Association between Smartphone Attachment and Mental Health in Adolescents

Abstract

Background:

The surge in adolescent smartphone use has coincided with the rise in the adolescent mental health crisis, raising public health concerns. Moving beyond the traditional focus on screen time, this study examined the association between smartphone attachment and mental health among adolescents.

Methods:

Data were analyzed from 137 community-dwelling adolescents (aged 16.5–18 years). Smartphone attachment was measured by the Mobile Phone Involvement Questionnaire (MPIQ). Patient Reporting Outcome Measures (PROMIS) pediatric short forms were used to measure anxiety, depression, sleep disturbance and general health. T-tests and ANOVAs compared anxiety and depression scores between groups above and below the MPIQ threshold for moderate-to-severe attachment. Multivariable regressions estimated associations between smartphone attachment and mental health outcomes, adjusting for demographics and health factors.

Findings:

Participants were mean age 17.7±0.6 years, 51.1% female, and 79.6% White and 96.4% non-Hispanic. Mean MPIQ score was 28.90±8.85, with females scoring higher than males (30.5 vs. 27.2, p = 0.03). Adolescents above the MPIQ threshold for moderate-severe attachment (≥32) reported significantly higher anxiety (52.9 vs. 46.3, p < 0.001) and depression (51.4 vs. 46.3, p = 0.002). In regression models, MPIQ scores were significantly associated with anxiety (adj. ß = 0.26, p < 0.01, CI = [0.099, 0.41] ) and depression (adj. ß = 0.15, p < 0.05, CI = [0.16, 0.28]), adjusting for demographics and health factors.

Conclusion:

Findings highlight a clinical concern for adolescents with elevated smartphone attachment, particularly among females. Targeted interventions are needed to prevent worsening mental health related to smartphone attachment.

1. Introduction

As of 2022, 95% of United States adolescents reported owning a smartphone, with 46% reporting constantly being online (Vogels, Gelles-Watnick, & Massarat, 2022). In August 2024, the Surgeon General called for warning labels for social media (Abbasi & Hswen, 2024). Concurrently, the prevalence of depression, self-harm, and suicide attempts among adolescents has risen globally by 13% (Shorey, Ng, & Wong, 2022; Twenge, Haidt, Joiner, & Campbell, 2020), exacerbated by the COVID-19 pandemic and projected to become the leading cause of disease burden by 2030 (Hock et al., 2012; U.S. Surgeon General, 2023). The ubiquity of smartphones in adolescents’ lives, accompanied with the ongoing mental health crisis, ignited significant public health concern (U.S. Surgeon General, 2023).

With the introduction of touchscreen devices, features on smartphones have expanded and provide numerous advantages in nurturing social connections, academic advancement, self-expression, and identity explorations – all pertinent to positive youth development outcomes (Granic, Morita, & Scholten, 2020; Nesi, Choukas-Bradley, & Prinstein, 2018; Valkenburg & Peter, 2011). The COVID-19 pandemic accentuated the importance of smartphones in adolescents’ lives by providing a “space” to maintain friendship and seek support regardless of the limitation of time and physical distances (Kline, Metcalf, Patel, Chang, & Nguyen, 2023; Liang, Kutok, Rosen, Burke, & Ranney, 2023; Rimel et al., 2023). This evidence underscored the social and emotional value that smartphones hold in adolescents’ lives – their smartphone is not just an object but an extension of the offline world (Granic et al., 2020; Navarro & Tudge, 2023; Paulus et al., 2023).

The omnipresence of smartphones in adolescents’ lives also posts significant mental health risks (De-Sola Gutiérrez, Rodríguez de Fonseca, & Rubio, 2016; Orben, Meier, Dalgleish, & Blakemore, 2024). Adolescence, a critical stage of active brain development, when the prefrontal cortex proliferates, allows advancement in social processing, impulse control, and emotional regulation – all closely related to mental health and increased vulnerability toward online risks (Blakemore, 2019; Casey, Duhoux, & Cohen, 2010; Orben & Blakemore, 2023; Tottenham, 2015). Many mental health issues such as anxiety, depression, and substance use disorders often initiate during adolescence and persist throughout adulthood (Blakemore, 2019). It has been suggested that closely related neural mechanisms behind social, emotional, and reward processing had increased adolescents’ tendency in sensation seeking, especially at the presence of peer influences (Albert, Chein, & Steinberg, 2013; Steinberg, 2007), which increased their tendency for both online risk taking and the development of problematic use.

Problematic smartphone use, though not yet an established diagnosis, is an emerging concept with heterogeneous definitions and labels in the literature (Bickham, 2021; Billieux, Maurage, Lopez-Fernandez, Kuss, & Griffiths, 2015; Sohn, Rees, Wildridge, Kalk, & Carter, 2019). Rooted in the addiction framework, problematic smartphone use includes symptoms of withdrawal, euphoria, loss of control, and conflict with other people and daily activities (Walsh, White, & Young, 2010) and was suggested to correlate with, depression and anxiety, especially among adolescents aged 15–19 years (Bickham, 2021; Sohn et al., 2019).

Adolescence also marks substantial changes in sleep-wake pattern, with teens generally going to bed later and sleeping less, increasing susceptibility to unhealthy sleep behaviors, such as short duration and delayed onset (Colrain & Baker, 2011; Gradisar, Gardner, & Dohnt, 2011). Unhealthy sleep is associated with a range of poor health outcomes, including mental health issues, cognitive challenges, and behavior problems (J. Liu, Ji, Rovit, Pitt, & Lipman, 2022; Wolfson & O’Malley, 2012). Prolonged smartphone use exacerbates sleep problems, particularly delayed bedtime, poor sleep, and daytime sleepiness (Hale & Guan, 2015; Hamilton & Lee, 2021; Riesch et al., 2019), placing adolescents at higher risks for mental health problems and decline in overall well-being (An, Ji, Zhou, & Liu, 2023; Hale & Guan, 2015; Riesch et al., 2019). Better understanding the relationship between smartphone involvement and adolescent mental health requires a comprehensive approach that accounts for sleep quality and overall well-being.

Despite the breath of research on adolescent smartphone use and metal health, findings remain inconclusive (Sohn et al., 2019; Valkenburg, Meier, & Beyens, 2022). Heterogeneity in the conceptualization and measurement of smartphone use was suggested as contributor to this inconsistency (Sohn et al., 2019; Valkenburg et al., 2022). Given smartphones’ social and emotional significance in adolescents’ daily lives, pathologizing use may overlook its benefit in fostering social connection and self-concepts (Nereim, Bickham, & Rich, 2019; Walsh et al., 2010). Conversely, solely measuring screentime provides a neutral but oversimplified view of adolescent smartphone use that misses its emotional and relational significance in adolescents’ lives. Hence, this study adopts a balanced position that conceptualizes problematic smartphone use as over-attachment instead of pathologizing it as addiction (Walsh et al., 2010). Further, the lack of consideration of mental-health related factors, such as overall well-being and sleep problems, may also contribute to the inconclusive findings. Moreover, few studies have attempted to identify thresholds of smartphone use associated with clinically significant mental health concerns to inform practical guidelines for adolescent psychiatric practitioners.

This study aimed to examine the association between adolescent smartphone attachment, including emotional and behavioral attachment, and self-reported anxiety and depressive symptoms, while accounting for sleep disturbance and general health. The focus on smartphone attachment rather than screentime marks a distinction that shifts the current conversation from quantity to quality and psychological dependence. Incorporating sleep disturbance and general health – factors closely interrelated with both smartphone attachment and mental health symptoms – adds a holistic perspective to the discussion.

Further, we examined the relationship between a threshold of moderate-to-severe smartphone attachment and above-normal-limits of anxiety and depressive symptoms among adolescents, allowing for a more practical interpretation of mental health risks associated with smartphone attachment. Our main hypothesis is that smartphone attachment is positively related to anxiety and depressive symptoms, after controlling for overall well-being and sleep quality. Determining these associations, particularly clinically relevant thresholds, allows nursing researchers and practitioners to identify emerging behavioral health concerns during routine assessment and in designing early prevention programs tailored to the mental health needs of adolescents today.

2. Methods

2.1. Data source

This present study is a cross-sectional, correlational study and includes data from a larger study with data collection from March 17, 2021 – November 4, 2024. The larger study included adolescents (aged 16.5–18 years) recruited through the Children’s Hospital of Philadelphia’s electronic health record system to examine driving after concussion. Uninjured controls were recruited as part of the larger study and is the sample included in this analysis; concussed adolescents are not included.

Inclusion criteria for the uninjured controls included: (1) no history of concussion, no recent concussion (within last year), or no previous traumatic brain injury or concussion with continuing symptoms; (2) aged 16.5–18 years; (3) valid drivers’ license; (4) driven four or more times in the past month; (5) personal smartphone capable of downloading smartphone applications used in the study; (6) able to read and write in English. Adolescents were ineligible to enroll in the parent study if they had (1) a musculoskeletal injury preventing them from standing or walking, or (2) self-reported claustrophobia, epilepsy, motion sickness, migraine history, and/or pregnancy (for participants who enrolled into the driving simulator arm only).

This analysis utilized data from the baseline survey. Demographic data, including sex, age, race, and ethnicity were collected via self-reported survey. In addition, surveys about a diverse range of health and health behavior factors, such as depression, anxiety, sleep, were completed online at home at the baseline timepoint using the Research Electronic Data Capture (REDCap) application (P. Harris, Taylor, Minor, et al., 2019; P. Harris, Taylor, Thielke, et al., 2009).

Of the 151 adolescents in the uninjured control group, 13 were excluded due to loss of follow-up and 1 was excluded due to missing responses to the primary independent variable. The final analytic sample included 137 adolescents.

2.2. Ethical consideration

The parent study was approved by the Institutional Reviewed Board at the Children’s Hospital of Philadelphia (IRB 20–017420) with a reliance agreement with the University of Pennsylvania. Parent consent and adolescent assent (<18 years) or adolescent consent (≥18 years) were obtained for all participants in the sample.

2.3. Smartphone attachment

The primary independent variable of interest, smartphone attachment, was measured by the Mobile Phone Involvement Questionnaire (MPIQ; Walsh et al., 2010). MPIQ is a self-report measure assessing behavioral and cognitive attachment to smartphones grounded in Brown’s (1986) behavioral addiction criteria and adolescents’ qualitative description of smartphone use behavior (Walsh, White, & Young, 2008). MPIQ is a one-dimensional 8-items measure scored on a Likert scale of 1 (strongly disagree)-7 (strongly agree), with sum scores ranging from 8 to 56. Higher scores mean stronger attachment to smartphones. Example items included: “I often think about my cell phone when I am not using it,” and “I feel connected to others when I use my cell phone.” As in a previous study, we used a sum score ≥ 32 as the threshold for moderate-to-severe attachment (Huang et al., 2023). The MPIQ showed good-moderate reliability in previous studies when applied on U.S. adolescent samples (α = 0.89; Argumosa-Villar et al., 2017) and in the current study sample (α = 0.84). Detailed information of the MPIQ in the context of the current study sample is available in supplementary materials.

2.4. Mental health outcomes

Mental health outcomes of anxiety and depressive symptoms were measured using the self-reported Patient Reporting Outcome Measures (PROMIS) Anxiety (7b v.2.0) and Depressive Symptoms (8a v.2.0) Pediatric Short Forms (Irwin et al., 2010; PROMIS, 2025). The PROMIS Anxiety (8-items) and Depressive Symptoms (8-items) short-form used a 7-days recall period and a 5-point response scale: 1 (never) – 5 (always). The PROMIS Depressive Symptoms 8-item measure assesses negative mood (e.g., sadness), anhedonia (e.g., loss of interest), self-perception (e.g., worthlessness), and social cognition (e.g., loneliness; Irwin et al., 2010; PROMIS, 2025). The PROMIS Anxiety 8-item measure focuses on fear, anxious misery (e.g., worry), and hyperarousal (e.g., nervousness; Irwin et al., 2010; PROMIS, 2025).

PROMIS measures were scored using t-scores, which centered around the mean 50 (SD = 10) based on the reference population. Previous literature established good test-retest reliabilities (correlation between 0.7–0.8) and internal consistencies (correlation between 0.7–0.8; Varni et al., 2014). Both measures also showed excellent internal consistency in the current sample (anxiety: α = 0.93; depression: α = 0.92). PROMIS raw scores were converted to t-scores and stratified into clinical categories based on cut-points: within normal limits (t-score ≤ 50), mild (t-score 51–55), and moderate-severe (t-score >55); t-score >50 was considered as above normal limits (t-score > 50; Fish et al., 2023; PROMIS, 2025).

2.5. Covariates

General health and sleep, measured by the PROMIS Global Health (v1.0) and Sleep Disturbance (8a v1.0) Pediatric Short Forms, were selected as covariates given their relationship with anxiety and depression (Paulus et al., 2023; Sohn et al., 2019). The PROMIS Sleep Disturbance Pediatric Short Form consists of 8-items, rated on a 5-point Likert scale: 1 (never) – 5 (always). It assesses sleep onset, continuity, quality, and parasomnia experienced in the past 7-days, and showed high measurement precision among the U.S. sample (marginal reliability > 0.9; Forrest et al., 2018; PROMIS, 2025). The PROMIS Global Health Pediatric Short Form consists of 7 core items that comprehensively assess physical, mental, and social health. Two additional items assess pain and fatigue separately from the 7 core items, and they were not included in the current analysis. All items were rated on a 5-point Likert scale: 1 (never) – 5 (always). Both measures demonstrated good internal consistency (α = 0.88) when applied on U.S. pediatric samples (Forrest et al., 2014; PROMIS, 2025) and the current sample (sleep disturbance: α = 0.88; global health: α = 0.73). For analysis, raw scores were converted to t-scores and categorized based on cut-points: 41 for global health (> 41 = good health; ≤41 = fair-poor health; Forrest et al., 2014) and 55 for sleep disturbance (≤55 = within normal limit (WNL); 56–59 = mild; > 59 = moderate-severe; PROMIS, 2025).

Other covariates included demographic variables, namely age, sex, and race, which were collected via self-report surveys. Race was dichotomized as White and non-White in the analysis as majority of the participants were White. Detailed breakdown of race and ethnicity were included for descriptive analysis.

2.6. Statistical analysis

Data was cleaned in R studio (version 4.3.1; R Core Team, 2023) and analyzed in STATA 18 BE (StataCorp, 2023). First, summary statistics described sample characteristics and all the key variables. Mean and standard deviation of PROMIS t-scores were obtained, as well as proportion of the sample by PROMIS cut-points. Given that depression and anxiety t-scores approximate a normal distribution, we used t-tests to compare mean anxiety and depression t-scores for groups above and below the MPIQ threshold (score 32). ANOVA assessed differences in MPIQ scores across cut-points of PROMIS health measures, followed by post-hoc Tukey’s tests for pairwise comparisons. Sex differences in smartphone attachment and mental health outcomes were analyzed using two-sample t-tests and Mann-Whitney U tests. Chi-squared tests were used to examine sex differences across PROMIS cut-points.

Spearman’s rank correlations examined bivariate associations between MPIQ sum scores and PROMIS t-scores. Subsequently, two multivariable regressions estimated the association between smartphone attachment and mental health outcomes, one each for anxiety and depression, controlling for demographics and the other health factors. PROMIS anxiety and depression t-scores were used as dependent variables in these two separate models. The regression followed a two-step process: First, PROMIS anxiety and depression t-scores were regressed upon MPIQ sum scores, respectively, controlling for demographic covariates. Second, health-related covariates, namely PROMIS sleep disturbance and global health t-scores were added into the model in a step-wised order. R-squared and Bayesian information criterion (BIC) were used to assess model fit. Models with higher R-squared and lower BIC value were deemed a better model fit. Multicollinearity was assessed using the variation inflation factors (VIFs). An VIF greater than 10 was deemed a cause of concern (O’brien, 2007). All statistical tests were two-sided and used p < 0.05 as the level of significance.

3. Results

Of the 137 participants, the mean age was 17.7±0.6 years, 51.09% were female, 79.56% identified as White, and 96.35% identified as non-Hispanic (Table 1).

Table 1.

Sample Characteristics (N = 137)

Characteristics
	Mean	SD
Age	17.74	0.60
	N	%
Sex
Male	67	48.91
Female	70	51.09
Race
White	109	79.56
Asian	12	8.76
Black/African American	6	4.38
Multiple Race	10	7.30
Ethnicity
Not Hispanic/Latino	132	96.35
Hispanic/Latino	5	3.65

Note. SD = standard deviation

3.1. Health measures results

PROMIS scores of sleep disturbance, global health, anxieties, and depression were examined in the total sample and stratified by sex (Table 2). Most adolescent participants perceived themselves as having good global health and no sleep disturbance. The mean depression t-score was 47.9±9.32, with 41.61% above normal limits (18.98% mild, 22.63% moderate-severe). The mean anxiety t-score was 48.5±10.14, with 39.42% above normal limits (13.14% mild, 26.28% moderate-severe). A significant sex difference was found; females scored significantly higher on anxiety (t-scores 51.10 vs 45.74, p-value = 0.001) than males, but not on depression, global health, or sleep disturbance. Subsequent Chi-squared test compared the proportion male and female by PROMIS cut-points and showed a higher proportion of female who reported above-normal-limit anxiety than males (WNL: 47.14 % female vs. 74.63% male, Mild: 17.14% vs. 8.96%, Moderate-severe: 35.71% vs. 16.42%, p = 0.004).

Table 2.

PROMIS outcomes stratify by sex

	Total (N = 137)	Female (N = 70)	Male (N = 67)	p-value
PROMIS depression
T-score mean (SD)	47.95 (9.35)	49.29 (9.08)	46.56 (9.50)	0.098
WNL N (%)	80 (58.39)	37 (52.86)	43 (64.18)	0.405
Mild N (%)	26 (18.98)	15 (21.43)	11 (16.42)
Moderate-severe N (%)	31 (22.63)	18 (25.71)	13 (19.40)
PROMIS anxiety
T-score mean (SD)	48.48 (10.14)	51.10 (9.48)	45.74 (10.15)	0.001
WNL N (%)	83 (60.58)	33 (47.14)	50 (74.63)	0.004
Mild N (%)	18 (13.14)	12 (17.14)	6 (8.96)
Moderate-severe N (%)	36 (26.28)	25 (35.71)	11 (16.42)
PROMIS global health
T-score mean (SD)	47.31 (6.78)	46.26 (6.16)	48.41 (7.25)	0.094
Good N (%)	113 (82.48)	58 (82.86)	55 (82.09)	0.906
Fair-poor N (%)	24 (17.52)	12 (17.14)	12 (17.91)
PROMIS sleep disturbance
T-score mean (SD)	49.78 (8.51)	50.19 (8.39)	49.36 (8.69)	0.385
WNL N (%)	99 (72.26)	48 (68.57)	51 (76.12)	0.612
Mild N (%)	17 (12.41)	10 (14.29)	7 (10.45)
Moderate-severe N (%)	21 (15.33)	12 (17.14)	9 (13.43)

Note. WNL = within normal limit; SD = standard deviation

3.2. Smartphone attachment results

Overall, the mean MPIQ sum score was 28.90±8.85 with females scoring higher than males (30.51 vs. 27.21, p = 0.03; Figure 1). The MPIQ sum score was stratified by PROMIS cut-points for anxiety, depression, global health, and sleep disturbance (Figure 1). A significant between-group difference was found among the PROMIS depression and anxiety groups (WNL, mild, moderate-severe). For depression, the mean MPIQ sum scores and standard deviations were 27.14±8.41, 29.77±9.40, and 32.71±8.44 for WNL, mild, and moderate-severe, respectively. The mean MPIQ sum scores for anxiety were 26.66±7.97, 30.56±8.19, and 33.22±9 .51 for WNL, mild, and moderate-severe, respectively. Post-hoc Tukey’s significant difference test showed significant differences in MPIQ mean score between the WNL group and the moderate-severe group for both depression (contrast = 5.57, p <0.008) and anxiety (contrast = 6.56, p-value <0.000). The mean MPIQ score of the moderate-severe groups also approximated the MPIQ threshold (≥32) for moderate-to-severe attachment. Despite insignificant between-group differences, the mean MPIQ scores for global health and sleep disturbance cut-points approximated those for anxiety and depression. For global health, the mean MPIQ sum scores were 28.41±8.26 (good) and 31.21±11.13 (fair-poor). For sleep disturbance, the scores were 28.11±8.66 (WNL), 30.06±8.23 (mild), and 31.67±9.93 (moderate-severe).

Figure 1.

MPIQ sum scores stratify by sex and PROMIS measures

3.2.1. Applying the smartphone attachment threshold

For MPIQ stratification by moderate-to-severe attachment threshold (≥ 32), 32.9% (n = 45) of the adolescents were above the threshold, with 62% (n = 28) of them being female. Of the total male (n = 67) and female (n = 70) participants, 25% of males and 40% of females were above the threshold. No significant difference was found in the proportion of male and females above and below the MPIQ threshold. Those above the MPIQ threshold scored significantly higher on reported anxiety (t-score 52.88 vs. 46.33, p < 0.000) and depressive (t-score 51.39 vs. 46.27, p <0.002) symptoms (Table 3).

Table 3.

Two-sample t-test using MPIQ median scores as cut point

	Less than 32		Greater than/Equal to 32		mean difference
	n	Mean (SD)	n	Mean (SD)	p-value [CI]
Anxiety	92	46.33 (8.79)	45	52.88 (11.34)	0.000 [−10.03, −3.06]
Depression	92	46.27 (8.92)	45	51.39 (9.38)	0.002 [−8.38, −1.85]

Note. CI = confidence interval; SD = standard deviation.

3.3. Bivariate correlation and stepwise regression results

A Spearman’s rank correlation matrix (Table 4) was used to examine bivariate associations between MPIQ score and PROMIS measures t-scores. Significant correlations were observed across all variables. MPIQ scores were positively correlated with depression, anxiety, and sleep disturbance t-scores, indicating that increased smartphone attachments correlate with greater self-reported depression, anxiety, and sleep disturbance symptoms. A significant negative correlation was observed between MPIQ scores and global health t-scores, where an increase in smartphone attachment is correlated with decrease general health. The correlations between MPIQ scores and mental health symptoms (anxiety and depression) were stronger than those with sleep disturbance and global health (Figure 2). However, sleep disturbance showed an even stronger correlation with depression and anxiety than MPIQ scores.

Table 4.

Spearman’s rank correlation matrix of key variables

Key variables	MPIQ	Anxiety	Depression	Global	Sleep
MPIQ	1
Anxiety	0.33***	1
Depression	0.27***	0.76***	1
Global Health	−0.18**	−0.49***	−0.61***	1
Sleep Disturbance	0.18**	0.42***	0.50***	−0.32***	1

Note.

^*p<0.05

^**p<0.01

^***p<0.001.

MPIQ = Mobile Phone Involvement Questionnaire; anxiety, depression, global health, and sleep disturbances refer to corresponding PROMIS measures t-scores.

Figure 2.

Scatter plots of MPIQ sum score and PROMIS measures t-scores

Stepwise multivariable regressions were conducted to explore the association between MPIQ scores and anxiety and depression t-scores while controlling for demographic and related health factors (Table 5). Results indicated that higher MPIQ scores were significantly associated with increased anxiety (adj. ß = 0.26, p < 0.01, CI = [0.099, 0.41]) and depressive (adj. ß = 0.15, p < 0.05, CI = [0.16, 0.28]) symptoms, adjusting for age, sex, race, sleep disturbance, and global health. Sleep disturbance and global health were also significantly associated with anxiety and depressive symptoms. These models, adjusted for sleep disturbance and global health, were compared against baseline models that adjusted only for demographic factors, and showed a significant increase in adjusted R-square (Anxiety: 0.17 vs. 0.40; Depression: 0.10 vs. 0.52) and decrease in BIC (Anxiety: 1018.18 vs. 981.30; Depression: 1006.13 vs. 928.42), indicating preferred model fit. VIF of all variables in the preferred model for anxiety and depressive symptoms were below 10, indicating no signs of multicollinearity.

Table 5.

Stepwise regression of anxiety and depression on MPIQ

	Anxiety			Depression
	Model1	Model2	Model3	Model1	Model2	Model3
MPIQ sum score	0.38***	0.28***	0.26**	0.29**	0.18*	0.15*
	(0.09)	(0.08)	(0.08)	(0.09)	(0.07)	(0.07)
Sex [Ref: Female]	−3.93*	−2.82	−2.96*	−1.53	−0.15	−0.30
	(1.62)	(1.43)	(1.38)	(1.55)	(1.22)	(1.14)
age	2.33	1.82	1.21	2.82*	2.19*	1.51
	(1.36)	(1.20)	(1.18)	(1.30)	(1.02)	(0.97)
White [Ref: other races]	2.44	3.40	3.34	3.69	4.89**	4.81**
	(2.42)	(2.13)	(2.06)	(2.31)	(1.81)	(1.70)
Global health		−0.67***	−0.56***		−0.84***	−0.72***
		(0.11)	(0.11)		(0.09)	(0.09)
Sleep disturbance			0.28**			0.31***
			(0.09)			(0.07)
Intercept	−3.91	38.22	30.89	−13.01	39.54*	31.32
	(25.07)	(23.05)	(22.37)	(23.99)	(19.59)	(18.45)
R-square	0.19	0.38	0.42	0.13	0.47	0.54
adj. R-square	0.17	0.35	0.40	0.10	0.45	0.52
BIC	1018.18	986.94	981.30	1006.13	942.44	928.42

Note.

^*p<0.05

^**p<0.01

^***p<0.001; Standard errors in parentheses.

MPIQ = Mobile Phone Involvement Questionnaire; Global health = PROMIS Global Health t-scores; Sleep disturbance = PROMIS Sleep Disturbance t-scores; Ref = reference group; BIC = Bayesian information criterion.

PROMIS Anxiety and Depression t-scores were used as dependent variables. Model 1 included MPIQ sum scores and demographic variables. Model 2 added PROMIS Global Health t-scores. Model 3 added PROMIS Sleep Disturbance t-scores.

Model 3 of both anxiety and depression demonstrated the best fit, as indicated by the adjusted R-squared and BIC.

4. Discussion

This study explored the association between smartphone attachment and depressive and anxiety symptoms in a sample of adolescents, considering additional health factors, namely sleep disturbance and overall well-being.

The overall MPIQ, which measures smartphone attachment, had a mean score of 28.90 (SD = 8.85), within the possible score range of 8–56, consistent with previous literature using the same measure in adolescent population (Bruni et al., 2015). A mean score <32 suggests, on average, behavioral traits of smartphone attachment are not severe enough to meet the moderate-to-severe attachment threshold. However, about one-third (n = 45, 32.8%) of the participants exceeded this threshold of 32, indicating a higher prevalence of problematic smartphone attachment than previously reported in the literature. Previous studies reported that approximately 10–20% of adolescents and young adult smartphone users reached the problematic threshold (Bickham, 2021; Nereim et al., 2019), a proportion lower than our findings. Although the discrepancy may be attributed to our small sample size, the MPIQ scoring mechanism, or the effect of COVID-19 pandemic, the health and developmental risks posed by the prevalence of smartphone over-attachment warrants attention.

While adolescents could engage in smartphones for constructive purposes, such as connecting with peers, adolescents’ sensitivity to social rejections can make smartphone use a double-edged sword, blurring the boundaries between normative and problematic use (Nereim et al., 2019; Orben et al., 2024). Online interactions, while fostering connections, may feel “in-authentic” and compromise offline relationships (Biernesser, Montano, Miller, & Radovic, 2020; Shankleman, Hammond, & Jones, 2021; Weinstein, 2018). Adolescents are also more prone to engage in risky behavior, particularly in the loosely monitored online space (Vannucci, Simpson, Gagnon, & Ohannessian, 2020). Additionally, smartphones over-attachment can induce family conflicts and displace developmentally-essential offline activities like sleep and physical activity (Jensen, George, Russell, Lippold, & Odgers, 2021; Scott & Woods, 2019). A longitudinal study found that cyberbullying, inadequate sleep, and physical inactivity mediated the association between social media use and mental health problems (Viner et al., 2019). Other studies suggest that family conflict and weakened parent-adolescent relationships partially mediate the association between prolonged screen media use and mental health challenges (X. Liu, Liu, Flores, & McDonald, 2024). Collectively, these findings suggest that prolonged smartphone engagement impacts mental health by exacerbating developmental vulnerabilities and displacing offline health-promoting activities (Viner et al., 2019), warranting the need for clinical vigilance in recognizing early signs of smartphone over-attachment among adolescents.

Significant sex differences in smartphone attachment and anxiety were observed, with adolescent females scoring significantly greater than males on both, underscoring challenges that adolescent females face online. These findings align with evidence from nationally represented U.S. and UK cohorts: compared to adolescent males, adolescent females tended to engage in more smartphone-based media activities, include a higher proportion of addictive social media users, and report greater magnitude of depressive symptoms (Kelly, Zilanawala, Booker, & Sacker, 2018; Twenge & Martin, 2020; Xiao, Meng, Brown, Keyes, & Mann, 2025). Social media use, despite its potential in bolstering social connections, increases the opportunities for online harassment and peer comparisons, especially among adolescent females (Twenge & Martin, 2020; Weinstein, 2018). Adolescent females spent more time crafting their social media posts and tended base their self-value on the amount of “likes” they received on their posts, which has been associated with body dissatisfaction, loneliness, fear of missing out, and reduced self-esteem (Nesi & Prinstein, 2015; Yau & Reich, 2019). This observed sex differences corresponds with adolescents’ developmental needs for social connections, identity development, and peer acceptance (Burnette, Kwitowski, & Mazzeo, 2017; Nesi & Prinstein, 2015), particularly as adolescent females’ sensitivity to social feedback makes them more vulnerable to online risks (Twenge & Martin, 2020).

Smartphone attachment, reflected by MPIQ sum score, was significantly associated with all PROMIS health measures, including anxiety, depression, sleep disturbances, and global health. The association is particularly strong with anxiety and depression, indicating that an increase in smartphone attachment correlates with higher reports of depressive and anxiety symptoms, even when accounting for the factor of sleep and overall well-being in the multivariable models. These findings align with literature that applied MPIQ measure among youth (Harwood, Dooley, Scott, & Joiner, 2014; Xiao et al., 2025). A recent longitudinal study also found higher risks of suicidal behavior and ideation among adolescents with high addictive mobile phone use trajectory (Xiao et al., 2025). Notably, the same study found that baseline screen time was not associated with suicidal outcomes, underscoring the importance of examining smartphone attachment over screen time (Xiao et al., 2025).

However, the regression coefficients for MPIQ in our results were relatively moderate (adj. β_depression = 0.15; adj. β_anxiety = 0.26). This moderate correlation is consistent with literature showing moderate-to-low correlations between general screen media use and mental health outcomes (Orben & Przybylski, 2019; Valkenburg et al., 2022). A recent meta-analysis similarly reported moderate correlation between smartphone addiction and anxiety (r = 0.39) and depression (r = 0.36) among college students (Li, Li, Liu, & Wu, 2020). Notably, most studies reported in the meta-analysis used convenient sample of college students, highlighting the value of our current analysis that focused exclusively on an adolescent sample.

Although the moderate correlations may reflect limitations of cross-sectional design and uncaptured individual-level variation in smartphone use motivation, content, and quality (Beyens, Pouwels, van Driel, Keijsers, & Valkenburg, 2020), MPIQ sum score and demographic covariates together explained <17% of the models’ variances in the current study. The low R-square suggests that smartphone use alone is not the primary contributor to anxiety and depression in adolescents. Unlike previous studies, our analysis accounted for sleep disturbance and overall health. Sleep disturbance and global health were significantly associated with anxiety and depressive symptoms in our models and substantially increased our models’ explained variances, indicating importance of taking a comprehensive approach in the discussions on smartphone use and mental health. In addition, individual factors, such as motivation for smartphone use, content consumed, and comorbid health factors like delayed bedtime, risk-taking behavior, and sedentary behavior, may likewise contribute to the link between smartphone overuse and mental health symptoms among adolescents (Beyens et al., 2020; J. Liu, Riesch, et al., 2022; Valkenburg & Peter, 2013; Vannucci et al., 2020). Further research is needed to clarify these relationships by incorporating these intrapersonal and health factors into the study design.

Previous psychometric efforts for measuring internet use-related attachment have either lacked established threshold to dissect severity or based thresholds solely on problematic use symptoms (B. Harris, Regan, Schueler, & Fields, 2020; Laconi, Rodgers, & Chabrol, 2014). To our knowledge, this analysis is the first that aims to identify smartphone use attachment threshold that corresponds with self-reported above-normal-limit anxiety and depressive symptoms. Using the midpoint MPIQ score of 32 as the threshold (Huang et al., 2023), we found that adolescents scoring above this threshold reported above-normal-limit anxiety and depressive symptoms. Conversely, adolescents with self-reported moderate-severe depressive and anxiety symptoms scored above the moderate-severe attachment threshold on the MPIQ.

Although our small sample size limited further exploration of MPIQ scoring across anxiety and depression cut-points, current findings indicate a bidirectional relationship, with smartphone attachment potentially exacerbating mental health symptoms or vice versa. Understanding smartphone attachment thresholds that correspond to above-normal-limit mental health challenges could aid in targeted screening and timely intervention. This analysis uses MPIQ as an example to encourage further research with larger samples size to clarify the directionality between smartphone use and mental health and to pinpoint clinically relevant thresholds.

4.1. Limitations

The study is subjected to several limitations. First, it utilizes data from a parent study, subjected to inclusion and exclusion criteria of the parent study, as well as limited to available covariates. Specifically, our sample size is relatively small and were drawn from a specific geographic area and met particular eligibility criteria, which limits the generalizability of the findings to the broader U.S. adolescent population. Additionally, given that the majority of participants identified as White and non-Hispanic, the findings may not be generalizable to more racially diverse populations, which underscores the need for studies including such groups. The cross-sectional design also limited causal inference to understand the directionality between smartphone attachment and mental health outcomes. A more nuanced understanding of the relationship between smartphone attachment, depression, and anxiety, while considering the role of sleep disturbance and global well-being, as well as the potential demographic interactions (e.g., sex and race), necessitates a larger sample size and longitudinal design. Lastly, although validated and reliable measures were employed, adolescent self-reporting may introduce bias, potentially leading to under- or over-estimation of key variables. To address this challenge, future research may consider adopting subjective and objective measures to assess smartphone involvement and a multi-informant approach in assessing mental health and related health concerns. Despite the limitations posted by our data, this analysis provides preliminary evidence linking phone attachment to mental health symptom outcome – advancing the conversation beyond measuring screen time alone and providing a step toward more nuanced assessment tools in clinical settings.

4.2. Implications

Findings from this study hold important implications for nurses and other health practioners working with adolescents in today’s rapidly evolving digital landscape. As smartphones hold an increasingly central position in adolescents’ daily lives, psychiatric nurses and other health professionals need to be cognized to screen for smartphone use behavior, motivation, and content to facilitate early conversation about healthy use and identification of problematic patterns. Incorporating screening tools for smartphone attachment in clinical settings could help identify and monitor adolescents who are at risk for smartphone over-attachment. Screening efforts should also consider emotional, physical, and behavioral health, such as mental health symptoms, sleep disturbance, overall well-being, and physical activity, to aid in the early identification of smartphone over-attachment. Further, practioners should account for the potential bidirectional relationships between smartphone use and mental health challenges, screening adolescents with mental health challenges for problematic smartphone use and vice versa. Integrating comprehensive smartphone use screening can help identify at-risk adolescents early, potentially preventing the escalation of mental health issues associated with smartphone attachment.

Additionally, as smartphones continue to play a central role in adolescents’ lives, strategies for healthy use and intervention are paramount. Digital literacy education should be integrated into the broader health literacy promotion efforts within both K-12 education and clinical settings to help distinguish healthy and over-attached phone use patterns. Rather than being viewed solely as a health risks, smartphones can also be leveraged as tools for promoting digital and health literacies (Mancone, Corrado, Tosti, Spica, & Diotaiuti, 2024). Adolescent psychiatric nurses and school-based health practioners are uniquely positioned to provide timely education and intervention regarding smartphone ownership, use, and literacy. By fostering open communication and offering targeted education, adolescent psychiatry nurses and school-based health practioners can support families in early conversations around smartphones or social media account ownership, setting healthy smartphone boundaries, and engaging in meaningful discussions about the potential risk and benefits.

5. Conclusions

This cross-sectional analysis of an adolescent’s cohort examines the relationship between smartphone attachment and symptoms of anxiety and depression, aiming to identify clinically relevant thresholds of smartphone attachment correspond to mental health concerns. Smartphone attachment was significantly associated with mental health symptoms, even after accounting for sleep disturbance and general health. Adolescents with moderate-severe smartphone attachment reported above-normal-limits anxiety and depressive symptoms, with adolescent females reporting higher smartphone attachment and anxiety than males, suggesting greater vulnerability. These findings have important clinical implications for adolescent psychiatric nurses and other health practitioners, supporting targeted interventions and early identification of at-risk adolescents to prevent worsening mental health related to smartphone attachment. A comprehensive approach that integrates intrapersonal and health-related factors is essential for understanding the nuances between smartphone attachment, mental health, and health-related factors, and for refining the threshold of smartphone involvement for clinical significance.