INTRODUCTION
Globally, there has been a significant increase in Internet/smartphone use among children and adolescents, and smartphones have become essential in the daily lives of children and adolescents in South Korea [1-3]. Recent research showed that the exponential penetration of smartphones has led to a greater prevalence of problematic excessive Internet/smartphone use [4-6]. The prevalence of Internet/smartphone addiction risk in South Korea was 20.7% for children (2.0% high risk, 20.7% mild risk) and 29.3% for adolescents (3.6% high risk, 25.7% mild risk) [7,8]. In addition, a recent previous study also reported that 22.1% (n=12,147) showed a potential risk of problematic smartphone use (PSU), and 3.0% (n=1,633) had a high risk of PSU among the 54,948 Korean adolescents [5].
Problematic users of Internet/smartphone/social network services (SNS) have been described with behavioral addiction as specified in the eleventh revision of the International Classification of Diseases-11 (ICD-11) for gaming disorder [9,10]. Internet addiction is described as an individual’s inability to control their Internet usage, leading to adverse effects in their daily life [11]. In particular, the addictive behavior may lead to significant distress and impairment in individual, educational, and social functions [12-14]. When evaluating the impact of Internet gaming disorder (IGD) on physical health, mental health, and daily functioning, it is crucial to use appropriate assessment tools [15].
As assessing the severity of Internet/game/smartphone use problems has become important, we developed a semi-structured interview to measure Internet/game/smartphone addiction. In Korea, the Diagnostic Interview for the Internet Addiction Scale (DIA) was developed based on the nine criteria of Diagnostic and Statistical Manual of Mental Disorders-Fifth Edition (DSM-5) Section III for diagnosing IGD [16]. Until now, the IGD criteria may be vulnerable to criticism for excluding the concept of “craving,” which has traditionally been considered significant in the context of addiction [17,18]. In particular, the “craving” factor has been shown to have a higher predictive rate for IGD compared to other criteria such as withdrawal, preoccupation, and escapism [17].
Thus, the DIA added the craving factor and finally consisted of a total of 10 Internet/game/smartphone addiction diagnosis factors (e.g. withdrawal, tolerance, cognitive salience, difficulty in regulating use, persistent use despite negative results, loss of interest in other activities, use of Internet/games/SNS to avoid negative feelings, deception regarding Internet/games/SNS use, interference with role performance, and craving).
In a previous study, the psychometric properties of DIA were analyzed and presented for Korean children and adolescents, however, there was a limitation in that the factor analysis of DIA was not performed [19]. Exploratory factor analysis (EFA) is a necessary stage in assessment development to examine the underlying factor structure and analyze how items correlate. It is needed to measure according to validated tools to understand the phenomenon of addiction and specifically assess it among children and adolescents. In addition, researchers need a proved assessment tool which showed validity and reliability for use across various populations [20]. To the best of our knowledge, no study of EFA on DIA among Korean children and adolescents has been published and few specific diagnostic interviews are available yet despite high research and public interest.
To bridge the gap, we developed a semi-structured interview that measures Internet/game/smartphone use, pilot-tested it in clinical samples, and investigated the internal factor structure. In addition, we validated the development of the DIA and the results of the EFA.
METHODS
Study population and data collection
A multicenter cohort study (clinic-Cohort for Understanding of Internet addiction Rescue factors in Early life, c-CURE) was performed in South Korea from 2015 to 2019 [7]. This clinical cohort study aimed to investigate and track Internet, gaming, and smartphone addiction in children and adolescents [2]. Details about the c-CURE study are presented in previous studies [2,7]. We recruited and conducted screenings on children and adolescents, aged 7-18 years, who showed excessive use of Internet games and/or smartphones at three university hospitals in South Korea. Participants completed screening using questionnaires specifically aimed at evaluating Internet and smartphone addiction. The enrollment screening criteria (cut-off values) in this study were as follows: Korean scale for Internet addiction for adolescents (K-scale; children≥82, adolescents≥95) [21], Korean smartphone addiction scale (S-scale; ≥42) [22], Smartphone addiction scale-short form version (SAS-SV; boys≥31, girls≥33) [23], and Internet addiction proneness scale (children≥40, adolescents≥32) [24]. Individuals who had a score higher than the cut-off on at least one of the assessments were included in the cohort study. Finally, a total of 194 children, adolescents, and their caregivers were included in the study.
Measurement
DIA
The DIA is a semi-structured diagnostic interview tool for IGD. The DIA is conducted by interviewing both children, adolescents, and their caregivers. Psychiatrists and clinical psychologists assessed the semi-structured in-person interviews with participants and their caregivers to identify DIA [7]. Each item in the interview includes a standardized representative question and multiple examples [19]. It includes 10 items according to the DSM-5 IGD diagnostic criteria [16]. These elements consist as follows: 1) cognitive salience, 2) withdrawal, 3) tolerance, 4) difficulty in regulating use, 5) loss of interest in other activities, 6) persistent use despite negative feelings, 7) deception regarding Internet/games/SNS use, 8) use of Internet/games/SNS to avoid negative feelings, 9) interference with role performance, and 10) craving.
The average length of DIA interviews generally ranges between 10 and 20 minutes for each participant and their caregivers. Each component is accompanied by standardized representative questions and multiple examples, which enables clinicians to assess the score more effectively. The script of the standardized interview for 10 items was shown in the previous study [19]. After conducting interviews, the clinician assessed a total score to ascertain whether the participants showed addiction to the Internet/games/SNS. The rating for each item was based on a 4-point scale, with the following categories: 0 for no information, 1 for no symptoms, 2 for subthreshold, and 3 for threshold level. The number of components with a score of 3 (threshold) is determined as the DIA total score (range: 0-10) and a higher score means a more severe level of IGD [19].
K-scale
The K-scale was developed for evaluating and addictive Internet behaviors [21]. It consists of 40 items, and the responses are evaluated using a 4-point Likert scale ranging from 1 (“not at all”) to 4 (“always”) [25]. The cutoff point of the K-scale is children ≥82, adolescents ≥95, and a higher total score is interpreted as a higher level of Internet addiction [21].
SAS-SV
The SAS-SV consists of 10 items, and the responses are evaluated using a 6-point Likert scale, ranging from 1 (“not at all”) to 6 (“it really is”) [23]. The score of each question was summed to obtain the total score of the SAS-SV and scores higher than the cut-off value (31 for boys, 33 for girls) were considered the high-risk user group [23].
S-scale
The S-scale is used to measure smartphone addiction and consists of 15 items. The responses are scored based on a 4-point Likert scale (1: “not at all” to 4: “it really is”). A score ranges from 15 to 60 points; a higher score means a more severe level of smartphone addiction. S-scale scores are categorized into general users (0-41), potential-risk (42-44), and high-risk groups (45-60) [22].
Young’s Internet Addiction Test
Young’s Internet Addiction Test (YIAT) consists of 20 items, and the responses are scored based on a 6-point Likert scale (0: “does not apply,” 1: “rarely,” 2: “occasionally,” 3: “frequently,” 4: “often,” 5: “always”) [26,27]. A score ranging from 0 to 100 points and a higher score means a more severe level of Internet addiction. YIAT scores are categorized into general user (0-49), potential-risk (50-79), and high-risk groups (80-100) [27,28].
Internet addiction proneness scale for adolescents
The Internet addiction proneness scale for adolescents (O_A) consists of 15 items, and it is administered to caregivers to corroborate the self-report K-scale [24]. The responses are scored based on a 4-point Likert scale (1: “not at all” to 4: “it really is”). The cutoff point of the O_A are as follows: children≥40, adolescents≥32 [24].
Statistical analysis
Sample characteristics were analyzed using ranges, frequencies, percentages, mean, and standard deviations (SD). Factor analysis, correlation analysis was conducted to verify the validity of DIA. Cronbach’s alpha correlation coefficient was conducted for the DIA scale of 10 items. Statistical tests were analyzed using SPSS software ver. 22.0 (IBM Corp.).
We performed parallel analysis using 1,000 randomly selected data sets. The determination of the number of factors to be obtained was performed by applying the eigenvalues greater than one rule, comparing model fit statistics, and analyzing the scree plot. Based on the factor structure we obtained, we conducted an oblique rotation (promax) to allow the extracted factors to correlate with each other [29]. The Kaiser-Meyer-Olkin (KMO) and Bartlett’s test demonstrated that the dataset was suitable for conducting EFA [20]. Based on results of EFA, confirmatory factor analysis (CFA) was performed utilizing structural equation modeling in Amos 19.0 program (SPSS Inc.). The fit indices assessed included Tucker-Lewis Index (TLI), comparative fit index (CFI), and root-mean square error of approximation (RMSEA). In addition, we performed the sensitivity analysis for correlations between the subscales of DIA and self-report scales among the smartphone-only user group. A significance level of p<0.05 was used for all statistical tests.
Ethics statement
All participants and their caregivers provided informed consent before participating in this study. The study was performed by the Declaration of Helsinki. This protocol was approved by the Institutional Review Board (IRB) of Eulji University Eulji Hospital (EMCS2015-05-020-001), SMG-SNU Boramae Medical Center (16-2016-4), and Uijeonbu St. Mary’s Hospital (UC150NMI0072).
RESULTS
Sample characteristics
The sociodemographic characteristics of study participants (n=194) are shown in Table 1. Of these, 146 were boys and 48 were girls, with children ages ranging from 7 to 18 years (13.17±2.46). According to the analysis of the content specifier, the combined type was the most common (63.9%, n=124), followed by IGD (22.2%, n=43), other Internet addiction (7.2%, n=14), and SNS addiction (5.7%, n=11). In the content specifier, the combined type was the most common at 46.4% (n=90), followed by using a smartphone (39.7%, n=77), and using a PC (11.9%, n=23). The average score of self-reported scales was 74.5 (SD=19.0) for K-scale, 28.4 (SD=12.1) for SAS-SV, 33.1 (SD=8.9) for S-scale, 47.1 (SD=16.4) for YIAT, 45.6 (SD=5.8) for O_A (children), and 41.9 (SD=7.3) for O_A (adolescents).
Correlations between DIA and self-report scale
The results of the correlation analysis are shown in Table 2. The DIA was highly correlated with the scores of the K-scale (r=0.327, p<0.01), YIAT (r=0.243, p<0.01), O_A for children (r=0.566, p<0.01), O_A for adolescents (r=0.383, p<0.01).
Factor structure and internal consistency of DIA
The results of factor analysis are shown in Table 3. Factor 1 included “Q1: cognitive salience,” “Q2: withdrawal,” “Q3: tolerance,” “Q4: difficulty in regulation,” “Q5: decrease in other activities,” “Q7: lying about Internet/game/SNS use,” “Q8: use Internet/game/SNS to avoid negative emotions,” and “Q10: craving.” In addition, factor 2 included “Q6: persistent use in spite of negative results,” and “Q9: interference in role performance and function.” They explained 49.15% of the whole scale. The Cronbach’s alpha correlation coefficient for the total scale was 0.806, and those for the two factors were 0.790 and 0.613, respectively. The overall sampling suitability of the 10-item scale was tested using KMO, resulting in a high value of 0.861. The p-value of the Bartlett test was <0.001, indicating that factor analysis was appropriate.
Correlations between the factor 1, 2 and self-report scale
The results of factor analysis are shown in Table 4. The factor 1 of DIA was highly correlated with the scores of the K-scale (r=0.348, p<0.01), SAS-SV (r=0.152, p<0.05), S-scale (r=0.175, p<0.05), YIAT (r=0.298, p<0.01), O_A for children (r=0.543, p<0.01), and O_A for adolescents (r=0.215, p<0.05). The factor 2 of DIA was highly correlated with the scores of the K-scale (r=0.195, p<0.01), YIAT (r=0.152, p<0.05), O_A for children (r=0.424, p<0.01), and O_A for adolescents (r=0.395, p<0.01), except for SAS-SV and S-scale.
CFA of DIA
The results in CFA of DIA are shown in Table 5. There are no definitive cut-off points for acceptable model fit when using these indices, but the following criteria are frequently used to indicate the goodness of fit for a particular model: CFI >0.90, TLI >0.90, and RMSEA <0.05. Factor loadings are generally considered strong if >0.40; however, even factors that load at lower levels but contribute to good model fit (i.e., the absence of those factors would reduce the strength of fit indices) may be appropriate to include if theoretically relevant. Results from the CFA indicated that the two-factor model demonstrated good model fit: TLI=0.919, CFI=0.950, RMSEA=0.058.
Sensitivity analysis—correlations between the subscales of DIA and self-report scale in the smartphone—only user group
Since the factor 2 was not correlated with smartphone addiction scales such as SAS-SV and S-scale, we conducted the subgroup analysis for the smartphone-only users. The results of the correlation analysis that was conducted in the smartphone-only users between the subscales of DIA and self-report scale are shown in Table 6. The results show that all the subscales of DIA were significantly related to those of the self-report scale among the smartphone-only user subgroups.
DISCUSSION
The DIA was developed for evaluating the level of Internet addiction for children and adolescents. A total of 194 participants were included and the validity of DIA was evaluated in the 10 items through factor structure, internal consistency reliability, and CFA.
The correlation between DIA and self-report scale showed a high correlation with DIA was the highly correlated with the scores of the K-scale, YIAT, O_A for children, and O_A for adolescents. Especially, the factor 1 of DIA was highly correlated with the scores of the K-scale, SAS-SV, S-scale, YIAT, O_A for children, and O_A for adolescents. The factor 2 of DIA was highly correlated with the scores of the K-scale, YIAT, O_A for children, and O_A for adolescents, except for SAS-SV and S-scale. Based on these results, the DIA’s association with other self-report scales emphasized the potential value of identifying factors contributing to IGD.
The DIA showed good internal consistency and validity. The Cronbach’s alpha in the current study was 0.806. Other prior studies have also found that existing measures of smartphone addiction or Internet addiction were shown to be reliable, with Cronbach’s alphas of above 0.7 [30]. Accordingly, our results suggest that the DIA is a reliable assessment for evaluating the severity of IGD in children and adolescents. Future comprehensive studies are needed to conduct using the DIA in children and adolescents in clinical settings, as well as community and school environments.
Our findings demonstrated the two-factor structure through a clinical sample of children and adolescents and verified good validity and convergence, similar to a prior study [19]. The 8 items (1-5, 7-8, and 10) that loaded onto the first factor related to cognitive salience, withdrawal, tolerance, difficulty in regulation, decrease in other activities, lying about Internet/game/SNS use, use Internet/game/SNS to avoid negative emotions, and craving. The 2 items (6 and 9) that loaded onto the second factor related to persistent use in spite of negative results and interference in role performance and function.
Previous findings suggested that the Internet/smartphone/SNS addiction process may influence problematic health behaviors and mental health in children and adolescents [31,32]. IGD is linked to higher levels of psychological distress and the presence of other mental health conditions such as sleep disturbances, depression, anxiety disorders, and attention-deficit hyperactivity disorder [3,17,33]. In light of these results, increasing evidence implicates IGD as a significant factor. Thus, since Internet/smartphone use patterns in children and adolescents are related to overall health behaviors and mental health, it is important to form appropriate and healthy daily habits at this developmental stage and to intervene actively in the prevention of IGD [34,35].
To be diagnosed with gaming disorder according to the ICD-11, an individual must have three specific symptoms: an increasing priority given to gaming, impaired control over gaming, and continued or escalated gaming despite the occurrence of negative consequences [7,9,10]. ICD-11 proposes diagnosis criteria as three factors for a gaming disorder, future studies should be necessary to develop the corresponding diagnosis assessment.
The limitations of our study were as follows. This survey was performed in a specific region and did not fully control for demographic factors, as the gender ratio was not one-to-one. Our study was conducted by a cross-sectional design. Therefore, we need to require designed longitudinal studies to assess and identify whether follow-up outcomes of IGD. Additionally, we did not analyze the other individual and environmental risk factors that could potentially impact the development and clinical symptoms of IGD. Further studies could be conducted to analyze potential risks or protective factors through comprehensive gathering of information. Also, additional studies are required to validate the biological markers and control study design.
Despite those aforementioned limitations, the sample collection of our study was conducted in well-controlled clinical settings. We collected data by performing face-to-face diagnostic interviews with clinicians. Therefore, this semi-structured 10-item scale could be considered to be an effective assessment to predict Internet addiction based on the professionals’ diagnoses.
In conclusion, this study demonstrated that the DIA is an effective tool to investigate Internet addiction in children and adolescents. The DIA showed a high convergent validity and good internal consistency with measures of IGD. DIA can be used to identify a potentially high-risk group for Internet addiction among children and adolescents in the clinical setting. To prevent Internet addiction, it is important to conduct additional research on the characteristics of addiction, develop effective interventions, and cautiously plan their implementation in the future.
PDF Links
PubReader
ePub Link
XML Download
Full text via DOI
Download Citation
Print
