How Has the COVID-19 Pandemic Impacted Internet Use Behaviors and Facilitated Problematic Internet Use? A Bangladeshi Study

Introduction

The outbreak of the coronavirus disease 2019 (COVID-19) has already spread across the entire world and has curtailed most individuals’ daily life activities and movements. In response to mitigate the ongoing pandemic, the authority of Bangladesh (where the present study was carried out) rapidly took some preventive and control strategies such as home confinement, closing down all educational institutions and implementing online learning, closing non-essential businesses, and imposing and mandatory spatial distancing.^1,2 Such measures, such as staying confined at home for a long time, can lead to negative psychological states and psychological vulnerability because of (i) loneliness due to reduced social interaction, (ii) fear of losing family members or loved ones to the virus, (iii) uncertainty of future or careers, and (iv) despair due to social and economic disruption.^3–5

Like other psychological impacts, the ongoing “stay-at-home” and confinement situation appears to have facilitated individuals’ increased engagement with technology.⁶ For example, an Indian study reported that 67.2% of participants reported an increase in their internet use since the start of the pandemic.⁷ However, higher engagement with technology use might become problematic or addictive for some individuals.^8,9 Additionally, problematic internet engagement is also associated with loneliness, and various psychological and mental health issues, all of which may be heightened by the ongoing pandemic.^10,11 For instance, a recent case report highlighted uncontrolled PUBG-gaming apparently lead to suicide in Pakistan during the ongoing pandemic.¹² Additionally, previous research has indicated that problematic internet use is associated with mental illness such as depression, anxiety, stress, and sleep problems,^9,13,14 and these mental health disorders have also increased during the ongoing COVID-19 pandemic.⁵ Psychoactive substance use and behaviors such as using social media, video gaming, surfing the internet, and watching sexually explicit material are all frequently used for relieving psychological distress (eg, daily life stressors, problems, and difficulties) in the form of “escapism”.^15–17

However, in line with the ongoing stressful situation caused by the COVID-19 pandemic, mental health studies have been conducted mostly assessing psychological disorders and issues. Extreme mental health impacts (ie, suicidal behavior) have been associated with problematic internet use particularly in relation to online gaming,¹² which is also consistent with a few reports prior to the non-COVID-19 pandemic.^18,19 However, only a few studies have examined the impact of problematic internet use in the context of the pandemic. For instance, a Mexican study reported that 2% of those surveyed might have had internet addiction (ie, scoring ≥70 [out of 100] on the Internet Addiction Test),²⁰ whereas an internet addiction prevalence rate of 14.4% was reported among Indonesian individuals (≥108 out of a total 264 score on a self-developed scale).²¹ However, given the low cutoff score reported to indicate internet addiction, the findings do not appear to have good face validity. In China, a prevalence rate of 2.68% for internet addiction was reported (score on the Internet Addiction Test [IAT] ≥ 70).²² A study was also conducted in Bangladesh during the COVID-19 pandemic assessing problematic internet use predictors using the nine-item Internet Disorder Scale-Short Form (IDS9-SF), but did not report the prevalence of internet addiction.²³ Therefore, the present study investigated changes regarding problematic internet use comprising a Bangladeshi student’s sample. A student sample was used in the present study because all previous studies conducted prior to the pandemic in Bangladesh had used student samples [see Griffiths and Mamun²⁴ for a very recent review on internet addiction-related studies in Bangladesh], and the present authors wanted to compare the present findings with those of previous studies in the country.

Methods

Study Procedure and Participants

A cross-sectional study was carried out among Bangladeshi students between October 7 and November 2 (2020) utilizing an online-based data collection platform (ie, Google Forms). A structured questionnaire was developed following previous studies conducted in Bangladesh, which were circulated on social media. To participate in the survey, inclusion criteria were being a Bangladeshi student (high school or above), having internet access, and an interest in participating the study. The sample size was calculated based on the following formula which estimated a sample size of 385. Utilizing a convenience sampling approach, a total of 617 individuals initially began completing the survey, and after removing incomplete questionnaires, 601 participants’ data were analyzed in the final sample. Therefore, the sample size was more than adequate.

[Here: N = population size, infinite; e= Margin of error, 0.05; z = z-score, 1.96 (95% confidence level)]

Ethics

Study participation was voluntary, and online informed consent was taken from the respondents by exploring the study objectives. Additionally, the confidentiality and anonymity of the data were also assured to them while taking part in the survey. Following the Helsinki Declaration 2013, the study protocol was approved for implementation by the Institute of Allergy and Clinical Immunology of Bangladesh [Reference: IRBIACIB/CEC/03202030]).

Measures

Sociodemographic Factors

Basic sociodemographic information was collected in the survey, including gender, educational status (eg, university, medical college, high school), present residence (eg, urban or rural), relationship status (ie, single, in a relationship, married), monthly family income [eg, lower-class = less than 15,000 BDT, middle class = 15,001–30,000 BDT, upper class = more than 30,000 BDT; based on Mamun et al⁹] and type of family (eg, nuclear or extended family). Additionally, participants were also asked if they were currently living with their families or not.

Behavioral Health-Related Measures

The survey included behavioral health-related variables, including cigarette smoking status, drug use status, sleep status, and physical exercise. For assessing sleeping patterns, the study followed prior Bangladeshi studies comprising three categories (eg, normal sleeping status = 6–7 hours¹⁰). Physical exercise was defined as walking, cycling, swimming, or other activities for at least 30 minutes daily. Perceived health status was assessed by asking participants whether they suffered from a number of illnesses on a list (eg, asthma, heart problems, kidney problems, diabetes, etc.).

Online Use Behaviors

Several online use behaviors were assessed in the present study. Considering the prior Bangladeshi studies, the duration of online use was assessed utilizing categories (eg, less than 2 hours, 2 to 3 hours, 4 to 5 hours, and more than 5 hours). The online activities included educational activities, chatting/texting, online gaming, watching/streaming videos/films, social media browsing, watching sexual materials/pornography, and online shopping.

Smartphone Addiction

The Smartphone Application-Based Addiction Scale was used for assessing the risk of smartphone addiction.²⁵ The scale comprises six items (eg, “My smartphone is the most important thing in my life”), which are responded based on a 6-point Likert scale from 1 (strongly disagree) to 6 (strongly agree). The total score ranges from 6 to 36. Based on previous recommendations, the risk of smartphone addiction was determined using a cutoff of 21 out of 36.²⁶ In the present study, the Cronbach’s alpha was 0.70.

Facebook Addiction

The Bergen Facebook Addiction Scale was used for assessing the risk of Facebook addiction.²⁷ The scale comprises six items (eg, “How often in the last year have you spent a lot of time thinking about Facebook or planned use of Facebook?”), which are responded to on a 5-point Likert scale from 1 (very rarely) to 5 (very often). The total score ranges from 6 to 30, where ≥18 was considered as the cutoff point for being at risk of Facebook addiction.²⁷ In the present study, the Cronbach’s alpha was 0.84.

Depression

The two-item Patient Health Questionnaire (PHQ-2) was used for assessing the presence of depression. Participants are asked how often they experienced the two core criteria for depressive disorders over the past two weeks (ie, “Little interest or pleasure in doing things”, and “Feeling down, depressed, or hopeless”), which are responded to on a 4-point Likert scale (0= not at all, 1= several days, 2=more than half the days, 3=nearly every day).^28,29 The total score ranges from 0 to 6, where ≥3 was considered as the cutoff point indicating the presence of depression.²⁸ In the present study, the Cronbach’s alpha was 0.77.

Anxiety

The two-item Generalized Anxiety Disorder (GAD-2) scale was used for assessing the presence of anxiety. Participants are asked how often they experienced the two core criteria for anxiety disorders over the past two weeks (ie, “Feeling nervous, anxious or on edge”, and “Not being able to stop or control worrying”), which are responded to on a 4-point Likert scale (0= not at all, 1= several days, 2=more than half the days, 3=nearly every day).^28,30 The total score ranges from 0 to 6, where ≥3 was considered as the cutoff point indicating the presence of anxiety.²⁸ In the present study, the Cronbach’s alpha was 0.77.

Internet Addiction

Young’s Internet Addiction Test (IAT) was used for assessing the risk of internet addiction. The scale comprises 20 items (eg, “Do you choose to spend more time online over going out with others”) which are responded to on a 6-point Likert scale from 0 (Not applicable) to 5 (Always).³¹ The total score ranges from 20 to 100. As the prior Bangladeshi studies used different cutoff scores, the present study followed these schemes for a better comparison even though they are not consistent across studies and different operational definitions apply to different cutoffs. At least four cutoff classification systems have been used to assess problematic internet use in Bangladesh. The first set of cutoff scores were those originally reported by Widyanto and McMurran:³¹ <20 [absence of addiction], 20–39 [low level of addiction and average online user], 40–69 [moderate addiction], and 70–100 [severe internet addiction]. The second cutoff score was ≥50 for “problematic internet use”.⁹ The third cutoff score was ≥60 for “excessive internet use”.³² Finally, the fourth cutoff score was ≥80 for “severe internet dependency”.⁸ In the present study, the Cronbach’s alpha was 0.91.

Statistical Analysis

From the responses in the Google Forms, the data were coded and prepared for final analysis in Microsoft Excel 2019. Formal analyses were performed by the IBM SPSS Statistics version 25. Descriptive statistics such as frequencies, percentages, means, and standard deviations were calculated. One-way ANOVAs were carried-out to identify if there were any significant IAT mean score differences within the studied variables. The p-value for significance was p<0.01. Finally, socio-demographic and behavioral health-related variables, online use behaviors, smartphone addiction, Facebook addiction, depression and anxiety were included in the hierarchical regression analyses with problematic internet use as the dependent variable. The normality of distribution (skewness and kurtosis values) and multicollinearity (VIF and tolerance values) were tested, and no issues were found.

Results

Characteristics of the Participants

In the total sample (N=601), more than half of the respondents were male students (57.2%) and 65.2% reported that they were currently studying at university. A larger proportion of the participants came from a nuclear family (78.0%), were single in relationship status (79.5%), and lived with the family (87.0%) during the time of the survey. Additionally, 44.6% belonged to a family having more than 30,000 BDT monthly family income (ie, upper class). Half of the participants performed physical activity, and 10.2% suffered from chronic illnesses. More than half of the participants reported using the internet for more than five hours every day (53.2%). Most participants reported using the internet for texting or communication (96.7%), social media browsing (95.5%), video streaming (92.5%), and engaging in educational purposes (84.2%). Using the thresholds outlined in the “Measures” section, a large proportion of the sample was reported as being at risk of problematic smartphone use (86.9%) and problematic Facebook use (39.4%) although the cutoffs for both instruments were arguably very low. Finally, approximately one-third of the sample reported as being at risk of probable depression (43.3%) and anxiety (32.6%) (Table 1).

Table 1 Distribution of the Studied Variables with Problematic Internet Use Score

Prevalence Rates of Problematic Internet Use

A total of 4% of the sample were classed as being at risk of internet addiction (ie, “severe internet dependency” scoring over ≥80 out of 100 on the IAT), whereas 49.1% scored as being problematic internet users (scoring ≥50 out of 100 on the IAT).

Problematic Internet Use Within the Studied Variables

Table 1 presents the distribution of the variables with problematic internet use (here, PIU is a continuous variable based on IAT score). There was no significant gender difference in problematic internet use scores (p=0.653). In relation to student status, medical students had higher problematic internet use scores compared to university and high school students (p<0.001). Similarly, students in a relationship were significantly more likely to be problematic internet users, followed by single and married participants (54.98 [SD±16.70], 50.29 [SD±15.97], and 42.01 [SD±16.17], respectively; p<0.001). The problematic internet use score was also reported higher among participants who did not exercise regularly (54.14 [SD±15.40] vs 45.80 [SD±16.18]; p<0.001) The more time spent online, the more likely individuals were of being problematic internet users (p<0.001). In relation to types of online use, messaging/chatting (p<0.001), gaming (p=0.018), video streaming (p=0.051), social media use (p<0.001), news sites surfing (p=0.003), and other (eg, job searching, scholarship searching, etc.) (p=0.047) were significantly associated with problematic internet use. Additionally, all of the psychopathological variables, including smartphone addiction, Facebook addiction, depression, and anxiety, were significantly associated with problematic internet use (Table 1).

Correlations of the Variables with Problematic Internet Use

Table 2 shows the correlation matrix of the continuous variables with internet addiction. All the variables showed a significant positive correlation. IA was significantly associated with smartphone addiction (r= 0.608), Facebook addiction (r= 0.762), depression (r = 0.509), and anxiety (r= 0.536).

Table 2 Correlations Among Selected Continuous Variables

Predictive Models for Problematic Internet Use

Table 3 presents four models predicting problematic internet use, which were analyzed by using multiple hierarchical regression. Model 1 included only socio-demographic variables, whereas behavioral health-related variables were added with socio-demographics in Model 2. Model 3 considered socio-demographic, behavioral health-related variables and online activities, and the final model (ie, Model 4) additionally added psychopathological variables. All models were associated with problematic internet use except Model 1 (p=0.404). Model 2 explained 8.2% of the variance for problematic internet use. This variance rose to 26.2% in Model 3 after online use behaviors were added. The final model explained 70.6% of the variance for problematic internet use after smartphone addiction, Facebook addiction, depression was added (Table 3).

Table 3 Predictive Models for Problematic Internet Use

Discussion

Over the past two decades, the internet has become essential in people’s daily lives. In the COVID-19 context, the internet is being used as the main source of COVID-19-related information and suggests there has been greater engagement with the internet than prior to the pandemic.³³ Additionally, internet engagement has increased among students because face-to-face interaction and activities have been restricted. More specifically, students have had to engage in online teaching, and because of the reduced face-to-face contact, are more likely to be engaging in other online activities such as social media use and online gaming.^16,34 Therefore, problematic internet-related coping behavior appears to have increased during the COVID-19 pandemic, leading to a greater risk of internet addiction across different cohorts.^34–36 One study reported that the prevalence of severe internet dependence rose 23% during COVID-19 pandemic.³³

Table 4 provides a comparison of results in the present study with online use behaviors in previous Bangladeshi studies assessing similar variables. The findings indicate that internet use in the present study appears to have greatly increased during the COVID-19 pandemic. For instance, 53.2% of the participants in the present study reported using the internet more than five hours daily, compared to 20.7% in a previous study.⁹ Similarly, using the internet for educational purposes, instant messaging, video streaming, and social media browsing were higher than in previous studies, although the amount of time spent on gaming online and shopping online was lower during the lockdown compared to other studies.^8–10,37 Due to home confinement, there may be increased face-to-face interaction with their family members and/or parents may have monitored their children’s online activity more than usual, which may have decreased the amount of time their children spent gaming online. In addition, given the ongoing economic disruption and crisis due to the pandemic, it is not surprising that a lower proportion of the participants spent time online shopping than before the pandemic.

Table 4 Comparison of Results in the Present Study with the Bangladeshi Studies Carried Out Prior to the COVID-19 Pandemic

The present study found 49.1% of the participants were classed as being problematic internet users (scoring 50 or more out of 100 on the IAT), and 4% were classed as being at risk of being addicted to the internet (ie, scoring over ≥80 on IAT). The main problem in trying to make comparisons across studies is that all the studies (i) comprise self-selected samples, (ii) comprise different cohort samples, and (iii) even when using the same instrument (mostly the IAT), the study authors used different cutoffs.^24,38 Table 5 provides a comparison of problematic internet use prevalence rates between the present study and the previous ones despite these problems. As aforementioned, based on the present study’s findings, internet engagement appears to have been increased as of the ongoing pandemic, and that may have resulted in an increase in problematic internet use.

Table 5 Comparison of the Problematic Internet Use Rates with the Prior Bangladeshi Studies Prior to the COVID-19 Pandemic (Adapted from Griffiths and Mamun24)

Problematic use of the internet can disrupt individuals’ quality of life and reduce the amount of time spent on offline social activities and educational/occupational duties. Severe dependency on internet use may also lead to adverse psychological consequences.^13,16 The present study found that using the internet for educational purposes was not significantly associated with problematic internet use, which reflects their controlled use. However, other online activities such as texting, social media browsing, and watching online videos were significantly associated with problematic internet use. Unsurprisingly, smartphone addiction and Facebook addiction were independently reported to be the significant risk predictors of problematic internet use and is the first time that this has been reported in Bangladesh studies. Other psychological variables, such as depression and anxiety were risk factors of problematic internet use as has been reported in the previous Bangladeshi studies.^9,39 Such findings were expected given that individuals frequently engage in excessive internet use to cope up with and help alleviate psychological distress.^17,32

One previous Bangladeshi study examining adolescents, reported that 36% of the variance of problematic internet use was explained by demographic factors (ie, academic performance, pocket money, father’s and mother’s education, and mother’s occupation) and internet use behavior variables (ie, weekly use frequency, daily use frequency, place where internet is used, device on which internet is used, most frequently used apps, and frequent internet activities), and which increased to 43% variance after adjusting for parental mediation.⁸ Another study assessing problematic internet use during the COVID-19 pandemic found that only 6% variance of problematic internet use was explained by socio-demographic factors (ie, educational qualification, marital status, and family type), lifestyle factors (ie, smoking status, sleeping hours, physical exercise, doing household chores), online use behavior-related factors (ie, internet using hours, playing online games, social media purposes, and recreational activities).²³ The present study found 26.2% variance of problematic internet use was explained by socio-demographic factors, behavioral health-related variables, and online use behaviors. The variance for problematic internet use was 70.6% when smartphone addiction, Facebook addiction, depression, and anxiety were added to the model.

The present study has a number of limitations that should be considered when interpreting the findings. The study was (i) cross-sectional, (ii) comprised online self-reporting data, and (iii) comprised a non-representative student sample. Additionally, other factors (such as the degree of loneliness due to the lockdown, sleep problems, etc.) were not considered in this study.⁴⁰ All of these factors limit the generalizability of the findings. Future research would benefit from longitudinal research using more representative samples.

Conclusion

The present study provides a comparison of Bangladeshi internet use behaviors and problematic internet use prevalence rates during the COVID-19 pandemic with the prior studies. Based on the present findings, it appears that the ongoing COVID-19 pandemic has increased the prevalence of problematic internet use behaviors. Therefore, health-and-control use strategies and programs should be implemented to decrease problematic internet use among vulnerable individuals to problematic internet use. Educational institutes should implement interventions to reduce problematic internet use among the students.