Internet addiction among adolescents during the COVID-19 pandemic: Associations with sociodemographic and psychological distress

This study used a cross-sectional design and utilized a simple random sampling technique to select secondary schools in the Melaka districts that are under the administration of the Ministry of Education.

The study participants comprise students aged 13 to 17 years. Participation is voluntary, and electronic consent is obtained from parents or guardians through an anonymous, self-administered online questionnaire. Adolescents who are confirmed by their parents to be suffering from brain diseases (such as seizures) and psychiatric diseases or are undergoing psychiatric treatment and adolescents from Special Education Classes are excluded from the study. Responses lacking completeness in online questionnaires are deemed invalid and are excluded from the analysis.

Data collection tools MVIAT is used to measure the presence of IA that is described elsewhere (Young, 2015). The questionnaire contains 20 items graded on a five-point Likert scale that examine several patterns of symptoms over the last 30 days, including salience, excessive use, neglect of work, anticipation, lack of control, and neglect of social life. The maximum total achievable is 100, with a cutoff point of 43. Any result in excess of 43 classifies an individual as IA (Chong Guan et al., 2015). The MVIAT has been validated in Bahasa Melayu by Chong Guan et al. (2015) and has good internal consistency (Cronbach’s alpha = 0.91), parallel reliability (intra-class coefficient correlation = 0.88, p < 0.001) and concurrent validity with the Compulsive Internet Use Scale (Pearson’s correlation = 0.84, p < 0.00) (Chong Guan et al., 2015).

The M-DASS-21 version was used to assess the level of depression, anxiety, and stress symptoms in the reference period of the past one week. It consists of 21 items graded by a 4-point Likert scale, with each domain surveyed (depression, anxiety, stress) comprising seven questions, and the highest graded score is 21. Questions numbers 3, 5, 10, 13, 16, 17, and 21 represented depression subscales that were divided into people that do not have depression (0–5), mild (6–7), moderate (8–10), severe (11–14), and extremely severe (15 and more). Questions numbers 2, 4, 7, 9, 15, 19, and 20 represented anxiety subscales, which were divided into people that do not have anxiety (0–4), mild (5–6), moderate (7–8), severe (9–10), and extremely severe (11 and more). Questions numbers 1, 6, 8, 11, 12, 14, and 18 represented stress subscales that were divided into people that do not have stress (0–7), mild (8–9), moderate (10–13), severe (14–17), and extremely severe (18 and more). The M-DASS-21 version has been translated by Osman et al. (2014) and the validity obtained Cronbach’s alpha values of (Depression = 0.84; Anxiety = 0.74; and Stress = 0.79) respectively.

The Coronavirus Impacts Questionnaires were used to assess individuals’ experiences with the COVID-19 disease and how it had affected their lives, including on a financial scale, resource scale, and psychological scale. The questionnaire consists of 6 items and the answers are graded on a 7-point Likert scale. The scale is designed to be used in a continuous fashion, and averages for each scale for each participant will be derived by statistical analysis. The Coronavirus Impacts Questionnaire (Short) has been developed by (Conway, Woodard & Zubrod, 2020) and the questionnaire performed well in Confirmatory Factor Analyses, suggesting that it can be utilized as a separate coherent questionnaire set (CFI = .95, TLI = .93, RMSE = .10, p < .001). The short version 6-item scale also yielded high internal reliability for each component (Financial = .76, Resource = .93, Psychological = .89) that showed a similar range as such long version questionnaire (Conway, Woodard & Zubrod, 2020).

The questionnaire was distributed through representative teachers at selected schools in tandem with social media and WhatsApp telecommunication platforms for optimal coverage. Respondents were required to answer a set of questionnaires containing five major domains, starting with sociodemographic background followed by habitual internet use questions. The remaining three domain questions are based on the Malay Version Internet Addiction Test (MVIAT), the Malay Version Depression, Anxiety, and Stress Scale-21 (M-DASS-21), and the Coronavirus Impacts Questionnaires, all of which the authors have obtained permission to use from the respective copyright holders.

The data was analyzed using IBM SPSS Statistics version 23 (SPSS Inc., Armonk, NY, USA). Descriptive data is presented as frequency and percentage. Demographic characteristics, habitual internet use, level of depression, anxiety, and stress of respondents with and without IA were compared by using the Chi-square test. An independent-sample t-test was used to compare the mean of the Coronavirus Impacts scores between the types of internet users. Simple and multiple logistic regression analyses were performed to observe the relationship between the studied factors and IA. A p-value of less than 0.05 was regarded as statistically significant in this study.

The study was approved by both the Ethical Committee Board, Faculty of Medicine, Universiti Kebangsaan Malaysia (UKM) (Reference no: UKM PPI/111/8/JEP-2021-336) and the Educational Policy Planning and Research Division, Ministry of Education (MOE) Malaysia.

The mean score for the total MVIAT was 43.49 ± 12.59 (Male, 43.26 ± 11.78; Female, 43.59 ± 12.93) (Table 1). Based on the total MVIAT score, 45.5% (male = 43.9%; female = 46.1%) of the respondents were classified as having IA while the remainder, 54.5% (male = 56.1%; female = 53.9%), were considered normal internet users (NIU). The mean IAT score in the IA group was 54.65 ± 9.34, while in the NIU was 34.18 ± 5.21. There was a significant difference between age and education status among IA and NIU (p < 0.05). Meanwhile, there were no statistically significant differences in the rates of gender, ethnicity, being an only child, mother and father’s education level, and median monthly family income among both groups. Thus, the first hypothesis was only partially accepted.

Smartphones or tablets were found to be the most favorable electronic devices (IA = 95.3%; NIU = 88.2%), followed by laptops, desktops, and notebooks (IA = 4.7%; NIU = 11.8%). The frequency and duration of internet use, as well as the use of electronic devices after midnight (00:00 h), were all significantly higher in IA than in the NIU group (p < 0.001) (Table 1).

Depression, anxiety, and stress were found to have an overall prevalence of 38.4% (n = 161), 45.0% (n = 189), and 35.2% (n = 148), respectively. Table 2 shows that the prevalence of different levels of depression, anxiety, and stress was significantly higher in the IA group than in the NIU group (p < 0.001). Therefore, the third hypothesis was supported.

Interestingly, there was no significant difference in COVID-19 Impact values manifested by the IA group in comparison with the NIU group in terms of Impact Financial of 3.43 (±1.65) as compared to 3.14 (±1.69) in the NIU (p = 0.079), Impact Resource (IA = 2.84(±1.70); NIU = 2.66 (±1.57) and p = 0.246), and Impact Psychological (IA = 3.72 (±1.92); NIU = 3.66 (±1.77) and p = 0.755) (Table 3). Hence, the second hypothesis was not supported.

Table 4 shows the multiple logistic regression analysis results of the factors involved in the development of IA. Age, depression, stress, and smartphones were identified as significant associated factors with IA. The odds of IA in older adolescents were 1.16 times higher than in younger adolescents (aOR = 1.16, 95% CI [1.00–1.35], p = 0.048). The odds of IA in adolescents with mild or moderate depression domains of the DASS-21 score were 2.43 times higher than those who had a normal score (aOR = 2.43, 95% CI [1.36–4.34], p = 0.003). The odds of IA in adolescents with severe or extremely severe stress domains of the DASS-21 score were 6.41 times higher than those who had a normal score (aOR = 6.41, 95% CI [2.18–18.82], p = 0.001). Adolescents who use smartphones or tablets were 3.52 times more likely to develop IA as compared to those using other devices (aOR = 3.52, 95% CI [1.43–8.67], p = 0.006).

Pre-pandemic studies using similar diagnostic tools of IAT revealed that the overall prevalence of IA ranged from 2.6% in Northern and Western Europe to 10.9% in Middle Eastern countries (Cheng & Li, 2014). Meanwhile, the prevalence of IA was between 1.0 and 9.0% among children and adolescents in Europe, 1.0 to 12.0% in the Middle East, and 2.0 to 18.0% in Asia (Christakis, 2010). Local data in Malaysia, however, demonstrated a higher range of IA, ranging from 2.4% to 49.2% (Abdul Aziz et al., 2018; Lai et al., 2015). Specifically, in the region of Malacca, the incidence of IA stands at 29.4% (Ministry of Health Malaysia, 2017). This indicates a substantial variation in IA rates within the country, emphasizing the importance of localized studies to capture the nuances of internet usage patterns and associated addictive behaviors. Understanding the prevalence of IA at the regional level, such as in Malacca, contributes to more targeted and effective interventions to address the specific needs of the local population in managing and preventing internet-related addictive behaviors.

Meanwhile, recent studies conducted amid the pandemic indicate a notable rise in the occurrence of IA compared to pre-pandemic rates, with figures ranging between 24.4% and 46.8% in various countries (Dong et al., 2020; Lin, 2020; Priego-Parra et al., 2020; Sun et al., 2020). The present study adds to the increasing amount of evidence by demonstrating that 45.5% of the respondents were classified as addicted users. Interestingly, this surge in IA during the pandemic aligns with historical patterns observed in previous disaster events such as during the severe acute respiratory syndrome (SARS) epidemic, natural disasters, and war (Dimaggio, Galea & Li, 2009; Lin, 2020).

The surge in IA during the COVID-19 pandemic can be attributed to several interconnected factors. Firstly, the finding that the lockdown period had a higher prevalence rate of overall behavioral addiction (e.g., internet addiction, gaming addiction, gambling addiction, smartphone addiction, social media addiction) than the non-lockdown period suggests that the advancement of the internet and individuals’ coping strategies during the lockdown played significant roles. Specifically, the lockdown measures may have heightened individuals’ psychological distress, prompting some to resort to potentially addictive behaviors as coping mechanisms (Alimoradi et al., 2022). Additionally, the closure of schools and the shift to online learning due to COVID-19 have transformed adolescents’ educational experiences, resulting in heightened use of digital devices and increased online activities (Besalti & Satıcı, 2022; Kusnadi & Kaloeti, 2022). With limited outdoor activities and reduced face-to-face interactions, adolescents have redirected their focus to online pursuits, encompassing social networking, online gaming, entertainment and educational tasks (Onukwuli et al., 2023; Patel, Patel & Patel, 2022; Putra, Fithriyah & Zahra, 2023; Woon, Daud & Razak, 2021).

The finding has shown that the occurrence of IA increased with age, as seen in the current study population aged 13 (13.6%) as compared with aged 17 (43.5%). The findings of this study’s regression analysis indicated a statistically significant association between age and internet addiction, with the likelihood of being classified as an addictive internet user increasing by 16% for each additional unit in age. (aOR = 1.16, 95% CI [1.00–1.35], p = 0.048). This result aligns with the observations made in seven European countries and Taiwan, which also indicate that the prevalence of dysfunctional internet behavior is greater among adolescents aged 16–17 in comparison to those aged 14–15 (Lin, 2020; Tsitsika et al., 2014). Malaysia, on the other hand, has described the prevalence of IA as increasing by 38% among 13 to 17 years old students (Ministry of Health Malaysia, 2018), and (Ooi et al., 2020) suggested that age is a significant predictor of IA as the prevalence would increase with the increasing age of the user. One plausible explanation for the observed association between age and addictive internet use may be attributed to the considerable academic pressure and studying load faced by older students in preparation for a more established future career pathway; thus, older adolescents utilize the internet more than younger ones to ensure better understanding and efficacy of learning (Chen et al., 2022; Hu et al., 2021).

Gender, on the other hand, has revealed no specific sex-related predominantity in this current finding when compared to other previous literature. The possible reasons for the gender difference observed are different online activities, duration of online surfing, and dissimilar self-control and emotional regulation between boys and girls that distinguish the two groups’ proneness to become addicted to the internet. (Dong et al., 2020; Liu et al., 2011) also emphasized the gender-linked pattern, suggested that girl’s maturity reflects a greater understanding of their actions and consequences, resulting in a higher prevalence of IA among female respondents.

A dissimilar pattern seen in the present study reflects equal habitual online activities in both genders in terms of types of electronic devices used, frequency of electronic device use, hours of surfing the internet, and frequency of online activity after midnight. Moreover, it is worth noting that these habitual internet use factors were all significantly higher among IA groups than among NIU groups. As such, the pandemic situation altered the pattern of online surfing due to a variety of factors, including various psychological effects and stress, a high level of virtual support, and lower functioning families experienced by adolescents. These factors all contributed significantly to the prediction of IA (Lin, 2020).

Research indicates a reciprocal relationship between addiction behaviors, such as internet addiction, and mental symptoms (Afifi et al., 2016). Psychiatric symptoms and disorders can instigate addictive behaviors, while engaging in addictive behaviors can also elevate the risk of experiencing psychiatric symptoms and disorders (Cai et al., 2023). The current study unveiled those individuals in the IA groups showed significantly higher levels of depression, anxiety, and stress during assessments compared to those in the NIU group. From the regression analysis, mild or moderate depression (aOR = 2.43, 95% CI [1.36–4.34], p = 0.003), (OR = 4.50, p < 0.001), severe or extremely severe stress (aOR = 6.41, 95% CI [2.18–18.82], p = 0.001) (OR = 7.03, p = 0.001), and anxiety (OR = 4.66, p < 0.001) were significantly related with IA. Indeed, the confirmation of a significant positive relationship between depression, anxiety, and stress with IA in our analysis aligns with numerous prior studies in the literature. An abundance of empirical research, spanning from cross-sectional studies (Abdul Aziz et al., 2018; Dong et al., 2011; Lai et al., 2015; Li et al., 2019) to longitudinal research, strongly supports the notion that increased time spent on internet either smartphones, social media, or gaming has not only worsened psychological distress (Chen et al., 2021; Chen et al., 2022), but pre-existing psychological distress also significantly and positively predicts IA (Zhao et al., 2023) and positively predicted internet Gaming Disorder severity (Teng et al., 2021).

Primary school students in China have documented notably elevated levels of psychological distress throughout the COVID-19 pandemic in comparison to the period preceding the outbreak due to their increased time spent on internet-related activities (Chen et al., 2021). The increased psychological distress that was noted during the pandemic period may be ascribed to the concerning rates of COVID-19 transmission and mortality that were reported (Ahorsu et al., 2020; Chen et al., 2021). As people contend with the continuous cascade of distressing news and the concrete consequences of the virus on their localities, emotions of apprehension, dread, and unpredictability inherently heighten. Adolescents were vulnerable to the overabundance of infodemic and negative feelings, such as fear and worries that were constantly spread and expressed on social media. In due course, the information obtained may intensify stress responses, heightened anxiety, and depression, increase their concern towards the situation, resulting in media consumption and further distress. This cycle is vicious and difficult to break, as the internet is always applicable, and information readiness is able to feed the curiosity enquired by adolescents (Biedroń et al. 2021; Priego-Parra et al., 2020). Meanwhile, Hu et al. (2021) suggested that IA among adolescents was the result of stressors induced by the pandemic situation as a maladaptive coping mechanism to overcome these negative emotions, which had to occur via the presence of a second mediator: learning burnout. Learning burnout develops when adolescents are kept socially distant at home due to fear of COVID-19 transmission, thus increasing online engagement that leads to negative effects on their academic performance and daily learning process. Subsequently, it promotes emotional disturbance among the vulnerable (Hu et al., 2021).

The current investigation nevertheless failed to establish a significant association between pandemic stressors, including impact financial, impact resource, and impact psychological with IA. This finding suggests that, in the context of the examined population, the heightened stressors brought about by the pandemic may not directly contribute to an increased likelihood of developing IA. Several factors could contribute to this lack of significant association. It’s possible that other individual, social, or environmental factors play a more prominent role in the development of IA in this specific population, mitigating the impact of pandemic-related stressors. Additionally, coping mechanisms, resilience, and pre-existing conditions may influence how individuals respond to stressors and their susceptibility to IA.

Our multivariate analysis confirmed that the use of smartphones or tablets has greater odds of IA. The types of devices are usually related to the likelihood of a certain individual remaining online, either for academic purposes or recreational use. According to Dong et al. (2020), the primary internet devices used among internet-addicted users (90.9%) during the COVID-19 pandemic were smartphones and tablets. Although desktop computers are commonly used by students for online learning, their inclination towards smartphones is frequently influenced by an extensive array of user-friendly attributes. Young students find smartphones to be an appealing and practical educational aid due to their inherent portability, accessibility, and the wide variety of learning applications that are available to them (Chen et al., 2021; Uther & Ylinen, 2018). Pre-pandemically, consistent findings yielded locally have shown smartphones as the device of choice for internet access, particularly adolescents, and the wide availability of electronic devices was reported to have significantly contributed to the soaring number of IA (Ooi et al., 2020).

The advantage of this study came from its considerably large sample size, involving 420 respondents. Furthermore, the direct focus of this study on the prevalence of IA and its associated factors at the time of the COVID-19 pandemic has supplemented new knowledge and information, particularly in the context of adolescence in this country. Nevertheless, several limitations highlighted above can be used as added value in future related studies. Firstly, this study was only conducted among adolescents in Malacca state, which means it could not represent the large proportion of the adolescent population in Malaysia as a whole. Secondly, the cross-sectional study design used in this survey does not permit the cause and effect to be shown from the risk factors for IA among adolescents. Thirdly, the definite diagnostic criterion of IA is still lacking globally, and the use of the Malay version of Young’s internet Addiction Test is based on the DSM-IV addiction criteria, which in this study has set a cutoff point that may differ from other countries. There are also different assessment tools available that yield varying results accordingly. Considering the limitations listed, recent findings should be interpreted cautiously. Despite that, this study carries an equal weight of benefits for future research and intervention planning.

The following information was supplied regarding data availability:

The raw data is available in the. Supplemental File