**1. Introduction**

Internet gaming disorder (IGD) has been identified as a potential psychiatric disorder because of its negative effects on multiple domains of functioning [1–3]. The international prevalence ranges from 0.3% to 12%, and it is an increasing public health concern, especially in Asian countries. As well as cognitive and behavioral symptoms similar to those of substance-use disorder, IGD also reportedly has associations with psychiatric symptoms, including attention-deficit hyperactivity disorder, depression, anxiety, and psychosomatic symptoms [4–7]. High comorbidity of emotional symptoms suggests that individuals with IGD might use gaming to escape emotional difficulties [8,9].

Addictive behaviors are often initiated as a maladaptive mechanism for coping with stress [10]. Stress may enhance abstinent individuals' memories of addictive behaviors as stress relievers then increase the risk of relapse to addictive behaviors after abstinence [11]. Tao et al. [12] were first to use the escape from stress through gaming as a criterion for Internet addiction. It was subsequently

listed as a diagnostic criterion for IGD in the Diagnostic and Statistical Manual of Mental Disorders (DSM-5) [13]. "Negative escapism" describes gaming being negatively reinforced as a means of avoiding stress and was reported among 77.8% of internalizing patients [14]. Kim et al. [6] found escape from negative emotions to be associated with depression in IGD [9]. Furthermore, high levels of escapism were also reported to be associated with more IGD symptoms [15]. Moreover, the motives for escapism mediate the association between psychiatric distress and problematic online gaming [16]. Thus, negative escapism is associated with the symptoms of Internet addiction, psychological distress, and poor life satisfaction among massively multiplayer online role-playing gamers [17], and escapism under stress could play a major role in excessive gaming behavior.

Convincing evidence indicates that stress is a risk factor for addiction and triggers relapses [18]. The negative reinforcement model of addiction is defined as drug-taking or addictive behavior that alleviates a negative emotional state. According to some hypotheses, the negative emotional state that drives such behaviors as negative reinforcement is derived from dysregulation of brain stress systems involved in addiction processes [19]. Thus, investigating the perceived stress among adults with IGD could contribute to understanding its role in developing addictive online gaming behavior.

Not all individuals who face stress develop addictions. Therefore, resilience factors may protect these mentally healthy individuals [20]. For example, psychosocial resilience, such as positive emotions, optimism, humor, cognitive flexibility, and reappraisal could attenuate stress-induced psychopathology [21]. Thus, resilience was reported to be a buffering factor against Internet addiction [22]. Furthermore, resilience was reportedly lower among adolescents with IGD according to questionnaire assessments [15]. Its moderating role in the association between stress and IGD severity was demonstrated by an online questionnaire survey [23]. However, the difference in resilience was not evaluated among adults with IGD based on diagnostic interviewing. Further, resilience included a variety of personal characteristics, such as acceptance, problem-solving skills, capacity to recover, self-regulation, personal competence, and self-efficacy [23,24]. Which characteristics of resilience plays an important role involving development of IGD and should be intervened firstly has not been well-evaluated.

Depression is one of the most reported correlates of IGD [25]. Although the causal relationship had not been confirmed between depression and IGD, a longitudinal study suggested that depression could be an outcome of pathological gaming [26]. Prolong stressful events contribute to depression and resilience plays an important role in stress-related disorder and depression [27]. Resilience has been also reported to decrease the likelihood of stress-induced depression [21]. Both stress and depression were reported to be associated with IGD [23,25]. However, the associations among stress, depression, and resilience have not been well examined through interview studies among adults with IGD. Further, we might hypothesize that resilience could be associated with IGD and contribute to perceived stress and depression.

According to the aforementioned studies, resilience enables people to adapt successfully to stress or emotional difficulty and to avoid stress-related disorders [28]. Thus, we hypothesized that an individual with lower resilience could have higher perceived stress and depression. As IGD could be a maladjustment behavior to depression or stress, we hypothesized that individuals with low resilience were more likely to have IGD. Further, the higher perceived stress and depression could involve the association between low resilience and IGD. Moreover, among individuals of IGD, we hypothesized that those with lower resilience had higher perceived stress and depression. Thus, the aim of the study was to evaluate: (1) the difference in resilience, perceived stress, and depression between IGD and healthy controls; (2) the difference in depression and perceived stress between individuals with lower resilience and those with adequate resilience; (3) the confounding effect of perceived stress and depression in association between resilience and IGD; (4) the difference in depression and perceived stress between individuals with lower resilience and those with adequate resilience among the IGD group; and (5) the most associated resilience characteristics of IGD.

#### **2. Methods**

#### *2.1. Participants*

Our participants comprised individuals who had IGD at the time of the study (the IGD group) and individuals who had never had IGD (the control group) based on a design of case-control study. All participants were recruited by advertisement from September 2012 to October 2013. The criteria for the IGD group specified that participants be: (1) young adults aged 20 to 30 years and with more than 9 years of education; (2) individuals spending either ≥40 h per week or ≥4 h per day on weekdays and ≥8 h per day on weekends engaged in Internet gaming; and (3) individuals who maintained a pattern of Internet gaming for more than 2 years. Participants who met all of three criteria underwent an additional interview with a psychiatrist using the criteria of the DSM-5 [13] for IGD diagnosis. Those who had IGD at the time were classified into the IGD group.

For each participant in the IGD group, we matched a participant to be included in the control group by gender, education level, and age (within a range of 1 year). The recruitment criterion of participants in the control group was that their daily nonessential Internet use was less than 4 h. These participants were classified into the control group after a diagnostic interview with a psychiatrist.

All participants underwent the interview, which comprised two steps: (1) a diagnostic interview based on the Mini-International Neuropsychiatric Interview (MINI) to assess psychotic disorders, bipolar I disorder, and substance-use disorders; and (2) a history-taking interview to evaluate psychotropic medication use, mental retardation, severe physical disorder, and brain injury. Those who had psychotic disorders, bipolar I disorder, substance-use disorders, mental retardation, severe physical disorder, or brain injury or used psychotropic medication were excluded.

#### *2.2. Measures*

The diagnostic criteria of IGD were those of the DSM-5 [13]. The nine criteria comprised preoccupation, withdrawal, tolerance, unsuccessful attempts to control others, loss of interests other than gaming, continued excessive use despite psychosocial problems, deceit regarding online gaming, escape, and functional impairment [13]. We developed a semi-structured interview schedule to assess the DSM-5 criteria for IGD among our participants. Those who met five or more criteria for IGD were classified into the IGD group.

#### 2.2.1. Chinese Version of the MINI

We conducted a diagnostic interview based on the psychotic disorder, bipolar I disorder, and substance-use disorder modules in the Chinese version of the MINI [29] to detect those excluding psychiatric disorders.

#### 2.2.2. 14-Item Resilience Scale (R14)

The R14, which provides reliable internal consistency and external validity, was developed to evaluate the levels of resilience in the general population. Participants' resilience was assessed using this scale [24], in which scores are calculated through summation of the response values for each item, enabling scores to range from 14 to 98. The internal consistency reliability (Cronbach's alpha) of the total scale is 0.93 in the current study. Scores less than 65 indicate low resilience, scores between 65 and 81 indicate moderate resilience, and scores of more than 81 indicate high levels of resilience [24]. In this study, participants scoring more than 64 were classified into an adequate resilience group, and those scoring 64 or less were classified into a low resilience group.

#### 2.2.3. Perceived Stress Scale

The Perceived Stress Scale (PSS) was designed to measure the extent to which situations in one's life are perceived as stressful. The PSS score was correlated with life-event scores, depression, and physical symptomatology [30]. It is suggested that the scale possesses adequate reliability for as outcome measure for experienced stress level. In this study, the 10-item PSS-10 was used to evaluate the level of stress experienced by participants and its internal consistency reliability (Cronbach's alpha) was 0.86.

#### 2.2.4. Center for Epidemiological Studies' Depression Scale

The 20-item Mandarin Chinese version [31] of the Center for Epidemiological Studies' Depression Scale (CES-D) [32] is a self-administered evaluation of the frequency of depressive symptoms during the past week. This was used to evaluate depression. Its internal consistency reliability (Cronbach's alpha) was 0.92 in the current study.

#### 2.2.5. Clinical Global Impression Scale for IGD

The Clinical Global Impression (CGI) scale [33] asks "Considering your total clinical experience with this particular population, how mentally ill is the patient at this time?" Possible responses were 1 for normal, not at all ill; 2 for borderline mentally ill; 3 for mildly ill; 4 for moderately ill; 5 for markedly ill; 6 for severely ill; and 7 for among the most ill patients. We modified the scale for IGD to 1 for normal, not at all ill; 2 for excessive online gaming without fulfilling the IGD criteria; 3 for fulfilling the IGD criteria with mild functional impairment; 4 for moderate functional impairment in health or one field such as academics, socializing, or profession; 5 for moderate functional impairment in multiple dimensions; 6 for severe impairment in one field; and 7 for severe impairment in multiple dimensions of daily life.
