Sleep disturbances negatively affect many health and functioning outcomes [1]. In adolescence and emerging adulthood, sleeping fewer than 7 h/night [2] or otherwise experiencing poor sleep health [3] is associated with anxiety/depression [4], illnesses [5], and worse academic outcomes [6]. Biological changes in circadian timing, early school start times, and social pressures are broadly associated with sleep problems in high school and college students [7]. Marginalized and minoritized students are at elevated risk for experiencing chronic sleep disturbances (sleep disparities) [8], indicating a need for targeted sleep interventions.
Across the lifespan, females experience sleep disturbances such as being unable to fall asleep or stay asleep disproportionately relative to males [9]. Many factors contribute to gender-based disparities in sleep including affective disorders and school-related stressors [9][10][11]. Gender-based discrimination has also been associated with shorter sleep durations [12]. In addition, balancing home responsibilities and work responsibilities may increase gender-based disparities in sleep [11]. Specifically, having high work demands, lower social support, and being a single parent are associated with poorer sleep quality in females [11].
Racial/ethnic minoritized students who are under-represented (URM) in higher education also experience sleep disturbances at a disproportionate rate (e.g., Black, Hispanic, American Indian/Native American) [13]. URM individuals are more likely to experience excessive daytime sleepiness, shorter sleep duration, obstructive sleep apnea, and shallower sleep stages [14]. Evidence indicates that sleep disparities may contribute to health disparities and academic achievement gaps [10]. Reasons why sleep disparities are present in URM individuals include experienced and anticipated discrimination, sociocultural influences, and environmental factors [15]. Experiences with discrimination and unfair treatment induce a physiological stress response that negatively affects sleep, health, and academic outcomes [16]. Furthermore, URM individuals have less access to healthy food/nutrition [17] and healthcare [18], and they are more likely to be exposed to environments with excessive noise, air pollution, or inopportune light (e.g., artificial light at night) [19].
Psychological and educational interventions (hereafter "psychoeducational interventions") may help to mitigate the effects of poor sleep among adolescents and emerging adults. For the purpose of the current work, our definition of "psychoeducational" is broad so as to capture the range of sleep programs attempted in educational settings. Generally speaking, psychoeducational interventions aim to increase sleep knowledge, reduce stress, or provide active cognitive and behavioral strategies so as to improve sleep quality. For example, interventions may expose students to the science and biology of sleep to increase sleep knowledge (e.g., sleep stages, sleep deprivation effects), and then teach specific strategies for improving their sleep quality. Such strategies have included a focus on sleep hygiene, relaxation/ mindfulness-based approaches, and cognitive behavioral therapy for insomnia (CBT-I). Sleep hygiene programs typically instruct students to maintain a consistent bed and wake time, avoid caffeine and other stimulants, have relaxing activities before bedtime, and create a good sleep environment. Relaxation/mindfulness-based approaches aim to reduce stress. Programs that draw upon CBT-I strategies typically provide instruction on stimulus control (e.g., behaviors that strengthen the bed as a cue for sleep and weaken it as a cue for wakefulness), time-inbed restriction, sleep hygiene, and relaxation techniques (note that these programs were providing strategy instruction rather than CBT-I clinical treatment). Previous work suggests that CBT-I-based interventions yield the best improvements in sleep in student groups [20].
Most of the psychoeducational sleep programs discussed thus far improved sleep knowledge, sleep duration, or sleep quality in students. However, it is unclear if these interventions are equally effective among minoritized student groups. A study by Tavernier and Adam [21], for example, showed that a text-message based sleep intervention only improved sleep duration for non-Hispanic White students but showed no benefit in racial/ethnic minority students. One interpretation was that a person can know the steps to obtain better sleep but experience socioeconomic or neighborhood barriers that prevent them from obtaining better sleep (e.g., lack of quiet space, lack of sufficient personal space for implementing stimulus control, unsafe neighborhood). Another possible interpretation is that there are cultural differences in beliefs about the importance of sleep (i.e., whether one believes that sleep is related to one's health) [22]; by this view, if the program is not culturally tailored in an effective manner, then it may have no impact on the individual.
Despite the issues discussed above, psychoeducational sleep programs have rarely investigated whether the intervention is similarly effective across demographic subgroups; some studies have not reported the demographic composition of the student sample (see Table S1 for details). To begin to address these gaps in the literature, we contacted the corresponding authors of published psychoeducational sleep programs in students with the request to provide effect sizes separated by gender and race/ethnicity groups. We then conducted a meta-analysis to determine whether the interventions were similarly effective across two gender groups (cis males versus cis females) and two racial/ethnic general classifications (non-URM versus URM students); note that we relied on general groups/classifications because the sample sizes in this literature would not be sufficiently large to reliably distinguish outcomes across all minoritized gender groups (e.g., non-binary), racial/ ethnic subgroups, and intersections [23](Table S1). To complement the meta-analytic approach, we systematically rated the content of each sleep program using a standardized rubric (See Table S2) to identify areas in which future interventions might be improved.
Across our meta-analysis and structured content review, we considered three hypotheses. First, the "null" hypothesis was that psychoeducational sleep programs would be similarly effective across groups; this outcome could occur, for example, if sleep health is of sufficient interest across demographic groups that it encourages people to form intentions to change nighttime behaviors [24]. Second, according to the view that programs should be individually or culturally tailored to be maximally effective [18], psychoeducational sleep programs that are not culturally tailored will lead to less of a benefit among minoritized groups. Third, according to the view that socioeconomic and neighborhood barriers prevent some students from obtaining healthy sleep [19], psychoeducational sleep programs should produce similar improvements in sleep knowledge across groups but have a differential impact on sleep quality. Investigating these possibilities will contribute to understanding of intervention strategies to address sleep disparities in student populations.
Article selection began with examination of two systematic review papers on sleep interventions in students [20,25], both of which included comprehensive searches of databases (Medline, PsycInfo, EBSCOhost). We identified all the intervention studies included in the two reviews and then forward/backward tracked citations from those papers to identify additional interventions through January 2022. We then repeated this forward/backward tracking process on the newly identified papers. Inclusion criteria were enrolling students as participants (grade school, college, or graduate school), reporting empirical data on sleep knowledge, sleep quality, or sleep duration, and having a psychoeducational sleep intervention design. Baseline sleep status was not a requirement for inclusion in the meta-analysis (5.5 % of studies set insomnia/sleep disorders as inclusion criteria, 22.2 % set insomnia/sleep disorders as exclusion criteria, and 72.2 % did not specify). Studies were excluded if they were not based in the United States to limit sources of variability that would emerge from international differences in educational systems, gender inequality, and racial/ethnic compositions across regions. Thirty-six papers met inclusion criteria [21, (see Table S1 and Fig. S1).
We emailed the authors of each study to request sharing, beginning with the corresponding author and repeating twice. Specifically, we requested the raw data file or pre-and post-intervention data for calculating effect sizes for sleep knowledge, sleep duration, and sleep quality measures, separated by a) males and females, and b) URM and non-URM participants. We followed the National Science Foundation's [61] definition of non-URM as students who identified as non-Hispanic White or Asian and URM as students who identified as Hispanic or Latinx, Black or African American, American Indian or Native Alaskan, or multiracial.
We used pre/post-intervention outcomes to compute Cohen's d for the sleep intervention effect size for each group in each study. Cohen's d represents the standardized difference between the intervention and control groups. We converted Cohen's d values to Hedges' g values, which are less biased values when the sample sizes are small [62]. Next, we compared the overall Hedges' g values across gender and racial/ethnic groups to determine whether the sleep intervention was similarly effective for different groups.
Analyses were conducting using the Comprehensive Meta-Analysis software (version 3, BioStat Inc.). We used random-effects models by assuming the true effects differed across interventions and the final estimated effect size was the average of the true effect sizes [62]. Since studies with smaller variances yield more precise estimates, we assigned weights to studies by the inverse of the variances. To quantify heterogeneity of effect sizes we used Q statistic, which is computed as the sum of squared deviations of individual effect sizes from the overall effect size, weighted by the inverse of the study variances. When there was significant heterogeneity, we conducted moderation analyses to investigate whether study or participant characteristics moderated the effect sizes (mean age of sample, sample size, study duration, journal impact factor, and intervention type). To determine whether effect sizes were similar across groups, we conducted a series of subgroup meta-regressions. Last, we used the trim-and-fill non-parametric test to assess and adjust for publication bias [63]. The trim-and-fill test statistically estimates the number of studies that may exist but were not included in the meta-analysis due to publication bias. The adjusted overall effect sizes take the estimated unpublished studies into account. All data are available at Open Science Framework (osf.io/c3ya8).
In addition to the meta-analytic approach, we systematically examined the content of sleep education programs, using a standardized rubric (See Table S2). In our literature review (Fig. S1 and Table S3), we identified 21 distinct psychoeducational sleep programs and requested the original intervention materials from study authors [21,26-28,30-39, 41,64-69]. The full intervention materials were shared by study co-authors or available online for 11 studies (50 %); partial intervention materials were available for the other 10 studies (see Table S3). Each set of intervention materials/descriptions was evaluated by two raters (raters included: AEN, CG, CL, AP). After independent ratings, all raters met to discuss and resolve any disagreements. Table S2 shows the standardized rubric, which had three general categories-student-tailoring, sensitivity to diverse groups, and structural barriers to implementing the intervention-which are detailed below.
Content ratings began with evaluating the degree to which interventions were clearly designed for student groups. Raters evaluated if the intervention was tailored to individuals, which is recommended for behavioral change [70] and is typical in goal attainment scaling interventions [71]. One example of an individually-tailored intervention is to identify each student's person-specific sleep problems/goals (e.g., difficulty initiating sleep versus difficulty maintaining sleep), and then tailor recommendations to that student's needs/goals. In addition, raters identified whether the intervention included a technological component (e.g., smart devices, text messaging), which can be advantageous to interventions in youth [72]. Raters also examined whether each sleep education intervention was being led/co-led by another student (undergraduate or graduate level), as past work indicated that intervention effectiveness improved when interventions were delivered by trained peer groups [73].
Intervention content was also rated on whether there were explicit displays of sensitivity to the diversity of their student participants. Sensitivity to different gender groups could be demonstrated, for example, by including education on sex differences in sleep disorders, sleep and women's health, or factors that impact sleep in non-binary individuals [74]. Sensitivity to racial/ethnic minoritized groups was shown, for example, by providing education on lifetime exposures that cause racial/ethnic health disparities [75]. Sensitivity to cultural differences in sleep was demonstrated by acknowledging differences in sleeping practices across cultures, for example co-sleeping [76]. Sensitivity to international students manifested, for example, when the intervention program was available in multiple languages. Lastly, the rubric included items on whether the intervention directly considered financial, structural, and neighborhood-level barriers to sleep quality that could interfere with implementing stimulus control or sleep hygiene practices. Specifically, we evaluated if the interventions explicitly considered barriers to sleep (and/or offered solutions) that arise from students a) not always having access to their own personal bedroom, b) living in a safe or quiet neighborhood, c) having access to healthy food, and d) otherwise acknowledging or taking steps to address socioeconomic barriers to implementing sleep recommendations [77].
When full intervention materials were available, we rated each item as explicitly present ("yes") or absent ("no"). When partial intervention materials were available, we replaced absent/no ratings with "cannot determine." Therefore, the primary metric for interpretation across interventions with full or partial materials was the proportion of content that was explicitly present ("yes" ratings).
Of the 36 studies that met inclusion criteria, 26 study authors responded to the request for data (72.2 % of studies). Of these respondents, data were available to be shared for 15 studies (41.7 % of total studies; for respondent rate comparison, see Ref. [78]). Table 1 displays the characteristics of each included study. The combined sample size was N = 599, including 63.4 % female participants and 23.7 % URM participants.
To determine whether the 15 included studies were representative of the broader psychoeducational intervention literature, we compared the sample size, mean age, gender proportion, race/ethnicity proportion, journal impact factor, and study publication year across the 15 studies with data available [21,[26][27][28][29][30][31][32][33][34][35][36][37][38][39] to the 21 studies without data available (Table S4) . The studies differed in sample size such that the excluded studies had larger sample sizes (Fig. S2), t(34) = 12.55, p = .001, d = 0.90, but similar durations (Fig. S3), t(34) = 1.45, p = .156, d = 0.50. As described further below, sample size did not moderate intervention effect sizes.
Among the entire study sample, psychoeducational interventions significantly improved sleep knowledge (g = 0.69, p < .001, 95 %CI: 0.42 to 0.95). Fig. S4 illustrates that the interventions improved sleep knowledge to a similar degree (p = .82) in males (g = 0.64, p = .02, 95 % CI: 0.12 to 1.15) and females (g = 0.73, p = .003, 95 %CI: 0.25 to 1.22). In addition, sleep knowledge improved in URM students (g = 1.14, p = .01, 95 % CI: 0.24 to 2.03) to a level that was similar to (p = .78) non-URM comparison students (g = 0.91, p < .001, 95 % CI: 0.42 to 1.41). There was not significant heterogeneity in effect sizes for sleep knowledge (Fig. S4; Q = 4.00, p = .55), and the effect size remained statistically significant when adjusting for potential publication bias using the trim-and-fill method (g = 0.74, 95 % CI: 0.52 to 0.96; see Fig. S5). Furthermore, similar results were observed when excluding studies that lacked an explicit sleep education component (Fig. S6).
Participants of psychoeducational sleep interventions also had significantly improved sleep quality (g = 0.25, p < .001, 95 % CI: 0.14 to 0.37). Subanalyses showed similar results when excluding studies that lacked any explicit sleep education component (g = 0.26, 95 %CI: 0.16 to 0.36, p <.001; Fig. S6) and when excluding education-only studies that lacked an active component such as CBT-I or mindfulness (g = 0.15, p = .001, 95 % CI: 0.06 to 0.24; Fig. S7). As shown in Fig. 1, the sleep interventions improved sleep quality to a similar degree (p = .63) in male students (g = 0.26, p = .002, 95 % CI: 0.10 to 0.43) and female students (g = 0.32, p < .001, 95 % CI: 0.17 to 0.46). Furthermore, sleep quality significantly improved in both non-URM students (g = 0.28, p < .001, 95 %CI: 0.13 to 0.43) and URM students (g = 0.30, p = .004, 95 % CI: 0.10 to 0.51), and to a statistically similar level (p = .83; Fig. S4). After adjusting for publication bias using the trim-and-fill non-parametric test, the effect size remained significant (g = 0.17, 95 % CI: 0.13 to 0.22).
The sleep quality results showed significant heterogeneity in the true effect sizes across studies (Fig. 1; Q = 114.29, p < .001). Therefore, we entered study duration, journal impact factor, mean age of sample, sample size, and intervention type into a meta-regression as moderators. Effect sizes were not significantly moderated by study duration (β = 0.005, 95 %CI:
-0.03 to 0.04, SE = 0.02, z = 0.31, p = .76), sample size (β = -0.0001, 95 %CI: 0.003 to 0.002, SE = 0.001, z = 0.08, p = .94), or mean age (β = 0.04, 95 %CI: 0.01 to 0.08, SE = 0.02, z = 1.53, p = .13). However, effect sizes were moderated by journal impact factor (Q = 14.34, p = .0002) such that interventions with larger effect sizes were published in journals with higher impact factors (β = 0.13, 95 %CI: 0.06 to 0.20, SE = 0.03, z = 3.79, p = .0002).
effect sizes for sleep quality were also moderated by intervention type (Q = 7.05, p = .03) such that CBT-I based approaches (β = 0.62, 95 %CI: 0.30 to 0.94, SE = 0.16, z = 3.77, p = .0002) led to more favorable outcomes than relaxation/mindfulness-based approaches (β = -0.45, 95 %CI: 0.81 to -0.10,SE = 0.18, z = 2.52, p = .01) and sleep hygiene- based approaches (β = -0.31, 95 %CI: 0.65 to -0.03, SE = 0.17, z = 1.80, p = .07). Interestingly, Fig. S8 shows that the benefits of the type of approach differed across groups: intervention type did not significantly influence sleep quality in males (F = 1.39, p = .29, η p 2 = 0.202) or non-URM students (F = 3.72, p = .06, η p 2 = 0.426), but it did significantly influence sleep quality outcomes in females (F = 5.05, p = .03, η p 2 = 0.479) and URM students (F = 10.31, p = .006, η p 2 = 0.721; Fig. S8). In females and URM students, CBT-I based interventions were associated with the most improvement in sleep quality (p's < 0.01), with no additional differences between sleep hygiene and relaxation/ mindfulness-based interventions (p's > 0.35). In other words, females and URM students benefited from CBT-I based approaches, but not sleep hygiene or relaxation/mindfulness-based interventions. We were also interested in outcomes across gender and URM groups by intervention types after removing studies that lacked an explicit sleep education component but there were too few papers for this subanalysis (k = 0 relaxation/mindfulness-based, k = 1 CBT-I).
Sleep duration, which showed no significant improvements in the overall sample (g = 0.13, p = .13, 95 % CI: 0.04 to 0.31), did not show significant improvements in any demographic subgroup (Fig. S4). Overall improvements to sleep duration were only observed when excluding education-only interventions (g = 0.21, p = .003, 95 % CI: 0.071 to 0.344; Fig. S7).
The materials for 21 interventions (Table S3) were systematically rated on content (Table S2). Ratings were similar across the studies included in the meta-analysis and those that were not, t(19) = 1.49, p = .25, and proportion of "yes" ratings was not significantly associated with sleep quality outcomes in any demographic group (all p's >0.05). Fig. S9 illustrates some possible areas for improvement in intervention content. Regarding designing the intervention for student groups, nearly half of the interventions (47.6 %) were individually focused and included a technological component, however, 0 % employed student peers in coleading the intervention. Regarding sensitivity to the diversity of students who might participate in sleep interventions, only a single intervention program explicitly included content (4.8 %) that addressed racial/ethnic and cultural factors, and none explicitly included content to address gender groups or international students. Regarding structural barriers to poor sleep, none (0 %) of the interventions addressed or provided possible solutions for financial, socioeconomic, and neighborhood-level barriers to sleep in students. Given the modest effect sizes of sleep education interventions on sleep quality/duration, these gaps may represent areas in which future interventions can be more inclusive and more effective.
Our central question was whether students from different URM statuses and gender subgroups benefited similarly from the broad range of psychoeducational sleep programs that have been attempted in student groups. Sleep knowledge and sleep quality improved post-intervention to a similar level across males, females, URM students, and non-URM students (collapsed across different intervention types). These findings were generally consistent with the viewpoint that educating people about sleep can increase intentions to change aviorsthough it is important to recognize that one's intentions to change behaviors do not always correspond to actual behavioral change [6], and that simply accruing more knowledge about sleep (without improving sleep attitudes/prioritization) is unlikely to lead to better sleep quantity or quality. Correspondingly, when collapsing across types of interventions, the effect sizes were modest for sleep quality outcomes (small-medium), and non-significant for sleep duration outcomes (except when excluding education-only studies). However, we observed that CBT-I based interventions showed large effect sizes. In the following sections, we discuss the study outcomes within the context of the overall literature and recommend steps for future work. Prior research has shown that females are more likely to experience poorer sleep quality [81] and both too short [82] and too long sleep durations [83], but the degree to which gender-specific difficulties with sleep are modifiable by sleep programs in educational settings has received minimal attention. We observed that female students benefited as much as male students from psychoeducational sleep interventions. This finding contrasted with the hypothesis that gender-differences in clinical conditions such as anxiety or depression would lead to greater difficulties for females to translate their new sleep knowledge into better behavioral health outcomes [9][10][11]. Both males and females benefited strongly from CBT-I based interventions, a finding that converges with evidence from randomized controlled trials of CBT-I for persons with insomnia that documented comparable efficacy across gender groups [84]. Future work is needed to determine whether post-intervention gender differences emerge over longer follow-up periods, as a function of the level of engagement by the participant (which can differ across gender groups) [85], and/or as a function of whether there is a match in gender between the educator/interventionist and the student. Similar outcomes were observed when comparing URM students to non-URM students. Prior work had indicated that URM students are at greater risk for poor sleep quality, clinical sleep disorders, and daytime sleepiness [14], which are broadly a manifestation of structural racism [8]. From this framework, we hypothesized that URM students would not benefit as strongly from psychoeducational sleep programs either at the level of initial knowledge acquisition (e.g., due to insufficient cultural tailoring to make the program content applicable/interesting) or at turning the new knowledge into better sleep quality (e.g., even if one adopts an earlier bedtime there may be unchangeable neighborhood-level noise or light factors that disrupt sleep). However, we found that URM students were just as likely as non-URM students to show improvements in sleep knowledge following CBT-I based interventions. In addition, CBT-I based approaches led to improved sleep quality in both URM and non-URM students. While these outcomes do not indicate that CBT-I can narrow or remediate racial/ethnic sleep disparities (i.e., both groups improved at the same rate, but started from different baselines), CBT-I based approaches may be one part of a scalable solution that begins to address problems with sleep quality at universities with diverse student bodies.
Though the overall findings are suggestive of psychoeducational interventions improving sleep quality to a similar degree across demographic groups, they require further consideration given Tavernier and Adam's [21] observation that adolescent URM students did not benefit from a sleep intervention to the same degree as non-URM students. One potential explanation is that the current meta-analysis was largely based on college students, a group that tends to have more control over their routines and may have greater healthcare access, better food quality, and safer physical living environments if they live on campus at an institution that provides such resources [10]. High school students, on the other hand, have nighttime schedules that reflect their parents/caregivers who may have variable work schedules and their own health concerns [86], which can have a downstream impact on the students' routines, healthcare access, food quality, and living environment. Therefore, an important direction for future research will be to examine sleep disparities longitudinally from high school into college, coupled with individual-level sleep interventions and structural-level interventions (e.g., safe and quiet living environments on campus).
The structured review of study demographics and content suggested several pathways by which psychoeducational sleep interventions might be improved. For example, these studies have over-sampled female students (63.4 %) and under-sampled URM students (23.7 %) relative to 2020 United States census data (50.5 % females and 34.6 % URM individuals) [87]. The under-representation of URM students was particularly noticeable relative to rates of URM individuals in the 2020 National Student Clearinghouse [88] dataset that aggregates approximately 3600 higher education institutions in the United States (59.4 % females and 41.5 % URM individuals). Therefore, future sleep interventions should invest in ensuring broader representation in student participants.
The content of sleep programs can also be improved. Future programs could explicitly incorporate cultural tailoring by modifying didactic content to directly speak to the cultural and social contexts of specific groups and by using visual content that reflects the audience [89]. Such approaches may be particularly favorable for recruiting, retaining, and benefiting individuals from groups that have historically been under-represented in psychoeducational sleep programs (see Table S1).
There is also a need for future student-focused interventions to address system-level financial and related socioeconomic barriers to sleep, which could include higher-pay campus jobs that reduce total hours and need for night shift work [90], improvement to dormitories such as blackout curtains, white noise machines, and improved air quality [91], health food pantries across campus [92], and elimination of early morning class times that could disadvantage those who have later circadian preferences or need to stay up late to work [79]. Many of these system-level barriers also apply to high school students (part time employment, living in loud neighborhoods with too much light exposure in the bedrooms, inconsistent access to healthy nutrition, and early class times); however, for high school and younger students, interventions to address these system level barriers would need to be directed at the parent-level (e.g., improving wages, reducing night shift work, altering home sleeping conditions), as parents are instrumental in setting bedtimes for the family [86]. Greater effort is also needed for addressing social barriers to sleep, including making sleep health socially desirable amongst students, using credible and relatable peers to connect with the participants and provide sleep health education, and using role modeling to help change campus culture [93]. In a college setting, peer student sleep ambassadors can be trained to deliver the information, model healthy sleep behaviors, and be examples of better sleep habits being associated with better health and academic outcomes.
The current work has limitations. Data and study materials were not available from all published studies in the literature and studies employed different methods for intervention delivery, study duration, and outcome measures (actigraphy or self-report and not all measures were on the same scale). Our definition of psychoeducational interventions was broad and included studies with only educational components as well as studies with active cognitive and behavioral strategies; this broad approach to inclusion reflects the range of sleep programs attempted in student groups, but lacks the precision afforded by focusing on a single, homogenous intervention approach. At the same time, it is also possible that some studies in this literature were not captured by the literature search process.
Another limitation was that individuals were categorized as males or females in individual studies; therefore, this meta-analysis does not inform whether psychoeducational sleep interventions are effective in students who have non-binary or transgendered identities [74]. Students were also categorized as URM versus non-URM individuals, an approach that was statistically necessary because the studies in the literature had small or absent samples sizes for individual racial/ethnic subgroups (Table S1); only three included studies enrolled at least 10 students who identified as Black or African American and only two studies recruited any participants who identified as Native American or Alaskan Native. Thus, there is room for progress in recruiting representative samples in future studies. With larger, more representative samples, future research will be powered to test for intersectionality (including the intersection of specific combinations of multi-racial groups) and to identify if there are specific minority groups that are more disadvantaged in implementing psychoeducational programs than others (e.g., socioeconomic or academic factors). Since "race" is a sociocultural construct [94], a better approach in future work may be to focus on specific exposures that contribute to disparities and/or produce barriers to implementing interventions, including neighborhood factors, food insecurity, and discrimination experiences, among others [15].
In conclusion, the current meta-analysis indicates that males, females, URM students, and non-URM students on average benefit similarly from psychoeducational sleep interventions (collapsed across intervention types). These interventions improved sleep quality, though typically not sleep duration. The largest improvements were observed for interventions that were based on CBT-I principles. Further work is needed to recruit broader samples, focus on exposures that produce disparities, and modify content in a manner that is tailored to the individual and sensitive to their sociocultural context. Collectively, the findings indicate that an investment by university administration in sleep programs that are based on (or explicitly train) CBT-I strategies could benefit students at their institution, especially if those investments were coupled with system-level changes to housing, food plans, healthcare, and class start times.