Effects of Short-Term Exercise Interventions on Behavioral and Psychological Symptoms in Patients with Dementia: A Systematic Review

INTRODUCTION

With 75 million people worldwide expected to suffer from dementia in 2030 and 132 million in 2050, dementia presents one of the biggest global public health challenges at present and the future [1]. The decline of cognitive functioning is the common denominator of different dementia syndromes. In the course of the disease, behavioral disturbances affect almost all patients [2]. Behavioral and psychological symptoms of dementia (BPSD) include affective disturbances with depression and anxiety, psychotic features with delusions and hallucinations, hyperactivity including irritability, and aggression or euphoria [3]. Caregivers perceive these symptoms as a much greater burden and challenge than the cognitive decline itself [4]. Periods of exacerbated BPSD are one of the leading causes for admission to specialized acute dementia care units in hospitals, geriatric psychiatry hospitals, or nursing homes.

The treatment of BPSD represents a main target for pharmacological and non-pharmacological interventions in acute dementia care units. Due to potential severe side-effects of antipsychotic medication [5], non-pharmacological approaches are increasingly being implemented for the treatment of BPSD [6]. Scherder et al. [7] describe a positive correlation between physical inactivity and agitation in patients suffering from dementia. Thus, physical activation should be considered more often, especially in acute dementia care settings, while the use of sedative medication and physical constraints should bereduced.

According to the Cochrane review by Forbes et al. [8], focusing on the effects of exercise programs for people with dementia, there is growing evidence that training of activities of daily living has substantial positive effects in these patients. With regard to the effects on BPSD, Forbes et al. [8] conclude that currently, there is not sufficient evidence to assess the effects of exercise on BPSD as the existing studies lack methodological quality. On a closer look, there are positive effects of exercise on some BPSD: Eggermont and Scherder [9] report positive effects of walking programs on affective behavior, when carried out several times a week for a minimum of 30 minutes. Thune ´-Boyle et al. [10] point out the positive effects of exercise on depressed mood, agitation, and restlessness in people with dementia. A current systematic review and meta-analysis by Souto Baretto et al. [11] confirms these results: structured exercise programs positively affect depression and aberrant motor behavior in patients suffering from dementia.

Though exercise therapy is implemented in routine care, there is a lack of evidence-based recommendations for physical activation programs in inpatient dementia care. Existing reviews focused on patients with mild to moderate dementia, mostly in community-dwelling or usual nursing home settings. These results cannot be translated into acute dementia care units (DCU) of geriatric hospitals, geriatric psychiatry hospitals or nursing homes. Patients are most often admitted to dementia care units (DCU) due to exacerbation of BPSD. From a clinical point of view, this exacerbation can often at least partly be explained by minor somatic factors (e.g., mild dehydration, hyponatremia, urinary tract infection, etc.) that further destabilize psychopathologically fragile demented patients. From our experience, these minor somatic complications do not prevent patients from participating in exercise programs. Both long-term nursing homes and community-dwelling settings cover different dementia populations as compared to acute DCU: patients with acute exacerbated BPSD in the acute DCU versus patients with less compromising or less acute BPSD in community-dwelling or long-term nursing home care. Both patient populations may respond differently to structured physical exercise programs [10].

Additionally, current literature reviews have excluded trials with a program period shorter than three months [12]. As the usual length of stay of patients with dementia in specialized geriatric psychiatry wards ranges from two to six weeks [13], results from exercise trials with a maximum of three months should be considered separately. This paper aims to give an overview and analysis of short-term exercise trials in acute dementia care settings. Thereby the main focus is on the effects on BPSD.

MATERIAL AND METHODS

A protocol of this systematic review has been published in the PROSPERO database (CRD42016033399). The reporting of this investigation has been conducted according to the PRISMA statement [14].

RESULTS

The systematic search resulted in 1,582 hits (see Fig. 1). From this sample, 97 records were screened and 18 full-texts were assessed for inclusion.

Finally, five trials were included in this review: three RCTs [19–21] and two non-randomized trials [22, 23].

Three of the included trials were conducted in the USA [21–23], one in Israel [19], and one in Australia [20]. Patients were recruited from acute geriatric or geriatric psychiatry hospital care in two of these trials [19, 21] and three of these trials were conducted in special dementia care units in nursing homes [20, 23]. In total N = 206 patients were included in these five trials: 153 women (74%) and 53 men (26%). The mean age of the included patients ranged from 64 (±6) years [19] to 85 (±6) years [23]. The levels of the patients’ cognitive impairment, as measured by the Mini-Mental State Examination mean value, ranged from 4.6 (±4.9) points [21], indicating severe cognitive impairment, to 22.3 (±4.5) points [19], indicating mild to moderate impairment of cognitive functioning.

Within this sample, three trials focused on the broad spectrum of BPSD measured by using complex instruments [20–22], whereas two of these trials focused on depressive and affective symptoms measured with specific instruments [19, 23]. Aman and Thomas [22] concentrated on both BPSD and depressive symptoms as measured with the Cornell Scale for Depression [24]. Only one of the included trials used one of the defined principal summary measurement, the CMAI [22], in order to determine the effects on BPSD. The NPI, as the most established instrument for measuring BPSD in this field of research, was not used in the included trials. Most of the trials applied proxy-rating instruments, with one trial using self-rating instruments for the assessment of affective symptoms [19]. The evaluation periods ranged from ten minutes of a single day [23] to a retrospective rating of the preceding eight hours of a nursing shift [22] as measured by using the Pittsburgh Agitation Scale (PAS) [25] or two weeks respectively measured with the CMAI [22]. None of the five included studies differentiated between exacerbated BPSD and acute delirium. Further information about the trial-setups is summarized inTable 2.

The duration of the exercise interventions ranged from three to twelve weeks. All five trials conducted three exercise sessions per week. The length of each session ranged from 30 to 45 minutes. Three of the included trials focused on endurance and strengthening exercises [19, 23], while two conducted a more general physical activation program [20, 21]. These exercise programs most commonly used mild intensity physical activation programs. One trial explicitly avoided higher intensities [19]. Three exercise programs were conducted in a sitting position for most of the training sessions [19, 23]. However, it remains unclear to what extent the patients followed these exercise programs due to inappropriate reporting of the exercise adherence in four of the five trials. Edwards et al. [23] used an advanced way of assessing and reporting the exercise adherence, implying not only the patient’s presence, but also his or her active participation in the exercises. 70% of all patients participated in more than 75% of the exercise sessions, implicating a high adherence to the exercise protocol. Detailed information about the applied interventions, the adherence to the intervention protocols, and the results on BPSD are summarized inTable 3.

The reported results on BPSD of each trial are specified within the following sections:

Aman and Thomas [22] reported pre-test mean values of 5.8 (±4.8) points and post-test mean values of 4.5 (±3.7) points on the PAS [25] ranging from 0–16 points (0 meaning no agitation). These differences were reported as significant (p < 0.05). The results on the CMAI [18] with 29–203 points (29 points = no agitation) were reported as follows: pre-test mean values 51.7 (±21.4) points and post-test mean values 52.4 (±28.0) points. There were no significant differences (p = 0.85). The patients’ depressive symptoms were assessed by the Cornell scale for depression in dementia [24] ranging from 0–30 points (0 meaning no depression). The authors reported a mean value of 7.0 (±7.9) points for the pre-test and 6.9 (±7.9) points for the post-test. These results showed no significant differences (p = 0.86).

The group of Hollimann et al. [21] used the second subscale of the Psychogeriatric Dependency Rating Scale [26] for measuring dementia-related problem behaviors like wandering, active aggression, and restlessness. For the study group, a mean of 7.4 points in the pre-test and a mean of 6.0 points in the post-test was reported. The control group showed a mean of 5.4 points at the pre-test and a mean of 4.6 points at the post-test. The authors reported a significant between-group difference of p = 0.03 at post-tests and concluded that the control-group was “slightly” better than the study-group at the post-test. There were no significant between-group differences at baseline [21].

The research group of Netz et al. [19] reported the results on the Geriatric Depression Scale [27]: within the geriatric sample, the exercise group showed a mean of 5.0 (±3.31) points at the pre-test and a mean of 4.0 (±3.27) points at the post-test. The geriatric control group showed a mean of 3.6 (±2.99) points at the pre-test and a mean of 2.3 (±1.6) points at the post-test. The variance analysis (ANOVA) revealed no significant differences, neither for group main effect, time main effect nor for group×time interaction. The psychogeriatric sample showed a mean of 12.87 (±3.31) points at baseline and a mean of 7.37 (±4.43) at post-test for the exercise group. The control group of the psychogeriatric sample showed a mean of 6.62 (±5.15) points at baseline and a mean of 4.54 (±4.53) at the post-test. The ANOVA showed a significant time main effect (F(1; 14) = 25.50; p < 0.01) a significant group main effect (F(1; 14) = 5.44, p < 0.05) and a significant group×time interaction (F(1;14) = 4.54; p < 0.05). Unfortunately Netz et al. [19] did not report any results of subsequent post-hoc tests that would allow deriving an interpretation based on the ANOVA. Based on the descriptive results of the psychogeriatric sample, Netz et al. [19] conclude higher decreases in the exercise group (Δ –5.50 points) as compared to the control group (Δ –2.08 points). Additionally the Brief Psychiatric Rating Scale [28] was used for the rating of the behavioral symptoms. The authors concluded that there were no differences detected, but no results were reported that allow following this conclusion.

Edwards et al. [23] reported the effects of the exercise intervention on negative affect symptoms by the Philadelphia Geriatric Center Apparent Affect Rating Scale [29]. For the anxiety symptoms, pre-test mean values of 1.8 (±1.2) points and 1.2 (±0.5) points for the post-test were reported. These differences were reported as significant (p = 0.01). For depressive symptoms, Edwards et al. reported pre-test mean values of 1.5 (±0.9) points and post-test mean values of 1.1 (±0.5) points. These differences were reported as significant (p = 0.03). Within the domain investigating the anger symptoms, the research group reported mean values of 1.1 (±0.3) points in the pre-test and 1.3 (±0.7) points in the post-test. These differences were reported as not being significant (p = 0.51).

Within the trial of Stevens and Kileen [20], the patients’ behavioral and psychiatric problems were assessed by the Revised Elderly Persons Disabilities Scale [30]. The research group reported no pre- and post-test means. The following differences between pre- and post-test of the behavioral problems were reported: group I (control group, no intervention) Δ 1.7 points (p = 0.98); group II (social visit group) Δ 2.1 points (p = 0.63), and group III (exercise intervention) Δ 4.6 points (p = 0.31). The between group analysis showed no significant differences (p = 0.42). The following differences were reported for the domain rating the psychiatric problems: group I Δ 0.4 points (p = 0.75); group II Δ 0.8 points (p = 0.42); and group III Δ –2.9 points (p = 0.64). The reported between-group analysis for the psychiatric problems showed no significant differences (p = 0.82). No significant differences within- or between-group comparison of the sub-domains related to age, level of dementia, or gender have been detected.

Due to the different trial setups, rating scales, and interventions applied, the results of the five included trials cannot be compared or aggregated in detail. However, the following tendencies can be derived: Two trials focusing on the broad spectrum of BPSD reported significant reductions and differences as compared to the pre-test and the control groups [21, 20]. One trial reported indifferent results on BPSD— results focusing on long-term trends (CMAI) did not change significantly, whereas the short-term scale (PAS) showed significant decreases in comparison to the pre-test clinical situation [22]. Due to insufficient information reported, the results of Netz et al. [19] cannot be included in the synthesis of effects on BPSD. Out of the three trials assessing depressive symptoms using an extra scale, two have reported no differences in pre-post-measurements [19, 22], while one trial reported significant reduction of depressive symptoms in pre-post comparison [23]. No deteriorations of BPSD and depressive symptoms were identified within the five trials and no adverse events were reported.

The risk of bias assessment (see Fig. 2 and Supplementary Material) indicates a very wide variety of methodological issues. One high-quality study [23] with an overall low risk of bias has been included in our analysis together with trials exhibiting severe methodological problems. For example, we have found an increased risk of bias due to selective reporting [20] as well as potential bias resulting from a high amount of drop-outs (≥20%) from the interventions [22] and effects of the co-interventions that have not been reported appropriately [21].

DISCUSSION

The primary objective of this systematic review was to give an overview and analysis of the effects of short-term exercise interventions on BPSD in acute dementia care settings.

Within the process of reviewing, we only found five trials meeting the inclusion criteria. This indicates that only few short-term exercise trials have been conducted in acute dementia care settings. This seems somehow surprising as exercise therapy and physical activation programs have been implemented in routine dementia care since some time. Furthermore, clinicians increasingly use non-pharmacological methods for the treatment of BPSD, as there is growing evidence of severe adverse effects like increased mortality and increased cardiovascular risk under psychopharmacological (i.e., neuroleptic) treatment [6].

The methodological heterogeneity among the five included trials, especially concerning the measurement of BPSD, limits a detailed synthesis of specific effects on BPSD. It is important to note that none of the five included trials controlled for the concurrent presence of an acute delirium. In general, delirium in dementia and exacerbation of BPSD without delirium are clinical conditions that are very difficult to differentiate. Anyhow, the pharmacological as well as non-pharmacological treatment is basically very similar in both conditions, and accompanying somatic treatment is key to successful treatment in both groups.

In general, the analysis of the included trials indicates positive effects of exercise programs on BPSD. To which extent short-term exercise interventions positively affect single sub-domains of BPSD still remains rather unclear due to the fact that most of the included trials have globally measured BPSD. The research group led by Edwards et al. [23] is the only group who not only report global results on BPSD, but also performed sub-domain analyses. Their analysis indicated significant reductions in “anxiety” and “depression” within this group, but not in the BPSD “anger” domain. None of the trials found deterioration of BPSD due to physical exercise.

Within the field of exercise science and dementia care, this review is the first to give an overview of the effects that short-term exercise programs have on BPSD in acute dementia care settings. Current systematic reviews and meta-analyses [8, 12] did not include any of the trials that were included and analyzed in this review. Most of them excluded these trials due to the shorter duration of the intervention [12] or due to insufficient reporting [8]. The general results of these studies are comparable with the results of this review, with physical activation having only small effects on the BPSD scores. The results on affective symptoms are in line with Potter et al. [12]: out of the three trials investigating the effects of exercise interventions on depressive symptoms, only one reported a significant reduction, whereas Potter et al. [12] reported significant decreases in one out of four trials. Eggermont and Scherder [9] found sustained walking programs to be beneficial for affective symptoms in dementia. These results can hardly be compared as the trials included within this systematic review did not apply any walking programs. As the other systematic reviews and meta-analyses focus on physical exercise programs with a longer duration, a detailed comparison of results is notpossible.

Scherder et al. [7] hypothesize a close link between the patients’ levels of physical activity and the incidence of behavioral disturbances. This link may explain the reported effects on the global scores of BPSD within the trials that we have analyzed. Rimmele et al. [31] reported increased cortisol responses to stress in physical inactive subjects as compared to the physical active age- and sex-matched subjects. A higher stress response in physical inactive patients could be an underlying explanation for the observed positive effects of exercise programs on BPSD but has to be investigated and confirmed by empirical investigations in elderly and especially in patients suffering from dementia. According to Scherder et al. [7], participation in exercise interventions could increase the patients’ level of physical fitness and may subsequently counter further neuropathological decline and deterioration of BPSD. A transfer of the hypothesis the more physical inactivity, the more agitation in dementia [7] into the acute dementia care setting leads to the question about how to promote physical activity and how to conduct exercise interventions within these settings and with these patients. The included trials have applied strength exercise with weights of max. 1 lb (0.45 kg), endurance exercises via walking in groups, psychomotor activation, and functional exercises for improving basic motor skills. According to Littbrand et al. [32], a combined functional weight-bearing exercise approach seems to be applicable for patients with dementia regarding exercise attendance and adverse events. Following the recommendations of Littbrand et al. [32], exercises for people with dementia should be performed task-specific and should challenge the individual’s physical capacity.

The fact that most of the included trials applied low intensity exercises leads to the question whether the results on BPSD might have been caused by social stimulation during the exercise sessions and not by the exercise itself. Stevens et al. [20] and Netz et al. [19] reported no between-group differences in BPSD before and after an exercise intervention. In both trials, the control group received a social stimulation program. Therefore, the social stimulation aspect might have contributed to the non-significant differences in BPSD in these intervention groups. The intensity of the exercise programs is a key aspect when investigating the effects of exercise programs on BPSD. Due to insufficient reporting of the trials, it was not possible to compare the outcomes of low intensity exercise programs with the trials that conducted more intensive exercise programs. With regard to the training in the included trials, we found exercise programs with light intensity where patients were mostly in a sitting position. This implies a very careful way of physical activation. One trial explicitly avoided to increase the targeted exercise intensity [19]. These seated exercises with very low intensity have been applied as control intervention for investigating the effects of more intensive exercise programs, depending on the patients’ physical capacities [33]. The studies included in this review applied low intensity exercise sessions in a patient population where more intense physical activation might have been feasible. However, it is likely that low intensity exercise may present a low threshold for participation. Furthermore, it requires less personnel and time resources. Therefore it is more easily applicable as compared to high intensity exercise programs. Referring to the guidelines of the American College of Sport Medicine [34], exercise intensity is the key factor in geriatric sports medicine. Especially relevant to these acute care settings, it is well known that high-intensity exercises are more effective on depressive symptoms in depressed geriatric patients [35]. The differences in the exercise protocols regarding exercise contents and very low exercise intensities might be key aspects explaining the reported differences in the effects on BPSD.

As observed by Littbrand et al. [32], only few of theincluded trials of this systematic review reported the adherence to the exercise protocol. Furthermore, the achieved intensity and the grade to which the participants have conducted the exercises were not been reported in any of the included studies. Assessing and reporting these details of the exercise protocol may be difficult taking into account the patient population and the overall research settings: However, it is essential to answer this research question. Edwards et al. [23] not only assessed the physical presence but furthermore reported the extent to which participants took part in the exercise sessions. Using such an approach may further contribute to answer the research question of dose-effect relationships in the field of exercise science in people with dementia.

The analysis of the methodological quality and risk of bias within the included studies reveals a wide heterogeneity. Trials with adequate methodological quality and innovative instrumental set-ups [23] have been included together with trials showing methodological problems and higher risk of bias. These problems were mainly due to incomplete data reporting [21] and selective reporting of the results [20]. Hereby, the reporting of exercise adherence turned out to be a problem (see Table 3). Most of the trials have reported to take adherence protocols, but only few of them have reported them appropriately. The methodological heterogeneity of the study sample limits the results and conclusions of this systematic review. RCTs and non-randomized studies of interventions were regarded within the study selection and data analysis. The final study sample of five trials including N = 206 patients additionally limits the findings especially their transferability into routine acute dementia care settings.

Taking these limitations into account, the following general implications for acute dementia care settings can be stated: 1) Promoting physical activity represents an achievable and worthwhile treatment aspect in acute dementia care settings; 2) Applying structured short-term exercise interventions in acute dementia care settings offers a feasible and safe treatment approach for BPSD; 3) No side-effects of physical exercise in dementia have been reportedso far.

Given the lack of dementia specific evidence-based guidelines, the well-established American College of Sport Medicine recommendations of 150 minutes of exercise in moderate intensity per week could be a well applicable goal. The implementation of this goal requires adapted ways of structuring physical activation in acute dementia care settings. Therefore, studies using low-threshold and interdisciplinary physical activation concepts like geriatric psychiatry in motion [36] may in the future provide evidence to tailor and structure short time interventions in acute dementia care. In order to be able to define appropriate recommendations for physical activation and exercises in acute dementia care settings, more RCTs are required.

It is important to reliably assess and rate BPSD. Therefore psychopathometric scales like the NPI or CMAI should further be promoted. Future studies should not only focus on global scores of BPSD but should also look at the different sub-domains. Technical innovations might be used in the future to objectively measure aspects of BPSD [37–39]. Furthermore, the effects of exercise interventions on the use of on-demand and routine psychotropic medication should be taken into account. Designing the interventions for exercise trials in acute dementia care settings should follow well-established results and findings in the field of geriatric exercise science. This includes especially the selection of exercises and the scheduled intensity [32, 34]. Investigating the effects of exercise interventions on BPSD should also take into account the basic level of the patients’ physical activity. This is an important issue, especially for acute dementia care settings, as patients with different levels of present physical activity (e.g., patients with wandering phenomenon or overall physical inactive patients) might respond to exercise interventions in different ways with respect to their former physical activity pattern.

CONCLUSIONS

Increasing patients’ physical activity through structured exercise programs presents a worthwhile approach for the treatment of BPSD in acute dementia care settings. Given the scarcity of available studies, more randomized controlled short-term exercise trials in acute dementia care settings are needed to define appropriate exercise recommendations for clinicians treating these patients.