The association between internet addiction and psychiatric co-morbidity: a meta-analysis

Internet addiction (IA) was initially considered a new psychiatric disorder in the fifth edition of the Diagnostic and Statistical Manual of Mental Disorders (DSM-5) [1]. Following publication of the DSM-5 in 2013 however, internet gaming disorder was instead, contemplated [2]. The proposed diagnostic criteria for internet gaming disorder mirror the core criteria of substance misuse disorders, and include the following: 1) preoccupation with internet gaming; 2) occurrence of withdrawal symptoms when internet gaming access is removed; 3) the need to spend increasing amounts of time on internet gaming; 4) unsuccessful attempts to control internet gaming; 5) continued excessive internet gaming despite negative psychosocial consequences; 6) loss of previous interests hobbies and entertainment as a result of excessive internet gaming; 7) the use of internet gaming to relieve dysphoria; 8) deceiving others about internet gaming; and 9) loss of relationship, educational opportunity or career as a result of internet gaming. IA is a broader construct which comprises internet gaming and other forms of addictive internet usage (e.g. addictive downloading, excessive use of social networking sites and addictive online shopping). As a result, the construct of IA warrants empirical attention for further consideration as a formal psychiatric disorder in the prospective diagnostic nomenclature.

With the growth of internet users worldwide, IA has become a pandemic in the new era [1]. Ko et al. [3] reported that the prevalence of IA ranged from 1% to 36.7%. The variability in prevalence rates across the studies that were reviewed may be attributed to variations in accessibility of the internet in different countries, definitions of IA and diagnostic instruments [4]. IA and psychiatric co-morbidity may co-occur as a dual diagnosis and engender significant impact on patients and existing treatment services. There is a pressing need to assess the strength of the association between IA and psychiatric co-morbidity. Excessive internet use was found to be associated with psychiatric conditions such as depression [5], insomnia [6], attention deficit and, hyperactivity and social phobia [7]. Ko et al. [3] published a review of the literature and reported that IA was associated with attention deficit and hyperactivity disorder (ADHD), major depressive disorder and social anxiety disorder. It was however, a descriptive review which lacked statistical analyses to support such associations. Conversely, a meta-analysis is able to estimate the aggregate prevalence and odds ratio (OR) of psychiatric co-morbidity in patients suffering from IA as compared to healthy controls. Carli et al. [8] published a systematic review on the association between pathological internet use and combined psychopathology. According to the review, the prevalence rates of psychiatric co-morbidity varied from 57% with symptoms of anxiety to 100% with attention deficit and hyperactivity. The high prevalence rate reported in this review has to be interpreted with caution because the authors did not specify the statistical method that was used to calculate the aggregate prevalence. Furthermore, the pooled OR was not calculated for each psychiatric co-morbidity. Carli et al. also concluded that depression and ADHD demonstrated the strongest correlations with pathological internet use. However, these correlations need to be corroborated by a methodologically robust meta-analysis. The amalgam of studies on IA and psychiatric co-morbidity seemed to elucidate heterogeneity yet both reviews did not report the level of heterogeneity. The objective of this meta-analysis was to evaluate the association between IA and psychiatric co-morbidity. We performed a meta-analysis of cross-sectional, case-control and cohort studies to determine the overall strength of the putative association between IA and psychiatric co-morbidity. We also computed the aggregate prevalence of psychiatric co-morbidity among people with IA as compared to controls without IA, and reported the level of heterogeneity for each psychiatric co-morbidity. As IA is a confluence of substance use and addictive disorder, we hypothesized that the prevalence of psychiatric co-morbidity in IA would be similar to the prevalence of psychiatry co-morbidity in substance use and addictive disorders in general.

Online databases were searched from inception to June 2012: PubMed (from 1966), Embase (from 1980), PsychINFO (From 1806), BIOSIS (from 1926), Science Direct (from 2006) and Cochrane CENTRAL (from 1993). The search terms used were permutations of keywords for internet addiction (internet addiction, problematic internet use) and psychiatric co-morbidity (depress*, mood, bipolar, ADHD, attention deficit hyperactivity, conduct, anxiety, phobia, panic, psychosis, schizo*, eating, anorexia, bulimia, personality, antisocial, borderline, narcissistic, histrionic, schizoid, schizotypal, paranoid, dependent, anxious, avoidance, obsessive compulsive, anankastic, histrionic, alcohol, cannabis, marijuana, amphetamine, cocaine, stimulant, ecstasy, hallucinogen, ketamine, phencyclidine and heroin) where the symbol* indicates truncation. Abstracts presented in major international conferences and unpublished dissertations were manually searched, and the authors of selected correspondences were contacted for further information. References from the retrieved papers were hand-searched.

Eight studies comprising a total of 1641 patients with IA and 11210 controls without IA were included in the analysis (Figure 5). The studies consisted of patients mainly from Asian countries. Descriptive characteristics of the study populations are shown in Table 1. Subgroup analyses are shown in Table 2.

To our knowledge, this is the first meta-analysis to investigate the relationship between IA and psychiatric co-morbidity with consideration of heterogeneity. Our findings suggest that IA is associated with alcohol abuse, ADHD, depression and anxiety. Amongst these significant psychiatric co-morbidities, alcohol abuse has the strongest association with IA. Carli et al. [8] reported that 75% of pathological internet users suffer from depression, 57% from anxiety and 100% from ADHD. In the present study, 26.3% of patients with IA suffer from depression, 23.3% from anxiety and 21.7% from ADHD. Our findings are similar to the prevalence of coexisting psychiatric disorders in patients suffering from alcohol and drug disorders which ranges between 20% and 30% [25]. Resultant findings support our hypothesis that the prevalence of psychiatric co-morbidity in IA is similar to that in substance use and addictive disorders. A previous study by Carli et al. may have overestimated the prevalence of psychiatric co-morbidity in IA, especially that of ADHD. Contrary to alcohol abuse, depression and ADHD, the aggregate prevalence of anxiety demonstrates significant heterogeneity. Subgroup analysis showed that the aggregate prevalence of the respective psychiatric symptoms varied significantly based on the age stratification of the sample.

Our results suggest that IA and psychiatric co-morbidity may co-occur as a result of complex interaction between various aetiological factors. The intricacies of the genetic transmission of IA remains exploratory. Montag et al. [26] found that the CC genotype of the rs1044396 polymorphism on gene coding for the nicotinic acetylcholine receptor subunit alpha 4 (CHRNA4) occurred significantly more frequently in patients suffering from IA. Moreover, nicotinic receptors play a key role in nicotine addiction. Lee et al. [27] reported that the homozygous short alleles (SS) of the serotonin transporter gene promoter region (5HTTLPR) are more prevalent among excessive internet users, and the genotype was found to be associated with depressive disorder [28]. The involvement of the serotonin genotype in IA and depression suggests that these two conditions may share similar neurochemical changes thereby warranting further investigation.

Patients suffering from IA are more likely to be non-compliant with psychotropic medication and psychotherapy because they are preoccupied with internet usage. Furthermore, depression and anxiety often occur as part of the internet withdrawal syndrome. Excessive internet usage may serve as a maintaining factor for anxiety by reinforcing the avoidance of anticipatory anxiety stemming from stressful situations and life events. The relationship between IA and alcohol is complicated. The biological reinforcement models [25] suggest that alcohol may alter neuroanatomical pathways that are involved in the positive reinforcement of internet use. Although internet-based treatment has been used for detrimental alcohol use [29], there is a paucity of research in the effects of alcohol on internet use. Wu and Delva [30] reported that the use of internet at home has no effect on drinking but the use of computers in internet cafés was a strong predictor of drinking among women. Yen et al. reported that fun-seeking is a shared characteristic of IA and alcohol abuse [19]. The above-mentioned findings support our subgroup analysis which demonstrated that alcohol abuse is more prevalent among younger subjects (10-18 years of age), as internet cafés and fun-seeking are typically more common among youths. While it is unlikely that IA causes ADHD, clinical impressions posit that internet usage may improve ADHD. In our subgroup analysis, ADHD is more prevalent in young adults (19-39 years of age). Clinical observations point to a predominance of inattention over and above hyperactivity in adults. The self-treatment hypothesis thus postulates that adult patients with ADHD use the internet excessively to control their inattention. Furthermore, the onset of ADHD (at age 7 years by clinical definition) usually predates the incipience of IA [31], thereby suggesting the potential role of ADHD as a predisposing factor of IA.

Present emergent findings confer significant clinical implications as they aid mental health professionals in appreciating that IA may not present as a singular diagnostic entity, but co-occur with alcohol abuse, ADHD, depression and anxiety. It is thus reasonable to expect that all patients who present with IA be adequately screened for alcohol abuse, ADHD, depression and anxiety. In a similar vein, it is recommended that patients who present with the above-mentioned psychiatric condition be interviewed about their internet usage. Our findings also inform the prospective treatment of IA because psychiatric co-morbidity may reinforce, if not maintain the pathological pattern of internet usage. Therefore, the treatment of IA and psychiatric co-morbidity should be integrated into a cohesive service which focuses on the minimization of harm [25]. The overarching goals of treatment would encompass the abstinence of internet usage, activity scheduling to supplant online activity with face-to-face interpersonal activity, and motivational interviewing. The reduction of internet usage may lead to a corresponding reduction of the severity of comorbid psychiatric symptoms. Pharmacotherapy involving the opioid receptor antagonist, naltrexone may diminish euphoria or the rewarding experience that is associated with alcohol abuse and IA. Similarly, selective serotonin reuptake inhibitors (SSRIs) may attenuate the impulsivity that is associated with internet use, depression and anxiety. For patients who tend to use the internet in internet cafés which serve alcoholic beverages, avoidance of those locations or short-term hospitalisation may treat both conditions simultaneously.

A strength of this meta-analysis is the minimal publication bias. Notwithstanding, this meta-analysis is not without its limitations. Firstly, we could not assess cause-and-effect mechanisms underpinning IA and psychiatric co-morbidity because most studies included are cross-sectional in nature. Secondly, the association between IA and psychiatric co-morbidity should be interpreted with caution because there exist confounding factors, for example attachment [32], environmental stress, parenting styles [33] family structure [34], and gender. In extant case-control studies [19, 20, 24], the proportion of males is significantly higher in the IA groups as compared to control groups. ADHD and alcohol abuse are also known to be more common in males than females [35]. Furthermore, the number of studies focusing on alcohol abuse is small (n = 3 studies). Thirdly, studies focusing on other psychiatric co-morbidities such as eating disorders and abuse of other recreational drugs did not meet the inclusion criteria and were not included. Although there exist other studies which reported the severity levels of depression, anxiety and ADHD, these studies were excluded because different questionnaires were used and the scores could not be combined. Most studies included in the meta-analysis employed self-report questionnaires with the exception of a single study which had established the diagnosis of psychiatric co-morbidity using a structured interview format [23]. Due to the small number of studies, we could not perform a sensitivity analysis to investigate the differences in psychiatric co-morbidity that was established by self-reported questionnaires and structured interviews as well as the differences in pooled OR between varying study designs. The meta-analysis was also limited in that the scales and structured interviews administered focused primarily on IA but not IA-specific behaviours such as gaming, shopping and social media. Consequently, we could not establish the psychiatric co-morbidity associated with specific types of IA behaviour. A fifth limitation resides in a high and statistically significant level of between-study heterogeneity that was found in the pooled OR for anxiety. This warrants further meta-regression to identify moderators that may be attributing to the significant heterogeneity. Meta-regression was not performed in this study because Gagnier et al. has recommended at least 10 studies per moderator in meta-regression to avoid spurious findings. Finally, most of the subjects included in this meta-analysis were young Asians from China and Korea. Further studies are required to investigate other ethnic groups in Europe and North America, as well as older adults. It is important to note that the patterns of internet use and recreational drug use may vary between Eastern and Western populations.

Further research is necessitated to arrive at a consensus on the definition of IA and examine the unique interactions between IA and psychiatric co-morbidity such as common aetiology, illness trajectory and treatment outcomes. In this meta-analysis, the definition of IA was based on two instruments, namely Young’s Internet Addiction Test and Chen Internet Addiction Scale. Although there are overlapping characteristics between two questionnaires, further research is required to arrive at a consensus on the diagnostic criteria of IA.

With the exception of ADHD (age of onset is known to be earlier than 7 years) [31], the sequence of development of IA and other psychiatric co-morbidity remains unclear. Prospective studies are required to determine if psychiatric co-morbidities such as depression and anxiety originate from IA and abate with reduced usage of internet or otherwise. Also, as patients suffering from IA and psychiatric co-morbidity may not respond to standard treatment approaches, further research is required to investigate the effect of psychotropic medications (e.g. antidepressant, methylphenidate) on the severity of IA and psychiatric co-morbidity.

The present meta-analysis engendered a significant association between IA and alcohol abuse, ADHD, depression and anxiety. Psychiatric co-morbidity occurs in a range from 13.3% to 26.3% among patients suffering from IA. To enhance understanding of the relationship between IA and psychiatric co-morbidity, prospective studies involving other psychiatric co-morbidities are required to establish the cause-and-effect mechanisms between IA and psychiatric co-morbidity. In view that the complex interactions between IA and psychiatric co-morbidity may lead to significant health burden, concerted mental health services need to be developed in anticipation of the recognition of IA as a diagnosable psychiatric disorder in the future.