Physical Therapy Exercises for Sleep Disorders in a Rehabilitation Setting for Neurological Patients: A Systematic Review and Meta-Analysis

Background: Sleep occupies one-third of human life and is essential for health and for emotional, physical, and cognitive well-being. Poor or insufficient sleep is associated with a wide range of dysfunctions that involve different body systems, such as the endocrine, metabolic, and immune systems, thus compromising the higher cortical functions, cognitive performance, mood, and post-physical activity recovery. The present systematic review and meta-analysis aimed to explore the effectiveness of physical therapy exercises on sleep disorders in patients with neurological disorders. Our systematic review identified 10 articles that investigated the effects of physical therapy on sleep disorders in patients with neurological disorders, 6 of which were included in the meta-analysis. Results suggest that physical therapy exercises are a safe and useful strategy for managing sleep disorders in neurorehabilitation.


Introduction
Sleep is a common function of living species. It occupies one-third of human life, and it is shown to be essential for health and for emotional, physical, and cognitive wellbeing [1][2][3][4]. Poor or insufficient sleep is associated with a wide range of dysfunctions that involve different body systems, such as the endocrine, metabolic, and immune systems, thus compromising the higher cortical functions, cognitive performance, mood, and postphysical activity recovery [1,4]. Sleep disturbance can affect both the duration and the quality of sleep, and when it occurs, it reduces the functionality and quality of life (QoL) of the person. Additionally, it represents a risk factor for secondary diseases and medical conditions [5]. Sleep quantity and quality can be affected by age, physical and psychological conditions, and environmental factors [4]. Several studies have shown that sleep disorders (insufficient sleep, excessive amount of perceived sleep, abnormal movements during sleep) are common among the non-motor symptoms in patients with neurological disorders [1,[6][7][8].
Excessive daytime sleepiness, likely caused by a combination of alterations in pathophysiological mechanisms involved in the regulation of sleep/wakefulness, effects of dopaminergic drugs, and nocturnal sleep disruption, is common in patients with Parkinson's disease (PD) [7][8][9][10][11]. Likewise, in patients with multiple sclerosis (MS), daytime tive protocols, manual therapy, and robotic rehabilitation; the comparison was evaluated considering no intervention, unsupervised home-based exercises, standard medical care, and other types of therapies/protocols different from the supervised physical one; and outcomes included any changes shown by the patients in sleep disorders, assessed instrumentally and/or clinically. Both studies that considered changes in sleep disorders as a primary outcome and studies that investigated changes in sleep quality and quantity as a secondary outcome were included in the present review.
Controlled and non-controlled clinical trials (i.e., randomized and non-randomized trials), retrospective studies, case reports, case series, and observational studies were included. No restrictions related to publication date, sex, and country were applied. Studies with a published full text in English or Italian were considered eligible.
Reviews, studies based on pharmacological treatments, and studies that included patients without neurological diseases were excluded.

Study Selection and Data Collection Process
Duplicate records were identified and removed using EndNOTE software. Study eligibility assessment and the data extraction process were carried out by two independent co-authors (SDA and MCC). In the case of any disagreement, the opinion of a third author (MT) was used to reach an agreement. The first selection of studies was initially conducted considering the title and abstract; afterward, full-text articles were examined.
The summary of results was reported following the Preferred Reporting Items for Systematic Reviews and Meta-Analyses (PRISMA) statement [39]. Two authors (DL and RMB) independently extracted the following relevant features of the included studies: name of the first author and publication year, study type, participants, rehabilitative intervention, and outcome measures.

Risk of Bias
Following the instructions in the Cochrane Handbook for Systematic Reviews of Interventions, the risk of bias was assessed using six criteria that were individually rated for each study. In this context, selection bias, performance bias, detection bias, and attrition and reporting bias were considered by the reviewer and assessed using the PEDro score.
The risk of bias was assessed using the Cochrane risk of bias [41] for the controlled trials and using a modified version of the Newcastle-Ottawa Scale (NOS) [42] for the observational studies. The assessment was performed by two authors (SDA and MCC); discrepancies were resolved by consensus with a third reviewer (MT). The modified NOS ranges from 0 to 7. In both scales, the higher is the score, the better is the methodological quality.

Data Synthesis
Data concerning qualitative synthesis were reported in a descriptive way by using means, DS, percentages, and ranges.
Quantitative analysis was conducted by comparing the outcomes used in the included studies. The studies' follow-up results were also considered and pooled.
When available, data for continuous variables were reported as mean differences (MDs), along with their 95% confidence intervals (CIs), while dichotomous outcomes were reported as relative risks (RRs), along with their 95% CIs. In the case of missing data, the authors of the studies were contacted for further information.
A meta-analysis of either dichotomous outcomes or continuous outcomes was carried out whenever possible; when a meta-analysis was not possible, results were presented using summary and descriptive statistics. Meta-analyses were carried out using Review Manager (version 5.2.6, Cochrane Collaboration, Oxford, England), and the p-value was considered statistically significant at <0.05.

Results
Electronic searches identified 1020 studies. Titles and abstracts were examined according to eligibility criteria. The full texts of the articles were read to determine the eligibility. Comparison of the retrieved titles identified 20 duplicates, which were excluded. The result consisted of 1000 articles eligible for inclusion. After a full-text analysis, 990 did not match the inclusion criteria, and 10 studies were included in the present systematic review, as reported in Figure 1. Review Manager (version 5.2.6, Cochrane Collaboration, Oxford, England), and the pvalue was considered statistically significant at <0.05.

Results
Electronic searches identified 1020 studies. Titles and abstracts were examined according to eligibility criteria. The full texts of the articles were read to determine the eligibility. Comparison of the retrieved titles identified 20 duplicates, which were excluded. The result consisted of 1000 articles eligible for inclusion. After a full-text analysis, 990 did not match the inclusion criteria, and 10 studies were included in the present systematic review, as reported in Figure 1.   Table 1 presents a narrative summary of results including studies with their associated characteristics and patient features. In particular, the following data are reported: first author's name, publication year, study type, participants, intervention, intervention duration, and outcome measures.  The included studies were all published in English and were conducted in different countries: three studies came from the United States, three came from Brazil, and Japan, Iran, Jordan, and Switzerland contributed to this review with one study each. Of the five investigated neurological diseases, four studies included patients with a diagnosis of PD, three studies included patients with stroke, two studies involved patients with a diagnosis of multiple sclerosis (MS), and one study included patients with spinal cord injury (SCI) between T7 and T12; no studies regarding patients with TBI were found.
In addition, 747 patients with neurological diseases and sleep disturbance were included in the review, of whom 490 had a stroke diagnosis, 133 had a clinical diagnosis of PD, 111 were persons with MS, and 13 were patients with SCI between T7 and T12.
The primary aim of the included studies was to evaluate the effect of a physical therapy intervention on sleep disorders in patients with neurological diseases.
All the included studies carried out a supervised physical therapy intervention. Different types of protocols in terms of proposed exercises and duration were performed. Summarizing the data, the physical therapy interventions lasted between 6 weeks and 3 months and with one to five sessions per week.
Concerning the outcomes, either instrumental or clinical assessments were performed to evaluate the sleep quality and quantity. The instrumental assessments consisted of polysomnography, electroencephalogram (EEG), electromyogram (EMG), electrooculogram (EOG), and actigraphy. Clinical scales, tests, and questionnaires were used to clinically assess sleep disorders and to investigate the patients' self-assessment of sleep quality. All the outcomes are displayed in Table 1.
The modified NOS scale was used to assess the quality of non-RCTs. The NOS scale of the included studies ranged between 5 to 6, with a mean score of 5.4 points out of 7 ( Table 2). None of the included studies reached the maximum score. The Cochrane risk of bias [41] was used for the RCTs (Figure 2).

Meta-Analysis
Quantitative analysis was carried out by comparing outcomes and follow-ups. This pool was based on comparable outcomes, and comparable times of follow-up allowed consideration of six studies in the meta-analysis (Figure 2). A description of the experimental protocol and control therapy is given in Table 1.
These six studies are as follows.
Al-Sharman et al. [45] compared the effects of a moderate-intensity aerobic exercise program (MAE) with a home exercise program (HEP) in individuals with MS. Both the interventions were conducted for 18 sessions, three times a week for 6 weeks, and each session lasted approximately 50-60 min + 15 min of stretching exercises before and after each exercise session. In Amara et al. [44], the effects of a resistance training (RT) intervention lasting three times a week for 16 weeks were compared with a sleep hygiene intervention (SHI) of a duration of 30-60 min + a telephone call every 4 weeks in patients with PD. Silvia-Batista et al. [47] compared an RT intervention of 24 sessions, two times per week for
Amara et al. [44] compared the effects of the RT intervention lasting three times a week for 16 weeks, with an SHI of a duration of 30-60 min + a telephone call every 4 weeks in patients with PD. In Sadeghi Bahmani et al. [46], an endurance training (ET) intervention and a coordinative training (CT) intervention were compared with an active control condition (ACC) in patients with MS. The three interventions were carried out three times a week for about 45-60 min for 8 consecutive weeks.
Amara et al. [44] compared the effects of the RT intervention lasting three times a week for 16 weeks with the SHI of a duration of 30-60 min + a telephone call every 4 weeks in patients with PD. In Sadeghi Bahmani et al. [46], the ET intervention and the CT intervention were compared with an ACC in patients with MS. The three interventions were carried out three times a week for about 45-60 min for 8 consecutive weeks
In Amara et al. [44], the effects of the RT intervention lasting three times a week for 16 weeks were compared with the SHI of a duration of 30-60 min + a telephone call every 4 weeks in patients with PD. Al-Sharman et al. [45] compared the effects of an MAE with an HEP in individuals with MS. Both the interventions were conducted for 18 sessions, three times a week for 6 weeks, and each session lasted approximately 50-60 min + 15 min of stretching exercises before and after each exercise session.

Comparison Assessed with Sleep Efficiency
The studies by Amara et al. [44] and Al-Sharman et al. [45] were considered. Metaanalysis revealed statistically significant results (p = 0.0008; mean difference = 8.80, 95% confidence interval (CI) = 3.66, 13.94) (Figure 2). Amara et al. [44] compared the effects of the RT intervention lasting three times a week for 16 weeks with the SHI of a duration of 30-60 min + a telephone call every 4 weeks in patients with PD. Al-Sharman et al. [45] compared the effects of an MAE with an HEP in individuals with MS. Both the interventions were conducted for 18 sessions, three times a week for 6 weeks, and each session lasted approximately 50-60 min + 15 min of stretching exercises before and after each exercise session.
Amara et al. [44] compared the effects of the RT intervention lasting three times a week for 16 weeks with the SHI of a duration of 30-60 min + a telephone call every 4 weeks in patients with PD. Al-Sharman et al. [45] compared the effects of an MAE with an HEP in individuals with MS. Both the interventions were conducted for 18 sessions, three times a week for 6 weeks, and each session lasted approximately 50-60 min + 15 min of stretching exercises before and after each exercise session. Figure 3 Show the results of the meta-analysis carried out on six studies by comparing different outcomes and follow-ups. This pool was based on comparable outcomes and comparable times of follow-up. The mean, standard deviation (SD), total number of participants, and data for continuous variables were reported as the mean difference, along with their 95% confidence intervals (CIs) for each study. The risk-of-bias summary for RCTs was reported for each of the six included randomized controlled trials. Scores from the Cochrane Assessment of Bias were reported for each clinical scale: a: PSQI; b: FSS; c: Insomnia; d: WASO; e: Sleep Efficiency; f: Total Sleep Time; experimental arm: supervised physical therapy; control arm: unsupervised exercises; no physical therapy; standard medical care; and other types of therapies/protocols different from the supervised physical one.
Brain Sci. 2021, 11, 1176 13 of 18 Figure 3 Show the results of the meta-analysis carried out on six studies by comparing different outcomes and follow-ups. This pool was based on comparable outcomes and comparable times of follow-up. The mean, standard deviation (SD), total number of participants, and data for continuous variables were reported as the mean difference, along with their 95% confidence intervals (CIs) for each study. The risk-of-bias summary for RCTs was reported for each of the six included randomized controlled trials. Scores from the Cochrane Assessment of Bias were reported for each clinical scale: a: PSQI; b: FSS; c: Insomnia; d: WASO; e: Sleep Efficiency; f: Total Sleep Time; experimental arm: supervised physical therapy; control arm: unsupervised exercises; no physical therapy; standard medical care; and other types of therapies/protocols different from the supervised physical one.

Discussion
The present systematic review and meta-analysis was performed to analyze the role of physical therapy performed in a clinical setting in the improvement of sleep disturbances in neurological patients in order to identify specific protocols that could be included in individualized neurorehabilitation programs.
Results suggest that physical therapy exercises could represent a beneficial interventional for improving sleep disorders in neurological patients. However, due to the relatively few studies and the heterogeneity of the interventions, it is difficult to generalize the results. Moreover, protocols differ in duration, intensity, and required tasks.
Most of the included studies explore the effects of physical therapy on sleep disorders in patients with PD [44,47,49,52], showing positive results. Specifically, multimodal physical therapy programs that stimulate aerobic metabolism and muscle endurance seem to be useful in improving sleep quality assessed by PSQI and the Mini-Sleep Questionnaire (MSQ), in addition to enhancing the sleep efficiency objectively measured with polysomnography [44,47,48]. Moreover, a decrease in the use of drugs promoting sleep was  Figure 3 Show the results of the meta-analysis carried out on six studies by comparing different outcomes and follow-ups. This pool was based on comparable outcomes and comparable times of follow-up. The mean, standard deviation (SD), total number of participants, and data for continuous variables were reported as the mean difference, along with their 95% confidence intervals (CIs) for each study. The risk-of-bias summary for RCTs was reported for each of the six included randomized controlled trials. Scores from the Cochrane Assessment of Bias were reported for each clinical scale: a: PSQI; b: FSS; c: Insomnia; d: WASO; e: Sleep Efficiency; f: Total Sleep Time; experimental arm: supervised physical therapy; control arm: unsupervised exercises; no physical therapy; standard medical care; and other types of therapies/protocols different from the supervised physical one.

Discussion
The present systematic review and meta-analysis was performed to analyze the role of physical therapy performed in a clinical setting in the improvement of sleep disturbances in neurological patients in order to identify specific protocols that could be included in individualized neurorehabilitation programs.
Results suggest that physical therapy exercises could represent a beneficial interventional for improving sleep disorders in neurological patients. However, due to the relatively few studies and the heterogeneity of the interventions, it is difficult to generalize the results. Moreover, protocols differ in duration, intensity, and required tasks.
Most of the included studies explore the effects of physical therapy on sleep disorders in patients with PD [44,47,49,52], showing positive results. Specifically, multimodal physical therapy programs that stimulate aerobic metabolism and muscle endurance seem to be useful in improving sleep quality assessed by PSQI and the Mini-Sleep Questionnaire (MSQ), in addition to enhancing the sleep efficiency objectively measured with polysomnography [44,47,48]. Moreover, a decrease in the use of drugs promoting sleep was low risk of bias, Brain Sci. 2021, 11, 1176 13 of 18 Figure 3 Show the results of the meta-analysis carried out on six studies by comparing different outcomes and follow-ups. This pool was based on comparable outcomes and comparable times of follow-up. The mean, standard deviation (SD), total number of participants, and data for continuous variables were reported as the mean difference, along with their 95% confidence intervals (CIs) for each study. The risk-of-bias summary for RCTs was reported for each of the six included randomized controlled trials. Scores from the Cochrane Assessment of Bias were reported for each clinical scale: a: PSQI; b: FSS; c: Insomnia; d: WASO; e: Sleep Efficiency; f: Total Sleep Time; experimental arm: supervised physical therapy; control arm: unsupervised exercises; no physical therapy; standard medical care; and other types of therapies/protocols different from the supervised physical one.

Discussion
The present systematic review and meta-analysis was performed to analyze the role of physical therapy performed in a clinical setting in the improvement of sleep disturbances in neurological patients in order to identify specific protocols that could be included in individualized neurorehabilitation programs.
Results suggest that physical therapy exercises could represent a beneficial interventional for improving sleep disorders in neurological patients. However, due to the relatively few studies and the heterogeneity of the interventions, it is difficult to generalize the results. Moreover, protocols differ in duration, intensity, and required tasks.
Most of the included studies explore the effects of physical therapy on sleep disorders in patients with PD [44,47,49,52], showing positive results. Specifically, multimodal physical therapy programs that stimulate aerobic metabolism and muscle endurance seem to be useful in improving sleep quality assessed by PSQI and the Mini-Sleep Questionnaire (MSQ), in addition to enhancing the sleep efficiency objectively measured with polysomnography [44,47,48]. Moreover, a decrease in the use of drugs promoting sleep was high risk of bias.

Discussion
The present systematic review and meta-analysis was performed to analyze the role of physical therapy performed in a clinical setting in the improvement of sleep disturbances in neurological patients in order to identify specific protocols that could be included in individualized neurorehabilitation programs.
Results suggest that physical therapy exercises could represent a beneficial interventional for improving sleep disorders in neurological patients. However, due to the relatively few studies and the heterogeneity of the interventions, it is difficult to generalize the results. Moreover, protocols differ in duration, intensity, and required tasks.
Most of the included studies explore the effects of physical therapy on sleep disorders in patients with PD [44,47,49,52], showing positive results. Specifically, multimodal physical therapy programs that stimulate aerobic metabolism and muscle endurance seem to be useful in improving sleep quality assessed by PSQI and the Mini-Sleep Questionnaire (MSQ), in addition to enhancing the sleep efficiency objectively measured with polysomnography [44,47,48]. Moreover, a decrease in the use of drugs promoting sleep was observed in patients following a physical therapy program, supporting physical exercises as a valid alternative in the non-pharmacological treatment of sleep disorders in patients with PD [44]. Interestingly, Amara et al. [44] highlighted how even an intervention based on sleep hygiene can lead to an improvement in the sleep quality assessed by the PSQI.
Although the PSQI is a subjective tool that can be conditioned by a placebo effect [53], the results shown by Amara et al. [44] leave open the possibility of linking a physical therapy program with discussion time with a board-certified sleep medicine physician to provide patients with suggestions for improving sleep hygiene. Further studies could investigate the effectiveness of this combined approach, and furthermore, they could identify the possible role of the physiotherapist in sleep hygiene programs.
Tidman et al. [52] showed an improved willingness to participate in social situations, perceived improvements in flexibility, and perceptions of improved daytime sleepiness in PD patients who performed a supervised physical exercise program, but no significant improvements were found in the sleep quality. These can lead to better adherence and responsiveness of the patient to the rehabilitation treatment, to an increased participation in ADL, and therefore to an improvement in the QoL [54].
Among the included studies, only one [45] evaluated sleep-related biomarkers (melatonin, serotonin, cortisol, highlighting that in patients with MS, aerobic training of moderate intensity leads to an increase in serotonin values. Moreover, this result, associated with the clinical and instrumental evaluation of the quality and quantity of sleep, shows a correlation between the increase in serotonin and the improvement in the quality of sleep assessed by the PSQI and the Insomnia Severity Index (ISI) after 6 weeks of physical training.
These findings are also supported by the other included study concerning MS, which showed a decrease in subjective sleep complaints after 8 weeks of physical exercise programs [46]. Although the lack of further studies prevents us from defining the efficacy of treatment in patients with MS, these results suggest that 45-60 min of exercise sessions, three times a week for a total of 18-24 sessions, may lead to an improvement of insomnia in patients with MS.
Three studies explored the effect of physical therapy treatment on post-stroke patients with sleep disorders, without clinically significant results. Specifically, they compared the effects of conventional motor rehabilitation [48], robotic-based rehabilitation [43], a home exercise program [43], and specific and individualized supervised motor training [51]. The difference between the four rehabilitation protocols and the heterogeneity of the population (different types of lesions and different times from stroke onset) do not allow us to draw conclusions.
One study [50] observed the effect of an aerobic physical exercise protocol in patients with SCI between T7 and T12 associated with periodic leg-movement-related sleep disorders. Although results identified a significant improvement in sleep quality in these patients, the lack of other studies and the small number of patients (n = 13) do not allow conclusions, although the data can be considered as a promising observation for further studies.
The heterogeneity of the studies makes it difficult to identify a single protocol that could be useful in improving sleep disorders in patients with neurological disorders. However, the presence of different protocols in terms of proposed exercises, settings, and populations could be considered as evidence of the great versatility and applicability of physical exercise in patients with neurological diseases associated with sleep disorders. Although physical therapy exercises for sleep disorders could be easily implemented in neurological patients, they are often neglected in conventional neurorehabilitation programs [55].

Limitations of the Current Review
Several limitations in the present review and meta-analysis are acknowledged. First was the small number of studies for each investigated pathology. Second, the methodologi-cal heterogeneity (e.g., study designs, outcome measures) restricted the number of studies eligible for quantitative analysis. In addition, data reporting was frequently incomplete or not always provided in a useful way to perform meta-analysis. Third, the variability of the interventions did not allow us to identify a single rehabilitative protocol that verifies the effectiveness. The internal validity of studies was also limited, and the methodological quality was low to medium, on average, as a consequence of the study designs (lack of randomization and blinding, small or uncontrolled groups). Despite the fact that sleep has a great impact on rehabilitation outcomes, there is a lack of primary research that considers sleep disturbance as a primary outcome in neurorehabilitation. Thus, we considered studies that investigated changes in sleep quality and quantity as a secondary outcome, and this aspect might have limited our findings.

Conclusions
Our review identified 10 articles that investigated the effects of physical therapy on sleep disorders in patients with neurological disorders, and 6 of them were included in a meta-analysis. Results suggest that physical therapy exercises could be a useful strategy for managing sleep disorders in neurorehabilitation. However, due to the heterogeneity of the interventions, it is difficult to generalize the results with a clinical recommendation. Future research in this area would benefit from a higher methodological quality of the study and interventions in physical therapy targeted specifically at sleep disorders.