|Home | About | Journals | Submit | Contact Us | Français|
Dementia is a common syndrome in the geriatric population. Subsequent impairment of cognitive functioning impacts the patient’s mobility, ADLs, and IADLs. It is suggested that older persons with lower levels of cognition are less likely to achieve independence in ADLs and ambulation (1–2). Frequently, nursing home residents are viewed as too frail or cognitively impaired to benefit from exercise rehabilitation. Often, persons with Mini Mental State Score (MMSE) score below 25 are excluded from physical rehabilitation programs. However, Diamond (3) and Goldstein (4) concluded that geriatric patients with mild to moderate cognitive impairment were just as likely as cognitively intact patients to improve in functional abilities as a result of participation in exercise rehabilitation programs.
The objective of this study is to compare, through a meta-analysis endurance and strength outcomes of Cognitively Impaired (MMSE <23) and Cognitively Intact (MMSE >24) older adults who participate in similar exercise programs.
Published articles were identified by using electronic and manual searches. Key search words included exercise, training, strength, endurance, rehabilitation, cognitive impairment, cognition, Mini Mental State Exam (MMSE), older adult, aged, and geriatrics. Articles were included if the were from RCTs or well-designed control studies.
A total of 41 manuscripts met the inclusion criteria. We examined 21 exercise trials with cognitively impaired individuals (CI=1411) and 20 exercise trials with cognitively intact individuals (IN=1510). Degree of cognitive impairment is based on the reported MMSE score. Moderate to large effect sizes (ES = dwi, Hedges gi) were found for strength and endurance outcomes for the CI groups (dwi = .51, 95% CI=. 42–.60), and for the IN groups (dwi =. 49, 95% CI=. 40 –.58). No statistically significant difference in ES was found between the CI and IN studies on strength (t=1.675, DF= 8, P=.132), endurance (t=1.904, DF= 14, P=.078), and combined strength and endurance effects (t=1.434, DF= 56, P=. 263).
These results suggest that cognitively impaired older adults who participate in exercise rehabilitation programs have similar strength and endurance training outcomes as age and gender matched cognitively intact older participants and therefore impaired individuals should not be excluded from exercise rehabilitation programs.
Cognitive impairment in old age severely impacts people’s ability to sustain common occupational and social functions. Increased age is the greatest risk factor for the development of Alzheimer’s disease. Most of the older adults with Alzheimer’s disease are 65 years old or older, although younger older adults can also develop this fatal disease. With the increased incidence of cognitive impairment in old age, there are more and more older individuals facing the difficulties of living (either alone or institutionalized) with a debilitating level of mental function that hinders their ability to interact with others or perform simple activities of daily living (i.e. going to the bathroom or getting out of bed). Although, planned physical activities and rehabilitation programs could be an effective and beneficial approach to maintain and restore the health and independence of the growing cognitively impaired senior population, studies have shown mixed results about the influence of cognitive function on rehabilitation outcomes (3–20). Early studies argued that older patients with cognitive impairments could not benefit from rehabilitation programs (21, 22) and that cognitive impairment can be an important predictor of functional outcome and discharge destination (20) in elderly patients. As a result of these studies, some health professionals and payer sources have questioned the value of providing rehabilitation services to cognitively impaired patients (15–17, 23, 24). Other studies (25–29) reported comparable functional gains in elderly rehabilitation patients with and without cognitive impairments. It is not surprising that study rehabilitation outcomes are mixed, given the heterogeneous characteristics of the population such as the different diagnoses, healthy status, levels of cognitive impairments, injuries, rehabilitation settings, small sample sizes, and limited number of facilities. Heyn (30) found that many physical rehabilitation and exercise training studies excluded patients with significant cognitive impairments. The argument has been that cognitively impaired individuals are unable to follow instructions and will therefore have little to no benefit from the attention and time of a physical therapist. It is because of this stigma that persons with a Mini Mental State Score (MMSE) score below 24 are often excluded from rehabilitation programs.
However, Diamond (3) and Goldstein (4) concluded that geriatric patients with mild to moderate cognitive impairment were just as likely as the cognitively intact patients to improve in functional abilities because of participation in physical rehabilitation. There is little information about the effects of exercise training in nursing home residents, given that a small percentage of nursing home residents are eligible to receive rehabilitation services. Moreover, it is suggested that more than half of nursing home residents have some degree of cognitive impairment with a significant percentage being undiagnosed (31). Another recent published study found significant functional gains from acute rehabilitation when investigating cognitively impaired stroke patients (32). Chronic health problems associated with the geriatric population are due, in part; to age-related loss of muscle mass and strength (sarcopenia) associated with physical inactive and sedentary lifestyles leading to severe chronic and acute frailty levels (33). Recent research has also shown that midlife obesity, measured by body-mass index (BMI), increases the risk for dementia, Alzheimer’s disease, and neurodegenerative disorders (34). Whitmer and colleagues (34) showed that midlife BMI is a strong predictor of dementia, conferring a 75% increased risk among those with the highest BMI. Mounting evidence suggests that not only should we be developing effective exercise recommendations and guidelines for the cognitively impaired older adult at risk for increased disability and comorbidity but also providing exercise training as a standard of care for this population.
The objective of this study is to compare through meta-analysis method endurance and strength outcomes of Cognitively Impaired (CI=MMSE <24) and Cognitively Intact (IN=MMSE >24) older adults who participate in similar randomized exercise trials. Such an analysis permits us to aggregate results across the extant literature to provide a quantitative answer concerning the relative benefits of exercise for impaired and non-impaired elderly.
We identified published articles using both electronic and manual searches. Electronic searches were conducted by using several databases that include the following: Abstracts International, Ageline, CINAHL, Cochrane Library, Dissertation, Educational Resources Information Center, MEDLINE, PEDro, PsycINFO, PsycLIT, PubMed, and Sport Discuss (SIRC/CDC). An extensive manual search and cross-referencing from review and original articles was also performed. Key search words included exercise, training, strength, endurance, rehabilitation, cognitive impairment, cognition, Mini Mental State Exam (MMSE), aged, older adult, and geriatrics.
In order to be included in the summary effect size analysis, the article must described a study that 1) is a randomized trial that includes a nonintervention control or comparison group or control or comparison period; 2) include only subjects older than 65 years; 3) reports a baseline Mini-Mental Status Examination (MMSE) score of less than 25 for cognitively impaired subjects (subjects may alternatively be diagnosed by a physician as having some degree of cognitive impairment or preexisting diagnosis of dementia reported by the original author) and greater than or equal to 25 for cognitively intact subjects (cognitive assessment may be performed using another assessment tool); 4) include an exercise program or form of rehabilitative exercises, physical activity, fitness, or recreational therapy; 5) report means, standard deviations (SDs), t test or F test, and n values; 6) have a minimum of 5 subjects in each group; 7) have at least 1 dependent variable from health-related physical fitness measures (cardiovascular, endurance, strength, flexibility, body mass index [BMI]); and 8) be a peer reviewed journal article published in English and indexed between January 1970 and December 2006.
Studies were excluded if they were not written in English, if they did not contain enough statistical information to compute an effect size (ES), or if the outcomes did not involve strength and endurance physical fitness measures. Studies that were based on qualitative designs or narrative case reports, or involved 5 or fewer subjects were also excluded.
The methodologic quality of the study was assessed by two reviewers independently (PH, KJ) using the UTMB/TLC Interventions Trial Quality Form (30). The form included information on randomization, institutional review board approval, subject characteristics, blinding, sampling method, measurement, data collection, intervention, interpretation of results, and reporting of biases and limitations. It also included a grade for internal validity, external validity, and level of evidence generated. Assessment scores were 1 for yes, 0 for no and for unclear. The range of possible scores was 0 to ≥22. Very high-quality studies were defined as those with a score of 22 or greater; high-quality studies were those with a score of 19 to 21; medium-quality studies were those with a score of 16–18; and low-quality studies were those with a score of 15 or less. Afterwards, scoring differences of opinion were discussed until consensus was reached.
The original electronic search yielded more than 500 articles (Figure 1). There was, however, significant overlap among the articles. To ensure comprehensive coverage, the ancestry search approach was used. The procedure for including studies in the final sample included reading each article in the sample to assess the presence of the inclusion and/or exclusion criteria. After exclusion, 41 trials were eligible for analysis.
The quantitative method of review proposed by Glass (35, 36), involves an estimation of a population ES, obtained from ES values generated from individual studies. Individual ES values were aggregated to a summary effect size (SES). To decide whether an SES was statistically significant, a fixed effects model was used (37). The SES is the weighted mean of the unbiased individual ES values, calculated according to the fixed effects model. Most studies reported sufficient statistics to calculate the standardized mean difference. Using Hedge’s gi(38), each individual study ES was calculated by dividing the difference between the experimental and control groups by the estimated pooled standard deviation (SDi) of the baseline outcome measure in the treatment and control groups. For each outcome, we reported Hedge’s gi and the 95% confidence interval (CI). The ES values were coded such that positive numbers always reflected improvements in performance, and negative numbers reflected deteriorations in performance. Basing decisions on Cohen’s “rules-of-thumb” interpretation of ES results (37), we adopted the 49 to .79; and large > .80 or greater. Following standardized mean difference ES for this study interpretation (39): small < .49; medium >.
The overall approach to ES calculations and moderator variables analyses were conducted using computer programs, Excel 2003 software and SPSS software, version 11.0. The ES values and summary effect sizes (SES) were calculated using the statistical program called Comprehensive Meta-Analysis software package (Biostat Inc). This computer program computes standardized mean difference ES and correlation coefficients (r) from summary statistics, such as means and SDs, t tests, F tests, homogeneity, and frequencies. SES statistics were computed through the Hedge’s formula (38), as previously described.
Intervention characteristics (exercise modality and outcome measures) and the subject’s age, gender, and levels of cognitive impairment were characteristics of primary interest (Table 1). Characteristics that were collected among the exercise intervention designs were exercise modality, duration, intensity, and frequency. When available, MMSE scores also were coded. These training characteristics and outcome measures were chosen a priori as variables that may influence the primary study outcome (ES).
In the event that important information about methods or relevant statistics was missing from a study (i.e., sample size, age, and exercise mode), that study was dropped from further consideration in the analysis. Fortunately, the selected studies provided most of the information needed for the statistical analysis. Two studies did not report session duration and MMSE scores although the author reported the sample as having cognitive impairments, dementia, or Alzheimer’s disease per clinical diagnosis.
A total of 41 studies met the inclusion criteria and were therefore included in the meta analysis (Figure 1). We examined 21 exercise trials with cognitively impaired (CI) individuals and 21 exercise trials with cognitively intact (IN) individuals (degree of cognitive impairment is based on the MMSE score). For those studies without a reported MMSE score or those studies in which an alternate cognitive assessment tool was used, we assigned a MMSE score >23 if the subjects were said to be cognitively intact and a mean MMSE score of <24 if the subjects were said to be cognitively impaired. For those studies in which the cognitive assessment tool was not reported, they were included if a clinical diagnosis of cognitive impairment or dementia was reported by a physician
Two main groups were analyzed: one group that contains subjects that have mild, moderate, or severe cognitive impairment and another group that contains subjects that have no cognitive impairment (cognitively intact). Of these 2 groups, subjects were randomized (for most studies) into either exercise intervention groups or control (non-exercising) groups. The subjects that were randomized to the exercising groups were subjected to either endurance or strength training regimens and sometimes to a combination.
Table 1 shows the characteristics of the 41 studies. It includes total number of subjects (N Total), the total number of subjects in the analysis (N Included: subjects included in the summary effect size analysis based on strength and endurance collected data), the total number of subjects who were randomized to either control groups (N. Control) or exercise/intervention groups (N. Experimental), and the mean age and mean MMSE scores for the included subjects. Tables 2 reports the data on the exercise trials for intact older subjects (IN= MMSE ≥24) while table 3 reports the 21 studies on subjects that are cognitively impaired (CI; MMSE <24). The Intact (IN) and Cognitively Impaired (CI) studies data are described in detail in tables 2 and and3.3. Tables 2 and and33 also describe the exercise interventions (e.g. endurance, strength effects) and moderator factors such as frequency (sessions/week) intensity (minutes/session), and duration (total number of weeks). Thirty-nine of the included studies were randomized controlled trials while the remaining two trials were placebo-controlled without a randomization procedure. The data on specific characteristics (mean subject age, MMSE scores, female/male ratios, exercise frequencies, intensities, and durations, mean study quality scores, etc.) of the included studies is summarized in table 1.
The included studies were published between 1974–2004 and received a mean quality score of 18 with a SD of 2 and a range of 13–23. There are 2921 older adults included in the analysis and although not all the studies reported male/female subject data, the studies gender sample were based approximately on 815 males and 2090 females. The exercise groups were compromised of a total of 1472 subjects while 1449 subjects served as controls. The mean age of the included subjects is 81 with a SD of 5 years and a range of 68–91. The mean MMSE score of the included subjects is 21 with a SD of 7 and a range of 6–29. This represents, however, the MMSE score range for the entire subject population and does not represent the MMSE scores of the cognitively impaired versus the non-impaired groups. Of the non-cognitively impaired group, the mean MMSE score is 28 with a SD of 1 and a range of 26–29. For the CI studies, the mean MMSE score is 16 with a standard deviation of 6 and a range of 6–23. The exercise interventions included a range of endurance measures (such as timed-up-and-go testing, aerobics, gait distance assessment, heart rate evaluation, cardiovascular system analysis, and related measurements) and strength training measures (such as arm curl, knee extension, grip strength, leg extension, quadriceps force, and related measures). The mean duration of the exercise training for all included studies is 16 weeks with a SD of 8 weeks and a range of 2–40 weeks. The mean exercise frequency is 3 sessions per week with a SD of 1 and a range of 1–6 sessions per week. Finally, the mean intensity of the exercise sessions is 50 minutes per session with a SD of 16 minutes per session and a range of 20–90 minutes per session.
Table 1 also shows the mean effect size for all the exercise intervention groups that underwent strength training. The ES for the older participants who were tested for strength training outcomes is a large effect (0.9+0.8) based on this study ES interpretation (Lipsey 2001): small <.49; medium >.49 to .79; and large > .80 or greater. This indicates that, overall, there is a net beneficial effect for individuals who participated in strength exercise training just as there was a moderate effect net benefit for participants who were tested on endurance outcomes (0.6+ 0.5). Again, these results reflect the combined effect sizes for cognitively impaired and non-impaired subjects. How do the results compare between these two groups? The mean endurance ES for the cognitively impaired individuals is moderate range (0.5 + 0.2), as well as for the intact individuals (0.7+ 0.6). In regards to the strength outcomes, the mean ES for the cognitively impaired individuals resulted in a moderate effect (0.7+f 0.2). Meanwhile, the intact individuals showed a large ES on strength outcomes (1.0 + 1.0.)
We also found a larger ES variance in the IN studies (figure 2) as compared to the CI studies (figure 3), and we assume that this variance is due to the different types of strength training and workloads (intensity) implemented in the exercise designs from some of the IN studies included in this meta-analysis. In general, the trend is that for healthy and cognitively intact individuals the exercise paradigm is based on high performance gains and for the CI individuals the exercise paradigms are designed more cautiously and with lower levels of intensity, however these different training doses did not result in statistical significant ES differences between the selected studies. Detailed information about the exercise paradigms is described on tables 2 and and3.3. We also combined the endurance and strength outcomes to compare SES of the cognitively intact and impaired adults who participated in similar exercise programs.
We found similar moderate effect sizes (ES = dwi, Hedges gi) when we combined the strength and endurance outcomes for the CI studies (dwi = .51, Effects=31, 95% CI=.42–.60), and for the IN studies (dwi =. 49, Effects = 33, 95% CI=. 40 –.58) (table 2). No statistically significant difference in ES were found between the CI and IN studies on strength (t=1.675, DF= 8, P=.132), endurance (t=1.904, DF= 14, P=.078), and combined strength and endurance effects (t=1.434, DF= 56, P=. 263). Most of the studies (78%) received a medium quality score of 18 points (SD + 2). The remaining studies received a high quality score.
The results support the hypothesis that there is a significant endurance and strength training benefit for cognitively impaired elderly subjects who participate in exercise trials and related physical rehabilitation programs. In addition, the findings support that cognitively impaired older adults have similar strength and endurance outcomes as cognitively intact participants. Therefore, there does not appear to be a significant deficiency in the ability of the cognitively impaired subjects to attend and adhere to an exercise program whether it is an endurance-based program and/or strength based program).
It has been assumed that individuals with cognitive impairment are considered too frail and too impaired to benefit from exercise rehabilitation (2). In addition, it has been assumed that cognitively impaired individuals are unable to follow simple directions and are therefore unable to carry out simple instructions during a physical therapy session; thus making physical rehabilitation sessions less effective and diminishing the positive benefits that could otherwise be seen if cognitive impairment was not present (21, 22). Because of this, many individuals do not have access to physical exercise programs as part of their therapeutic management or are simply placed on a skilled nursing facility (40). Frequently, nursing home residents are viewed as too frail or too cognitively impaired to benefit from rehabilitation programs (1, 19).
However, as our data indicates, cognitively impaired individuals benefit just as much as cognitively intact individuals when provided with opportunities to participate in exercise rehabilitation programs and therefore should not be excluded from rehabilitation services considering their health need to improve the physiological factors associated to muscle mass and physical function as well as o slow down frailty.
We also found significant outcome gains on both strength and endurance outcomes supporting the need to develop exercise rehabilitation programs that target multi-level systems and functions, as is the case for good cardiovascular function and body composition levels for this population. Unfortunately, these conclusions are based mostly on the performance of patients with mild-moderate CI and not those with severe CI. More research is needed to directly compare these two groups and to consider the need for different exercise guidelines and recommendations for varying degrees of cognitive impairment. More research is also needed to define the predictors of maximal benefit and to determine the cost of implementing exercise rehabilitation programs for the older cognitively impaired adult.
In conclusion., based on our meta-analysis results as well as the controlled randomized clinical trials investigating the effects of exercise training on cognitively impaired older adults (3, 4, 32), there is a need to evaluate best rehabilitation practice protocols and reconsider current geriatric healthcare policy and practice such that cognitively impaired older individuals should be provided with physical rehabilitation as a standard of care practice as well as with opportunities to participate in community-based exercise training programs with the aim to improve physical health. In addition, the geriatric exercise rehabilitation program should be aimed to positively affect participants’ quality of life. By doing so, there should be significant functional improvement in activities of daily living in cognitively impaired individuals and a decline in the costs associated to caring for the disabilities of the cognitively impaired older adult.
Suppliers: a. SPSS Inc, 233 S Wacker Dr, 11th Fl, Chicago, IL 60606. b. Biostat Inc, 14 N Dean St, Englewood, NJ 07631.