There is a known increased morbidity and mortality due to illnesses among people with mental disorders. A large Nordic study has recently shown that psychiatric patients have a reduced life expectancy of 15–20 years compared to the general population (1). This excess mortality is to some extent explained by suicide and accidents, but most of lost year of life is due to excess mortality from somatic disease (2). Detailed analysis of mortality from specific causes of death shows that this excess mortality is caused by virtually any physical illness (3).

A study of patients treated with antipsychotic medication in outpatients’ services in Copenhagen has shown that the prevalence of metabolic syndrome is twice as high in these patients compared with the general population (4). The metabolic syndrome is a cluster of symptoms related to increased risk of cardiovascular events, type II diabetes, and all-cause mortality. The definition of metabolic syndrome varies, but most includes central obesity, raised blood pressure, high fasting glucose levels, and dyslipidemia (raised triglycerides and lowered high-density lipoproteins) (5). The high prevalence of metabolic syndrome among mentally ill is presumably partly due to weight gain and impaired glucose tolerance, which can be side effects of antipsychotic medication (6). Another reason for the high morbidity and mortality may be the lacking treatment of physical illnesses (7). However, there is also reason to believe that lifestyle factors such as smoking, diet, and exercise habits play a role in the excess mortality, morbidity, and high rates of illnesses. Sedentary lifestyle and poor diet is also more frequent than in the general population (8, 9). Low levels of physical activity are associated with increased risk of cardiovascular disease, diabetes, and all-cause mortality (10, 11). However, adherence to behavioral interventions in both the general population as well as in patients with mental illness is a major issue (12). The excess morbidity in patients with severe mental illness can be attributed to unhealthy lifestyle, medication, and that patients with mental illness do not receive same treatment for physical illnesses as mentally healthy patients. In theory, these are all modifiable factors.

With our current knowledge, there is no reason to believe that the physiological benefit of exercise should be different in mentally ill compared to healthy subjects. However, psychiatric symptoms (e.g., anxiety and lack of energy), medication adverse events (e.g., weight gain and sedation), and factors such as social isolation and difficulties in planning, are issues for many patients with either depression or schizophrenia. Due to these challenges, results from lifestyle interventions in the mentally healthy populations cannot be extrapolated to mentally ill populations.

Exercise, which has a low number of side effects, is affordable for most health care systems and is increasingly an objective for research in this patient group. Furthermore, isolated exercise interventions are simple to implement, less costly, and demands less education of staff compared to more complex interventions. The aim of this study was to review the literature on the effect of exercise interventions on fitness variables such as maximal oxygen uptake and strength. Secondly the aim was to assess the effect of exercise interventions on weight and body mass index (BMI). Also, adherence to exercise interventions will be reviewed.

This study was as a systematic review for the primary aims. We included randomized clinical trials allocating patients diagnosed with schizophrenia, bipolar disorder, or depression according to a diagnostic system (i.e., DSM or ICD) to an exercise intervention or a control group/treatment as usual or to an exercise intervention as add-on treatment. The trials had to report outcomes on change in maximal oxygen uptake, repetition maximum or similar in order to be included. We also searched references in relevant papers for additional trials. Secondly, we included studies aiming to identify exercise participation.

The updated search in March 2014 resulted in 5742 hits. Five (13–17) trials reporting the effect of exercise interventions for schizophrenia fulfilled the inclusion criteria and were included. Nine trials (18–26) reporting the effect of exercise in patients with depression fulfilled the inclusion criteria and were included. No trials including patients with bipolar disorder fulfilled the inclusion criteria.

Trials examining the effect of exercise in patients with schizophrenia are small in number and size. There is little evidence that exercise interventions will increase cardiovascular fitness or lead to decreased weight in this patient group. There is convincing evidence that depressed outpatients participating in strength or aerobic training will improve their fitness in terms of maximal oxygen uptake or repetition maximum. However, there is little evidence that exercise interventions in patients will depression will decrease their weight.

Four of five trials assessing exercise in patients with schizophrenia found a positive effect on fitness levels and lack of significance is likely to be a type II error due to the small sample sizes. However, in the larger study by Scheewe et al. (13), the significant differences in aerobic fitness were due to a decrease in fitness in the control group and not an increase in the intervention group. In six trials including patients with depression to aerobic exercise, five had a significant increase maximal oxygen uptake. The trials that did not improve aerobic fitness was a 10-day intervention of inpatients by Knubben et al. (24), which is probably too short an intervention to gain physiological improvement. In a Finnish study of 1639 men (42–60 years) without a history of type II diabetes or atherosclerotic cardiovascular maximal oxygen uptake was associated with 16% reduction of all-cause mortality per one increment in MET (3.5 ml oxygen/min/kg) during an average follow-up of 16 years (28). In the DEMO-I trial, the aerobic exercise groups increased their aerobic capacity equivalent to an increase of 0.83 METs and in the DEMO-II trial, the increase was equivalent to 0.97 METs (19, 20). This suggests that the increase in cardiovascular fitness observed in exercise trials of depressed patients is clinically relevant. On the other hand, no data exists on maximal oxygen uptake beyond the intervention and therefore the current evidence does not allow conclusions or speculations whether the observed increases in cardiovascular fitness is transformed to any benefit for the patients on hard endpoints.

Patients with schizophrenia are unlikely to decrease weight in response to exercise interventions. For patients with depression only one trial reported on weight and found no weight reducing effect of exercise participation (19). However, in that trial a reduction of waist circumference was found in the intervention group reflecting a potential redistribution of weight from abdominal fat to muscle. A recent meta-analysis confirms the effect of both moderate to high intensity aerobic training and strength training for reduction of visceral fat (29). Isolated aerobic exercise as an intervention for weight loss in overweight or obese populations has in meta-analysis been found to reduce weight by 1.6 kg, waist circumference of 2.1 cm, diastolic blood pressure with 1.8 mmHg, and total cholesterol with 1.54 mg/dl for 6 month programs (30). These results suggest that even in mentally healthy populations an isolated aerobic intervention is not likely to induce dramatic changes in weight or waist circumference.

These findings open a discussion whether multiple interventions should be employed, i.e., combined exercise and diet interventions. In mentally healthy populations, the combination of exercise and diet interventions for weight loss has been shown to be superior to exercise alone (31). A trial of patients with severe mental illness allocated to an intervention containing both of advice on diet as well as exercise obtained a weight loss of 3.2 kg after 18 months duration (32). A meta-analysis of 17 non-pharmacological interventions in patients with severe mental illness supports these findings with an overall weight reduction of 3.12 kg (33). However, only five trials reported the effect beyond the intervention with conflicting results on weight and BMI (33). For schizophrenia, only a few and mainly very small trials fulfilled the inclusion criteria for this review. The main reasons for exclusion of trials were lack of diagnosis (e.g., not specifying diagnostic instrument) and the use of mixed interventions (e.g., exercise and diet versus treatment as usual). The results from this review should be interpreted in that context. Taken together, isolated exercise interventions in patients with schizophrenia or depression are unlikely to induce clinically relevant weight loss. On the other hand, there is evidence in favor of mixed interventions for weight loss.

Regarding adherence to exercise intervention in patients with schizophrenia, there was a trend for high attendance in small trials whereas larger trials had lower attendance. Previous studies have tried to identify barriers to physical activity and limited experience of physical, activity engagement, impact of the illness and effects of medication, and anxiety is mentioned by one study (34) and partially confirmed in a smaller study (35). The authors have recently investigated predictors of adherence in exercise trials including patients with depression and found that age and satisfaction with the intervention was associated with higher adherence (36). The age aspect could explain some of the variation observed in attendance in trials with depressed patients. Unfortunately, the majority of included trials do not mention considerations regarding attendance.

Based on the current evidence isolated exercise interventions are unlikely to improve cardiovascular fitness or induce weight loss in patients with schizophrenia, but there are few studies and thus a rather small evidence base. In patients with depression, exercise interventions are likely to induce clinically relevant short term effects, however, due to lack of reporting, little is known about the weight reduction. Future exercise trials in patients with mental illness should preferably measure changes in cardiovascular strength, repetition maximum, and anthropometric outcomes. Ideally, participants should be assessed beyond the intervention to obtain knowledge on lasting effects.

The authors declare that the research was conducted in the absence of any commercial or financial relationships that could be construed as a potential conflict of interest.