Study design, participants and procedure

In this cross-sectional study, the study participants were those students who were taking the university entrance test at Jahangirnagar University, Dhaka, Bangladesh. The entrance test was held between June 18 to 25, 2023, and data collection took place within this time frame. The participants deemed eligible for inclusion in this study were those individuals partaking in the university entrance test while concurrently residing in the university dormitories during the stipulated entrance test period. Students were approached for participating in this study at night who resided in the dormitories. The selection of the participants was facilitated through a convenience sampling approach, wherein all students present during the data collection phase were approached for solicit their participation. A noteworthy response was garnered from a total of 2,533 students who actively contributed to the study by responding to the survey. A key facet of the recruitment process involved a strategic briefing session conducted by the research team prior to participants’ active involvement. This session served the purpose of acquainting participants with the intricate terms and concepts embedded within the survey questionnaire, as well as on the study’s aims and objectives, potential benefits, and risks. However, after eliminating incomplete questionnaires, data from 2,157 participants were kept for final analysis.

MeasuresSociodemographic factors

Sociodemographic information, including variables such as gender, permanent residence (rural and urban), religion, family type, and monthly family income. Participant’s family was categorized into three distinct groups: those with incomes less than 15,000 Bangladeshi Taka (BDT), those falling within the 15,000–30,000 BDT range, and those with monthly family incomes exceeding 30,000 BDT as following the previous study within the similar population [17]. For religion, others represent those studies belonging to a religion other than Muslim, that is, Hindu, Christianity, Buddhism, etc.

Admission-related variables

In Bangladesh, it is a common practice for most universities to permit students to attempt the entrance test at most twice. Therefore, data regarding the participants’ test-taking status were collected, distinguishing whether they were first-time test takers or had appeared on the test for a second time. Additionally, information related to their educational background in high school (Science, Commerce, and Arts), and GPA of their previous public exams at high schools were collected. Participants were queried about whether they had sought guidance from professionals or coaching centers during their test preparation, as well as, their satisfaction on mock tests was recorded. Their average monthly expenditures related to test preparation and the specific type of university to which they aspired for admission were also asked.

Digital device-related variables

Digital device-related information is primarily divided into two key categories: the utilization of digital media platforms and the assessment of screen time. In particular, participants were asked if they used a number of digital devices or gadgets including TV, PC, Smartphone, and Gaming device. Whereas, the screen time of different digital activities was collected and categorized as per a previous study conducted by Garmy et al. [18]. And following this study, students were classified into two clear-cut groups: those who exceeded 2 h of daily digital device usage and those who maintained a usage duration of less than or equal to 2 h per day.

Patient health questionnaire

The Patient Health Questionnaire (PHQ-9) was used to evaluate levels of depression [19]. The PHQ-9 comprises of nine distinct items, each requiring respondents to indicate their experiences over the past two weeks using a four-point Likert scale (0 = not at all, 1 = several days, 2 = more than half of the days, and 3 = nearly every day). The cumulative scale ranges from 0 to 27, with higher scores signifying greater severity of depression [19]. A score equal to or more than 10 is typically indicative of depression. For this study, Cronbach’s alpha coefficient, which measures internal consistency, demonstrated a good level of reliability at 0.83.

Generalized anxiety disorder

The Generalized Anxiety Disorder (GAD-7) was used to evaluate levels of anxiety [20]. The GAD-7 comprises a total of seven items, with participants responding to them using a four-point Likert scale (0 = not at all, 1 = several days, 2 = more than half of the days, and 3 = nearly every day), reflecting their experiences over the past two weeks. The overall scale ranges from 0 to 21, and higher scores are indicative of more severe anxiety. Typically, a score equal to or more than 10 is considered an indicator of anxiety [20]. For this study, Cronbach’s alpha coefficient, which measures internal consistency, demonstrated an excellent level of reliability at 0.90.

Digital addiction scale

The Digital Addiction Scale (DAS) was developed using a modified version of Bergen Facebook Addiction Scale [21]. In this study, “Facebook” was replaced with digital devices and/or gadgets for each of the items, e.g., “Used digital device in order to forget about personal problems?”. For this modification, we performed Exploratory Factor Analysis (EFA), and Confirmatory Factor Analysis (CFA). The KMO measure of sampling adequacy was 0.845 (p < 0.001). In the EFA, the six items have explained a 48.68% variance. In CFA, the goodness of fit indices generated excellent values (χ2/df = 5.89, RMSEA = 0.04 (90% CI [0.03, 0.06]), SRMR = 0.02, CFI = 0.98, NFI = 0.98, GFI = 0.99]. Both EFA and CFA results suggest that the scale is valid and reliable in assessing digital addiction among university students. The items 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. The cutoff point for the scale was determined using the ROC curve. The Area Under Curve had a value of 1, indicating a perfect classifier at a cutoff point of 18.5. The sensitivity for this cutoff score was 81.8% and specificity was 100% [22]. For this study, Cronbach’s alpha coefficient demonstrated a good level of reliability at 0.79. The scale items are presented in the Supplementary material.

Statistical analysis

Data collection and entry procedures were conducted using Google Forms, following which the collected data were formatted for the final analysis utilizing SPSS 25 and AMOS 23 software. The analysis consisted of the application of both descriptive and inferential statistics, with all analyses being conducted for the entire sample and subgroups categorized by the student status (first-time test takers vs. repeat test takers). Descriptive statistics were calculated, utilizing frequencies and percentages, mean and SD, skewness, and kurtosis. Normality assumptions were checked using skewness (<|3|) and kurtosis (<|10|) values of the variables [23]. Besides, Kolmogorov-Smirnov test was performed to check the normal distribution of the data. Both the EFA and CFA were performed to validate the scale. In the CFA, we used the criteria suggested by Hu and Bentler, where RMSEA < 0.05, SRMR < 0.05, GFI > 0.95, CFI > 0.95, and NFI > 0.95 indicated excellent model fit [24]. Furthermore, mean comparisons, including independent sample t-tests and one-way ANOVA tests analyzed within the data. Linear regression was also employed using the backward selection method to identify the potential predictors of digital addiction. Besides, a spatial analysis of digital addiction was conducted using ArcGIS 10.8 software. For this purpose, geographic data was sourced from https://www.diva-gis.org/. Initially, data from the total participants were aggregated according to districts. Subsequently, post-hoc analyses were performed and the outcomes were visualized using the maps, stratifying the results by gender and student status. In all statistical tests, a significance level of 0.05 was adopted.

Ethics statement

This study was conducted in full compliance with the principles outlined in the Helsinki Declaration of 1975, with its subsequent revisions in 2008. Besides, all steps involving human participants and patients received formal approval from the review board of CHINTA Research Bangladesh [ref: chinta/2023/5]. Prior to enrolling participants in this study, they were provided with a briefing regarding the study’s objectives and purposes. Importantly, they were made fully aware of their right to decline participation or withdraw from the study at any point, thus emphasizing the paramount importance of informed and voluntary consent. This study participation required informed written consent from the participants.