Sleep and Stroke-Related Delirium: A Systematic Review

: Study objectives : Sleep and circadian rhythms disorders are frequent in the acute stroke. Sleep modiﬁcations are likely to contribute to the development of stroke-related delirium, a common neuropsychiatric complication of acute stroke. This systematic review aimed to clarify the association between sleep modiﬁcations and the occurrence of delirium in patients with acute stroke. Methods : The current systematic review was performed according to the Preferred Reporting Items for Systematic reviews and Meta-Analyses (PRISMA) guidelines. The search was performed on PubMed and Scopus databases. Only studies that provided data concerning sleep, or pre-existing sleep disorders, in acute stroke and performed a formal evaluation of delirium were included. Results : The literature search enabled the identiﬁcation of 15 studies, which exhibited high heterogeneity in terms of study design, settings, sleep assessments, delirium measures, and types of sleep intervention. In the study quality assessment, the majority of the studies were rated as weak or moderate. In most of the cases, sleep was subjectively assessed by the patients or rated by clinicians. None of the studies performed polysomnography for the evaluation of sleep. Only four of the studies assessed the impact of a sleep intervention on delirium, suggesting the potentially protective role of sleep promotion in reducing the prevalence and severity of stroke-related delirium. Conclusions : The evidence arising from the present systematic review supports that sleep disruption is a potential promoting factor for stroke-related delirium. We suggest that a formal sleep assessment and sleep promotion should be included in routine stroke care.


Introduction
Delirium is a neuropsychiatric syndrome characterized by a disturbance of attention, awareness, and cognition, with an acute onset and a fluctuating course [1]. Delirium frequently occurs as a complication of the acute phase of stroke, affecting approximately one in four patients [2]. Delirium has a negative impact on prognosis, leading to increased mortality [3], disability [4], and prolonged hospitalization [5]. The pathogenesis of delirium is believed to be multifactorial, with its onset dependent on the interaction between predisposing risk factors and precipitating factors occurring during hospitalization [6]. A bidirectional relationship between sleep and delirium is commonly hypothesized. In fact, sleep deprivation and delirium share several common risk factors and clinical manifestations. In particular, sleep and circadian rhythms disruption has been closely associated with delirium in the Intensive Care Unit (ICU) [7]. Several mechanisms affecting sleep in critically ill patients have been described, such as noise, altered light-dark cycle, sedation, mechanical ventilation, pain, and stress response. All these factors may have potential bidirectional effects on sleep disruption and delirium occurrence [7][8][9].
Sleep is disrupted in the acute phase of stroke as well [10]. Sleep architecture and quality are particularly compromised after a stroke, with thalamic and cortical strokes exhibiting the most severe alterations [11,12]. Additionally, circadian rhythm is disrupted in the acute phase of stroke: melatonin is significantly reduced after an ischemic stroke when compared to controls [12]. Furthermore, a high prevalence of sleep disorders has been reported in acute stroke patients. Sleep-disordered breathing (SDB) affects around 60% of patients, with nearly one-third of them experiencing severe SDB [13]. Other highly prevalent sleep disorders in acute stroke include insomnia, periodic limb movements, and restless leg syndrome, affecting, respectively, 40%, 30%, and 10% of patients [14]. Also, hypersomnia and excessive daytime sleepiness are frequent in the first month after a stroke [15].
The existing evidence arising from the literature points towards the role of pre-existing sleep disorders and sleep modifications occurring in the acute phase of stroke as contributing factors to the development of stroke-related delirium [16]. As sleep disturbances are potentially modifiable factors, a growing interest has been directed to sleep interventions aimed at preventing delirium in acute stroke patients [17].
The purpose of the current study is to systematically review the medical literature that has evaluated the role of sleep disturbances associated with delirium in acute stroke patients and to investigate the efficacy of sleep interventions on stroke-related delirium.

Methods
The current systematic review was performed according to the latest Preferred Reporting Items for Systematic reviews and Meta-Analyses (PRISMA) statement [18] and was registered on PROSPERO International Prospective Register of Systematic Reviews (ID: CRD42022346332).

Eligibility Criteria for Systematic Review
The eligibility criteria were established according to the PICO methodology, detailed as follows: Population: Patients within one month from acute stroke. Intervention: Where available, sleep interventions to prevent or treat delirium in patients with acute stroke.
Comparison: Comparisons were made between sleep interventions and no interventions, or between patients with subjective or objective disruption of nocturnal sleep and patients without subjective or objective disruption of nocturnal sleep.
Outcome: The considered outcome measures were the incidence, duration, and severity of the stroke-related delirium.
The following search criteria were applied for study eligibility: human studies, English language, and involving adults (≥18 years). Case reports, reviews, comments, editorials, perspectives, erratum, unpublished manuscripts, and conference abstracts were excluded. Only studies that reported data about sleep, and in which delirium was formally evaluated with a validated tool, were included. No time limit was applied to the search.

Information Sources and Search Strategy
The search was performed on PubMed and Scopus databases. In order to reduce the risk of bias due to missing results, we also performed a search for unpublished studies on clinicaltrials.gov. The last search was conducted on 15 July 2022. The following string was used to perform the search: "(((stroke) OR (cerebrovascular disease) OR (cerebral bleeding) OR (brain hemorrhage) OR (brain ischemia) OR (brain hematoma)) AND ((delirium) OR (agitation) OR (confusional state) OR (confusion) OR (attention)) AND ((sleep) OR (circadian rhythm) OR (sleep-wake cycle) OR (insomnia) OR (hypersomnia) OR (somnolence) OR (sleepiness) OR (hypnotics) OR (melatonin) OR (nocturnal) OR (nighttime) OR (apnea)))".
In order to find other eligible studies, we screened the reference lists of selected articles and of other reviews and meta-analyses focused on the same topic.

Selection Process
The search strategy identified 2894 reports. All the authors of the current research screened the identified reports for the title and abstract. Subsequently, three of the reviewers (VB, ER, and GDM) read the full text of the retrieved papers to assess their eligibility. At the end of this process, the reference list of eligible articles, and of other reviews and meta-analyses focused on the same topic of the current research, were screened to identify further eligible reports. The choice to include or exclude a report in this systematic review was solved by consensus of the three reviewers (VB, ER, and GDM).
The search strategy and study selection are described in detail in the PRISMA flow diagram ( Figure 1).

Data Collection Process
VB, ER, IS, and GDM independently extracted the data from the selected studies. The following data were obtained: first author, year of publication, study design, sample size of the study population and of the control group, age, study setting, outcome measures, sleep measures, type of sleep intervention, type of stroke, incidence and measures of delirium, and main findings.

Outcomes
The main endpoint of the present systematic review was to establish whether sleep modifications were associated with the occurrence and/or severity of delirium in patients with acute stroke. The secondary endpoint was to evaluate whether sleep interventions were effective in reducing the incidence and severity of delirium in patients with acute stroke.

Quality and Certainty of Evidence Assessment
The eligible reports were independently evaluated for a quality assessment by VB, ER, IS, and JM. In case of disagreement among the reviewers' scoring, further reviewers (GDM and CI) were consulted. Studies were evaluated by means of the "Quality assessment tools for quantitative studies" developed by the Effective Public Healthcare Practice Project (EPHPP) [19]. This tool consists of six components (selection bias, design, confounders, blinding, data collection methods, and withdrawals and drop-outs) which are rated as strong, moderate, or weak according to a standardized guide ("Dictionary for the Effective Public Health Practice Project Quality Assessment Tool for Quantitative Studies"). After assessing each component, we obtained a global rating as follows: strong for studies with no weak and at least four strong ratings, moderate for studies with one weak and/or less than four strong ratings, and weak for studies with two or more weak ratings.
Two more reports were added after screening the reference lists of the eligible articles and of other reviews and meta-analyses focused on the same topic of the current research [63,64]. At the end of this process, 15 reports were included in this systematic review [16,17,[63][64][65][66][67][68][69][70][71][72][73][74][75]. Two reports included in this systematic review evaluated the same study population, although the main outcome measure differed between the studies [16,69].
No relevant unpublished studies were found from the search on clinicaltrials.gov.
Concerning sleep assessment, none of the reviewed studies performed an instrumental evaluation of sleep by means of polysomnography. Two studies objectively assessed sleep apnea by means of cardiorespiratory polygraphy [72,73]. The study by Sandberg et al. evaluated the effectiveness of nasal-CPAP by means of nocturnal finger-oximetry, and assessed compliance with treatment, as well as sleep duration, from the pressure-sensitive bed recordings [72]. In most of the studies, sleep evaluation was based on the medical records or nurse observations [63,64,67,70,73], or from a patient-subjective point of view [63,64,67]. In three studies the method for sleep evaluation was not specified [16,69,74]; the study by Mengel et al. did not provide any sleep measure [17].
Sleep interventions were heterogeneous across the studies. A multicomponent intervention, which included sleep-related items, was performed by Song et al. [74]. Nakamizo et al. [16] and McLaughlin et al. [68] evaluated sleep disruption in terms of frequent night-time care. A pharmacological intervention was the most common strategy, with various drugs acting on sleep and circadian rhythms [17,[63][64][65]67,70]. A four-week treatment with nasal CPAP was performed by Sandberg et al. [72]. Four studies did not perform any type of sleep intervention [66,71,73,75].

Study Quality Assessment
The EPHPP scores for each selected study are detailed in Figure 2 (upper panel). Following the EPHPP quality criteria, four studies received a strong rating [17,63,64,72], eight studies a moderate rating [16,66,67,69,71,[73][74][75], and three studies a weak rating [65,68,70]. The main weakness of the studies included in this systematic review was the presence of a selection bias in the collection of the study sample; in fact, none of the studies showed a strong score in this component. A proper study design was observed only in five studies. Other frequently observed weaknesses of the selected studies included the lack of blinding of delirium assessors or patients (13 studies) and the lack of controls for possible confounders (5 studies).
On the other hand, most of the papers showed a satisfactory data collection method and a low number of withdrawals and drop-outs.
The global score for each component of all the studies included in this systematic review is shown in Figure 2 (lower panel).

Outcome Measures
The main endpoint of the current study was to systematically review the existing literature focused on the role of sleep modifications in triggering or improving strokerelated delirium. A total of six studies evaluated sleep modifications as predictive factors of delirium occurrence in acute stroke [66,68,69,71,73,75]. Notably, these studies present a high heterogeneity for what concerns the "measure" of sleep.
Wang et al. reported that sleep deprivation is an independent predictor of delirium in patients with acute cerebrovascular diseases (OR: 8.03) [75]. Reznik et al. reported that sleepwake disturbance, rather than being a risk factor, is a prominent characteristic of delirious patients with intracerebral hemorrhage [71]. Two reports observed that sleep disruption as result of frequent nocturnal care may result in a higher incidence of delirium [68,69].
On the other hand, Hosoya et al. observed that "somnolence" after a stroke is a predictor of delirium [66]. However, in all these reports a formal assessment of sleep was missing; therefore, the definitions of "sleep deprivation" and "somnolence" remain unclear.  A specific sleep disorder (e.g., SDB) has been investigated by Sandberg et al. [72,73]. The authors observed that sleep apneas (defined as an Apnea-Hypopnea Index ≥ 10 events/h) are a risk factor for delirium. Remarkably, these are the only research studies in which an instrumental evaluation of sleep was performed. However, the sleep study was limited to a cardio-respiratory polygraphy; therefore, in this case as well, any information regarding the sleep-wake cycle or sleep architecture is missing.
Two studies have highlighted that a history of sleep disturbances prior to stroke (e.g., insomnia, irregular nocturnal lifestyle, and the use of sleep medications) is a risk factor for delirium in the acute stroke phase [66,69]. In the study from Song et al., such an association remains controversial [74].
The secondary endpoint of the current review was to evaluate whether sleep interventions could reduce the impact of delirium in acute stroke. A total of eight studies evaluated sleep interventions to prevent or treat delirium in acute stroke [17,[63][64][65]67,70,72,74]. We observed that the proposed interventions differed significantly: in particular, sleep management included the use of exogenous circadian rhythm synchronizers [17,64,67,70], sleep medications [63,67,70], naloxone [65], a multicomponent intervention including sleep hygiene measures [74], and nasal CPAP [72]. Conversely, the studies by Nakamizo et al. [16,69] and McLaughlin et al. [68] evaluated the role of iatrogenic sleep disruption induced by frequent night-time care in delirium precipitation.
Concerning the use of exogenous circadian rhythm synchronizers, melatonin and ramelteon, a selective melatonin receptors agonist, have been tested for stroke-related delirium prevention and treatment. In the DREAMS study, prophylactic treatment with melatonin at 2 mg nightly started upon admission was shown to reduce the incidence of delirium in patients with acute ischemic stroke (OR: 0.60), while no significant effect on its duration was observed [17]. Also, a nightly administration of 8 mg ramelteon was effective, compared to placebo, in delirium prevention (OR: 0.07) in a small cohort of acute patients, regardless of significant modifications to subjective sleep parameters [64]. Moreover, ramelteon showed a higher efficacy as compared to other sedative drugs for reducing the clinical severity of delirium in elderly patients with acute stroke and concomitant insomnia [70].
Suvorexant, a selective orexin receptor antagonist, was tested by Hatta et al. in an elderly population of acute patients at a daily dose of 15 mg on three consecutive evenings, starting upon admission; the percentage of acute stroke patients in this cohort was 31% [63]. No patients developed delirium in the study group (0/36) compared to a prevalence of delirium of 17% (6/36) in the placebo group, thus suggesting a potential role for suvorexant in delirium prevention [63]. In a subsequent retrospective study, suvorexant, in addition to ramelteon, improved subjective sleep quality and reduced the incidence of delirium in comparison to GABAR agonists [67].
Similarly, sleep hygiene measures (i.e., reducing night-time care, avoiding daytime naps, and applying earplugs and eyepatches) alone [16], or in the context of a multicomponent intervention [74], may be protective regarding delirium incidence and severity.
On the other hand, the treatment of sleep apnea with nasal-CPAP is effective for improving the depressive symptoms of patients with acute stroke, but does not reduce the incidence of delirium in this population [72]. Rather, delirium per se reduces compliance to ventilatory support in patients with acute stroke [72].

Discussion
The purpose of this study was to review the effect of sleep modifications on the occurrence of delirium in acute stroke patients ("stroke-related delirium"). In this systematic review, we identified 15 studies.
The evidence from these studies suggests that sleep disruption is a precipitating factor for the occurrence of delirium. However, this evidence should be considered weak. According to the EPHPP scores, only 4 of the studies included could be evaluated as strong, whereas the remaining 11 were considered to be of moderate or weak quality. The main flaw was the lack of standardized tools for the evaluation of sleep: only the studies by Sandberg et al. [72,73] performed an instrumental evaluation of sleep, but it consisted of cardio-respiratory monitoring for sleep apnea. Thus, no study provided information based on direct measures of sleep architecture. Finally, only in the studies by Hatta et al. [63] and Sandberg et al. [72] were the evaluators blinded with respect to the study measures and outcome.
The studies were highly inconsistent with regards to the criteria used for the definition of sleep disruption, the availability of objective sleep measures and the tools used to assess sleep, the types of stroke, the time interval from stroke to evaluation of delirium, and the tools used to assess delirium. In particular, most of the studies did not use validated sleep measures, and only two studies used objective measures. A further element of inconsistency are the settings of the studies: only seven were performed in a stroke unit.
Importantly, none of the studies properly investigated the complex interaction between delirium and sleep. An unsolved issue is the disentanglement of the interrelation between delirium and sleep. In fact, sleep disturbances cannot only precipitate delirium, but delirium, once present, can itself be associated with disruption of the sleep-wake cycle [6]. Thus, it should be taken into the account that sleep disorders may also be an early sign of delirium, and not the trigger [61]. A bidirectional relationship between sleep and delirium has been proposed. Sleep deprivation affects various areas of the brain, including key brain areas involved in delirium, such as the pre-frontal cortex, thalamus, and posterior parietal cortex [76]. From a molecular point of view, a potential link between sleep disturbances and the pathogenesis of delirium may exist in the Ascending Reticular Activating System (ARAS), a neuronal network located in the brainstem responsible for the arousal response. A hypothesized pathway for delirium involves dysfunction in the ARAS system, resulting in a depletion of acetylcholinergic projections and excessive dopaminergic production, which causes disturbances in alertness and attention [77]. Delirium may be the consequence of the failure of the arousal system to sustain full wakefulness. In this view, delirium can be interpreted as a sleep-wake state dissociation [78].
Aside from acetylcholine depletion and dopamine, another molecular link between delirium and sleep can be found in tryptophan, a precursor of serotonin and melatonin. Serotoninergic pathways from dorsal raphe nuclei play a central role in the maintenance of wakefulness and REM sleep suppression. Recent evidence points towards a protective role of selective serotonin reuptake inhibitors (SSRI) in ICU-related delirium [79]. Melatonin is the major regulator of circadian rhythm and, in the hospital setting, circadian misalignment is a recognized precipitating factor for delirium; moreover, the reduction of plasmatic levels of melatonin was associated with post-operative delirium [80].
The secondary aim of our study was to assess the effect of sleep interventions on the incidence of delirium in acute stroke. Current available data from the literature suggest that sleep interventions may reduce the incidence of delirium in patients with acute stroke. Also, this evidence should be considered weak. Only four such studies were available in the literature [16,17,72,74]; other studies performed sleep interventions and measured the outcome as 'delirium', but either no data were available concerning the number of enrolled patients with stroke [63,64], or an untreated control group was not enrolled [65,67,70]. The results comes from a few studies [16,17,72,74], only two of which were rated as strong according to the EPHPP scores (see Figure 2); nevertheless, the interventions were highly dissimilar: only one used a pharmacological treatment, whereas in the case of Nakamizo, the intervention consisted of reducing night-time care. None of the interventional studies adopted objective or validated sleep measures, and the evaluators were not blinded with respect to the endpoints, study measures, or outcome. It is not possible to establish a direct correlation between the changes in sleep patterns resulting from the sleep-aimed intervention and the occurrence of delirium, due to insufficient sleep measurements both prior to and after the intervention. However, to date, the limited evidence coming from the existing literature suggests that a combined pharmacological (exogenous circadian rhythms synchronizers, sleep promoting medications) and non-pharmacological approach (avoidance of iatrogenic sleep disruption, adoption of multicomponent sleep promoting interventions) should be considered in stroke units to prevent or treat delirium.
Overall, our findings suggest that sleep disruption is a possible risk factor, or a trigger factor, in the development of delirium in acute stroke patients (stroke-related delirium). Even when the sleep disorder (such as insomnia, use of sleeping pills, or circadian misalignment) was pre-existent to the cerebrovascular event, it appeared to be associated with delirium development in the acute phase of stroke. Indeed, acute medical illness and hospitalization are known to exacerbate or worsen a pre-existing sleep disturbance, thus having a role in delirium predisposition or precipitation.
We believe that this grade of evidence is sufficient to justify the adoption of measures aimed at improving sleep, or at avoiding sleep disruption, in acute stroke. As concerns the specific sleep interventions used in the literature, no definite evidence is available on their efficacy, and no evidence-based recommendations on their use can be made.
Sleep disruption in an ICU setting and its relationship with delirium is a widely explored issue, whereas little evidence is available in stroke units' settings, due to their recent development.
Future studies should consider the weaknesses that have emerged from this systematic review. In particular, randomized, controlled clinical trials are necessary to assess the effectiveness of sleep interventions and to establish evidence-based recommendations for delirium prevention and treatment in stroke units. Moreover, future studies should adopt validated sleep measures (i.e., polysomnography) to clarify whether sleep disruptions and sleep modifications promote delirium, and to identify specific sleep patterns that could be considered as predictors of delirium, in order to identify those patients that would benefit from preventive sleep therapy. Ideally, sleep should be assessed prospectively before and after delirium onset to demonstrate its association with stroke-related delirium and its pathogenic role. Finally, it should be kept in mind that sleep disruption may also be an initial sign of delirium and not the trigger [61]. The challenge of disentangling this interrelation requires thoroughly designed prospective studies with sensitive delirium and sleep primary outcomes.
Author Contributions: V.B. and E.R.: drafting of the manuscript; study concept and design; analysis, acquisition, and interpretation of the data; critical revision of the manuscript for important intellectual content; and final approval of the manuscript. I.S., J.M., A.C. and C.I.: analysis, acquisition, and interpretation of the data; critical revision of the manuscript for important intellectual content; and final approval of the manuscript. G.D.M.: study supervision; study concept and design; analysis, acquisition, and interpretation of the data; critical revision of the manuscript for important intellectual content; and final approval of the manuscript. All authors have read and agreed to the published version of the manuscript.
Funding: This research received no external funding.

Institutional Review Board Statement:
This review has been registered on PROSPERO International Prospective Register of Systematic Reviews (ID: CRD42022346332).

Data Availability Statement:
All data used in the present manuscript are available on request.

Conflicts of Interest:
The authors declare no conflict of interest.