Skip to main content
NCBI home page
NIHPA Author Manuscripts logoLink to NIHPA Author Manuscripts
. Author manuscript; available in PMC: 2020 Aug 1.
Published in final edited form as: Health Psychol. 2019 Jun 3;38(8):716–726. doi: 10.1037/hea0000753

Associations of Longitudinal Sleep Trajectories with Risky Sexual Behavior During Late Adolescence

Wendy M Troxel 1, Anthony Rodriguez 2, Rachana Seelam 2, Joan S Tucker 2, Regina A Shih 3, Elizabeth J D’Amico 2
PMCID: PMC6675653  NIHMSID: NIHMS1035614  PMID: 31157533

Abstract

Objective:

The current study examines longitudinal sleep patterns in relation to risky sexual behaviors in a racially/ethnically diverse sample of adolescents.

Methods:

The sample is comprised of 1850 youth (mean age at first wave = 16.21; 57% female). Sleep duration, sleep variability (difference between weekend and weekday sleep duration), and sleep quality were collected over 4 annual assessments from 2013 to 2017. Risky sexual behaviors (i.e., sex without condom use or sex after using drugs or alcohol) were examined at the fourth follow-up assessment when youth were 19 years old. Longitudinal latent class analysis characterized patterns of individual sleep dimensions over time as well as the combination of sleep dimensions, and examined how emergent sleep classes associated with subsequent risky sexual behavior, after adjustment for sociodemographics and mental health.

Results:

After covariate adjustment, persistent “short” weekend sleepers were 2.2 times more likely to engage in risky sexual behaviors, compared to youth with sufficient weekend sleep duration. Contrary to expectations, adolescents with more consistent weekend/weekday sleep were 1.6 to 2 times more likely to engage in risky sexual behaviors, compared to those with greater variability; however, lack of variability may be an indicator of chronic insufficient sleep, both weekdays and weekends. There were no significant differences in risky sexual behavior according to classes of weekday sleep duration or quality. In the combined class model, those with persistently short and poor quality sleep were at marginally greater risk for engaging in risky sexual behaviors.

Conclusion:

Insufficient sleep in adolescents may increase risk for sexual risk-taking, and may set the stage for accelerated health risk trajectories into adulthood.

Keywords: Sleep, adolescence, sexual activity, risky behavior, latent class analysis

Adolescence is characterized by dramatic changes in socioemotional, cognitive, physical, and neurobiological development. Biological changes in sleep timing (circadian phase delay), as well as the conflict between societal obligations (e.g., early school start times) and teen social pressures, put adolescents at high risk for sleep and circadian disruptions, including insufficient sleep duration, insomnia, and circadian misalignment (mismatch between sleep schedule and internal circadian timing) (Carskadon, Wolfson, Acebo, Tzischinsky, & Seifer, 1998; Dahl, 1996). An emerging body of cross-sectional and longitudinal research demonstrates that sleep problems increase risk for alcohol and other drug (AOD) use among adolescents (Wong, Brower, Fitzgerald, & Zucker, 2004; Wong, Roberson, & Dyson, 2015), although evidence for bidirectional associations also exist (Fakier & Wild, 2011; Giannotti, Cortesi, Sebastiani, & Ottaviano, 2002; Gromov & Gromov, 2009; Johnson & Breslau, 2001). Poor decision-making, as well as disruptions in impulse control, reward sensitivity, and emotional regulation, are purported mechanisms underlying a causal role of sleep disturbance in the development of risk-taking behaviors (Forbes & Dahl, 2005; Hasler & Clark, 2013), such as AOD use as well as other health risk behaviors, including risky driving, aggressive behaviors, and smoking (McKnight-Eily et al., 2011; O’Brien & Mindell, 2005). For example, sleep-deprived adolescents and adolescents who report high levels of daytime sleepiness show impairments in higher-level executive functioning, as well as reductions in emotion regulation, impulse control, and effortful control of attention (Anderson, Storfer-Isser, Taylor, Rosen, & Redline, 2009; Beebe, DiFrancesco, Tlustos, McNally, & Holland, 2009). Furthermore, studies have shown that greater variability in weekday and weekend sleep durations (i.e., “weekend oversleep”) is associated with changes in reward-related brain function among adolescents as well as increased risk for AOD use (Hasler, Bootzin, Cousins, Fridel, & Wenk, 2008; Hasler et al., 2012).

Sleep disturbances may also contribute to risky sexual behaviors in adolescents, such as sex without a condom or sex with multiple partners, perhaps through similar mechanisms. However, research to date has been mostly cross-sectional and results have been equivocal. For instance, in a predominantly (78%) Caucasian sample of 388 high school students, O’Brien and Mindell (2005) found cross-sectional associations of self-reported sleep disturbances and greater delay in weekend bedtimes (relative to weekday bedtimes) with sexual risk-taking behaviors (e.g., having sex with multiple partners, sex without condom use). Other analyses from the Youth Risk Behavioral Surveillance System (YRBSS) have shown that insufficient sleep duration (i.e., less than 8 hours) is associated with having had sex with one or more partners in the past 3 months (McKnight-Eily et al., 2011; Meldrum & Restivo, 2014). In contrast, in a rural sample of 322 teens from North Dakota, Reichenberger and colleagues (Reichenberger, Hilmert, Irish, Secor-Turner, & Randall, 2016) did not find significant cross-sectional associations of sleep duration with indicators of risky sex, such as having sex multiple partners, getting someone pregnant, not using condoms during sex, or using drugs while having sex. Although the authors did not report the racial/ethnic composition of the sample, according to Census data, the state of North Dakota is predominantly (88%) white/Caucasian (“https://www.census.gov/quickfacts/ND,” 2018).

Data in this area are further limited by the limited assessment of sleep, which has primarily focused on sleep duration. However, it is important to consider other types of sleep issues, including poor sleep quality and variability in sleep duration, in relation to risky sexual behavior, as these sleep issues are also highly prevalent in adolescents and are associated with health risk behaviors (Catrett & Gaultney, 2009; Shibley, Malcolm, & Veatch, 2008). Finally, most of the existing studies of sleep and risky sexual behavior have focused on univariate associations, or have included a limited number of covariates, generally gender or age. Given that sleep and risky sexual behaviors are known to covary with other psychosocial characteristics, including mental health symptoms and sociodemographics, the question remains whether sleep issues uniquely contribute to adolescent sexual risk-taking, above and beyond these known risk factors (Shochat, Cohen-Zion, & Tzischinsky, 2014). In a notable exception, however, Wong and colleagues (Wong et al., 2015) found that short sleep duration and greater sleep difficulties predicted greater risk of “having regretted sex due to drinking” at a subsequent wave, even after controlling for alcohol-related problems and sociodemographic characteristics.

In summary, despite compelling evidence that a variety of sleep problems including, but not limited to, sleep duration, are associated with a range of health risk behaviors in adolescence, there has been little work addressing the association between sleep problems and risky sexual behavior. Furthermore, despite the dynamic nature of sleep, and the fact that transient sleep problems are fairly normative in adolescents (Short, Gradisar, Gill, & Camfferman, 2013), the extant literature mostly captures sleep at a single snap-shot in time, and utilizes constrained measures of sleep. Finally, most of the literature has focused on predominantly Caucasian samples, and has failed to statistically control for important confounders, including mental health symptoms. Importantly, sexual risk-taking in adolescence poses serious health risks that can have enduring effects on adolescents’ health. For instance, sexually active adolescents have higher rates of sexually transmitted disease (STD) acquisition than any other age group (Weinstock, Berman, & Cates, 2004), and having an STD increases the susceptibility of contracting human immunodeficiency virus (HIV) (Newbern et al., 2013).

The present study addresses some of these limitations by examining the degree to which longitudinal trajectories of multiple types of common sleep disturbances in adolescents are associated with risky sexual behaviors, including sexual activity without a condom or sexual activity in combination with drugs or alcohol in a large, diverse sample of adolescents (ages 16–19). Recognizing the multidimensional nature of sleep and that acute or temporary sleep disruptions are fairly normative in adolescence (Short et al., 2013), the current study includes longitudinal assessments of key sleep domains that have been identified as particularly important for adolescent health and functioning (National Sleep Foundation, 2000): sleep duration, variability in sleep duration (weekdays vs. weekends), and trouble sleeping, and examines sleep profiles over time, during adolescence and late adolescence. To assess longitudinal sleep patterns, we conducted a longitudinal latent class analysis (LLCA; latent class analysis applied to longitudinal data). LLCA is a type of mixture model that models patterns of states across time rather than scaled change (i.e., change across time is along a single continuous scale) as is common in growth mixture models. For instance, in a simple linear growth mixture model, one might have three classes (groups) of individuals where each class is characterized by a distinct intercept and slope (trajectory). However, in LLCA, intercepts and slopes are not estimated; instead individuals are grouped into classes by similar patterns across time, which do not assume any particular growth function (e.g., linear). Therefore, in LLCA, assignment to classes is based on similarity of response patterns over time. These models are useful for determining longitudinal profiles of behaviors, such as sleep, particularly when it is not known a priori how the behavior is expected to change over a given developmental period. With regard to sleep during adolescence and late adolescence, whereas there is evidence that sleep problems, including higher rates of insufficient sleep duration increase across the high school years (Eaton et al., 2010), there is also evidence that for those who transition to college, there may be improvements in sleep relative to high school (Doane, Gress-Smith, & Breitenstein, 2015), which may be due, at least in part, to the removal of a known exogenous contributor to insufficient sleep among high school teens: early school start times (Hale & Troxel, 2018; Owens, 2014). Therefore, LLCA is a particularly useful approach as it can identify patterns of sleep across these developmental periods, and it does not assume a particular form of change over time (e.g., linear), but rather allows sleep patterns or profiles to emerge based on similarity of responses across respondents over time (Feldman, Masyn, & Conger, 2009). We hypothesized that adolescents who showed persistent (over time) sleep problems including shorter sleep duration, greater variability between weekend versus weekday sleep duration, and more trouble sleeping would be more likely to report risky sexual behaviors at the follow-up assessment, as compared to those with transient sleep problems or “good sleepers”. We further hypothesized that the combination of insufficient sleep duration (weekdays and weekends) combined with poor sleep quality would be associated with greater risk of sexual risk-taking.

Given that sleep disturbances are known to covary with a number of factors that are themselves risk factors for sexual risk-taking behavior, notably mental health symptoms (Gregory et al., 2005; Gromov & Gromov, 2009), the current analyses also examined the independent association between sleep trajectories and risky sexual behavior, after statistically controlling for these factors and sociodemographic characteristics.

Method

Participants

The current study utilized four waves of data from wave 6 (2013) to wave 9 (2017) from a large multi-wave study of adolescents and young adults in southern California. Participants were initially recruited in 6th and 7th grades (ages 11 to 13) for an alcohol and other drug prevention program, CHOICE, and were representative of the 16 middle schools in southern California from which they were recruited (D’Amico et al., 2012). The study has a Certificate of Confidentiality and all procedures were approved by the RAND Human Subjects Protection Committee.

Adolescents completed waves 1 through 5 in middle school; follow up rates ranged from 74% to 90%, excluding new youth that could have come in at a subsequent wave. Adolescents transitioned from 16 middle schools to over 200 high schools and were re-contacted and re-consented to complete annual web-based surveys. At wave 6, 61% of teens participated in the follow up survey. From wave 6 to 7, 80% of the sample was retained, 91% of the sample from wave 7 to wave 8, and 89% of the sample from wave 8 to wave 9. If a participant did not complete a wave of data collection, they were still eligible to complete all subsequent waves. That is, we fielded the full sample at every wave so that all participants had an opportunity to participate in each survey. In this study, youth who did not complete at least one sex-related outcome measure at wave 9 or who did not have sleep data for at least three timepoints were excluded (N= 689). The current analytic sample included 1,850 youth (mean age 16.23 at Wave 1 and 19.33 at Wave 4). Compared to excluded youth, the analytic sample had slightly more females (57% versus 47%), more mothers with a college degree (52% versus 46%), and fewer youth who reported mixed race/ethnicity (11% versus 14%). No other significant differences on demographics were found.

Measures

Sleep measures.

Self-reported sleep items were added to the ongoing survey in 2013 (wave 6) and continued through 2017 (wave 9). The primary sleep measures for the current study were: calculated sleep duration for weekdays and weekends separately (derived from reported bedtimes and waketimes), sleep variability (calculated as the difference between weekend and weekday sleep duration); and trouble sleeping (as an indicator of sleep quality). Similar measures have been used in several prior studies with adolescents (Larson, Laska, Story, & Neumark-Sztainer, 2015; Pasch, Laska, Lytle, & Moe, 2010), and have been shown to significantly correlate with diary and actigraphy assessments of sleep (Wolfson et al., 2003). Specifically, for weekday or weekend sleep duration, respectively, youth were asked “What time do you usually go to bed on school days (or weekend days)?” and “What time do you usually wake-up on school days (or weekends)?”. The difference between bedtime and waketime was computed and used in analyses for weekday or weekend sleep duration. Notably, because this calculated measure does not account for wakefulness during the sleep period, it is actually a measure of “time in bed”. However, to be consistent with the literature, which has used similar methods and used the term “sleep duration”, we refer to this variable as “sleep duration” (Kurina et al., 2013; Miller et al., 2015). Weekday and weekday sleep durations were calculated separately given substantial prior research showing considerable variability between weekday and weekend sleep durations for adolescents, largely due to the constraints of early school start times during the week (Owens, 2014). Sleep variability was calculated as the difference between weekday and weekend sleep duration (23). Trouble sleeping was assessed using an item taken from the 15-item Patient Health Questionnaire (PHQ) Somatic Symptom Severity Scale measure (Kroenke, Spitzer, & Williams, 2002). Youth were asked “During the past 4 weeks, how much have you been bothered by trouble sleeping”. Responses included 1= Not bothered, 2= Bothered a little, and 3=Bothered a lot. These procedures were applied at each assessment wave 6 through 9.

Risky sexual behavior

(Centers for Disease Control and Prevention, 2014) was assessed by combining information obtained from youth who reported on whether they used alcohol/marijuana/other drugs right before (within an hour or so) or during sex (0=No, 1=Yes) and whether they used a condom during sex (0=No, 1=Yes) (Centers for Disease Control and Prevention, 2014). Responses were then combined such that if a youth reported either substance use before/during sex or not using a condom, they were scored as engaging in risky sexual behavior (1=Yes) whereas not using substances before/during sex AND using a condom would be scored as not engaging in risky sexual behavior (0=No).

Covariates1

We adjusted for covariates (assessed at wave 6) known to be associated with sleep and/or risky sexual behavior, such as age, race/ethnicity, gender, maternal education (as a proxy for socioeconomic status), and mental health. Participants were classified into one of five racial/ethnic groups: non-Hispanic White, non-Hispanic Black, Hispanic, Asian, and any other ethnicities. Mental health was assessed using the Mental Health Inventory (MHI-5) (Berwick et al., 1991), which comprises five items focused on past month anxiety and depression symptoms (e.g., felt downhearted and blue, been a very nervous person). Scores on the MHI-5 were scaled such that they ranged from 0 to 100, with higher scores indicating better mental health (α =.77). Finally, we included an indicator for whether the student had attended one of the original intervention schools in 2008.

Analytic Methods

To estimate LLCA models, we used the manual three-step approach (Nylund-Gibson, Grimm, Quirk, & Furlong, 2014) using Mplus version 8.0 (Muthén & Muthén, 2012–2018) to extract classes. First, models were developed separately for each individual sleep measure, resulting in four final best class solutions. These class solutions are based on individual sleep dimensions and characterize distinct developmental patterns associated with each sleep characteristic across the data collection waves, which span adolescence through late adolescence. To determine the best fitting solution for each individual sleep measure across time, we estimated separate models that fit one to five latent class solutions. Second, we used LLCA to model the developmental patterns of several sleep measures simultaneously, consisting of3 out of 4 of the individual sleep dimensions: both weekday and weekend sleep duration, and sleep quality. In the multi-dimensional sleep model, we did not include sleep variability, as it is a variable constructed from weekday and weekend sleep duration, and therefore conflated with those variables. Consistent with prior literature using LLCA (Laska, Pasch, Lust, Story, & Ehlinger, 2009), we determined the best final class structure for the individual sleep dimensions and the multi-dimensional sleep measure, by evaluating the combination of several fit indices including: negative two log likelihood (−2LL), Akaike Information Criteria (AIC), Bayesian Information Criteria (BIC), the sample size adjusted Bayesian Information Criteria (aBIC), the Vuong-Lo-Mendell-Rubin adjusted likelihood ratio test (VLMRT), and the bootstrapped likelihood ratio test (BLRT). The −2LL, AIC, BIC, and aBIC are all log likelihood measures for which lower values indicate better fit (Nylund, Asparouhov, & Muthén, 2007). Likelihood ratio testing was determined by the VMLRT and BLRT tests which compare the improvement between sequential class models (i.e., compare 2 class versus 3 class models) and provides a significance test to evaluate the improvement in model fit by including an additional class (Nylund et al., 2007). In addition, theoretical consideration was given to the meaning of the emergent classes as well as the final resulting sample sizes of each class. Thus, decisions on final best fitting solutions were based on both statistical and theoretical rationale(Lanza, Patrick, & Maggs, 2010).

Next, we sought to understand the association between the emergent classes of longitudinal sleep patterns for each of the sleep dimensions individually (i.e., weekday sleep duration, weekend sleep duration, sleep variability, and trouble sleeping) as well as the multi-dimensional sleep measure (i.e., consisting of sleep duration and quality) and the outcome: risky sexual behavior assessed at Wave 9. We assessed differences in risky sexual behavior by implementing the manual three-step auxiliary BCH approach which tests for mean differences in the outcome between classes using a Wald chi-square test (Asparouhov & Muthen, 2014). This method fixes the parameters of latent classes to ensure that the measurement of latent classes is not influenced by the inclusion of covariates.

Missing data was handled by Mplus such that it adjusts for missing data using a maximum likelihood estimator under the assumption that data are missing at random and uses all data that are available for each participant.

Results

Sample Description

On average, participants were 16.21 (SD = 0.72) years old at wave 6 (Table 1). The sample was 57% female and 44% Hispanic. About 28% of participants had mothers with a high school education or less. Average amount of self-reported sleep across waves ranged from 7.40–7.78 hours for weekdays and 8.73–9.28 hours for weekends, with sleep quality rated between 1.57–1.61 (with 1=not bothered at all and 2 = bothered a little). At wave 9, 37% of youth reported engaging in risky sexual behavior.

Table 1.

Sample Descriptive Statistics

Variable M(SD) or %>(n) N = 1,850
Demographics
 Age 16.21 (0.72)
 Female 57% (n=1,054)
 White 21.6% (n=399)
 Hispanic 44.0% (n=814)
 Asian 21.1% (n=390)
 Black 2.5% (n=46)
 Other-race 10.9% (n=201)
 Mother’s Education
  Did not finish high school 12.7% (n=235)
  High school 15.4% (n=284)
  Some college 12.7% (n=234)
  College 52.4% (n=970)
  Don’t know 6.9% (n=127)
Mental Health (MHI5) 66.28 (19.55)
Weekday Sleep Duration (hours)
 Wave 6 7.62 (1.08)
 Wave 7 7.40 (1.19)
 Wave 8 7.71 (1.39)
 Wave 9 7.78 (1.43)
Weekend Sleep Duration (hours)
 Wave 6 9.28 (1.45)
 Wave 7 9.12 (1.43)
 Wave 8 8.89 (1.40)
 Wave 9 8.73 (1.42)
Sleep Variability (hours)
 Wave 6 1.65 (1.60)
 Wave 7 1.70 (1.62)
 Wave 8 1.16 (1.60)
 Wave 9 0.94 (1.72)
Trouble Sleeping
 Wave 6 1.61 (0.74)
 Wave 7 1.60 (0.73)
 Wave 8 1.58 (0.71)
 Wave 9 1.57 (0.70)
Any Risky Sexual Behavior 36.5% (n=638)
Open in a new tab

Note: Demographics, mental health and sleep measures are taken from Wave 6 in 2013. Risky sexual behavior is measured at Wave 9 in 2017.

Based on results from the likelihood ratio test (e.g., VLMRT) from the LLCAs, in combination with theoretical consideration, a three-class solution best represented longitudinal patterns of sleep for weekday sleep duration, sleep variability, and sleep quality. Model fit indices and likelihood ratio tests are presented in Table 2. For weekend sleep duration, a four-class structure was also significantly identified; however, it was likely to be a smaller subset of another class already identified, and as such resulted in a very small proportion of individuals (< 5%), thus we kept our 3-class solution across all individual sleep measures. For the multi-dimensional sleep models, a 4-class solution best fit the data. Figure 1 shows the pattern of latent classes for each individual sleep dimension.

Table 2.

Model Fit Statistics and Entropy for LLCAs

Model (−2LL) Free Parameters AIC BIC aBIC VLMRT BSLRT Entropy
Weekday Sleep Duration
1 Class 23693.24 8 23709.24 23753.43
2 Classes 22850.48 13 22876.48 22948.28 22906.98 842.76* 842.76* 0.64
3 Classes 22658.89 18 22694.89 22794.30 22737.12 191.59* 191.59* 0.62
4 Classes 22594.36 23 22640.36 22767.39 22694.32 64.53 64.53 0.66
5 Classes 22509.03 28 22565.03 22719.67 22630.71 85.33 85.33 0.67
Weekend Sleep Duration
1 Class 25349.35 8 25365.35 25409.53 25384.11
2 Classes 24505.78 13 24531.78 24603.58 24562.28 843.57* 843.57* 0.59
3 Classes 24276.23 18 24312.23 24411.64 24354.46 229.55* 229.55* 0.68
4 Classes 24196.88 23 24242.87 24369.91 24296.84 79.35* 79.35* 0.66
5 Classes 24151.02 28 24207.02 24361.66 24272.70 45.86 45.86 0.71
Sleep Variability
1 Class 27037.12 8 27053.12 27097.30 27071.89
2 Classes 26437.34 13 26463.34 26535.14 26493.84 599.78* 599.78* 0.59
3 Classes 26326.78 18 26362.78 26462.20 26405.01 110.56* 110.56* 0.61
4 Classes 26236.31 23 26282.31 26409.34 26336.27 90.47 90.47 0.66
5 Classes 26181.62 28 26237.62 26392.26 26303.31 54.69 54.69 0.66
Trouble Sleeping
1 Class 15771.35 8 15787.35 15831.53 15806.11
2 Classes 14261.81 13 14287.81 14359.61 14318.30 1509.54* 1509.54* 0.82
3 Classes 14027.91 18 14063.91 14163.32 14106.14 233.90* 233.90* 0.79
4 Classes# 14027.91 23 14073.91 14200.94 14127.87 - - 0.83
5 Classes# 14027.80 28 14083.91 14238.55 14149.60 - - 0.85
Multi-Dimensional Sleep Measure
1 Class 64813.93 24 64861.94 64994.49 64918.24
2 Classes 63035.63 37 63109.63 63313.98 63196.43 1778.30* 1778.30* .731
3 Classes 62081.46 50 62181.46 62457.61 622298.76 954.17* 954.17* .755
4 Classes 61662.56 63 61788.56 62136.51 61936.36 418.89* 418.89* .754
5 Classes 61449.62 76 61601.62 62021.37 61779.92 212.94 212.94 .760
Open in a new tab

Note. AIC = Akaike Information Criterion; BIC = Bayesian Information Criterion; aBIC = Sample size-adjusted BIC; BLRT = Parametric Bootstrapped Likelihood Ratio Test; VLMRT = Vuong-Lo-Mendell-Rubin Likelihood Ratio Test.

(*)

denotes the current class structure (k classes) was a significant improvement over the previous class structure (k-1 classes).

(#)

denote models with non-convergence perturbations.

Figure 1–

Figure 1–

Open in a new tab

Longitudinal latent classes for individual sleep dimensions

Weekday Sleep Duration

Patterns of weekday sleep duration over 4 years revealed three clear latent classes. Group 1 is called short weekday sleepers (n=250, 14%), and includes adolescents who slept an average of 6.35 hours per weekday with a range between 5.8 to 6.8 hours across waves. This is the smallest group, and includes adolescents who had the poorest (i.e., shortest) weekday sleep duration pattern over assessment waves. Group 2 is called intermediate weekday sleepers (n= 1,095, 60%), and includes adolescents who slept an average of 7.53 hours per weekday with a range of 7.3 to 7.7 hours across waves. This is the most common cluster; however, this cluster notably still falls short of the 8–10 hours of sleep recommended for this age group. Group 3 is called sufficient weekday sleepers (n= 484, 26%), and includes adolescents who consistently slept an average of 8.55 hours per weekday with a range of 8.5 to 8.6 across waves. After accounting for covariates, there were no significant differences between the three weekday sleep duration classes on risky sexual behavior (Table 3).

Table 3.

Risky Sexual Behavior Estimates and Standard Errors by Sleep Measure

Weekday Sleep Duration Risky Sexual Behavior
Estimate SE χ2 p OR (CI)
Short sleepers 0.33 0.04
Intermediate sleepers 0.38 0.02
Sufficient sleepers 0.35 0.03
Short vs Intermediate 0.92 .337 0.94 (0.68, 1.29)
Short vs Sufficient 0.10 .757 1.14 (0.91, 1.42)
Intermediate vs Sufficient 0.57 .451 0.82 (0.62, 1.10)
Weekend Sleep Duration
Short sleepers 0.38 0.03
Intermediate sleepers 0.38 0.02
Long sleepers 0.21 0.04
 Short vs Intermediate 0.02 .901 1.02 (0.82, 1.28)
 Short vs Long* 10.76 .001 2.24 (1.48, 3.39)
 Intermediate vs Long* 11.99 .001 2.28 (1.55, 3.36)
Sleep Variability
Low variability 0.43 0.02
Moderate variability 0.32 0.02
High variability 0.27 0.06
 Low vs Moderate* 8.53 .004 1.61 (1.32, 1.96)
 Low vs High* 5.24 .022 1.97 (1.19, 3.26)
 Moderate vs High 0.34 .560 1.22 (0.74, 2.03)
Trouble Sleeping
Good sleepers 0.36 0.02
Moderate sleepers 0.39 0.04
Poor sleepers 0.37 0.03
 Good vs Moderate 0.61 .435 0.86 (0.66, 1.12)
 Good vs Poor 0.19 .667 0.93 (0.72, 1.21)
 Poor vs Moderate 0.11 .740 0.92 (0.66, 1.28)
Multi-Dimensional Sleep Measure
Sufficient, poor sleepers (1) 0.33 0.04
Short, poor sleepers (2) 0.40 0.04
Short, good sleepers (3) 0.40 0.03
Sufficient, good sleepers (4) 0.34 0.02
 1 v 2 1.70 .192 0.72 (0.49, 1.04)
 1 v 3 2.97 .085 0.71 (0.52, 0.97)
 1 v 4 0.14 .713 0.93 (0.69, 1.25)
 2 v 3 0.00 .970 0.99 (0.72, 1.37)
 2 v 4 1.85 .174 1.29 (0.96, 1.77)
 3 v 4 3.03 .082 1.31 (1.05, 1.64)
Open in a new tab

Note. Covariates in the models included: age, race/ ethnicity, gender, maternal education (as a proxy for socioeconomic status), mental health, and an indicator for whether the participant was in one of the original intervention schools.

Weekend Sleep Duration

Three distinct classes were identified for weekend sleep duration over the 4 assessment waves. Group 1 is called short weekend sleepers (n= 436, 24%), and includes adolescents who consistently slept, on average, 7.8 hours on weekend nights with a range of 7.6 to 8.0 hours across waves. Group 2 is called intermediate weekend sleepers (n= 1,213, 67%), and includes adolescents who slept on average approximately 9.2 hours per night on the weekend with a range of 8.9 to 9.5 hours across waves. Group 3 is called long weekend sleepers (n = 169, 10%), and includes adolescents who reported the greatest amount of weekend sleep across the assessments with an average sleep time of 10.64 hours on the weekend with a range of 9.9 to 11.1 hours across waves. As shown in Table 3, after controlling for covariates, significant differences were found such that short weekend sleepers were 2.24 times more likely to engage in risky sexual behavior than long weekend sleepers (χ2 = 10.756, p=0.001). Similarly, intermediate weekend sleepers were 2.28 times more likely to engage in risky sexual behavior than long weekend sleepers (χ2 = 11.995, p=0.001).

Sleep Variability

Group classification for sleep variability (i.e., differences in weekday/weekend sleep duration) revealed three distinct classes. Group 1 is called low variability sleepers (n= 784, 43%), and includes adolescents with a relatively small difference in weekday/weekend sleep duration with an average difference of approximately 30 minutes. Group 2 is called moderate variability sleepers (n= 937, 52%), and includes adolescents with weekday/weekend sleep duration differences of 1.83 hours (~1 hour and 50 minutes). Group 3 is called high variability sleepers (n = 82, 5%), and includes adolescents with relatively large differences between weekday/weekend sleep duration with an average difference of 3.59 hours (~ 3 hours and 36 minutes). On average, sleep durations across the week for Class 1, 2, and 3 are 8.2 hours, 8.39 hours, and 8.59 hours, respectively, indicating that the low variability class had consistently shorter sleep than the other classes [F(2,1847)=18.741, p<.001]. As shown in Table 3, after controlling for covariates, significant differences were noted such that low variability sleepers were 1.6 times more likely to engage in risky sexual behavior than moderate variability sleepers (χ2= 8.526, p=0.004) and 1.96 time more likely than high variability sleepers (χ2= 5.244, p=0.022).

Trouble Sleeping

Three distinct classes of sleep quality (i.e., trouble sleeping) were identified over the 4 assessment waves. Group 1 is called good sleepers (n = 1,250, 68%), and includes adolescents who reported on average “little to no” (mean = 1.3) trouble sleeping across the assessments. Group 2 is called moderate sleepers (n = 297, 16%) and includes adolescents who on average reported “little” (mean = 1.9) trouble sleeping across the assessments. Group 3 is called poor sleepers (n = 299, 16%) and includes adolescents who on average reported “little to a lot” (mean = 2.4) of trouble sleeping across assessments. Table 3 shows that after controlling for covariates, there were no significant differences between the three sleep quality classes on risky sexual behavior.

Multi-Dimensional Sleep Patterns

Four distinct longitudinal classes emerged for the multi-dimensional sleep pattern. Class 1 is called “sufficient, poor sleepers” (n=260, 14%), and includes adolescents who slept between 7.6 to 8.18 hours across waves (mean = 7.95 hours) on weekdays and 9.5 hours, on average, on weekends and who reported “little to a lot” (mean = 2.2) trouble sleeping. In contrast, Class 2 is called “short, poor sleepers” (n=216, 12%), who were characterized by relatively short longitudinal patterns of both weekday and weekend sleep durations (means = 6.7 and 8.2 hours, respectively), but similar levels of trouble sleeping (mean =2.4). Classes 3 and 4 were differentiated from Classes 1 and 2, by an overall lack of trouble sleeping, despite differing patterns of sleep duration. Specifically, Class 3, called “short, good quality sleepers” (n=539, 29%), was characterized by relatively short weekday (mean = 7.1 hours) and weekend sleep duration (mean =8.2 hours) and minimal trouble sleeping (mean = 1.4). Class 4, called “sufficient, good sleepers” (n=833, 45%), includes adolescents who slept, on average 8.1 hours during weekdays, and 9.6 hours on average, on weekends, and who reported minimal trouble sleeping (mean = 1.3). As shown in Table 3, after controlling for covariates, statistical comparisons for class differences in risky sexual behavior revealed that Class 3 was marginally (p=0.08) more likely to engage in risky sexual behavior than Class 1 (“sufficient, poor sleepers”) and Class 4 (“sufficient, good sleepers”). Given that Classes 2 and 3 both represented “short sleepers” and Classes 1 and 4 “sufficient sleepers” we conducted exploratory analyses of the differences between the two groups of classes. Results indicated that “short sleepers” were 1.25 times more likely to engage in risky sexual behavior than “sufficient sleepers”, (χ2= 5.12, p=0.024).

Discussion

The current study is only the second to examine longitudinal patterns of specific adolescent sleep issues as well as a multi-dimensional sleep measure, consisting of duration and quality, in relation to risky sexual behavior in a racially/ethnically diverse sample of adolescents. Class analyses revealed three distinct longitudinal patterns for each of the individual sleep measures and four distinct patterns for the multi-dimensional sleep measure. Consistent with prior reports suggesting that insufficient sleep is highly prevalent among adolescents (Matthews, Hall, & Dahl, 2014; Owens, 2014), even the “intermediate” weekday sleep duration group did not meet current recommendations for optimal sleep duration for adolescents. Importantly, the terminology for naming the different classes is somewhat arbitrary and intended primarily to be descriptive in nature.

We hypothesized that greater weekend/weekday sleep differences would be associated with increased risk of engaging in risky sexual behaviors, based on prior cross-sectional work indicating that the tendency for “weekend oversleep” among adolescents may be a marker of circadian misalignment (Owens, 2014). Contrary to our hypothesis, we found that smaller weekday/weekend differences (i.e., lesser variability) in sleep duration were associated with greater likelihood of engaging in risky sexual behavior. This may be due to the low variability group having a shorter sleep duration across the week than the groups with greater variability.

We also found that the persistently short weekend sleep duration group was more likely to engage in risky sexual behavior. This suggests that adolescents who fail to “catch up” on sleep on the weekends may be chronically sleep-deprived, and therefore may be at-risk for engaging in risky health behaviors. This interpretation is plausible given that two out of three classes of weekday sleep duration fell below recommended guidelines for sufficient adolescent sleep, and only 27% of youth were in the persistently sufficient sleep class (i.e., 8.5 hours). Moreover, our multi-dimensional sleep models lend further support to this interpretation, as two classes emerged from these models that could be characterized by short weekend and short weekday sleep, with or without trouble sleeping; these “short weekend and weekday sleepers” were disproportionately likely to be characterized as “low variability” sleepers (50%), as compared to the two other classes with relatively sufficient sleep, at least on the weekends (39% characterized as low variability sleepers). Furthermore, the pattern of results suggested that those with short weekend and weekday sleep duration (Class 3) were at marginally greater risk for risky sexual behavior, as compared to the sufficient sleep classes. Therefore, there may be competing pathways through which chronic insufficient sleep versus weekend “catch-up” sleep are associated with increased risk for poor health behaviors among adolescents.

In contrast to prior research showing that self-reported sleep difficulties are associated with “having regretted sex due to drinking”(Wong et al., 2015), we did not find significant associations between the “trouble sleeping” classes or the multi-dimensional class which included short weekend and weekday sleep duration and trouble sleeping (Class 2) and the sexual risk behaviors examined in this study. The reasons for this discrepancy are unclear, but is likely due to differences across studies in the measurement of self-reported sleep quality, differences in the type of sexual risk behavior examined (e.g., regretting sex versus reporting sex without a condom), or other factors. Given limited work in this area, it is important for future studies to better understand the conditions under which insomnia-related symptoms may impact sexual risk behavior.

Based on the emergent classes from this longitudinal data in adolescence, lack of variability is likely an indicator of the risk associated with chronic sleep insufficiency, rather than a marker of the relative stability or instability of sleep-wake patterns. Indeed, from an intervention perspective, this poses a significant challenge, as the standard sleep hygiene recommendation is to maintain consistent bedtimes and wake-up times throughout the week. However, for the majority of U.S. adolescents, whose weekday sleep opportunity is constrained due to early school start times, maintaining consistency in sleep-wake schedules throughout the week, may not only be unrealistic, but also may be unhealthy—if it perpetuatesa pattern of chronic sleep insufficiency. On the other hand, excessive weekend “oversleep” can exacerbate teens’ biologically driven phase delay, making it even harder to fall asleep at a reasonable hour during the school week. Therefore, finding a “middle ground” which allows for some weekend catch-up sleep, while maintaining some level of consistency in sleep-wake patterns may be warranted, although more research is needed to determine the most beneficial (and attainable) balance for teenagers (Crowley & Carskadon, 2010).

Importantly, all analyses statistically adjusted for factors known to be associated with both sleep problems and risky sexual behavior, including baseline mental health symptoms. Therefore, findings suggest that persistent sleep problems may uniquely contribute to adolescent risky sexual behavior. Consequently, improving adolescent sleep health may be an important target of prevention and intervention. Alternatively, it may be that weekend insufficient sleep is a marker, rather than a predictor of risky sexual behaviors, as perhaps adolescents who are staying out later on the weekends, presumably without parental supervision, are more likely to engage in a variety of risky behaviors, including sexual risk-taking. Other unmeasured factors may also play a role, such as greater time spent with peers, greater susceptibility to peer influences and peer pressure, and less parental involvement.

Nevertheless, there are several plausible pathways through which sleep disturbances, including insufficient sleep or poor sleep quality, could influence sexual behavior outcomes among adolescents. Consistent with the literature on sleep disturbances and other risk-taking behaviors including AOD use, sleep loss and problems may increase the likelihood for risky sexual behaviors via changes in reward-related neural circuitry, and by influencing impulsivity and decision-making processes (Hasler & Clark, 2013). For instance, in a study which utilized a task that simulated real-life decision-making, under sleep-deprived conditions, participants made dramatically riskier decisions as compared to their own performance under rested conditions (Killgore, Balkin, & Wesensten, 2006). Thus, sleep loss or problems in adolescents may increase risk for sexual risk-taking, by compromising decision-making and by altering reward-related brain function. In addition, sleep problems may also be implicated in adolescent sexual risk-taking by influencing affective and behavioral regulation (Dahl & Lewin, 2002). Although we statistically adjusted for mental health symptoms, it is possible that other unmeasured mental health symptoms or problems with emotion or behavior regulation may underlie some of the associations between sleep problems and risky sexual behaviors.

Our findings add to the small literature on longitudinal sleep patterns and health risk behaviors among adolescents by examining trajectories of sleep patterns in association with sexual risk taking. Strengths of the study include the large, racially-ethnically diverse sample of youth, the longitudinal assessment of sleep, the inclusion of multiple domains of sleep relevant to adolescent health behaviors, and inclusion of key covariates known to be related to sleep and/ or risky sexual behaviors.

It is also important to note study limitations. First, findings cannot establish causality or temporality between sleep problems and sexual behaviors, as items used to assess risky sexual behavior were only examined at wave 9. In addition, due to concerns about insufficient power and parameter and standard error bias, biological sex was treated as a covariate rather than as a multigroup indicator (Nylund et al., 2007). However, it will be important for future studies to assess larger samples and consider whether differential associations between sleep classes and sexual behaviors emerge. An additional limitation is the use of self-reported measures of sexual behavior and sleep patterns, which may have introduced bias or common method variance. Moreover, self-reported sleep duration was calculated from self-reported bedtimes and wake-up times, which may overestimate actual sleep time, given it does not account for sleep latency or middle of the night awakenings. However, similar methods are used in other epidemiologic sleep studies (Hasler et al., 2004; Miller et al., 2015) and have shown modest correlations with actigraphy-assessed sleep time spent asleep (Cespedes et al., 2016). As mentioned, a strength of our study is the broader assessment of longitudinal sleep patterns using assessments that have been validated in prior work with adolescents (Wolfson et al., 2003), particularly given that prior work has primarily on sleep duration only. However, research that includes objective measures of sleep patterns via wrist actigraphy or polysomnography would reduce recall or other biases inherent to self-report and provide a more comprehensive assessment of sleep patterns and variability. Furthermore, our measure of trouble sleeping was based on a single-item measure, rather than a validated measure of sleep quality, such as such as the Pittsburgh Sleep Quality Index (Buysse, Reynolds, Monk, Berman, & Kupfer, 1989), which may have limited its utility in predicting risky sexual behaviors.

In summary, current findings add to the growing literature suggesting a potential role of sleep disturbances in contributing to adolescent risk-taking behaviors, including, in this case, sexual risk-taking. The epidemic nature of insufficient sleep among adolescents has been well-documented. Findings further demonstrate the persistence of insufficient sleep across the adolescent years. Multi-pronged strategies that consider both individual/ family-level interventions, including increased parental monitoring/ rule-setting around bedtimes and use of technology, as well as policy-levers, such as delaying school start times, are needed to reduce the epidemic of teen sleep problems, and mitigate the impact of sleep problems on risk-taking behaviors, including sexual risk-taking.

Figure 2–

Figure 2–

Open in a new tab

Multi-Dimensional Sleep Classes

Funding Source:

Work on this article was supported by a grant from the National Institute of Alcohol Abuse and Alcoholism (R01AA016577, R01AA020883) to Elizabeth D’Amico.

Financial Disclosure: The authors have no financial relationships relevant to this article to disclose.

Footnotes

Potential Conflicts of Interest: The authors have no conflicts of interest relevant to this article to disclose.

1

We ran sensitivity models, additionally controlling for past month alcohol or other drug use (AOD use); however, analyses were virtually the same. Given the overlap with the outcome, models are presented without AOD use as a covariate.

References

  1. Anderson B, Storfer-Isser A, Taylor HG, Rosen CL, & Redline S (2009). Associations of executive function with sleepiness and sleep duration in adolescents. Pediatrics, 123(4), e701–707. doi: 10.1542/peds.2008-1182 [DOI] [PMC free article] [PubMed] [Google Scholar]
  2. Asparouhov T, & Muthen B (2014). Auxiliary Variables in Mixture Modeling: Using the BCH Method in Mplus to Estimate a Distal Outcome Model and an Arbitrary Secondary Mode. Mplus Web Notes, 21(2), 1–22. [Google Scholar]
  3. Beebe DW, DiFrancesco MW, Tlustos SJ, McNally KA, & Holland SK (2009). Preliminary fMRI findings in experimentally sleep-restricted adolescents engaged in a working memory task. Behavioral and Brain Functions : BBF, 5, 9–9. doi: 10.1186/1744-9081-5-9 [DOI] [PMC free article] [PubMed] [Google Scholar]
  4. Berwick DM, Murphy JM, Goldman PA, Ware JE Jr., Barsky AJ, & Weinstein MC (1991). Performance of a Five-Item Mental Health Screening Test. Medical Care, 29(2), 169–176. [DOI] [PubMed] [Google Scholar]
  5. Buysse DJ, Reynolds CF 3rd, Monk TH, Berman SR, & Kupfer DJ (1989). The Pittsburgh Sleep Quality Index: a new instrument for psychiatric practice and research. Psychiatry Research, 28(2), 193–213. [DOI] [PubMed] [Google Scholar]
  6. Carskadon MA, Wolfson AR, Acebo C, Tzischinsky O, & Seifer R (1998). Adolescent sleep patterns, circadian timing, and sleepiness at a transition to early school days. Sleep, 21(8), 871–881. [DOI] [PubMed] [Google Scholar]
  7. Catrett CD, & Gaultney JF (2009). Possible insomnia predicts some risky behaviors among adolescents when controlling for depressive symptoms. The Journal of Genetic Psychology, 170(4), 287–309. doi: 10.1080/00221320903218331 [DOI] [PubMed] [Google Scholar]
  8. Centers for Disease Control and Prevention. (2014). Youth Risk Behavior Surveillance System (YRBSS). Retrieved from https://www.cdc.gov/healthyyouth/data/yrbs/index.htm?s_cid=tw-zaza-1016
  9. Cespedes EM, Hu FB, Redline S, Rosner B, Alcantara C, Cai J, … Patel SR (2016). Comparison of Self-Reported Sleep Duration With Actigraphy: Results From the Hispanic Community Health Study/Study of Latinos Sueño Ancillary Study. American Journal of Epidemiology, 183(6), 561–573. doi: 10.1093/aje/kwv251 [DOI] [PMC free article] [PubMed] [Google Scholar]
  10. Crowley SJ, & Carskadon MA (2010). Modifications to weekend recovery sleep delay circadian phase in older adolescents. Chronobiology International, 27(7), 1469–1492. doi: 10.3109/07420528.2010.503293 [DOI] [PMC free article] [PubMed] [Google Scholar]
  11. D’Amico EJ, Green HDJ, Miles JNV, Zhou AJ, Tucker JA, & Shih RA (2012). Voluntary after school alcohol and drug programs: If you build it right, they will come. Journal of Research on Adolescence 22(3), 571–582. [DOI] [PMC free article] [PubMed] [Google Scholar]
  12. Dahl RE (1996). The regulation of sleep and arousal: Development and psychopathology. Development & Psychopathology, 8, 3–27. [Google Scholar]
  13. Dahl RE, & Lewin DS (2002). Pathways to adolescent health sleep regulation and behavior. Journal of Adolescent Health, 31(6 Suppl), 175–184. [DOI] [PubMed] [Google Scholar]
  14. Doane LD, Gress-Smith JL, & Breitenstein RS (2015). Multi-method assessments of sleep over the transition to college and the associations with depression and anxiety symptoms. Journal of Youth and Adolescence, 44(2), 389–404. doi: 10.1007/s10964-014-0150-7 [DOI] [PubMed] [Google Scholar]
  15. Eaton DK, McKnight-Eily LR, Lowry R, Perry GS, Presley-Cantrell L, & Croft JB (2010). Prevalence of insufficient, borderline, and optimal hours of sleep among high school students - United States, 2007. Journal of Adolescent Health, 46(4), 399–401. doi: 10.1016/j.jadohealth.2009.10.011 [DOI] [PubMed] [Google Scholar]
  16. Fakier N, & Wild LG (2011). Associations among sleep problems, learning difficulties and substance use in adolescence. Journal of Adolescence, 34(4), 717–726. doi: 10.1016/j.adolescence.2010.09.010 [DOI] [PubMed] [Google Scholar]
  17. Feldman BJ, Masyn KE, & Conger RD (2009). New approaches to studying problem behaviors: a comparison of methods for modeling longitudinal, categorical adolescent drinking data. Developmental Psychology, 45(3), 652–676. doi: 10.1037/a0014851 [DOI] [PMC free article] [PubMed] [Google Scholar]
  18. Forbes EE, & Dahl RE (2005). Neural systems of positive affect: Relevance to understanding child and adolescent depression? Developmental Pyscyopathology, 17, 827–850. [DOI] [PMC free article] [PubMed] [Google Scholar]
  19. Giannotti F, Cortesi F, Sebastiani T, & Ottaviano S (2002). Circadian preference, sleep and daytime behaviour in adolescence. Journal of Sleep Research, 11(3), 191–199. [DOI] [PubMed] [Google Scholar]
  20. Gregory AM, Caspi A, Eley TC, Moffitt TE, O’Connor TG, & Poulton R (2005). Prospective Longitudinal Associations Between Persistent Sleep Problems in Childhood and Anxiety and Depression Disorders in Adulthood. Journal of Abnormal Child Psychology, 33(2), 157–163. doi: 10.1007/s10802-005-1824-0 [DOI] [PubMed] [Google Scholar]
  21. Gromov I, & Gromov D (2009). Sleep and Substance Use and Abuse in Adolescents. Child and Adolescent Psychiatric Clinics of North America, 18(4), 929–946. doi: 10.1016/j.chc.2009.04.004 [DOI] [PubMed] [Google Scholar]
  22. Hale L, & Troxel W (2018). Embracing the School Start Later Movement: Adolescent Sleep Deprivation as a Public Health and Social Justice Problem. American Journal of Public Health, 108(5), 599–600. doi: 10.2105/AJPH.2018.304381 [DOI] [PMC free article] [PubMed] [Google Scholar]
  23. Hasler BP, Bootzin RR, Cousins JC, Fridel K, & Wenk GL (2008). Circadian phase in sleep-disturbed adolescents with a history of substance abuse: A pilot study. Behavioral Sleep Medicine, 6(1), 55–73. doi: 10.1080/15402000701796049 [DOI] [PubMed] [Google Scholar]
  24. Hasler BP, Buysse DJ, Klaghofer R, Gamma A, Ajdacic V, Eich D, … Angst J (2004). The association between short sleep duration and obesity in young adults: a 13-year prospective study. Sleep, 27(4), 661–666. [DOI] [PubMed] [Google Scholar]
  25. Hasler BP, & Clark DB (2013). Circadian misalignment, reward-related brain function, and adolescent alcohol involvement. Alcoholism: Clinical and Experimental Research, 37(4), 558–565. doi: 10.1111/acer.12003 [DOI] [PMC free article] [PubMed] [Google Scholar]
  26. Hasler BP, Dahl RE, Holm SM, Jakubcak JL, Ryan ND, Silk JS, … Forbes EE (2012). Weekend-weekday advances in sleep timing are associated with altered reward-related brain function in healthy adolescents. Biological Psychiatry, 91(3), 334–341. [DOI] [PMC free article] [PubMed] [Google Scholar]
  27. https://www.census.gov/quickfacts/ND. (2018).
  28. Johnson EO, & Breslau N (2001). Sleep problems and substance use in adolescence. Drug and Alcohol Dependence, 64(1), 1–7. [DOI] [PubMed] [Google Scholar]
  29. Killgore WD, Balkin TJ, & Wesensten NJ (2006). Impaired decision making following 49 h of sleep deprivation. Journal of Sleep Research, 15(1), 7–13. doi: 10.1111/j.1365-2869.2006.00487.x [DOI] [PubMed] [Google Scholar]
  30. Kroenke K, Spitzer RL, & Williams JB (2002). The PHQ-15: Validity of a New Measure for Evaluating the Severity of Somatic Symptoms. Psychosomatic Medicine, 64(2), 258–266. [DOI] [PubMed] [Google Scholar]
  31. Kurina LM, McClintock MK, Chen J-H, Waite LJ, Thisted RA, & Lauderdale DS (2013). Sleep duration and all-cause mortality: a critical review of measurement and associations. Annals of Epidemiology, 23(6), 361–370. doi: 10.1016/j.annepidem.2013.03.015 [DOI] [PMC free article] [PubMed] [Google Scholar]
  32. Lanza ST, Patrick ME, & Maggs JL (2010). Latent Transition Analysis: Benefits of a Latent Variable Approach to Modeling Transitions in Substance Use. Journal of drug issues, 40(1), 93–120. [DOI] [PMC free article] [PubMed] [Google Scholar]
  33. Larson N, Laska MN, Story M, & Neumark-Sztainer D (2015). Sports and Energy Drink Consumption are Linked to Health-Risk Behaviours among Young Adults. Public Health Nutrition, 18(15), 2794–2803. doi: 10.1017/S1368980015000191 [DOI] [PMC free article] [PubMed] [Google Scholar]
  34. Laska MN, Pasch KE, Lust K, Story M, & Ehlinger E (2009). Latent Class Analysis of Lifestyle Characteristics and Health Risk Behaviors among College Youth. Prevention Science, 10(4), 376–386. doi: 10.1007/s11121-009-0140-2 [DOI] [PMC free article] [PubMed] [Google Scholar]
  35. Matthews K, Hall M, & Dahl R (2014). Sleep in healthy black and white adolescents. Pediatrics, 133(5), e1189–e1196. [DOI] [PMC free article] [PubMed] [Google Scholar]
  36. McKnight-Eily LR, Eaton DK, Lowry R, Croft JB, Presley-Cantrell L, & Perry GS (2011). Relationships between hours of sleep and health-risk behaviors in US adolescent students. Preventive Medicine, 53(4), 271–273. doi: 10.1016/j.ypmed.2011.06.020 [DOI] [PubMed] [Google Scholar]
  37. Meldrum RC, & Restivo E (2014). The behavioral and health consequences of sleep deprivation among U.S. high school students: Relative deprivation matters. Preventive Medicine, 63, 24–28. doi: 10.1016/j.ypmed.2014.03.006 [DOI] [PubMed] [Google Scholar]
  38. Miller CB, Gordon CJ, Toubia L, Bartlett DJ, Grunstein RR, D’Rozario AL, & Marshall NS (2015). Agreement between simple questions about sleep duration and sleep diaries in a large online survey. Sleep Health, 1(2), 133–137. doi: 10.1016/j.sleh.2015.02.007 [DOI] [PubMed] [Google Scholar]
  39. Muthén LK, & Muthén BO (2012–2018). Mplus User’s Guide. Eight Edition. . Los Angeles, CA:: Muthén & Muthén. [Google Scholar]
  40. National Sleep Foundation. (2000). Adolescent sleep needs and patterns. Retrieved from http://sleepfoundation.org/sites/default/files/sleep_and_teens_report1.pdf [Google Scholar]
  41. Newbern EC, Anschuetz GL, Eberhart MG, Salmon ME, Brady KA, De Los Reyes A, … Schwarz DF (2013). Adolescent Sexually Transmitted Infections and Risk for Subsequent HIV. American Journal of Public Health, 103(10), 1874–1881. doi: 10.2105/AJPH.2013.301463 [DOI] [PMC free article] [PubMed] [Google Scholar]
  42. Nylund-Gibson K, Grimm R, Quirk M, & Furlong M (2014). A Latent Transition Mixture Model Using the Three-Step Specification. Structural Equation Modeling: A Multidisciplinary Journal, 21(3), 439–454. doi: 10.1080/10705511.2014.915375 [DOI] [Google Scholar]
  43. Nylund KL, Asparouhov T, & Muthén BO (2007). Deciding on the Number of Classes in Latent Class Analysis and Growth Mixture Modeling: A Monte Carlo Simulation Study. Structural Equation Modeling: A Multidisciplinary Journal, 14(4), 535–569. doi: 10.1080/10705510701575396 [DOI] [Google Scholar]
  44. O’Brien EM, & Mindell JA (2005). Sleep and risk-taking behavior in adolescents. Behavioral Sleep Medicine, 3(3), 113–133. [DOI] [PubMed] [Google Scholar]
  45. Owens J (2014). Insufficient Sleep in Adolescents and Young Adults: An Update on Causes and Consequences. Pediatrics. doi: 10.1542/peds.2014-1696 [DOI] [PMC free article] [PubMed] [Google Scholar]
  46. Pasch KE, Laska MN, Lytle LA, & Moe SG (2010). Adolescent Sleep, Risk Behaviors, and Depressive Symptoms: Are They Linked? American Journal of Health Behavior, 34(2), 237–248. doi: 10.5993/AJHB.34.2.11 [DOI] [PMC free article] [PubMed] [Google Scholar]
  47. Reichenberger DA, Hilmert CJ, Irish LA, Secor-Turner M, & Randall BA (2016). Associations Between Sleep and Health-Risk Behaviors in a Rural Adolescent Population. Journal of Pediatric Health Care, 30(4), 317–322. doi: 10.1016/j.pedhc.2015.08.003 [DOI] [PubMed] [Google Scholar]
  48. Shibley HL, Malcolm RJ, & Veatch LM (2008). Adolescents with insomnia and substance abuse: consequences and comorbidities. J Psychiatr Pract, 14(3), 146–153. doi: 10.1097/01.pra.0000320113.30811.46 [DOI] [PubMed] [Google Scholar]
  49. Shochat T, Cohen-Zion M, & Tzischinsky O (2014). Functional consequences of inadequate sleep in adolescents: a systematic review. Sleep Med Rev, 18(1), 75–87. doi: 10.1016/j.smrv.2013.03.005 [DOI] [PubMed] [Google Scholar]
  50. Short MA, Gradisar M, Gill J, & Camfferman D (2013). Identifying Adolescent Sleep Problems. PloS One, 8(9), e75301. doi: 10.1371/journal.pone.0075301 [DOI] [PMC free article] [PubMed] [Google Scholar]
  51. Weinstock H, Berman S, & Cates W Jr. (2004). Sexually transmitted diseases among American youth: incidence and prevalence estimates, 2000. Perspectives on Sexual Reproductive Health, 36(1), 6–10. doi: 10.1363/psrh.36.6.04 [DOI] [PubMed] [Google Scholar]
  52. Wolfson AR, Carskadon MA, Acebo C, Seifer R, Fallone G, Labyak SE, & Martin JL (2003). Evidence for the Validity of a Sleep Habits Survey for Adolescents. Sleep, 26(2), 213–216. [DOI] [PubMed] [Google Scholar]
  53. Wong MM, Brower KJ, Fitzgerald HE, & Zucker RA (2004). Sleep problems in early childhood and early onset of alcohol and other drug use in adolescence. Alcoholism: Clinical and Experimental Research, 28(4), 578–587. doi: 10.1097/01.Alc.0000121651.75952.39 [DOI] [PubMed] [Google Scholar]
  54. Wong MM, Roberson G, & Dyson R (2015). Prospective relationship between poor sleep and substance-related problems in a national sample of adolescents. Alcoholism: Clinical and Experimental Research, 39(2), 355–362. doi: 10.1111/acer.12618 [DOI] [PMC free article] [PubMed] [Google Scholar]

RESOURCES