; Huang, Li, & Tao, 2010; Jorgenson, Hsiao, & Yen, 2016; King & Delfabrro, 2014; King, Delfabbro, & Griffiths, 2012; King et al., 2011; Kuss & Lopez-Fernandez, 2016; Przepiorka, Blachnio, Miziak, & Czuczwar, 2014) as well as one meta-analysis of treatments for Internet addiction (Winkler, Dörsing, Rief, Shen, & Glombiewski, 2013). Winkler et al. (2013) concluded that “effect sizes were high, robust, and maintained over follow-up” for both psychological and pharmacological treatments for Internet addiction (both gaming and non-gaming Internet activities), although they caution that these results should be considered preliminary. This study made an important contribution to the field by presenting pooled effect sizes across studies, but the inclusion of studies without control groups (i.e., pretest-posttest studies) in the calculation of effect sizes likely led to an overestimate of the effectiveness of these treatments. Further, past reviews have not distinguished between treatment studies for IGD and the more broad classification of Internet addiction, despite a now growing literature indicating important differences in these problems. Many of the reviews are also limited by the relatively small number of studies, especially randomized controlled trials, available at the time of their publication. The Kuss and Lopez-Fernandez (2016) and Jorgenson et al. (2016) reviews are recent; however, unlike the current review, Kuss and Fernandez (2016) do not differentiate between IGD and Internet addiction and Jorgenson et al. (2016) review a broader literature on etiology, prevalence rates, and risk factors for Internet addiction without providing a systematic review of treatment research.

The current systematic review is unique from earlier ones and has several aims. First, it provides an update on the state of the science, as many treatment studies have been published recently. Second, compared to past reviews, this analysis sets a more rigorous standard for inclusion of trials, with less emphasis on very small scale reports that are not powered to detect differences between interventions, and a greater emphasis on randomized trials. It still describes pre-post evaluations of interventions without control groups, but with the consideration that changes over time and with any intervention are likely in the natural course of these conditions and their treatment. Third, IGD studies are considered separately from those of Internet addiction more generally because these conditions differ on important features, including their status in the DSM-5 and likely the ICD-11 (see Aarseth et al., 2016; World Health Organization, 2016), severity, prevalence, and risk factors, which in turn may influence treatment approaches and clinical responses to them. Finally, past reviews have not included an analysis of whether any of the treatments evaluated for IGD or Internet addiction should be considered evidence-based treatments using established criteria. The current review evaluates whether any treatments meet the criteria for Well-Established or Probably Efficacious outlined in Chambless et al. (1998)’s guidelines for the establishment of evidence-based treatments.

Method

Inclusion and Exclusion Criteria

We first specified the scope and limits of the review. Inclusion criteria were that studies: 1) evaluate a treatment for IGD and/or Internet addiction; 2) use a design that is either multiarmed (randomized or non-randomized) or pretest-posttest; 3) include at least 10 participants per group to exclude very small pilot feasibility studies or single case designs; and 4) include an outcome measure related to severity of problems or duration of gaming or Internet use behaviors. We did not specify a start date due to the relatively recent emergence of research on Internet and gaming-related problems. The end date of the review was set at September 28, 2016, when the database searches were conducted. Studies were excluded if they: 1) focused on prevention rather than treatment of gaming or Internet addiction; 2) were review or theoretical papers; 3) focused on treatments for online gambling or use of online pornography exclusively or as the primary presenting problem; or 4) were not available in English.

Search Strategy

We searched PudMed and PsychInfo using the following combination of search terms: [‘Internet’ OR ‘gaming’ OR ‘video game’ OR ‘online’ OR ‘Facebook’ OR ‘social media’ OR ‘smartphone’] AND [‘addiction’ OR ‘pathological’ OR ‘excessive’ OR ‘problem’ OR ‘disorder’] AND [‘treatment’ OR ‘intervention’]. Secondary reference searching was conducted on all included studies. Finally, the reference lists of several reviews of similar topics were hand searched for relevant studies (Jorgenson et al., 2016; King et al., 2011; King & Delfabbro, 2014; Kuss & Lopez-Fernandez, 2016; Przepiorka et al., 2014; Winkler et al., 2013).

Screening Abstracts

Titles, abstracts, citation information, and descriptor terms of citations identified through the search strategy were screened in a two-step process. First, a study team member screened records individually to remove clearly non-relevant records. Full text articles were obtained for all records that remained after the initial review. Second, two authors screened records independently and compared results. All discrepancies were resolved through consensus and, when needed, a third reviewer.

Data Extraction and Management

For each included study, data were extracted by a trained coder and cross-checked by a second coder, with consensus or a third coder addressing differences. The following information was collected from each study: type of treatment, sample size, mean age of the sample and standard deviation (or range when mean was not available), study design, nature of the comparison groups (when applicable), method of diagnosing IGD or Internet addiction related to inclusion criteria, primary outcome variables related to IGD or Internet addiction, and study results. Study results were recorded for the primary outcome variables related to IGD or Internet addiction (e.g., severity of addiction, time spent online or gaming). When a study included follow-up assessments past the immediate post-treatment assessment, results from both the post-treatment assessment and the longest follow-up were recorded. Meta-analysis was not conducted due to the notable heterogeneity in study design, treatment modality, method of diagnosing IGD and Internet addiction, and methods of assessing primary outcome measures.

Criteria for Evidence-Based Treatments

The criteria identified by Chambless et al. (1998) for the identification of evidence-based treatments was used to evaluate the overall status of the scientific literature on treatments for IGD and Internet addiction. These criteria are used to categorize treatments into Well-Established, Probably Efficacious, and Experimental, and are detailed in Table 1.

Table 1.

Criteria for identification of evidence-based treatments described in Chambless et al. (1998)

Level 1: Well-established Treatments I. Two or more well-designed between-group experiments that demonstrate that the treatment is:  A. More effective than a pill or therapy placebo or another treatment  B. As effective as an already established treatment, using adequate sample sizes OR II. A large series of single-case design experiments that demonstrate efficacy and:  A. Are well designed experiments  B. Compare the treatment to an established treatment III. A treatment manual or the equivalent must be used IV. Sample characteristics must be provided V. At least two different investigators or teams must demonstrate these treatment effects

Level 2: Probably Efficacious Treatments I. At least two experiments demonstrating the treatment to be more effective than a wait-list control group OR II. At least one experiment that meets all well-established criteria except V (replicated by two different investigators or teams) OR III. A series of single-case design experiments that meet well-established treatment criteria but is a small rather than large series of experiments

Level 3: Experimental Treatments Not enough evidence that the treatment meets criteria for Well-established or Probably Efficacious

Results

The initial database search yielded 11,031 records; 2 additional records were identified through other means (see Figure 1). Once duplicates were removed, 9,265 records underwent initial screening, and 45 were retained for screening in duplicate and underwent full-text review. Of those, 6 were not available in English, 5 were not multi-armed or pretest-posttest designs, 2 did not evaluate a treatment specifically for either IGD or Internet addiction, 4 did not include outcome measures related to IGD or Internet addiction severity, and 2 presented results from a sample already presented in another paper in this review. The remaining 26 studies were deemed eligible for inclusion: 13 focused on IGD and 13 on Internet addiction more generally.

Figure 1.

Disposition of study records

Internet Gaming Disorder Studies

Table 2 describes the included IGD studies. Five of the 13 studies evaluated medication, 4 evaluated cognitive-behavioral therapy (CBT) approaches, and 4 evaluated other types of interventions (i.e., a speaking and writing course, family therapy, eclectic psychotherapy, self-discovery camp). Among the non-medication interventions, there was substantial variability in the number of sessions, ranging from as few as 5 to as many as 21. Two of the 4 CBT studies focused on young adults, while the other 2 recruited adolescent samples. Studies evaluating other types of psychosocial interventions all recruited adolescent samples. Most of the medication trials focused on young adults and older adolescents, except for Han et al. (2009) who recruited children with a mean age of 9.3 years.

Table 2.

Studies included in the review of treatments for Internet Gaming Disorder

First author, year	Country	Treatments	N	Tx duration (sessions)	Age M(SD) or range	Study Design	Diagnostic Method for IGD Inclusion Criteria	Primary Outcome Variable(s)	Significant Results (follow-up timeframe)
		Medication
Han, 2010	South Korea	Bupropion	11	6wks	21.5±5.6	Pre-post*	YIAS >50 + gaming >30 hr/wk + impairment or distress	Weekly gaming (hrs); Gaming addiction (YIAS)	Decrease in gaming time and YIAS (post-tx)
Han, 2009	South Korea	Methyl-phenidate	62	8wks	9.3±2.2	Pre-post**	YIAS was used but all participants were included regardless of score	Weekly gaming (hrs); Gaming addiction (YIAS)	Decrease in gaming time and YIAS (post-tx)
Han, 2012b	South Korea	Bupropion+ education Placebo+ education	25 25	8wks 8wks	21.2±8.0 19.1±6.2	RCT***	YIAS >50 + gaming >30 hr/wk + impairment or distress	Weekly gaming (hrs); Gaming addiction (YIAS)	Greater reductions in gaming time and YIAS in the bupropion compared to placebo group (post-tx) Reductions and group differences maintained at follow-up (4 wks post-tx)
Park, 2016a	South Korea	Atomexetine Methylphenidate	42 44	12wks 12wks	17.1±1.0 16.9±1.6	RCT**	DSM-V criteria; no information on how this was assessed	Gaming addiction (YIAS)	No differences on YIAS between groups (post-tx). YIAS decreased in both groups but a test statistic was not presented.
Song, 2016	South Korea	Bupropion Escitalopram No Treatment Control	44 42 33	6wks 6wks 6wks	20.0±3.6 19.8±4.2 19.6±4.0	RCT†	DSM-V criteria; no information on how this was assessed	Gaming addiction (YIAS)	Decreased YIAS for active groups, but not control. Greater decrease for bupropion than escitalopram (post-tx).
		CBT-based Psychotherapy
Zhang, 2016b††	China	Craving behavioral intervention group No intervention control	23 17	6wks(6) N/A	21.9±1.8 22.0±1.9	Non-RCT	CIAS ≥67 + gaming >14hr/wk for ≥1yr; plays 1 of 4 most popular gamesˆ	Weekly gaming (hrs) Gaming addiction (CIAS)	Lower CIAS and less gaming in intervention group than controls (post-tx). No test statistic to compare pre-post change between groups.
Li et al., 2013	China	CBT group therapy Basic counseling	14 14	6wks(12) 6wks(12)	12–19 yrs	RCT	Gaming >30hr/wk + OGCAS >35 + IAS >3 + distress or maladaptive bx	Gaming addiction (YIAS)	YIAS decreased in both groups but no group differences (post-tx)
Kim, 2012	South Korea	CBT+ Bupropion Bupropion only	32 33	8wks(8) 8wks	16.2±1.4 15.9±1.6	RCT***	YIAS >50 + gaming >30hr/wk + maladaptive bx or distress	Weekly gaming time; Gaming addiction (YIAS)	Greater reduction in CBT+med than med only (post-tx) Group differences maintained at follow-up (4 wks post-tx)
Park, 2016b	South Korea	CBT group therapy Virtual Reality group therapy	12 12	4wks(8) 4wks(8)	24.2±3.2 23.6±2.7	RCT*	Gaming >30hr/wk + disruption of life + distress or maladaptive bx + YIAS >50	Gaming addiction (YIAS)	Both groups had reductions in YIAS but no between group differences (post-tx)
		Other Interventions
Han, 2012a	South Korea	Family therapy	15	3 wks(5)	14.2±1.5	Pre-post*	Gaming >4hrs/day and >30hrs/wk + YIAS >50 + impaired bx or distress	Weekly gaming (hrs); Gaming addiction (YIAS)	Decrease in gaming time and YIAS score (post-tx)
Palleson, 2015	Norway	Eclectic psychotherapy intervention (CBT, family, motivational interviewing, solution-focused)	12	(13)	15.7 (1.3)	Pre-post	GASA ≥3 on all items (self-report) or PVP score of 4 or 5 on all items (parent report)	Gaming addiction (patient-reported GASA and PVP, parent-reported PVP)	Decrease in parent-reported PVP but not patient reported GASA or PVP (post-tx)
Sakuma, 2017	Japan	Self-Discovery Camp	10	9 days	16.2±2.2	Pre-post	Satisfied Griffith’s 6 components of addictionˆˆ + met DSM-V IGD criteria through clinical interview	Daily gaming (hours) and weekly gaming (hours and days)	Decrease in hours/day and hours/week of gaming but not days/week (3 months post-tx)
Kim, 2013	South Korea	MMORPG speaking and writing course General education	27 32	8wks(21) 8wks(21)	17.4±0.6 17.5±0.6	RCT	Playing DF ≥4hr/day	Average daily gaming in the past month (min)	Both groups showed reductions in gaming but no difference between groups (post-tx)

Note. Bx = behavior; CBT = Cognitive Behavioral Therapy; CIAS = Chinese Internet Addiction Scale; DF = Dungeon & Fighter; GASA = Gaming Addiction Scale for Adolescents; IAS = Internet Addiction Scale; MMORPG = Massive Multiplayer Online Role-Playing Games; OGCAS = Online Game Cognitive Addiction Scale; PVP = Problem Video Game Playing Scale; RCT = randomized control trial; tx = treatment; YIAS = Young Internet Addiction Scale

^*a sample of healthy controls was recruited by not included in primary analyses;

^**all participants had a diagnosis of attention deficit hyperactivity disorder;

^***all participants had a diagnosis of major depressive disorder;

^†participants were randomly assigned to the two medication groups but the control group were participants who declined medication treatment (not randomly assigned).

^††similar results were presented in two other publications by the same authors; the paper that presented on the largest sample size was chosen for inclusion (Zhang, Ma, et al., 2016; Zhang, 2016a).

^ˆparticipants reported playing one of the following games as their primary online activity: 1) Cross Fire; 2) Defense of the Ancient version 1; 3) Defense of the Ancient version 2; or 4) World of Warcraft.

^ˆˆGriffith’s six components of addiction are salience, mood modification, tolerance, withdrawal, conflict, and relapse (Griffiths, 2005).

There was also substantial variability in the methodological rigor of the studies. Three of the 5 medication studies were RCTs, 3 of the 4 CBT studies were RCTs, and only 1 of the 4 studies evaluating other types of approaches was an RCT. Other studies were quasi-experimental or pretest-posttest designs.

The approach to assessing inclusion criteria related to IGD varied across studies. Eight studies used scores from at least one instrument for assessing symptoms, with the majority of these (5) using the Young Internet Addiction Scale (YIAS) to evaluate IGD. Some of these studies used a combination of scores on assessment instruments and other criteria (e.g., weekly gaming hours). Two studies did not provide enough information to determine how the gaming-related inclusion criteria were assessed. Of the remaining three studies, one provided general information about the inclusion criteria but did not clearly specify how it was assessed (Han et al., 2009; Sakuma et al., 2017) and one used number of gaming hours per day as the only gaming-related inclusion criteria (Kim, Kim, Shim, Im, & Shon, 2013). The third recruited a sample of children with attention deficit/hyperactivity disorder (ADHD) and did have inclusion criteria related to IGD (Han et al., 2009).

Pharmacotherapy

The medication trials for IGD examined psychotropic drugs typically used for treating depression or ADHD. Two studies with pretest-posttest designs found significant decreases in gaming time and IGD symptoms using a 6-week trial of bupropion in 11 young adults (Han, Hwang, & Renshaw, 2010) and an 8-week trial of methylphenidate in 62 children (Han et al., 2009). The latter recruited a sample with comorbid IGD and ADHD and found that improvements in IGD was positively correlated with improvements in ADHD symptoms.

The three RCTs of pharmacotherapies included 25 to 44 participants per study arm. Two studies found that bupropion, a drug commonly used for treating depression, was superior to both a no medication control group (Song et al., 2016) and a placebo condition (Han & Renshaw, 2012) in reducing IGD symptoms in young adult samples. One of these studies found that these effects were maintained at 4 weeks post-treatment (Han & Renshaw, 2012), and the other reported that escitalopram, another commonly prescribed antidepressant, was superior to no medication, but inferior to bupropion in reducing IGD (Song et al., 2016). The Han and Renshaw (2012) study recruited a sample that was comorbid for IGD and Major Depressive Disorder (MDD). They found bupropion to be superior to placebo in decreasing depression symptoms from pre-to post-treatment, but these effects did not persist at 4-week follow-up.

The third RCT compared two medications typically prescribed for ADHD and found that adolescents receiving either atomoxetine or methylphenidate showed reductions in IGD symptoms over a 12-week trial (Park, Lee, Sohn, & Han, 2016). All adolescents in this study had a diagnosis of ADHD, and there was an advantage for methylphenidate over atomoxetine in improving ADHD symptoms. The results of this study are difficult to interpret, as no placebo group was used.

Three of the five medication studies did not clearly specify how the IGD inclusion criteria were assessed. Two of these (Park, Lee, et al., 2016; Song et al., 2016) stated that DSM-5 criteria were used but did not provide information about how they were assessed. The third (Han et al., 2009) did not have inclusion criteria related to gaming, rendering cross-study comparisons speculative.

CBT-based psychotherapy

All four studies evaluating CBT-based psychotherapy employed two-group clinical trial designs (three randomized, one not randomized). The non-randomized clinical trial compared a group-based CBT intervention focusing on craving to a notreatment control group among 40 young adults (Zhang et al., 2016b). After 6 weeks, the group that received CBT reported significantly fewer weekly gaming hours and IGD symptoms compared to the no-intervention control group, but no test statistic was provided for the changes from pre-to post-treatment. Two other studies by this research group were identified during the literature search (Zhang et al., 2016a; Zhang, Ma, et al., 2016), but because their results were from the same sample, only the paper that presented the largest proportion of the sample was included in this review.

One of the randomized trials compared a group CBT treatment to basic supportive counseling with an adolescent sample (Li & Wang, 2013). After six weeks of twice weekly sessions, both conditions showed significant reductions in IGD symptoms, but there was not a significant difference between groups.

A second randomized trial compared bupropion plus 8 sessions of CBT to bupropion alone for adolescents with comorbid IGD and MDD (Kim, Han, Lee, & Renshaw, 2012). There were no statistically significant between-group differences in reductions in depression symptoms. The CBT group showed significantly greater reductions on time spent gaming and IGD symptoms compared to medication-only at post-treatment and this effect was maintained at 4-weeks post-treatment. However, there was no control for therapist time (i.e., CBT sessions were between 90 and 120 minutes weekly whereas medication management consisted of 10 minute weekly check-ins).

The most recent RCT compared 8 sessions of traditional group CBT to 8 sessions of virtual reality group therapy, a CBT-based approach that uses virtual reality technology to teach relaxation skills, simulate high-risk situations for gaming, and provide sound-assisted cognitive restructuring (Park, Kim, et al., 2016). Not only did the content and format of the treatments differ, but so did the duration of the weekly sessions, with the traditional groups lasting 2 hours and the virtual reality groups only 25 minutes. After 4 weeks, young adults in both groups showed reductions in IGD symptoms, but there were no significant differences between groups.

Other approaches

Of the four additional studies examining interventions for IGD, three employed pretest-posttest designs, and only one was an RCT. One of the pretest-posttest studies found that 5 sessions of family therapy over 3 weeks was related to significant decreases in gaming time and IGD symptoms in a sample of adolescents (Han, Kim, Lee, & Renshaw, 2012). This family intervention targeted family function and cohesion and encouraged families to engage in new shared activities unrelated to gaming. In another study, 13 sessions of eclectic psychotherapy that borrowed from CBT, family therapy, motivational interviewing, and solution-focused therapy was found to be related to significant decreases in parent-reported but not adolescent-reported IGD symptoms (Palleson, Lorvik, Bu, & Molde, 2015). Detailed information about the treatment protocol used in this study was not provided. Finally, adolescents attending a 9-day self-discovery camp showed significant decreases in gaming time at 3-month follow-up (Sakuma et al., 2017). The camp experience included 14 sessions of CBT, 8 sessions of ‘personal counseling’, 3 medical lectures, a workshop about gaming, engagement in positive non-gaming activities, and a prohibition against gaming devices during the 9-day stay. None of these three studies had a control or comparison condition.

The RCT evaluated a writing and speaking course using content borrowed from massive multiplayer online role-playing games (MMORPGs) compared to a general education course in a sample of 59 adolescents (Kim et al., 2013). Unlike the other studies, the inclusion criteria for gaming problems was based solely on number of hours/day playing a specific video game (i.e., ≥4 hours) rather than symptoms related to gaming. The authors hypothesized that harnessing adolescents’ interest in a specific MMORPG could engage them in tasks aimed at improving writing and speaking ability and could also decrease IGD symptoms. Adolescents in the treatment group participated in a series of tasks in which they wrote and then spoke about various aspects of the MMORPG. The control group followed the same procedures but the content was not related to gaming. Both groups showed decreases in average daily gaming after 8 weeks of intervention, and there was not a significant difference between groups.

Internet Addiction Studies

Table 3 describes the Internet addiction studies that met the inclusion but not exclusion criteria. Of these 13 studies, 1 evaluated a medication, 6 evaluated a CBT-based treatment, 3 evaluated family-based approaches, and 3 evaluated other types of approaches (i.e., an online program focused on healthy Internet use; reality therapy; daily journaling). Among the non-medication interventions, there was substantial variability in the number of sessions, ranging from 1 to 30. The Internet addiction studies tended to focus on older samples compared to the IGD studies. The medication trial recruited an adult sample with a mean age around 38 years. All but one of the CBT studies recruited adults, while the remaining study recruited adolescents with a mean age around 16 years. Studies of family-based approaches primarily recruited adolescents. Studies of other types of interventions focused on either adolescents or young adults.

Table 3.

Studies included in the review of treatments for Internet Addiction

First author, year	Country	Treatments	N	Tx duration (sessions)	Age M (SD) or range	Study Design	Diagnostic Method for Internet Addiction Inclusion Criteria	Primary Outcome Variable(s)	Significant Results (follow-up timeframe)
		Medication
Dell’Osso, 2008	USA	Open-Label Phase: Escitalopram Double-Blind Discontinuation: Escitalopram Placebo	19 7 7	10wks 9wks 9wks	38.5±12.0 37.5±11.5	Ph 1: Pre-post Ph 2: RCT	IC-IUD defined as uncontrollable, distressing, time consuming, and resulting in difficulties. Assessment method unclear.	Nonessential Internet use (hrs/wk); Obsessive/compulsive Internet use (IC-IUD-YBOCS)	Ph 1: decrease in hrs/wk and IC-IUD-YBOCS Ph 2: no group differences (post-tx)
		CBT-based therapy
Santos, 2016	Brazil	CBT (for Internet use) + pharmacotherapy (for anxiety)	39	10wks (10)	28.6(5.9)	Pre-post**	IAT > 50	Addiction severity (IAT)	Reduction in IAT (post-tx)
Wölfling, 2014	Germany	Short-term Treatment for Internet and Computer Game Addiction	42	Group (15) + Individual (8)	26.1(6.6)	Pre-post	AICA-S ≥ 7 + AICA-C ≥13	Addiction severity (AICA-S); weekend hrs online	Reduction in AICA-S and hrs spent online (post-tx)
Young, 2013	USA	CBT-IA	128	12wks (12)	22–56	Pre-post	Meets 5 or more criteria on IAT or 4 criteria but exhibited serious pxs related to Internet use	Addiction severity (IADQ)	Reduction in IADQ (post-tx) Reduction maintained (6 months post-tx)
Young, 2007	USA	CBT	114	(12)	Female:38 Male:46	Pre-post	Used IAT but cut-off was not stated.	Abstinence from problematic Internet applications	Increased abstinence but no test stat presented (post-tx) Results maintained but no test stat presented (6 months post-tx)
Du, 2010	China	Multimodal school based group CBT No treatment control	32 24	(8) N/A	15.4±1.7 16.6±1.2	RCT	Endorsed the first 5 items of the BDQ and at least 1 of the next 3 items.	Internet overuse (IOSS)	Overall reduction in IOSS but no group differences (post-tx) Same results at 6 months post-tx
Zhu, 2012	China	Electroacupuncture (EA) CBT EA + CBT	39 36 37	(20) (10) (30)	21.0±2.0 22.5±2.1 22.5±2.3	RCT	Met criteria for at least 3 of 7 behaviors related to Internet addiction. Assessment method unclear.	Addiction severity (IAT)	IAT reduction for all groups. Combined tx had lower IAT scores than CBT or EA alone. EA alone had lower IAT than CBT alone.
		Family Therapy
Shek, 2009	China	Multilevel therapy intervention (includes motivational interviewing and family therapy)	59	15–19 months	11–15 (n = 29); 16–18 (n = 27); 18+ (n = 3)	Pre-post	Endorsed ≥4 items on CIA-Young10 or ≥5 items on CIA-Young8 or ≥3 items on CIA-Young7 or ≥3 items on CIA-Goldberg	Addiction severity – 4 versions of the CIA (Young10, Young8, Young7, Goldberg)	Reductions in CIA-Young, CIA-Young8, CIA-Young7, and CIA-Goldberg (post-tx)
Liu, 2015	China	Multi-family group therapy WL	21 25	2wks(6) N/A	15.0(1.7) 15.7(1.2)	Non-RCT	APIUS >3	Addiction severity (APIUS); Internet use (hrs/wk)	Greater reductions in APIUS and Internet use in tx vs WL (post-tx) Results maintained at 3 months post-tx
Zhong, 2011	China	Group-based family intervention Group-based treatment as usual	28 29	(14) NR	17.9±3 18.3±2	RCT	Internet addiction criteria proposed by Tao et al. (2009) ††	Addiction severity (OCS)	Overall reductions in OCS but no statistical comparison between groups (post-tx) Lower OCS for family tx than control (3 months post-tx)
		Other Approaches
Lee, 2016	South Korea	Home-based daily journaling intervention	46	2wks	13.2±0.8	Pre-post*	On SKYPSA, scored >42 on total or >14 on adjustment to daily life or >12 on withdrawal or >13 on tolerance.	Addiction severity (SKYPSA)	Decrease in SKYPSA (post-tx)
Su, 2011	China	Noninteractive HOSC HOSC in laboratory HOSC in natural Environment WL	16 17 16 16	(1) (1) (1) N/A	Undergrad and grad students; no ages provided	RCT	YDQ ≥3 and spends >14 hrs/wk online	Addiction severity (YDQ); online hours/week	Greater reductions in hrs/wk and YDQ in tx groups than WL but no differences between the 3 tx groups (1 month post-tx)
Kim, 2008	South Korea	Reality therapy (group) No treatment control	13 12	5wks(10) 5wks	24.2	RCT	NR	Addiction severity (K-IAS)	Greater reduction in K-IAS in tx vs control group (post-tx)

Note. AICA-C = Scale for the Assessment of Internet and Computer Game Addiction, Clinician Rated; AICA-S = Scale for the Assessment of Internet and Computer Game Addiction, Self-Report; APIUS = Adolescent Pathological Internet Use Scale; BDQ = Beard’s Diagnostic Questionnaire for Internet Addiction; CBT = Cognitive Behavioral Therapy; CIA = Chinese Internet Addiction Scale; HOSC = Healthy Online Self-Helping Center; IA = Internet Addiction; IADQ = Internet Addiction Diagnostic Questionnaire; IAT = Young Internet Addiction Test; IC-IUD-YBOCS = Impulsive-compulsive Internet usage disorder Yale-Brown Obsessive Compulsive Scale; IOSS = Internet Overuse Self-Rating Scale; K-IAS = Korean-Internet Addiction Scale; N/A = not applicable; NR = not reported; OCS = Online Cognition Scale; Ph = Phase; SKYPSA = Scale of Korean Youth Proneness to Smartphone Addiction; tx = treatment; YDS= Young Diagnostic Questionnaire; WL = waitlist.

^*a sample of healthy controls was recruited by not included in primary analyses;

^**all participants were diagnosed with an anxiety disorder.

^†Behaviors were: tolerance, withdrawal, increasing frequency/using more than the planned amount, unable to cut-back, spent excessive amounts of time on the Internet or engaging in behaviors to facilitate Internet use; neglecting important responsibilities because of Internet use; continued use despite consequences.

^††Preoccupation with the Internet and withdrawal (both present) plus one of more of the following: tolerance; unsuccessful attempts and/or persistent desire to reduce Internet use; continued excessive use despite physical or psychological problems; loss of interest in previous hobbies/interests; use of the Internet to escape a dysphoric mood. Must have clinically significant functional impairment.

Similar to the IGD studies, there was substantial variability in methodological rigor; the one medication trial included an RCT component, 2 of the 6 CBT studies were RCTs, 1 of the 3 family therapy studies was an RCT, and 2 of the 3 studies evaluating other approaches were RCTs. The other 7 studies used pretest-posttest or non-randomized trial designs.

Studies also varied in methods for assessing inclusion criteria related to Internet addiction. Seven used assessment instruments (with or without other criteria) and provided information on the cut-offs used. Three studies provided a general description of the inclusion criteria but did not provide details about how they were assessed (Dell’Osso et al., 2008; Zhu et al., 2012; Zhong et al., 2011). Two used assessment instruments but did not provide the cut-offs used (Young, 2007; Liu et al., 2008). One study did not provide any information about inclusion criteria related to Internet addiction (Kim, 2008).

Pharmacotherapy

Unlike the IGD field, in which medication trials were relatively common, our search uncovered only one medication trial for Internet addiction. This study consisted of a 10-week open-label trial of escitalopram followed by a 9-week double-blind discontinuation phase comparing escitalopram to placebo (Dell’Osso et al., 2008). There were reductions in non-essential Internet use and compulsions for Internet use during the open trial phase, but escitalopram was not significantly better than placebo in the double-blind phase.

CBT-based psychotherapy

Four studies used pretest-posttest designs to evaluate CBT for Internet addiction, all in adult samples. There was substantial variability in the treatment approaches, which included combined CBT for Internet use and pharmacotherapy for anxiety (Santos et al., 2016); Short-term Treatment for Internet and Computer Game Addiction, which incorporates 15 sessions of group and 8 sessions of individual CBT (Wölfling, Beutel, Dreier, & Müller, 2014); and individual CBT (Young, 2007, 2013). All four studies found significant improvements in symptoms related to Internet addiction between pre- and post-treatment, and the one study that also evaluated amount of time spent on the Internet found improvements on that outcome (Wölfling et al., 2014). The two studies of individual CBT also included follow-ups 6-months post-treatment and found that gains made during treatment were maintained over time (Young, 2006, 2013). The Santos et al. (2016) study recruited a sample of adults with comorbid Internet addiction and anxiety and found a significant reduction in anxiety symptoms at follow-up. None of these studies employed a comparison condition.

Two of the studies evaluating CBT approaches were RCTs. Du, Jiang, and Vance (2010) evaluated a multimodal school-based group CBT intervention that included group parent training and teacher psychoeducation compared with a no treatment control condition. At post-treatment and 6 months after treatment completion, adolescents in both conditions showed reductions in Internet use, but there was not a significant difference between groups.

The second RCT was a three-group design comparing electroacupuncture, CBT, and a combination of electroacupuncture and CBT in a sample of 112 young adults (Zhu et al., 2012). All three groups showed significant reductions in Internet addiction symptoms over time. Participants in the combined treatment condition had superior outcomes compared to the single treatment groups, and the electroacupuncture alone group’s results were superior to the CBT alone group. However, the lack of a control group and different intensities of treatments across groups present challenges to understanding these between-group differences. The combined group received 30 contacts with the treatment providers (20 sessions of electroacupuncture and 10 sessions of CBT) while the electroacupuncture alone group received 20 sessions and the CBT alone group received only 10 sessions, rendering modality comparisons confounded by intensity differences.

Family therapy

Three studies evaluated a family-based intervention for Internet addiction. In a pretest-posttest study, a sample of 59 adolescents receiving a multilevel intervention that included individual motivational interviewing, behavioral contracting, development of a career plan, and family therapy showed significant reductions in addiction severity after 15–19 months of treatment (Shek, Tang, & Lo, 2009). The family therapy component focused on resolving conflicts, improving communication, reframing symptoms of addiction, and discussing stages of change.

A second study used a non-randomized design to compare a 6-session multi-family group therapy relative to a waitlist control condition (Liu et al., 2008). Families were assigned to the treatment group if their schedule allowed them to attend the therapy sessions or to waitlist if they were unable to attend. Family therapy focused on improving parent-child communication skills about Internet addiction as well as more general communication skills, the association between unmet needs of the adolescent and their use of the Internet, and establishing healthy expectations for the family system. Compared to waitlist, adolescents in the treatment condition had significantly greater reductions in time spent on the Internet and addiction severity. A difference in time spent on the Internet and addiction severity was again observed at a 3-month follow-up, but families were self-selected into the two conditions.

Finally, one RCT compared a 14 session group-based family intervention to treatment as usual (also group-based) in a sample of 57 families (Zhong et al., 2011). Seven sessions were for adolescents only, 4 for parents only, and 3 included both adolescents and parents. Adolescent sessions focused on pros/cons of Internet use and enhancing communication skills using dream interpretation, sandplay, psychodrama, and role play. Parent sessions focused on communication skills and how to deal with the adolescent’s Internet use. Parent-adolescent sessions focused on communication skills and improving family functioning. They found significant reductions in addiction severity in both groups post-treatment, but a statistical comparison between groups was not presented. Adolescents in the family intervention group had significantly fewer addiction symptoms at the 3-month follow-up compared to the control condition. The control group received treatment as usual that included military training and group therapy focused on addictions; however, the number of sessions of group therapy was not clearly stated. It is possible that the treatment effects were due to the frequency or intensity, rather than the content, of the family-based treatment.

Other approaches

One of the three studies of other approaches used a pretest-posttest design, and two were RCTs. In a pre-post study, Lee, Seo, and Choi (2016) found that a sample of 46 adolescents who completed a home-based daily journaling intervention showed significant decreases over time in Internet addiction symptoms. Adolescents tracked their smartphone use daily for two weeks (amount of time spent, content, location where used, and reflective self-evaluations). They were asked to discuss their problems related to smartphone use with their parents and were encouraged to modify their behaviors.

One RCT (Kim, 2008) evaluated group-based reality therapy compared to a no treatment control condition in a sample of 25 young adults. Reality therapy is rooted in choice theory and uses the WDEP model (W=wants, D=direction and doing, E=evaluation, P=planning and commitment; Wubbolding, 2000) to help patients to control their behavior and make new choices related to Internet use. The content of the group sessions was not described in detail. Compared to the control group, participants who received reality therapy showed significantly greater reductions in Internet addiction severity following 5 weeks of treatment.

Su, Fang, Miller, and Wang (2011) evaluated an interactive self-help website based on principles of motivational interviewing. Undergraduate and graduate students (n = 65) were randomly assigned to 1 of 4 groups: 1) using the website in a lab setting; 2) using the website in their natural environment; 3) a non-interactive online program of the same content; or 4) a waitlist condition. The website had four modules: an introduction; a self-assessment of Internet use that includes a decisional balance exercise; setting goals related to change using a readiness to change ruler; and a CBT-based module focused on methods to change Internet behavior. All three of the treatment groups showed significantly greater reductions in the amount of time spent on the Internet and Internet addiction severity compared to the waitlist condition, with no significant differences between any of the treatment groups.

Evaluation of Evidence-Based Treatments

Studies in the IGD and Internet addiction categories were evaluated separately to determine if any treatment modality met criteria for Well-Established or Probably Efficacious treatments using the criteria described by Chambless et al. (1998). In order to be considered Well-Established, two or more well-designed between-group experiments must demonstrate that the treatment is either more effective than a pill, therapy placebo, or another treatment or as effective as an already established treatment. These treatment effects must be demonstrated by two different investigators or research teams. To be considered Probably Efficacious, at least two experiments must demonstrate that the treatment is more effective than a wait-list control group or at least one experiment must meet the criteria for the Well-Established group. Pre-post and non-randomized designs do not count towards these criteria. In the case of IGD, the studies that were RCTs (Han & Renshaw, 2012; Kim et al., 2012; Kim et al., 2013; Li et al., 2013; Park, Kim, et al., 2008; Park, Lee, et al., 2016; Song et al., 2016) had relatively small samples, making them underpowered to detect all but large effect sizes, which are uncommon in psychotherapy studies. Many did not find between-group differences (Kim et al., 2013; Li et al., 2013; Park, Kim, et al., 2008; Park, Lee, et al., 2016). The same was true for the Internet addiction studies, with RCTs being underpowered and most finding no between-group differences Thus, for both the IGD and Internet addiction groups, none of the treatments met the minimum requirements for either Well-Established or Probably Efficacious, primarily due to lack of a rigorous design, small sample sizes, and lack of between group differences. Thus, all treatments are currently considered Experimental.

Discussion

The studies reviewed here cover a diverse array of treatment approaches for IGD and Internet addiction. All of them were published in the past 10 years, indicating a recent spike in interest, likely driven by clinical demand for effective treatments for these problems. Thus, despite controversies over the diagnostic legitimacy of IGD and Internet addiction, researchers and patients have shown interest in developing and accessing treatments for problems associated with these behaviors. However, even though progress has been made in recent years, it is still difficult to draw conclusions from this literature due to methodological limitations and a lack of a critical mass of studies on any specific treatment.

This review set stricter inclusion and exclusion criteria compared to past reviews on this topic and also conducted a more rigorous evaluation using the Chambless et al. (1998) criteria to determine whether any treatments can be considered evidence-based. As a result, past reviews resulted in a more positive evaluation of the extant treatments (e.g., Kuss & Lopez-Fernandez, 2016; Winkler et al., 2013), whereas the current review found that no treatments for IGD or Internet addiction yet meet the standard for a Well-Established or Probably Efficacious treatment described by Chambless et al. (1998) and, thus, all are currently considered Experimental.

However, similar to recent reviews (Kuss & Lopez-Fernandez, 2016; Winkler et al., 2013), the results of the current analysis also highlight the paucity of well-designed treatment research studies for IGD and Internet addiction. Despite these limitations, this review can inform areas for future work. In terms of medication treatments, clinical trials suggest that the antidepressant bupropion may hold some promise for the treatment of IGD, but larger RCTs involving double blinded placebo controls are needed before definitive conclusions can be drawn. Escitalopram, methylphenidate, and atomoxetine have also been applied to treat IGD, but they have not yet been compared to a placebo condition. In the case of medication treatments for Internet addiction, very little is known. There was a single study of escitalopram, but, given the study’s limitations and the lack of other medication trials, much more work will be needed to determine whether medication exerts effects beyond those that occur naturally over time or in response to help seeking in general.

The evidence for CBT-based interventions in treating both IGD and Internet addiction is also limited. In the case of IGD, there were four clinical trials but only two had active treatment controls and all had relatively small sample sizes. There were six studies evaluating CBT for Internet addiction, indicating substantial interest from researchers in using CBT to treat this condition. However, four of them did not have control groups. Little is known about the natural progression of these conditions, and some proportion is likely to get better without treatment. Therefore, it is difficult to draw conclusions about the efficacy of treatments for these problems without an appropriate control group. Of the other two studies on Internet addiction, one compared three active treatments (Zhu et al., 2012) while the other used a no treatment control and found no advantage of CBT (Du et al. 2010). Thus, although results from most published pre-post studies showed improvements over time, there has yet to be a well-designed RCT of CBT that supports its effectiveness.

Only one study evaluated a family-based treatment for IGD. This was an unexpected finding. Treatments for IGD have been most commonly evaluated with adolescent samples, and adolescent IGD shares many similar characteristics and risk factors with adolescent substance abuse problems (e.g., Park, Kim, & Cho, 2008; Yen, Yen, Chen, Chen, & Ko, 2007). Thus, it is surprising that more researchers have not modeled IGD treatments on evidence-based treatments for adolescent substance abuse, the most effective of which are family-based (Tanner-Smith, Wilson, & Lipsey, 2013; Waldron & Turner 2008). The three studies on family-based treatments for Internet addiction were all conducted with adolescent samples. Two were randomized trials: one that compared family treatment to waitlist (Liu et al., 2015) and another to treatment as usual (Zhong et al., 2011). Both had relatively small samples but still found an advantage for the treatment with a family component. However, the Liu et al (2015) study did not randomize participants to group; rather, participants were assigned to the waitlist if they were currently unavailable to participate in the treatment group. Thus, it is likely that the two groups differed in ways that influenced the results. Larger well-designed randomized trials with longer follow-ups are needed to confirm these effects, and it is imperative that these studies include controls for therapist time and expectancy effects.

Finally, there was an eclectic group of treatment studies for both IGD and Internet addiction that were not captured in the categories described above. All of these studies had methodological limitations, including lack of control groups and small sample sizes, and a few failed to find significant treatment effects.

Methodological Limitations

This review highlights the lack of methodological rigor in the treatment research for IGD and Internet addiction. As noted above, many studies utilized small sample sizes and/or did not compare the active treatment to a control group that allows for meaningful conclusions about treatment efficacy. Pretest-posttest studies have an important role in establishing feasibility, safety, and initial effectiveness of treatment approaches. At this point, this study design has been utilized to evaluate many of the treatment approaches summarized in this review, and the field is ready to move on to well-controlled studies to make stronger conclusions about treatment efficacy. Researchers and clinicians should be cognizant that most any intervention is likely to lead to some reductions in symptoms over time in persons seeking treatment (e.g., Dew & Bickman, 2005; Meyer et al., 2002), and treatment expectancy and intensity effects should be considered. In addition, it is crucial that studies be adequately powered to detect treatment effects, as it is otherwise impossible to determine whether null effects are due to the lack of treatment efficacy or inadequate sample sizes.

Another major barrier is the lack of standardized measures or even standardized definitions of IGD and Internet addiction across studies. In the case of IGD, this may improve given its inclusion as a condition for further research in the DSM-5 (APA, 2013) as well as recent publications guiding assessment of IGD (Petry et al., 2014, 2016). Standardization of Internet addiction definitions and measures have even further to go. In the studies reviewed here, there was substantial variability in what types of symptoms were included in the Internet addiction study inclusion criteria, and at least 10 different instruments were used to measure Internet addiction across studies. This variability and the failure of Internet addiction to be considered for inclusion as a condition for additional research in the DSM-5, as well as likely the ICD-11 (Aarseth et al., 2016; World Health Organization, 2016), point to the need for a consensus on the defining symptoms of Internet addiction. Once symptoms are clearly defined, validated measures of these symptoms should be evaluated and used consistently across studies to allow for comparison of results. Finally, study results highlight the variability in information reported across studies, with some published papers omitting key information about inclusion criteria related to IGD and Internet addiction. Improving reporting standards across journals would allow for more meaningful comparisons across studies and improved understanding of study implications.

We currently know very little about the long-term effects of any treatment for IGD or Internet addiction. The majority of studies only measured effects directly following the end of treatment, with a few exceptions. The majority of these exceptions included follow-ups at 3 months post-treatment end or less (Han & Renshaw, 2012; Kim et al., 2012; Liu et al., 2015; Sakuma et al., 2017; Su et al., 2011; Zhong, 2011) and only three followed up at 6-months (Du et al., 2010; Young, 2007, 2013). The remainder of the studies only reported outcomes immediately post-treatment. Future studies should include long-term follow-ups whenever feasible to generate information on durability of treatment effects.

Many of the studies also contained substantial limitations related to statistical analyses. For example, intent-to-treat analyses were rarely conducted, with most studies dropping non-completers from the analyses. In addition, several studies did not fully report statistical results (e.g., leaving out p values, means, or standard deviations) or did not conduct key statistical tests (e.g., group differences), making the results difficult to interpret.

Five of the studies (4 on IGD, 1 on Internet addiction) focused on a comorbid sample (i.e., with ADHD, MDD, or anxiety disorders). This is an important area for exploration, as IGD and Internet addiction are likely to be comorbid with mental disorders. However, there are currently not enough published clinical trials focused on comorbid conditions to draw conclusions about treatment efficacy for this more complex presentation. Further, it may be premature to begin developing and evaluating treatments for comorbid conditions, as the prevalence of such conditions in IGD and Internet addiction samples has not been firmly established. Instead, it may be advantageous to include careful assessments of comorbid conditions in all clinical trials for IGD and Internet addiction as well as to conduct well-designed epidemiological studies to estimate the prevalence of such comorbidities. These studies can inform whether it is necessary to develop specific interventions for comorbid conditions in the future.

Finally, many studies did not adequately describe the treatment under evaluation, rendering it impossible to ascertain the content of the intervention. It was often unclear whether the treatments were manualized and how therapists were trained and supervised on treatment protocols. Further, studies did not report on therapist adherence to the treatment models, making it difficult to draw conclusions about whether the treatments were delivered with fidelity.

Limitations of the Review

The results of the review should be considered in light of certain limitations. First, for logistical reasons, we excluded papers that were not available in English. This exclusion may be problematic for this literature, as IGD and Internet addiction has received substantial attention from researchers in Asian and European countries, many of whom publish in non-English language journals. Nevertheless, we only excluded 6 papers for the language criteria. Based on the English translations of their abstracts, three of these were not RCTs and, while the other three were RCTs, all were authored by a research group that was included in this review (Zhu et al., 2012) and may have been from the same sample. Second, there is likely a “file drawer” effect, wherein manuscripts are not submitted or accepted for publication if there are null findings. Thus, there may be more studies that did not show any positive effects. Finally, although we made a distinction between studies of IGD and Internet addiction more broadly, treatment studies in the Internet addiction group lumped IGD and Internet addiction together. Among those that reported on subtypes of Internet use (i.e., only 4 of the 13 studies), the proportion of participants who reported gaming as the primary problem varied substantially from 10% (Young, 2007) to 78% (Wolfling et al., 2014). Although these four studies report descriptive information about Internet use, none of them examined treatment outcomes separately for these subgroups.

Conclusions and Future Directions

Treatment development and evaluation for IGD and Internet addiction is a new but developing area of research with ample room for growth. Several treatment modalities have been evaluated, including medications, CBT, family-based treatments, and an eclectic array of other approaches. Unfortunately, methodological problems limit the conclusions that can be drawn about any of these approaches, and there are currently no treatments for IGD or Internet addiction that meet the criteria for an evidence-based treatment or even a possibly efficacious intervention. As noted above, future studies should aim to: a) include a control group that will allow for conclusions about the efficacy of the treatment under study; b) be adequately powered to detect treatment effects; c) include long-term follow-ups to assess durability of effects; d) use manualized treatment approaches and measure therapist adherence; and e) employ intent-to-treat statistical analysis and ensure adequate reporting of statistical results. The CONSORT guidelines (Schulz, Altman, & Moher, 2010) indicate reporting standards for clinical trials and the guidelines from Chambless et al. (1998) can be used to establish evidence-based treatments. In addition, it will be important to conduct well-designed psychometric studies of measures of both IGD and Internet addiction that can serve as gold standards for the field and allow for consistent assessment of treatment outcomes across studies. In the case of Internet addiction, additional work is needed to establish a consistent and agreed upon definition of the condition and what it constitutes.

There is room for additional treatment development for these two problems. IGD and Internet addiction both disproportionately affect adolescent and young adult populations (e.g., Haagsma, Pieterse, & Peters, 2012; Mentzoni et al., 2011), and researchers can draw on the more established literature of efficacious treatments for other disorders among adolescents in developing treatments for technology-related addictions. This literature is fairly robust, with evidence that family-based approaches seem to offer some treatment advantage for adolescent substance abuse (Tanner-Smith et al., 2013) and that “boot-camp” style correctional programs are largely ineffective (Pearson & Lipton, 1999). Cognitive-behavioral strategies, and perhaps even motivational interventions, may be useful for adolescent and adult populations alike, given their place in treatment of other mental health conditions (Dutra et al., 2008; Waldron & Turner, 2008). Finally, much more research is needed on epidemiology, etiology, risk factors, and outcomes of both IGD and Internet addiction. There has been very little research on the natural progression of these conditions, making it difficult to determine when and how much treatment is necessary to improve upon natural recovery rates. A better understanding of these issues is critical for establishing the efficacy of any intervention.