Motivating factors and barriers towards exercise in severe mental illness: a systematic review and meta-analysis

1People with severe mental illness (SMI) experience a premature mortality of around 15–20 years, largely due to inequalities in physical health (Ribe et al. ). For instance, people with SMI have a significantly higher risk of obesity, hyperglycaemia and metabolic syndrome, all of which contribute towards the development of cardiovascular diseases (Gardner-Sood et al. ). Many of these physical health issues are related to modifiable risk factors which can be treated and attenuated through lifestyle changes, including exercise and diet (McNamee et al.; Curtis et al. ). This is particularly important for those receiving antipsychotic treatment since these medications greatly increase cardio-metabolic risk when combined with a sedentary lifestyle (McNamee et al.; Vancampfort et al. ).

2People with SMI engage in significantly less vigorous exercise, and significantly greater amounts of sedentary behaviour than health controls (Stubbs et al.,; Vancampfort et al. ). This inactivity is predictive of a range of adverse health outcomes including obesity, diabetes and medical co-morbidity among people with SMI (Vancampfort et al.,; Suetani et al. ). It is also associated with more severe negative symptoms and poor socio-occupational functioning (Vancampfort et al.; Suetani et al. ).

3An increasing body of research demonstrates that exercise interventions can improve physical health and reduce psychiatric symptoms in people with major depression and psychotic disorders (Rosenbaum et al.; Firth et al. ). Exercise has also been found to reduce negative symptoms and cognitive deficits in schizophrenia (Firth et al.; Kimhy et al. ); aspects of the illness which are often left untreated and particularly influential on long-term functioning (Galletly,; Arango et al. ). Thus, proper implementation of exercise within the care of people with SMI could reduce cardio-metabolic risk and the associated mortality, while also facilitating functional recovery.

4The optimal modality of exercise interventions for people with SMI is yet to be established. A recent meta-analysis suggests that various exercise modalities can be effective for improving outcomes in SMI, although only if a sufficient total volume of activity is achieved (Firth et al. ). Clinical trials have also found that significant benefits for depressive and psychotic symptoms only occur among participants who achieve sufficient amounts of exercise (Hoffman et al.; Scheewe et al. ). Therefore, training programmes which can maximize adherence to exercise in SMI may be the most effective.

5Meta-syntheses of the qualitative literature have previously examined the factors which may encourage or prevent exercise participation among people with SMI (Mason & Holt,; Soundy et al. ). For instance, improving self-identity and body image is a valued outcome of exercise programmes, while the sedative effects of psychotropic medications can inhibit physical activity (Mason & Holt,; Soundy et al. ). Although valuable, qualitative investigations can be influenced by interviewers’ biases, and results may only represent a subset of the population. Data from survey-based studies may therefore provide a more accurate representation of the entire patient group.

6Improving our understanding of desired outcomes of exercise among people with SMI could enhance health promotion initiatives, and inform the development of interventions that are both motivating and rewarding for patients. Furthermore, determining the most common barriers would help to optimize resource allocation when delivering exercise services in clinical practice. Thus, we conducted a systematic review of studies reporting quantitative data on motivating factors and barriers towards exercise for people with SMI. We also quantified patients’ responses in these surveys using meta-analytical techniques to determine which were most pertinent for this patient group.

7An electronic database search of Ovid Medline, Allied and Complementary Medicine Database (AMED), PsycINFO, EMBASE, and the Health Management Information Consortium (HMIC) database, using the search algorithm: ‘exercise’ or ‘physical activity’ or ‘sport*’ AND ‘psychiatric’ or ‘severe mental’ or ‘serious mental’ or ‘schizophrenia’ or ‘psychosis’ or ‘bipolar’ or ‘manic depress*’ or ‘major depress*’ or ‘clinical depress*’ or ‘depressive disorder’ AND ‘motiv*’ or ‘barriers’ or ‘incentives’ or ‘attitudes’ or ‘preferences’ or ‘advantages’ or ‘disadvantages’ was conducted in May 2016, considering articles published from database inception. A search of Google Scholar was conducted using the same key words to identify any additional relevant articles. The reference lists of retrieved articles were also searched.

8Only English-language research articles in peer-reviewed journals were included in this review. Eligible samples were those in which >80% of the sample had a diagnosis of a SMI (i.e. schizophrenia, schizoaffective disorder, other psychotic disorders, bipolar disorder or major depressive disorder) and/or were currently receiving treatment for SMI. Studies which inferred the presence of SMI solely from participants’ response to screening questionnaires were excluded if no diagnosis or current treatment for SMI could be confirmed. Eligible studies were those reporting proportional data on motivating factors and/or barriers towards physical activity among people with SMI, from questionnaires, surveys or other quantitative methods. Studies which used only qualitative methods were not eligible for inclusion, as these have been comprehensively reviewed elsewhere (Mason & Holt; Soundy et al. ). ‘Motivating factors’ were defined as any outcome of exercise perceived by patients to be a reason for increasing physical activity. ‘Barriers’ were defined as any physiological, psychological or socio-ecological conditions reported to reduce patients’ participation in exercise.

9Articles were screened by two reviewers (J.F. and S.R.) to assess eligibility. Disagreements on eligibility were resolved through discussion. A systematic tool was developed (see Supplementary Table S1) to extract all relevant quantitative data from each study into the following categories: Motivating factors for exercisePhysical: physical health; fitness; strength; weight loss.Psychological: well-being; enjoyment; reduce distress; mood; self-esteem.Socio-ecological: socializing; health professional advice; routine.Barriers to exercisePhysical: physical illness; tiredness/fatigue.Psychological: distress; depression; motivational; self-confidence; safety.Socio-ecological: cost; access to facilities; time; support; insufficient information.Information on study characteristics (sample size, demographics, location, care setting) was also extracted from each study, and is summarized in Table 1.

10Motivating factors for exercise

11Physical: physical health; fitness; strength; weight loss.

12Psychological: well-being; enjoyment; reduce distress; mood; self-esteem.

13Socio-ecological: socializing; health professional advice; routine.

14Barriers to exercise

15Physical: physical illness; tiredness/fatigue.

16Psychological: distress; depression; motivational; self-confidence; safety.

17Socio-ecological: cost; access to facilities; time; support; insufficient information.

18We sought to establish the overall prevalence of motivating factors or barriers towards exercise proportion among people with SMI. Therefore, where any specific motivating factor/barrier had been examined by ⩾3 independent studies, data was pooled using proportional meta-analysis in StatsDirect 2.7 (StatsDirect, ). A random-effects model was applied in all meta-analyses, in order to account for expected heterogeneity between studies (DerSimonian & Laird, ). The degree of variance between studies was assessed with Cochran's Q and indexed as I2, which estimates the amount of variance caused by between-study heterogeneity, rather than chance. As wording of questions can differ between studies, combinability of study data for meta-analyses was first established through agreed selection by two reviewers (J.F. and S.R.).

19Fig. 1 shows the full study selection process. The initial database search returned 1534 results. This was reduced to 1163 after duplicates were removed. A further 1109 articles were excluded after reviewing the titles and abstracts for eligibility. Full text versions were retrieved for 54 articles, of which nine were eligible for inclusion. A further three articles were identified from a similar search of Google Scholar. A total of 12 different studies articles, each with unique samples were eligible for inclusion (Faulkner et al.; Ussher,; Sylvia et al.; Gorczynski et al.; Kane et al.; Wynaden et al.; Carpiniello et al.; Bassilios et al.; Deighton & Addington; Fraser et al.; Klingaman et al.; Firth et al. ). Additional data was obtained for four studies from the corresponding authors (Sylvia et al.; Gorczynski et al.; Deighton & Addington,; Firth et al. ).

20Characteristics of included studies are detailed in Supplementary Table S2. Three were conducted in the United States, three in Canada, three in Australia, two in the UK, and one in Italy. There were a total of 6431 psychiatric patients within these studies; 85.5% with schizophrenia, 6.2% with an unspecified SMI, 2.3% with bipolar or major depression, and 6% other/unknown diagnosis. Where specified, 65% were community-based outpatients while 35% were inpatients within psychiatric units. The median age was 42.6 years (range = 19.8–55 years). Samples ranged from 26–86% male (median = 62%). Of 5757 subjects, 50% belonged to minority groups within their respective countries, while 50% were white. Five studies (n = 470) also reported employment, showing that 68% of participants were unemployed. All survey items which were combined for meta-analyses are highlighted in Tables 1 and 2.

21Meta-analyses of proportional data are displayed in Fig. 2. The most endorsed reason for exercising was to improve general physical health; endorsed by 91% of people with SMI (N = 6, n = 790, 95% CI 80–94, Q = 81, p < 0.01, I2 = 94%). Two studies which examined motivations for exercise using Likert scales also found that general health improvement ranked higher than all other options (Faulkner et al.; Gorczynski et al. ).

22Increasing fitness/energy was the most widely assessed physical health motivation (N = 5, n = 549). This was a motivating factor for 75% of respondents (95% CI 64.9–83.4, Q = 19, p < 0.01, I2 = 79%) and ranked as ‘highly important’ in three Likert-scale studies (Faulkner et al.; Sylvia et al.; Gorczynski et al. ). ‘Improving appearance’ and ‘losing weight’ were examined in only three studies each, but received high rates of endorsement of 77% (n = 465, 95% CI 64–88, Q = 13.3, p < 0.01, I2 = 85%) and 83%, respectively (n = 169, 95% CI 54–99, Q = 30, p < 0.01, I2 = 93%). ‘Improving strength’ averaged 72% endorsement (N = 3, n = 169, 95% CI 55–87, Q = 10, p < 0.01, I2 = 81%).

23As shown in Fig. 2, overall mental health, reducing stress and managing mood were equally popular motivating factors, with 80% (N = 6, n = 788, 95% CI 62–93, Q = 134, p < 0.01, I2 = 96%), 78% (N = 4, n = 520, 95% CI 59–92, Q = 50, p < 0.01, I2 = 94%) and 81% (N = 3, n = 464, 95% CI 62–93, Q = 32, p < 0.01, I2 = 94%) of patients agreeing, respectively. Improved sleeping patterns was a motivating factor for 72% of patients (N = 3, n = 464, 95% CI 55.6–86, Q = 20, p < 0.01, I2 = 90%). Enjoyment of exercise was only endorsed by 54% of respondents (n = 807, 95% CI 42.5–64.6, Q = 53, p < 0.01, I2 = 89%). Likert scales studies also found that mental health benefits and enjoyment of exercise scored moderate-to-high for importance as reasons for exercise. The benefits of exercise for self-confidence were assessed in five studies. Although unsuitable for meta-analysis, five studies which assessed the benefits of exercise for self-confidence showed that this is a broadly accepted and valued reason to exercise (See Table 1).

24Social aspects of exercise seen as motivating factors by 27% of patients (N = 3, n = 452, 95% CI 23–32, Q = 0.1, p < 0.097, I2 = 0%). In Likert-scale studies, social aspects scored the lowest of all options presented (Sylvia et al.; Gorczynski et al.; Kane et al. ). Similarly, only a minority of participants saw ‘improving daily routine’ as an important reason for exercise (Sylvia et al.; Wynaden et al. ). In contrast, three independent studies found that ‘professional support’ was perceived as a motivating factor for increasing exercise by the majority of patients (Ussher,; Sylvia et al.; Carpiniello et al. ).

25Fig. 3 shows meta-analyses of barriers towards exercise. Physical illness and poor health was a barrier for 25% of participants (N = 3, n = 359, 95% CI 10–41, Q = 64, p < 0.01, I2 = 92%). Tiredness/low energy was more common, reported by 45% of patients (N = 5, n = 6080, 95% CI 25–67, Q = 322, p < 0.01, I2 = 99%) and rated as 7.4/10 on relevance scales (Sylvia et al. ). Two studies also showed that patients with long-term schizophrenia were more affected by tiredness than healthy controls (Carpiniello et al.; Klingaman et al. ). However, this difference did not exist between patients with first-episode psychosis and healthy controls (Deighton & Addington, ).

26Proportional meta-analyses showed substantial differences in psychological barriers. ‘Stress/depression’ was a barrier to exercise for 61% of respondents (N = 3, n = 5646, 95% CI 43–77, Q = 48, p < 0.01, I2 = 96%), whereas ‘disinterest in exercise’ was a barrier for only 32% (N = 3, n = 5822, 95% CI 16–51, Q = 96, p < 0.01, I2 = 98%). Feeling unsafe and fears of injury were even less common, at 12% (N = 4, n = 5747, 95% CI 9–16, Q = 7, p = 0.07, I2 = 57%) and 8% (N = 3, n = 359, 95% CI 5–11, Q = 0.9, p = 0.64, I2 = 0%), respectively. Data from five studies assessing confidence-related barriers was unsuitable for meta-analyses, but collectively showed that this was only a concern for a minority of participants (7–36%), and to a limited extent; consistently scoring <2/5 on Likert scales of importance (Table 2).

27Data on ‘low motivation’ was also unsuitable for proportional meta-analysis. However, all three studies which assessed this found that motivational deficits were among the most common psychological barriers towards exercise (Carpiniello et al.; Deighton & Addington,; Fraser et al. ). Furthermore, patients with long-term schizophrenia experienced motivational barriers significantly more than healthy controls (Carpiniello et al. ). Again, however, there was no significant difference in the early stages of illness (Deighton & Addington, ).

28The most frequently experienced practical barrier was a ‘lack of support’, reported by 50% of respondents (N = 3, n = 5646, 95% CI 15–86, Q = 240, p < 0.01, I2 = 99%). This was significantly more prevalent among schizophrenia patients than healthy controls (Carpiniello et al.; Klingaman et al. ). People with first-episode psychosis also scored these items higher than controls, although differences were not statistically significant (Deighton & Addington, ). ‘Lack of training partner’ was a moderately ranked barrier, but was regarded as significantly more important by those patients who were interested in increasing their exercise (Faulkner et al. ).

29‘Lack of time’ was the most widely investigated practical barrier, although only 19% of respondents identified this as a barrier (N = 5, n = 6078, 95% CI 11.3–27.2, Q = 68, p < 0.01, I2 = 94%). Three studies using Likert scales also found that time-related barriers were mostly unimportant (Faulkner et al.; Gorczynski et al.; Deighton & Addington, ). Furthermore, ‘lack of time’ was significantly less of a barrier for people with SMI than for healthy controls (Deighton & Addington,; Klingaman et al. ). Only 10% of patients felt that ‘lack of exercise information’ was a barrier (n = 589, 95% CI 7–14, Q = 3.4, p = 0.18, I2 = 42%). Additional data (unsuitable for meta-analysis) on cost and accessibility of exercise services indicated these were of low importance (See Table 2).

30The purpose of this study was to examine the motivating factors and barriers towards exercise among people with SMI, in order to inform the design and delivery of interventions aiming to increase exercise participation. A total of 12 studies (of 6431 psychiatric patients with predominantly schizophrenia/schizoaffective disorders) were identified. As nine of the 12 studies reviewed had been conducted from 2013 onwards, the evidence/data presented can be considered timely and up-to-date.

31Our results show that the primary incentive for engaging in exercise was to improve physical health (Fig. 2). Specifically, weight loss was the single most popular reason for participating in exercise, comparable to the motivating factors identified by the general population (Sherwood & Jeffery, ), and unsurprising given the high rates of overweight and obesity among people with SMI (Vancampfort et al. ). Although weight management can be a key motivating factor for initiating an exercise programme, it is important to note (a) the relatively modest contribution of physical activity to weight loss beyond that achieved through dietary interventions (Haskell et al. ), and (b) that improvements in mental and physical health outcomes in response to exercise interventions are often achieved independent of weight loss (Firth et al. ). While weight management may be an important motivating factor for people with SMI to commence an exercise programme, education and support should be provided to ensure long-term adoption and maintenance regardless of any change in body weight achieved. Furthermore, if weight loss is a primary aim, dietary interventions must be provided as part of best-practice lifestyle interventions (Ward et al. ).

32The high endorsement of ‘fitness’ as an incentive is encouraging, since this is readily improved by exercise interventions in SMI (Vancampfort et al., ), and is more predictive of cardiovascular disease than any other aspect of metabolic health (Myers et al.; Hu et al. ). Health promotion programmes should therefore emphasize the benefit of fitness in order to maximize uptake of exercise in this patient group. Furthermore, interventions should ideally be designed by exercise professionals to ensure that they meet basic principles of exercise prescription, in order to exert significant physiological effects and enable patients to achieve realistic fitness goals.

33Patients also valued the psychological effects of exercise, and 75% of patients viewed stress reduction/mood enhancement as motivating factors. Recent meta-analyses have shown that exercise can significantly improve psychological well-being among people with SMI and reduce depression (Rosenbaum et al.; Firth et al. ). However, the present study also found that stress, depression and low energy often also act as barriers towards exercise.

34The most prominent socio-ecological barrier identified across the studies included in this review was a ‘lack of support’. Nonetheless, the majority of patients felt that exercise supervision would enable them to exercise more (Ussher,; Sylvia et al.; Carpiniello et al. ). This is congruent with the qualitative literature, within which patients with SMI have stipulated that adequate support can overcome many of the barriers faced towards exercise (Soundy et al.; Firth et al. ).

35Although unsupervised interventions which use less resource-intensive methods (such as education or behavioural change techniques) may seem more cost effective than supervised exercise, this may not be the case for people with SMI. Several recent meta-analyses of exercise interventions in this population have shown that interventions which provide professional support have better adherence to physical activity and significantly greater effects on cardiorespiratory fitness (Vancampfort et al.,; Stubbs et al. ). Since both physical activity and fitness are strong predictors of cardiovascular risk and all-cause mortality (Hu et al.; Kodama et al. ), supervised interventions which effectively target these variables may ultimately prove more financially worthwhile for improving long-term health outcomes (Vancampfort et al., ).

36Previous intervention studies have further shown that whereas exercise access and advice is ineffective for increasing physical activity in SMI (Archie et al.; Bartels et al. ), providing adequate social support does enable patients to achieve sufficient levels of moderate-to-vigorous exercise (Bartels et al.; Firth et al. ). Although there is currently a lack of cost-effectiveness research examining supervised exercise in SMI, financial reports of exercise interventions for diabetes, mild depression and heart disease indicate that professionally delivered training programmes produce large economic benefits from avoided health system costs (Deloitte Access Economics, ).

37A strength of these findings is the large number of patients (n = 6431) included in the review. Within this, there was also substantial ethnic diversity within the included samples, with 50% belonging to minority groups. However, all of the studies were conducted in western, developed countries, and thus no studies have examined barriers towards exercise among people with SMI in Asia or developing countries. Furthermore, no studies examined differences in motivations or barriers towards exercise between the different ethnic groups within their respective samples. This gap in the literature should be given further consideration in future research, as studies in the general population have shown that beliefs about exercise, and primary reasons for engaging in physical activity, differ significantly between ethnic groups even within the same country (Dergance et al.; Shiu-Thornton et al. ). Specifically, those in minority ethnic groups may face additional challenges towards exercise, such as feeling unsafe in their neighbours (Fahlman et al. ) or lacking opportunity to engage in culturally appropriate physical activity (Caperchione et al. ). Thus, efforts should be undertaken to identify and provide acceptable physical activity interventions for ethnically diverse populations.

38Despite the large total sample, one limitation of this review is that some of the motivations and barriers assessed in meta-analyses were examined by as few as three studies. Additionally, some eligible studies did not provide any proportional data, and thus were not included in the meta-analysis at all. Nonetheless, a full systematic review of each eligible study was also undertaken, for consideration alongside the meta-analytic outputs, in order to provide a complete account of all relevant findings.

39It should also be considered that the large majority of patients (85%) in this meta-analysis had a diagnosis of schizophrenia, while bipolar disorder and major depressive disorder were relatively under-represented among the eligible studies. Thus, future research should examine if the same motivations and barriers towards exercise identified in this review also generalize to patients with SMIs other than schizophrenia. An online survey study of individuals with high depressive symptoms (but without a confirmed SMI) indicates that our findings will generalize beyond schizophrenia, as the most common barriers towards exercise reported by these individual were again low mood and fatigue (Busch et al. ), as was observed in our SMI samples (Fig. 3).

40A final limitation is that results are based on self-reported data, derived from questionnaires and surveys administered to patients. Therefore, the results could be affected by response bias, or participants lacking sufficient interest/experience with exercise to accurately describe the barriers faced. The findings from patients’ self-report in this study are also congruent with health professionals’ opinions, who also acknowledge the importance of social support in overcoming various barriers towards regular exercise (Soundy et al. ).

41People with SMI value exercise for its ability to improve physical health and appearance, and the psychological benefits. However, mental health symptoms, tiredness and insufficient support present substantial barriers for the majority of patients. Taking this into account, exercise training programmes for people with SMI should be designed to improve exercise capacities and cardiorespiratory fitness, while also providing the necessary levels of supervision or assistance for each patient to overcome psychological barriers and achieve their goals. Such interventions would be motivating and rewarding for patients, resulting in higher levels of exercise engagement. This, in turn, could improve physical health outcomes and facilitate functional recovery in SMI.