Effects on Sleep Quality of Physical Exercise Programs in Older Adults: A Systematic Review and Meta-Analysis

Background: Given the beneficial effects of exercise in different populations and the close relationship between healthy ageing and sleep quality, our objective was to determine if physical exercise delivered through a structured program improves sleep quality in older adults. Methods: Embase, PubMed/MEDLINE, Web of Science, and Cochrane Register of Clinical Trials (CENTRAL) were searched to 15 January 2023. Studies that applied physical exercise programs in older adults were reviewed. Two independent reviewers analysed the studies, extracted the data, and assessed the quality of evidence. Results: Of the 2599 reports returned by the initial search, 13 articles reporting on 2612 patients were included in the data synthesis. The articles used interventions based on yoga (n = 5), multicomponent exercise (n = 3), walking (n = 2), cycling (n = 1), pilates (n = 1), elastic bands (n = 1), and healthy beat acupunch (n = 1). In the intervention group, we found significant improvement in Pittsburgh sleep quality index of −2.49 points (95% CI −3.84 to −1.14) in comparison to the control group (p = 0.0003) and sleep efficiency measured with objective instruments (MD 1.18%, 95% CI 0.86 to 1.50%, p < 0.0001). Conclusion: Our results found that physical exercise programs in older adults improve sleep quality and efficiency measured with objective instruments.


Introduction
The number of older people is growing worldwide. Currently, about 10% of the global population is over 65 years old [1], and it is projected that between 2015 and 2050, the proportion of the world's population over 60 years old will nearly double from 12 to 22% [2]. The challenge for this population is to achieve healthy ageing [2]. In addition, the clinical guidelines widely recommend physical activity as a central component of A systematic review was reported according to the Preferred Reporting Items for Systematic Reviews and Meta-Analyses (PRISMA) guidelines [26]. The review was registered in the International Prospective Register of Systematic Reviews (PROSPERO) CRD42022369802.

Criteria for Considering Studies in This Review
Randomised controlled trials (RCTs) that included older adults (>65 yr) [27] performing a structured exercise program were searched. The included studies aimed to determine the effects of exercise on sleep quality. The search strategy was based on the PICO model (population: older adults; intervention: exercise program; control: no intervention or placebo; and outcome: sleep quality, sleepiness, apnoea-hypopnoea index) [28].

Search Strategies and Data Resources
The Embase, PubMed/MEDLINE, Web of Science, and Cochrane Register of Clinical Trials (CENTRAL) were reviewed on 23 October 2022 and updated on 15 January 2023. Manual searches with the following terms were conducted: ((elderly) OR (older people) OR (older adults) OR (ageing)) AND ((Exercise) OR (physical training) OR (physical activity) OR (resistance training) OR (aerobic) OR (gym) OR (gymnastic) OR (strength) OR (balance)) AND ((sleep quality) OR (sleepiness) OR (apnea-hypopnea index)) (supplementary material 1). No language or publication restrictions were imposed.
The terms selected were combined using Boolean logical operators (OR, AND, NOT). Additionally, a manual search of the references included in the selected articles was conducted. All references were analysed using Rayyan web software [29].

Reviewing Procedure and Data Extraction
The selected articles were reviewed independently by investigators with experience in meta-analysis and training in the literature review. First, the titles and abstracts of all identified studies were reviewed in pairs by four investigators (LSN-MOY-OM-RTC). Studies deemed irrelevant based on the title and abstract were excluded. Any possible disagreements were resolved by a third reviewer (MSR). Second, the first step's full-text versions of the articles selected were reviewed and rechecked against the eligibility criteria (LSN-RTC). Again, any potential disagreements were resolved by a third reviewer (MSR). There were no disagreements between the investigators. Finally, study authors were contacted as required to determine study eligibility.
Two authors (OM-RTC) extracted the data using a standardised protocol and reporting forms. The following information was extracted from each included study: design, population characteristics, and exercise program characteristics (type of exercise, setting, duration, intensity, frequency). If some relevant data were not in the article, the author was contacted to request the information.

Methodological Quality Assessment
The primary articles' methodological quality was assessed using the Cochrane Collaboration tool to assess the risk of bias (the Cochrane Handbook for Systematic Reviews of Interventions) [28]. The tool included seven items: generation of a random sequence, allocation concealment, blinding of participants and personnel, blinding of outcome assessment, completeness of outcome data, selection of reports, and other biases. For each item, the risk of bias for the study was rated according to three categories: low, high, or unclear risk of bias. Two reviewers (CFJ-MSM) independently assessed the risk of bias in the studies. A third author (RTC) was consulted for discrepancies that could not be resolved.

Data Synthesis and Analysis
A narrative synthesis was used to summarise the characteristics of studies, including population and intervention characteristics. Summaries of the association between the outcomes for each study in terms of mean differences (MDs) or standard mean differences (SMDs) using Review Manager 5 (RevMan, Copenhagen, Denmark: The Nordic Cochrane Centre, The Cochrane Collaboration, 2014) were reported. Absolute values and obtained combined measures of the effect of each primary outcome through meta-analysis with a random-effect model due to the expected heterogeneity between studies were compared [28]. Statistical heterogeneity was measured with the I 2 statistic and classified as low (I 2 < 25%), moderate (I 2 = 25-50%), or high (I 2 > 50%) [28]. A sub-analysis was carried out by type and duration of training and setting for the main outcome. Additionally, the overall certainty of evidence was assessed independently by two reviewers (RTC, LSN) using the GRADE approach [30]. Disagreements were solved by consensus. Publication bias was assessed by visualising a funnel plot and Begg's and Egger's tests for the possible existence of study bias using the Jamovi software (version 2.3) [31].

Study Selection
The initial search yielded 2599 potential studies. In total, 376 duplicate records were deleted. We screened 2223 titles and abstracts and excluded 2169 records that did not meet the inclusion criteria. Fifty-four of these were assessed as full text. Of these, 30 were  excluded for including people under 65 years old, 14 for being conference abstracts, 3 for  not being RCTs, 3 for not including exercise as intervention, and 1 for duplicated data. Ultimately, 13 studies met the criteria for eligibility and were included in the review [32][33][34][35][36][37][38][39][40][41][42][43][44]. The flow chart of the study selection process is shown in Figure 1.

Study Selection
The initial search yielded 2599 potential studies. In total, 376 duplicate reco deleted. We screened 2223 titles and abstracts and excluded 2169 records that did the inclusion criteria. Fifty-four of these were assessed as full text. Of these, 30 cluded for including people under 65 years old, 14 for being conference abstracts, being RCTs, 3 for not including exercise as intervention, and 1 for duplicated d mately, 13 studies met the criteria for eligibility and were included in the review The flow chart of the study selection process is shown in Figure 1.

Methodological Quality Assessment
All studies had a high or unclear risk of bias in at least one domain. The majority of studies claimed to be randomised. However, only half of them explain how the randomisation was conducted [34,35,38,40,41,43]. No study reported that participants and personnel were blinded. However, this is a common feature in non-pharmacological studies, particularly in rehabilitation studies, in which it is complex to blind professionals or patients due to the nature of the intervention. One-third of studies reported that researchers and outcome assessments were blinded [35,37,38,40]. Five studies had a low risk of bias on attrition rates and selection reporting [34,35,38,41,43]. Finally, the majority of the studies had a low risk of other potential sources of bias ( Figure 2). Funnel plots and Begg's and Egger's tests indicated a publication bias for the estimation of the effect of exercise on sleep quality (Begg = 0.08; Egger = 0.009) (Figure 3).
When performing the analysis by setting, the programs carried out at home did not improve sleep quality (SMD −0.95, 95% CI −1.91 to 0.02, p = 0.06). However, the programs carried out in the facilities showed a significant improvement in sleep quality (SMD −1.39, 95% CI −2.45 to −0.33, p = 0.01) (Figure 8). No significant differences were found when performing the analysis by type of training. When performing the analysis by time duration, the programs with duration less than 12 weeks improve the sleep quality (SMD −0.66, 95% CI −1.10 to −0.22, p = 0.003) with a high heterogeneity (I 2 = 64%). However, the programs carried out between 12 weeks and 12 months (SMD −1.76, 95% CI −3.67 to 0.15, p = 0.07) and more than 12 months (SMD −1.47, 95% CI −4.18 to 1.23, p = 0.29) did not improve sleep quality. The heterogeneity was high in both cases (I 2 = 97, and I 2 = 95%, respectively) ( Figure 6). If we analysed by time duration, but including intermediate assessment points of high-duration programs, the programs between 12 weeks and 12 months improve the sleep quality (SMD −1.94, 95% CI −3.51 to −0.38, p = 0.02) with a high heterogeneity (I 2 = 97%), and the programs of more than 12 months (SMD −0.11, 95% CI −0.23 to 0.02, p = 0.09) did not improve sleep quality (Figure 7).  When performing the analysis by setting, the programs carried out at home did not improve sleep quality (SMD −0.95, 95% CI −1.91 to 0.02, p = 0.06). However, the programs carried out in the facilities showed a significant improvement in sleep quality (SMD −1.39, 95% CI −2.45 to −0.33, p = 0.01) (Figure 8). No significant differences were found when performing the analysis by type of training.
Sleep efficiency: Five studies examined sleep efficiency [32,35,37,38,42]. The instruments used were accelerometry/actigraphy [35,37] and a specific domain of PSQI [32,38,42]. These studies compared 108 participants in the IG versus 100 participants in the CG. The heterogeneity was moderate (I 2 = 38%). Older people in the IG had no change after intervention (SMD 0.60, 95% CI −0.12 to 1.31) in comparison to CG (p = 0.10). The heterogeneity of the comparison was high (I 2 = 82%) (Figure 9). If we analysed only the studies with an objective measure of sleep efficiency [35,37], there were 59 older people compared in the IG versus 58 in the CG. Older people in the IG had a higher sleep efficiency post-intervention (MD 1.18%, 95% CI 0.86 to 1.50%) in comparison to CG (p < 0.0001) (Figure 10). The heterogeneity of the comparison was low (I 2 = 0%). The certainty of evidence, according to the GRADE methodology, was low.      Sleep efficiency: Five studies examined sleep efficiency [32,35,37,38,42]. The instruments used were accelerometry/actigraphy [35,37] and a specific domain of PSQI [32,38,42]. These studies compared 108 participants in the IG versus 100 participants in the CG. The heterogeneity was moderate (I 2 = 38%). Older people in the IG had no change after intervention (SMD 0.60, 95% CI −0.12 to 1.31) in comparison to CG (p = 0.10). The heterogeneity of the comparison was high (I 2 = 82%) ( Figure 9). If we analysed only the studies with an objective measure of sleep efficiency [35,37], there were 59 older people compared in the IG versus 58 in the CG. Older people in the IG had a higher sleep efficiency post-intervention (MD 1.18%, 95% CI 0.86 to 1.50%) in comparison to CG (p < 0.0001) (Figure 10). The heterogeneity of the comparison was low (I 2 = 0%). The certainty of evidence, according to the GRADE methodology, was low.     Sleepiness: Only one study examined the Epworth sleepiness scale (ESS) post-intervention [40]. No significant benefits of exercise programs were shown for prevalent cases of poor sleep quality or sleep-wake behaviours that were evaluated using the ESS.

Discussion
This systematic review provides new information about the effects of exercise on sleep quality measured exclusively in the older adult population, showing that exercise improves sleep quality, particularly in facilities programs, and improves sleep efficiency measured with objective instruments. However, these results must be analysed with caution because there is a high heterogeneity in interventions and baseline characteristics of the population.

Effects on Sleep Quality
Our meta-analysis reports that exercise improves sleep quality. One possible explanation could be that physical activity regulates sleep patterns [34]. With increasing age, there are substantial changes in sleep quality and quantity, including slow-wave sleep, spindle density, and sleep continuity [45]. Furthermore, it seems that for older adults with sufficient mobility and no pre-existing disease, a structured physical activity program may help reduce insomnia [46].
Poor sleep quality is closely related to mental health problems in older adults. Although this research did not explore anxiety or depression as an outcome, exercise in older Sleepiness: Only one study examined the Epworth sleepiness scale (ESS) post-intervention [40]. No significant benefits of exercise programs were shown for prevalent cases of poor sleep quality or sleep-wake behaviours that were evaluated using the ESS.

Discussion
This systematic review provides new information about the effects of exercise on sleep quality measured exclusively in the older adult population, showing that exercise improves sleep quality, particularly in facilities programs, and improves sleep efficiency measured with objective instruments. However, these results must be analysed with caution because there is a high heterogeneity in interventions and baseline characteristics of the population.

Effects on Sleep Quality
Our meta-analysis reports that exercise improves sleep quality. One possible explanation could be that physical activity regulates sleep patterns [34]. With increasing age, there are substantial changes in sleep quality and quantity, including slow-wave sleep, spindle density, and sleep continuity [45]. Furthermore, it seems that for older adults with sufficient mobility and no pre-existing disease, a structured physical activity program may help reduce insomnia [46].
Poor sleep quality is closely related to mental health problems in older adults. Although this research did not explore anxiety or depression as an outcome, exercise in older adults has already been reported to have a positive effect in reducing depression and anxiety [32]. Another aspect to consider is whether the improvement in physical capacity can reduce the use of medications. Although some reports show no statistically significant association between the change in physical capacity and the use of benzodiazepines and z-hypnotics, a decrease in psychotropic drugs and antidepressants has been observed when physical capacity improves in community-dwelling older adults [47]. The high use of psychotropics, like benzodiazepines, may have detrimental effects on older adults (i.e., risk for fractures, falls, and premature mortality, among others) [48][49][50], and it is reported that only about half of the older population who take psychotropics respond to these medications [51].
The mechanisms that explain the improvement in sleep quality due to exercise in older adults are still not completely clear. One possible explanation is that exercise training provides a benefit on sleep quality by decreasing the sleep latency and the use of sleep medication in older adults [52]. Another possible mechanism is that during sleep there is a fluid accumulation in the neck, which causes an increase in the pressure on the upper airway, which can cause the onset of obstructive sleep apnoea. After exercise, there is a significant reduction in the amount of fluid in the neck [53,54]. Other proposed mechanisms include body temperature changes by effects on adenosine levels, cytokine concentration changes, increased energy consumption/metabolic rate, changes in mood/anxiety symptoms, growth hormone secretion, improved fitness level, and body composition change, among others [54][55][56][57][58].

Sleep Quality Assessment
Sleep quality is a concept that includes quantitative aspects of sleep and more subjective aspects, such as "depth" or "restfulness" of sleep [59]. The most used instrument was PSQI, an index created in the psychiatric field [59]. In our meta-analysis, we ob-served a change of −2.49 units in the PSQI. Limited studies evaluate the minimal clinically important difference (MCID) for this instrument, suggesting a change between 1.54 and −3 [60,61]; however, these data have not been determined for older adults. Due to the lack of universal acceptance of an MCID for PSQI, any statistically significant effect between groups at the end of treatment was considered important in this study.
Sleep is a complex phenomenon, and the literature suggests that objective and subjective indices measure different components of sleep in older adults [62]. Specifically, subjective sleep quality, according to the PSQI, appears to be poorly correlated with sleep measured by actigraphy or polysomnography, and the lack of relationship between objective and subjective sleep measures appears to be unrelated to age, gender, education, or cognitive status [35]. Therefore, both types of evaluation should complement each other. Based on the evidence, an objective and subjective assessment of sleep quality is suggested, particularly because among the selected studies, only one [35] concluded that a multimodal lifestyle could improve subjective, but not objective, sleep in older adults with mild cognitive impairment and poor sleep.
One of the outcomes analysed was sleep efficiency, which was objectively reported through accelerometry and actigraphy [35,37] and subjectively through one of the PSQI dimensions [32,38,42]. We found no differences between the control group and the intervention group. However, when analysing only the objective instruments, there was a significant difference in favour of the intervention. Some authors recommend using subjective data as an adjunct to actigraphic data in estimating total sleep time and sleep efficiency in sleep-disordered patients, especially those with disorders of excessive somnolence [62]. Therefore, articles that only report subjective measures should be analysed with caution.

Exercise Programs
Meta-analyses showed very high heterogeneity. This is probably because the populations are very different and range from healthy subjects to older adults with pathologies. However, one aspect that was even more heterogeneous was the intervention. Although most of the programs had a similar frequency (three to five times per week), the duration was highly variable, with programs that lasted between 8 and 12 weeks [36,43,44], up to programs that far exceeded 12 months [35,38,40]. On the other hand, work intensities were not reported in most of the studies. Although it is possible to work at a relatively known intensity in aerobic or strength exercise, it is not so easy to determine the intensity with activities such as pilates, healthy beat acupunch, or yoga, which was used in almost half of the studies [32,36,38,39,42,43].
The most common interventions included were yoga [32,36,38,39,43] and multicomponent exercise [34,40,44]; however, one of the interventions that caught our attention is acupunch [42]. Acupunch is a non-invasive practice where a natural parabola is produced by swinging of the relaxed wrist, elbow, and shoulder joints to direct cuffing and tapping of the fist/palm onto the targeted acupoint along the meridians to transport "qi" [63]. The acupunch itself cannot be considered an exercise or an incentive for physical activity. However, the authors describe the healthy beat acupunch as a modification of the original acupunch by adding the physical fitness guidelines for older adults and essential elements of a comprehensive exercise program for older adults [64].
When analysing by setting, we found that home-based programs (defined as at least 50% of their completion at home) were not effective in improving sleep quality [35,38,40], whereas programs carried out in a facility, such as a nursing home or community centre, were effective if they managed to improve the quality of sleep significantly [32,34,41,43,44]. One of the critical points that could explain the improvement in this type of setting is the degree of supervision. One point in which the programs in facilities agree is the close supervision and guidance in the exercise program [34,44]. Another interesting finding is that when analysing the duration of the programs, those that lasted less than 12 weeks were more effective than those that lasted more than 12 weeks. On the other hand, by adding the analysis by intermediate evaluation points, we were able to find that programs lasting between 12 weeks and 12 months were also effective. However, we believe that it is risky to carry out an analysis since programs older than 12 weeks are few and we think that with the increase in the sample size more conclusive results can be obtained.
An important point to consider is the type of physical training. Although most of the studies comply with the principles of a structured exercise program, with frequency, intensity, time, and duration, as usually happens, physical exercise programs are confused with those that encourage physical activity that consists of increasing moderate to vigorous physical activity (MVPA) [65]. These programs usually increase physical activity through walking. Although both programs show benefits in chronic diseases [65], the principles of physical training cannot be fully applied, so they cannot be analysed as a dose response.

Clinical Implications
This study shows that exercise has effects on sleep quality. Clinicians and decision makers can review these data and carry out exercise programs focused on older adults who report sleep problems with or without pathologies and who report the use of medications for the management of sleep disorders or to improve mental health. This may be particularly relevant at community care levels, such as nursing homes or primary care. On the other hand, it is recommended to carry out supervised programs that comply with the principles of training to achieve the best possible effect on the population.

Limitations
Our study has some limitations. Most of the studies included populations with different diseases or comorbidities, which may limit the extrapolation of the results and recommendations to the entire spectrum of older people, even though our results were statistically significant. The heterogeneous nature of populations implies that many subjects have different pathophysiological behaviours and conditions, a broad spectrum of severity, and the implication or impact that suffering from associated comorbidities may influence the magnitude of reported results. On the other hand, we tried to sub-analyse by health status; however, it was very complex since the definition of healthy is not uniform among the different authors and, therefore, it loses reliability. Additionally, the studies did not report adherence, which is a fundamental factor in determining the effectiveness of an intervention. Future studies must describe the adherence to exercise, since this factor must be analysed together with the results when evaluating the effectiveness of an intervention.

Conclusions
Physical exercise programs in older adults improve sleep quality, particularly in facilities, and sleep efficiency measured with objective instruments.