Community Mental Health Journal

, Volume 50, Issue 1, pp 81–95

The Effectiveness of Lifestyle Interventions to Reduce Cardiovascular Risk in Patients with Severe Mental Disorders: Meta-Analysis of Intervention Studies

Authors

    • Àmbit Atenció Primària Barcelona, Institut Català de la Salut
  • Luis Miguel Martín-López
    • Hospital del Mar, Parc de Salut Mar
  • Roser Masa-Font
    • Primary Health Care Team 10DÀmbit Atenció Primària Barcelona, Institut Català de la Salut
  • Noemí Olona-Tabueña
    • Àmbit Atenció Primària Barcelona, Institut Català de la Salut
  • Yuani Roman
    • Àmbit Atenció Primària Barcelona, Institut Català de la Salut
  • Jaume Martin-Royo
    • Primary Health Care Team 10IÀmbit Atenció Primària Barcelona, Institut Català de la Salut
  • Silvia Oller-Canet
    • Mental Health Care Centre MaragallÀmbit Atenció Primària Barcelona, Institut Català de la Salut
  • Susana González-Tejón
    • Primary Health Care Team 1DÀmbit Atenció Primària Barcelona, Institut Català de la Salut
  • Luisa San-Emeterio
    • Mental Health Care Centre Sant Martí NordÀmbit Atenció Primària Barcelona, Institut Català de la Salut
  • Albert Barroso-Garcia
    • Primary Health Care Team 10HJÀmbit Atenció Primària Barcelona, Institut Català de la Salut
  • Lidia Viñas-Cabrera
    • Primary Health Care Team 10CÀmbit Atenció Primària Barcelona, Institut Català de la Salut
  • Gemma Flores-Mateo
    • Departament de SalutInstitut Investigació Atenció Primària JordiGol, Institut Català de la Salut
Original Paper

DOI: 10.1007/s10597-013-9614-6

Cite this article as:
Fernández-San-Martín, M.I., Martín-López, L.M., Masa-Font, R. et al. Community Ment Health J (2014) 50: 81. doi:10.1007/s10597-013-9614-6

Abstract

Patients with severe mental illness have higher prevalences of cardiovascular risk factors (CRF). The objective is to determine whether interventions to modify lifestyles in these patients reduce anthropometric and analytical parameters related to CRF in comparison to routine clinical practice. Systematic review of controlled clinical trials with lifestyle intervention in Medline, Cochrane Library, Embase, PsycINFO and CINALH. Change in body mass index, waist circumference, cholesterol, triglycerides and blood sugar. Meta-analyses were performed using random effects models to estimate the weighted mean difference. Heterogeneity was determined using i2 statistical and subgroups analyses. 26 studies were selected. Lifestyle interventions decrease anthropometric and analytical parameters at 3 months follow up. At 6 and 12 months, the differences between the intervention and control groups were maintained, although with less precision. More studies with larger samples and long-term follow-up are needed.

Keywords

Psychotic DisordersSchizophreniaLife StyleFood HabitsExercise

Introduction

Schizophrenic disorders and bipolar disorders included in the group of severe mental disorders (SMD) (Generalitat de Catalunya 2003) have an overall prevalence of approximately 3 % of the general population (Perälä et al. 2007).

The presence of cardiovascular risk factors (CRF) like obesity, dyslipidemia, diabetes mellitus, tobacco and metabolic syndrome is clearly greater in these patients (McEvoy et al. 2005). Thus, even though it is a predominantly young population, life expectancy is 15–30 years less than the rest of the population in its setting (Colton and Manderscheid 2006; Van Citters et al. 2010). The interaction of multiple etiological factors causes this increased the cardiovascular risk (Goff et al. 2005): genetic predisposition, unhealthy lifestyle, clinical and psychosocial factors as well as the increasingly evident use of antipsychotics.

Obesity as an independent CRF and together with the rest of the aforementioned factors is associated with multiple pathologies (Kwon et al. 2006) that cause functional impairment and decreased quality of life. The physical impact associated with the taking of anti-psychotic drugs may reduce adherence to treatment and promote relapse (Álvarez-Jiménez et al. 2006).

Proper control of CRF has been shown to reduce cardiovascular morbidity and mortality (Semergen 2005). Although it is difficult for patients with SMD to increase their physical activity and decrease their food intake due to the limitations of the disease and the adverse effects of antipsychotics (Centorrino et al. 2006), the literature suggests that physical exercise and diet change are possible under the supervision of specific programs.

Some of the studies on SMD patients to assess the effectiveness of different interventions to reduce their high cardiovascular risk have been contradictory and inadequate. Some systematic reviews suggest that schizophrenic patients might benefit from diet and exercise programs, but methodological limitations do not make it possible to reach clear conclusions (Faulkner et al. 2007; Veit and Barnas 2009). A meta-analysis published in 2008 was conclusive and recommended lifestyle interventions in order to reduce weight (Alvarez-Jiménez et al. 2008). Several European medical associations agree to recommend that all programs for SMD patients should include exercise, advice on a healthy diet and behavioral interventions to ensure changes in lifestyle (De Hert et al. 2009).

The present study aims to determine whether interventions to modify lifestyles in patients with severe mental illness decrease anthropometric and biochemical parameters related to the risk of cardiovascular disease in comparison with routine clinical practice.

Methodology

It is a systematic review of the literature, in which there have been several meta-analyses. Below, how each stage was developed is explained.

Search Strategy and Study Selection

The selection criteria were defined to select the diagnosis, type of study, type of interventions and outcome variables measured. The diagnoses included in the group of severe mental illness are as follows: schizophrenia, schizoaffective disorder, bipolar disorder, delusional disorder, recurrent major depressive disorder, obsessive–compulsive disorder and the borderline and schizotypal personality disorder. The type of study admitted was a randomized trial or non-randomized intervention study with a control group. The outcome variables should be changes in at least one of the following parameters: blood pressure, cholesterol, triglycerides, blood glucose, glycosylated hemoglobin, weight, body mass index (BMI), waist circumference (WC), cardiovascular disease incidence (newly diagnosed mortality).

A literature search was conducted of published literature in databases: Pubmed, Cochrane library, CINALH, PsycINFO and EMBASE. The search period was from the beginning of each database until December 2010. There was no limitation on language. Keywords used referring to each element of the research question were: psychiatric diagnosis (“Psychotic Disorders”, “Schizophrenia”, “Bipolar Disorder”, “Obsessive–Compulsive Disorder”, “Borderline Personality Disorder”, “Depressive Disorder, Major”, “Schizophrenia, Paranoid “), and interventions (“Exercise”, “Physical Fitness”, “Life Style”, “Food Habits”, “Cardiovascular Diseases/prevention and control”, “Obesity/prevention and control”, “Hypertension/prevention and control”, “Diabetes Mellitus/prevention and control”, “Dyslipidemias/prevention and control”). Moreover, MeSH terms referring to the type of design selected were included (“Controlled Clinical Trial”, “Cohort Studies”, “Randomized Controlled Trial”, “Intervention Studies”, “Comparative Study”, “Multicenter Study”). Finally, a manual review of the references of the selected articles and review articles was carried out.

Data Extraction

Each selected article was reviewed independently by two researchers who extracted the data needed to describe the study. Discrepancies were resolved through consensus between the two or a third researcher if no agreement was reached. The methodological quality of each study was assessed, also independently, by two investigators. In order to do this, the criteria of the Cochrane collaboration’s tool for assessing risk of bias (Higgins and Green 2011) were used. Each of the criteria was categorized as: Clearly yes/not sure/Clearly no. The criteria for which there were differences between the two evaluators were discussed until a joint decision was reached. A score of between 0 and 24 to the total scale was given in order to perform a subgroup analysis of quality (0 was non-compliance with any criteria and 24 was the fulfillment of all criteria).

Statistical Analysis

For the meta-analysis, the mean change from baseline to follow-up for the intervention and control groups was calculated. This was done by using a paired Student’s t test to obtain the group mean and pooled standard deviation (SD). Standards errors and confidence intervals were converted to SD for the analyses. The measurements of the different variables were converted to the same metric units (kg, cm, mg/dl). Given that the variables are continuous outcome measures, the weighted mean difference (WMD) was estimated using random effects meta-analyses with 95 % CIs. Clinical trials were analyzed according to the intention-to-treat principle. Doing an analysis of different follow-up periods (3, 6 months and 1 year follow up) was decided on. Heterogeneity was quantified with the I2 statistical, which describes the proportion of total variation in study estimates due to heterogeneity (Higgins and Thompson 2002). A meta-regression and subgroup analyses were used to evaluate whether results were different results by the quality of the study (≤13 or >13) or study design (randomized or not randomized). The relative influence of each study on pooled estimates was assessed by omitting one study at a time. Finally, publication bias was assessed using Begg’s test, Egger’s test and funnel plots. Statistical analyses using STATA version 11 was conducted (STATA Corp, College Station, TX, USA).

Results

Of all the references that were identified (Fig. 1), the full text of 55 studies were reviewed, of which 29 were excluded. The most frequent reasons for exclusion were: no control group (7 articles), intervention with a drug treatment, but not lifestyle (9 articles), not being an original study, but of revisions (4 articles), not having any of the response variables analyzed (4 articles) and other reasons (5 articles). The 26 original articles were included in the review.
https://static-content.springer.com/image/art%3A10.1007%2Fs10597-013-9614-6/MediaObjects/10597_2013_9614_Fig1_HTML.gif
Fig. 1

Selection of studies in different stages of the review

Description of Studies Included (Table 1)

In general, the studies are similar to each other with respect to population, age, diagnosis and type of study. There are more differences in terms of sample size, intervention type and the duration of follow up.
Table 1

Description of general characteristics of included studies

Author

Country

Design

Population

Analyzed sample size

Follow-up duration

Intervention

Outcomes

May et al. (1985)

USA

NRCT

Psychiatric inpatients. 20–64 years old

Diagnosis: schizophrenia, schizoaffective disorder, depression, personality disorders

Control = 26

Intervention = 28

24 months

Physical activity, nutritional advice and others

Blood pressure, cardiac frequency

Ball et al. (2001)

USA

RCT

Psychiatric outpatients. 18–60 years old

Diagnosis: schizophrenia and olanzapine, and weight increase of ≥7 %

Control = 11

Intervention = 11

16 weeks

Physical activity, nutritional advice and education. 16 weeks of duration

Weight, BMI

Littrell et al. (2003)

USA

RCT

Psychiatric outpatients. 18 and more years old

Diagnosis: schizophrenia, schizoaffective disorder, and antipsychotic treatment

Control = 35

Intervention = 35

6 months

Educational program of physical activity, nutrition, and lifestyle. 4 months of duration

Weight, BMI

Vreeland et al. (2003)

USA

NRCT

Day hospital

Diagnosis: schizophrenia, schizoaffective disorder, and atypical antipsychotic treatment + weight increase

Control = 15

Intervention = 27

12 weeks

Physical activity and lifestyle. 12 weeks

Weight, BMI

Menza et al. (2004)

USA

NRCT

Outpatients. 18–65 years old

Diagnosis: schizophrenia, schizoaffective disorder, and atypical antipsychotic treatment + weight increase

Control = 25

Intervention = 28

12 months

Educational program of physical activity and nutrition. 12 months of duration

Weight, BMI, blood pressure, waist and abdominal circumference

Brar et al. (2005)

USA

RCT

Inpatient or outpatient. 18 –65 years old

Diagnosis: schizophrenia, schizoaffective disorder, and risperidona +BMI >26

Control = 37

Intervention = 28

14 weeks

Behavioral program to decrease weight.

14 weeks.

Weight, BMI, blood pressure, waist circumference

Evans et al. (2005)

Australia

Virginia

RCT

Psychiatric outpatients

Diagnosis: schizophrenia, schizoaffective disorder, bipolar disorder, depression. Treatment with olanzapine.

Control = 11

Intervention = 23

6 months

Nutritional education during 3 months.

Weight, BMI, waist circumference

McCreadie et al. (2005)

Scottish

RCT

Psychiatric outpatients. Day centre

Diagnosis: schizophrenia

Control = 26

Intervention = 37

18 months

Free fruit and vegetables + nutrition educational program

BMI, CVRF

Skrinar et al. (2005)

USA

RCT

Psychiatric outpatients. Day hospital. 18–55 years old

Diagnosis: schizophrenia, schizoaffective, disorder, bipolar disorder, depression with antipsychotic treatment + weight increase

Control = 11

Intervention = 15

12 weeks

Physical activity and health education. 12 weeks.

Weight, BMI, cardiac frequency, HDL-cholesterol

Álvarez-Jiménez et al. (2006)

Spain

RCT

Primary care outpatients and emergency patients. New psychotic diagnosis. 15–60 years old.

Control = 33

Intervention 28

3 months

Educational program of physical activity and nutrition. 3 months.

Weight, BMI

Beebe et al. (2005)

USA

RCT

Hospital outpatients. 40–63 years old

Diagnosis: schizophrenia

Control = 4

Intervention = 4

16 weeks

Physical activity program

BMI

Kwon et al. (2006)

Seoul Chorea

RCT

Hospital outpatients. 19–64 years old

Diagnosis: schizophrenia, schizoaffective disorder. Olanzapine and weight increase of ≥7 %

Control: 14

Intervention: 29

3 months

Physical activity and diet program. 8 weeks

Weight, BMI

McKibbin et al. (2006)

USA

RCT

Psychiatric outpatients. Day hospital. 40–81 years old

Diagnosis: schizophrenia and diabetes mellitus

Control = 29

Intervention = 28

6 months

Health education program about physical activity, diet and lifestyles.

Weight, BMI, blood pressure, waist circumference

Cholesterol (Total, HDL, LDL), triglycerides.

Scocco et al. (2006)

Italia

RCT

Outpatients

Diagnosis: schizophrenia, schizoaffective disorder. Olanzapine

Control = 8

Intervention = 9

6 months

Diet educational program

Weight

Weber and Wyne (2006)

USA

RCT

Outpatients. 18–65 years old

Diagnosis: schizophrenia, schizoaffective disorder with antipsychotic treatment + BMI ≥25

Control = 7

Intervention = 8

16 weeks

Cognitive-behavioral program of nutrition and physical activity.

16 weeks

Weight, BMI, waist circumference

Jean-Baptiste et al. (2007)

USA

RCT

Outpatients. Adults

Diagnosis: schizophrenia, schizoaffective disorder with antipsychotic treatment + BMI ≥30

Control = 6

Intervention = 8

22 weeks

Educational program of physical activity and nutrition. Walking. 22 weeks

Weight, triglycerides.

Poulin et al. (2007)

Canada

RCT

Outpatients. 18 and more years old

Diagnosis: schizophrenia, schizoaffective disorder, bipolar disorder and antipsychotic treatment. Low physical activity.

Control = 51

Intervention = 59

18 months

Physical activity and nutritional advice. 18 months

Weight, BMI, waist circumference

Cholesterol (Total, HDL, LDL), triglycerides

Wu et al. (2007)

Taiwan

RCT

Hospital inpatients. 18–65 years old

Diagnosis: schizophrenia. Clozapine and BMI >27.

Control = 25

Intervention = 28

6 months

Physical activity and diet

Weight, BMI, waist and abdominal circumference, cholesterol, triglyceride

Forsberg et al. (2008)

Switcher land

RCT

Outpatients

Diagnosis: schizophrenia, schizoaffective, disorder, bipolar disorder.

Control = 33

Intervention = 41

12 months

Physical activity and nutritional advice. 12 months

Weight, BMI, waist circumference, blood pressure, cardiac frequency

Melamed et al. (2008)

Israel

NRCT

Hospital inpatients. Adults.

Adultos

Diagnosis: schizophrenia, schizoaffective, disorder.

Control = 31

Intervention = 28

12 months

Health education program about physical activity, diet and lifestyles. 3 months

Weight, BMI

Wu et al. (2008)

China

RCT

Outpatients. 18–45 years old

Diagnosis: schizophrenia. Antipsychotic treatment and weight increase of ≥10 %.

Control = 29

Intervention = 89

12 weeks

Health education program about physical activity and diet. 12 weeks.

Weight, BMI, waist circumference

Blouin et al. (2009)

Canada

NRCT

Outpatients. 18–65 years old

Diagnosis: schizophrenia, schizoaffective disorder, and second generation antipsychotic treatment.

Control = 10

Intervention = 6

12 weeks

Physical activity and nutritional advice. 12 weeks

Weight, BMI, waist circumference

Marzolini et al. (2009)

Canada

RCT

Outpatients

Diagnosis: schizophrenia, schizoaffective with ≥1 CVRF

Control = 6

Intervention = 7

3 months

Physical activity. 3 months

Weight, BMI, waist circumference, blood pressure

Iglesias-Garcia et al. (2010)

Spain

RCT

Outpatients

Diagnosis: schizophrenia. Antipsychotic treatment and BMI >27

Control = 7

Intervention = 7

3 months

Theoric health education program about physical activity, diet and lifestyles. 3 months

Weight, BMI, waist circumference, blood pressure, cardiac frequency

McKibbin et al. (2010)

USA

RCT

Psychiatric outpatients. Day hospital. 40–81 years old

Diagnosis: schizophrenia and diabetes mellitus

Control = 29

Intervention = 28

12 months

Health education program about physical activity, diet and lifestyles.

Weight, BMI, blood pressure, waist circumference

Cholesterol (Total, HDL, LDL), triglycerides.

Porsdal et al. (2010)

Denmark Norway, Switcher land.

NRCT

Outpatients

Diagnosis: psychosis, and antipsychotic treatment. Weight problems.

Control = 59

Intervention = 278

6 months

Theoric health education program about physical activity, diet

Weight, waist circumference

RCT Randomized controlled Trial, NRCT Non Randomized controlled Trial, BMI Body Mass Index

The region of origin of the studies is, for most the part, North America. However, there are also 3 Asian (Kwon et al. 2006; Wu et al. 2007, 2008), 6 European studies (Álvarez-Jiménez et al. 2006; Forsberg et al. 2008; Iglesias-Garcia et al. 2010; McCreadie et al. 2005; Porsdal et al. 2010; Scocco et al. 2006), 1 Australian (Evans et al. 2005) and 1 Israeli (Melamed et al. 2008) study.

From a design point of view, there are 20 randomized intervention studies and six non-randomized with a control group (Blouin et al. 2009; May et al. 1985; Melamed et al. 2008; Menza et al. 2004; Porsdal et al. 2010; Vreeland et al. 2003).

Most studies are performed on an outpatient basis, although some of them were carried out with only hospitalized patients (May et al. 1985; Melamed et al. 2008; Wu et al. 2007).

As for the target population, some studies only used diagnostic criteria (Beebe et al. 2005; Forsberg et al. 2008; May et al. 1985; McCreadie et al. 2005; Mckibbin et al. 2006; Melamed et al. 2008). Others combine diagnosis and antipsychotic treatment (Álvarez-Jiménez et al. 2006; Blouin et al. 2009; Evans et al. 2005; Littrell et al. 2003; Marzolini et al. 2009; Poulin et al. 2007; Scocco et al. 2006) and several have weight gain and/or BMI as inclusion criterion in addition to diagnosis and treatment (Ball et al. 2001; Brar et al. 2005; Iglesias-Garcia et al. 2010; Jean-Baptiste et al. 2007; Kwon et al. 2006; Menza et al. 2004; Porsdal et al. 2010; Skrinar et al. 2005; Vreeland et al. 2003; Weber and Wyne 2006; Wu et al. 2007, 2008).

Most investigations study less than 30 patients in each group. There are 10 studies that do follow up on more than 30 patients (Álvarez-Jiménez et al. 2006; Brar et al. 2005; Forsberg et al. 2008; Littrell et al. 2003; McCreadie et al. 2005; McKibbin et al. 2006; Melamed et al. 2008; Porsdal et al. 2010; Poulin et al. 2007; Wu et al. 2008) The Wu study (Wu et al. 2007) includes the largest sample in the intervention group (n = 96).

The duration of follow up ranged from 3 months (Álvarez-Jiménez et al. 2006; Blouin et al. 2009; Brar et al. 2005; Iglesias-Garcia et al. 2010; Kwon et al. 2006; Marzolini et al. 2009; Skrinar et al. 2005; Vreeland et al. 2003; Wu et al. 2008), 3–6 months (Ball et al. 2001; Beebe et al. 2005; Evans et al. 2005; Littrell et al. 2003; McKibbin et al. 2006; Porsdal et al. 2010; Scocco et al. 2006; Weber and Wyne 2006; Wu et al. 2007), 12 months (Forsberg et al. 2008; Jean-Baptiste et al. 2007; McKibbin et al. 2010; Melamed et al. 2008; Menza et al. 2004) to more than 1 year (May et al. 1985; McCreadie et al. 2005; Poulin et al. 2007). Some of the articles provide results from several points throughout the follow up period.

Interventions usually lasted 3 months, although some of them were extended to 6 months (Jean-Baptiste et al. 2007; McCreadie et al. 2005; McKibbin et al. 2006, 2010; Menza et al. 2004) and 18 months (Poulin et al. 2007). Most combined physical activity and nutritional advice, although some only focused on diet (Evans et al. 2005; McCreadie et al. 2005; Scocco et al. 2006) or exercise (Beebe et al. 2005; Marzolini et al. 2009). There are interventions that were for individuals, others groups and studies that used both techniques.

The median scores obtained on the quality scale was 13 points (maximum/19, minimum/3). The criteria that were rated worst are those related to the blinding of the patient or of the evaluators. None of the studies meet these criteria. The best rated criteria are those related to the definition of the intervention (only one study had no well-defined intervention: May et al. 1985), the definition of outcome variables (only one study did not have clearly defined variables: May et al. 1985) and the definition of the selection criteria (7 studies did not have them clearly defined: Littrell et al. 2003; May et al. 1985; Melamed et al. 2008; Menza et al. 2004; Porsdal et al. 2010; Scocco et al. 2006; Vreeland et al. 2003).

Meta-Analysis of Body Mass Index (BMI) (Fig. 2)

Fifteen studies evaluated the effect of the intervention on the decrease of the BMI in comparison to routine clinical practice at 3 months follow up. Except for five items, the rest provided conclusions in favor of a greater reduction in BMI in the intervention group. This effect may be due to a weight decreased in the intervention group, while weight in the control group either stayed the same or increased. However both groups showed increased weight in two studies (Evans et al. 2005; Álvarez-Jiménez et al. 2006), but more so far the control group. Nevertheless the difference between them was statistically significant.
https://static-content.springer.com/image/art%3A10.1007%2Fs10597-013-9614-6/MediaObjects/10597_2013_9614_Fig2_HTML.gif
Fig. 2

Weighted mean differences (WMD) of body mass index (BMI) between the intervention group and the control group, and the CIs of 95 % (95 % CI). Result at 3, 6 and 12 months follow-up

At 6 months follow-up the intervention group frequently showed a decrease or stabilization in weight relative to baseline. Average weight tended to increase in the control group. Scocco et al. (2006) shows a similar increase en BMI in both groups.

Of the published works with a 12 month follow-up, only one of the articles (Forsberg et al. 2008) shows similar results in both groups.

Pooled results indicated a significant reduction in BMI in the intervention group versus the control group (WMD = −1.16 kg/m2, 95 % CI −1.72, −0.58, i2 = 92.7 %). To explore the source of heterogeneity, a meta-regression and subgroup analysis were performed (Table 2). We did not find statistically significant differences in body weight reduction based on the quality scale score of the study (P of interaction = 0.264) or if the study was randomized or not (P of interaction = 0.277).
Table 2

Analysis of subgroups according to methodological quality (score greater or less tan 13) and type of design (randomized clinical trial or not)

 

BMI

Waist circumference

Glucose (mg/dl)

Triglyceride

Cholesterol

N

WMD

I2

P value interaction

N

WMD

I2

P value interaction

N

WMD

I2

P value interaction

N

WMD

I2

P value interaction

N

WMD

I2

P value interaction

(95 % CI)

(95 % CI)

(95 % CI)

(95 % CI)

(95 % CI)

Quality (3 months follow-up)

≤13

7

−1.60 (−2.24, −0.97)

77.2 %

 

3

−2.90 (−3.86, −1.94)

1 %

 

1

−1.16 (−4.85, 2.52)

 

2

−48.69 (−131.78, 34.39)

85.8 %

 

3

−4.39 (−13.19, 4.41)

46.7 %

 

>13

8

−0.744 (1.67, 0.17)

95.5 %

0.261

5

−2.44 (−3.91, 0.96)

72.1 %

0.992

2

−7.45 (−10.78, − 4.11)

87.4 %

0.052

1

−17.8 (−24.37, −11.23)

0.630

1

−13.53 (−16.85, − 10.22)

 

0.315

Randomization (3 months follow-up)

Yes

11

−0.90 (−1.61, 0.190)

93.4 %

 

5

2.20 (−3.61, 0.78)

56.3 %

 

2

−5.29 (−12.96, 2.37)

93.9 %

 

1

−10.8 (−31.39, 9.79)

 

2

−4.72 (−17.18, 7.73)

66.4 %

 

No

4

−1.82 (−3.03, 0.62)

87.7 %

0.277

5

−3.20 (−4.12, −2.33)

1 %

0.608

1

−5.58 (−7.80, −3.35)

0.952

2

−51.14 (−127.04, 24.76)

84.7 %

 

2

−9.35 (−21.35, 2.65)

60.3 %

0.799

Quality (6 months follow-up)

≤13

4

−1.52 (−2.25, −0.79)

45.5 %

 

1

−4.34 (−6.61, −2.06)

 

1

0.46 (−3.26 4.18)

 

1

−63.1 (−85.63, −40.57)

 

1

−10.3 (−15.87, −4.73)

  

>13

2

−1.26 (−1.77, −0.76)

26.5 %

0.738

2

−4.17 (−5.22, −3.13)

1 %

0.858

1

−7.38 (−9.43, −5.33)

 

1

−30.26 (−36.98, −23.54)

 

1

−20.49 (−23.71, −17.28)

 

Randomization (6 months follow-up)

Yes

4

−1.52 (−2.25, −0.79)

45.5 %

 

1

−4.34 (−6.61, −2.06)

 

1

0.46 (−3.26 4.18)

 

1

−63.1 (−85.63, −40.57)

 

1

−10.3 (−15.87, −4.73)

 

No

2

−1.26 (−1.77, −0.76)

26.5 %

0.738

2

−4.17 (−5.22, −3.13)

1 %

0.858

1

−7.38 (−9.43, −5.33)

 

1

−30.26 (−36.98, −23.54)

 

1

−20.49(−23.71, −17.28)

 

Quality (12 months follow-up)

≤13

1

−3.3 (−3.89 −2.71)

 

2

−5.52 (−7.98, −3.06)

73.7 %

 

2

−9.32 (−13.29, −5.34)

69.4 %

2

−5.62 (−6.15, −5.08)

1 %

 

 

>13

3

−1.56 (−2.60, − .53)

73.0 %

0.070

Randomization (12 months follow-up)

Yes

1

0.10 (−1.53, 1.73)

 

2

−5.52 (−7.98, −3.06)

47.3 %

 

2

−9.32 (−13.29, −5.34)

69.4 %

 

2

−5.62 (−6.15, −5.08)

1 %

 

No

3

−2.48 (−3.32, −1.63)

81.6 %

0.028

 

 

At 6 months follow up and combining results from six studies, the weighted mean difference (WMD) is −1.42 kg/m2 (95 % CI −1.83 to −1.00, i2 = 37.3 %). Neither were any differences found in the subgroup analysis (P = 0.738 interaction of both).

Finally, four studies did a follow-up of 12 months. The difference between the two groups is −2.03 kg/m2 (95 % CI −3.01 to −1.05, i2 = 85.2 %). In this case, there are differences in the analysis based on the quality scale score (P of interaction = 0.07) and in accordance with randomization (P of interaction = 0.028).

Meta-Analysis of Waist Circumference (Fig. 3)

The eight studies in which waist circumference was calculated at 3 months of follow-up refer a reduction of the measure in the intervention group. On the other hand the value in the control group stayed the same or increased. This was true except for the Marzolini et al. (2009) study, in which a reduction in waist circumference in both groups was described.
https://static-content.springer.com/image/art%3A10.1007%2Fs10597-013-9614-6/MediaObjects/10597_2013_9614_Fig3_HTML.gif
Fig. 3

Weighted mean differences (WMD) of waist circumference between the intervention group and the control group, and the CIs of 95 % (95 % CI). Result at 3, 6 and 12 months follow-up

The results show a decrease a decrease in waist circumference in the intervention group and an increase in the control group in the remaining follow-up periods.

At 3 months follow up, the WMD is −2.58 cm (95 % CI −3.52 to −0.87, i2 = 61.9 %), at 6 months −4.20 cm (−5.15 to −3.26 95 %, i2 = 0.0 %) and at 12 months follow-up −5.52 cm (−7.98 to −3.06 95 %, i2 = 73.7 %). The latter average is calculated by combining two studies. There were no differences in any of the subgroup analyses based on the quality scale score or randomization, at three and 6 months (P > 0.60).

Meta-Analysis of Lipid Levels and Blood Glucose (Figs. 4, 5, 6)

Total cholesterol, triglycerides and blood glucose were measured in different studies (although few). The values at follow-up were variable in terms of blood glucose, Blood glucose went down in the intervention group but it was similar or rose in the control group in the majority of the studies. With respect to triglycerides all the studies refer to a decrease in the intervention group and an increase in the control group. Only Wu et al. (2007) describes a reduction of the triglycerides value in the control group. Average cholesterol was lowest at follow- up in the intervention group, although it was also reduced in the control group in some studies (Wu et al. 2007; Blouin et al. 2009; Skrinar et al. 2005).
https://static-content.springer.com/image/art%3A10.1007%2Fs10597-013-9614-6/MediaObjects/10597_2013_9614_Fig4_HTML.gif
Fig. 4

Weighted mean differences (WMD) of glucose between the intervention group and the control group, and the CIs of 95 % (95 % CI). Result at 3 and 6 months follow-up

https://static-content.springer.com/image/art%3A10.1007%2Fs10597-013-9614-6/MediaObjects/10597_2013_9614_Fig5_HTML.gif
Fig. 5

Weighted mean differences (WMD) of triglycerides between the intervention group and the control group, and the CIs of 95 % (95 % CI). Result at 3, 6 and 12 months follow-up

https://static-content.springer.com/image/art%3A10.1007%2Fs10597-013-9614-6/MediaObjects/10597_2013_9614_Fig6_HTML.gif
Fig. 6

Weighted mean differences (WMD) of cholesterol (mg/dl) between the intervention group and the control group, and the CIs of 95 % (95 % CI). Result at 3, 6 and 12 months follow-up

The WMD between the intervention and control groups was calculated. The estimation of cholesterol after 3 months of follow-up was −8.32 mg/dl (95 % CI −14.57 to −2.08, i2 = 63.5 %). At 6 months, this difference increases for the intervention group: −15.66 mg/dl (95 % CI −25.64 to −5.68, i2 = 89.7 %). Something similar occurs with triglycerides up to 6 months of follow up (weighted mean difference −24.64 mg/dl at 3 months and −44.85 mg/dl at 6 months). At 12 months, the magnitude of the difference between groups decreases, although it is still statistically significant (weighted mean difference −5.62 mg/dl, 95 % CI −6.15 to −5.08). The i2 is greater than 86 % in the three analyses.

In terms of glucose, the results are less homogeneous in the different follow-up periods. The mean difference between the intervention and control groups is no longer significant at 6 months, although again becomes significant for the intervention group at 1 year of follow up.

Subgroup analysis according to the quality scale score and based on the randomized allocation of patients to study groups did not show statistically significant differences between the two groups, except for the glucose outcome variable at 3 months (according to their score on the quality scale: P value for interaction = 0.052).

Publication Bias

Asymmetry was not observed in any of the five funnel plots calculated for each measurement result. Neither did tests that assess asymmetry turn out statistically significant (minimum value of the Begg’s test = 0.133, minimum value of Egger’s test = 0.07), except in the analysis done with studies that evaluate the BMI as a result measure. In this case, although the Begg’s test was not significant (P = 0.498), the Egger′s test was (P = 0.001).

Discussion

Patients with severe mental illness have a higher prevalence of CRF. There are different studies in which different types of intervention achieved some reduction in the risk factors in these patients. This meta-analysis evaluated the effectiveness of different interventions to reduce CV risk factors in patients with an SMD.

All the studies have some characteristics in common: the outcome variables studied are similar measures and their reading is objective and little susceptible to bias. However, there are methodological issues such as the measurement of some of the variables, the type of intervention applied and the lack of blinding which may have influenced the results. The measurement of the variables that may be most subject to error are the variables of weight and waist circumference if there are no standardized instructions available. The lack of a homogeneous measurement standard among different evaluators can lead to systematic bias. In all the studies except for one (May et al. 1985), the measurement of outcome variables was well-defined. The analytical parameters are laboratory measurements and their estimation may differ from study to study, However, what is compared is the difference between before and after in subjects from the same study, in which the values were measured in the same way. This can lead to a decrease in the variability of the measure between studies.

Interventions can be different in terms of duration, the techniques used and if the type of intervention was individual or group. All have required a lifestyle intervention that, in general, has combined diet and exercise. There are those in which only diet and not physical activity is taken into account (Evans et al. 2005; McCreadie et al. 2005; Scocco et al. 2006; Porsdal et al. 2010). Some interventions were carried out with behavioral treatments and others were done with problem resolution techniques.

None of the trials is blinded to the intervention received. Being an educational intervention cannot cover up this aspect. Being able to perform a concealed allocation and to evaluate the outcome variables in a blinded way are important in these designs. As seen in the results, this last criterion is not fulfilled by any of the clinical trials reviewed.

The clearest results and those obtained by a great number of studies are those that refer to the greater reduction (even though there is less gain in some studies) in BMI at three, at 6 and at 12 months follow-up in the group receiving the intervention group compared to the control group. This result agrees with the meta-analysis published by Alvarez-Jiménezet al. (2008), which analyzed weight loss at 3 months. In our study, we were able to recognize that this result holds after a longer follow-up.

As for the waist circumference, although it loses precision in the weighted estimate, the result is also a greater reduction for the intervention group. This parameter is very important because it is currently considered a cardiovascular risk factor that is more predictive than the BMI, especially in its relationship to the metabolic syndrome (Grundy et al. 2005).

The analytical parameters were not included in all the studies. In studies that measure cholesterol and triglycerides, a weighted mean difference in favor of the intervention group and therefore a reduction in the values of these parameters was seen. Blood glucose behaves less clearly in different follow-up periods. These analytical parameters are as important to measure as outcome variables in clinical trials as they make up a part of the cardiovascular risk tables from which the risk of cardiovascular events is quantified (Marrugat et al. 2003).

As for the analysis of how they affect different types of intervention, the implementation of interventions in which advice and/or dieting are combined with physical exercise are those that derive significant outcomes.

The studies which evaluate interventions that are not combined are those that are not conclusive in terms of results. In McCreadie et al. (2005), the intervention is only dietary and uses the increased consumption of fruits and vegetables and no physical or analytical parameters primary variables. That fact is what did not allow it to be included in the meta-analysis. The Scocco intervention (Scocco et al. 2006) is dietary but not physical, and do not show differences between the study groups. Marzolini et al. (2009) includes a physical activity intervention but not diet intervention in its inquiry. The results are favorable in terms of an increase in the physical capacity but the anthropometric measurements are not modified.

Other studies in which the results of change are not significant have small samples that have led to this result: Skrinar et al. (2005) does show differences in the BMI for the intervention group involved in a diet and exercise program for 12 weeks, but it does not prove to be significant. It is probably due to reduced sample size (20 patients included in the analysis). The same applies to the study by Weber and Wyne (2006) which found a greater reduction in the BMI in the intervention group but with such a small sample (15 patients total) that it does not reach statistical significance. Significant results were not obtained in the study by Iglesias-Garcia et al. (2010) but it included very few patients (14 in total) and the intervention consisted in informing and advising about diet and exercise, but without physical activity.

The aim was to avoid heterogeneity by including clinical trials with a control group. Other designs were excluded (cross-sectional or case–control studies). However, heterogeneity was present for several outcomes, which were not explained by subgroup analyses. In this analysis, the following is noteworthy: the studies that measured the BMI outcomes at 12 months were mostly non-randomized and presented a positive result for the intervention group (McKibbin et al. 2010; Melamed et al. 2008; Poulin et al. 2007). The only randomized study with 12 months follow up had a similar result relative to the change in the BMI for both groups (Forsberg et al. 2008). With the other outcomes, the results by subgroups at 12 months follow-up are not assessable as there are only one or two studies included.

Publication bias is another concern with meta-analyses. This meta-analysis only includes studies that are actually published. With tests and visual inspections, such as the Egger’s method used in this review, it was possible to exclude publication bias with some confidence except in body mass index outcomes.

One advantage of pooling data from clinical trials to investigate this important clinical issue is better generalizability, as this analysis combines data from heterogeneous populations. In this review, the populations are quite similar in terms of illness (schizophrenia and other SMD) and the type of treatment received (antipsychotic drugs).

It can be concluded that there is a difference in BMI, WC, cholesterol, glucose and triglycerides in the group receiving a combined intervention of diet and physical activity than in the control group at 3 months follow up. This difference is a greater reduction o less increase in the intervention group relative to the control group. This change is maintained at 6 and 12 months in terms of BMI. For the other parameters, although the measurement result is also in favor of intervention group, the precision of the measurement is less. This leads to recommending further studies to make measurements at longer-term follow-ups with larger samples. Moreover, it is important to design and carry out clinical trials with standardized measurements of their variables and to apply techniques to the intervention in order to foment patient compliance with follow-up plan.

Acknowledgments

This study supported by a grant from the National Institute Carlos III, FIS PI11/01861.

Copyright information

© Springer Science+Business Media New York 2013