Introduction
With the rapid advancement of technology, the use of computers, smartphones, and other electronic devices has increased dramatically. While this surge in technology has brought numerous benefits, it has also resulted in negative health consequences associated with excessive use [
1,
2]. The problematic use of smartphones and the internet has given rise to smartphone dependence, which is characterized by excessive reliance on digital devices. This dependence has led to several unintended consequences, including smartphone addiction, impaired social interactions, reduced productivity, and mental health concerns such as anxiety and depression [
3,
4]. Research continues to accumulate, demonstrating a strong association between problematic smartphone use (PSU) and mental health conditions, including depression and anxiety [
5–
9].
The World Health Organization has officially recognized technology addiction as a global issue, noting that excessive internet use can impair an individual’s ability to manage time, energy, and focus [
1,
2,
10]. The pervasive integration of smartphones into daily life has made it increasingly difficult for adolescents to regulate their use, often resulting in overdependence or addiction. Globally, the median prevalence of PSU among children and young people has been reported as 23.3% (range, 14.0%–31.2%) [
11]. In the Republic of Korea, the prevalence of smartphone overdependence (SOD) among adolescents reached 40.1% in 2023, nearly double the prevalence observed in adults (22.7%) [
3]. The risk of SOD has increased sharply, rising from 29.3% in 2018 to 40.1% in 2023, including 5.2% in the high-risk group and 34.9% in the potential risk group [
3,
12]. Anxiety is also increasingly prevalent among adolescents. Keyes and Platt [
13] reported a significant rise in adolescent anxiety and identified digital technology, including smartphone use, as a novel risk factor contributing to these trends.
Numerous studies have explored the association between SOD and anxiety [
4,
6,
7,
9,
14–
17]. Some studies have examined anxiety as a psychological correlate of SOD, while others have focused on anxiety as a mental health outcome associated with excessive smartphone use [
6–
9,
16–
18]. Examining reciprocal associations between SOD and anxiety is essential for understanding how these conditions may interact within adolescent populations. Such evidence may inform screening approaches for identifying at-risk groups and guide the development of future public health strategies. Considering smartphone dependence and anxiety as interconnected conditions may also contribute to improved therapeutic outcomes and enhanced mental well-being. The purpose of this study was to investigate the relationship between SOD and anxiety among adolescents. Specifically, this study examined the association between SOD and anxiety from a reciprocal perspective.
By examining these associations, this study sought to contribute empirical evidence that may support future research and intervention planning focused on adolescent mental health and digital behaviors. Given the cross-sectional design, this study emphasized identifying reciprocal associations rather than inferring causal relationships. Ultimately, this study aimed to contribute to improving mental health outcomes and promoting healthier digital habits among adolescents.
Materials and Methods
Design
We conducted a secondary analysis using a nationally representative cross-sectional dataset derived from the 19th Korea Youth Risk Behavior Survey (KYRBS). The data were collected via a web-based survey administered between August and November 2023, and the dataset was downloaded from the Korea Disease Control and Prevention Agency (KDCA)’s KYRBS website [
19].
Participants
The survey targeted Korean middle and high school students aged 12 to 18 years. Using stratified random sampling, 400 middle schools and 400 high schools were selected nationwide. Within each selected school, 1 classroom per grade was randomly chosen, and all students in those classrooms were invited to participate in the survey. To ensure voluntary participation and anonymity, standardized procedures were implemented. School teachers trained by the KDCA explained the study purpose and participation process. Students then anonymously completed a web-based, self-administered questionnaire in their school computer laboratories during class time. Of the initial sample of 56,935 students across 800 schools, 52,880 students from 799 schools participated, yielding a response rate of 92.9% [
20]. After excluding participants with missing data on key variables, including smartphone usage time, SOD, and generalized anxiety disorder (GAD), a total of 50,975 adolescents were included in the final analysis. This exclusion process ensured that the analytic dataset permitted a more accurate and reliable assessment of the relationship between smartphone use and anxiety.
Measurements
Generalized anxiety disorder
GAD was assessed using the GAD 7-item scale (GAD-7), a widely validated instrument developed by Spitzer et al. [
21]. The GAD-7 evaluates symptoms such as nervousness, anxiety, excessive worry, and irritability through 7 items assessing the frequency of these symptoms over the preceding 2 weeks. Participants responded to each item using a 4-point Likert scale ranging from 0 (not at all) to 3 (nearly every day), yielding a total score ranging from 0 to 21. Scores were categorized as follows: 0–4 (minimal), 5–9 (mild anxiety), 10–14 (moderate anxiety), and 15–21 (severe anxiety). In the present study, the GAD-7 demonstrated high internal consistency, with a Cronbach’s α of 0.91, indicating excellent reliability.
Smartphone overdependence
SOD, as defined by the Ministry of Science, ICT and Future Planning and the Korea Information Society Development Institute (2016), refers to excessive reliance on smartphones that interferes with daily functioning. The SOD scale (SOS) was used to assess this construct and consists of 10 items across 3 subscales: salience (prioritizing smartphone use over other activities), loss of control (difficulty regulating smartphone use), and problematic outcomes (physical discomfort, social conflicts, and impairments in family, school, or work settings). Each item is scored on a 4-point Likert scale, producing total scores ranging from 10 to 40. Based on SOS scores, SOD was classified into 3 categories: normal risk (10–23), potential risk (24–30), and high risk (31–40), with higher scores indicating greater levels of smartphone reliance. In this study, the SOS demonstrated strong internal consistency, with a Cronbach’s α of 0.91. The psychometric properties of the SOS have been recently revalidated among Korean adolescents, confirming its 3-factor structure and robust internal consistency [
22]. Although the validated version proposed a 9-item structure, the present study employed the original 10-item version to ensure comparability with the national monitoring system and to allow classification into the 3 official SOD risk categories.
Covariates
Covariates included demographic and psychosocial factors: sex (male or female), school level (middle school or high school), living area (urban or rural), academic performance (categorized as high, intermediate, or low), and socioeconomic status (upper, moderate, or lower). Smartphone usage time was assessed by asking participants to report their average daily smartphone use on weekdays and weekends, including hours and minutes. These values were converted into minutes and used to calculate a weighted average daily smartphone usage, accounting for 5 weekdays and 2 weekend days. The resulting daily average was categorized into 5 groups: less than 2 hours per day, 2–5 hours per day, 5–10 hours per day, 10–15 hours per day, and more than 15 hours per day.. This approach enabled a more accurate estimation of habitual smartphone use by accounting for differences between weekday and weekend patterns. Additional covariates included perceived stress (classified as “a little” or “a lot”), depressive feelings (yes or no), and loneliness (yes or no).
Statistical Analysis
All analyses were conducted using SAS ver. 9.4 (SAS Institute Inc.), with sampling weights applied to account for the complex survey design and to ensure national representativeness. Descriptive statistics were used to summarize participants’ general characteristics, and Chi-square tests were performed to compare the prevalence of anxiety and SOD across participant characteristics. To examine associations between SOD and anxiety, 2 multivariable logistic regression models were specified: one with anxiety as the dependent variable and the other with SOD as the dependent variable. Covariates were selected a priori based on prior literature and theoretical relevance and included sex, school level, living area, socioeconomic status, academic performance, perceived stress, depressive symptoms, and loneliness. Before model estimation, multicollinearity among independent variables was assessed using variance inflation factors (VIFs) and collinearity diagnostics derived from ordinary least squares regression. All VIF values were close to 1 (range, 1.01–1.56), and intercept-adjusted condition indices were well below commonly accepted thresholds, indicating no evidence of problematic multicollinearity. Odds ratios (ORs) and 95% confidence intervals (CIs) were estimated to quantify associations. Model performance was evaluated using percent concordance, and Somers’ D statistics were examined to assess model discrimination.
IRB Approval
This study utilized publicly available data from the 19th KYRBS, conducted by the KDCA in collaboration with the Ministry of Education. Informed consent was obtained from all participants by the KDCA prior to data collection. Ethical approval for this secondary analysis was exempted by the Institutional Review Board of Jeju National University (approval number: JJNU-IRB-2024-086), as the study involved analysis of anonymized national survey data.
Results
General Characteristics of the Participants
The general characteristics of the participants in the study are summarized in
Table 1. Among the 50,975 participants, 50.9% were male and 49.1% were female. Participants were relatively evenly distributed by school level, with 51.4% attending middle school and 48.6% attending high school. Most participants resided in urban areas (94.2%), whereas 5.8% lived in rural areas. Regarding socioeconomic status, 43.2% of participants were classified as upper class, 45.2% as middle class, and 11.6% as lower class. In terms of academic performance, 37.8% were categorized in the upper test score group, 29.4% in the middle group, and 32.7% in the lower group. Smartphone use varied across participants: 6.9% reported using smartphones for less than 2 hours per day, 49.8% for 2 to 5 hours per day, 37.2% for 5 to 10 hours per day, 4.8% for 10 to 15 hours per day, and 1.3% for more than 15 hours per day.
For perceived stress, 74.2% of participants reported experiencing “a little” stress, whereas 25.8% reported experiencing “a lot” of stress. With respect to depressive symptoms, 74.2% reported no depression, while 25.8% reported experiencing depression. Similarly, 81.8% of participants reported not experiencing loneliness, whereas 18.2% reported experiencing loneliness. Regarding GAD, 63.8% of participants were classified as having minimal anxiety, 23.6% as mild anxiety, 8.2% as moderate anxiety, and 4.4% as severe anxiety. For SOD, 71.6% of participants were classified as normal, 25.0% as being at potential risk, and 3.3% as being at high risk.
Prevalence Rates of Anxiety and SOD by General Characteristics
The prevalence of anxiety differed significantly according to participants’ general characteristics (
Table 2). Overall, female adolescents exhibited a substantially higher prevalence of anxiety (16.4%) compared with male adolescents (8.9%) (
p<0.001). School level was not significantly associated with anxiety prevalence, with middle school students reporting a prevalence of 13.0% and high school students 12.2% (
p=0.051). Similarly, living area was not associated with anxiety prevalence, with urban residents reporting a prevalence of 12.6% and rural residents 11.7% (
p=0.283). Economic status showed a significant association with anxiety, as adolescents in the lower economic group exhibited the highest prevalence (20.9%), compared with those in the upper (11.2%) and middle (11.8%) economic groups (
p<0.001). Academic performance was also associated with anxiety prevalence, with the lower performance group reporting a higher prevalence (15.3%) than the upper (11.7%) and middle (10.7%) performance groups (
p<0.001).
Smartphone use was significantly associated with anxiety prevalence. The lowest prevalence of anxiety was observed among adolescents who used smartphones for less than 2 hours per day (9.4%), whereas the highest prevalence was observed among those who used smartphones for more than 15 hours per day (22.8%) (p<0.001). Perceived stress showed a strong association with anxiety, with individuals reporting “a lot” of stress exhibiting an anxiety prevalence of 27.9%, compared with 3.5% among those reporting lower stress levels (p<0.001). In addition, adolescents reporting depression (31.5%) and loneliness (40.9%) demonstrated markedly higher anxiety prevalence compared with those without depression (6.0%) and loneliness (6.3%) (p<0.001). Anxiety prevalence also differed significantly across SOD categories. Adolescents in the high-risk SOD group exhibited an anxiety prevalence of 40.2%, compared with 19.0% in the potential risk group and 9.1% in the normal group (p<0.001).
SOD also differed significantly according to general characteristics (
Table 2). Female adolescents exhibited a higher prevalence of SOD (4.1%) than male adolescents (2.5%) (
p<0.001). School level was not significantly associated with SOD prevalence, with rates of 3.3% among middle school students and 3.4% among high school students (
p=0.591). Economic status showed a significant association with SOD, with the lower economic group exhibiting a prevalence of 5.6%, compared with 3.0% in both the upper and middle economic groups (
p<0.001). Academic performance was also significantly associated with SOD prevalence, with adolescents in the lower performance group reporting a higher prevalence (4.8%) than those in the upper (2.8%) and middle (2.3%) performance groups (
p<0.001).
Regarding smartphone usage time, adolescents who used smartphones for more than 15 hours per day exhibited the highest prevalence of SOD (15.2%), whereas those who used smartphones for less than 2 hours per day exhibited the lowest prevalence (0.4%) (p<0.001). Perceived stress was also significantly associated with SOD, with adolescents reporting higher stress levels showing a prevalence of 5.6%, compared with 2.0% among those reporting lower stress levels (p<0.001). Depression and loneliness were similarly associated with SOD prevalence, with higher prevalence observed among adolescents experiencing depression (6.1%) and loneliness (7.7%), compared with those without depression (2.3%) and loneliness (2.3%) (p<0.001). Notably, SOD prevalence increased progressively with greater anxiety severity. Adolescents with minimal anxiety exhibited a low prevalence of SOD (1.5%), whereas those with severe anxiety exhibited a substantially higher prevalence (17.0%) (p<0.001), indicating a clear graded association between anxiety severity and SOD.
The Multivariable Logistic Model of Anxiety and SOD
The multivariable logistic regression analysis included multiple independent variables to assess their associations with anxiety and SOD. The variables examined were sex, school level, living area, socioeconomic status, student test scores, smartphone usage, perceived stress, depression, loneliness, and GAD. These factors were analyzed simultaneously to evaluate their relationships with both anxiety and SOD, thereby allowing for a comprehensive assessment of the interrelationships among these variables.
In the analysis of anxiety, several variables emerged as significant predictors. Female adolescents had 1.28 times higher odds of experiencing anxiety compared with male adolescents (OR, 1.28; p<0.001). High school students exhibited a slightly lower likelihood of anxiety than middle school students (OR, 0.93; p=0.041). Adolescents from lower economic backgrounds had 27% higher odds of anxiety compared with those from upper economic backgrounds (OR, 1.27; p<0.001). Perceived stress, depression, and loneliness were the strongest predictors of anxiety. Adolescents reporting high levels of stress had nearly fivefold higher odds of anxiety (OR, 4.90; p<0.001), while those reporting depression had almost threefold higher odds (OR, 2.86; p<0.001), and those reporting loneliness had nearly fourfold higher odds (OR, 3.88; p<0.001). In addition, adolescents classified as being at potential risk for SOD had 1.76 times higher odds of anxiety (OR, 1.76; p<0.001), whereas those classified as being at high risk for SOD had 3.51 times higher odds of anxiety (OR, 3.51; p<0.001).
In the analysis of SOD, female adolescents were not significantly more likely to exhibit SOD compared with male adolescents (p=0.112). Adolescents residing in rural areas had 27% lower odds of SOD compared with those residing in urban areas (OR, 0.73; p=0.021). Economic status demonstrated a protective association, with adolescents from middle economic backgrounds being 16% less likely to exhibit SOD compared with those from upper economic backgrounds (OR, 0.84; p=0.004). Smartphone usage time was the strongest predictor of SOD. Adolescents who used smartphones for more than 15 hours per day had 26.66 times higher odds of SOD compared with those who used smartphones for less than 2 hours per day (OR, 26.66; p<0.001). Loneliness was also a significant predictor of SOD (OR, 1.25; p=0.004). In addition, GAD was significantly associated with SOD, with adolescents experiencing severe anxiety showing 7.68 times higher odds of SOD (OR, 7.68; p<0.001).
Regarding model performance, the anxiety model demonstrated a percent concordance of 86.5% and a Somers’ D value of 0.737, indicating strong predictive ability. The Akaike information criterion (AIC) for the anxiety model was 1,883,104.2, suggesting good overall model fit. In comparison, the SOD model demonstrated a percent concordance of 77.3% and a Somers’ D value of 0.566, reflecting weaker predictive performance relative to the anxiety model. The AIC for the SOD model was 722,196.71, indicating acceptable model fit but lower predictive capacity. Overall, both models were statistically significant; however, the model predicting anxiety demonstrated superior fit and discrimination compared with the model predicting SOD (
Table 3).
Discussion
This study found that anxiety was significantly associated with SOD. These findings are consistent with previous studies reporting significant associations between anxiety and SOD [
8,
17,
18]. In addition, higher levels of SOD were associated with increased odds of anxiety, aligning with prior studies that examined anxiety in relation to SOD [
6,
7,
15,
16,
23]. Notably, recent longitudinal evidence further supports this association. A prospective cohort study among Chinese adolescents reported that new-onset or persistent smartphone dependence was associated with an increased subsequent risk of anxiety symptoms, whereas cessation of dependence was not associated with elevated risk, suggesting that sustained problematic use may precede later emotional difficulties [
24]. In the multivariate logistic regression analyses, the magnitude of the associations between anxiety and SOD was attenuated compared with univariate models (
Table S1). This finding is consistent with the results reported by Jo et al. [
9], who observed a reduction in the effect size of GAD after adjusting for relevant covariates. This pattern suggests that the association between anxiety and SOD may be moderated by other psychosocial factors, such as stress, loneliness, and smartphone usage time, reflecting a multifactorial and context-dependent interaction. After adjustment for key covariates, statistically significant associations between anxiety and SOD persisted, indicating robust reciprocal associations rather than causal effects. This interpretation is consistent with longitudinal findings suggesting reciprocal pathways between problematic mobile phone use and internalizing symptoms. For example, a longitudinal study among Korean adolescents demonstrated bidirectional associations between mobile phone addiction and depressive symptoms over time, indicating that psychological distress may both contribute to and result from problematic phone use [
25]. Taken together, these findings underscore the importance of considering the broader psychosocial context in which these variables are associated, as focusing on a single factor may obscure the complexity of adolescent mental health.
Anxiety has been shown to be strongly associated with psychological factors such as stress, loneliness, and depression, as well as with SOD and sociodemographic characteristics including female sex and lower socioeconomic status. Our findings are consistent with those of Yang et al. [
23], who reported significant associations between PSU and depression and anxiety in their systematic review. Similarly, Lee and Lee [
7] reported significant associations among SOD, anxiety, and suicidal ideation across genders. Specifically, girls with PSU demonstrated substantially higher odds of anxiety compared with their non-PSU counterparts [
7]. These findings further support the interpretation that SOD is closely associated with adverse mental health outcomes. This pattern is also consistent with prior research indicating that females tend to report higher levels of anxiety and depression [
6,
7]. Taken together, the consistency across studies highlights the close association between excessive smartphone use and adolescent mental health, suggesting that interventions addressing smartphone use may represent relevant components of broader mental health strategies for both genders. However, longitudinal evidence also suggests that these associations may not be uniform across subgroups. A recent longitudinal study from the Republic of Korea reported that increased smartphone use predicted poorer mental health outcomes particularly among female adolescents, highlighting potential heterogeneity by gender and contextual factors [
26].
In contrast, SOD was most strongly associated with the amount of time spent using smartphones, with anxiety, loneliness, and stress also showing significant associations. In the present study, smartphone usage time demonstrated the strongest association with SOD. This finding is consistent with the results reported by Cha et al. [
27], who observed that the prevalence of smartphone dependence was 17.9% among adolescents using smartphones for less than 4 hours per day and increased to 30.5% among those using smartphones for more than 4 hours per day, indicating that higher usage time is associated with a greater likelihood of smartphone dependence. Similarly, previous studies have reported significant associations between smartphone usage time and smartphone addiction among adolescents [
28–
30]. Prior studies have also reported female sex as being significantly associated with SOD [
28,
29], whereas sex was not significantly associated with SOD in the present study. Chu et al. [
31] reported significant associations between depressive symptoms and smartphone dependency, and Lee [
32] similarly reported associations between depressive symptoms, loneliness, and SOD regardless of sex. In contrast, depression was not significantly associated with SOD in this study. Elhai et al. [
4] reported that, without adjusting for other relevant variables, depression was consistently associated with PSU with at least medium effect sizes. In the present study, however, depression did not emerge as a significant factor influencing SOD. Conversely, Oh and Heo [
33] reported significant associations between SOD and depression. With respect to loneliness, this study identified a significant association between loneliness and SOD, whereas Lapierre et al. [
30] reported contrasting findings, suggesting that smartphone addiction significantly influenced loneliness in a longitudinal context. Taken together, these findings suggest that the relationships among SOD, depression, and loneliness are complex and potentially reciprocal, warranting further investigation using longitudinal designs [
30,
33]. Furthermore, previous research has reported divergent findings regarding the association between anxiety and SOD. In the present study, anxiety was significantly associated with SOD, whereas Hussain et al. [
34] reported no significant association between anxiety and PSU in analyses that included personality traits. These discrepancies may reflect differences in the covariates included and the conceptual frameworks applied across studies. The present study incorporated psychosocial variables such as stress, loneliness, anxiety, and smartphone usage time, whereas Chu et al. [
31] emphasized factors including self-esteem, aggressiveness, affective parenting attitudes, peer attachment, and resilience, and Hussain et al. [
34] focused primarily on personality traits.
The findings from this study, together with those of previous studies, suggest that the statistical significance of variables associated with SOD may vary depending on the set of factors included in the analysis. These findings indicate that the association between anxiety and SOD is more complex than a simple unidirectional pattern. Although anxiety was significantly associated with SOD, an association in the opposite direction was also observed after accounting for other relevant factors. This highlights the need for future research to explore the diverse and interacting variables associated with SOD more thoroughly. By incorporating a broader range of psychological and interpersonal factors, future studies may provide a more nuanced understanding of how these variables collectively relate to smartphone dependency. From a practical perspective, these findings suggest that interventions addressing SOD may be relevant for adolescent mental health. While it remains important to manage anxiety directly, addressing smartphone use may represent a complementary consideration within broader mental health strategies. Accordingly, integrated approaches that combine psychological support with behavioral guidance on smartphone use warrant further evaluation in future interventional studies. These findings underscore the importance of considering a wider range of factors in both research and practice when examining the interconnected nature of technology use and mental health. Importantly, this study identified a significant disparity in anxiety prevalence by socioeconomic status. Adolescents from lower-income households exhibited higher levels of psychological distress, suggesting that future prevention efforts may benefit from prioritizing socioeconomically disadvantaged populations.
This study highlights the importance of addressing SOD and anxiety as interconnected issues in adolescent populations. The findings suggest that schools may consider comprehensive approaches that integrate emotional regulation support with guidance on healthy smartphone use. Given the observed reciprocal associations, focusing on a single dimension may be insufficient to fully address adolescent well-being. Multitiered systems of support, involving school nurses, counselors, and mental health professionals, may facilitate the identification of students at risk and the delivery of appropriate support. Anxiety has been shown to be significantly associated with suicidal ideation, suicide planning, and suicide attempts among adolescents [
35]; therefore, addressing both anxiety and PSU may have implications that extend beyond emotional well-being.
This study contributes to a deeper understanding of the relationship between anxiety and SOD; however, several limitations should be acknowledged. First, the cross-sectional nature of the data limits the ability to infer causality between SOD and anxiety. Although bidirectional associations were observed, longitudinal designs are required to establish temporal precedence. Second, all variables, including smartphone usage, anxiety symptoms, and SOD levels, were assessed using self-reported questionnaires, which may be subject to recall bias or social desirability bias. Future studies would benefit from incorporating objective measures, such as digital usage tracking or clinician-administered anxiety assessments. In addition, several key psychosocial constructs, including stress, loneliness, and depressive symptoms, were measured using brief or dichotomized items, which may not fully capture their multidimensional nature. Furthermore, this study did not account for several well-established confounders, such as sleep duration and quality, physical activity, family environment, academic stress, and experiences including cyberbullying, all of which may influence both smartphone use and anxiety. Finally, although the sample is representative of Korean adolescents, cultural, educational, and technological differences may limit the generalizability of these findings to adolescents in other countries. Cross-national research is needed to determine whether similar patterns are observed across diverse sociocultural contexts.
