Background

There is now a greater focus on patients’ physical health by mental health services [1]. People with serious mental illness have consistently higher levels of mortality and morbidity than the general population. Mortality rates remain persistently high around twice those of the general population [2]. The life expectancy of people with serious mental illness is shortened by between 11 and 18 years [3]. Despite a steady reduction in mortality rates in the general population no significant change has been observed in people with serious mental illness [4]. The widening differential gap in mortality suggests that people with schizophrenia have not fully benefited from the improvements in health outcomes available to the non-mentally ill population [2].

The underlying causes for the health problems of this population are both complex and multi-factorial [5]. People with serious mental illness such as schizophrenia have higher levels of cardiovascular disease [6, 7], metabolic disease [8], diabetes [9, 10], and respiratory illness [11, 12]. Although genetics may have a role in the physical health problems of these patients, lifestyle and environmental factors such as smoking, obesity, poor diet, and low levels of physical activity play a prominent part [13]. Some of the treatments given to people with serious mental illness contribute to the health problems in this population. For example, neuroleptic medication can lead to significant metabolic problems such as weight gain, lipid abnormalities, and changes in glucose regulation [14]. Evidence is becoming clearer that long term exposure to these medications may contribute to the higher mortality levels in this population [5].

There is now a greater focus on attempting to improve the physical health of these patients. However they have less access to medical care, poorer quality of care, and preventative health checks are less commonly completed in both primary and secondary care compared with the general population [15–17]. The nature of their mental illness may also affect their motivation as many individuals may be not ready to change their lifestyle [18, 19]. Individuals with serious mental illness have higher levels of smoking, weight problems, poor dietary intake, and low levels of physical activity. Rates of smoking of up to 70% have been found in patients with schizophrenia [7, 20, 21]. The prevalence of smoking in the general population is approximately 20% [22]. Individuals with serious mental illness have a poorer diet than the non-mentally ill population [20, 23]. Levels of obesity range from 40-60%, up to four times that of the non-mentally ill population [24–26]. These individuals are less active with lower levels of physical activity compared with the general population [27–29].

Regular physical activity has been found to be beneficial for both physical and mental health in the general population. There is a direct relationship between physical activity and a reduction in cardiovascular disease, cerebrovascular disease, stroke, and hypertension [30, 31]. Physical activity also leads to a reduction in the risk of metabolic health problems such as diabetes and metabolic syndrome [30]. A routine level of at least 150 minutes of physical activity per week is required to achieve a consistent reduction in risk [31]. In the general population about 60% of men and 70% of women self-report less than the recommended levels and objective measures of activity suggest that far more of these individuals are failing to meet the recommendations [32]. Physical activity has beneficial effects on mental health. It has been shown to be effective in the treatment of depression [33] and anxiety disorders [34]. It has positive effects on psychological well-being [35], quality of life [36] and in the reduction of stress [37].

The aim of this review was to determine the effectiveness of exercise programmes for people with serious mental illness. Two main objectives of this review were to firstly determine the effect of these programmes on levels of exercise activity, and secondly the effect of exercise on mental health and well-being.

Methods

Eligibility criteria

Studies met the following criteria for inclusion in the review:

  1. 1.

    Adults with schizophrenia or other types of schizophrenia-like psychosis, schizoaffective disorders, and bipolar affective disorder irrespective of the diagnostic criteria used, age, ethnicity and sex.

  2. 2.

    All patients, adults, clients, in the community or in hospital.

  3. 3.

    All relevant randomised controlled trials.

  4. 4.

    Interventions where a primary or secondary aim was to promote exercise or physical activity.

Search Methods, and study selection

We searched the following electronic databases: Ovid MEDLINE, Embase, CINAHL, PsycINFO, Biological Abstracts on Ovid, and The Cochrane Library (January 2009, repeated May 2013). The systematic search included hand searching of journals, books, cross-referencing and bulletins (e.g. brief reports/brief statement of facts). The search filter, the Cochrane Highly Sensitive Search Strategy, was used to assist in the identification of randomised trials in MEDLINE [38].

The abstracts of studies were examined by RP. Full text of the studies that potentially met the eligibility criteria was obtained. Discrepancies were discussed with co-investigators. We checked articles that met the inclusion criteria for duplication of the same data.

Data extraction and analysis

Data was extracted by one author (RP) and checked for accuracy by the second (DS). Data was extracted onto prepared forms to include: participants and setting, location, description of the intervention, type of exercise, study size, methodological issues, risk of bias, results, and general comments. All analyses were conducted using Revman Manager version 5.1. We performed a PRISMA evaluation of our meta-analysis using a standard checklist of 27 items that ensure the quality of a systematic review or meta-analysis [39]. A summary measure of treatment effect was used as different outcome measures were found. The standardized mean difference (SMD) with 95% confidence intervals was calculated as the difference in means between groups divided by the pooled standard deviation. If no standard deviations were found they were calculated from standard errors, confidence intervals, or t values [40]. Authors were contacted for missing data if analyses could not be completed. Statistical heterogeneity was assessed by the I2 test. The degree of heterogeneity was categorised as the following [36]: 0% to 40% low level of heterogeneity; 30% to 60% moderate heterogeneity; 50% to 90% substantial heterogeneity; 75% to 100%: considerable heterogeneity. Standard mean differences were based on the random-effects model as this would take into account any differences between studies even if there was no statistically significant heterogeneity [40].

Quality assessment

There is no agreed standardised method to assess the quality of studies in systematic reviews. We adapted the Cochrane Collaboration’s tool for assessing the risk of bias [40]. The following recommended domains were considered: sequence generation, allocation concealment, blinding, incomplete outcome data, selective outcome reporting, and other sources of bias. Each item was rated according to the level of bias and categorised into either low, high, or unclear. The category unclear indicated unclear or unknown risk of bias [40].

Results

The electronic search identified 1284 potentially eligible reports. Nine hundred and thirty two were excluded on the basis of the title or abstract alone. We retrieved the full text of 216 articles and excluded a further 209 studies. The review excluded a large number of studies as they included additional components such as dietary or weight programmes (Figure 1).

Figure 1
figure 1

Flowchart of the results of the systematic search.

Full size image

All included studies had been published between 2005 and 2013. The studies varied in their setting, size, age, study intervention type, and use of outcome measures.

Study characteristics

Characteristics of the sample, interventions, outcomes assessment and results are shown in Tables 1, 2 and 3.

Table 1 Systematic review – exercise interventions (Description of programmes)
Full size table
Table 2 Systematic review – exercise interventions (Methodological issues/Risk of bias)
Full size table
Table 3 Systematic review – exercise interventions (Results/Comments)
Full size table

Setting and participant characteristics

Eight studies met the inclusion criteria. Of these 5 were based solely in either a community or outpatient setting. The remainder comprised a combination of inpatient hospital together with community or outpatient programmes. Diversity of setting may have had an impact on the delivery and generalisability of the exercise interventions. Exercise change may be more easily achievable within a more closely supervised setting such as an inpatient ward. The sample size of studies in this review varied considerably. The largest study had 118 patients in their programme the smallest 10 individuals. The mean age of individuals ranged from 27 years to 52 years. The majority were in the age group 30–40 years. One study [41] used an older age population with a mean age of 46.9 years and one a younger group with a mean of 29.2 years [42]. The variation in age group may have affected the implementation of the programme. A younger population may have been more able to improve their level of exercise. The ethnicity of participants was described in only two studies. Beebe et al. [43] found that 54% of participants were Caucasian, 44% African-American, while Beebe et al. [41] in a small study of 10 participants, 80% were Caucasian and 20% African-American.

Exercise interventions

Each study used a different type of aerobic exercise incorporating cardiovascular exercise and resistance training. Three studies used walking as their exercise activity [41, 43, 44], while four studies used a combination of general aerobic and cardiovascular exercise [42, 45–47]. Cycling was the method of exercise activity used in the remaining study [48]. In addition to the exercise activity general information was given to all participants about exercise. Only two studies used more specific advice and guidance [43, 45]. Skrinar et al. [45] offered seminars on a range of topics such as adequate individual levels of exercise, healthy eating, stress relief, spirituality and wellness. In the remaining study Beebe et al. (2011) compared the effect of adding specific exercise advice and motivational skills to a simple walking programme [43].

Only one study [42] used a standardised programme of exercise comprising of cardiovascular and muscle strength exercises [49]. Moderate levels of exercise intensity were described in 7 out of the 8 programmes. Only Methapatara et al. [44] described the specific amount of exercise activity used in their programme. The actual amount or dose of exercise activity was not measured in the remaining studies.

Each programme varied in their frequency and duration. Some were twice a week [46] while Skrinar et al. [45] described a programme 4 times per week. The duration of the programmes lasted between 10 weeks [47] to 24 weeks [42].

Outcomes

A variety of different outcome measures were used (Table 4). This made it especially difficult to compare the results of individual interventions. Two studies used a validated measure of exercise, namely the 6-Minute Walking Distance [41, 46]. Others used measures such as body mass index [46], the number of exercise sessions attended [45], or the number of minutes walked [43].

Table 4 Comparison: exercise versus standard care
Full size table

Six studies compared the effect of exercise with usual care. A significant increase in the distance walked in 6 minutes was found in one RCT (n = 13, SMD = 1.81, CI 0.44 to 3.18, z = 2.58, p = 0.01) (Figure 2) [46]. A small non-significant reduction was found on body mass index (n = 151, SMD = -0.24, CI -0.56 to 0.08, p = 0.14; heterogeneity, Chi2 = 1.45, I2 = 0%, p = 0.69) (Figure 3). No effect was found comparing the effect of exercise on body weight (n = 77, SMD = 0.13, CI -0.32 to 0.58, p = 0.57; heterogeneity, Chi2 = 0.07, I2 = 0%, p = 0.79) (Figure 4).

Figure 2
figure 2

Effect of exercise versus usual care: 6-minute walking distance.

Full size image
Figure 3
figure 3

Effect of exercise versus usual care: body mass index.

Full size image
Figure 4
figure 4

Effect of exercise versus usual care: weight.

Full size image

There was no overall beneficial effect on negative symptoms (n = 84, SMD = -0.54, CI -1.79 to 0.71, p = 0.40) or positive symptoms of schizophrenia (n = 44, SMD = -1.66, CI -3.78 to 0.45, p = 0.12). Exercise did not lead to an improvement in anxiety and depressive symptoms (n = 94, SMD = -0.26, CI -0.91 to 0.39, p = 0.43; heterogeneity, Chi2 = 3.92, I2 = 49%, p = 0.14) (Figure 5).

Figure 5
figure 5

Effect of exercise versus usual care: anxiety & depression.

Full size image

Data from single RCTs was available for the following outcome measures. Overall, there was no clear evidence that exercise interventions led to significant improvements in quality of life. A small non-significant increase in points of physical and mental domains were found (physical domain: n = 30, SMD = 0.45, CI -0.27 to 1.18, z = 1.22, p = 0.22: mental domain: n = 30, SMD = 0.65, CI -0.09 to 1.39, z = 1.73, p = 0.08).

The final study compared the effect of adding specific exercise advice and motivational skills to a simple walking programme [43]. No significant change was found in the attendance at walking groups, persistence, or minutes walked compared with the control.

Methodological design and quality

The review found a lack of standardisation in terms of the exercise intervention, setting and outcomes measures (Tables 1, 2 and 3). Six studies described an adequate method of randomisation. Four out of the 8 studies used satisfactory methods to conceal the allocation of treatment. Beebe et al. [41] was the only study to incorporate outcome assessors blinded to the treatment group. There was considerable variation in sample size. No study described a sample size calculation. We used the Cochrane Collaboration’s tool for assessing the risk of bias [40] (Figure 6).

Figure 6
figure 6

Risk of bias graph: review authors’ judgements about each risk of bias item presented as percentages across all included studies.

Full size image

One study only adequately addressed the analysis of incomplete data [44]. Data was analysed on an intention-to-treat basis with last observation carried forward on the following measures, bodyweight, body mass index, and waist circumference.

Levels of attrition varied in the included studies between 0% and 33% with a mean of 18.9%. No dropout was observed in two studies [44, 46]. Methapatara et al. [44] had no attrition in a larger study of 64 individuals, perhaps suggesting that participants in this study were more motivated to change.

Discussion

Exercise programmes found in this review had a modest beneficial effect on levels of exercise activity. Exercise did not lead to an improvement in body mass index, negative or positive symptoms of schizophrenia, or the individual’s quality of life. No effect was found on body weight or symptoms of anxiety or depression. No beneficial effect was found by the addition of specific exercise advice with motivational techniques to a simple exercise programme. The review found that many studies varied in the type of exercise programme used, setting, age group, sample size, and outcome measures. Comparison of these outcomes proved difficult. Diversity of setting may have had an impact on the delivery and generalisability of the exercise interventions. Exercise change may be more easily achievable within a more closely supervised setting such as an inpatient ward.

The exercise programmes used a variety of outcome measures. Only two studies used a validated measure of exercise [41, 46] the 6-Minute Walking Distance [50]. The majority of studies did not measure the quantity or intensity of the activity programme. It therefore proved difficult to compare the equivalent benefits of each study. We could not estimate the amount of exercise required to achieve an improvement in levels of exercise activity or mental health symptoms.

Exercise programmes resulted in only moderate or little change in measured outcomes. It is unclear why these programmes did not lead to a more significant improvement. The physical activity component of the interventions may not have been sufficiently intense. The duration of the programme may have been too short to bring about change. Delivery of the uptake of the sessions may have been inadequate to achieve a change in activity levels. The difficulties bringing about change in this population may reflect the inherent problems in this population. There are greater levels of smoking, weight problems, and co-morbid illness [51]. Individuals with greater risk factors and co-morbid physical illness may have more difficulty achieving higher levels of physical activity.

The review found that attrition levels were low with one study having no drop-out of participants. Analysis of dropout rates can give valuable information about patient characteristics and trial characteristics that affect the overall uptake of an intervention [52]. Loss to follow up after recruitment and attrition in randomised controlled trials affects the generalisability and the reliability of their results [52, 53]. Studies of health-behaviour change [54] in the general population have found varying levels of attrition. Smoking cessation interventions have been found to have attrition rates of up to 49%. Attrition rates of between 27% [55] and 32.8% [56] have been observed in weight loss studies. In people with serious mental illness similar levels of attrition have been found. Khan et al. [57] found in 45 trials of antidepressants with a total of 19,000 subjects the mean drop out rate was 37%. Explanation for the low level of attrition in this review is unclear. Low drop-out rates may suggest the participation of motivated individuals willing to change their behaviour. Several studies included individuals based in hospital wards. The structured setting of a ward environment may have reduced the levels of attrition.

Despite the health problems of this population we found only 8 randomised controlled trials assessing the effectiveness of exercise programmes for people with serious mental illness. However the number of studies in this field has steadily grown over the past few years. For example, Ellis et al. [58] in a systematic review of exercise programmes in psychosis identified one randomised controlled trial [41], while Gorczynski et al. [59] found three studies [41, 46, 60] following a systematic search completed in December 2008. However one of these studies included the effects of yoga therapy in people with schizophrenia [60]. A recent literature review by Faulkner et al. [61] found a total of 7 randomised controlled trials, however 4 of these included yoga therapy. Faulkner et al. [61] identified one additional small study [48] which was included in the review in this paper.

It is unclear why so few studies have been conducted in this field? In our review we found that many studies were excluded as they contained additional components such as dietary advice or measures to reduce weight. The focus on physical activity alone may be being lost in interventions designed to address the current increasing general concerns about obesity and metabolic problems in this population [61]. This is in addition to a general lack of well designed studies aiming to address the health problems and risk factors in this population. For example, smoking levels in people with serious mental illness remain about two to three times levels found in the general population. However relatively few randomised trials have been conducted with the primary aim of achieving smoking cessation in this population [62].

There are strengths and limitations to the results we have presented. This review found modest changes in levels of exercise activity, but no effect on symptoms of mental health. A number of limitations need to be acknowledged. Research in this field has been so far been limited with only a small number of randomised controlled trials. These tended to be small in size and of short duration. The heterogeneity of programmes affected the impact and generalisability of studies found in the review. Studies failed to quantify the amount and intensity of exercise in their programmes. Interventions tended to use non-standardised exercise programmes and a variety of outcome measures. It proved difficult to recommend from this review the most suitable and effective programme of exercise to individuals with serious mental illness. Until further research is conducted individuals with serious mental illness should be encouraged to meet the general recommendations currently advised to the general population.

Research in the future needs to focus on methods to improve levels of exercise in individuals with serious mental illness. Programmes have proved successful in the general population. There is a need to conduct well designed randomised trials of physical activity programmes. Research needs to incorporate a standardised exercise programme and outcome assessment. The duration and intensity of interventions needs to be sufficient to achieve change in levels of physical activity, mental health symptoms, or weight. New programmes need to take into account the specific needs and potential barriers to exercise of those with serious mental illness. Levels of motivation, mental health symptoms, and weight enhancing medication add additional complexity and difficulties to this process. Potential barriers to and benefits of physical activity for people with serious mental illness have been shown in previous research. McDevitt et al. [63] found that in individuals with serious mental illness symptoms of mental illness (e.g. lack of energy or volition), medication, weight gain from medication, and safety concerns restricted their ability to be active. Johnstone et al. [64] found that limited experience of previous physical activity reduced self esteem and confidence, and lack of structure or planning to their day, limited engagement in exercise activity. In the review we described three studies identified specific barriers affecting participation in their programmes. Marzolini et al. [50] found that medical and health reasons, supervised trips and family visits, and medical appointments affected participation. Skrinar et al. [45] identified several barriers such as problems with transport, financial issues, treatment factors, and conflicting schedules with other treatment programmes affected participation. Beebe et al. [43] found similar problems with the most common reasons given for non-attendance being transportation problems (22.2%), physical illness (20.6%), and conflict with another appointment (12.7%).

For clinicians there remains no clear standardised method to improve levels of physical activity in this population. The health problems of this population continue to be highlighted in many leading publications [1]. An effective method needs to be developed in this population to reduce the persistently highs levels of cardiovascular risk and mortality.

Conclusion

In conclusion we found that exercise programmes can lead to an improvement in exercise activity but had no significant effect on symptoms of mental health or body weight. However, it is clear that further research is needed with studies of larger size using comparable interventions and outcome measures.