Sleep-Related Predictors of Risk for Alcohol Use and Related Problems in Adolescents and Young Adults

Introduction

Abundant cross-sectional data indicate that alcohol use and related problems are accompanied by disruptions to sleep and circadian rhythms.¹ Alcohol's negative impacts on sleep are well established, especially in adults, and a smaller body of literature also reports alcohol's disruption of circadian rhythms.^2-4 Growing evidence supports sleep and circadian factors as influencing alcohol use and related problems, including as risk factors for the initial development of use and problems, as predictors for relapse in individuals with alcohol use disorder (AUD), and as targets for intervention.^2,5-7 Given the marked changes in sleep and circadian rhythms that occur throughout adolescence into young adulthood,⁵ paralleling the time frame when initial alcohol use and development of alcohol-related problems are most likely to occur,⁸ there may be particular value to studying the association between sleep/circadian rhythms and alcohol during this developmental stage.

Sleep and Circadian Changes in Adolescence and Beyond

As a result of living on a rotating planet with alternating light and dark periods, humans and most other living organisms have evolved to experience internal biological rhythms lasting approximately 24 hours.⁹ These circadian rhythms modulate the timing of many, if not most, physiological, behavioral, and psychological processes, including the sleep-wake cycle, with the goal of optimizing temporal relationships with the environment and with one another. Notably, the timing of circadian rhythms is not static but shows developmental changes. Starting with the onset of puberty, the timing of sleep and circadian rhythms shifts later throughout adolescence, peaking around age 20 before reversing course and slowly shifting earlier over the rest of the life span.^10,11 The changes in sleep timing are driven by both biological and sociocultural factors and thus can vary based cross-nationally¹² and on sociodemographic characteristics.^11,13 Biological factors include the changes in circadian rhythms as well as changes in homeostatic sleep propensity, which accumulates more slowly during adolescence.¹⁴ Exposure to blue light (e.g., via electronic devices) in the evening can exacerbate these tendencies toward later sleep and circadian timing.^15,16

Although the need for sleep remains relatively stable during this period—with recommendations for 8 to 10 hours/night in youth ages 13 to 17 and for 7 to 9 hours/night in people age 18 and older—actual sleep duration tends to decrease, especially on school/work nights.^14,17 This reduction in sleep duration is driven in part by a mismatch between the tendency for later sleep/circadian timing and relatively early school schedules, particularly during middle school and high school. This mismatch, termed circadian misalignment or social jet lag, not only results in insufficient sleep duration, but also can contribute to difficulty falling asleep on school nights, daytime sleepiness on school days, and large swings in sleep timing and duration on weekdays versus weekends.¹⁴ Such swings tend to manifest as later sleep timing and shorter sleep duration, especially for those with later circadian timing.¹⁸ Although the effects of early school start times are most systematic during secondary education, circadian misalignment and the associated constellation of sleep problems can persist well after high school. Regardless of etiology, insomnia, insufficient sleep, and social jet lag remain prevalent in the years after high school graduation into people's twenties,^18-20 although prevalence varies based on sociodemographic characteristics.²¹

Sleep is multifactorial, and as illustrated above, different facets of sleep are interrelated in complex ways.^22,23 Circadian misalignment and social jet lag are often accompanied by a constellation of sleep-related problems and thus cannot be adequately captured by only assessing sleep quality, sleep duration, or sleep timing, especially if not distinguishing between school days or workdays and free days.

Alcohol Trajectories in Adolescence and Beyond

The developmental span from adolescence to young adulthood is a time of increasing alcohol use and related problems.⁸ Alcohol use then tends to decline in early adulthood as individuals begin to "mature out" due to increases in adult responsibilities.²⁴ Further, both earlier initiation of alcohol use and more rapid progression from initiation to intoxication have been found to predict problematic alcohol use later on.^25-27 Multiple explanatory mechanisms are thought to underlie the onset and progression of risky alcohol use in adolescence through early adulthood. In particular, heightened sensation seeking and impulsivity have been consistently identified as potential risk factors for problematic alcohol use^28-33 and are related to sleep and circadian factors.^34,35

Overview of Alcohol's Effects on Sleep

The effects of alcohol on sleep and, to a lesser extent, circadian rhythms in adult samples have been thoroughly and recently reviewed,^2-4 so are only briefly discussed here. Given the bidirectional relationships between sleep and alcohol use, a brief summary of the evidence for alcohol's effects on sleep and circadian rhythms is warranted as it provides important context in interpreting observational data where it is impossible to fully parse these bidirectional effects.

Alcohol administration studies in adults have assessed alcohol's acute effects on sleep via polysomnography, which measures brain activity (electroencephalography [EEG]), eye movements, muscle activity, and cardiac activity. These studies found that during the first half of the night, alcohol tends to shorten the time it takes to fall asleep (sleep onset latency [SOL]), reduce nighttime wakefulness (i.e., decrease wake after sleep onset [WASO]), decrease rapid eye movement (REM) sleep, and increase the deepest of the non-REM sleep stages (i.e., slow-wave sleep).² (See Box: Glossary of Sleep-related Terms for more detailed definitions.) However, during the second half of the night, alcohol tends to acutely increase WASO and reduce sleep efficiency (the percentage of time spent asleep relative to the time spent attempting to sleep), while leading to a rebound in REM sleep.² Overall, polysomnography studies suggest that adults spend more time awake on nights after consuming alcohol.² Some sex differences in the acute effects of alcohol have been noted, as described below.

Acute alcohol effects in adolescents have been much less studied, but findings suggest some distinctions from the effects observed in adults. In a study with 24 participants ages 18 to 21 (12 women) with a mean breath alcohol content of 0.084% at lights out, alcohol's effects were broadly similar but with less evidence of benefits for sleep. Specifically, adolescents did not exhibit the decrease in SOL or the REM rebound,³⁶ and although alcohol appeared to increase delta power (EEG activity in the 1–4 Hz range; typically highest during slow-wave sleep) during the first few sleep cycles, it simultaneously increased alpha power (EEG activity in the 8–13 Hz range; associated with quiet wakefulness) in frontal regions.³⁷ This alpha-delta pattern in response to alcohol has been observed in some but not all prior studies^38,39 and is thought to reflect disrupted sleep. No sex differences were reported.

As reviewed by Koob and Colrain,² alcohol's effects on sleep—when alcohol use is more chronic and/or when people who chronically use alcohol (i.e., patients with AUD) abstain from drinking—can diverge from the acute effects of alcohol in complex ways too nuanced to adequately review here. Generally, chronic alcohol use is associated with worse sleep (e.g., more insomnia, longer SOL and WASO), although sleep may intermittently improve on drinking nights; similarly, abstinence is typically associated with initial worsening of sleep with some incremental improvement over time.² Various sleep abnormalities persist in individuals with AUD, even with long-term abstinence (> 30 days). A recent meta-analysis of cohort studies in broader samples underscores the general conclusion that chronic alcohol use does not improve sleep overall, and likely increases the likelihood of developing sleep disorders over time.⁴⁰

Although intensive longitudinal studies cannot confirm causality or directionality, analyses of day-to-day alcohol–sleep associations in young adults suggested that drinking on a given day was associated with later sleep timing that night.^41,42 Interestingly, such analyses offered mixed evidence for whether drinking worsened⁴³ or improved⁴² sleep. Additionally, some studies in young adults have shown that cannabis use may mitigate alcohol's effects on sleep,^43,44 although these studies require replication, and the relevant mechanisms remain unknown.

Alcohol's effects on sleep also depend on the timing of alcohol consumption; for example, a study in middle-aged men administered alcohol 6 hours before bedtime found no benefit for SOL.⁴⁵ This likely was due to a combination of the temporal dynamics of the biphasic response to alcohol and circadian variation in the response to alcohol. While the literature on alcohol effects on circadian rhythms is more limited, particularly in humans,^3,4 studies have suggested disruption of melatonin and core body temperature rhythms. Multiple animal studies have indicated that acute and chronic alcohol use disrupted the circadian system's response to light, which is the most important cue (zeitgeber, or time giver) for entraining to the 24-hour day.^46,47

Although parallel effects in humans were not supported by one study in healthy adults reporting light or regular but not heavy alcohol use,⁴⁸ more recent work suggested reduced retinal responsivity to light in a group of adults who drank heavily. Light or regular drinking has previously been defined as "consumption of one to five standard alcoholic drinks/week" and no more than three episodes of binge drinking in the past year.⁴⁹ Heavy drinking has been defined by the National Institute on Alcohol Abuse and Alcoholism as five or more drinks on any day or 15 or more drinks per week for men, or four or more drinks on any day or eight or more drinks per week for women (see https://go.nih.gov/TiogZz9). However, there is no standardized definition of either "light/regular drinking" or "heavy drinking" across the studies described in this article.

The present paper reviews current evidence for prospective associations between sleep/circadian factors and alcohol involvement during adolescence and young adulthood, with an emphasis on the effects of sleep/circadian factors on alcohol use and related outcomes. This focus was selected in part because identifying modifiable sleep–alcohol relationships during this developmental period offers the potential for shifting the trajectory for alcohol-related problems before they develop into chronic AUD. This article also describes and discusses potential mechanisms by which sleep may influence alcohol use and problems, as well as potential important differences in sleep–alcohol associations based on key moderators, such as assigned sex at birth; lesbian, gay bisexual, transgender, queer/questioning, intersex, and asexual (LGBTQIA+) identities; and racial and ethnic identities.

Methods

Search Methods

The initial search of the existing literature was conducted on July 18, 2022, in PsycInfo, PubMed, and Web of Science using the search terms "sleep" and "alcohol" paired with "adolescent" or "adolescence" or "young adult" or "emerging adult," in the title or abstract fields; results were restricted to English-language articles but had no restriction by date. Next, the search was narrowed by including only articles that had a prospective/longitudinal or experimental design, included a sleep-related measure as a predictor, assessed an alcohol-related measure as an outcome, and had a sample primarily composed of adolescents and/or young adults. Based on these search terms, the resulting ages of participants in the articles ranged from ages 12 to 30. Table 1 offers information on age ranges in specific studies. Two of the authors completed this step by conducting a joint review of candidate article abstracts.

Results of the Literature Search

The initial search resulted in 720 articles (174 in PsycInfo, 305 in PubMed, and 241 in Web of Science). After review of the abstracts to identify articles that met all the key search criteria, the list was narrowed to 27 articles reporting on observational longitudinal studies and three articles reporting on experimental studies (specifically, two intervention trials). Noted for potential inclusion were 35 additional articles that reported on studies with alcohol-related predictors and sleep-related outcomes, and/or reported on candidate moderators or mediators of sleep–alcohol associations. An additional 104 articles cited here were identified via a variety of methods, including review of relevant article reference lists and prior exposure based on the authors' previous work in this area. Finally, while this review focused on sleep/circadian–alcohol associations in human studies, a few select findings from three animal studies^46,47,50 were included when they appeared particularly complementary to the human findings and/or helped speak to a gap in the human literature.

Results of the Reviewed Studies

Longitudinal Sleep and Alcohol Studies

Overall, the existing literature—based on 27 articles, including three intensive longitudinal studies—provides consistent evidence that a range of sleep/circadian factors during adolescence predicts later alcohol involvement. These included difficulties with falling or staying asleep, lower overall sleep quality, shorter sleep duration, daytime sleepiness, later sleep timing and/or chronotype (i.e., tendency for relatively earlier or later sleep-wake timing), and variable sleep timing and/or social jet lag (see Box: Glossary of Sleep-related Terms). Alcohol-related outcomes assessed included metrics of both quantity and frequency of use, binge or heavy drinking episodes, alcohol intoxication, alcohol-related consequences/problems, AUD symptoms, and alcohol craving. Table 1 provides a summary of the longitudinal studies, including sample composition, study design, and timescale; which multidimensional sleep variables were predictive of alcohol outcomes; and whether differences across assigned sex, gender identity, and racial/ethnic identity were assessed.

*Age was rounded to 1 decimal place; demographic (sex/gender and race/ethnicity) percentages were rounded to whole numbers. No studies clarified whether assigned sex at birth or gender identity was assessed and reported. Therefore, the language (e.g., sex vs. gender) in the original article was retained. When possible, race/ethnicity terms were standardized to be consistent with National Institutes of Health categories. In some cases, the published papers did not specify the racial/ethnic identities beyond "White" and "non-White" (or "Caucasian" and "non-Caucasian," for which "White" and "non-White" were substituted).

†Sleep variables based on insomnia symptoms without numerical data to calculate efficiency were categorized under "Satisfaction."

Note: ADD Health, National Longitudinal Study of Adolescent to Adult Health; AUD, alcohol use disorder; ECHO, Etiology of Childhood Obesity studies; IDEA, Identifying the Determinants of Eating and Activity; M, mean; MLS, Michigan Longitudinal Study; NCANDA, National Consortium on Alcohol and Neurodevelopment in Adolescence; SD, standard deviation; SUD, substance use disorder.

A majority of the articles^5,6,51-63 also reported on other substance outcomes, particularly use of nicotine/tobacco and cannabis/marijuana, with findings suggesting that sleep-related risk for substance use may not be specific to alcohol. Indeed, the overall literature suggests a transdiagnostic scenario where multiple aspects of sleep/circadian disturbance (e.g., insomnia, sleep loss, delayed phase) increase the risk for alcohol and other substance use disorders as well as for other psychiatric disorders.⁶⁴

Although this review focuses primarily on the period of adolescence through young adulthood, two papers based on the Michigan Longitudinal Study^61,62 and one paper based on a study in Hong Kong⁶⁵ reported that childhood sleep problems predicted later substance use, indicating that relationships between sleep and substance use are not specific to adolescents. Notably, childhood sleep tends to predict adolescent sleep,^62,65 which could partially explain the association with adolescent substance use, but also suggests the potential value of starting early with sleep-focused prevention and/or intervention efforts. Indeed, one study reported prospective sleep–substance use associations entirely within the fourth through sixth grades, and implicated inhibitory control as a potential mediator.⁶⁶ Although that study's findings contrasted with one of the papers from the Michigan Longitudinal Study (which did not support inhibitory control as a mediator in the sleep–alcohol associations),⁶² changes in mood regulation, impulsivity, and/or poor decision-making remain plausible mechanisms in the longitudinal associations between childhood sleep problems and later substance use.

Several caveats are important to consider when interpreting the existing literature. First, multiple articles relied on the same longitudinal datasets; thus, 14 out of 27 longitudinal papers were based on six studies. Second, earlier studies tended to focus on only one or two sleep characteristics and were thus unable to treat sleep as a multidimensional construct. Third, papers based on more recent studies, seemingly designed to specifically consider sleep, were more likely to employ a multidimensional sleep framework.^{5,6,41,53,67-69} Fourth, except for two intensive longitudinal studies^41,42 that used actigraphy—a wearable device containing an accelerometer to measure rest-activity patterns—most studies relied on self-reported sleep and are subject to the relevant biases. For example, beyond typical retrospective biases associated with self-report, there are also longstanding observations of subjective-objective discrepancies in sleep, particularly in individuals with insomnia disorder.⁷⁰ Also, none of the studies included objective circadian predictors (e.g., dim light melatonin onset) despite cross-sectional evidence that circadian timing is related to alcohol outcomes.^49,71 Fifth, observational designs cannot assess causation and directionality and therefore must be interpreted with caution. Relatedly, one recent co-twin study indicated that sleep-related risk for alcohol misuse exists over and above genetic and environmental factors.⁷² However, other emerging research using genetic methods has yielded more mixed results whether the relationships between sleep/circadian characteristics and substance use should be attributed to shared genetic variance or pleiotropy^73-75 or suggests a causal relationship from sleep to substance misuse.⁷⁶

Some of the included studies tested putative mediators of the sleep–alcohol relationship, such as behavioral inhibition, attention problems, and internalizing/externalizing symptoms; however, the results have been inconsistent (see below for further discussion). Furthermore, given that a tendency toward relatively late timing of the sleep-wake schedule (i.e., a later chronotype) is often associated with worse sleep among adolescents and young adults,⁷⁷ sleep characteristics are a putative mediator of the association between chronotype and alcohol-related risk. However, existing studies often have not supported this for alcohol⁷⁸ or other outcomes such as depression.^79,80 One study in late adolescents and young adult veterans reported that insomnia severity statistically mediated the association between depression or symptoms of post-traumatic stress disorder and alcohol use and related consequences.⁸¹

The time frame of assessment varied substantively across the studies, with intensive longitudinal designs narrowing the focus to day-to-day relationships whereas the more traditional longitudinal studies ranged from months to multiple years between assessments. These varying time frames are important when considering that distinct mechanistic pathways may be operating within different timescales. For example, studies with annual or multiannual time points may be speaking more to the cumulative effects of sleep/circadian characteristics, although few studies have directly tested this.⁶⁵ Interestingly, the intensive longitudinal designs (e.g., ecological momentary assessment [EMA]) appear more likely to find more nuanced associations between sleep and alcohol. For example, some EMA evidence from young adult samples suggests that better sleep efficiency⁴² or quality⁴¹ on a given night predicts more alcohol use the following day, although those findings emerged from samples with participants with sleep problems who consume alcohol. EMA findings from a much wider age range (ages 20 to 73) suggest that age may moderate sleep–alcohol associations; the younger group (age < 49) showed associations between worse sleep quality and more subsequent alcohol use whereas the older group (age > 50) drank more following nights of better sleep quality.⁸²

The complex findings in EMA studies speak to the relevance of considering the multidimensional nature of sleep. In one study of undergraduate students who consumed alcohol (mean age = 20.5 years), shorter sleep and earlier wake times (based on actigraphy) and better sleep quality (based on self-report) all predicted more alcohol use the next day.⁴¹ In the combined model that included all the sleep predictors simultaneously, only waking earlier and better perceived sleep quality upon waking predicted more alcohol use. One interpretation of this is that shortened sleep led to deeper, more consolidated sleep, perceived in turn as higher quality, although it remains possible that shorter sleep may have impacted other intervening mechanisms (e.g., impaired cognitive control). Alternatively, as the authors suggested, late adolescents and young adults may be more likely to socialize and drink when feeling refreshed, especially given that drinking among adolescents and young adults primarily occurs in social contexts.⁸³ Collectively, these findings suggest the value of considering multidimensional sleep relationships with alcohol using designs that allow consideration of both short-term (i.e., day-to-day) and longer-term (i.e., months-to-years) timescales, such as embedding an EMA burst design within a longitudinal study, as done by Graupensperger and colleagues.⁵¹ Relatedly, such designs allow parsing of between-person and within-person effects, which may well reveal distinct sleep–alcohol associations at the between-person and within-person levels.

In summary, the published longitudinal data indicate that multiple sleep and/or circadian characteristics prospectively predict alcohol-related outcomes during adolescence through young adulthood. However, the current literature is limited by overreliance on a relatively small number of longitudinal studies, largely relying on self-report measures, and insufficient consideration of the multidimensional nature of sleep. Important next steps include, but are not limited to, consideration of different timescales, including within the same study design, and examination of key mediators and moderators of sleep–alcohol associations.

Experimental Sleep and Alcohol Studies

At present, experimental evidence of causal effects of sleep on alcohol-related outcomes is based solely on insomnia treatment studies in individuals with heavy alcohol use and/or AUD, most of which are from samples older than adolescents or young adults. A systematic review and meta-analysis of nine studies of primarily middle-aged adults⁷ concluded that insomnia treatment, particularly behavioral treatment, improved sleep quality and reduced depression in individuals with AUD. The authors found no definitive benefit of insomnia treatment for reducing alcohol use, although the relapse rates in two trials of cognitive-behavioral therapy for insomnia (CBT-I) were considerably lower (11% and 15%) than might be expected for adults in AUD treatment.⁸⁴ Caution is warranted in drawing strong conclusions about the potential impact on alcohol-related outcomes based on these studies, however, as the review also noted limitations related to small samples, relatively short follow-up periods, and not focusing on participants who were concurrently engaged in AUD treatment. These limitations reflect the fact that the studies generally were designed to focus on sleep outcomes rather than alcohol outcomes. Moreover, these studies varied in whether the patients were seeking or engaged in AUD treatment, and whether they were required to be abstinent at study start.

The limited published data from two sleep treatment studies in late adolescents and emerging adults are broadly consistent with the prior literature in adults, suggesting that sleep disturbance in the context of heavy alcohol use is amenable to nonpharmacological interventions; however, it remains unclear whether improving sleep measurably reduces alcohol involvement. A novel web-based intervention including both sleep and alcohol content improved sleep quality and sleep-related impairment in heavy-drinking college students,⁸⁵ although it did not outperform a control condition (psychoeducation about sleep hygiene) and did not significantly improve actigraphy-based sleep outcomes. Interestingly, although alcohol use through a 3-month follow-up declined in both conditions, reductions were larger in the control condition. The results suggested that greater reductions in sleep-related impairment may predict greater reductions in drinking (medium-to-large effect size), but those findings were not statistically significant.

Related work by Fucito and colleagues suggested that heavy-drinking college students were more receptive to sleep-focused interventions (even if they included content related to drinking) than to purely alcohol-focused interventions.⁸⁶ This may be due to less stigma associated with sleep treatment. Aside from direct effects of sleep treatment on alcohol outcomes, this could mean that sleep treatment may provide a "foot in the door" for individuals with sleep and alcohol problems. Accordingly, Fucito and colleagues are currently conducting a sleep intervention trial that focuses on sleep hygiene in young adults ages 18 to 25 who drink heavily.⁸⁷

A more recent study tested the efficacy of CBT-I in 56 young adults ages 18 to 30 who reported monthly binge drinking and met criteria for insomnia disorder.^88,89 The study differed from prior CBT-I and alcohol studies in the sample age and that participants were still actively drinking at the start of CBT-I. The only alcohol-related treatment component was the standard sleep hygiene recommendation to reduce alcohol use before bedtime. With regard to sleep outcomes, CBT-I reduced self-reported insomnia severity relatively better than the sleep hygiene control condition, although neither treatment significantly improved actigraphy-based sleep efficiency.^88,89 Although drinking quantity and drinking-related consequences both decreased over time, these outcomes were not differentially better during CBT-I.⁸⁸ However, although insomnia improvements were not related to changes in drinking, they did mediate the reduction in alcohol-related consequences in the CBT-I group. A secondary analysis reported greater (albeit modest) reductions in alcohol craving for the CBT-I group than for the control group that, again, were statistically mediated by improvements in insomnia.⁸⁹ However, those reductions in alcohol craving were not sustained at the 1-month follow-up assessment.

In summary, the existing experimental literature on sleep predictors of alcohol outcomes during adolescence and young adulthood is confined to a handful of trials testing nonpharmacological sleep interventions in individuals reporting heavy drinking and/or AUD. Consistent with the parallel literature in adult samples, such interventions appear beneficial for sleep-related outcomes but with no clear impact on alcohol-related outcomes. However, a preliminary finding of CBT-I reducing alcohol craving is worth further investigation, as is the further development of sleep-focused treatments, perhaps including more consideration of circadian factors.

Potential Mediators and Moderators

Prior reviews have examined plausible mechanisms linking sleep/circadian disturbances to alcohol use and alcohol-related problems, with a particular emphasis on reward function.^90-92 A recent review by the authors⁹¹ proposed a broader conceptual model that considered both positive and negative reinforcement pathways, and noted that elevated impulsivity may exacerbate either pathway. While this model may have heuristic value, it is not without limitations. These include not explicitly addressing bidirectional effects (i.e., alcohol effects on sleep/circadian function) or incorporating plausible factors that influence which pathway is most salient for a particular individual or at a given time. Further, research on sleep–alcohol associations has largely been conducted with samples of predominantly White-identifying individuals and has largely not explored possible differences in associations between sleep and alcohol across assigned sex, racial and ethnic identities, and for LGBTQIA+ individuals. The following sections offer some preliminary evidence of the importance of including diverse samples in future investigations and of examining differences in associations to ensure generalizability of future treatments and to inform culturally responsive interventions for both sleep/circadian disturbances and AUD.

Mechanisms related to positive reinforcement

Extensive cross-sectional, longitudinal, and experimental evidence from both human and preclinical studies has supported the influence of sleep/circadian factors on reward-related processes and underlying physiology⁹² and, in turn, the relevance of reward-related processes to risk for alcohol use and related problems.^28,93,94

Although relevant human experimental studies probing sleep/circadian effects on reward-related processes have been more scarce than animal models, experimental sleep deprivation protocols have demonstrated causal effects on reward-related brain function in healthy adolescent and young adult samples.^95-97 For example, experimentally imposed circadian misalignment reduced the neural response to monetary reward and during response inhibition in healthy adolescents without regular substance use.⁹⁸ The analyses included objective measures of sleep duration and alertness, thus suggesting circadian effects on reward function beyond those of insufficient sleep. However, these studies have focused on non-alcohol rewards. In contrast to emerging animal research suggesting circadian misalignment during adolescence alters reward circuitry function and increases alcohol use during adulthood,⁵⁰ almost no published human studies have examined sleep/circadian effects on alcohol cue reactivity and/or its neural correlates. Furthermore, few existing studies have combined sleep/circadian effects, reward, and alcohol outcomes, although one cross-sectional analysis found that "eveningness"—the self-reported preference for relatively later timing of sleep and activity—was associated with altered neural processing of reward, which in turn is associated with greater alcohol use and AUD symptoms.⁹⁹ A longitudinal analysis in the same study found that the prospective association between eveningness and AUD symptoms was statistically mediated by the medial prefrontal cortex response to monetary reward.⁵² Most recently, a study reported that an objective measure of circadian misalignment (measured on a Thursday) prospectively predicted a lower neural response to monetary reward (measured on a Friday) in late adolescents with regular alcohol use.¹⁰⁰ However, the reduced neural response to reward did not prospectively predict alcohol use that weekend, but rather was associated with more binge drinking episodes at baseline. Finally, in the aforementioned CBT-I trial in adolescents and young adults reporting heavy alcohol use and insomnia, the investigators found evidence of relatively larger reductions in delay discounting (large rewards only) in the CBT-I group, although this was not mediated by insomnia severity. However, there was no apparent effect on negative affect, suggesting that improved sleep may have relatively greater effects on reward-related processes.⁸⁹

Some evidence suggests sleep/circadian modulation of the stimulating effects of alcohol (e.g., increases in energy and excitement). This may be particularly relevant during adolescence, when alcohol may be relatively more stimulating and less sedating than in adulthood.^101-103 Notably, a relatively more stimulating response to alcohol is a risk factor for AUD. Thus, adolescents at high risk for AUD endorsed greater alcohol-induced stimulation and stronger wanting for alcohol compared to adolescents at low risk for AUD.¹⁰⁴ Moreover, young adults reporting greater stimulation after alcohol administration were more likely to have developed AUD by 10-year follow-up.¹⁰⁵ In laboratory-based sleep studies in late adolescents and emerging adults, acute alcohol administration did not reduce SOL,³⁶ especially when consumed in the evening,¹⁰⁶ suggesting the stimulating rather than sedating effects also may be influenced by time of administration. Furthermore, later sleep timing was associated with greater self-reported stimulation response following alcohol administration in the laboratory (at least in White male participants).¹⁰⁷

Lastly, sleep/circadian factors may be relevant to positive reinforcement-related alcohol cognitions. Adolescents and young adults tend to report more motives attributed to improving their social experiences and enhancing enjoyment versus motives related to attenuating negative affect (i.e., coping).¹⁰⁸ Given that eveningness is associated with increased alcohol motives across the board,¹⁰⁹ including enhancement and social motives, it is possible that the tendency toward later sleep/circadian timing in this age group contributes to reasons for using alcohol.

Mechanisms related to negative reinforcement

Adverse life events and stress levels disrupt sleep and prospectively predict AUD outcomes, both on a longitudinal basis during adolescence into adulthood,^110,111 and more proximally (day to day).^112,113 Furthermore, demonstrating sleep- or drinking-related reactivity to stress heralds the risk for sleep-¹¹⁴ or alcohol-related problems¹¹⁵ in the future.

Several lines of evidence indicate that sleep problems, perhaps driven by stress and/or anxiety, may lead to using alcohol as a coping method, thus implicating negative reinforcement pathways. Studies suggest that about 10% (range 6% to 16%) of adolescents and young adults report using alcohol as a sleep aid, with higher rates in individuals with heavier alcohol use and/or worse sleep.^116-118 Interestingly, one longitudinal study of adolescents with and without AUD found that their use of alcohol as a sleep aid declined over time, dropping by half from baseline to 5-year follow-up; this may reflect adolescents' learning that alcohol's effectiveness at promoting sleep declines with regular use.¹¹⁹

Compared to "good sleepers," adults with insomnia may experience relatively greater tension reduction and deeper sleep (based on slow-wave sleep) in response to alcohol, underscoring why they might initially turn to alcohol as a sleep aid. Although experimental evidence suggests they rapidly develop tolerance to these effects, these individuals often persist in choosing alcohol as a sleep aid.^120,121 Similarly, young adults with insomnia who regularly use alcohol reported better sleep efficiency on drinking days, seemingly due to shorter SOL, in a recent EMA study,⁴² and reported sleeping worse on nights when they avoided alcohol in the 2 hours before bed.¹²² In contrast with the experimental study,¹²⁰ the association with better sleep efficiency remained even after accounting for number of consecutive drinking days.⁴² Notably, these associations were not observed for actigraphy-based sleep efficiency.

Sleep also may modulate effects of stress on alcohol use. Along with associations with drinking motives in general (see above), eveningness in college students was associated with worse coping with stress, which in turn may predict drinking to cope.¹⁰⁹ Another study found that late chronotypes had both more adverse childhood experiences and greater alcohol use during young adulthood.¹²³

Craving-related mechanisms

Craving—a criterion for diagnosis of AUD and widely studied as a proximal predictor of alcohol use—is a complex construct, with apparent contributions of both positive- and negative-reinforcement processes.¹²⁴ Recent studies have offered preliminary evidence that alcohol craving is influenced by sleep/circadian factors. Two studies reported the presence of a 24-hour rhythm in alcohol craving,^125,126 suggesting modulation by circadian rhythms, although the studies were mixed in whether sleep characteristics predicted the timing or amplitude of the craving rhythm. Lower sleep quality was associated with elevated tonic (i.e., long-term) craving as determined using the Obsessive-Compulsive Drinking Scale, but not with cue-induced craving (as measured using the Alcohol Urge Questionnaire) during a cue reactivity paradigm in patients with AUD.¹²⁷ Finally, less sleep predicted more alcohol craving the next day in an EMA study,⁵¹ and reductions in insomnia severity mediated reductions in alcohol craving in a CBT-I trial.⁸⁹

Relatedly, growing evidence implicates a role for the orexin/hypocretin system in sleep-alcohol associations via both negative reinforcement and reward-related processes. Orexin/hypocretin regulates wakefulness, reward seeking, and other motivated behavior, including alcohol craving and alcohol seeking; in turn, the orexin/hypocretin system is modulated by acute and chronic stress.^128,129 Ongoing trials are testing whether suvorexant, a dual orexin receptor antagonist, can reduce both alcohol craving and insomnia symptoms.^130,131

Impulsivity-related mechanisms

Similar to craving, the multifaceted construct of "impulsivity" may be relevant to both positive and negative reinforcement pathways in understanding sleep/circadian-related risk for alcohol involvement. In general, facets of impulsivity are considered a key risk factor for the development of heavy alcohol use and related problems.^29,32 Importantly, impulsivity facets may differentially relate to alcohol use through both positive and negative reinforcement pathways. For example, negative urgency, or acting rashly in response to strong negative mood, may reflect drinking to cope with negative mood/stress whereas positive urgency may reflect expecting alcohol to increase arousal.¹³²

Multiple sleep/circadian characteristics have been linked to impulsivity domains (e.g., Kang et al.^34,35). For example, recent prospective evidence in adolescents suggested that both sleep duration and insomnia were bidirectionally associated with impulse control.¹³³ Recent studies found that later chronotype was associated with greater impulsivity overall (e.g., Kang et al.³⁴), including greater self-reported trait- and state-level impulsivity across multiple subdimensions in White male drinkers.¹³⁴ Also, as noted above, experimentally imposed circadian misalignment reduced neural activation in the right inferior frontal gyrus during response inhibition in healthy and non–substance-using adolescents.⁹⁸

Moderation by assigned sex and gender identity

Studies found that both sleep/circadian characteristics and risk for problematic alcohol use vary by assigned sex at birth (sex); however, there has been insufficient attention to the role of sex in sleep/circadian–alcohol associations. This is important as rates of AUD among female individuals have risen 84% in the past decade, compared to a 34% increase among male individuals.^135,136 Consistent with this trend, alcohol use has risen for women but not men.¹³⁷ Prior research found that female individuals reported higher levels of disturbed sleep (e.g., insomnia),¹³⁸ while male individuals tended to report later sleeping times.¹³⁹ Recent findings suggest that sleep/circadian characteristics differentially contribute to alcohol risk for male and female individuals. Indeed, recent longitudinal studies found that male individuals in particular may be at heightened alcohol-related risk attributed to sleeplessness^138,140 and later weekday/weekend bedtime.⁶ However, other studies observed stronger associations between multiple sleep characteristics (e.g., total sleep time, sleep efficiency, nighttime awakenings) and alcohol-related risks among female individuals.^5,141 Factors that may contribute to increases in alcohol use and sleep disturbance among female individuals may include heightened drinking to cope with negative affect and stress.^142-144 However, these studies did not clarify whether they were measuring assigned sex or gender identity (the term "identity" is used to reflect that race and gender are social constructs¹⁴⁵ and that the vast majority of research on humans asks participants to self-identify their race and gender).

Inequities in sleep^146,147 and alcohol use¹⁴⁸ exist for individuals with minoritized gender identities (e.g., transgender, nonbinary, gender-fluid). Importantly, a recent study examining factors that influenced sleep among individuals who identified as transgender found that one-third of the sample endorsed feelings of internalized shame (i.e., distress, anxiety, and dysphoria attributed to their identity) as reason for sleep disturbance.¹⁴⁹

Inequities in sleep duration¹⁵⁰ and alcohol use¹⁵¹ also exist among individuals with minoritized sexual orientations (e.g., lesbian, gay, queer, bisexual). However, only one cross-sectional study has examined whether sleep/circadian characteristics contribute to inequities in alcohol problems and whether these associations present differently among subgroups of people with minoritized sexual orientations (e.g., bisexual women, gay men).¹⁵² The study found that compared to heterosexual men, gay men were less likely to experience short sleep duration and reported consuming fewer alcohol drinks per day. Lesbian and bisexual women, when compared to heterosexual women, reported a greater number of alcoholic drinks per day and were more likely to use sleep medication. Further, bisexual women were more likely to experience short sleep duration and to be diagnosed with a sleep disorder compared to heterosexual women.

It is important to place these findings within a minority stress model framework, where individuals with minoritized identities are exposed to identity-based stressors¹⁵³ that occur at both interpersonal and systemic levels.¹⁵⁴ Identity-based stressors—defined as chronic modes of stress attributed to discrimination and internalized stigma directed at one's minoritized identity (e.g., sexual, gender, or racial identities)—are prominent predictors of health inequities, including alcohol behaviors and sleep disturbances, among individuals with minoritized sexual and gender identities.^146,155,156 However, further examination of possible differential associations between sleep indices and alcohol behaviors is needed.

Moderation by racial and ethnic identities

As a function of sociohistorical context and multiple levels of discrimination, inequities in sleep health and alcohol problems have been shown for individuals with minoritized racial and ethnic identities.^157-162 Significantly less research has examined if sleep disturbances related to discrimination contribute to the inequities in alcohol problems and whether the associations between sleep and alcohol differ among individuals with different racial or ethnic identities. Structural racism affects neighborhood-level factors that impact sleep (e.g., noise pollution) and alcohol use (e.g., alcohol outlet density), and neighborhood socioeconomic indicators (i.e., income, crime rates, discrimination) have been implicated in inequities in sleep, which may contribute to downstream poor health outcomes.¹⁶³ Specifically, studies have identified that individuals with low socioeconomic status tend to inhabit urban areas, which may be more hazardous and noisier and may have higher levels of crime. Such neighborhood characteristics have been found to be associated with greater rates of chronic sleep disturbance,¹⁶⁴ which in turn have been linked to heightened alcohol consumption among adolescents as reviewed above (also see Edwards, Reeves, and Fishbein¹⁶⁵). As individuals with minoritized racial and ethnic identities may be more socioeconomically disadvantaged as a result of sociohistorical structural and interpersonal discrimination, these youth may be at greater risk for poor sleep quality in addition to elevated risk for alcohol use. These environmental factors may also affect associations between sleep and alcohol differently for individuals with minoritized racial or ethnic identities. All of these potential associations have direct implications for prevention and treatment.

Cross-sectional evidence suggests that alcohol use may be more disruptive to sleep for Black individuals relative to White individuals. Among men with AUD, Black men had more severe sleep disturbances compared to White men.¹⁶⁶ Based on National Health Interview survey data collected between 2004 and 2015, sleep duration and sleep quality were highest in Black individuals who never consumed alcohol (i.e., lifetime abstention) and worsened as alcohol use involvement increased.¹⁶⁷ For White individuals, this pattern was more variable. Importantly, the racial differences in this study were more pronounced for women than men, demonstrating the importance of examining intersectionality.

Research examining associations between sleep and alcohol use in minoritized racial or ethnic groups beyond Black or African American individuals is nascent. However, consistent with research with predominantly White samples, binge drinking in adolescence has been shown to relate to poorer sleep quality in young adulthood for Mexican American and American Indian (as defined in the article) individuals.¹⁶⁸

Studies examining how sleep may differentially affect alcohol use and experiences while drinking across racial and ethnic groups are even more sparse. Preliminary research found that later sleep timing was related to increased sensitivity to the stimulating effects of alcohol for White men but not Black men;¹⁰⁷ however, no differences existed in associations with 24-hour rhythms in alcohol craving for Black and White young adults.¹²⁵

Other possible moderators

Multiple other moderators of the relationship between sleep/circadian factors and alcohol use are plausible but have received little attention to date, including the role of age and/or developmental stage. An exploratory analysis of the longitudinal data from the National Consortium on Alcohol and Neurodevelopment in Adolescence study⁵ found a different pattern of sleep/circadian predictors of binge alcohol severity at middle- and high-school age time points versus post–high-school age time points. This difference could reflect context, given systematic early school start times versus more flexibility in schedules after high school (i.e., college and/or employment), but more research is needed to replicate and further clarify this finding.

Sleep/circadian-related risk for alcohol outcomes also may be moderated by the stage of alcohol use and related problems, potentially varying as individuals progress through the three-stage cycle framework of AUD—binge/intoxication, negative affect/withdrawal, and preoccupation/anticipation as described by Koob and Colrain.² The shift from enhancement motives/positive reinforcement in the binge/intoxication phase to coping motives/negative reinforcement in the withdrawal/negative affect stage could be paralleled by a shift in relevant sleep/circadian pathways. That is, accumulating alcohol use/problems may contribute to more chronic and/or more distinct sleep/circadian disturbances, which in turn may maintain or exacerbate alcohol involvement. Additionally, sleep problems have been identified as a risk factor for relapse during early abstinence in individuals with AUD.²

Conclusions and Future Directions

Based on the above discussion, future research on the intersection between sleep and alcohol should address existing gaps related to both research methodology and specific questions addressed. For example, future studies should employ assessment batteries able to assess multidimensional sleep/circadian characteristics and should include both self-report and objective measures, particularly objective assessments not yet sufficiently leveraged in this literature, such as the Multiple Sleep Latency Test to assess daytime sleepiness. Research also can benefit from the use of combined longitudinal and intensive longitudinal designs, such as EMA bursts within a larger longitudinal study framework, which will allow consideration of both different timescales and parsing of between-person (trait) and within-person (state) effects.

Such studies should further explore the role of relevant moderators, with particular attention to sleep–alcohol associations for individuals with minoritized identities. Equally important is consideration of the association between sleep and cannabis use, including simultaneous use with alcohol, given the high prevalence of this practice in late adolescents and young adults and evidence suggesting somewhat opposing effects of both substances on sleep. Examination of potential differences in sleep–alcohol associations across international samples could help determine how varying cultural contexts may differentially influence sleep, alcohol use, and their association.

Furthermore, experimental research is needed to demonstrate causal effects of sleep/circadian manipulations on alcohol-related risk. Additionally, experimental studies using approaches such as forced desynchrony or ultradian sleep-wake protocols could help parse the role of circadian versus sleep homeostatic contributions in modulating alcohol-related processes (e.g., alcohol craving).

Other research gaps to be addressed include the clarification of potential shared genetic variance and/or pleiotropic contributions to sleep–alcohol associations, which should further clarify trait- versus state-level effects, as well as investigation of different mechanistic pathways linking sleep to alcohol outcomes. These ideally should allow for comparison of distinct pathways within the same dataset and include not only the putative mechanisms described above (e.g., reward function, negative reinforcement, impulsivity) but also others that may well be worth consideration, such as hypothalamic-pituitary-adrenal axis function.

Finally, research gaps exist with respect to treatment of adolescents and young adults with both alcohol problems and sleep problems. Rigorous treatment studies in this population are needed that go beyond CBT-I to include attention to circadian factors, and with sufficient follow-up periods to better elucidate differential effects on alcohol.

Overall, the existing longitudinal and experimental evidence indicates that a range of sleep/circadian characteristics during adolescence and young adulthood influence risk for the development of alcohol use and/or related problems. Although studies in late adolescents and young adults engaging in regular and/or heavy drinking show that sleep treatment can improve sleep in those individuals, as well as potentially reduce alcohol craving and alcohol-related consequences, no studies in any age group have yet demonstrated that improving sleep reduces drinking behavior. Future research embedding intensive longitudinal studies within prospective research studies is needed to understand the underlying mechanistic pathways from sleep and circadian rhythm to differential alcohol use behaviors and problems as there is evidence that specific sleep indices may relate to certain AUD criteria.¹⁶⁹ Such studies could hold promise for informing treatment for both sleep problems and AUD.