Daily-level associations between sleep hygiene, cannabis use, and sleep health outcomes

Abstract

Objective:

Many use cannabis to promote sleep despite insufficient evidence to recommend cannabis as a sleep-aid. Sleep health may instead be promoted through sleep hygiene, though research among those using cannabis is nascent. This study evaluated daily within-person associations among aggregate and individual sleep hygiene behaviors, cannabis use, and sleep outcomes.

Methods:

Adults with regular cannabis use (N = 85) completed baseline measures and up to 14 daily diaries (n = 1021) on cannabis use and sleep.

Results:

Multilevel models indicated that aggregate sleep hygiene engagement was associated with greater same-night total sleep time, sleep efficiency, and sleep quality. Absence of pre-bedtime arousal was associated with greater same-night total sleep time, sleep efficiency, and sleep quality; only using the bed for sleep/sex was associated with greater same-night sleep efficiency and sleep quality; and greater sleep timing variability was associated with poorer sleep quality. There were significant interactions between cannabis and alcohol use on total sleep time and cannabis use and sleep timing variability on sleep quality. The effect of alcohol on same-night total sleep differed based on cannabis use; total sleep was shorter following alcohol-only use and longer following co-use. The negative association between sleep timing variability and sleep quality was attenuated on cannabis use nights.

Conclusions:

This study highlights associations between sleep hygiene behaviors and more favorable sleep health outcomes among those with frequent cannabis use, supporting sleep hygiene as a first-line approach for this population.

Introduction

Cannabis use is prevalent, with 43.6 million aged 12 or older reporting past month use in the United States (Substance Abuse and Mental Health Service Administration [SAMHSA], 2023). Further, sleep problems are prevalent among those using cannabis (Conroy et al., 2016), and cannabis is commonly used to promote sleep (Doremus et al., 2019). Despite frequent use for this purpose and subjective reports of improved sleep following cannabis use (Goodhines et al., 2019), studies with objective sleep measures (e.g., polysomnography) do not support consistent sleep-related benefits (Gates et al., 2014). Cannabis may produce short-term improvements to total sleep time and sleep onset latency (i.e., time needed to fall asleep); however, these effects abate with chronic use, likely due to tolerance (Babson et al., 2017). Further, sleep disturbances are common during cannabis withdrawal (Gates et al., 2016), and cannabis sleep-aid use is prospectively associated with increased cannabis dependence symptoms (Goodhines et al., 2022). While objective sleep measures are needed to clarify cannabis’ effects on sleep, behavioral sleep interventions often rely on self-reported sleep behaviors and outcomes for assessment and treatment monitoring. Given the prevalence of sleep problems among those using cannabis and lack of evidence to support cannabis as a routine sleep-aid, there is a need to understand whether those using cannabis may benefit from established behavioral sleep strategies, such as sleep hygiene.

Sleep hygiene consists of a variety of behavioral and environmental recommendations to promote sleep health (Irish et al., 2015). Common sleep hygiene behaviors include maintaining a regular sleep schedule, managing stress, optimizing the sleep environment (e.g., light, temperature), avoiding pre-sleep substance use (e.g., caffeine, nicotine, and alcohol), and avoiding daytime naps (Irish et al., 2015). In general, sleep hygiene is associated with improved sleep (Irish et al., 2015) and preliminary studies indicate that cognitive behavioral therapy for insomnia (CBT-I), which includes sleep hygiene components, reduces insomnia symptoms among those with cannabis use (Arnedt et al., 2023; Miller et al., 2022). However, numerous barriers exist that prevent many from accessing CBT-I (see review by Koffel et al., 2018), whereas sleep hygiene is a valuable first-line approach to supporting sleep health across the general population given its wide accessibility. Furthermore, to our knowledge, no research has examined if sleep hygiene is associated with improved sleep among those using cannabis, whether cannabis may modulate these behavioral strategies, or whether these associations are related at the daily-level given day-to-day variability inherent to sleep hygiene implementation and cannabis use. Given sleep hygiene’s accessibility and scalability within clinical settings and public health efforts, examining these associations at the daily level is essential for providing evidence-based sleep recommendations to those using cannabis and for informing whether interventions may benefit from addressing daily fluctuations in sleep hygiene behaviors.

Cannabis use may interact with several sleep hygiene behaviors. For instance, bed stimulus control (i.e., using the bed for sleep/sex only) aims to strengthen the conditioned association between bed and sleepiness (Verreault et al., 2023). However, pre-sleep cannabis use may interfere with this association by providing an alternative cue for sleep. Next, consistent sleep-wake timing facilitates circadian entrainment (Irish et al., 2015); however, cannabis may disrupt circadian rhythm regulation (Babson et al., 2017) and attenuate the benefits of consistent sleep-timing. Both pre-bedtime physiological and cognitive arousal disrupt sleep (Bonnet & Arrand, 2010; Carney et al., 2013), though cannabis is commonly used to regulate negative affect (Livingston et al., 2023) and may modulate the effects of pre-bedtime arousal. Finally, avoiding substance use before bedtime (e.g., alcohol, stimulants) is particularly salient, as cannabis and alcohol co-use may differentially impact sleep compared to use of either substance alone (Graupensperger et al., 2021). Understanding whether cannabis modulates associations between sleep hygiene and sleep health outcomes is critical for determining whether sleep hygiene should be recommended to improve sleep health among those using cannabis, as well as whether modifications to recommendations are warranted among this population.

The present study evaluated whether sleep hygiene was associated with same-night sleep outcomes (i.e., total sleep time, sleep efficiency, and sleep quality) at the daily-level among adults with frequent cannabis use. We evaluated models with aggregate daily sleep hygiene (i.e., the sum of 11 daily sleep hygiene behaviors reported) as the focal predictor to understand if overall implementation was associated with more favorable sleep outcomes. Separate models evaluated five individual sleep hygiene behaviors (i.e., pre-bedtime arousal, bed stimulus control, pre-bedtime alcohol use, pre-bedtime stimulant use [caffeine and/or nicotine], and sleep timing variability) as focal predictors to assess relations with specific behaviors. We hypothesized that greater aggregate sleep hygiene and engagement with recommended individual sleep hygiene behaviors would be associated with more favorable same-night sleep. Given prior research supporting positive associations with self-reported sleep constructs (Goodhines et al., 2019), we hypothesized that cannabis use would be positively associated with same-night sleep outcomes. Finally, we evaluated whether cannabis use moderated associations between sleep hygiene behaviors and sleep outcomes. Given the absence of prior research examining these associations, we treated moderation analyses as exploratory and, therefore, did not specify directional hypotheses.

Method

Participants and Procedures

Participants were recruited from a psychology department subject pool at a United States university in the Mountain West and online postings on various platforms (e.g., social media, listservs). The study recruitment and participation transpired from October 2023 to May 2024. Eligibility was assessed using an online survey based on the following criteria: a) 18 years or older, b) past-month cannabis frequency at ≥1x per week, and c) no plans to cut down or quit using cannabis within the next month. Eligible participants provided informed consent and completed a baseline survey on cannabis use and sleep, followed by once daily diaries on their prior day cannabis use and sleep for 14 consecutive days. Daily diaries were hosted on Qualtrics and participants were sent a daily link via text message and email at a time of their choosing before noon, and a reminder message was sent each afternoon. Participants were instructed to complete each diary prior to midnight on the corresponding day to receive applicable compensation. Participants were offered research participation credit and/or an electronic gift card for completing the study. All procedures were approved by the University of Wyoming Institutional Review Board.

Eighty-nine individuals completed the baseline survey, though four participants discontinued participation prior to initiating the daily surveys. The final sample (N = 85; M_age = 24.28, SD = 7.71; range: 18–60) was predominantly assigned female at birth (75.3%) and college students (77.6%), with the following racial/ethnic identities: Non-Hispanic White (75.3%), Hispanic White (11.8%), non-Hispanic multiracial (7.1%), non-Hispanic Black (2.4%), Hispanic multiracial (2.4%), and non-Hispanic Middle Eastern (1.2%).

Measures

Cannabis Use

Cannabis use was assessed using the Marijuana Use Grid (MUG; Pearson and Marijuana Outcomes Study Team, 2021). At baseline, participants used a daily report grid divided into six 4-hr time blocks (e.g., 8 am to 12 pm) to report on typical weekly use instances and grams/milligrams of cannabis use across four types of cannabis products (i.e., flower, edibles, concentrates, other). In the daily survey, participants reported on whether they used cannabis the prior day (yes/no) and if so, were asked to report on type of cannabis products used (i.e., flower, concentrates, edibles and ‘other’) and were given free-response options to enter quantity and use timing. Cannabis use disorder risk was evaluated at baseline with the Cannabis Use Disorders Identification Test-Revised (CUDIT-R; Adamson et al., 2010). Individual items index frequency of cannabis use and CUD symptoms over the past six months, with individual items varying in terms of response options. Items were summed to create an overall risk score, ranging from 0–31 with higher scores indicating greater risk and scores ≥ 13 representing a cutoff for probable CUD. Internal consistency in the present study was questionable (α = .67). Cannabis negative consequences were assessed at baseline with the Brief Marijuana Consequences Questionnaire (B-MACQ; Simons et al., 2012). Individual items index the presence (yes/no) of 21 cannabis negative consequences over the past-month. Items are summed to create a total cannabis consequences score. Internal consistency in the present study was good (KR-20 = .83).

Alcohol and Other Substance Use

At baseline, participants reported on past-month frequency of alcohol and nicotine product use. An additional item indexed typical alcohol quantity (i.e., number of standard drinks) per drinking occasion. Those endorsing nicotine use were asked to indicate the types of products (e.g., cigarettes, chewing tobacco) used within the past month and were provided a free response box to indicate typical quantity consumed on use days. A modified version of the Caffeine Consumption Questionnaire (CCQ; Landrum, 1992) indexed past-month caffeine use. Participants were asked to select the types of caffeinated beverages/products (e.g., coffee, tea) consumed on a typical day in the past month. For each selected, typical quantity consumed on a typical day was entered a free-response option. The daily surveys indexed prior day (yes/no) alcohol, nicotine, and caffeine use and time of last use for each substance. Additional items assessed the types of caffeinated beverages consumed, caffeine quantity for each beverage (i.e., in ounces), and alcohol use quantity (i.e., in standard drinks).

Sleep Outcomes

Past-month sleep outcomes were assessed at baseline with the Pittsburgh Sleep Quality Index (PSQI; Buysse et al., 1989). The PSQI is an 18-item inventory that produces component scores for sleep duration, sleep disturbance, sleep latency, sleep efficiency, sleep quality, sleep medication use, and daytime dysfunction. The sum of the seven component scores (range 0–3) yields an overall score ranging from zero to 21. Higher scores indicate poorer sleep quality, with scores ≥ 6 distinguishing poor-quality sleepers among college students (Dietch et al., 2016). Internal consistency in the present study was acceptable (α = .77). Daily sleep outcomes were assessed using the Consensus Sleep Diary (Carney et al., 2012). Participants reported their total sleep time, in hours and 15-minute increments, each night. Sleep efficiency was computed by dividing total sleep time by time in bed, multiplied by 100 to yield a percentage. Sleep quality was rated on a scale of 1 = very poor to 5 = very good.

Sleep Attitudes

Attitudes regarding sleep were assessed at baseline using the Charlotte Attitudes Towards Sleep Scale (CATS; Peach & Gaultney, 2017). The CATS consists of 10 items on sleep attitudes and beliefs (e.g., “I am inclined to skip sleep in order to socialize longer”). Individual items are responded to using a seven-point scale (1 = strongly disagree, 7 = strongly agree) and are averaged, with higher scores indicating more favorable attitudes. Internal consistency in the present study was acceptable (α = .74).

Sleep Hygiene

Baseline sleep hygiene behaviors were assessed using the Sleep Hygiene Index (SHI; Mastin et al., 2006). The SHI consists of 13 items on sleep hygiene behaviors (e.g., “I go to bed at different times from day to day”), with participants responding to how true each statement is for them on a five-point scale (0 = never, 4 = always). Items are summed, with higher scores indicating poorer sleep hygiene. The SHI demonstrates superior reliability to predating sleep hygiene measures (α = .66; Mastin et al., 2006). Internal consistency for the present study was questionable (α = 0.68), albeit comparable to original measure development research (Mastin et al., 2006).

For daily assessment of sleep hygiene, SHI items were modified to evaluate the presence (yes/no) of 11 sleep hygiene behaviors on the prior day (see Table 1 for an overview of baseline/daily sleep hygiene behaviors). Several modifications were made to the daily assessment version to increase parsimony and alignment with current research. Three original SHI items were excluded (i.e., getting out of bed and going to bed at different times from day to day and staying in bed longer than one should) due to item phrasing not aligning with daily assessment and to reduce redundancy. Instead, sleep timing variability was evaluated as the deviation from typical sleep midpoint (i.e., midpoint between sleep onset and wake-time) at baseline (collected from PSQI) and sleep midpoint (collected from Consensus Sleep Diary) for each diary day in hours. To allow for dichotomous evaluation, deviations ≥ 60 minutes on a given day compared to typical sleep midpoint were coded as present for sleep-wake variability, aligning with operationalizations of sleep-wake regularity for the general population (Ravyts et al., 2019). Prior day exercise, rather than intense exercise close to bedtime (original SHI item), was assessed because routine exercise is associated with improved sleep and pre-bedtime exercise may not disturb sleep (Chennaoui et al., 2015), contrary to original recommendations. Finally, the daily survey asked about any prior day naps rather than naps > 2 hours (original SHI item) because nap duration has limited effects on nocturnal sleep (Irish et al., 2015). All other original items were retained.

Table 1.

Endorsement of Sleep Hygiene Behaviors

Sleep Hygiene Behavior	Baseline % “Frequently” or “Always”	Daily % “Yes”
I take daytime naps lasting two or more hours	4.7%	N/A
I took a daytime napa	N/A	18.6%
I exercise to the point of sweating within 1 hour of going to bed	9.4%	N/A
I exerciseda	N/A	32.3%
I sleep on an uncomfortable bed (e.g., poor mattress or pillow, too many or not enough blankets)a	11.8%	8.7%
I sleep in an uncomfortable bedroom (e.g., too bright, too stuffy, too hot, too cold, or too noisy)a	14.2%	11.5%
I get out of bed at different times from day to day	20.0%	N/A
I go to bed at different times from day to day	22.3%	N/A
I stay in bed longer than I should two or three times a week	30.6%	N/A
Sleep timing variability > 60 minutesa	N/A	35.0%
Sleep timing variability – continuous M (SD)b	N/A	0.94 (1.00)
Pre-bedtime arousalb	N/A	64.8%
I do important work before bedtime (e.g., pay bills, schedule, or study)a	30.6%	19.8%
I go to bed feeling stressed, angry, upset, or nervousa	54.1%	19.2%
I do something that may wake me up before bedtime (e.g., play video games, use the internet, or clean)a	60.0%	56.3%
I use alcohol, nicotine, or caffeine within 4 hours of going to bed or after going to beda	33.0%	49.3%
Alcohol use within 4 hours of going to bed or after bedb	N/A	21.4%
Stimulant use within 4 hours of going to bed or after bedb	N/A	35.5%
I think, plan, or worry when I am in beda	51.8%	22.5%
I use my bed for things other than sleep or sex (e.g., watch television, read, eat, or study)a,b	54.1%	57.9%

Note. “Baseline” reflects the proportion of participants (N = 85) who responded “frequently” or “always” to the corresponding Sleep Hygiene Index item. “Daily” reflects the proportion of diary days (n = 1,021) the corresponding item was indicated as present. “N/A” denotes items that were not assessed at the corresponding timepoint.

^aItems comprising daily aggregate sleep hygiene score.

^bItems evaluated as individual sleep hygiene behaviors in separate multilevel models.

To evaluate aggregate sleep hygiene engagement, the 11 dichotomous items were summed for each day to create a total score of the number of sleep hygiene behaviors endorsed. This score demonstrated moderate convergent validity with the baseline SHI sum score (r = .54). For main analyses, this score was then reverse-coded to allow for higher values to reflect more adaptive sleep hygiene. For models evaluating individual sleep hygiene behaviors, alcohol and stimulant use (i.e., caffeine and/or nicotine) were specified as separate items given their differential effects on sleep (Irish et al., 2015). Pre-bedtime arousal was assessed dichotomously, such that responding “yes” to any of the following three items constituted presence: going to bed with negative affect, engaging in important work before bedtime, engaging in alerting tasks before bedtime. Sleep timing variability was represented continuously. Bed stimulus control was assessed dichotomously.

Statistical Analyses

Multilevel models examined associations among sleep hygiene, cannabis use, their interaction, and sleep outcomes (i.e., total sleep time, sleep efficiency, sleep quality). Two sets of models were tested: one examining aggregate sleep hygiene (i.e., sum of sleep hygiene behaviors implemented on a given day) and another evaluating relations with individual sleep hygiene behaviors: 1) pre-bedtime arousal (yes/no), 2) bed stimulus control (yes/no), pre-bedtime alcohol use (yes/no), pre-bedtime stimulant use (i.e., caffeine and/or nicotine; yes/no), and sleep timing variability (in hours). Separate models were tested for each predictor and outcome. Level two was defined by participants, whereas level one was defined by study days. Time-variant predictors included dichotomous cannabis use, sleep hygiene engagement, and a dummy variable representing weekend (i.e., Friday/Saturday) to account for potential differences in sleep behaviors by day of the week (Forquer et al., 2008). Cannabis use and sleep hygiene engagement were person-mean and grand-mean centered to disaggregate and control for between person effects in order to isolate within-person effects. An interaction term between person-centered cannabis use and sleep hygiene was incorporated. Time-invariant predictors included sex assigned at birth given differences in cannabis use (Cuttler et al., 2016) and sleep-related attitudes given associations with sleep hygiene behaviors (Peach & Gaultney, 2017).

Unconditional models for each outcome were implemented to compute intraclass correlation coefficients. Next, separate unconditional models with autoregressive covariance structure evaluated the presence of serial correlation. Initial models evaluated an intercept-only structure and random effects were introduced and evaluated iteratively to select the best fitting model (Bell & Jones, 2015). Determinations were made based on a combination of fit statistics (e.g., Akaike information criterion), model convergence, and likelihood ratio tests (LRTs). LRT compares a neighboring model containing one additional parameter, where significant p values (p < .05) indicate improved fit for the larger model. All models were estimated using restricted maximum likelihood (REML), which utilizes all available data and appropriately handles missing observations when data are missing at random (Enders, 2011). To address Type I error from multiple comparisons, the Benjamini-Hochberg procedure was applied to control for false discovery rate (FDR; Benjamini & Hochberg, 1995). Corrections were applied separately within each outcome, as each was conceptualized as a distinct inferential family comprising within-person fixed effects of sleep hygiene and cannabis use across models. Between-person fixed effects and interaction terms were not included in correction, as this study focused on within-person effects and interactions were treated as exploratory given the absence of a priori hypotheses.

Transparency and Openness

We report how we determined our sample size, all data exclusions, and measures included in this study. Additional measures collected at baseline that were determined to not be central to the present analyses are provided as supplemental material. Given the absence of prior research on these relations, which yields challenges in estimating parameters needed for reliable power calculations, no a priori power analysis was conducted (Bacchetti, 2010). Instead, sample size was informed by analogous research detecting within-person cannabis-sleep associations using multilevel modeling of diary data (Goodhines et al., 2019), resulting in a proposed sample of ≥ 80 participants. Analysis code is provided as supplemental material. Data and non-proprietary materials for this study are available upon reasonable request by emailing the corresponding author. Data were analyzed in SPSS (version 29.0) using the MIXED procedure. This study was not preregistered. Strengthening the Reporting of Observational Studies in Epidemiology (STROBE) was used as the reporting guideline (Von Elm et al. 2007).

Results

Descriptives

At baseline, participants’ average past-month cannabis use frequency was 20.09 days (SD = 9.55). Average cannabis use disorder risk score (CUDIT-R) was 10.92 (SD = 5.07) and average number of past-month negative cannabis consequences (B-MACQ) was 4.20 (SD = 4.83). Past-month sleep characteristics included average total sleep time of 7.43 hours (SD = 1.06), average sleep efficiency of 87.74% (SD = 9.87), average sleep quality of “fairly good” (M = 1.12, SD = 0.66), and average PSQI global score of 6.04 (SD = 3.02). On average, participants reported favorable sleep attitudes (CATS M = 5.26, SD = 0.74) consistent with prior college samples (M = 4.76, SD = 0.80; Peach et al., 2018) and reported lower engagement in poor sleep hygiene behaviors over the past-month (SHI M = 21.18, SD = 6.13) when compared to other young adult samples (SHI M = 34.66, SD = 6.60; Mastin et al., 2006). Finally, nine participants (10.6%) self-reported a current sleep disorder.

Diary completion rate was 85.8% (n = 1,021) and participants completed 12.01 (SD = 3.58) of 14 possible daily surveys, on average. Percentage of missing days was not significantly correlated with any primary variable at the between-person level: cannabis use (r = .01, p = .928), aggregate sleep hygiene (r = −.07, p = .518), total sleep time (r = −.18, p = .095), sleep efficiency (r = −.21, p = .050), and sleep quality (r = .21, p = .053). Cannabis use was endorsed on 70.1% of days (n = 716). Of cannabis use days, the following cannabis products were used: flower (45.3%), edibles (19.1%), and concentrates (50.0%). Average time of last cannabis use per day was 09:12 PM (SD = 2.61 hours). Across all days, average total sleep time was 7.73 hours (SD = 1.54) which was greater than normative data for young adults (6.84 hours; Boulos et al., 2019), average sleep efficiency was 82.63% (SD = 12.02), and average sleep quality was “fair” (M = 3.64, SD = 0.91). Participants reported implementing an average of 7.36 (SD = 1.85) adaptive sleep hygiene behaviors per day. Table 1 depicts the frequencies of sleep hygiene behaviors.

Multilevel Models

Unconditional models suggested that 25%, 30%, and 17% of the observed total variability in total sleep time (ICC = 0.25), sleep efficiency (ICC = 0.30) and sleep quality (ICC = 0.17) was due to between-person differences, respectively. In turn, this indicated substantial within-person variability for each outcome. Separate unconditional models suggested no serial correlation was present for total sleep time (ρ = −0.01, p = .875) and sleep efficiency (ρ = 0.05, p = .168), but serial correlation was indicated for sleep quality (ρ = 0.08, p = .029). Thus, models evaluating sleep quality were specified with autoregressive one covariance structure. Aggregate sleep hygiene model results are depicted in Table 2 and individual behavior models are depicted in Tables 3–5.

Table 2.

Multilevel Models of Aggregate Daily Sleep Hygiene Behaviors and Cannabis Use Predicting Sleep Outcomes

	Total Sleep Time	Sleep Efficiency	Sleep Quality
	Estimate (SE)	Estimate (SE)	Estimate (SE)
Fixed Effects
Within-Person
Weekend	0.10 (0.08)	0.83 (0.71)	0.13 (0.06)*
Cannabis Use	0.14 (0.13)	2.15 (1.00)*	0.28 (0.08)***
Aggregate Sleep Hygiene	0.14 (0.04)**	1.58 (0.23)***	0.14 (0.02)***
Sleep Hygiene x Cannabis Use	−0.09 (0.09)	−0.04 (0.70)	−0.03 (0.06)
Between-Person
Intercept	5.85 (0.70)***	86.93 (5.81)***	3.24 (0.34)***
Sex	−0.11 (0.23)	−1.47 (1.91)	−0.11 (0.11)
Sleep Attitudes	0.36 (0.14)**	−0.68 (1.14)	0.09 (0.07)
Cannabis Use	−0.26 (0.27)	−1.07 (2.28)	0.04 (0.13)
Aggregate Sleep Hygiene	−0.01 (0.08)	1.77 (0.66)**	0.13 (0.04)***
Random Effects
Residual	1.52 (0.08)***	96.56 (4.47)***	---
Intercept1	0.62 (0.13)***	40.69 (7.91)***	0.12 (0.03)***
Sleep Hygiene2	0.05 (0.02)*	---	0.01 (0.01)*
Weekend3	0.56 (0.21)**	---	0.04 (0.05)
Covariance1,2	0.01 (0.04)	---	−0.01 (0.01)
Covariance1,3	−0.17 (0.12)	---	−0.02 (0.03)
Covariance,2,3	−0.07 (0.05)	---	−0.02 (0.01)
AR1 Covariance	---	---	0.61 (0.03)***
Autocorrelation	---	---	0.07 (0.04)

Note. Sex was coded as 0 = male and 1 = female. The aggregate sleep hygiene score reflects the sum of each sleep hygiene behavior. AR = autoregressive. Total sleep time was evaluated in hours. Sleep efficiency represents total sleep time divided by time in bed, producing a percentage from 0–100. Sleep quality was rated on a scale from 1 = very poor to 5 = very good. Within-person fixed effects of sleep hygiene and cannabis use were subject to FDR correction;

^†denotes values that were significant prior to but did not survive FDR correction. Between-person effects, interactions, covariates, and random effects were not subject to correction.

^*p < .05

^**p < .01

^***p < .001.

Table 3.

Multilevel Models of Individual Daily Sleep Hygiene Behaviors and Cannabis Use Predicting Total Sleep Time

	No Pre-Bedtime Arousal	Bed Stimulus Control	Alcohol Use	Stimulant Use	Sleep Timing Variability
	Estimate (SE)	Estimate (SE)	Estimate (SE)	Estimate (SE)	Estimate (SE)
Fixed Effects
Within-Person
Weekend	0.03 (0.13)	0.04 (0.13)	0.05 (0.13)	0.06 (0.13)	0.16 (0.12)
Cannabis Use	0.19 (0.13)	0.19 (0.13)	0.16 (0.13)	0.19 (0.13)	0.11 (0.13)
Sleep Hygiene Behavior	0.29 (0.11)**	0.26 (0.11)†	−0.01 (0.13)	−0.07 (0.14)	−0.20 (0.08)†
Interaction	−0.12 (0.35)	0.10 (0.39)	1.06 (0.42)*	0.36 (0.51)	0.16 (0.15)
Between-Person
Intercept	6.14 (0.69)***	5.93 (0.67)***	5.89 (0.66)***	5.94 (0.66)***	5.95 (0.67)***
Sex	−0.09 (0.22)	−0.13 (0.23)	−0.11 (0.22)	−0.12 (0.22)	−0.12 (0.22)
Sleep Attitudes	0.31 (0.13)*	0.36 (0.13)**	0.36 (0.13)**	0.35 (0.13)**	0.35 (0.13)**
Cannabis Use	−0.27 (0.27)	−0.26 (0.28)	−0.28 (0.27)	−0.31 (0.28)	−0.30 (0.27)
Sleep Hygiene Behavior	0.38 (0.35)	−0.08 (0.31)	0.42 (0.36)	0.17 (0.26)	−0.15 (0.17)
Random Effects
Residual	1.62 (0.08)***	1.63 (0.08)***	1.62 (0.08)***	1.64 (0.08)***	1.42 (0.07)***
Intercept1	0.60 (0.12)***	0.61 (0.13)***	0.61 (0.13)***	0.60 (0.13)***	0.64 (0.13)***
Weekend2	0.72 (0.24)**	0.69 (0.23)**	0.69 (0.43)**	0.67 (0.23)**	0.38 (0.18)*
Sleep Hygiene Behavior 3	---		---		0.23 (0.07)**
Covariance1,2	−0.19 (0.13)	−0.18 (0.13)	−0.19 (0.13)	−0.17 (0.13)	0.38 (0.18)*
Covariance1,3	---		---		0.07 (0.08)
Covariance2,3	---		---		0.14 (0.08)

Note. Separate models were tested for each sleep hygiene behavior listed on the top row. Interaction = person-centered sleep hygiene behavior X person-centered cannabis use. Weekend was specified as Friday and Saturday. Sex was coded as 0 = male and 1 = female. All predictors were assessed dichotomously, except for sleep attitudes and sleep timing variability. Cannabis use and sleep behavior were grand-mean centered (between-person) and person-mean centered (within-person). Total sleep time was evaluated in hours. Within-person fixed effects of sleep hygiene and cannabis use were subject to FDR correction;

^†denotes values that were significant prior to but did not survive FDR correction. Between-person effects, interactions, covariates, and random effects were not subject to correction.

^*p < .05

^**p < .01

^***p < .001.

Table 5.

Multilevel Models of Individual Daily Sleep Hygiene Behaviors and Cannabis Use Predicting Sleep Quality

	No Pre-Bedtime Arousal	Bed Stimulus Control	Alcohol Use	Stimulant Use	Sleep Timing Variability
	Estimate (SE)	Estimate (SE)	Estimate (SE)	Estimate (SE)	Estimate (SE)
Fixed Effects
Within-Person
Weekend	0.08 (0.06)	0.08 (0.06)	0.12 (0.06)*	0.10 (0.06)	0.16 (0.06)*
Cannabis Use	0.32 (0.08)***	0.31 (0.08)***	0.31 (0.08)***	0.32 (0.08)***	0.28 (0.08)***
Sleep Hygiene Behavior	0.20 (0.07)**	0.29 (0.07)***	−0.15 (0.08)	−0.09 (0.09)	−0.17 (0.04)***
Interaction	−0.28 (0.22)	−0.03 (0.24)	0.43 (0.26)	−0.10 (0.32)	0.24 (0.09)*
Between-Person
Intercept	2.96 (0.36)***	2.95 (0.53)***	2.88 (0.34)***	2.90 (0.35)***	2.71 (0.35)***
Sex	−0.20 (0.12)	−0.17 (0.12)	−0.21 (0.11)	−0.21 (0.12)	−0.23 (0.12)
Sleep Attitudes	0.16 (0.07)*	0.15 (0.07)*	0.17 (0.07)*	0.17 (0.07)*	0.20 (0.07)**
Cannabis Use	0.06 (0.14)	0.03 (0.14)	0.09 (0.14)	0.05 (0.15)	0.04 (0.14)
Sleep Hygiene Behavior	0.12 (0.19)	−0.20 (0.16)	−0.27 (0.19)	−0.01 (0.13)	−0.02 (0.08)
Random Effects
Intercept1	0.12 (0.03)***	0.12 (0.03)***	0.12 (0.03)***	0.13 (0.03)***	0.13 (0.03)***
Sleep Hygiene Behavior2	---	---	---	---	0.04 (0.02)*
Covariance1,2	---	---	---	---	0.04 (0.02)
AR1 Covariance	0.68 (0.03)***	0.67 (0.03)***	0.68 (0.03)***	0.68 (0.03)***	0.64 (0.03)***
Autocorrelation	0.10 (0.04)**	0.08 (0.04)*	0.09 (0.04)*	0.08 (0.04)*	0.09 (0.04)*

Note. Separate models were tested for each sleep hygiene behavior listed on the top row. Interaction = person-centered sleep hygiene behavior X person-centered cannabis use. Weekend was specified as Friday and Saturday. Sex was coded as 0 = male and 1 = female. All predictors were assessed dichotomously, except for sleep attitudes and sleep timing variability. Cannabis use and sleep behavior were grand-mean centered (between-person) and person-mean centered (within-person). AR = autoregressive. Sleep quality was rated on a scale of 1 = very poor to 5 = very good. Within-person fixed effects of sleep hygiene and cannabis use were subject to FDR correction;

^†denotes values that were significant prior to but did not survive FDR correction. Between-person effects, interactions, covariates, and random effects were not subject to correction.

^*p < .05

^**p < .01

^***p < .001.

Aggregate Sleep Hygiene

Total Sleep Time.

At the within-person level, greater aggregate sleep hygiene use was associated with longer same-night total sleep time (γ = 0.14, p < .001), whereas cannabis use was not associated with same night total sleep time (γ = 0.14, p = .294). The interaction between within-person aggregate sleep hygiene and cannabis use was not significant (γ = −0.09, p = .329).

Sleep Efficiency.

At the within-person level, aggregate sleep hygiene use (γ = 1.58, p < .001) and cannabis use (γ = 2.15, p = .032) were associated with greater same-night sleep efficiency. At the between-person level, higher average aggregate sleep hygiene use was associated with greater sleep efficiency (γ = 1.77, p = .009). The interaction between within-person aggregate sleep hygiene and cannabis use was not significant (γ= −0.04, p = .956).

Sleep Quality.

At the within-person level, aggregate sleep hygiene use (γ = 0.14, p < .001) and cannabis use were associated with greater same-night sleep quality (γ = 0.28, p < .001). At the between-person level, higher average aggregate sleep hygiene was associated with greater sleep quality (γ = 0.13, p < .001). The interaction between within-person aggregate sleep hygiene and cannabis use was not significant (γ = −0.03, p = .560).

Individual Sleep Hygiene Behaviors

Total Sleep Time.

At the within-person level, absence of pre-bedtime arousal was associated with longer same-night total sleep time (γ = 0.29, p = .006). Implementation of bed stimulus control (γ = 0.26, p = .021; not significant following FDR correction), sleep timing variability (γ = −0.20, p = .017; not significant following FDR correction), alcohol use (γ = −0.01, p = .930), and stimulant use (γ = −0.07, p= .639) were not associated with same-night total sleep. Further, cannabis use was not associated with same-night total sleep time in any model. However, there was a significant interaction between alcohol and cannabis use on same-night total sleep time (γ = 1.06, p = .012). Model visualization suggested that the effect of alcohol on total sleep time depended on whether cannabis was used, such that same-night sleep was shorter following alcohol-only use and longer following co-use (see Figure 1).

Figure 1: Interactions between cannabis use and sleep hygiene on sleep outcomes.

Note. Depicts plots of significant interactions between person-centered cannabis use and sleep hygiene behaviors on sleep outcomes, with values depicted at +/− 1 SD from the mean. Total sleep time represented in hours (hr). Sleep quality was rated on a scale of 1 = very poor to 5 = very good.

Sleep Efficiency.

At the within-person level, absence of pre-bedtime arousal (γ = 2.87, p = .006), and implementation of bed stimulus control (γ = 5.90, p < .001) were associated with greater same-night sleep efficiency. Alcohol use (γ = 2.05, p = .044; not significant following FDR correction), stimulant use (γ = 0.34, p = .748), and sleep timing variability (γ = −1.01, p = .142) were not associated with same-night sleep efficiency. At the between-person level, higher average bed stimulus control was associated with greater sleep efficiency (γ = 8.52, p < .001). Across all models, cannabis use was associated with greater same-night sleep efficiency. There were no significant interactions between sleep hygiene behaviors and cannabis use on same-night sleep efficiency.

Sleep Quality.

At the within-person level, absence of pre-bedtime arousal (γ = 0.20, p = .002) and implementation of bed stimulus control (γ = 0.29, p < .001) were associated with greater same-night sleep quality, while greater sleep timing variability was associated with poorer same-night sleep quality (γ = −0.17, p < .001). Across all models, cannabis use was associated with greater same-night sleep quality. There was a significant interaction between sleep timing variability and cannabis use on same-night sleep quality (γ = 0.24, p = .010). Model visualization suggested that the negative association between sleep timing variability and sleep quality was attenuated on cannabis use nights (see Figure 1).

Discussion

Although many report using cannabis for sleep promotion (Doremus et al., 2019), there is insufficient evidence to recommend cannabis as an objective sleep-aid (Gates et al., 2014). Instead, sleep hygiene may promote sleep health among those with cannabis use, though research in this area is nascent. To address this gap, the present study examined daily associations among sleep hygiene, cannabis use, and same-night sleep outcomes. Greater aggregate sleep hygiene use was associated with more favorable same-night sleep outcomes. Evaluation of individual behaviors demonstrated similar relations, with variability depending on the strategy and sleep outcome. Cannabis use was positively associated with self-reported same-night sleep efficiency and sleep quality, while not associated with total sleep time. Only two significant interactions between sleep hygiene and cannabis use on sleep outcomes emerged. Overall, findings support the benefits of sleep hygiene for sleep health among those with cannabis use and suggest that these associations are robust within the context of cannabis use.

In a sample of adults with frequent cannabis use, greater within-person aggregate sleep hygiene was associated with greater same-night total sleep time, efficiency, and quality. Further, greater between-person aggregate sleep hygiene was associated with greater sleep efficiency and quality. Absence of pre-bedtime arousal was positively associated with each sleep outcome, implementing bed stimulus control was positively associated with sleep efficiency and quality, and greater sleep timing variability was associated with lower same-night sleep quality. These behaviors may be particularly salient to sleep health as only using the bed for sleep and sex and consistent sleep timing are common stimulus control practices that are integral to CBT-I (Furukawa et al., 2024). Still, the mechanisms of action underlying stimulus control interventions have been debated and require future study (Verreault et al., 2024). Although several techniques to reduce arousal are outlined within behavioral sleep treatments (e.g., cognitive therapy, relaxation; Perlis et al., 2005), this study did not evaluate mechanisms underlying pre-bedtime arousal. Future research should examine strategies implemented among those using cannabis, especially as cannabis may be used to reduce negative affect (Livingston et al., 2023). Taken together, implementation of greater nightly number of sleep hygiene behaviors and several specific strategies are associated with greater sleep health. This aligns with prior research that has demonstrated the benefits of sleep hygiene (Irish et al., 2015), while also extending these findings to those with frequent cannabis use and highlighting the relevance of day-to-day fluctuations in sleep hygiene behaviors. Beyond encouraging overall sleep hygiene implementation, interventions may benefit from incorporating strategies to help individuals identify and address daily barriers to sleep hygiene adherence.

Alcohol use was not associated with same-night sleep efficiency following FDR correction. While a positive association between alcohol and self-reported sleep efficiency has been reported previously (Miller et al., 2021), this may reflect reductions to sleep onset latency (Ebrahim et al., 2013) or perceptions that sleep is more consolidated, as these relations are not evidenced when objective sleep measures are employed (Miller et al., 2021). Nonetheless, it is well-documented that alcohol overall is detrimental to sleep (He et al., 2019). Finally, stimulant use (i.e., caffeine, nicotine) was not associated with any same-night sleep outcome. Although these substances have been associated with sleep-disrupting effects in prior research (Drake et al., 2013; Jaehne et al., 2009), much of this evidence comes from experimental studies that administer large doses, and tolerance to caffeine’s sleep disrupting effects may develop among those with regular intake (Irish et al., 2015).

Cannabis use was associated with greater self-reported same-night sleep efficiency and quality, while not associated with total sleep time. Notably, the positive association with quality in the present investigation appeared to be more robust than the association with sleep efficiency. This aligns with prior reviews reporting mixed relations between cannabis use and total sleep time and efficiency, whereas more consistent positive associations with sleep quality (Gates et al., 2014). These mixed relations may be influenced by use of objective versus subjective sleep measures, with the latter often overestimating total sleep time and efficiency when compared to gold-standard sleep measures (i.e., polysomnography; Lehrer et al., 2022). Positive beliefs about cannabis’ effects on sleep (Livingston et al., in press) may also contribute to more robust relations with sleep quality, given the subjectivity of this domain. Studies with objective sleep measures are needed to clarify cannabis’ effects on sleep.

In exploratory analyses, only two significant interactions between sleep hygiene and cannabis use on sleep outcomes emerged. First, the negative association between sleep timing variability and sleep quality was attenuated by same-day cannabis use, such that sleep quality was lower on nights with higher variability but to a lesser extent on cannabis use days. It is possible that individuals with regular cannabis use may experience poorer sleep quality on non-use days due to withdrawal Gates et al., 2016) and/or expectancy effects (Livingston et al., in press), with such effects exacerbated within the context of sleep disruption stemming from variable sleep timing. Next, there was a significant interaction between alcohol and cannabis use on total sleep time, such that total sleep time was lower on alcohol-only nights and higher following co-use. Similar findings have been reported in other daily diary research (Graupensperger et al., 2021), supporting the possibility that cannabis may temper alcohol-related arousals. However, this does not necessarily support co-use as beneficial to sleep, and rather may highlight the interplay between alcohol’s detrimental effects on sleep and perceptions that cannabis improves sleep. Importantly, the lack of interactions overall indicates that engagement in adaptive sleep hygiene behaviors is robustly associated with more favorable sleep health outcomes and is not disrupted by cannabis use. Given insufficient evidence to support cannabis as a routine sleep-aid, it is unlikely that its use provides incremental sleep-related benefits above and beyond healthy sleep practices. Accordingly, sleep hygiene recommendations should be considered as a first-line approach to promote sleep health among those using cannabis. While sleep hygiene recommendations should be personalized to maximize sleep improvements (Irish et al., 2015), existing strategies appear effective among those using cannabis and may not necessitate substantial modifications.

The present study is not without limitations. First, while both college and non-college students across a range of ages were included, the sample predominantly consisted of healthy (i.e., average total sleep time was greater than normative young adult data, average CUD risk was below the cutoff) young adult white female college students. Though results may not generalize to those with marginalized identities, older adults, and those with clinically significant sleep disturbances and cannabis use, college students represent a particularly relevant population for this research. Namely, young adults are disproportionately impacted by sleep problems (McArdle et al., 2020) and cannabis use (SAMHSA, 2023), and navigate unique environmental and social barriers to sleep hygiene implementation (e.g., irregular schedules, shared living spaces, limited control over sleep environment). Future studies should examine whether these findings extend to more diverse samples. Next, this investigation was limited to self-report measures of sleep and cannabis use. As noted, cannabis-sleep relations may vary based on use of objective versus subjective sleep measures (Gates et al., 2014). Studies implementing both types of measures are needed to comprehensively evaluate cannabis’ effects on sleep. Similarly, we examined cannabis use dichotomously. Although there have been proposals to standardize cannabis quantity (Freeman & Lorenzetti, 2019), approaches have yet to be developed that allow for standardization across cannabis product types. Examining relations between substance dose and sleep outcomes remains an important area for future study. Finally, our models emphasized a subset of sleep outcomes and sleep hygiene behaviors; however, other sleep health outcomes (e.g., daytime alertness, timing) and behaviors warrant investigation.

The present research indicates that greater daily and overall implementation of adaptive sleep hygiene behaviors is associated with longer, more efficient, and higher quality same-night sleep among those with cannabis use, and that these effects are robust within the context of cannabis use. Absence of pre-bedtime arousal and stimulus control behaviors (i.e., only using the bed for sleep/sex, consistent sleep timing) were notable predictors of greater sleep health, highlighting the salience of these particular strategies. Although additional research is needed to clarify reciprocal cannabis-sleep relations, these findings suggest that existing sleep hygiene recommendations may be a viable approach to promoting sleep health within clinical and public health efforts among those using cannabis.

Table 4.

Multilevel Models of Individual Daily Sleep Hygiene Behaviors and Cannabis Use Predicting Sleep Efficiency

	No Pre-Bedtime Arousal	Bed Stimulus Control	Alcohol Use	Stimulant Use	Sleep Timing Variability
	Estimate (SE)	Estimate (SE)	Estimate (SE)	Estimate (SE)	Estimate (SE)
Fixed Effects
Within-Person
Weekend	0.17 (0.72)	0.08 (0.71)	0.17 (0.75)	0.57 (0.73)	1.64 (0.74)*
Cannabis Use	2.29 (1.01)*	2.21 (1.00)*	2.18 (1.03)*	2.35 (1.03)*	2.29 (0.99)*
Sleep Hygiene Behavior	2.87 (1.00)**	5.90 (1.05)***	2.05 (1.01)†	0.34 (1.06)	−1.01 (0.68)
Interaction	2.82 (2.66)	−0.39 (3.66)	2.86 (3.21)	−0.22 (3.97)	0.36 (1.16)
Between-Person
Intercept	85.93 (5.61)***	85.90 (5.09)***	82.05 (5.74)***	82.13 (5.74)***	80.85 (5.83)***
Sex	−2.68 (1.81)	−0.93 (1.74)	−2.90 (1.91)	−2.81 (1.92)	−3.06 (1.95)
Sleep Attitudes	−0.28 (1.08)	−0.55 (0.99)	0.47 (0.13)	0.43 (1.10)	0.65 (1.11)
Cannabis Use	−1.34 (2.21)	−1.58 (2.11)	−0.94 (2.39)	−0.86 (2.43)	−1.39 (2.39)
Sleep Hygiene Behavior	4.75 (2.93)	8.52 (2.41)***	−0.87 (3.14)	−0.94 (2.22)	−0.15 (1.46)
Random Effects
Residual	96.51 (4.61)***	91.40 (4.40)***	100.90 (4.67)***	101.45 (4.70)***	91.07 (4.38)***
Intercept1	42.43 (8.28)***	39.39 (7.68)***	44.27 (8.60)***	44.15 (8.59)***	45.23 (8.66)***
Sleep Hygiene Behavior2	22.95 (10.97)*	24.33 (12.59)			17.54 (5.31)***
Interaction 3	---	146.34 (103.76)	---
Covariance1,2	−18.13 (7.11)*	−21.00 (8.22)*			4.09 (5.14)
Covariance1,3	---	-24.44 (26.71)	---
Covariance2,3	---	-19.95 (29.08)	---

Note. Separate models were tested for each sleep hygiene behavior listed on the top row. Interaction = person-centered sleep hygiene behavior X person-centered cannabis use. Weekend was specified as Friday and Saturday. Sex was coded as 0 = male and 1 = female. All predictors were assessed dichotomously, except for sleep attitudes and sleep timing variability. Cannabis use and sleep behavior were grand-mean centered (between-person) and person-mean centered (within-person). Total sleep time was evaluated in hours. Within-person fixed effects of sleep hygiene and cannabis use were subject to FDR correction;

^†denotes values that were significant prior to but did not survive FDR correction. Between-person effects, interactions, covariates, and random effects were not subject to correction.

^*p < .05

^**p < .01

^***p < .001.

Public Health Significance.

This study indicates that greater overall implementation of sleep hygiene and use of specific strategies (i.e., avoiding pre-bedtime arousal, only using the bed for sleep and sex, and consistent sleep timing) is related to more favorable same-night sleep in adults with regular cannabis use. Given limited evidence supporting cannabis as a routine sleep-aid, sleep hygiene should be considered as a first-line approach to promote sleep health within this population.