Professional Documents
Culture Documents
DOI 10.1007/s11920-014-0508-2
Abstract Internet gaming addiction was included in the latest discussed in light of their implications to the fulfilment of
version of the DSM-V as a possible disorder recently, while the validation criteria.
debate is still on-going as to whether the condition called
“Internet Addiction” (IA) could be fully recognised as an Keywords Internet addiction . Risk factor . Protective factor .
established disorder. The major contention is how well IA Health effect . Adolescents . Systematic review
could fulfil the validation criteria as a psychiatric disorder as
in other well-established behavioural addictions. In addition to
various proposed validation criteria, evidence of risk and Introduction
protective factors as well as development of outcomes from
longitudinal and prospective studies are suggested as impor- It has long been noticed that excessive use of information and
tant. A systematic review of available longitudinal and pro- communication technologies could become problematic be-
spective studies was conducted to gather epidemiological haviour that requires more attention [1]. In the past decade or
evidence on risk and protective factors of IA and the health more, different terms have been used to describe such behav-
effect of IA on adolescents. Nine articles were identified after iour including: “compulsive computer use”; “Internet depen-
an extensive search of the literature in accordance to the dency”; “pathological Internet use”; and “Internet addiction”
PRISMA guidelines. Of these, eight provided data on risk or [2–5]. Some other terms have also been used to describe
protective factors of IA and one focused solely on the effects specific Internet use behaviours, such as: “Internet gaming
of IA on mental health. Information was extracted and disorder”, “Internet gaming addiction”, “cybersex addiction”
analysed systematically from each study and tabulated. Many [6–9]. While the debate is still on-going as whether the con-
exposure variables were studied and could be broadly classi- dition called “Internet addiction” (IA) could be fully
fied into three main categories: psychopathologies of the recognised as an established disorder, Internet gaming addic-
participants, family and parenting factors, and others such as tion (IGA) has been included in Sect. III of the latest version of
Internet usage, motivation, and academic performance. Some the Diagnostic and Statistical Manual of Mental Disorder V
were found to be potential risk or protective factors of IA. It (DSM-V) as an emerging disorder that requires further inves-
was also found that exposure to IA had a detrimental effect on tigation [10]. While acknowledging that there may be many
the mental health of young people. These results were different facets of the phenomenon of human behaviours
related to excessive and unhealthy use of the Internet, for ease
This article is part of the Topical Collection on Psychiatry in the Digital of communication the term IA will be used loosely, covering
Age the broad range of the aforementioned behaviours.
As there are many aspects of the phenomenon, there are
L. T. Lam
Discipline of Paediatrics and Child Health, Sydney Medical School, also different thoughts on the definition of IA and as a result,
The University of Sydney, Sydney, Australia different models of assessment based on various conceptual-
isation of the phenomenon [11, 12]. Lui and Potenza provided
L. T. Lam (*)
a historical perspective of the early development in the con-
The Hong Kong Institute of Education, 10 Lo Ping Road, Tai Po,
N.T, Hong Kong, SAR, China ceptualisation of IA from the late 1990s to the early 2000s
e-mail: ltlam@ied.edu.hk [11]. They analysed the diagnostic criteria of IA proposed by
508, Page 2 of 9 Curr Psychiatry Rep (2014) 16:508
Young which was based on the DSM-IV criteria for patholog- Methods
ical gambling characterised by three main features: preoccu-
pation, loss of control, and harmful effect [5]. Lui and Potenza Search Strategies
further noted that Shapira et al. [13] had used a broader
concept for the definition of IA as a behaviour that is uncon- The PRISMA guidelines for systematic reviews and meta-
trollable; markedly distressing, and resulting in difficulties in analysis were employed for the conduct of the literature search
terms of social life, occupation, and finance [11]. This view following a systematic and structured approach [19]. Major
was also advocated by Ha et al. when their research team medical, health, and psychological literature databases includ-
defined IA as: “the inability of individuals to control their ing PubMed, MEDLINE with Full Text, Cochrane Database
Internet use, resulting in marked distress and/or functional for Systematic Review, and PsychINFO were used and the
impairment in daily life” [14]. Such a definition was also search included all publication years (till June 2014). The
adopted by Pies in his analysis of the conceptualisation and keywords used for the systematic search were: (“excessive
the diagnosis of IA [12]. Internet use” or “problematic Internet use” or “pathological
Among all the debate on IA is the major contention of how Internet use” or “Internet addiction” or “excessive computer
well IA could fulfil the validation criteria as a psychiatric use” or “Internet gaming” or “computer gaming” or “Internet
disorder, as in other well-established behavioural addiction gaming addiction” of “ Internet gaming disorder”) AND
[15]. Robins and Guze, who first proposed a set of formal (“longitudinal study” or “ prospective study” or “cohort
criteria for establishing the validity of psychiatric diagnoses, study” or “follow-up study”). Included in the search were
suggested five criteria [16]. These included: a clear clinical articles published as reviews for the purpose of identifying
description of the disorder; evidence from laboratory studies; relevant studies. However, these review articles were not
exclusion of other disorders; evidence from follow-up studies; included in this systematic review. Limitations were imposed
and family studies [16]. Pies echoed this view by advocating on the search for publications in the English language only
his set of criteria that at least one of the three needed to be and the study subjects were adolescents loosely defined as
satisfied, namely a genetic linkage on a specific locus to the young people aged between 13 to 18 years. Upon completion
condition; a well-known aetiology or pathology; and the of the search on the electronic database, titles and abstracts of
course, prognosis, stability, and response to treatment should the identified articles were assessed for their suitability to be
be relatively consistent across different populations [12]. A included in the review. Additional searches were also con-
similar set of criteria was also put forward by Gentile et al. in ducted on other “grey” literature databases such as Google
which, first and foremost, information on the aetiological risk scholar. After assessing the titles and abstracts, full text of the
and protective factors of IA as well as the comorbidities of IA articles deemed suitable were retrieved for further examina-
were important for the determination of its validity as a psy- tion of the contents of the studies to determine their final
chiatric disorder [17]. In attempting to gather evidence for the inclusion in the review. Furthermore, the reference lists of
satisfaction of the aforementioned criteria, various studies were the selected articles were also examined for additional suitable
reported. For example, to examine the issue of comorbidities of publications that might have been overlooked in the previous
IA, Ko et al. conducted a systematic review of the literature on search.
the association between IA and psychiatric disorders [18•]. A
number of disorders were identified as comorbidities of IA. Selection Criteria
These included: substance use, attention-deficit hyperactivity
disorder, depression, and social anxiety disorder [18•]. How- The following selection criteria were applied for suitable
ever, in terms of aetiological risk and protective factors, as well articles: 1) epidemiological studies with appropriate study
as the course and prognosis of IA, information could only be design that included a cohort of participants recruited with
provided by well-designed epidemiological studies of a longi- the exposure variable assessed at baseline and then followed
tudinal or prospective nature, as highlighted as one of the for a period of time for outcome measures; 2) examined the
criteria by Robins and Guze [16]. relationship between some potential risk factors (exposure) of
The aim of this study is to examine the available informa- IA (outcome), or the effect of IA (exposure) on the health
tion through a systematic review of longitudinal or prospec- (outcome) of the participants; 3) the assessment of IA, as an
tive studies on possible risk and protective factors of IA, as exposure or an outcome variable, was conducted using a
well as the effect of IA on the health of adolescents. The validated and standardised instrument based on a sound the-
reasons for choosing adolescents as the target population of oretical framework; 4) the assessment of risk factors and
this study are twofold: 1) it has been demonstrated that IA is health was conducted using a validated and standardised
more prevalent in adolescents than in other age groups; 2) instrument; 5) provided information on the risk estimate of
there are far more studies available in this age group than in the relationship. Conference proceedings, abstracts, and non-
others. peer-reviewed journal articles were excluded from this review.
Curr Psychiatry Rep (2014) 16:508 Page 3 of 9, 508
Information Extraction and Analysis smaller sample size of about 500 [24, 26, 39•]. For assessment
of IA, the reported studies mainly used three validated instru-
Information was extracted from the selected articles and tab- ments: The Compulsive Internet Use Scale designed by Van
ulated for further analysis. Of interest was information on the Rooij et al., Young’s Internet Addiction Test, and Chen’s
relationship between potential risk or protective factors and Internet Addiction Scale. In terms of the risk factors, many
IA. These might include personal, familial, parental, Internet exposure variables were studied and could be broadly classi-
usage, psychopathologies, and personality factors. Addition- fied into three main categories: psychopathologies of the
ally, the effect of IA on the health of young people was also participants, family and parental factors, and others such as
relevant to the review. Information extracted from the selected Internet usage, motivation, and academic performance. In all
studies included the location of the study, study design, study these studies, appropriate statistical analytical approaches
sample, assessment of the exposure and outcome variables, were employed including multivariate logistic regression
analytical approach, estimated effects, and other information modelling, Cox proportional hazard regression modelling,
or remarks relevant to the study. This information was then and the Structural Equation modelling with adjustments to
summarised and analysed by the risk and protective factors of some potential confounding factors in all studies but one.
IA and the effects of IA on health separately. However, it was also noted that nearly all of these studies
had utilised schools as the sampling unit, but none considered
the effect of school in their analyses. The results of these
studies suggested a number of potential risk factors of IA.
Results For psychopathologies, Ko et al. found that Taiwanese young
people who had been assessed with attention deficit and
Following the aforementioned procedures, an extensive search hyperactivity disorder, and hostility at baseline were at a
was conducted resulting in 23 articles reporting on a longitu- higher risk of IA at follow-up with a hazard ratio of 1.72
dinal or prospective study in the English language on Internet (95 % C.I.=1.21-2.43) for the former and 1.67 (95 % C.I.=
use among adolescents [14, 20–38, 39•, 40–42]. After exam- 1.17-2.68) for the latter [25]. Cho et al. also found that
ining the abstract of these reports, 14 were possible candidates children who had been assessed with withdrawal, anxious/
for inclusion in the review. A closer examination of the full text depressed, and thought problems at childhood were also at a
of these articles revealed that only nine satisfied the selection higher risk of IA seven years later in adolescence with odds
criteria and contained sufficient information. Of these nine ratios of 1.16, 1.07, and 1.19 respectively [39•]. In terms of
articles, eight provided data on risk or protective factors of IA familial and parental factors, Ko et al. reported that conflict
and one focused solely on the effects of IA on mental health between parent and child due to the Internet at baseline was
[24–26, 28, 30, 35, 38, 39•, 41]. Among the articles reporting significantly related to IA 12 months later (OR=2.31, 95 %
on risk or protective factors of IA, a cohort of Taiwanese C.I.=1.28-4.18) [25]. At the same time, not living with mother
adolescents had been used three times resulting in three articles was also identified as a potential risk factor of IA in the same
on different risk factors [25, 35, 41]. These were included as cohort of adolescents (OR=1.66, 95 % C.I.=1.03-2.69). Other
three separate reports in the analysis. The main reason for the potential risk factors were also identified including usage of
exclusion of other articles was that they focused on other the Internet with a strong association between online gaming
Internet use but not IA. As a result, eight reports on the risk and IA (β=0.15, p<0.001) [24], and young people who had
or protective factors and IA and one on the effects of IA on adopted the Behaviour Approach System as a motivation style
mental health were included. Detailed information was extract- were found to be more at risk of IA (OR=1.05 95 % C.I.=
ed from these articles and summarised in Tables 1 and 2. 1.02-1.09) [35]. On the contrary, some protective factors were
also reported from these studies. In the study by Van den
The risk or Protective Factors of Internet Addition Eijnden et al., it was found that quality communication be-
tween parents and children on Internet use at baseline signif-
In terms of the study design, nearly all but one utilised self- icantly reduced the risk of compulsive Internet use at follow-
reported questionnaire surveys to collect baseline data and the up among young people in the Netherlands (β = -0.10,
cohort was resurveyed for the outcome measures with follow- p<0.05) [26]. A recent study in Hong Kong also revealed that
up periods ranging from 6 to 36 months. The only exception higher scores of positive youth development significantly
was the report by Cho et al. in which parents had been related to a lower risk of IA (OR=0.67) [38].
recruited as informants to provide data on the psychopathol-
ogy of their children at childhood with a follow-up period of The Effect of IA on Mental Health
7 years [39•]. The majority of the reported studies utilised a
large sample size of more than 1000 participants recruited The systematic search of longitudinal or prospective studies
from primary to junior high schools except three with a on the effects of IA on health of young people yielded only
Table 1 Information extracted from individual longitudinal or prospective cohort studies on risk factors of internet addiction in adolescents
Reference (author, year, Study design & methods Participants Exposure & outcome variable Method of analysis Variables adjusted Results Comments
place)
508, Page 4 of 9
Van den Eijnden et al. Design: Eighth grade students from 4 Exposure variable: Data were analysed using Instant messaging, depression, Instant messaging at time 1 was • The same cohort was followed
(2008), Netherlands Self-reported questionnaire schools recruited form the • Internet usage assessed by the structural equation loneliness, and compulsive positively and significantly with low attrition rate.
[24] surveys at two times. southern part of the frequency of different modelling (SEM). Of Internet use from time 1 associated with compulsive • No adjustment made for school
Follow-up duration: Netherlands (N=660) Internet usages; interest to this review was were included in the model Internet use for girls at time 2 effect.
6 months in between two • Depressive symptoms assessed the cross-lagged analysis as predictor of the same (β=0.14, p<0.05), but not for
surveys by the Depressive Mood with panel data. More four variables at time 2. males.
List; specifically was the effect of Separate analyses were
• Loneliness measured by the the exposure variable on the conducted for males and
Loneliness Scale. outcome variable. females for testing the
Outcome variable: moderating effect of
Compulsive Internet use gender.
assessed by the Compulsive
Internet Use Scale
Ko et al. (2009), Taiwan Design: Seventh grade students Exposure variable: Data were analysed using Cox Age and gender ADHD was significantly • The relationship between
[25] Self-reported questionnaire randomly selected from 10 • Attention Deficit/hyperactivity proportional hazard associated with Internet exposure and outcome
surveys at baseline and junior high schools with 4 disorders assessed by the regression model. Addiction (HR=1.72, 95 % variables was examined in a
other three time points urban, 4 suburban, and 2 ADHD Self-rated Scale; C.I.=1.21-2.43). Hostility was cohort without Internet
Follow-up duration: rural schools in southern • Depression assessed by the also associated with Internet Addiction at baseline.
6, 12, and 24 months. Taiwan in 2005. Only those CED-D; Addiction (HR=1.67, 95 % • No adjustment made for school
who were assessed without • Social phobia measured by C.I.=1.17-2.38) effect.
Internet Addiction at Fear of Negative Evaluation
baseline were included in the Scale;
analysis (N=1848) • Hostility assessed by Buss-
Durkee Hostility Inventory-
Chinese Short Form.
Outcome variable:
Internet Addiction measured by
the Chen Internet Addiction
Scale.
Van den Eijnden et al. Design: Adolescents aged 10 to 15 who Exposure variable: Data were analysed using Internet parenting practices at Quality communication about • High attrition rate at follow-up
(2010), Netherlands Self-reported online used the MSN instant Internet specific parenting structural equation T2 and Compulsive Instant use at time 1 with only 31 % responded
[26] questionnaire surveys at messenger were recruited practices assessed by a modelling (SEM). Of Internet use at T1 were significantly decreased the risk both surveys. Drop-outs
baseline in 2006 and a through an advertising newly developed Internet interest to this review was included in the analysis. of compulsive Internet use at analysis indicated that drop-
follow-up banner with 1647 parenting practice scale the cross-lagged analysis Age, gender and school time 2 (β=-0.10, p<0.05), outs were more likely to
Follow-up duration: respondents at baseline and designed and validate by the with panel data. More were also included in the follow a lower level of
6 months. 510 responded at follow-up. authors. specifically was the effect of model to control for their education than the
Response rate was 31 %. Outcome variable: the exposure variable on the effects. respondents of both surveys.
Compulsive Internet use outcome variable. • Potential sampling bias that
assessed by the Compulsive might cause problem in
Internet Use scale also generalising the results.
designed and validated by • Adjustment made for school
the authors effect.
Van Rooij et al. (2010), Design: Junior high school students age Exposure variable: The focus of analysis was on the Age, sex, education, and Changes of Internet activities, • The study aimed to investigate
Netherlands [28] 13-15 years, recruited from 9 relationship between the ethnicity except surfing and mailing, the change of Compulsive
Curr Psychiatry Rep (2014) 16:508
Table 1 (continued)
Reference (author, year, Study design & methods Participants Exposure & outcome variable Method of analysis Variables adjusted Results Comments
place)
Self-reported questionnaire secondary schools Internet usage, including game change of Internet usage were significantly associated Internet Use behaviour in
surveys in two consecutive (N=1421) playing and others, assessed over time and the change of with changes of Compulsive relationship to the change of
years in 2007 and 2008. by questions of frequency of Compulsive Internet use Internet use with the strongest Internet usage, but not a
Follow-up duration: use. over time. Data were association being online direct examination of the
12 months between two Outcome variable: analysed using multiple gaming (β=0.15, p<0.001) aetiological factors of
surveys Compulsive Internet use linear regression. Internet addictive
Curr Psychiatry Rep (2014) 16:508
Addiction at baseline.
study was a longitudinal study on healthy Chinese young
people who were depression free at baseline, with results
indicating that those who had been classified as having mod-
2009 [25].
Comments
effect.
bias.
develop depressive symptoms and scored high enough on the
Zung Depression Scale to be clinically significant in compar-
significantly related to Internet
2.31, 95 % C.I.=1.28-4.18).
1.03-2.69)
Discussion and Conclusions
Results
baseline as predictors.
exposure variables at
adolescents.
These results have shed some light on the aetiology of the
problem as well as the effect of the problem on the health,
particularly the mental health, of adolescents. It is apparent
that the aetiology of IA is rather complex and there are a range
Internet Addiction measured by
Internet Addiction measured by
of possible risk factors that may influence the onset and the
development of the problem. As indicated from the results,
family function
Exposure variable:
Outcome variable:
Internet Addiction at
Seventh grade students
data with complete
analysis (N=1630)
Self-reported questionnaire
Study design & methods
12 months.
suggesting no difference
systems. According to Jessor et al., familial and parental
retainers.
Comments
95 % C.I.=1.3-4.3), but
serious illness,
Zung Self-rating
Outcome variables:
Exposure variable:
problems including
psychiatric disorder.
follow-up
9 months.
Design:
of interest.
Lam & Peng
Human and Animal Rights and Informed Consent This article does
Reference
not contain any studies with human or animal subjects performed by any
of the authors.
508, Page 8 of 9 Curr Psychiatry Rep (2014) 16:508
38. Yu L, Shek DT (2013) Internet addiction in Hong Kong adoles- 43. Liu CY, Kuo FY (2007) A study of Internet Addiction through the
cents: a three-year longitudinal study. J Pediatr Adolesc Gynecol lens of the interpersonal theory. Cyberpsychol Behav 10:799–804
26(3 Suppl):S10–7 44. Snodgrass JG, Dengah HJ 2nd, Lacy MG, Fagan J (2013) A formal
39.• Cho SM, Sung MJ, Shin KM, Lim KY, Shin YM (2013) Does anthropological view of motivation models of problematic MMO
psychopathology in childhood predict internet addiction in male play: achievement, social, and immersion factors in the context of
adolescents? Child Psychiatry Hum Dev 44:549–55, Cho et al study culture. Transcult Psychiatry 50(2):235–62
on the aetiological relationship between childhood psychopatholo- 45. Kim H, Davis K (2009) Towards a comprehensive theory of prob-
gy and Internet Addiction in adolescents. This is the only longitu- lematic Internet use: evaluating the role of self-esteem, anxiety,
dinal study that had a follow-up period longer than 3 years and flow, and the self-rated importance of Internet activities. Comp
covered the entire population of children in a city. In fact the full Hum Behav 25:490–500
follow-up period of the cohort was 7 years since the baseline data 46. De Leo JA, Wulfert E (2013) Problematic Internet use and other
collection. Moreover, this is the only longitudinal study that risky behaviors in college students: an application of problem-
employed parents as informants instead of self-report by the par- behavior theory. Psychol Addict Behav 27:133–41
ticipants so far. The short-coming of the study was that it had a 47. Ko CH, Yen JY, Yen CF, Chen CS, Weng CC, Chen CC (2008) The
rather high drop-out rate association between Internet addiction and problematic alcohol use
40. Mittal VA, Dean DJ, Pelletier A (2013) Internet addiction, reality in adolescents: the problem behavior model. Cyberpsychol Behav
substitution and longitudinal changes in psychotic-like experiences 11:571–6
in young adults. Early Interv Psychiatry 7:261–9 48. Yen JY, Ko CH, Yen CF, Chen SH, Chung WL, Chen CC (2008)
41. Ko CH, Wang PW, Liu TL, Yen CF, Chen CS, Yen JY (2014) The Psychiatric symptoms in adolescents with Internet addiction: com-
bidirectional associations between the family factors and internet parison with substance use. Psychiatry Clin Neurosci 62:9–16
addiction among adolescents in a prospective investigation. 49. Jessor R (1987) Problem-behavior theory, psychosocial develop-
Psychiatry Clin Neurosci. doi:10.1111/pcn.12204 ment, and adolescent problem drinking. Br J Addict 82:331–42
42. Yang SJ, Stewart R, Lee JY, Kim JM, Kim SW, Shin IS, Yoon JS 50. Jessor R, Costa FM, Krueger PM, Turbin MS (2006) A develop-
(2014) Prevalence and correlates of problematic internet experi- mental study of heavy episodic drinking among college students:
ences and computer-using time: a two-year longitudinal study in the role of psychosocial and behavioral protective and risk factors. J
Korean school children. Psychiatry Investig 11:24–31 Stud Alcohol 67:86–94