Introduction

With the rapid development of Internet technology, electronic products such as mobile phones have become one of the main tools for individuals to access and supply information [1]; conduct interpersonal communication; obtain entertainment, diversion, and relaxation; receive monetary compensation (such as finding bargains on product and services to save money, getting profitable financial information, or working and doing tasks to make money) [2]; and pursue other activities (such as education and health management), by virtue of their devices’ convenience, accessibility, and powerful functions [3, 4]. However, in the process of using these electronic products and the functions they enable, human beings often experience a variety of problematic behaviors, including overuse and dependence [5], which have gradually become the focus of academic attention. Among groups that typically have access to their own phones, adolescents and young adults, especially college students, are more likely to experience problematic mobile phone use (PMPU) because they have more free time, lower levels of self-control, and increased identity and lifestyle needs (such as online learning, social interaction, games, and shop**) [6,7,8]. A meta-analysis pointed out that the prevalence of PMPU among Chinese college students was as high as 23% [9]. In a previous study, PMPU was defined as uncontrolled or excessive use of mobile phones by individuals that causes problems in daily life [10]. In some studies, it was also referred to as mobile phone dependence [11], mobile phone addiction [12], and smartphone addiction [13]. Similar to the symptoms of substance use disorder, uncontrolled and excessive use was the important symptoms and characteristics of PMPU [14]. In this concept, uncontrolled use was considered a core feature of PMPU, meaning, although aware of the adverse effects, individuals still used and had difficulty controlling the use of mobile phones [10]. Further, excessive use means that an individual’s mobile phone use exceeds a certain time and range. Obviously, this requires a demarcation point to determine whether an individual has excessive mobile phone use. Based on the existing research, it is not feasible to determine the cut-off point by using quantitative methods such as time and frequency, because the motivations and natures or modes of mobile phone use of different individuals are very heterogeneous [15]. For example, when mobile phones were used to contact families and friends, provide social support, or when learning or working with an aim to increase productivity or self-improvement (such as participating in online work conferences, browsing online learning resources, information retrieval, and schedules) [16], the use time of mobile phone can be long and may not have negative consequences; such situations should not be categorized as excessive mobile phone use [17]. Therefore, on the basis of considering the motivation, nature or mode of mobile phone use, current studies tend to use evaluation from the perspective of others (for example, my classmates say that I use mobile phones for too long and too often), rather than quantitative methods to judge excessive use. Specifically, when evaluating excessive use, it need to declare to participants that the mobile phone usage in the evaluation refers to mobile phone use patterns or content are uncontrolled online games, social media, or entertainment (e.g., watching movies and listening music)—and the reason or motivation for use is evasion of reality, failure to regulate stress and negative emotions, and boredom, fear of missing out [14] or specific personality traits such as shyness [10]. In addition, PMPU also showed two characteristics: tolerance (the frequency and duration of mobile phone use by individuals to achieve satisfaction have increased significantly) and withdrawal (individuals experience psychological withdrawal symptoms such as panic, restlessness, and irritability when separated from their mobile phone) [18]. Therefore, in this study, we define PMPU as an individual’s uncontrolled or excessive use of mobile phones and adverse effects when performing activities with the motivation and purpose of relieving negative emotions, relaxing oneself, and satisfying online social and entertainment needs, rather than activities with the motivation and purpose of self-improvement, increase productivity or search for social support, such as work, study, and communication with families and friends.

The concept of PMPU is somewhat controversial, because the main function of mobile phones was the operation of Internet-based applications [19], which indicates that PMPU has many similarities with Internet-based addictions, such as gaming disorders, and may have mutual influences. For example, individuals with Internet addiction are more likely to experience PMPU, and vice versa. Previous research has shown that there was a positive correlation between Internet addiction and PMPU [20]. However, studies have also shown differences between the two in risk factors such as gender and personality characteristics [21]. For example, men experience more Internet addiction, while women demonstrate more PMPU [22]. However, it should be noted that mobile phones not only provide functions such as the Internet and games, but also have various other services and functions such as communications, cameras, multimedia playback, painting, and e-book reading. These services may not be related to the Internet. In addition, individual Internet use relies not only on mobile phones but also on desktop laptops, computers, or digital tablets. Therefore, some symptoms of PMPU may be different from those of Internet addiction. In addition, a study found that, compared with using their phones for playing games, individuals with PMPU were more likely to use social networks [23], which indicates that there may also be differences between PMPU and gaming disorders. Therefore, current research tends to treat PMPU as an independent concept, and has developed some specific assessment tools [15, 24,25,26], which are widely accepted and recognized by scholars. However, it should be noted that none of the existing scales can fully consider the characteristics of uncontrolled or excessive, tolerance and withdrawal to evaluate PMPU [15]. Thus, this study used the Mobile Phone Addiction Tendency Scale (MPATS), which is widely used in mainland China, to focus on the evaluation of uncontrolled use, excessive use evaluated by classmates or friends, tolerance, withdrawal symptoms, and negative consequences [39]. This indirectly indicated that poor sleep quality may cause individuals to develop PMPU. Studies have shown that in individuals with PMPU, constant exposure to blue light can inhibit the secretion of melatonin, and cause sleep and circadian rhythm disorders [40, 41], which might be an important factor in the generation of psychopathological symptoms such as depression [42]. Previous studies have explored the mediating role of sleep quality between PMPU and depressive symptoms using cross-sectional study designs [Measurements

MPATS

PMPU was measured using the MPATS, developed by ** medication, and daytime dysfunction. A Likert 4-level scoring method (0–3 points) was used for each dimension, for a total score of 0–21 points; the higher the score, the lower the sleep quality of the subjects. We used the Chinese version of the PSQI in this study [67]. The Cronbach’s alpha of the scale was 0.70 and 0.73 at T1 and T2, respectively.

The patient Health Questionnaire-9 (PHQ-9)

The PHQ-9 was used to evaluate the frequency of depressive symptoms over the past two weeks [68]. This scale had been translated into Chinese, and showed good psychometric properties in the general population [69]. It is composed of nine items, with 0, 1, 2, and 3 points corresponding to “no,” “several days,” “more than half of the days,” and “almost every day,” respectively, for a total score of 0–27 points; the higher the score, the more serious the depressive symptoms. The Cronbach’s alpha of the scale was 0.86 and 0.89 at T1 and T2, respectively.

Data analysis

The descriptive statistics and Pearson’s correlation analysis were conducted using SPSS version 25.0 (IBM Corporation, Armonk NY, USA) for Windows. Data are presented as n (%) for categorical variables and mean ± SD for numerical variables. We used AMOS 23.0 software to perform cross-lagged panel analysis. First, we evaluated the longitudinal measurement invariance of the four scales used in this study, including configural invariance, metric invariance, and scalar invariance [70]. Second, adjusting gender and age, we constructed a cross-lagged model to test the longitudinal bidirectional relationships among PMPU, bedtime procrastination, sleep quality and depressive symptoms in college students. Model fit was evaluated using comparative fit index (CFI), Tucker–Lewis index (TLI), root–mean–square error of approximation (RMSEA) and standard root–mean–square (SRMR). According to previous studies [71], CFI and TLI greater than 0.90 and RMSEA and SRMR less than 0.08 indicate that the model fit is acceptable. Considering that chi-squared is sensitive to sample size, we did not use chi-squared as an indicator of model fit [72]. Meanwhile, we calculated 95% confidence intervals (CI) using a bias-corrected bootstrap sample that was repeated 5000 times. The 95% CI did not include zero, indicating that the effect was statistically significant (p-value < 0.05). In addition, we used the change values of CFI (ΔCFI) and RMSEA (ΔRMSEA) to evaluate the measurement invariance. When ΔCFI≤0.01 and ΔRMSEA≤0.015, the measurement invariance model was acceptable [73].

Results

Measurement invariance test

In order to test the longitudinal measurement invariance of the scales, we first established configural invariance models. The results showed that the configural invariance models of the four scales all fitted well. Subsequently, we set the factor loadings to be equal over time and established metric invariance models. All model fits were good. The fit results of the metric invariance models showed that ΔCFI and ΔRMSEA were both less than 0.01, indicating invariance of factor loadings on each scale over time. On the basis of the metric invariance model, we further restricted the equality of thresholds, to test scalar invariance; ΔCFI and ΔRMSEA were still within the acceptable range. These results indicate that the four scales have measurement invariance at two time points. More results about the model-fitting index are shown in Multimedia Appendix 1.

Descriptive and correlational analyses

Table 1 shows the means, standard deviations, and correlations of the variables. Correlation analysis showed statistically significant correlations among the four variables of PMPU, bedtime procrastination, sleep quality, and depressive symptoms at two time points.

Table 1 Descriptive and correlations analyses for variables
Full size table

Cross-lagged model

The cross-lagged model showed good fit to the data (χ2(df) = 6.682(3), p > 0.05; CFI = 0.999, TLI = 0.986, RMSEA = 0.032, SRMR = 0.010). Figure 1 and Table 2 display the results for the cross-lagged model. The results suggest that PMPU at T1 positively predicted bedtime procrastination, depressive symptoms at T2 and vice versa, while sleep quality was only significantly predicted one-way. Bedtime procrastination at T1 positively predicted sleep quality at T2, and vice versa and only significantly positively predicted depressive symptoms one-way. Moreover, sleep quality at T1 positively predicted depressive symptoms at T2 and vice versa.

Fig. 1
figure 1

Cross-lagged relationships among PMPU, bedtime procrastination, sleep quality, and depressive symptoms. Notes: *p < 0.05, **p < 0.01. Only statistically significant paths are displayed

Full size image
Table 2 Cross-Lagged Model
Full size table

Discussion

In this study, we adopted a two-wave longitudinal design and constructed a cross-lagged model to analyze the bidirectional associations among PMPU, bedtime procrastination, sleep quality, and depressive symptoms. These findings help to understand the longitudinal associations and direction of effects among these factors. The results suggest some longitudinal associations in both directions: (1) PMPU with bedtime procrastination and depressive symptoms; (2) sleep quality with bedtime procrastination and depressive symptoms. As far as we know, this is the first study in which bedtime procrastination was bidirectionally associated with problematic phone use and sleep quality. There were also some one-way associations: (1) PMPU predicted sleep quality, and (2) bedtime procrastination predicted depressive symptoms.

This study showed that depressive symptoms were bidirectionally associated with PMPU and sleep quality, consistent with previous studies [37, 46, 74]. However, a recent 6-month longitudinal study in China found that depressive symptoms at baseline predicted follow-up PMPU, as opposed to the converse [13]. A possible reason was that the length of follow-up was different, which may lead to different directions of effect between PMPU and depressive symptoms. Two previous longitudinal studies, of more than 1 year each, in South Korea found a bidirectional relationship between PMPU and depressive symptoms [37, 75]. However, a 3-month short-term longitudinal study in the United States found only a one-way association, in which PMPU at baseline predicted follow-up depressive symptoms [76]. This may mean that it will take longer to observe the bidirectional relationship between the two. However, there have also been longer than 1 year longitudinal studies that found only a one-way relationship [

Conclusion

This study further expands our understanding of the longitudinal and bidirectional relationship among PMPU, bedtime procrastination, sleep quality, and depressive symptoms. In addition, it will assist school mental health educators in designing targeted interventions to reduce PMPU, sleep problems, and depressive symptoms among college students.