Sleep Quality Types and Their Influences on Psychological and Physical Health in Chinese Adolescents: A Person-Centered Approach

Abstract

Poor sleep in adolescents is a global health problem. This study examined Chinese adolescents’ sleep quality types by using a person-centered approach and examined the associations of adolescent sleep quality types with psychological characteristics and health risk behaviors. Using data from 1,310 Chinese adolescents (53.7% male; mean age = 13.48 years) and latent class analyses, four distinct sleep quality types were identified: good sleepers (26%), fair sleepers (55%), short sleepers (8%), and poor sleepers (11%). These types of sleepers exhibited differences in their psychological characteristics and health risk behaviors. Good sleepers displayed the best psychological health (highest self-esteem, life satisfaction and positive well-being, and fewest depressive symptoms), followed by short sleepers, fair sleepers, and poor sleepers. Good sleepers were least likely to have suicidal thoughts. Short sleepers were most likely to be physically inactive. Poor sleepers were most likely to drink alcohol and have suicidal thoughts. The implications are discussed.

Keywords

sleep psychological health health risk behaviors gender differences Chinese adolescents

Adolescents are one of the most vulnerable age groups to poor sleep, due to the biological changes in their circadian rhythms, the developmental and social influences they experience (Chung, Chan, Lam, Lai, & Yeung, 2017). The prevalence of insufficient sleep, sleep disturbances, and insomnia among adolescents have been documented all over the world (Chung & Cheung, 2008; Gradisar, Gardner, & Dohnt, 2011; Kaneita et al., 2006; Short, Gradisar, Lack, Wright, & Dohnt, 2013) and is considered a global health concern (Gradisar et al., 2011). Furthermore, cross-cultural research demonstrated cultural difference in adolescent sleep. For example, Asian (e.g., Chinese and Japanese) adolescent bedtimes were later than those of their peers in North America and Europe, resulting in significantly shorter sleep durations and higher rates of daytime sleepiness than American and European adolescents (Gradisar et al., 2011; Olds, Blunden, Petkov, & Forchino, 2010).

The literature shows that sleep is a crucial factor affecting adolescent psychological health. For example, in a longitudinal study with the Finnish Twin Cohort (range of ages: 18-95 years), Paunio et al. (2008) demonstrated that sleep quality was a predictor of later life satisfaction. An experimental study by Dagys and colleagues (2012) also demonstrated that sleep influenced adolescents’ moods. Specifically, adolescents reported more positive feelings when they slept longer. On the other hand, among Australian adolescents, depression was related to worse sleep, including shorter sleep duration, longer sleep onset latency, and greater daytime dysfunction (Short et al., 2013). This association is likely subsequent, as demonstrated by Roane and Taylor’s (2008) study, which found that insomnia during adolescence predicted depression in young adulthood. Lovato and Gradisar’s (2014) meta-analysis also suggested that sleep disturbance was the predictor of current and future depression in adolescence.

There is also research examining the mechanisms underlying the association of adolescent sleep and psychological health. For example, both clinical and experimental findings supported that poor sleep at night negatively influenced the next day’s affective brain function (e.g., up or down regulation of affective states) and subsequent changes in affect (Walker & Harvey, 2010). An experimental study with early adolescents demonstrated that participants in the sleep-deprived condition reported lower levels of positive affect, compared with participants in rested condition (Talbot, McGlinchey, Kaplan, Dahl, & Harvey, 2010). A clinical study with early adolescents also found that good sleep quality was positively associated with the next day’s positive affect and negatively associated with the next day’s negative affect (Van Zundert, Van Roekel, Engels, & Scholte, 2015). In addition, both higher levels of positive affect and lower levels of negative affect during the day were associated with better sleep quality at night (Van Zundert at al., 2015). Therefore, the adverse affective consequences of poor sleep could propel adolescents into a vicious cycle where poor sleep and affective dysfunction continued to aggravate each other (Van Zundert at al., 2015). Such cycle could lead to worse psychological functioning (Van Zundert at al., 2015) and eventually contribute to the development of psychiatric disorders (Walker & Harvey, 2010).

Sleep also exhibits various extents of influence on physical health, via its effect on participation in different health risk behaviors. For instance, some studies with national samples have shown that self-reported insomnia symptoms were associated with greater alcohol use among Japanese adolescents (Kaneita et al., 2006) and American adolescents (McKnight-Eily et al., 2011; Roane & Taylor, 2008). Longitudinal studies provided further support for sleep as an antecedent of substance use. For example, childhood sleep problems predicted the onset of alcohol and cigarette use in adolescence (M. M. Wong, Brower, & Zucker, 2009). Adolescent sleep problems also predicted subsequent alcohol and cigarette use (Pieters, Burk, Van Der, Dahl, Wiers, & Engels, 2015). Furthermore, adolescent insomnia symptoms have been linked with suicidal ideation (Owens & Adolescent Sleep Working Group, 2014; Roane & Taylor, 2008). Psychological autopsies also showed a significant association between completed suicide in adolescents and higher rates of sleep disturbance and insomnia (Goldstein, Bridge, & Brent, 2008). In addition, previous studies have demonstrated a negative relationship between sleep quality and being physically inactive among adolescents. Insufficient sleep (<8 hours) was related to lower levels of physical activity among adolescents (McKnight-Eily et al., 2011). Ortega and colleagues (2010) also found that insufficient sleep (<8 hours) was associated with more sedentary habits (e.g., spending more time on watching television) among European adolescents.

These findings indicate that poor sleep may contribute to early onset or higher rates of health risk behaviors and therefore have an aversive effect on adolescent physical health. The effect of sleep on the executive functions was proposed as a potential mechanism underlying the associations of sleep deprivation and sleep problems with health risk behaviors (Pieters et al., 2015). For example, previous studies demonstrated that inhibitory control (a subdomain of executive functions) mediated the negative associations of sleep duration with cigarette and alcohol use among early adolescents (Warren, Riggs, & Pentz, 2017). Another study also found that both overall executive functions and inhibitory control mediated the negative association between sleep duration and sedentary behaviors among early adolescents (Warren, Riggs, & Pentz, 2016).

Although many studies have been conducted to investigate sleep and its impact during adolescence (for reviews, see Gradisar et al., 2011; Owens & Adolescent Sleep Working Group, 2014), several recommendations have been made to advance the progress in adolescent sleep research (for reviews, see Gradisar et al., 2011). The first recommendation is the inclusion of multiple components of sleep, such as sleep duration, difficulties in initiating sleep, and daytime sleepiness (Gradisar et al., 2011). Many previous studies assessed some of the components only. For example, Paunio and colleagues (2008) used two components of sleep quality (i.e., subjective sleep quality and sleep duration) to study the relationship between sleep and life satisfaction. However, according to Buysse, Reynolds, Monk, Berman, and Kupfer (1989), sleep quality has seven components, including subjective sleep quality, sleep onset latency, sleep duration, habitual sleep efficiency, sleep disturbances, use of sleeping medication, and daytime dysfunction. Using only few components of sleep quality may limit the ability to identify individual differences in sleep quality, and thus hinder the investigation of the relationships between sleep and its correlates (Magee, Reddy, Robinson, & McGregor, 2016).

The second recommendation is the usage of person-centered approaches to capture individual similarities and differences in adolescent sleep quality (Shochat et al., 2017). Person-centered approaches use the relationships among the components of sleep to classify individuals into groups with similar patterns of sleep quality (Shochat et al., 2017). While many studies have examined the associations of distinct components of sleep (e.g., duration, disturbances, onset latency) with adolescent psychological and physical health (e.g., Lovato & Gradisar, 2014; Pieters et al., 2015), these studies did not take the relationships among the components of sleep into account. However, previous studies demonstrated that the components of sleep were related (e.g., Lin, Xie, Yan, Chen, & Yan, 2017). Therefore, using person-centered approaches to account for the relationships among the components of sleep may provide additional insight into the association between sleep and its correlates (e.g., depressed mood; Shochat et al., 2017).

Latent class analysis (LCA), which is a person-centered approach, could be used to identify patterns of sleep quality (sleep quality types). This method has been adopted in recent studies (Magee et al., 2016; Ng, Luo, & Heng, 2014) because LCA can identify the homogeneous classes that best describe the associations between multiple components of sleep quality. For example, Magee et al. (2016) used LCA and several sleep quality components to study the sleep quality types of Australian employees. They identified five distinct sleep quality types and found significant differences in physical activity levels across the sleep quality types. These sleep quality types (i.e., good sleepers, minor sleep disturbances, frequent sleep disturbances, long sleepers, and poor sleepers) displayed different combinations of the sleep components (i.e., sleep duration, sleep quality, and frequency of various sleep problems). Compared with good sleepers, frequent sleep disturbances and long sleepers showed lower levels of physical activity. Compared with long sleepers, minor sleep disturbances reported greater levels of physical activity. Hence, adopting LCA could provide a better and comprehensive understanding of sleep quality and its associations with psychological and physical health. Given that the biological, social, and developmental changes affecting adolescent sleep quality (e.g., circadian phase delay, early school start times, and increased autonomy) are unique in adolescence (for review, see Chung et al., 2017), adolescents likely exhibit different sleep quality types from adults. Therefore, it would be informative to use LCA with multiple components of sleep quality to examine adolescent sleep quality and its associations with psychological and physical health.

The third recommendation is that more studies of adolescent sleep patterns and problems should be conducted in different regions of the world (for review, see Gradisar et al., 2011). Most studies of adolescent sleep were conducted in Western cultures (e.g., Dagys et al., 2012; Goldstein et al., 2008; M. M. Wong et al., 2009). Although some cross-cultural studies have shown that Chinese adolescents exhibited significantly shorter sleep durations and higher rates of daytime sleepiness than American and European adolescents (Gradisar et al., 2011; Olds et al., 2010), no studies, to our knowledge, have examined Chinese adolescents’ sleep quality types. However, Chinese adolescents likely exhibit distinct sleep quality types, due to the differences in living space and sleep practices across Western and Chinese cultures. For example, room-share is more common in Asian countries (Mindell, Sadeh, Kwon, & Goh, 2015). Due to the “unaffordable” house prices in Hong Kong (Forrest & Xian, 2018), sharing a room with siblings or other family members is common in Hong Kong. A previous study found that school-aged children with their own bedroom had better sleep quality (i.e., a shorter sleep onset latency), compared with children who shared a bedroom (Mindell, Meltzer, Carskadon, & Chervin, 2009). In addition, it is widely documented that Chinese societies and families have an emphasis on academic success and achievements (Huang & Gove, 2015). While such emphasis induces a great amount of academic stress, academic stress has been reported as the risk factor for poor sleep quality (e.g., sleep disturbances) among Hong Kong Chinese adolescents (Chung & Cheung, 2008). Thus, research examining Hong Kong Chinese adolescents’ sleep quality types is warrant.

On the other hand, there are only few studies examining the associations of sleep quality with some aspects of psychological and physical health among Chinese adolescents, such as poor sleep quality with depression (Cheung & Wong, 2011), and sleep duration with suicide attempts (Liu, 2004). In addition, little is known about how Chinese adolescent sleep quality types correlate with psychological and physical health. Given that sleep is an important predictor of adolescent psychological and physical health in Western cultures (for review, see Owens & Adolescent Sleep Working Group, 2014), knowledge of Hong Kong Chinese adolescents’ sleep quality types and the associations of sleep quality types with psychological and physical health can help develop culturally appropriate interventions for Hong Kong Chinese adolescents. Therefore, it is imperative to identify the sleep quality types of Hong Kong Chinese adolescents and examine their associations with psychological and physical health.

The Current Study

To address these recommendations, the first aim of this study was to use LCA, with all seven components of sleep quality, to identify sleep quality types in Hong Kong Chinese adolescents. In addition, this study also examined the predictive effect of gender on the probability that an individual belongs to one class over another class (sleep quality type). The second aim of this study was to explore whether these sleep quality types displayed different levels of eight commonly examined distal outcomes of sleep quality. These outcomes include four psychological characteristics: self-esteem, life satisfaction, positive well-being, and depressive symptoms, and four health risk behaviors: physical inactivity, smoking, drinking, and suicidal thoughts.

Based on the literature reviewed, distinct sleep quality types were expected to be identified. Given the higher prevalence of poor sleep quality in female adolescents (Galland et al., 2017), it is plausible that there would be a significant predictive effect of gender on the likelihood of class assignment (membership of sleep quality types). In relation to the second aim of this study, the sleep quality types were expected to be distinctively associated with psychological characteristics and health risk behaviors.

Method

Participants and Procedure

Ethical approval was obtained from the university review board. Formal consent was obtained from the parents of the participants. Participants were recruited from 13 secondary schools located in different districts of Hong Kong (N = 1,368; male: n = 738, 54.1%; mean age = 13.48 ± 0.75 years, in the range from 12 to 18). The self-report questionnaire containing all measures of interest was administered in classrooms during regular class sessions. Analyses were run for 1,310 participants who completed the whole measure of sleep quality (male: n = 703, 53.7%; mean age = 13.48 ± 0.75 years). Where data were available, comparisons between those included in the study and those excluded due to missing data indicated no significant differences in relation to age, gender, level of psychological health, participation in three of the four physical health risk behaviors, and six of the seven components of sleep quality. However, the excluded participants had slightly lower scores for daytime dysfunction (mean difference = .27; p = .03). Meeting the national guidelines for physical activity (American College Health Association, 2012) was more common among the excluded participants (p = .02).

Measures

Sleep quality

Sleep quality was assessed using the Pittsburgh Sleep Quality Index (PSQI; Buysse et al., 1989). The PSQI is widely used and has been validated in different cultures (Mindell, Sadeh, Kwon, & Goh, 2013). The Chinese version of the PSQI has been demonstrated to be a psychometrically sound measure (Ho & Fong, 2014; Tsai et al., 2005) and has been used for Hong Kong Chinese adolescents (Cheung & Wong, 2011). The PSQI consists of nine questions, which were developed to assess the seven components of sleep quality (subjective sleep quality, sleep onset latency, sleep duration, habitual sleep efficiency, sleep disturbances, use of sleeping medication, and daytime dysfunction). A sample question for PSQI is “During the past month, how would you rate your sleep quality overall?” The scores of the seven components were calculated according to the scoring guidelines (Buysse et al., 1989), with a possible range of 0 to 3. The composite score of the Global Sleep Quality Index was acquired by summing the seven scores of the sleep quality components, with a possible range of 0 to 21. Higher scores represented poorer sleep quality.

Self-esteem

Self-esteem was assessed using the 10-item Rosenberg Self-Esteem Scale (RSE; Rosenberg, 1965, 1979) with five items reverse-scored. A sample item is as follows: “I take a positive attitude toward myself.” The Chinese version of the RSE has been extensively used and demonstrated to be reliable and valid for Chinese populations (Zhao, Kong, & Wang, 2012). A 4-point scale ranging from 1 (strongly disagree) to 4 (strongly agree) was used, with higher scores reflecting more positive self-evaluations. Internal consistency of the RSE was good in this study (Cronbach’s α = .81).

Positive well-being

Positive well-being was assessed by the five-item scale of the World Health Organization–Five Well-Being Index (WHO-5; Topp, Østergaard, Søndergaard, & Bech, 2015). The scale captures positive mood, vitality, and interests as marker of positive well-being. A sample item is as follows: “I have felt cheerful and in good spirits.” Topp et al. (2015) developed evidence for the reliability and validity of the scale. A 6-point response scale ranging from 0 (never) to 5 (almost all of the time) was used, with higher scores reflecting a higher level of positive well-being. Internal consistency of the WHO-5 was excellent in this study (Cronbach’s α = .92).

Life satisfaction

Life satisfaction was assessed using five items of the Satisfaction With Life Scale (SWLS; Diener, Emmons, Larsen, & Griffin, 1985), which has been widely used in Chinese populations (e.g., Lai & Ma, 2016). A sample item is as follows: “I am satisfied with my life.” Participants responded to this scale on a 6-point Likert-type scale, ranging from 0 (strongly disagree) to 5 (strongly agree), with higher scores reflecting greater life satisfaction. Internal consistency of the SWLS was good (Cronbach’s α = .85).

Depressive symptoms

Depressive symptoms were assessed using Patient Health Questionnaire–9 (PHQ-9; Spitzer, Kroenke, Williams, & Patient Health Questionnaire Primary Care Study Group, 1999). A sample item is as follows: “Feeling down, depressed, or hopeless.” PHQ-9 has been used in Chinese contexts (e.g., Lai & Ma, 2016). A 4-point response scale, ranging from 0 (not at all) to 3 (nearly every day), was used, with higher scores reflecting more depressive symptoms. Internal consistency of the PHQ-9 was excellent in this study (Cronbach’s α = .91).

Substance use and suicidal thoughts

The frequency of cigarette and alcohol consumption as well as having suicidal thoughts over the previous 3 months was assessed using three questions: “Over the past 3 months, have you ever . . . (e.g., had suicidal thoughts)?” The response options ranged from 0 (never in my lifetime) to 6 (more than 5 times per week). The response options were re-coded (0 = 0 and 1-6 = 1) for analysis.

Physical inactivity

The frequency of participation in moderate to vigorous physical activity was assessed using one question: “Over the past 7 days, how often did you participate in moderate to vigorous physical activities (e.g., jogging, swimming, cycling, and playing basketball), (one time equals to at least 30 minutes)?” The response options ranged from 0 (never) to 7 (more than 7 times). The response options were re-coded for analysis (5-7 = 0 and 0-4 = 1) according to the national guidelines for physical activity (American College Health Association, 2012), with 0 = meeting the national guidelines and 1 = not meeting the national guidelines.

Statistical Analysis

This study had two analytic aims. The first aim was to identify meaningful subgroups with distinctive sleep quality types. LCA was used to extract classes of participants with similar configurations of the seven components of sleep quality. LCA was conducted in Mplus version 7.31, using maximum likelihood estimation with robust standard errors (MLR). Models were specified with 1 to 5 classes (k = 1-5; Nylund, Asparouhov, & Muthén, 2007) and estimated with 2,000 random sets of starting values, each allowing 200 iterations (Hipp & Bauer, 2006).

To identify the best model with a good number of classes, theoretical interpretability and several fit indexes were considered. These included the log-likelihood value, the Akaike information criterion (AIC; Akaike, 1987), the Bayesian information criterion (BIC; Schwarz, 1978), the sample-size–adjusted BIC (SSA-BIC), the Lo–Mendell–Rubin adjusted likelihood ratio test (LMR; Lo, Mendell, & Rubin, 2001), the bootstrap likelihood ratio test (BLRT), and entropy. Lower values for AIC, BIC, and SSA-BIC indicate a better fitting model (Nylund et al., 2007). Significant p values for LMR and BLRT suggest that the k-class model has a significantly better fit than the model with one less class (Nylund et al., 2007). A higher value of entropy suggests greater separation between classes. In addition, the average latent class probabilities were evaluated. Values larger than .8 on the main diagonal of the classification matrix indicate a high reliability of classification (Rost, 2006). The optimal latent class model was chosen by closely comparing the above model fit statistics with the theoretical meaning of the extracted classes.

After identification of the best model, gender was specified as the covariate to examine its predictive effect on the probability that an individual belongs to one class over another class.

The second aim was to examine whether there would be mean differences in psychological characteristics and probability differences in participation in health risk behaviors across the classes. These involved auxiliary variables (i.e., the distal outcomes) in relation to the final LCA solution. The distal outcomes (i.e., self-esteem, life satisfaction, positive well-being, depressive symptoms, probability of participation in physical inactivity, smoking, drinking, and having suicidal thoughts) were tested using Lanza’s model-based approach (Lanza, Tan, & Bray, 2013). Listwise deletion was used for auxiliary variables with missing data (see the percentages in Table 1).

Table 1.

Descriptive Statistics, Frequencies, and Correlations of Variables (n = 1,310).

Variable	1	2	3	4	5	6	7	8	9	10	11	12	13	14	15
1. Subjective sleep quality	—
2. Sleep onset latency	.38**	—
3. Sleep duration	.36**	.15**	—
4. Sleep efficiency	.14**	.27**	.56**	—
5. Sleep disturbances	.38**	.45**	.15**	.11*	—
6. Use of sleeping medication	.32**	.28**	.13*	.08	.37**	—
7. Daytime dysfunction	.53**	.21**	.29**	.07	.32**	.14*	—
8. Self-esteem	−.30**	−.22**	−.23**	−.07	−.31**	−.15*	−.36**	—
9. Life satisfaction	−.32**	−.22**	−.26**	−.06	−.22**	−.04	−.36**	.61**	—
10. Positive well-being	−.34**	−.26**	−.27**	−.10*	−.29**	−.14**	−.35**	.57**	.78**	—
11. Depressive symptoms	.42**	.29**	.27**	.09*	.42**	.20**	.46**	−.54**	−.43**	−.44**	—
12. Physical inactivity	−.02	.06	.03	.12*	−.01	−.10	.00	−.10**	−.12**	−.17**	.00	—
13. Smoke	.14*	.10	.22**	.02	.15*	.31**	.10	−.20**	−.17**	−.15**	.27**	−.09	—
14. Drinking	.23**	.13**	.16**	.04	.19**	.20*	.27**	−.22**	−.25**	−.18**	.30**	−.13*	.71**	—
15. Suicide	.33**	.18**	.28**	.12*	.36**	.10	.37**	−.51**	−.46**	−.38**	.49**	−.03	.32**	.42**	—
16. Gender	−.13**	−.03	−.19**	.05	−.08	−.10	−.13**	.11**	.03	.07	−.16**	−.25**	.07	−.07	−.25**
$\bar{X}$	—	—	—	—	—	—	—	2.75	2.78	2.96	1.19	—	—	—	—
SD	—	—	—	—	—	—	—	.50	1.15	1.14	.82	—	—	—	—
Skewness	—	—	—	—	—	—	—	−.36	−.33	−.40	.48	—	—	—	—
Kurtosis	—	—	—	—	—	—	—	.04	−.29	−.03	−.07	—	—	—	—
Endorsement (%)	—	—	—	—	—	—	—	—	—	—	—	24.6	4.8	24.2	21.9
Missing (%)	—	—	—	.3	—	—	—	1.1	1	0.9	2	0.2	0.6	0.7	0.5

Note. Missing data were 0.2% for gender. The estimates (.00 or –.00) are not exactly zero. They range from .004 to –.003.

p < .05. **p < .01.

Results

Descriptive Analyses

Table 1 presents the descriptive statistics, frequencies, and correlations of all of the variables used in this study. The results suggest that most of the seven components of sleep quality significantly covaried with correlation coefficients ranging from .11 to .56. In general, the components of sleep quality were negatively associated with self-esteem, life satisfaction, and positive well-being (r = −.36 to –.10, p < .05) and positively related to depressive symptoms (r = .09-.46, p < .05). The four health risk behaviors were positively associated with some of the components of sleep quality (r = .13-.37, p < .05).

Latent Class Analyses

Table 2 provides the fit statistics of the models with one to five classes. In all of the models, the log likelihood, AIC, and SSA-BIC incrementally decreased as the number of classes increased. Although the models featuring three to five classes had nonsignificant p values for LMR, they were considered the candidate model solutions, as BLRT is a more accurate indicator than LMR for determining number of classes (Nylund et al., 2007). The 5-class model showed the lowest values for AIC and SSA-BIC, with the second highest entropy and a significant p value for BLRT. However, close inspection of the substantive interpretability revealed that some classes of the 5-class model displayed primarily medium-sized conditional response probabilities for the seven components of sleep quality, making the solution difficult to interpret. Therefore, the 4-class solution was retained as the preferred model because it exhibited the second-lowest AIC, BIC, and SSA-BIC values, with the highest entropy and a significant p value for BLRT (see Table 2). Additional support for the 4-class solution was given by the average posterior probabilities of latent class, which ranged from .81 to .92, with low cross-probabilities, varying from .00 to .11, indicating a reasonable level of classification accuracy and a low average likelihood of misclassification.

Table 2.

Summary of Fit Indices for the Latent Classes.

Model	Log likelihood	Free parameters	AIC	BIC	Sample-size–adjusted BIC	LMR	BLRT	Entropy	The smallest estimated class size: n (%)
1 class	−8,205.985	21	16,453.970	16,562.703	16,495.996	—	—	—	1,310 (100%)
2 classes	−7,879.736	43	15,845.473	16,068.117	15,931.527	p = .000	p = .000	.61	617 (47%)
3 classes	−7,760.472	65	15,650.943	15,987.499	15,781.025	p > .05	p = .000	.66	203 (15%)
4 classes	−7,690.411	87	15,554.822	16,005.289	15,728.931	p > .05	p = .000	.76	99 (8%)
5 classes	−7,630.818	109	15,479.635	16,044.013	15,697.772	p > .05	p = .000	.73	84 (6%)

Note. The smallest estimated class size is the estimated class size of the smallest class in the models. AIC = Akaike information criterion; BIC = Bayesian information criterion; LMR = Lo–Mendell–Rubin adjusted likelihood ratio test; BLRT= bootstrap likelihood ratio test.

The 4-class solution yielded four distinct and readily interpretable classes that captured the sleep quality of the adolescent participants in this study (see Table 3). The first class, labeled good sleepers¹ (n = 343; 26% of the participants), mainly comprised teenagers with very good subjective sleep quality, long sleep duration (>7 hours), and high habitual sleep efficiency (≥85%). Among the good sleepers, about half experienced no sleep onset latency, and one third experienced no sleep disturbance or daytime dysfunction. Relative to the other classes, this class featured the lowest probability of reporting moderate to severe sleep onset latency (score: 3-6) and daytime dysfunction (score: 3-6), and some to many sleep disturbances (score: 10-27).

Table 3.

Probability of Sleep Quality Components in the Four Sleep Quality Classes.

Sleep characters	Class 1Good sleepers (n = 343; 26%)	Class 2Fair sleepers (n = 726; 55%)	Class 3Short sleepers (n = 99; 8%)	Class 4Poor sleepers (n = 142; 11%)
Subjective sleep quality
Very good	.71	.02	.19	.00
Fairly good	.27	.69	.53	.10
Fairly bad	.00	.29	.28	.52
Very bad	.02	.00	.00	.39
Score of sleep onset latency
0	.46	.17	.15	.08
1-2	.46	.58	.42	.36
3-4	.06	.20	.24	.34
5-6	.01	.04	.20	.23
Sleep duration
>7 hr	.73	.54	.05	.21
>6 hr and ≤7 hr	.19	.34	.23	.27
>5 hr and ≤6 hr	.04	.11	.42	.31
≤5 hr	.04	.02	.31	.21
Habitual sleep efficiency
≥85%	.95	.93	.00	.78
75%-84%	.04	.07	.45	.14
65%-74%	.01	.01	.26	.03
≤64%	.00	.00	.29	.05
Score of sleep disturbances
0	.29	.04	.09	.06
l-9	.68	.86	.76	.50
10-18	.03	.11	.15	.41
19-27	.00	.00	.00	.03
Use of sleeping medication
Not during the past month	1.00	.96	.97	.92
Less than once a week	.00	.04	.00	.04
Once or twice a week	.00	.01	.03	.01
Three or more times a week	.00	.00	.00	.02
Score of daytime dysfunction
0	.29	.03	.08	.01
1-2	.48	.40	.44	.00
3-4	.20	.48	.44	.44
5-6	.03	.09	.05	.56

The second class, labeled fair sleepers (n = 726; 55% of the participants), was the largest class. Members of this class presented the second-lowest probability of short sleep duration (≤6 hours) and the second highest probability of high habitual sleep efficiency (≥85%). Compared with good sleepers, fair sleepers showed a higher probability of declaring poorer subjective sleep quality and moderate to severe daytime dysfunction. The third class, labeled short sleepers (n = 99; 8% of the participants), was marked by the highest probability of short sleep duration (≤5 hours) and low habitual sleep efficiency (≤64%). However, short sleepers displayed better subjective sleep quality than fair sleepers.

The fourth class, labeled poor sleepers (n = 142; 11% of the participants), consisted of adolescents with bad subjective sleep quality and severe daytime dysfunction. Of the four classes, this class showed the lowest probability of good subjective sleep quality and no sleep onset latency. In general, almost all of the adolescents (in all classes) reported being nonusers of sleeping medication.

Regarding gender effect, gender significantly predicted the likelihood of membership in two of six class comparisons. Compared with Class 1 (good sleepers), being male was less likely to belong to Class 2 (fair sleepers; b = −.81, p < .01) and Class 4 (poor sleepers; b = −.52, p < .01).

Analyses of Auxiliary Variables (Distal Outcomes)

The mean levels of psychological characteristics and probabilities of participating in health risk behaviors significantly varied with the four classes, with the exception of smoking (see the overall chi-square values in Table 4). Table 4 and Figure 1 present the means and standardized means of the psychological characteristics by the four sleep quality classes. In general, good sleepers had the best psychological health, with the highest mean levels of self-esteem, life satisfaction, and positive well-being, and the lowest mean levels of depressive symptoms. Poor sleepers had the worst psychological health, with the lowest mean levels of self-esteem, life satisfaction, and positive well-being, and the highest mean levels of depressive symptoms. For those two classes between good and poor sleepers, short sleepers displayed better psychological health than fair sleepers, with above-average self-esteem and below-average depressive symptoms. Table 4 and Figure 2 show the probabilities of participating in health risk behaviors. Among all four classes, good sleepers had the lowest probability of having suicidal thoughts; short sleepers had the highest probability of being physically inactive, and poor sleepers had the highest probability of drinking alcohol and having suicidal thoughts. Generally, good sleepers had the lowest probability of participating in unhealthy behaviors, whereas poor sleepers had the highest.

Table 4.

Means of Psychological Characteristics (Continuous Variables) and Probabilities of Participating in Unhealthy Behaviors (Binary Variables) by the Four Sleep Quality Classes.

	(a) Class 1: Good sleepers	(b) Class 2: Fair sleepers	(c) Class 3: Short sleepers	(d) Class 4: Poor sleepers	Overall χ²
Self-esteem	3.04_b,c,d	2.64_a,c,d	2.79_a,b,d	2.27_a,b,c	363.503***
Life satisfaction	3.51_b,c,d	2.77_a,d	2.76_a,d	1.88_a,b,c	313.766***
Positive well-being	3.79_b,c,d	2.91_a,d	2.96_a,d	1.98_a,b,c	383.841***
Depressive symptoms	0.59_b,c,d	1.21_a,c,d	1.04_a,b,d	2.15_a,b,c	661.612***
Physical inactivity	0.76_c	0.74_c	0.87_a,b,d	0.75_c	8.276*
Smoking	0.03_d	0.05	0.04	0.09_a	5.153
Drinking	0.14_b,d	0.22_a,d	0.23_d	0.49_a,b,c	53.656***
Suicidal thoughts	0.07_b,c,d	0.23_a,d	0.23_a,d	0.52_a,b,c	51.995***

Note. Subscripts indicate class means or probabilities that are significantly different at least at p < .05. Due to the use of listwise deletion, available data were n = 1,296 for self-esteem, n = 1,297 for life satisfaction, n = 1,298 for positive well-being, n = 1,284 for depressive symptoms, n = 1,307 for physical inactivity, n = 1,302 for smoking, n = 1,301 for drinking, and n = 1,303 for suicidal thoughts.

p < .05. ***p < .001.

Figure 1.

Standardized means of psychological characteristics by types of sleepers.

Figure 2.

Probabilities of participation in health risk behaviors by types of sleepers.

Discussion

This study attempted to examine Hong Kong Chinese adolescents’ sleep quality using a person-centered approach (LCA), and to investigate the associations of sleep quality with psychological characteristics and health risk behaviors. As expected, distinct classes featuring different configurations of the seven components of sleep quality were identified. Importantly, the four sleep quality classes were meaningful and exhibited significant differences in the psychological characteristics and health risk behaviors.

This study is the first to explore individual differences in Hong Kong Chinese adolescents’ sleep quality, using LCA. Their sleep quality was captured in four classes: good sleepers, fair sleepers, short sleepers, and poor sleepers. Fair sleepers comprised the largest class (55% of participants) and exhibited modest subjective sleep quality and sleep duration, few sleep problems (i.e., sleep onset latency and sleep disturbance), and moderate frequency of daytime dysfunction. Good sleepers comprised the second-largest class (26% of participants) and displayed relatively good sleep quality with occasional sleep problems and low frequency of daytime dysfunction. Taken together, these results indicate that a vast majority of the adolescents in Hong Kong sleep well, which is consistent with Cheung and Wong’s (2011) finding that about 70% of Hong Kong adolescents were not insomniacs. Poor sleepers and short sleepers were the third largest and smallest classes, respectively (11% and 8% of participants). Short sleepers presented similar characteristics to fair sleepers, except for subjective sleep quality, sleep duration, and habitual sleep efficiency. Short sleepers were more likely to report better subjective sleep quality than fair sleepers. Among all four classes, short sleepers had the worst habitual sleep efficiency and were most likely to sleep less. Poor sleepers were the worst for all other components of sleep quality, by much greater margins. These results are broadly consistent with previous studies in non-Chinese samples for identifying different sleep quality subgroups (Magee et al., 2016; Ownby, Peruyera, Acevedo, Loewenstein, & Sevush, 2014). In general, this study provided evidence for the existence of distinct types of sleep quality among Chinese adolescents. The results of LCA also advanced the findings of variable-centered approaches by offering nuanced combinations of individuals’ sleep quality, for example, the two middle classes (i.e., fair sleepers and short sleepers).

There was partial support for the predictive effect of gender on the probability that an individual belongs to one class over another class. Gender affected the classification when good sleepers class was the comparison class. Compared with the good sleepers class, female had higher probabilities of being in the fair sleepers and poor sleepers classes. Therefore, consistent with previous studies using variable-centered approaches, this study provided further evidence for the higher prevalence of poor sleep quality in female adolescents (Galland et al., 2017).

Through accounting for the interaction of the seven components of sleep quality, LCA offers another way to understand the associations between sleep quality and its correlates. Good sleepers had the best psychological health, exhibiting higher self-esteem, life satisfaction, and positive well-being and fewer depressive symptoms, whereas poor sleepers had the worst psychological health. Fair sleepers and short sleepers showed moderate psychological health, which was significantly different from good sleepers and poor sleepers. Despite fair sleepers and short sleepers sharing some characteristics of sleep quality, they exhibited significant differences in self-esteem and depressive symptoms. Short sleepers appeared to have higher self-esteem and fewer depressive symptoms than fair sleepers. These findings provided extra evidence for the “four sleep quality classes” solution.

In terms of health risk behaviors, significantly different from the other three classes, short sleepers showed the highest probability of physical inactivity. This finding is inconsistent with previous research with Australian adults (Magee et al., 2016). Magee and colleagues found that long sleepers showed lowest levels of physical activity and poor sleepers (featuring short sleep duration) showed modest levels of physical activity. However, the negative effect of short sleep duration on physical activity was demonstrated by a cross-cultural study with adolescents from 12 countries (Lin et al., 2018). Therefore, it is possible that the adolescent sleep quality class featuring short sleep duration is associated with levels of physical activity differently from adults’ sleep quality class. In general, poor sleepers had the highest probability of smoking², drinking, and having suicidal thoughts. Therefore, beyond the adverse influence of poor sleep quality on health in itself, it could be the cause of different health risk behaviors (M. M. Wong et al., 2009).

Overall, these results showed the strengths of the adoption of LCA. Using LCA granted us unique and meaningful information about the heterogeneity in the sleep quality of Hong Kong Chinese adolescents, especially in discerning differences in the two subgroups with moderate sleep quality (i.e., fair sleepers and short sleepers). LCA also enabled us to understand the relationship between sleep quality and health that only emerges in specific subgroups (sleep quality types). For example, this study found that short sleepers appeared to have higher self-esteem and fewer depressive symptoms than fair sleepers. These findings contradict the results of previous studies using variable-centered approaches. A longitudinal study demonstrated that shorter sleep duration was related to subsequent lower self-esteem and higher depression (Fredriksen, Rhodes, Reddy, & Way, 2004). On the other hand, using the subscales of the PSQI, a longitudinal study showed that better subjective sleep quality was associated with subsequent higher self-esteem and lower depression (M. L. Wong et al., 2013). Therefore, the adoption of LCA may be the reason for the contradictive findings because LCA can account for the relationships among the components of sleep quality. Compared with fair sleepers, short sleepers displayed higher probability of declaring good subjective sleep quality. Therefore, it is plausible that short sleepers showed higher self-esteem and fewer depressive symptoms than fair sleepers because LCA accounted for the interaction of better subjective sleep quality with shorter sleep duration.

One major theoretical contribution of this study is the provision of meaningful information about the distinct configurations of the sleep quality components among Chinese adolescents. Despite the methodological differences in the literature on sleep quality types, the current classifications of sleep quality types are similar to previous findings of identifying subgroups with different sleep quality in Western cultures (Magee et al., 2016; Ownby et al., 2014). Therefore, this study extended the literature of sleep quality types to Chinese populations. This study is also a welcomed addition to the scarce literature examining the associations of sleep quality and psychological characteristics and health risk behaviors among Chinese adolescents.

Several practical implications also emerge from this study. In general, the results supported that the four adolescent sleep quality types were distinctively associated with psychological and physical health. Such findings provided support for the meaningfulness of these sleep quality types. The account of the interplay of the sleep quality components by LCA can reveal the dynamics that might undermine the success of a policy or intervention for improving adolescent health. For example, when studies examining the effectiveness of an intervention for depression (e.g., increasing sleep duration) yield inconsistent findings, using LCA can help researchers understand whether effectiveness of an intervention is affected by the interplay of the components of sleep quality. The findings of this study also indicate a need to implement policies or design interventions for improving adolescent sleep quality. For example, a systematic review suggested that delayed school start times increased weekday sleep duration of adolescents (Minges & Redeker, 2016). Therefore, policymakers may implement educational policies for promoting later school start times to improve adolescent sleep quality. In addition, while academic stress has been reported as the risk factor for sleep disturbances (Chung & Cheung, 2008), high expectations from parents have been reported as the leading source of academic stress (Sun, Dunne, & Hou, 2012). Therefore, schools may implement policies for encouraging good communication between parents and teachers to help parents set reasonable academic expectations. Furthermore, sleep hygiene training for reducing hot beverage intake and evening technology time (Galland at al., 2017) may be introduced as extracurricular activities or special workshops during class sessions for enhancing adolescent sleep quality.

Despite the above implications, this study has some limitations. As with any study using only self-reporting measures, the current findings may display same source bias. Future studies collecting data from parents and using objective measures (e.g., actigraphy) could provide a more comprehensive portrayal of adolescent sleep patterns. In addition, other factors (e.g., puberty socioeconomic status and bedroom-sharing) are associated with sleep quality (Mindell et al., 2009; Pieters, Van Der Vorst, Burk, Wiers, & Engels, 2010). Future research including these factors could provide valuable insight.

In conclusion, sleep literature has been confined by the dominant use of variable-centered approaches or the use of only some components of sleep quality. This study was the first attempt to examine Hong Kong Chinese adolescents’ sleep quality using a person-centered approach (LCA) with multiple components of sleep quality. This study successfully demonstrated the existence of distinct types of sleep quality, and that these sleep quality types could differentiate adolescent psychological characteristics and health risk behaviors. Adolescents, showing better types of sleep quality, had better psychological health and partook in fewer health risk behaviors. This study provided complementary information about the influence of sleep quality on psychological and physical health.

Footnotes

Declaration of Conflicting Interests

The author(s) declared no potential conflicts of interest with respect to the research, authorship, and/or publication of this article.

Funding

The author(s) disclosed receipt of the following financial support for the research, authorship, and/or publication of this article: This study was supported by funding from the Early Career Scheme of the Research Grants Council of the University Grants Committee of Hong Kong (Grant 25401414) and The Hong Kong Polytechnic University (Code: 1-ZVKM).

ORCID iDs

Catie C. W. Lai

Cecilia M. S. Ma

Notes

Author Biographies

Catie C. W. Lai is a junior researcher in the Department of Applied Social Sciences at the Hong Kong Polytechnic University. Her research interests include sleep, well-being, youth problems, positive youth development, personality, suicide, psychometric testing, cross-cultural differences, and animal-assisted therapy.

Cecilia M. S. Ma is an assistant professor in the Department of Applied Social Sciences at the Hong Kong Polytechnic University. Her research interests include youth problems, positive youth development, structural equation modeling, psychometric testing, growth curve modeling, multivariate statistics, and cultural socialization process.

References

Akaike

(1987). Factor analysis and AIC. Psychometrika, 52, 317-332.

American College Health Association. (2012). National College Health Assessment II: Reference group executive summary. Hanover, MD: Author.

Buysse

D. J.

Reynolds

C. F.

Monk

T. H.

Berman

S. R.

Kupfer

D. J.

(1989). The Pittsburgh Sleep Quality Index: A new instrument for psychiatric practice and research. Psychiatry Research, 28, 193-213.

Cheung

L. M.

Wong

W. S.

(2011). The effects of insomnia and internet addiction on depression in Hong Kong Chinese adolescents: An exploratory cross-sectional analysis. Journal of Sleep Research, 20, 311-317.

Chung

K. F.

Chan

M. S.

Lam

Y. Y.

Lai

C. S. Y.

Yeung

W. F.

(2017). School-based sleep education programs for short sleep duration in adolescents: A systematic review and meta-analysis. Journal of School Health, 87, 401-408.

Chung

K. F.

Cheung

M. M.

(2008). Sleep-wake patterns and sleep disturbance among Hong Kong Chinese adolescents. Sleep, 31, 185-194.

Dagys

McGlinchey

E. L.

Talbot

L. S.

Kaplan

K. A.

Dahl

R. E.

Harvey

A. G.

(2012). Double trouble? The effects of sleep deprivation and chronotype on adolescent affect. Journal of Child Psychology and Psychiatry, 53, 660-667.

Diener

E. D.

Emmons

R. A.

Larsen

R. J.

Griffin

(1985). The Satisfaction With Life Scale. Journal of Personality Assessment, 49, 71-75.

Forrest

Xian

(2018). Accommodating discontent: Youth, conflict and the housing question in Hong Kong. Housing Studies, 33, 1-17.

10.

Fredriksen

Rhodes

Reddy

Way

(2004). Sleepless in Chicago: Tracking the effects of adolescent sleep loss during the middle school years. Child Development, 75, 84-95.

11.

Galland

B. C.

Gray

A. R.

Penno

Smith

Lobb

Taylor

R. W.

(2017). Gender differences in sleep hygiene practices and sleep quality in New Zealand adolescents aged 15 to 17 years. Sleep Health, 3, 77-83.

12.

Goldstein

T. R.

Bridge

J. A.

Brent

D. A.

(2008). Sleep disturbance preceding completed suicide in adolescents. Journal of Consulting and Clinical Psychology, 76, 84-91.

13.

Gradisar

Gardner

Dohnt

(2011). Recent worldwide sleep patterns and problems during adolescence: A review and meta-analysis of age, region, and sleep. Sleep Medicine, 12, 110-118.

14.

Hipp

J. R.

Bauer

D. J.

(2006). Local solutions in the estimation of growth mixture models. Psychological Methods, 11, 36-53.

15.

Hirshkowitz

Whiton

Albert

S. M.

Alessi

Bruni

DonCarlos

Kheirandish-Gozal

(2015). National Sleep Foundation’s updated sleep duration recommendations. Sleep Health, 1, 233-243.

16.

R. T.

Fong

T. C.

(2014). Factor structure of the Chinese version of the Pittsburgh Sleep Quality Index in breast cancer patients. Sleep Medicine, 15, 565-569.

17.

Huang

G. H.

Gove

(2015). Asian parenting styles and academic achievement: Views from eastern and western perspectives. Education, 135, 389-397.

18.

Kaneita

Ohida

Osaki

Tanihata

Minowa

Suzuki

Hayashi

(2006). Insomnia among Japanese adolescents: A nationwide representative survey. Sleep, 29, 1543-1550.

19.

Lai

C. C.

C. M.

(2016). The mediating role of social support in the relationship between psychological well-being and health-risk behaviors among Chinese university students. Health Psychology Open, 3. doi:10.1177/2055102916678106

20.

Lanza

S. T.

Tan

Bray

B. C.

(2013). Latent class analysis with distal outcomes: A flexible model-based approach. Structural Equation Modeling: A Multidisciplinary Journal, 20, 1-26.

21.

Lin

R. M.

Xie

S. S.

Yan

Y. W.

Chen

Y. H.

Yan

W. J.

(2017). Perfectionism and adolescent sleep quality: The mediating role of repetitive negative thinking. Journal of Health Psychology, 22, 1-11.

22.

Lin

Tremblay

M. S.

Katzmarzyk

P. T.

Fogelholm

Lambert

E. V.

Standage

(2018). Temporal and bi-directional associations between sleep duration and physical activity/sedentary time in children: An international comparison. Preventive Medicine, 111, 436-441.

23.

Liu

(2004). Sleep and adolescent suicidal behavior. Sleep, 27, 1351-1358.

24.

Mendell

N. R.

Rubin

D. B.

(2001). Testing the number of components in a normal mixture. Biometrika, 88, 767-778.

25.

Lovato

Gradisar

(2014). A meta-analysis and model of the relationship between sleep and depression in adolescents: Recommendations for future research and clinical practice. Sleep Medicine Reviews, 18, 521-529.

26.

Magee

C. A.

Reddy

Robinson

McGregor

(2016). Sleep quality subtypes and obesity. Health Psychology, 35, 1289-1297.

27.

McKnight-Eily

L. R.

Eaton

D. K.

Lowry

Croft

J. B.

Presley-Cantrell

Perry

G. S.

(2011). Relationships between hours of sleep and health-risk behaviors in US adolescent students. Preventive Medicine, 53, 271-273.

28.

Mindell

J. A.

Meltzer

L. J.

Carskadon

M. A.

Chervin

R. D.

(2009). Developmental aspects of sleep hygiene: Findings from the 2004 national sleep foundation sleep in America poll. Sleep Medicine, 10, 771-779.

29.

Mindell

J. A.

Sadeh

Kwon

Goh

D. Y.

(2013). Cross-cultural comparison of maternal sleep. Sleep, 36, 1699-1706.

30.

Mindell

J. A.

Sadeh

Kwon

Goh

D. Y.

(2015). Relationship between child and maternal sleep: A developmental and cross-cultural comparison. Journal of Pediatric Psychology, 40, 689-696.

31.

Minges

K. E.

Redeker

N. S.

(2016). Delayed school start times and adolescent sleep: A systematic review of the experimental evidence. Sleep Medicine Reviews, 28, 86-95.

32.

C. W.

Luo

Heng

B. H.

(2014). Health status profiles in community-dwelling elderly using self-reported health indicators: A latent class analysis. Quality of Life Research, 23, 2889-2898.

33.

Nylund

K. L.

Asparouhov

Muthén

B. O.

(2007). Deciding on the number of classes in latent class analysis and growth mixture modeling: A Monte Carlo simulation study. Structural Equation Modeling, 14, 535-569.

34.

Olds

Blunden

Petkov

Forchino

(2010). The relationships between sex, age, geography and time in bed in adolescents: A meta-analysis of data from 23 countries. Sleep Medicine Reviews, 14, 371-378.

35.

Ortega

F. B.

Chillón

Ruiz

J. R.

Delgado

Albers

Álvarez-Granda

J. L.

Castillo

M. J.

(2010). Sleep patterns in Spanish adolescents: Associations with TV watching and leisure-time physical activity. European Journal of Applied Physiology, 110, 563-573.

36.

Owens

, & Adolescent Sleep Working Group. (2014). Insufficient sleep in adolescents and young adults: An update on causes and consequences. Pediatrics, 134, e921-e932.

37.

Ownby

R. L.

Peruyera

Acevedo

Loewenstein

Sevush

(2014). Subtypes of sleep problems in patients with Alzheimer disease. The American Journal of Geriatric Psychiatry, 22, 148-156.

38.

Paunio

Korhonen

Hublin

Partinen

Kivimäki

Koskenvuo

Kaprio

(2008). Longitudinal study on poor sleep and life dissatisfaction in a nationwide cohort of twins. American Journal of Epidemiology, 169, 206-213.

39.

Pieters

Burk

W. J.

Van Der Vorst

Dahl

R. E.

Wiers

R. W.

Engels

R. C.

(2015). Prospective relationships between sleep problems and substance use, internalizing and externalizing problems. Journal of Youth and Adolescence, 44, 379-388.

40.

Pieters

Van Der Vorst

Burk

W. J.

Wiers

R. W.

Engels

R. C.

(2010). Puberty-dependent sleep regulation and alcohol use in early adolescents. Alcoholism: Clinical & Experimental Research, 34, 1512-1518.

41.

Roane

B. M.

Taylor

D. J.

(2008). Adolescent insomnia as a risk factor for early adult depression and substance abuse. Sleep, 31, 1351-1356.

42.

Rosenberg

(1965). Society and the adolescent self-image. Princeton, NJ: Princeton university press.

43.

Rosenberg

(1979). Conceiving the self. New York, NY: Basic Books.

44.

Rost

(2006). Latent-class-analyse [Latent class analysis]. In Petermann

Eid

(Eds.), Handbuch der psychologischen diagnostik [Handbook of psychological assessment] (pp. 275-287). Göttingen, Germany: Hogrefe.

45.

Schwarz

. (1978). Estimating the dimension of a model. The Annals of Statistics, 6, 461-464.

46.

Shochat

Barker

D. H.

Sharkey

K. M.

Van Reen

Roane

B. M.

Carskadon

M. A.

(2017). An approach to understanding sleep and depressed mood in adolescents: Person-centered sleep classification. Journal of Sleep Research, 26, 1-14.

47.

Short

M. A.

Gradisar

Lack

L. C.

Wright

H. R.

Dohnt

(2013). The sleep patterns and well-being of Australian adolescents. Journal of Adolescence, 36, 103-110.

48.

Spitzer

R. L.

Kroenke

Williams

J. B.

, & Patient Health Questionnaire Primary Care Study Group. (1999). Validation and utility of a self-report version of PRIME-MD: The PHQ Primary Care Study. Journal of the American Medical Association, 282, 1737-1744.

49.

Sun

Dunne

M. P.

Hou

X. Y.

(2012). Academic stress among adolescents in China. Australasian Epidemiologist, 19, 9-12.

50.

Talbot

L. S.

McGlinchey

E. L.

Kaplan

K. A.

Dahl

R. E.

Harvey

A. G.

(2010). Sleep deprivation in adolescents and adults: Changes in affect. Emotion, 10, 831-841.

51.

Topp

C. W.

Østergaard

S. D.

Søndergaard

Bech

(2015). The WHO-5 Well-Being Index: A systematic review of the literature. Psychotherapy and Psychosomatics, 84, 167-176.

52.

Tsai

P. S.

Wang

S. Y.

Wang

M. Y.

C. T.

Yang

T. T.

Huang

C. J.

Fang

S. C.

(2005). Psychometric evaluation of the Chinese version of the Pittsburgh Sleep Quality Index (CPSQI) in primary insomnia and control subjects. Quality of Life Research, 14, 1943-1952.

53.

Van Zundert

R. M.

Van Roekel

Engels

R. C.

Scholte

R. H

. (2015). Reciprocal associations between adolescents’ night-time sleep and daytime affect and the role of gender and depressive symptoms. Journal of Youth and Adolescence, 44, 556-569.

54.

Walker

M. P.

Harvey

A. G.

(2010). Obligate symbiosis: Sleep and affect. Sleep Medicine Reviews, 14, 215-217.

55.

Warren

Riggs

Pentz

M. A.

(2016). Executive function mediates prospective relationships between sleep duration and sedentary behavior in children. Preventive Medicine, 91, 82-88.

56.

Warren

C. M.

Riggs

N. R.

Pentz

M. A.

(2017). Longitudinal relationships of sleep and inhibitory control deficits to early adolescent cigarette and alcohol use. Journal of Adolescence, 57, 31-41.

57.

Wong

M. L.

Lau

E. Y. Y.

Wan

J. H. Y.

Cheung

S. F.

Hui

C. H.

Mok

D. S. Y.

(2013). The interplay between sleep and mood in predicting academic functioning, physical health and psychological health: A longitudinal study. Journal of Psychosomatic Research, 74, 271-277.

58.

Wong

M. M.

Brower

K. J.

Zucker

R. A.

(2009). Childhood sleep problems, early onset of substance use and behavioral problems in adolescence. Sleep Medicine, 10, 787-796.

59.

Zhao

Kong

Wang

(2012). Self-esteem and humor style as mediators of the effects of shyness on loneliness among Chinese college students. Personality and Individual Differences, 52, 686-690.