This study examined the efficacy of exercise programs for individuals with intellectual disability (ID) based on experimental designs. Multiple databases were searched from inception up until March 2019. Randomized control trials were eligible for inclusion if they (a) included a population with ID; (b) used an exercise intervention; and (c) included performance, cardiovascular health, and/or psychological measures. All data were pooled using random effects models of standardized mean differences. The review included 18 studies which represented data from 799 individuals with ID. The largest effect was seen in relation to anxiety and depression symptoms (SMD = −3.07). This study represented the first comprehensive analysis on this topic area and illustrated that exercise can play an important therapeutic role for this population.
The benefits of physical activity (PA) are numerous and have been so widely accepted that even the World Health Organization (WHO) presents guidelines for PA. Although the evidence that PA is good for both mental and physical health and that it may help improve overall quality of life is strong, according to the WHO, few people actually engage in sufficient PA on a regular basis (World Health Organization, 2002). Policy makers work to address this issue through mass media and informational campaigns encouraging individuals to incorporate more PA into their lives. Too often however, programs are designed and resources designated in a way that best serves the general population. Such an approach ignores the unique needs of individuals who experience impairments due to cognitive deficits, such as those with intellectual disability (ID). ID is a disability which occurs before the age of 18 and results in adaptive, cognitive, and social impairments (Schalock et al., 2007)
Adults with ID already comprise a disproportionately large number of annual hospital visits and are dispensed more medications than adults without ID. One in three adults with ID in Canada are deemed high-cost patients, a designation used to refer to patients who rank within the top 10% of annual health care spending (Lunsky et al., 2018). Given the fact that this cohort tends to experience a disproportionate number of secondary health conditions compared to typically developing adults, it would be reasonable to suggest that they, in turn, have lower quality of life (May & Kennedy, 1998). Physical inactivity is one modifiable factor which is linked to many conditions and diseases experienced by this population including motor impairments, cardiovascular disease, obesity, diabetes, and types of cancer (Durstine et al., 2013). However, it is not enough to increase PA when attempting to create meaningful changes to specific outcomes. Instead it is important to increase exercise behavior which is meaningful and structured and therefore, intentionally targets health and fitness related outcomes (Caspersen & Christenson, 1985).
Exercise is defined as a physical activity which is planned, structured, repetitive, and purposeful, and focuses on improvement or maintenance of one or more fitness-related outcomes (Caspersen & Christenson, 1985). Exercise is one means by which individuals, including those with ID, can meet recommended guidelines for PA (Department of Health and Human Services, 2018). Regular engagement in exercise can also have positive benefits for an individual's mental well-being and overall quality of life which can also significantly deteriorate as this population continues to age (Ravindran et al., 2016). Approximately four in every 10 individuals with an ID are diagnosed with a secondary mental illness. The most common conditions are anxiety, depression, bipolar disorder, and schizophrenia. Often adults with ID are often prescribed multiple medications, as opposed to more holistic alternatives such as exercise (Finlayson, Turner, & Granat, 2011), which has continually shown to regulate and improve symptoms associated with mental illnesses (Ravindran et al., 2016). Therefore, a compelling argument can be made that increasing exercise behavior in this cohort may help to mitigate secondary conditions associated with physical and mental health.
The Canadian physical activity guidelines suggest a minimum of 150 min of PA per week to be considered sufficiently active (Dairo et al., 2016). Approximately 23% of typically developing Canadians are sufficiently active; comparatively, research has found that as few as 2% of individuals with ID are sufficiently active (Chow, Choi, & Huang, 2018), however this percentage does vary in number ranges from 2% to 9% (Dairo et al., 2016).
Although barriers to PA for this population have been acknowledged by researchers, the most effective means to increase PA behavior remains unclear. As Temple and colleagues (2017) stated in their recent review on physical activity promotion for adults with ID:
It is clear from this review that experimental research focused on increasing participation in physical activity and promoting physical activity to improve the health of adults with intellectual disabilities is in its infancy… Despite the potential benefits of physical activity and low levels among adults with intellectual disabilities, this review demonstrates that research to document the process and outcomes of physical activity interventions is sadly lacking. (pp. 451–452)
Although these findings apply to PA behaviors in general, how best to increase participation in exercise is also at this point unknown.
A meta-analysis by Shin and Park (2012) assessed the effect of exercise programs on individuals with ID. The authors focused on outcomes related to body composition, physiological outcomes (e.g., fitness), and physical performance (i.e., balance), and found an overall positive effect of exercise programs in this population. Additionally, it was found that programs which were shorter in duration (e.g., 10 min) and ran four times per week were more effective than those that ran for longer durations, but less frequently (e.g., three times per week). One of the major limitations of this review was related to the exclusion of individuals with Down syndrome, despite the fact that this cohort makes up a large percentage of individuals with ID. Additionally, the authors chose to focus on highlighting physical health outcomes only, despite the fact that the psychological domain which can include anxiety, depression and self-efficacy, can all be positively impacted by exercise. Furthermore, the studies reviewed in the previous meta-analysis included a broad range of research designs. Reducing the quality of evidence to only experimental or randomized control trials can be used to examine the efficacy of exercise-based intervention, however cross-sectional and longitudinal designs cannot. Therefore, the objective of the current study is to assess the effectiveness of exercise interventions based on experimental designs on individuals with ID. The specific research questions are:
Do exercise interventions lead to positive physical and/or psychological outcomes in individuals with intellectual disabilities when compared to a control?
What is the magnitude (i.e., effect size) of these changes?
Which of these outcomes, mental or physical, are most effected by exercise?
Search Strategy and Selection Criteria
A literature search of all relevant databases was conducted in the winter of 2019. Date limits were applied up to March 2019. The following databases were searched: Medical Literature Analysis and Retrieval System Online (MEDLINE), Psych Info, and SportDiscus. Search terms included physical activity and/or exercise (title and abstract), intellectual disabilities (title and abstract), adult (title and abstract), children (title and abstract). No specific publication format restrictions were set. Only studies written in English were included in the literature search. In order to ensure the data in this analysis is of the highest methodological rigor and contains studies with the highest quality evidence, only studies which randomly assigned individuals or clusters to an experimental or control group were used.
Given the limited research on this topic, this analysis included studies on individuals from all age groups (both children and adults). In order to be inclusive of the kinds of programs created for individuals with ID, studies which focused on any ID population were included. The intervention must have been one that was specific to exercise. Any and all programs were included regardless of setting. Again, due to the relatively small body of studies in this area, any and all exercise modalities were included. However, no cross-sectional studies were included. Primary outcome variables included physical and psychological variables. Physical outcome variables included were body composition measures (body mass index, waist circumference), blood pressure, oxygen consumption and aerobic capacity. Psychological outcomes included anxiety, self-rated depression, and self-efficacy.
Data Collection and Analysis
Selection of Studies
Two independent reviewers screened the titles and abstracts of all obtained articles. Of those identified for potential inclusion, the full texts were obtained and reviewed by two independent reviewers. The reviewers discussed and resolved any discrepancies that were found. A total of 18 studies were included in the review (see Figure 1).
Assessment of Risk of Bias in Included Studies
In order to grade the strength of the evidence, study quality was assessed using the Cochrane Risk of Bias 2.0 (RoB 2.0), specifically designed for cluster randomized trials. The RoB 2.0 was used to assess bias related to threats to internal validity such as flawed research design, poor study execution, and/or incomplete reporting of results. Specifically, the RoB 2.0 assesses risks associated with randomization and allocation sequence, blinding, incomplete or missing outcome data, and selective reporting (Sterne et al., 2017). The RoB 2.0 was completed by two independent reviewers. Any disagreements were discussed until a consensus was reached.
Measurement of Treatment Effects
To assess the effectiveness of the intervention on the various outcome measures, standardized mean difference (SMD) and standard error were calculated for all outcome variables. The use of SMD and standard error allows for the summary reporting of findings taking into account that different scales and measures were used across different studies. The degree of the SMD was assessed using Cohen's standardized conventions (Sterne et al., 2017) for effect size, small (0.2), medium (0.5), and large (0.8). Studies typically reported pre-and postintervention time points, however due to studies not reporting variability in the change scores (i.e., standard deviation of the change score), a comparison of the postintervention measurement score was used (Cohen, 1992). The postintervention mean score and standard deviation for measures of performance, body composition, cardiovascular, and psychological measures were all entered. A direct comparison of the exercise intervention and control group was then completed.
Assessment of Heterogeneity
For the purpose of this review, heterogeneity is defined as follows: “Statistical heterogeneity manifests itself in the observed intervention effects being more different from each other than one would expect due to random error (chance) alone.” (Higgins & Green, 2011, para.1). A visual and statistical examination of any study estimate inconsistencies was completed by visually examining forest plots and consideration of the X2 and I2 values. The proposed thresholds from the Cochrane handbook Chapter 9.5.2 (Higgins & Green, 2011) were used to interpret I2 values.
Random effects models were generated for each outcome using RevMan (Version 5.3, 2014) software. Forest plots of the main analyses and tables containing the results of the sensitivity analyses were also generated. In cases where outcomes were assessed through unique measurements (e.g., musculoskeletal fitness was measured in both upper and lower body), subgroups would be used to analyze the data. This form of data treatment allowed for independent analyses of effect sizes for varying measures within the same outcome. Sensitivity analyses investigated the impact of varying intraclass correlation coefficient (ICC) values.
Results of Search
A total of 715 records were obtained from all databases. After 129 duplicates were removed, 585 titles were screened. From this first level of screening (titles and abstracts), 489 studies were deemed irrelevant, leaving 96 full text articles to be reviewed for eligibility. Of those 96, 78 were excluded. Reasons for exclusion included wrong (a) study design (53 studies), (b) outcomes (11 studies), (c) intervention (7 studies), (d) patient population (4 studies), (e) route of administration (2 studies), and (f) setting (1 study). A total of 18 studies were included in this review (see Figure 1). The percent disagreement for the full text review was 7%, with discrepancy on seven articles. Of those seven articles, 0 were included in the final review.
The 18 studies included 799 individuals with ID from studies conducted in Europe, the United States, Australia, and South Africa. Sixteen of the trials randomized individual participants into the intervention or control group, whereas two studies used cluster randomization (e.g., Day Activity centers). Intervention length ranged from 5 weeks to 52 weeks. Most interventions ran for 10 to 12 weeks and were typically performed three times per week.
Risk of Bias in Included Studies
Table 1 documents data regarding the risk of bias assessments for all included studies. Overall, two studies were low risk of bias, nine studies were deemed to have some concerns, and the remaining seven were rated as having a high risk of bias. Out of the 18 studies, nine studies had at least some risk of bias regarding deviations due to intended intervention, usually due to lack of proper participant/trial personnel blinding.
Results of Pooled Sample
Participants in this review included individuals with ID with mild to moderate levels of impairment (n = 279), individuals with Down syndrome with ID (n = 130), individuals with ID with profound levels of impairment (n = 37), and individuals with ID of unspecified etiology (n = 353). Of those, 349 (44%) were male, however, the gender of 30 participants was unspecified (Beasley, 1982).
Effects of Intervention and Performance Measures
The following results come from 18 studies. All results for performance measures are outlined in Table 2.
In this review, submaximal exercise is defined as a type of exercise that is terminated before reaching ventilatory threshold or maximum heart rate (HR). It is used to estimate VO2 max or aerobic fitness (Heyward, 2009). Estimates of aerobic fitness were tested using the 6-min walk test (Boer et al., 2014; Boer, & Moss, 2016; Calders et al., 2011; Marks et al., 2013), and the shuttle run test (Ozmen et al., 2007; van Schijndel-Speet et al., 2014). Pooled estimates of all measures of aerobic fitness from eight studies (Beasley,1982, Boer et al., 2014; Boer, & Moss, 2016; Calders et al., 2011; Lee et al., 2016; Marks et al., 2013; Ozmen et al., 2007; van Schijndel-Speet et al., 2014) with a combined sample size of 333 participants showed almost no increase in aerobic fitness when comparing exercise interventions to sedentary controls, SMD = 0.13, 95% CI [-0.11, 0.37]. The range of effects shows moderate possible harm, no effect, and small benefit. The overall I2 = 17% indicated low heterogeneity.
For the purpose of this review, balance is defined as static balance or the ability to maintain the body in a fixed position (Rival et al., 2005). Three studies (Borji et al., 2018; Lee et al., 2016; van Schijndel-Speet et al., 2014; pooled n = 162) assessed static balance. Pooled estimates show that although the effect size is large the CI were wide, SMD = 1.25, 95% CI [-0.39, 2.90]; indicating no effect of the intervention. This range indicates that results show a small degree of harm, no effect, and large benefit.
Functional fitness reflects one's ability to perform physical activities of daily life with relative ease (Heyward, 2009). Functional fitness included five studies (Boer et al., 2014; Boer, & Moss, 2016; Calders et al., 2011; Lee et al., 2016; Marks et al., 2013) and a total of 325 participants, with pooled estimates indicating a minimal effect (SMD = −0.07). Overall, results were imprecise: showing high benefit, no effect, and a moderate degree of harm. When subgroup analysis was examined, Sit to Stand test was found to favor the controls (SMD = 0.37), whereas the Get up and Go test did have a large, benefit observed in the intervention group (SMD = −0.77. (The decrease in the Get up and Go scores does indicate an improvement on the test, as decreased time indicates better functional fitness.) The X2 test of subgroup differences was statistically significant (p = 0.02) indicating the effects differ between the two subgroups.
For this review, muscular strength was defined as any activity which elicited the maximum force that a muscle or muscle group can generate at a specific velocity. Studies looked at pooled estimates from all measures from six studies (Boer et al., 2014; Calders et al., 2011, Giagazoglou et al., 2013; Shields et al., 2013; Suomi, 1998; van Schijndel-Speet et al., 2014) of 351 participants that showed an effect estimate indicating strong, positive increase when comparing the intervention group to the control group, SMD = 0.70, 95% CI [0.24, 1.16], indicating benefit. Subgroup analyses show slightly less conclusive results for upper body strength, 95% CI [-0.17, 1.26], whereas the results for low body musculoskeletal strength indicate a strong effect, SMD = 0.86, 95% CI [0.30, 1.42]. The overall I2 = 74%, indicating a high degree of heterogeneity.
Body composition included body mass index (Boer et al., 2014, Calders et al., 2011; Melville et al., 2015; Ozmen et al., 2007), weight in pounds (Marks et al., 2013), weight in kilograms (van Schijndel-Speet et al., 2014). Pooled estimates from all measures from six studies of 343 participants showed a small effect size for weight, SMD = 0.13, 95% CI [-0.12, 0.37]. The 95% CI indicates the possibility of benefit, no effect, and possible harm. Only two studies (Melville et al., 2015; van Schijndel-Speet et al., 2014) measured waist circumference and results were inconsistent: possible benefit, no effect, and possible harm, 95% CI [-0.37, 0.50]. All results for body composition are outlined in Table 3.
All results for cardiovascular health are outlined in Table 4. Four studies (Boer & Moss, 2016; Boer et al., 2014; Calders et al., 2011; Rosety-Rodriguez et al., 2014) examined maximal oxygen uptake (n = 109); the pooled estimates indicated a medium effect (SMD = 0.55) when comparing the exercise intervention to the sedentary control; 95% CI [0.17, 0.94].
Blood pressure was examined in three studies (Boer & Moss, 2016; Calders et al., 2011; van Schijndel-Speet et al., 2014), and the results of pooled estimates revealed a medium effect, SMD = −0.30, 95% CI [−0.56, −0.03]. Subgroup analyses of systolic and diastolic blood pressure found a medium effect of exercise on systolic blood pressure, SMD = −0.47, 95% CI [−.085, −0.10], whereas exercise had a small effect on diastolic, SMD = −0.12, 95% CI [−0.48, 0.23]; with no significant differences between subgroups (p = 0.19).
Although results from the pooled estimates of three studies (Boer & Moss, 2016; Boer et al., 2014; Calders et al., 2011; n = 89) examining heart rate did show a small effect of exercise benefiting the intervention group (SMD = 0.11) the 95% CI show indefinite results: possible harm, no effect, and possible benefit.
All results for psychological outcomes are outlined in Table 5. Anxiety and depression was assessed by three studies (Carraro & Gobbi, 2012, 2014; van Schijndel-Speet, 2014). The pooled estimates from 140 participants showed a large effect of exercise on anxiety and depression related symptoms (SMD = −3.07). However, the CI were extremely wide and provides unspecified results, 95% CI [−6.81, 0.66]. Self-efficacy was examined in two studies (Marks et al., 2013; Melville et al., 2015; n = 152) and pooled estimates indicated a large effect of exercise when comparing the intervention to the sedentary control (SMD = 0.74, 95% CI [−0.33, 1.80].
Summary of Main Results and Certainty of Evidence
Interest in exercise as an intervention for various outcome measures for individuals with IDs has steadily increased, with a greater number of interventions appearing in the literature. However, the inclusion of all available trials provided inconsistent results. Some evidence did indicate improvements with regards to lower body muscular endurance, blood pressure, reaction time, and self-efficacy. Results for other outcomes were inconsistent, even potentially harmful. Furthermore, the best available evidence for other outcomes is uncertain due to the quality of evidence. It is possible, however that those with ID need extra support to fully and consistently maximize the benefits of exercise.
Overall Completeness and Applicability of Evidence
The studies within this review include individuals with various types of ID (Down syndrome, IDD, PDD-NOS), various exercise modalities, and a wide range of age groups. Very few of these studies are adequately powered, and a limited number had examined multiple outcomes in the same study. The problem of low sample size resulted in confidence intervals that were wide and could not provide quality, determinant results. Additionally, few studies reported ICC values, which also made it difficult to assess quality. It is important to consider, however the difficulty associated with recruiting this population, which could largely account for the small participant sample sizes. In general, studies had at least some concern or high concern with regards to risk of bias. Typically, studies did not have blinded outcome assessors, and as a result, the risk of bias assessment with regards to measurement outcome was high.
Reporting errors were also a major cause of concern. One study, in particular, did not include any information regarding the duration, frequency, intensity, or modality of the exercise intervention (van Schijndel-Speet, 2014). Additionally, some studies included information for some of these categories, but not all. As a result, it was not possible to conduct post hoc analyses exploring the effect of intervention duration, frequency, and intensity on outcomes. Therefore, the body of research must become more robust in order to identify detectable differences/effects estimates resulting from various durations, frequencies, and intensities. Specifically, increases in sample size and the quality surrounding the measurement of the interventions need to be improved.
Agreements and Disagreements With Other Studies or Reviews
Overall, the results from this meta-analysis show that although definitive and salient changes were not observed in all domains/outcomes, some improvements were apparent. Notably, there is at least some modest evidence that exercise may lead to positive changes in musculoskeletal strength, maximal oxygen uptake, and blood pressure. Previous research supports the findings from the current meta-analysis as previous work has also found that exercise programs improved muscle force, VO2, and self-esteem in a similar population (Shin & Park, 2012). Furthermore, in the general population, sustained and ongoing exercise behavior can help to improve total blood pressure score, and reduce symptoms associated with hypertension (Carpio-Rivera et al., 2016), and it appears this extends to populations with ID as well.
The current review was the first to examine the impact that exercise can have on mental health. The evidence showed that there were large (though, somewhat imprecise) gains in mental health outcomes for adults with ID, specifically when investigating exercises impact on anxiety and depressive symptoms. This finding not only points out the link between mental health and exercise, but the complexity of this link in those with ID, and the importance of exercise as a means of providing holistic treatment for secondary mental health conditions. However, the results from this study show strong, but variable results due to few studies actually investigating mental health in this population. This finding provides more reason for researchers and practitioners to continue their consideration of exercise and mental health in this group. More research needs to explore how these benefits are maintained beyond the duration of the intervention.
Although research has previously shown that exercise can indeed improve body composition outcomes (Swift et al., 2014, the current meta-analysis found only a small effect on body composition despite the fact that each of the six studies used exercise as a means to facilitate weight loss. Beyond the small effect size, confidence intervals indicated no effect and possibility of harm. However, when it comes to weight loss, exercise is only one small aspect related to weight loss. Individuals must also maintain a healthy diet and must reduce caloric intake while increasing energy input. Furthermore, weight loss is highly unique and differs from person to person (Swift et al., 2013). One factor which hinders weight loss and improvements in body composition for individuals with ID is medication. A large percentage of this population is prescribed antipsychotic and antidepressant medications (Doan et al., 2013), both of which can cause weight gain (Wharton et al., 2018). Medication-induced weight gain is significant and often difficult to reverse. Although none of the studies in the current review indicated whether or not participants were taking medication, it is possible that this, in combination with other individual factors (i.e., nutrition), resulted in vague findings.
Flexibility, step count, and heart rate also had unclear results, which aligns with previous work (Shin & Park, 2012). It is likely this is a result of small sample sizes as each of these outcomes had less than 100 participants included in the pooled estimates. However, outcomes with larger sample sizes (submaximal exercise, balance, functional fitness, anxiety/depression) also had imprecise results. Many of the studies included in these outcomes were 12-week exercise interventions, and it is possible that this length was not enough to produce any salient changes.
The results of this review, although inconsistent, do provide some evidence indicating the need for more research to determine the efficacy of fitness/exercise programs for this population. It is clear that within this population, exercise and exercise related behaviors can lead to some positive changes in specific outcomes. Current research is largely focused on physical outcomes although there is an extreme lack of quality evidence supporting exercise as an alternative therapy for mental health in this population. It is important to recognize that exercise may act as a primary treatment for many comorbid conditions that are prominent within this population such as anxiety and depression (Lunsky et al., 2018). Although this review was not able to identify best exercise practices for this population, it does provide evidence that exercise (through any method) may be of benefit for individuals with ID.
This review is largely limited by the quality of evidence, which justifies the need for future studies to employ methodologically sound, adequately powered interventions. Additionally, very little information could be drawn regarding optimal program frequency, timing, and length (which are fundamental to any exercise program) due to lack of evidence. Lack of evidence also extends to adherence, as no studies completed follow-ups to confirm if the changes they saw extended beyond the duration of the intervention. This is an important note for future researchers as evidence which supports whether or not these changes are resistant to time is lacking. Furthermore, no studies measured fidelity related to the implementation of the intervention, and if there were deviations from the intended program this may have affected the final results. Therefore, it is important that future research ensure that interventions are being carried out according to the initial design.
Insufficient evidence does not allow us to draw conclusions regarding several outcomes including functional fitness, submaximal exercise performance, and heart rate. Overall, this review serves a pertinent reminder that although individual studies have identified exercise as a prominent way to improve many lifestyle and health factors in those with ID, the data should be subject to reproduction before these results can be taken as fact.