Structure and stability of internet gaming disorder from childhood to late adolescence: A 5‐wave birth cohort study

BACKGROUND AND AIMS: The natural course and longer-term stability of internet gaming disorder (IGD) and its symptoms are largely unknown, yet such knowledge is essential to aid prevention, treatment and policy development. This study therefore examined four aspects of IGD symptom stability: of form, at the group level, rank-order and of within-person changes. DESIGN: Birth-cohort study. SETTING: Trondheim, Norway. PARTICIPANTS: A stratified sample (n = 812, 50.9% girls) of the 2003/2004 birth cohorts was examined biennially over five waves from age 10 to 18 (n = 3236 observations). MEASUREMENTS: IGD was assessed through a semi-structured clinical interview. Latent growth curves were employed to assess the stability of symptoms at the group level, and random intercept cross-lagged models were used to determine how changes in symptoms predicted subsequent changes in symptoms at the within-person level. FINDINGS: IGD symptoms loaded on two factors, termed 'Heavy involvement' and 'Negative consequences,' consistently across ages and sex. Strong measurement invariance was partly achieved across ages. The point prevalence of IGD diagnosis varied between 1.1%-2.2%, and 5.6% [95% confidence interval (CI) = 4.0-7.3) had an IGD diagnosis at least once across ages 10-18 (boys: 10.2%, CI = 6.9-13.5; girls: 1.6%, CI = 0.5-2.7). Symptoms capturing Heavy involvement increased slightly from age 10-16 and then declined sharply at age 18, whereas symptoms representing Negative consequences remained stable. Regarding stability relative to the group, moderate two-year stability was observed (Heavy involvement r = 0.33 to 0.56, P < 0.001; Negative consequences r = 0.19, P < 0.01 to 0.37, P < 0.001) with increasing stability with age. The stability from age 10-16 and 10-18 was weak for Heavy involvement (r = 0.14 and 0.17, P < 0.05) and absent for Negative consequences (r = 0.09 to 0.11, P > 0.05). Increases in Heavy involvement predicted further increases in Heavy involvement two years later (β = 0.17, P < 0.01 to 0.39, P < 0.001) and increased Negative consequences at ages 14 and 18 (β = 0.29 and 0.28, P < 0.001). CONCLUSIONS: Although the point prevalence of internet gaming disorder (IGD) diagnosis in the 2003/2004 birth cohorts of Trondheim, Norway, was low, 10% of boys fulfilled the criteria for a diagnosis at least once during preadolescence and adolescence. IGD symptoms at age 10 were only weakly related to IGD symptoms in mid- and late adolescence. However, as Heavy involvement symptoms predicted later increases in Negative consequences symptoms from age 12 onwards, early adolescence may offer a window of opportunity for when interventions might intersect a development toward a full-blown IGD diagnosis before symptoms become more intertwined with additional problems and resistant to change.