Vitamin D Supplementation and Sleep: A Systematic Review and Meta-Analysis of Intervention Studies

Background: Vitamin D deficiency is associated with sleep disorders and poor sleep quality. Whether vitamin D supplementation (VDS) helps resolve these problems remains unclear. Objective: To systematically review the effect of VDS on sleep quantity, quality, and disorders, and perform a meta-analysis of available data. Methods: The reporting of this review followed the PRISMA statement. VDS human interventions studies that reported on sleep quality, quantity, or disorders were included. Medline, CINAHL, EMBASE, PsycInfo, the Cochrane Library, Clinicaltrials.gov, and the ICTRP were searched, in addition to the references of the included articles and previous relevant reviews, without language or time restrictions. Included studies were critically appraised, findings were narratively synthesized, and a meta-analysis was conducted. Furthermore, the overall certainty of the evidence was assessed. Results: A total of 19 studies were included (13 randomized controlled trials (RCTs), 1 opportunistic addition to an RCT, 4 pre–post studies, and 1 pre–post study analyzed as a case series); 3 RCTs were meta-analyses. The risk of bias was generally low. Pre–post studies showed a significant improvement in sleep quality with VDS. Similarly, the results of the meta-analysis revealed a statistically significant decrease in the Pittsburgh Sleep Quality Index with VDS compared with placebo (mean difference, −2.33 (95% CI, −3.09, −1.57); p < 0.001; I2 = 0%), with a moderate certainty of evidence. The results regarding the effect of VDS on sleep-related impairment, difficulty, and disorders, as well as sleepiness and restless legs syndrome, were not unanimous. Conclusions: VDS is promising in improving sleep quality; however, its effect on sleep quantity and disorders needs to be further investigated.


Introduction
Inadequate sleep is a common public health problem of significant personal and societal burden [1]. Sleep disorders such as insomnia, obstructive sleep apnea (OSA), excessive daytime sleepiness (EDS) and fatigue, sleep deprivation, and restless legs syndrome (RLS) are increasingly being diagnosed in clinical practice [2]. It is estimated that 59% of young adults suffer from a sleep disorder and do not get enough sleep [3] and only 36% of this population reports being free of sleep disturbances [4]. Inadequate sleep is an underappreciated determinant of health [5,6] and can lead to short-term and long-term consequences. In the short run, inadequate sleep may result in cognitive and motor performance impairments, which can lead to decreased quality of life and reduced productivity [7,8]. In the longer term, cumulative sleep deprivation can serve as a factor in the development and exacerbation of cardiovascular and metabolic diseases, cancer, diabetes mellitus, gastrointestinal disorders, and mental illnesses [6,9]. Thus, the economic burden of inadequate sleep is substantial, warranting urgent investment in health measures to address this issue [1].
Low vitamin D status is a prevalent condition that has been linked to a wide range of adverse health outcomes [10,11]. Growing evidence has demonstrated that vitamin D has a role in sleep regulation [12]. Specifically, vitamin D deficiency (VDD) can increase risk of sleep disorders and is associated with sleep difficulties, shorter sleep duration, and

Study Selection
Studies meeting the inclusion criteria previously specified were identified by screening titles and/or abstracts from electronic scientific databases via Endnote, version X6. Full texts of potentially eligible studies were retrieved. Finally, records in the grey literature search zone were further assessed for eligibility of inclusion.

Data Extraction
For all eligible studies, data related to the features of the study, population groups, interventions given (type, form, and the dose of vitamin D in experimental groups, comparator, and duration), outcomes, and main findings were extracted and recorded in a data extraction form. When reported as nmol/L, the author converted serum 25OHD to ng/mL by dividing by a factor of 2.496. The author contacted the authors of some included studies to obtain additional data, when they were not reported in the published studies.

Quality Assessment
The Cochrane criteria (sequence generation, allocation concealment, blinding of participants and outcome assessors, incomplete outcome data, and selective outcome reporting) were used as the tool to assess the risk of bias of RCTs included in this review [20]. Furthermore, a modified version of the Cochrane Risk of Bias tool [21] (eligibility criteria, measurement of exposure and outcome, control confounding, and follow-up) was used to assess the risk of bias of non-randomized studies. Potential sources of bias for both RCTs and non-randomized studies were graded as low, high, or unclear risk.
The Grading of Recommendations Assessment, Development, and Evaluation (GRADE) criteria (risk of bias, inconsistency, indirectness, imprecision, and publication bias) were used to assess the overall certainty of the evidence presented using GRADE Evidence Profiles developed in the GRADEpro GDT software (www.gradepro.org; accessed on 8 November 2021).

Data Synthesis
A narrative composite of the study findings was provided when a meta-analysis was not feasible. Furthermore, author-recorded features of the study, population group characteristics, intervention provided, comparator, and the outcome were included in this composite.
A meta-analysis was conducted when participants, treatments, and outcomes shared similar characteristics to allow pooling. Standard meta-analyses comparing VDS with placebo were performed using RevMan version 5.4 (The Cochrane Collaboration, The Nordic Cochrane Centre). A random-effects model was used for the analysis of more than two studies. The results were reported as the mean difference with 95% confidence intervals. The I 2 statistic was used to assess heterogeneity among different studies.

Quality of Reporting
The Preferred Reporting Items for Systematic reviews and Meta-Analyses literature search extension (PRISMA-S) checklist and the PRISMA statement were followed for the literature search component [22], and the reporting of this systematic review [23].
Three out of thirteen RCTs generated data that could be combined in the meta-analysis.  Table 1 demonstrates the characteristics of the included studies. Six of the studies were conducted in Iran [25,27,30,31,34,39], five in the USA [28,29,32,33,42], one in Ireland [26], one in New Zealand [37], one in KSA [35], one in China [36], one in Turkey [40], and one in the Netherlands [24]. The number of participants varied from 5 [31] to 18,353 [32]. Two studies were conducted in the pediatric population: children with Autism Spectrum Disorder [40] and ADHD [31], one study was conducted on overweight postmenopausal women [28], another in elderly women [39], and one on community-dwelling older people [24]. One study was conducted in patients receiving maintenance methadone treatment [25], one in active-duty warfighters [29], one on veterans with multiple areas of chronic pain and low serum 25(OH)D [42], one on adult patients with urticarial [33], one on fibromyalgia syndrome patients [30], one in patients with chronic low back pain [41], and one on patients with depression with tied anxiety symptoms [36]. Finally, one study was conducted on adults with OSA [26], one on adults with sleep disorders [27], two in patients with RLS [35,38], and one on adults with vitamin D deficiency, abdominal obesity, and symptoms of insomnia [34].

Characteristics of Included Studies
In the majority of the studies, the intervention consisted of vitamin D3 supplementation [24,[26][27][28][29][32][33][34][35]37,38,42], two used vitamin D2 supplementation [40,42], and the form of vitamin D was unclear in six trials [25,30,31,36,39,41]. Only Sharifan et al. [34] assessed Three out of thirteen RCTs generated data that could be combined in the meta-analysis. Table 1 demonstrates the characteristics of the included studies. Six of the studies were conducted in Iran [25,27,30,31,34,39], five in the USA [28,29,32,33,42], one in Ireland [26], one in New Zealand [37], one in KSA [35], one in China [36], one in Turkey [40], and one in the Netherlands [24]. The number of participants varied from 5 [31] to 18,353 [32]. Two studies were conducted in the pediatric population: children with Autism Spectrum Disorder [40] and ADHD [31], one study was conducted on overweight postmenopausal women [28], another in elderly women [39], and one on community-dwelling older people [24]. One study was conducted in patients receiving maintenance methadone treatment [25], one in active-duty warfighters [29], one on veterans with multiple areas of chronic pain and low serum 25(OH)D [42], one on adult patients with urticarial [33], one on fibromyalgia syndrome patients [30], one in patients with chronic low back pain [41], and one on patients with depression with tied anxiety symptoms [36]. Finally, one study was conducted on adults with OSA [26], one on adults with sleep disorders [27], two in patients with RLS [35,38], and one on adults with vitamin D deficiency, abdominal obesity, and symptoms of insomnia [34].

Assessment of Risk of Bias
The assessment of the risk of bias of included studies is presented in Figure 2. Regarding RCTs, the risk of bias was low, except for the study by Zhu et al. [36]. As for non-randomized trials, measurement of exposure was unclear in the studies conducted by Arico et al. [38], Eshaghi et al. [39], and Maheshwari et al. [41]. Finally, risk of bias regarding incomplete follow-up was high in the studies carried out by Arico et al. [38] and Huang et al. [42].

Assessment of Risk of Bias
The assessment of the risk of bias of included studies is presented in Figure 2. Regarding RCTs, the risk of bias was low, except for the study by Zhu et al. [36]. As for nonrandomized trials, measurement of exposure was unclear in the studies conducted by Arico et al. [38], Eshaghi et al. [39], and Maheshwari et al. [41]. Finally, risk of bias regarding incomplete follow-up was high in the studies carried out by Arico et al. [38] and Huang et al. [42].
(a) Findings from the included studies are presented in Table 2. Findings from the included studies are presented in Table 2.

Sleep Quality
The pre-post studies by Huang et al. [42], Eshaghi et al. [39], and Maheshwari et al. [41] investigated the effect of VDS on sleep quality assessed by the PSQI. All three trials showed a significant improvement in overall sleep quality with VDS. However, the three trials were heterogeneous and did not contain ample information allowing pooling; hence, performing a meta-analysis of their results was impossible.
The four RCTs conducted by Ghaderi et al. [25], Majid et al. [27], Mason et al. [28], and Mirzaei et al. [30] explored the effect of VDS on sleep quality assessed by the PSQI. Only three [25,27,30] provided information and had similar characteristics to allow pooling, whereas the study by Mason et al. [28] did not report on numerical outcomes, and therefore was not included in the meta-analysis. This study showed no significant change in overall sleep quality with VDS and a deterioration in total PSQI among women who repleted their vitamin D levels, concluding that VDS of 2000 IU/d may result in overall worse sleep quality for postmenopausal women with low circulating vitamin D undergoing weight loss.
As for the results of the meta-analysis of the eligible RCTs [25,27,30], the forest plot for the mean difference in the PSQI based on VDS is presented in Figure 3. The three studies included patients undergoing maintenance methadone treatment [25], people with a PSQI ≥ 5 [27], and fibromyalgia syndrome patients [30]. The duration of VDS was short (8 [27,30] to 12 weeks [25]), and the dose was either 3571.42 [25,27] or 7142.85 IU [30]. A statistically significant decrease in the PSQI in the group receiving VDS as compared with placebo was shown by the meta-analysis (mean difference, −2.33 (95% CI, −3.09, −1.57); p <0.001). The heterogeneity of the analysis was null (I 2 = 0%). The overall certainty of the evidence of the meta-analysis was moderate (Supplementary S2).

Sleep Quality
The pre-post studies by Huang et al. [42], Eshaghi et al. [39], and Maheshwari et al. [41] investigated the effect of VDS on sleep quality assessed by the PSQI. All three trials showed a significant improvement in overall sleep quality with VDS. However, the three trials were heterogeneous and did not contain ample information allowing pooling; hence, performing a meta-analysis of their results was impossible.
The four RCTs conducted by Ghaderi et al. [25], Majid et al. [27], Mason et al. [28], and Mirzaei et al. [30] explored the effect of VDS on sleep quality assessed by the PSQI. Only three [25,27,30] provided information and had similar characteristics to allow pooling, whereas the study by Mason et al. [28] did not report on numerical outcomes, and therefore was not included in the meta-analysis. This study showed no significant change in overall sleep quality with VDS and a deterioration in total PSQI among women who repleted their vitamin D levels, concluding that VDS of 2000 IU/d may result in overall worse sleep quality for postmenopausal women with low circulating vitamin D undergoing weight loss.
As for the results of the meta-analysis of the eligible RCTs [25,27,30], the forest plot for the mean difference in the PSQI based on VDS is presented in Figure 3. The three studies included patients undergoing maintenance methadone treatment [25], people with a PSQI ≥ 5 [27], and fibromyalgia syndrome patients [30]. The duration of VDS was short (8 [27,30] to 12 weeks [25]), and the dose was either 3571.42 [25,27] or 7142.85 IU [30]. A statistically significant decrease in the PSQI in the group receiving VDS as compared with placebo was shown by the meta-analysis (mean difference, −2.33 (95% CI, −3.09, −1.57); p <0.001). The heterogeneity of the analysis was null (I 2 = 0%). The overall certainty of the evidence of the meta-analysis was moderate (Supplementary 2).

Disturbed sleeping
The only pre-post study [40] investigating sleep habits and disorders in children with ASD showed that VDS may be beneficial in these patients, as well as healthy individuals with sleep disturbances.
As for RCTs, the effects of VDS on sleep-related impairment, sleep difficulty, and sleep disorders were assessed by McCarthy et al. [29], Okereke et al. [32], and Zhu et al. [36], respectively; the results were not unanimous. While McCarthy et al. [29] showed a statistically significant improvement in sleep-related impairment with VDS, Okereke et al. [32] and Zhu et al. [36] did not report on such findings. In both RCTs, there were nonsignificant differences in likelihood of sleep problems with VDS compared with placebo after controlling for confounding variables).

Sleepiness
Two RCTs assessed the effect of VDS on sleepiness using different tools [26,34]. The study of Kerley et al. [26], which was conducted on patients with OSA, did not report any

Disturbed sleeping
The only pre-post study [40] investigating sleep habits and disorders in children with ASD showed that VDS may be beneficial in these patients, as well as healthy individuals with sleep disturbances.
As for RCTs, the effects of VDS on sleep-related impairment, sleep difficulty, and sleep disorders were assessed by McCarthy et al. [29], Okereke et al. [32], and Zhu et al. [36], respectively; the results were not unanimous. While McCarthy et al. [29] showed a statistically significant improvement in sleep-related impairment with VDS, Okereke et al. [32] and Zhu et al. [36] did not report on such findings. In both RCTs, there were non-significant differences in likelihood of sleep problems with VDS compared with placebo after controlling for confounding variables).

Sleepiness
Two RCTs assessed the effect of VDS on sleepiness using different tools [26,34]. The study of Kerley et al. [26], which was conducted on patients with OSA, did not report any difference in sleepiness between the group receiving VDS and those receiving placebo. In contrast, the study by Sharifan et al. [34], which was conducted on patients with insomnia, showed a beneficial effect of vitamin D3-fortified low-fat milk on insomnia symptoms compared with unfortified milk. No effect was detected with vitamin D3-fortified low-fat yogurt compared with unfortified one.

RLS
The pre-post study conducted by Arico et al. [38] found that long-term VDS (6 months) decreased RLS severity in a small population (5 patients). In contrast, the RCT by Wali et al. [35] found no effect of VDS on severity of RLS compared with placebo.

Sleep Problems as Adverse Events of VDS
Sleep problems as an adverse event of VDS were assessed in the two RCTs conducted by Mohammadpour et al. [31], and de Koning et al. [24]. Both studies showed no significant difference in sleep problems with VDS versus placebo in children with ADHD or community-dwelling people with depressive symptoms, respectively.

Discussion
This systematic review and meta-analysis investigated the effect of VDS on sleep quantity and quality, and sleep disorders. VDD is an emerging risk factor for suboptimal sleep and sleep disorders [12,15,43]. Such an association has been observed in several healthy and ill populations including factory workers, older community-dwelling men, hemodialysis patients, and pregnant women [12]. Specifically, through a meta-analysis of observational studies involving 9397 participants, Gao et al. found that participants with VDD had increased odds of sleep disorders and poor sleep quality by 1.5 fold, short sleep duration by 1.75 fold, and sleepiness by 1.36 fold. They also provided evidence that serum 25(OH)D below 20 ng/mL could significantly heighten the odds of unhealthy sleep [15]. Similarly, through a meta-analysis of observational studies conducted on 1864 subjects with sleep disorders and 1340 control participants, Yan et al. [43] found that the average serum vitamin D concentration in the group with sleep disorders was 0.75 ng/mL lower than that in the control group [43].
The association between VDS and sleep regulation is biologically plausible and worth investigation given its clinical and public health relevance. Nevertheless, we found a limited number of human interventional studies-especially RCTs-and most of the included studies focused on sleep quality compared with other sleep-related outcomes. The evidence from included studies was promising regarding the effectiveness of VDS on enhancing sleep quality; nevertheless, studies investigating sleep quantity and sleep disorders were scarce, heterogeneous in terms of included populations and methodologies, and their findings were not unanimous, preventing generating a solid conclusion. Hence, our results suggest that VDS is promising in improving sleep quality; however, its effect on sleep quantity and disorders needs to be further investigated.
Although the exact physiological mechanisms between vitamin D and sleep regulation have not yet been fully uncovered, several direct and indirect mechanisms have been suggested [12]. One potential mechanism is the extensive presence of vitamin D receptors in many parts of the brain, specifically areas that affect sleep [44]. Another theory involves the expression enzymes involved in vitamin D activation and degradation (25-hydroxylase and 1-hydroxylase and 24-CYP24A1) in areas in the brain known to be involved in sleep regulation including the supraoptic and paraventricular nuclei within the hypothalamus and the substantia nigra [44,45]. Another plausible theory considers the effect of sunlight. It is well known that vitamin D levels are regulated by exposure to sunlight and since sunlight also affects the circadian rhythm, it is highly plausible to assume that there is a link between those factors [46][47][48][49]. Furthermore, the production of melatonin-a hormone involved in the regulation of circadian rhythms and sleep-is regulated by vitamin D; thus, impaired vitamin D levels could decrease melatonin levels, potentially leading to sleep disorders [50,51]. One final plausible mechanism is that vitamin D, as an immunomodulatory molecule, plays a role in downregulating inflammatory markers that are involved in sleep regulation such as tumor necrosis factor α (TNF-α), cytokines and prostaglandin D2. In the case of VDD, such inflammatory markers would be raised, negatively affecting sleep [52,53]. All of these factors may explain our findings regarding the beneficial effect of VDS on sleep quality.
The only study involving patients with OSA was conducted by Kerley et al. [26]. VDD is a common finding in OSA patients compared with non-apneic subjects, and vitamin D levels were shown to be inversely correlated with the severity of OSA [54]. This association is likely to be mediated by complex pathogenetic mechanisms, such as immune system modulation, myopathy, and inflammation; nevertheless, these mechanisms are not fully understood yet. Additionally, this relationship seems to be confounded by numerous factors, such as obesity [54]. Kerley et al. [26] did not find a beneficial effect of VDS on improving sleepiness in this patient population. This might be due to the small sample of the trial and to the fact that 90% of the VDS group were stable on continuous positive airways pressure, which may have diluted any potential benefit of VDS [26].
Our findings concerning the effectiveness of VDS on RLS were contradictory [35,38]. VDD may play a role in this movement disorder through its link with dopaminergic dysfunction. Nevertheless, to date, the causality between VDD and RLS is only hypothesized. This null effect, despite the significant improvement in vitamin D levels in the intervention group, suggests that VDS may not have a therapeutic effect in RLS, although it may contribute to the pathophysiology of the syndrome [35]. The authors of this study argue that in RLS, vitamin D levels in the brain are more important than in the blood. Accordingly, the improvement in serum levels with VDS may not have sufficiently affected vitamin D levels in the brain [35].
It should be noted that vitamin D levels in response to supplementation depend on three factors, dose, frequency and interval [55]; accordingly, our findings could be interpreted in light of this fact. For example, the RCTs that found a beneficial effect of VDS on sleep disorders were conducted over a short period of time, i.e., 10 weeks in the study by McCarthy et al. [29], and 12 weeks in the studies by Rorie et al. [33] and Sharifan et al. [34]. In contrast, RCTs reporting no significant improvement in sleep disorders with VDS were those conducted over a long duration, i.e., 6 months in the study by Zhu et al. [36], 18 months in the study by Slow et al. [37], and a median follow-up duration of 5.3 years in the study by Okereke et al. [32]. This observation, although it could be attributed to seasonality, raises queries on whether extended supplementation with vitamin D may not always result in better outcomes.
Finally, human circadian rhythms-sleep and wakefulness cycles-are synchronized by environmental signals, especially light and dark intervals through sunlight [48]. Although intentional sun exposure could hence be recommended to enhance sleep and vitamin D levels, it remains highly challenging to titrate one's exposure, besides the documented negative side effects of ultraviolet irradiation.

Strengths and Limitations
Our study has numerous strengths. First, we followed a systematic approach in our search and analysis, using a highly sensitive search strategy, and followed recommended reporting approaches for the review [23] as well as the search strategy [22]. Second, we contacted the authors of some included studies to obtain additional data when they were not reported in the published studies. Unfortunately, we did not receive feedback from all authors. Finally, the risk of bias of the majority of included studies was low.
Nevertheless, the current analysis had some limitations. First, we were limited to study level rather than individual-level data, which would have been more accurate than the overall mean change in sleep. Second, there were variabilities between included studies, which complicates the comparisons as well as the interpretation of our results, especially in the study populations, the outcomes assessed, and the assessment methods. Third, some papers did not provide crucial information such as the form of vitamin D, the levels of vitamin D at the end of the study, the daily dose equivalent, the assessment method of vitamin D levels, nor compliance with VDS. Although we contacted respective authors, we could not obtain the needed information in some instances. This would have enabled us to better interpret our findings. Fourth, while we tried to make the literature search as exhaustive as possible, pertinent studies might have been missed; a common limitation to systematic reviews. Furthermore, we did not have access to some potentially eligible studies for full-text screening. Fifth, given the small number of studies investigating the effect of VDS on sleep disorders and their heterogeneity, we could not perform a metaanalysis of their findings and our conclusion remains limited to a qualitative synthesis. Finally, repeat screening, selection of studies, data extraction, and quality assessment was not possible.

Conclusions
In conclusion, the evidence presented in this review suggests a beneficial role of VDS in enhancing sleep quality. These results remain to be interpreted with caution given the limited number of included RTCs and their relatively small sample size. Nevertheless, the positive effects of such supplementation could be considered in clinical practice, especially in the context of beneficial skeletal [56] and pleiotropic extraskeletal effects [57,58] of vitamin D, as well as the relatively limited cost of VDS. As we could not find enough studies assessing the effect of VDS on sleep disorders, OSA, sleepiness, and RLS, this remains to be explored in future adequately powered, high-quality RCTs.
Funding: Funded by Cluster Grant R18030. College of Natural and Health Sciences, Zayed University, Dubai, United Arab Emirates. The funding body was not involved in study design, data collection and analysis, results interpretation, and manuscript write up.