s3-eu-west-1.amazonaws.com€¦ · web viewbehavioural interventions addressing lifestyle in...
TRANSCRIPT
Supplementary material
Trial protocol......................................................................................................... 2Methods/design...........................................................................................................................................3
Search strategy.................................................................................................... 10
Table S1: Trial, Population and Treatment Characteristics....................................12
Figure S1: Flow chart............................................................................................17
Table S2: Risk of bias assessments.......................................................................18
Table S3: ASPECT-R scores....................................................................................20
Table S4: Grades of Recommendation, Assessment, Development and Evaluation (GRADE): Summary of findings.............................................................................21
Trial Sequential Analysis......................................................................................23Figure S2: Trial sequential analysis of BMI:................................................................................................23
Figure S3: Trial sequential analysis of 5% weight loss................................................................................24
Figure S4: Trial sequential analysis of quality of life:..................................................................................25
Funnel plots.........................................................................................................26Figure S5: Funnel plot of weight (BMI).......................................................................................................26
Figure S6: Funnel plot of weight (kg)..........................................................................................................27
Figure S7: Funnel plot of weight (BMI) (maintenance):..............................................................................28
Figure S8: Funnel plot of weight quality of life...........................................................................................29
Figure S9: Funnel plot of weight cholesterol..............................................................................................30
Figure S10: Funnel plot of systolic blood pressure.....................................................................................31
Figure S11: Funnel plot of waist circumference.........................................................................................32
Figure S12: Funnel plot for all-cause discontinuation.................................................................................33
Forest plots..........................................................................................................34Figure S13: Forest plot of weight (kg).........................................................................................................34
Figure S14: Forest plot of quality of life......................................................................................................35
Figure S15: Forest plot of systolic blood pressure......................................................................................36
Figure S16: Forest plot of total cholesterol................................................................................................37
Figure S17: Glucose....................................................................................................................................38
Figure S18: Waist circumference................................................................................................................39
Figure S19: Forest plot of all-cause discontinuation:..................................................................................40
1
Trial protocol
Lifestyle interventions for weight management in patients with serious mental illness: A systematic review, meta-analysis and meta-regression analysis exploring the moderators and mediators of treatment effects
Background
Schizophrenia is a severe and life shortening disease, with life expectancy being reduced by up to 20 years for males and 15 years for females compared to the general population.1 About 60% of the excess mortality is due to physical illness, with cardiovascular disease being the single largest cause of death.2 The unhealthy lifestyle of patients with schizophrenia is a modifiable factor contributing to the high risk of cardiovascular disease3. Indeed, recent randomized clinical trials4–6 have succeeded in initiating weight loss through individual health promotion, leading to a number of reviews7, guidelines8,9 and meta-analyses10–12 concluding that lifestyle interventions for patients with SMI can now be called evidence-based practice. We fully recognize the urgent need for action, but we have concerns about high degree of heterogeneity and the real-world effectiveness of these interventions. To prevent premature recommendations, it is important to be aware of the different moderators and mediators of efficacy of healthy lifestyle interventions for patients with SMI. These will be explored in this analysis, with special focus on internal validity measured as risk of bias, and external validity measured as effectiveness in the real world as opposed to efficacy.
The quality, or risk of bias, can be assessed with several tools. Methodological studies show that trials with unclear or inadequate methodological quality regarding bias domains may be associated with bias (systematic error, the overestimation of benefits, and the underestimation of harms) when compared to trials using adequate methodology.13–15
The ability to address real-world effectiveness can be assessed using rating scales to quantify to which degree the design is pragmatic.16,17 RCTs can be categorized as explanatory (exploring efficacy) or pragmatic (exploring effectiveness), neither being intrinsically superior to the other, but answering different research questions. The explanatory trial investigates a potential causal mechanism under highly controlled settings, answering the question “Can this work?”, while the pragmatic trial investigates whether an intervention is feasible in a real-world setting, answering the question “Will this work?”. Most trials exist on a continuum between these two extremities. A clearly defined research question is crucial in planning a clinical trial, but similarly important in planning meta-analyses, as pooling all RCTs could be misleading due to heterogeneity in trial design.
Equally important questions are “In whom will this work?” and “What part of these complex interventions works?” To answer these questions, an exploratory approach to characteristics of populations and interventions is desired to identify leads for defining design that are associated with greater benefits or harms.
2
Behavioural interventions addressing lifestyle in patients with SMI need special considerations. The composition of the sample results from exclusion/inclusion criteria and the process of recruitment. Patients with severe symptoms of SMI, substance abuse, unstable medication and comorbid medical disorder, are often excluded, resulting in limited external validity. Furthermore, individuals volunteering to behavioural trials are likely to be more motivated and well functioning than the clinical population as a whole. Behavioural interventions are defined as interventions designed to affect the actions a person takes regarding health. Financial resources and human engagement in clinical trials will often exceed possibilities in clinical settings, enabling interventions that are not transferable to real world. Based on these considerations, explanatory trials are more likely to report positive results.
Aim and hypotheses
We aim to summarize the effect of non-pharmacologic weight loss and healthy lifestyle interventions compared to a randomized control condition in the severely mentally ill. We further aim to investigate moderators and mediators of treatment effects, including trial design, patient and intervention characteristics, with a particular focus of efficacy versus effectiveness trials and study bias/quality.
Based on prior literature, we hypothesized that healthy lifestyle interventions can be successful in reducing body weight and improving metabolic status in the severely mentally ill. However, we further hypothesized that the overall effect size will be modest and that the effect size is diminished in trials with a pragmatic as opposed to explanatory design, as well as trials with less bias as opposed to trials with high risk of bias.
Methods/designDesign
Systematic review and meta-analysis with sensitivity, subgroup and meta-regression analyses.
Search strategy
The search will be conducted in the following electronic databases: CENTRAL, MEDLINE, EMBASE, and Science Citation Index (Web of Science), using MeSH terms or similar when possible. Search terms will include the following:
Schizophrenia or schizophrenic or psychotic or schizoaffective disorder or serious mental illness or major depression or severe mental disorder or bipolar disorder or mania or severe mental illness
AND
3
Weight or weight loss or weight management or weight gain or BMI or body mass index or weight reduction or cardiovascular risk or obesity or abdominal obesity
AND
Lifestyle intervention or health promotion or physical activity or exercise or diet or nutrition
AND
Random or randomized or randomly
Additional Sources: reference lists of relevant reviews and included studies
Search Date: From database inception to September 2016
Inclusion criteria
Participants should be diagnosed with major depression, schizophrenia, schizoaffective psychosis or bipolar disorder
Participants aged >17 years of both sexes. Individual lifestyle interventions, defined as interventions designed to affect the action a person takes
regarding health from an individual level: Interventions to manage weight include efforts to modify energy balance through improved diet or
increased physical activity or both. The approaches might include elements theoretically founded techniques to modify behavior, like psychoeducation, psychological counseling, motivational interviewing, stages of change or cognitive therapy
Studies of lifestyle interventions for weight loss can be delivered across any type of setting, including community settings, mental health settings or primary care settings
Randomized clinical trials. Allocation is perceived as randomized when terms including ‘randomly’, ‘random’, and ‘randomization’ are used.
No restriction with regards to type of publication (that is, we will include abstracts and full text reports). No restriction with regards to language (that is, we will translate foreign language articles) The trials had to allocate participants to a lifestyle intervention versus a concurrent control group or using
lifestyle intervention as an add-on to treatment as usual. Reporting on either weight gain prevention or weight loss
Exclusion criteria
Trials investigating the effect of structural or environmental modifications Trials using combination of lifestyle interventions and any pharmacological agents
4
Outcomes
The two primary questions to be answered in this meta-analysis are:
1. Does lifestyle interventions work to manage weight, and is the effect clinically significant and sustainable?2. Are there any adverse effects?
Thus the two primary outcomes are 1) weight measured on a continuous scale and as number needed to treat (NNT) to reach clinically significant effect on weight (≥5%) using results immediately post intervention and follow up; 2) adverse events: self-reported (such as perceived health, quality of life) and weight gain. Secondary outcomes are1) systolic blood pressure; 2) hemoglobin A1c; 3) total cholesterol 4) waist circumference. Exploratory outcomes includes the analyses of mediators and moderators answering the following questions:
Can this work (internal validity)?
Will this work (external validity)?
For whom will it work (population characteristics)?
What kind of interventions will work (intervention characteristics)?
Is the effect sustainable?
1. Internal validitya. risk of bias (high vs low as per Cochrane Risk of Bias Tool)b. data analysis (Intent-to-treat, i.e., last-observation-carried-forward or mixed models, vs observed
cases analyses
2. External validitya. study origin (Europe, vs North-America vs Asia vs rest)b. % drop out in intervention vs control conditionsc. high vs low explanatory /pragmatic design (ASPECT-R score)
3. Population characteristicsa. Mean ageb. Gender distributionc. Diagnosesd. Baseline weighte. Illness durationf. Global assessment functioningg. Negative symptomsh. Cognitive functioningi. % living in supported housingsj. % using illegal drugs k. inpatient/daypatient vs outpatientl. Pattern of medicationm. Prevention (starting healthy lifestyle intervention together with a weight-gain producing psychotropic
medication) vs intervention (starting healthy lifestyle intervention after weight gain has already occurred on psychotropic treatment)
5
4. Interventions characteristicsa. Durationb. Number of attended sessionsc. Intervention modality (exercise, diet or both)d. Intervention setting (group vs individual vs mixed group and individual treatment)e. Program fidelity (number of sessions attended)f. Maintenance effect
Study selection
Two investigators (HS + KBJ) will independently examine titles and abstracts to remove obviously irrelevant reports. Two investigators (HS + KBJ) will examine the remaining full text reports to identify eligible studies. Any disagreement in the initial search and full text review process will be resolved by consensus or discussion with CH or JK. The final list of included studies will be reviewed independently by all researchers (HS, xx, xx,) to ensure consensus.
Data extraction
Two authors (HS, ASJ) will independently extract data, using pre-piloted forms and ASPECT-R working sheets. Any discrepancies in the data extraction or inclusion/exclusion of trials will be resolved by referring to the original papers. CH or JK will assist as adjudicators in case of disagreement. The authors have previously published trial reports assessing the effect of lifestyle counseling. To avoid academic bias, a third assessor (xx) will perform the assessment of bias and design for this trial.
Risk of bias assessment
Methodological studies have shown that trials with unclear or inadequate methodological quality regarding bias domains may be associated with bias (systematic error, the overestimation of benefits, and the underestimation of harms) compared to trials using adequate methodology.13–15 Definitions in the assessment of bias risk of a trial18 will be used according to the Cochrane Handbook for Systematic Reviews of Interventions including the following domains: allocation sequence generation, allocation concealment, blinding of participants and personnel, blinding of outcome assessors, incomplete outcome data, selective outcome reporting, for-profit bias, and other bias. Trials with low risk of bias in all domains are characterized as low risk. However, as lifestyle trials do not have the opportunity to blind participants and personnel, we defined that trials with low risk of bias in the domains of allocation concealment, blinding of outcome assessors and intention-to-treat will be characterized as trials with ‘lower risk of bias. To be conservative, trials with no report on any of these three domains will be characterized as high risk.
Aspect-R
The ASPECT-R (A Study Pragmatic-Explanatory Characterization Tool-Rating; ©2014 Janssen Pharmaceuticals, Inc.), assessing six domains that are specifically related to the explanatory-pragmatic spectrum, was developed to permit post-hoc evaluation of these domains in published RCTs.19–21 The six domains covered with ASPECT-R, are i) eligibility criteria ii) intervention flexibility iii) practice setting/practitioner experience iv) follow up intensity/duration v) outcome(s) vi) participant compliance assessment. The authors have provided definitions of terminology and descriptive anchors for each of the six domains. The domains are rated from 0 to 6, where 0 is considered extremely explanatory and 6 extremely pragmatic. Using an Excel®-based file with cover page cells for entry of the study
6
objective(s) and study population of interest, the ASPECT-R instrument has individual domain-related worksheets where the user rates each of the six domains. Each of the six domain worksheets has a section provided for the user to summarize and record the rationale for their domain scoring.
Data synthesis and analysis
This meta-analysis will be conducted using Comprehensive Meta-Analysis V3 (http://www.meta-analysis.com).
In order to pool results from studies reporting either body weight or BMI related outcomes, estimates of standardized mean difference (SMD) for each individual study will be carried out. SMD is the mean difference in weight between the healthy lifestyle intervention and control groups dived by the pooled standard deviation. The result is a unitless effect size measure. By convention, SMD effect sizes of 0.2, 0.5, and 0.8 are considered small, medium, and large, respectively.22 For the calculations, we will use either change from baseline to endpoint (preferred) or endpoint scores (only preferred if change score results were skewed, i.e., SD >twice the mean). Additionally, weighted mean difference (WMD) will calculated for weight change in kilograms (kg) and BMI (kg/m2) to generate clinically intuitive results. If long term follow up is reported, these will be used in subgroup analyses. For dichotomous variables, we will calculate the relative risks with a 95% confidence interval. Intention-to-treat (ITT) data will always be preferred, but observed cases (OC) data will also allowed. Results from random-effects analyses23 will be used throughout.
The degree of heterogeneity will be quantified using the I-squared statistic24 and the chi-square test of homogeneity) with I-squared >50% may represent substantial heterogeneity, and >75% may represent considerable heterogeneity. Heterogeneity will be explored by analysis of sub-groups and meta-regression.
Publication bias will be assessed by visual inspection of a funnel plot and by Egger’s test.25
Subgroup analyses and meta-regression
The possible effects of a number of variables on the primary outcome is categorized as exploratory outcomes. These will be explored in subgroup analyses for categorical variables and meta-regression for continuous variables.
1. Nordentoft M, Wahlbeck K, Hällgren J, et al. Excess mortality, causes of death and life expectancy in 270,770 patients with recent onset of mental disorders in Denmark, Finland and Sweden. PLoS One. 2013;8(1):e55176. doi:10.1371/journal.pone.0055176.
2. Lawrence D, Hancock KJ, Kisely S. The gap in life expectancy from preventable physical illness in psychiatric patients in Western Australia: retrospective analysis of population based registers. Bmj. 2013;346(may21 1):f2539-f2539. doi:10.1136/bmj.f2539.
3. Dipasquale S, Pariante CM, Dazzan P, Aguglia E, McGuire P, Mondelli V. The dietary pattern of patients with schizophrenia: a systematic review. J Psychiatr Res. 2013;47(2):197-207. doi:10.1016/j.jpsychires.2012.10.005.
4. Daumit GL, Dickerson FB, Wang N-Y, et al. A Behavioral Weight-Loss Intervention in Persons with Serious Mental Illness. N Engl J Med. 2013;368(17):1594-1602. doi:10.1056/NEJMoa1214530.
5. Green CA, Yarborough BJH, Leo MC, et al. The STRIDE weight loss and lifestyle intervention for individuals taking antipsychotic medications: a randomized trial. Am J Psychiatry. 2015;172(1):71-81. doi:10.1176/appi.ajp.2014.14020173.
6. Bartels SJ, Pratt SI, Aschbrenner K a, et al. Clinically significant improved fitness and weight loss among overweight persons with serious mental illness. Psychiatr Serv. 2013;64(8):729-736.
7
doi:10.1176/appi.ps.003622012.
7. McGinty EE, Baller J, Azrin ST, Juliano-Bult D, Daumit GL. Interventions to Address Medical Conditions and Health-Risk Behaviors Among Persons With Serious Mental Illness: A Comprehensive Review. Schizophr Bull. July 2015. doi:10.1093/schbul/sbv101.
8. Cooper SJ, Reynolds GP, Barnes T, et al. BAP guidelines on the management of weight gain , metabolic disturbances and cardiovascular risk associated with psychosis and antipsychotic drug treatment. J Psychopharmacol. 2016;30(8):717-748. doi:10.1177/0269881116645254.
9. Kuipers E, Yesufu-Udechuku A, Taylor C, Kendall T. Management of psychosis and schizophrenia in adults: summary of updated NICE guidance. BMJ. 2014;348:g1173.
10. Caemmerer J, Correll CU, Maayan L. Acute and maintenance effects of non-pharmacologic interventions for antipsychotic associated weight gain and metabolic abnormalities: a meta-analytic comparison of randomized controlled trials. Schizophr Res. 2012;140(1-3):159-168. doi:10.1016/j.schres.2012.03.017.
11. Gierisch JM, Nieuwsma J a, Bradford DW, et al. Pharmacologic and behavioral interventions to improve cardiovascular risk factors in adults with serious mental illness: a systematic review and meta-analysis. J Clin Psychiatry. 2014;75(5):e424-40. doi:10.4088/JCP.13r08558.
12. Bruins J, Jörg F, Bruggeman R, Slooff C, Corpeleijn E, Pijnenborg M. The effects of lifestyle interventions on (long-term) weight management, cardiometabolic risk and depressive symptoms in people with psychotic disorders: a meta-analysis. PLoS One. 2014;9(12):e112276. doi:10.1371/journal.pone.0112276.
13. Wood L, Egger M, Gluud LL, et al. Empirical evidence of bias in treatment effect estimates in controlled trials with different interventions and outcomes: meta-epidemiological study. BMJ. 2008;336(7644):601-605. doi:10.1136/bmj.39465.451748.AD.
14. Schulz KF, Chalmers I, Hayes RJ, Altman DG. Empirical evidence of bias. Dimensions of methodological quality associated with estimates of treatment effects in controlled trials. JAMA. 1995;273(5):408-412.
15. Moher D, Pham B, Jones A, et al. Does quality of reports of randomised trials affect estimates of intervention efficacy reported in meta-analyses? Lancet (London, England). 1998;352(9128):609-613. doi:10.1016/S0140-6736(98)01085-X.
16. Thorpe KE, Zwarenstein M, Oxman AD, et al. A pragmatic-explanatory continuum indicator summary (PRECIS): a tool to help trial designers. J Clin Epidemiol. 2009;62(5):464-475. doi:10.1016/j.jclinepi.2008.12.011.
17. Tosh G, Soares-Weiser K, Adams CE. Pragmatic vs explanatory trials: the pragmascope tool to help measure differences in protocols of mental health randomized controlled trials. Dialogues Clin Neurosci. 2011;13(2):209-215.
18. Higgins JPT, Altman DG, Gotzsche PC, et al. The Cochrane Collaboration’s tool for assessing risk of bias in randomised trials. BMJ. 2011;343:d5928.
19. Alphs LD, Bossie CA. ASPECT-R-A Tool to Rate the Pragmatic and Explanatory Characteristics of a Clinical Trial Design. Innov Clin Neurosci. 2016;13(1-2):15-26.
20. Bossie CA, Alphs LD, Williamson D, Mao L, Kurut C. Inter-rater Reliability Assessment of
8
ASPECT-R: (A Study Pragmatic-Explanatory Characterization Tool-Rating). Innov Clin Neurosci. 2016;13(3-4):27-31.
21. Bossie CA, Alphs LD, Correll CU. Long-acting injectable versus daily oral antipsychotic treatment trials in schizophrenia: pragmatic versus explanatory study designs. Int Clin Psychopharmacol. 2015;30(5):272-281. doi:10.1097/YIC.0000000000000082.
22. Cohen jacob. Statistical Power Analysis for the Behavioral Sciences. Routledge; 1988. doi:doi:10.4324/9780203771587.
23. DerSimonian R, Laird N. Meta-analysis in clinical trials. Control Clin Trials. 1986;7(3):177-188.
24. Higgins JPT, Thompson SG, Deeks JJ, Altman DG. Measuring inconsistency in meta-analyses. BMJ. 2003;327(7414):557-560. doi:10.1136/bmj.327.7414.557.
25. Egger M, Davey Smith G, Schneider M, Minder C. Bias in meta-analysis detected by a simple, graphical test. BMJ. 1997;315(7109):629-634.
Search strategy
9
Bibliographic search for PubMed:
Search ((random OR randomly OR randomized)) AND (((((weight OR obesity OR overweight OR BMI OR
"body mass index" OR "abdominal obesity" OR "weight gain")) OR (((("Obesity, Abdominal"[Mesh]) OR
"Obesity"[Mesh]) OR "Body Weight Changes"[Mesh]) OR "Body Mass Index"[Mesh]))) AND
(((((((("Depressive Disorder, Major"[Mesh]) OR "Bipolar Disorder"[Mesh]) OR "Schizophrenia"[Mesh]) OR
"Antipsychotic Agents"[Mesh] "antidepressants" OR mood stabilizers")) AND ((schizophrenia[Text Word] OR
bipolar[Text Word] OR psychosis[Text Word] OR major depression[Text Word] OR schizoaffective[Text
Word] OR psychotic[Text Word] OR antipsychotics[Text Word])))) AND ((((diet[Text Word] OR dietary[Text
Word] OR coaching[Text Word] OR counselling[Text Word] OR health education[Text Word] OR
nutrition[Text Word] OR physical activity[Text Word] OR exercise[Text Word] OR health promotion[Text
Word] OR behavior intervention[Text Word] OR "weight reduction programs"[Text Word]))) OR (((((((("Life
Style"[Mesh]) OR "Risk Reduction Behavior"[Mesh]) OR "Health Promotion"[Mesh]) OR "Health
Education"[Mesh]) OR "Counseling"[Mesh]) OR "Diet"[Mesh]) OR "Exercise"[Mesh]) OR "Weight Reduction
Programs"[Mesh]))))
Table S1: Trial, Population and Treatment CharacteristicsLegend: Trial characteristics. FEP: first episode patients; AP: Antipsychotics; AD: Antidepressants; ANX: Anxiolytica; MS: Mood Stabilisers; OP: Outpatients; IP: Inpatients; WL: Weight loss; Prev.: Prevention;
10
N Age mean
% Male
Diagnoses Medication
Duration weeks
Sessions,
numbers
Baseline BMI
Intervention Control Setting Aim
Alvarez-Jimenez et al. 2006
61 I:26 75.4 FEP AP: 100 % 12 12 24.2 Behavior therapy and psychoeducation.
Advice. Equal frequency of therapist contact
OP Pre
Europe C:27.5 AD: 1.6% Diet and exercise
ANX: 24.5%:
Individual
Attux et al. 2013
134 I:36.2 60 Schizophrenia: 88.1%
AP: 98.75%
12 12 29,1 Psychoeducation and behavioral techniques
Outpatient program. regular psychiatrist
OP Pre
South America
C:38.3 Others: 9.4% Diet and exercise
Individual and group
Bartels et al. 2013
133 43.8 38 Schizophrenia: 18%
52 52 36.8 Fitness club membership. Health coach. MI
Fitness club membership
OP WL
USA Schizoaffective:13%
Diet and exercise
Major depression: 34%
Individual
Bipolar: 35 %
Bartels et al. 2015
210 43.9 49 Scizophrenia: 23%
AP: 100% 52 50 36.8 Fitness club membership. Health coach. MI
Fitness club membership
OP Pre
USA Schizoaffective: 32%
AD: 16.5%
Diet and exercise
Major depression 16%
MS: 48.5%
Individual
Bipolar: 29 %
Battaglia et al. 2013
18 I:36 100 Schizophrenia: 100%
AP: 100 % 12 24 28.5 Football Treatment as usual
OP Pre
Europe C:35 Exercise
Group
Beebe et al. 2005
10 52 80 Schizophrenia: 100%
AP: 100 % 16 48 32.5 Treadmill exercise program
Outpatient program
OP WL
USA AD: 20% Exercise
ANX: 10%
Group
Brar et al. 2005
72 I:40 41 Schizophrenia: 53.5%
AP: 100 % 14 20 101.3 kg Didactic behavioral program
Monthly weighins
OP WL
USA C:40.5 Schizoaffective: 46.5%
AD: 17% Diet and exercise
ANX: 7.1%
Group and individual
Cordes et al. 2014
74 I:38.2 43 Schizophrenia: 100%
AP: 100 % 24 12 27.3 Psychoeducation and behavioral therapy
Monitoring OP Pre
Europe C:35.8 AD: 14% Diet and exercise
MS: 16.6%
Group
ANX: 64.5%
12
Daumit et al. 2013
291 45.3 (11.3)
49.8 Schizophrenia: 29.2%
AP: 89.7% 24 82 36.0 Tailored behavioural weight management.
Information about lifestyle
OP WL
USA Schizoaffective: 28.9%
AD: 60.1%
Diet and exercise
Major depression 12%
MS: 45.5%
Group and individual
Bipolar: 22 %
Evans et al. 2005
51 I:34.6 43.1 Scizophrenia: 31.4
AP: 100 % 12 6 28.8 Educational sessions
Treatment as usual
OP Pre
C:33.6 Schizoaffective: 23.5%
Diet
USA Major depression 9.8%
Individual
Bipolar: 15.7%
Font et al. 2015
332 I:46.3 55 Schizophrenia: 67.2%
AP: 100% 12 24 32.3 Sessions on exercise and nutrition
Treatment as usual
OP WL
Europe C:47.1 Schizoaffective: 17.2%
Diet and exercise
Bipolar: 15.6%
Group and individual
Forsberg et al. 2008
41 I:39.8 58.7 Schizophrenia: 73.2%
AP: 73.2 %
52 104 31.1 Sessions on nutrition and group exercise
Study circle on aesthetic techniques
OP Both
Europe C:42.8 Bipolar: 7.3% Diet and exercise
Group
Gaughran et al. 2018
406 44.1 55 Psychosis: 100%
36 31.2 Cognitive therapyAnd MI
Treatment as usual
OP/IP Both
Europe Diet and exercise
individual
Gillhoff et al. 2010
50 48 54 Bipolar: 100%
20 31 28.4 Behavior therapy Usual care OP WL
Europe Diet and exercise
Group
Goldberg et al. 2013
71 52 81 Schizophrenia: 37%
AP: 100 % 26 22 84.1 kg Behavior therapy Written information on helathhealth subjects
OP WL
USA Major depression: 14%
Diet and exercise
Bipolar: 25 % Group and individual
Green et al. 2015
200 47.2 28 Schizophrenia: 29%
AP: 91 % 24 24 38.3 Behavior therapy Treatment as usual
OP WL
USA Bipolar: 69% AD: 16.5%
Diet and exercise
MS: 48.5%
Group
Holt et al. 2018
412 40 50.9 Schizophrenia: 69.1%Schizoaffective: 16.0%
AP: 100% 52 7 35.7 Cognitive behavior therapy
Treatment as usual
OP WL
Europe First episode: 15.3%
Diet and exercise
Group
Iglesias-Garcia
15 39.9 68.8 Schizophrenia: 100%
AP: 100 % 12 12 32 Structured education
Weekly weighings
OP WL
13
et al. 2010Europe Diet and exercise
Group
Khazaahl et al. 2007
61 40.7 45.9 Schizophrenia: 73.8%
AP: 100 % 12 12 30 Cognitive behavioral therapy
Brief nutritional information
OP WL
Europe Bipolar: 8.3% MS: 23% Diet and exercise
Group
Kilbourne et al. 2012
68 45.3 39 Bipolar: 100%
AP: 11 % 24 9 35.2 Tailored psychotherapy and caremanagement
Quarterly wellness newsletters
OP WL
USA MS: 26% Diet and exercise
Group and individual
Kilbourne et al. 2013
118 52.8 83 Bipolar: 97.5 %
AP: 50 % 52 16 33.2 Tailored psychotherapy and caremanagement
Quarterly wellness newsletters
OP WL
USA Schizoaffective: 2.5%
MS: 77.1%
Diet and exercise
Group and individual
Kilbourne et al. 2016
45 55.3 84.6 Schizophrenia: 7.3%
AP: 38 % 24 9 33.3 Tailored psychotherapy and caremanagement
Quarterly wellness newsletters
OP WL
USA Major depression: 57.1%
AD: 83.5%
Diet and exercise
Bipolar: 24% MS: 52.1 Group and individual
Kwon et al. 2006
48 30.9 31 Schizophrenia: 100%
AP: 100 % 12 10 26.81 Cognitive behavioral therapy
Routine care with verbal recommendations
OP WL
Asia Diet and exercise
Individual
Lee et al. 2014
22 44.1 54.5 AP: 100 % 8 8 33.4 Pedometers and eight weekly phone calls
Written information on physical activity
OP WL
USA Exercise
Individual
Littrell et al. 2003
70 33.7 61.4 Schizophrenia: 77%
16 16 26.3 Psychoeducation Usual care OP Prev
USA 34.5 Schizoaffective: 33%
Diet and exercise
Group
Lovell et al. 2014
89 25.7 60 Schizophrenia: 83%
24 8 32.7 Motivational and behavioral therapy
Usual care OP Prev
Europe Schizoaffective: 2%
Diet and exercise
Group and individual
Marzolini et al. 2009
13 44.6 61 Schizophrenia: 100%
AP: 100 12 24 27.2 Exercise twice weekly
Usual care OP
canada AD: 15 Exercise
MS: 53 Group
Mauri et al. 2008
33 38.9 42.8 Schizoaffective: 10.2 %
12 12 30 Psychoeducation Usual care OP WL
14
Europe Major depression 2%
AP: 100 % Diet
Bipolar: 87.7 %
Individual
McKibbin et al. 2006
57 I:53.1 64.9 Schizophrenia: 84.2%
AP: 100 % 24 24 33.6 Health education classes
Written material on healthy lifestyle
OP WL
USA C:54.8 Schizoaffective: 15.8%
Diet and exercise
Group and individual
Methapatara et al. 2011
64 40.4 64 Schizophrenia: 100%
12 5 28.4 Group education. Pedometer walking
Written material on healthy lifestyle
OP WL
Asia Diet and exercise
Group and individual
Milano et al. 2007
36 45 (N/A)
44 AP: 100 12 27 27.2 Calorie restriction and exercise three times a week
Usual care OP Both
Europe Diet and exercise
Group and individual
Osborn et al. 2018
327 32 47 Schizophrenia:35%
AP: 72% 26 13 51 Behavioral wheel Usual care OP WL
Bipolar:46% Diet and exercise
Psycoses: 19%
Individual
Scheewe et al. 2013
54 I:29.2 73 Schizophrenia: 76%
AP: 100 % 24 48 26.6 Strict protocol physical exercise twice weekly
Occupational therapy
IP/OP WL
Europe 30.1 Schizoaffective: 24%
Exercise
Group
Scocco et al. 2006
18 46.4 55 Schizophrenia: 75%
AP: 100 % 8 29 Weight control education
Usual care OP Both
Europe Schizoaffective: 25%
Diet
Individual
Skrinar et al. 2005
20 I:39.7 33 Schizophrenia: 100%
AP: 100 12 48 32.9 Four weekly training sessions.
Recordkeeping on physical activity
IP/OP Both
USA C:36.3 Diet and exercise
Group and individual
Speyer et al. 2016
280 38.6 44 Schizophrenia: 90%
52 42 34.2 Tailored healthcoach. MI
Care coordination
OP WL
Europe Schizoaffective: 10%
Diet and exercise
Group and individual
Ratliff et al. 2012
30 50.1 35 Schizophrenia 60%
8 8 41.6 Financial reimbursment
Wait list OP WL
USA Diet and exercise
Group
Usher et 101 54.5 Schizophrenia AP: 100 % 12 12 33.3 Weekly health Usual care OP WL
15
al. 2013 : 84.2% educationAustralia
Major depression: 6.9%
Diet and exercise
Bipolar: 6.9% Group
Weber et al. 2006
157 29 Schizophrenia: 100%
AP: 100 % 16 16 33 Cognitive behavioral therapy
Usual care OP WL
USA Diet and exercise
Group
Wu et al. 2007
53 I:42.2 41.51 Schizophrenia: 100%
AP: 100 % 12 11 30.3 Psychoeducation Monthly treadmill test and weighing
IP WL
Asia C:39 Diet and exercise
Group and individual
Wu et al. 2008
64 26.3 64 Schizophrenia: 100%
AP: 100 % 24 72 24.5 Dietary and training program
Usual care IP WL
Asia Diet and exercise
Individual
16
Figure S1: Flow chart
17
Records identified through database searching
(n =1774 )
Scr
ee nin
g Scr
Elig
ibili
ty Inc
lud
Ide
nti fica
tio n
Additional records identified through other sources
(n = 6)
Records after duplicates removed(n = 1441)
Records screened(n =1441)
Records excluded(n = 1228)
Full-text articles assessed for eligibility
(n =213)
Additional search (n =3)
Studies included in quantitative synthesis
(meta-analysis)(n = 41)
Full-text articles excluded:Not randomized (n =98)
Not reporting body weight (n = 40)
Not lifestyle (n = 13)
Not SMI (n=11)
Secondary publications (n = 11)
Other (n = 2)
Table S2: Risk of bias assessments.
Study name Year
Random sequence
generationAllocation
concealment
Blinding of participants and
personnel
Blinding of outcome
assessmentIncomplete
outcome dataSelective reporting
low
high
unclear
low
high
unclear
low
high
unclear
low
high
unclear
low
high
unclear
low
high
unclear
Beebe 2005 X X X X X X Wu 2007 X X X X X XSkrinar 2005 X X X X X XMethapatara 2011 X X X X X X Bartels 2015 X X X X X x X Bartels 2013 X X X X X X Green 2015 X X X X X XLittrell 2003 X X X X X X X Khazaal 2007 X X X X X X Evans 2005 X X X X X XWu 2008 X X X X X X Alvarez-jiminez 2006 X X X X X Attux 2013 X X X X X X Battaglia 2013 X X X X x X X Cordes 2014 X X X X X X Daumit 2013 X X X X X XForsberg 2008 X X X X X XGillhoff 2010 X X X X X XIglesias-garcia 2010 X X X X X XKilbourne 2012 X X X X X X Kwon 2006 X X X X X X Lee 2014 X X X X X XLovell 2014 X X X X X XMarzolini 2009 X X X X X X XMauri 2008 X X X X X X XMckibbin 2006 X X X X X XScheewe 2013 X X X X X XScocco 2006 X X X X X XUscher 2013 X X X X X XWeber 2006 X X X X X XMilano 2007 X X X X x X XSpeyer 2016 X X X X X XKilbourne 2012 X X X X X XKilbourne 2016 X X X X X XGoldberg 2013 X X X X X XFont 2015 X X X X X XBrar 2005 X X X X X XGaughran 2018 X X X X x X
18
Table S3: ASPECT-R scores Studies Total score
(0-24)Eligibility(0-6)
Flexibility(0-6)
Expertise(0-6)
Duration(0-6)
Battaglia 2013 6 0 0 3 3Wu 2008 6 0 1 3 2Brar 2005 8 1 1 4 2Iglesias-Garcia 2010 10 1 3 3 3Wu 2007 7 1 1 3 2Attux 2013 12 2 2 4 4Cordes 2014 10 2 2 2 4Evans 2005 17 2 5 5 5Kwon 2006 12 2 3 4 3Methapatara 2011 7 2 2 1 2Milano 2007 9 2 2 3 2Scheewe 2013 7 2 1 2 2Bartels 2013 14 3 4 4 3Bartels 2015 14 3 4 4 3Daumit 2013 11 3 2 3 3Font 2015 13 3 4 4 2Littrell 2003 12 3 3 3 3Mauri 2008 11 3 2 3 3Scocco 2006 16 3 4 4 5Alvarez-Jimenez 2006 14 4 5 2 3Beebe 2005 11 4 1 4 2Forsberg 2008 18 4 5 5 4Goldberg 2013 15 4 4 4 3Khazaahl 2007 15 4 5 3 3Lee 2014 16 4 3 4 5Lovell 2014 18 4 6 4 4Usher 2013 14 4 3 3 4Weber 2006 14 4 3 4 3Gillhoff 2010 20 5 5 5 5Green 2015 17 5 5 3 4Kilbourne 2012 15 5 3 4 3Kilbourne 2013 16 5 4 4 3Kilbourne 2016 16 5 4 4 3Marzolini 2009 17 5 4 5 3Mckibbin 2006 15 5 3 4 3Skrinar 2005 11 5 2 3 1Speyer 2016 20 6 6 4 4Gaughran 2017 19 4 6 5 4Holt 2018 18 4 5 5 4Osborn 2018 29 5 6 4 5
20
Table S4: Grades of Recommendation, Assessment, Development and Evaluation (GRADE): Summary of findings Lifestyle intervention compared to treatment as usual for weight management in people with severe mental illnessOutcomes Anticipated absolute effects* (95% CI) Relative effect
(95% CI) № of participants (studies)
Quality of the evidence(GRADE)
Comments
Risk with treatment as usual
Risk with Lifestyle intervention
BMI post-intervention The mean BMI post-intervention was 0 kg/m2
The mean BMI post-intervention in the intervention group was 0,57 kg/m2 lower (-0,98 lower to -0,17 lower)
- 4237(40 RCTs)
⨁◯◯◯VERY LOW a,b,c
BMI long term The mean BMI long term was 0 kg/m2
The mean BMI long term in the intervention group was 0,63 kg/m2 lower (-1.3 lower to 0.04 higher)
- 1412(17 RCTs)
⨁◯◯◯VERY LOW a,d
Clinically significant weight loss 21.708 per 100.000
27819 per 100.000(17.782 to 40.746)
OR 1.5(1.07 to 2.13)
1121(9 RCTs)
⨁◯◯VERY LOW a
Quality of Life The mean quality of Life was 0 SMD
The mean quality of Life in the intervention group was 0.03 SMD higher (-0.11 lower to 0.17 higher)
- 1309(15 RCTs)
⨁◯◯◯VERY LOW a
Waist circumference The mean waist circumference was 0 cm
The mean waist circumference in the intervention group was 2.13 cm lower (-2.98 lower to -1,28 lower)
- 2128(24 RCTs)
⨁◯◯◯VERY LOW a,b
Fasting glucose The mean fasting glucose was 0 mg/dl
The mean fasting glucose in the intervention group was 0.04 mg/dl lower (-0.20 lower to 0.28 higher)
- 1056(12 RCTs)
⨁◯◯◯VERY LOW a,b
Cholesterol The mean cholesterol was 0 The mean cholesterol in the intervention group was 0,06 lower (-0.14 lower to -0.01 higher)
- 1658(15 RCTs)
⨁◯◯◯VERY LOW a,b
Systolic bloodpressure The mean systolic blood pressure was 0 mmHg
The mean systolic blood pressure in the intervention group was -0.58 mmHg lower (-1.70 lower to 0.54 higher)
- 1733(18 RCTs)
⨁◯◯◯VERY LOW a,b
*The risk in the intervention group (and its 95% confidence interval) is based on the assumed risk in the comparison group and the relative effect of the intervention (and its 95% CI). CI: Confidence interval; OR: Odds ratio GRADE Working Group grades of evidence
21
High quality: We are very confident that the true effect lies close to that of the estimate of the effectModerate quality: We are moderately confident in the effect estimate: The true effect is likely to be close to the estimate of the effect, but there is a possibility that it is substantially differentLow quality: Our confidence in the effect estimate is limited: The true effect may be substantially different from the estimate of the effectVery low quality: We have very little confidence in the effect estimate: The true effect is likely to be substantially different from the estimate of effect
22
Trial Sequential AnalysisFigure S2: Trial sequential analysis of BMI:
Trial sequential analysis for weight, DARIS = diversity-adjusted required information size
23
Funnel plotsFigure S5: Funnel plot of weight (BMI)
-5 -4 -3 -2 -1 0 1 2 3 4 5
0
1
2
3
4
5
Stan
dard
Err
or
Difference in means
Funnel Plot of Standard Error by Difference in means
26
Figure S6: Funnel plot of weight (kg)
-20 -10 0 10 20
0
2
4
6
8
10
Stan
dard
Err
or
Difference in means
Funnel Plot of Standard Error by Difference in means
27
Figure S7: Funnel plot of weight (BMI) (maintenance):
-4 -3 -2 -1 0 1 2 3 4
0
1
2
3
Stan
dard
Err
or
Difference in means
Funnel Plot of Standard Error by Difference in means
28
Figure S8: Funnel plot of weight quality of life
-2,0 -1,5 -1,0 -0,5 0,0 0,5 1,0 1,5 2,0
0,0
0,2
0,4
0,6
0,8
Stan
dard
Err
or
Std diff in means
Funnel Plot of Standard Error by Std diff in means
29
Figure S9: Funnel plot of weight cholesterol
-2,0 -1,5 -1,0 -0,5 0,0 0,5 1,0 1,5 2,0
0,0
0,1
0,2
0,3
0,4
0,5
0,6
Stan
dard
Err
or
Difference in means
Funnel Plot of Standard Error by Difference in means
30
Figure S10: Funnel plot of systolic blood pressure
-8 -7 -6 -5 -4 -3 -2 -1 0 1 2 3 4 5 6 7 8
0
2
4
6
8
10
Stan
dard
Err
or
Difference in means
Funnel Plot of Standard Error by Difference in means
31
Figure S11: Funnel plot of waist circumference
-10 -5 0 5 10
0
1
2
3
4
5
6
7
8
Stan
dard
Err
or
Difference in means
Funnel Plot of Standard Error by Difference in means
32
Figure S12: Funnel plot for all-cause discontinuation
-2,0 -1,5 -1,0 -0,5 0,0 0,5 1,0 1,5 2,0
0,0
0,5
1,0
1,5
2,0
Stan
dard
Err
or
Log risk ratio
Funnel Plot of Standard Error by Log risk ratio
33
Forest plotsFigure S13: Forest plot of weight (kg)
Study name Statistics for each study Difference in means and 95% CI
Difference Standard Lower Upper in means error Variance limit limit Z-Value p-Value
Alvarez-Jimenez 2006 -2,880 1,098 1,206 -5,032 -0,728 -2,622 0,009Attux 2013 -4,000 2,779 7,724 -9,447 1,447 -1,439 0,150Bartels 2013 0,300 4,868 23,697 -9,241 9,841 0,062 0,951Bartels 2015 -6,000 3,493 12,200 -12,846 0,846 -1,718 0,086Battaglia 2013 -4,200 5,545 30,753 -15,069 6,669 -0,757 0,449Brar 2005 2,400 5,728 32,810 -8,827 13,627 0,419 0,675Cordes 2014 2,900 3,500 12,250 -3,960 9,760 0,829 0,407Daumit 2013 -1,500 0,539 0,291 -2,557 -0,443 -2,782 0,005Evans 2005 -4,000 0,940 0,883 -5,842 -2,158 -4,256 0,000Forsberg 2008 6,900 9,095 82,725 -10,927 24,727 0,759 0,448Gillhoff 2010 -1,800 3,954 15,636 -9,550 5,950 -0,455 0,649Goldberg 2013 5,700 4,585 21,023 -3,287 14,687 1,243 0,214Green 2015 -1,520 3,868 14,959 -9,101 6,061 -0,393 0,694Iglesias-Garcia 2010 -3,000 9,175 84,179 -20,982 14,982 -0,327 0,744Khazaahl 2007 4,500 4,116 16,942 -3,567 12,567 1,093 0,274Kwon 2006 -2,460 1,035 1,072 -4,489 -0,431 -2,376 0,017Littrell 2003 -3,800 4,176 17,441 -11,985 4,385 -0,910 0,363Lovell 2014 1,400 1,465 2,145 -1,470 4,270 0,956 0,339Marzolini 2009 1,300 6,344 40,243 -11,134 13,734 0,205 0,838Mauri 2008 -3,800 0,968 0,937 -5,697 -1,903 -3,926 0,000Mckibbin 2006 -0,800 5,069 25,695 -10,735 9,135 -0,158 0,875Methapatara 2011 -2,210 0,960 0,921 -4,091 -0,329 -2,303 0,021Milano 2007 4,100 5,168 26,705 -6,028 14,228 0,793 0,428Ratcliff 2015 -1,910 1,054 1,112 -3,976 0,156 -1,812 0,070Scocco 2006 -1,970 1,532 2,346 -4,972 1,032 -1,286 0,198Skrinar 2005 0,000 2,659 7,073 -5,212 5,212 0,000 1,000Speyer 2016 -0,530 2,692 7,244 -5,805 4,745 -0,197 0,844Usher 2013 -1,100 3,784 14,318 -8,516 6,316 -0,291 0,771Weber 2006 -6,290 3,584 12,848 -13,315 0,735 -1,755 0,079Wu 2007 -5,200 1,090 1,188 -7,336 -3,064 -4,771 0,000Wu 2008 -3,800 0,702 0,493 -5,175 -2,425 -5,415 0,000Holt 2018 2,800 2,210 4,885 -1,532 7,132 1,267 0,205
-2,215 0,406 0,165 -3,011 -1,419 -5,455 0,000-8,00 -4,00 0,00 4,00 8,00
Favours intervention Favours control
Weight (kg)
Meta Analysis
34
Figure S14: Forest plot of quality of life
Study name Statistics for each study Std diff in means and 95% CI
Std diff Standard Lower Upper in means error limit limit p-Value
Gaughran 2017 -0,231 0,100 -0,427 -0,036 0,020Holt 2018 0,135 0,099 -0,059 0,328 0,172Alvarez-Jimenez 2006 0,109 0,179 -0,241 0,459 0,541Brar 200 -0,756 0,246 -1,238 -0,274 0,002Evans 2005 -0,749 0,378 -1,489 -0,009 0,047Font 2015 0,113 0,110 -0,102 0,329 0,302Goldberg 2010 0,354 0,193 -0,024 0,733 0,067Kilbourne 2012 -0,131 0,243 -0,607 0,345 0,591Kilbourne 2013 0,104 0,184 -0,257 0,465 0,574Kilbourne2016 0,023 0,128 -0,227 0,273 0,857Lovell 2014 0,771 0,224 0,332 1,211 0,001Marzolini 2009 -0,630 0,661 -1,925 0,665 0,340Speyer 2016 0,059 0,130 -0,195 0,313 0,650Usher 2014 0,050 0,199 -0,340 0,440 0,800Osborn 2018 0,134 0,111 -0,084 0,351 0,229
0,032 0,071 -0,108 0,171 0,657-1,00 -0,50 0,00 0,50 1,00
Fav ours controls Fav ours interv entios
Quality of life
Meta Analysis
35
Figure S15: Forest plot of systolic blood pressure
Study name Statistics for each study Difference in means and 95% CIDifference Standard Lower Upper in means error Variance limit limit Z-Value p-Value
Attux 2013 -1,500 2,195 4,818 -5,802 2,802 -0,683 0,494Bartels 2015 0,300 2,451 6,009 -4,504 5,104 0,122 0,903Brar 2005 -2,800 2,870 8,234 -8,424 2,824 -0,976 0,329Daumit 2013 -0,400 1,641 2,693 -3,617 2,817 -0,244 0,807Font 2015 -0,410 1,948 3,794 -4,227 3,407 -0,211 0,833Forsberg 2008 -1,800 8,307 69,013 -18,082 14,482 -0,217 0,828Gillhoff 2010 -1,700 4,115 16,932 -9,765 6,365 -0,413 0,680Killbourne 2012 2,100 3,520 12,390 -4,799 8,999 0,597 0,551Kilbourne 2013 -6,500 3,374 11,381 -13,112 0,112 -1,927 0,054Kilbourne 2016 0,810 2,127 4,523 -3,359 4,979 0,381 0,703Lee 2014 -3,750 5,806 33,708 -15,129 7,629 -0,646 0,518Marzolini 2009 -1,000 3,421 11,706 -7,706 5,706 -0,292 0,770Mckibben 2006 -7,300 5,086 25,868 -17,268 2,668 -1,435 0,151Scheewe 2013 -2,200 3,944 15,557 -9,930 5,530 -0,558 0,577Speyer 2016 1,100 1,655 2,741 -2,145 4,345 0,664 0,506Gaughran 2017 -1,900 2,212 4,893 -6,236 2,436 -0,859 0,390Holt 2018 3,000 1,683 2,834 -0,299 6,299 1,782 0,075Osborn 2018 -3,000 1,715 2,940 -6,361 0,361 -1,750 0,080
-0,578 0,570 0,325 -1,696 0,540 -1,013 0,311-1,00 -0,50 0,00 0,50 1,00
Fav ours interv entions Fav ours controls
Systolic blood pressure
Meta Analysis
36
Figure S16: Forest plot of total cholesterol
Study name Statistics for each study Difference in means and 95% CI
Difference Standard Lower Upper in means error Variance limit limit Z-Value p-Value
Attux 2013 -0,100 0,187 0,035 -0,467 0,267 -0,535 0,593Bartels 2015 -0,100 0,152 0,023 -0,398 0,198 -0,659 0,510Daumit 2013 -0,050 0,089 0,008 -0,224 0,124 -0,564 0,573Gillhoff 2010 -0,200 0,289 0,084 -0,766 0,366 -0,692 0,489Kilbourne 2012 0,000 0,147 0,022 -0,288 0,288 0,000 1,000Kilbourne 2013 -0,200 0,203 0,041 -0,597 0,197 -0,987 0,323Lee 2014 0,400 0,502 0,253 -0,585 1,385 0,796 0,426Font 2015 0,000 0,104 0,011 -0,204 0,204 0,000 1,000Mauri 2008 -0,300 0,383 0,147 -1,050 0,450 -0,784 0,433McKibbin 2006 -0,700 0,252 0,064 -1,194 -0,206 -2,775 0,006Skrinar 2005 -0,200 0,457 0,209 -1,095 0,695 -0,438 0,661Speyer 2016 0,000 0,142 0,020 -0,278 0,278 0,000 1,000Wu 2007 -0,200 0,248 0,061 -0,685 0,285 -0,808 0,419Holt 2018 -0,200 0,125 0,016 -0,445 0,045 -1,603 0,109Gaughran 2017 0,060 0,167 0,028 -0,267 0,387 0,360 0,719Osborn 2018 0,100 0,117 0,014 -0,129 0,329 0,857 0,392
-0,062 0,039 0,002 -0,139 0,015 -1,569 0,117
-1,00 -0,50 0,00 0,50 1,00Fav ours interv ention Fav ours controls
Cholesterol
Meta Analysis
37
Figure S17: Glucose
Study name Statistics for each study Difference in means and 95% CI
Difference Standard Lower Upper in means error Variance limit limit Z-Value p-Value
Wu 2008 -0,300 0,088 0,008 -0,473 -0,127 -3,394 0,001Attux 2013 -0,200 0,241 0,058 -0,672 0,272 -0,831 0,406Font 2015 0,000 0,199 0,040 -0,390 0,390 0,000 1,000Forsberg 2008 0,100 0,241 0,058 -0,373 0,573 0,414 0,679Gillhoff 2010 0,200 0,156 0,024 -0,105 0,505 1,284 0,199Lee 2014 -1,200 1,512 2,285 -4,163 1,763 -0,794 0,427Mauri 2008 0,200 0,210 0,044 -0,211 0,611 0,953 0,340McKibbin 2006 -0,900 1,022 1,045 -2,903 1,103 -0,881 0,379Scheewe 2013 0,000 0,179 0,032 -0,351 0,351 0,000 1,000Daumit 2013 -0,420 0,249 0,062 -0,909 0,069 -1,685 0,092Holt 2018 0,400 0,315 0,099 -0,217 1,017 1,271 0,204Osborne 2018 2,800 0,695 0,483 1,438 4,162 4,029 0,000
0,037 0,122 0,015 -0,202 0,277 0,306 0,760
-1,00 -0,50 0,00 0,50 1,00Fav ours interv ention Fav ours controls
Glucose
Meta Analysis
38
Figure S18: Waist circumference
Study name Statistics for each study Difference in means and 95% CIDifference Standard Lower Upper in means error Variance limit limit Z-Value p-Value
Wu 2007 -2,300 1,160 1,347 -4,574 -0,026 -1,982 0,047Methapatara 2011 -4,240 1,200 1,441 -6,593 -1,887 -3,532 0,000Bartels -4,600 2,499 6,244 -9,498 0,298 -1,841 0,066Evans 2005 -5,500 1,491 2,223 -8,422 -2,578 -3,689 0,000Wu 2008 -1,800 0,350 0,123 -2,486 -1,114 -5,143 0,000Attux 2013 -3,900 2,269 5,148 -8,347 0,547 -1,719 0,086Cordes 2014 2,100 2,964 8,787 -3,710 7,910 0,708 0,479Daumit 2013 -2,000 0,836 0,698 -3,638 -0,362 -2,394 0,017Font 2015 -0,080 1,537 2,362 -3,092 2,932 -0,052 0,958Gillhoff 2010 -4,500 3,559 12,669 -11,476 2,476 -1,264 0,206Iglesias-garcia 2010 -3,900 7,252 52,584 -18,113 10,313 -0,538 0,591Kilbourne 2013 -3,300 2,742 7,517 -8,674 2,074 -1,204 0,229Lee 2014 -1,800 7,830 61,313 -17,147 13,547 -0,230 0,818Lovell 2014 0,300 1,753 3,073 -3,136 3,736 0,171 0,864Marzolini 2009 0,900 5,523 30,499 -9,924 11,724 0,163 0,871McKibbin 2006 -3,600 3,847 14,799 -11,140 3,940 -0,936 0,349Scheewe 2013 -3,800 4,123 17,002 -11,882 4,282 -0,922 0,357Usher 2013 1,300 2,585 6,683 -3,767 6,367 0,503 0,615Speyer 2016 -1,900 1,978 3,913 -5,777 1,977 -0,961 0,337Kilbourne 2012 -9,400 4,427 19,596 -18,076 -0,724 -2,123 0,034Kilbourne 2016 3,400 1,952 3,811 -0,426 7,226 1,742 0,082Holt 2018 -2,400 1,667 2,778 -5,667 0,867 -1,440 0,150Gaughran 2017 -3,200 2,019 4,076 -7,157 0,757 -1,585 0,113Osborn 2018 -2,000 1,715 2,940 -5,361 1,361 -1,166 0,243
-2,123 0,411 0,169 -2,928 -1,318 -5,168 0,000-1,00 -0,50 0,00 0,50 1,00
Fav ours interv ention Fav ours controls
Waist circumference
Meta Analysis
39
Figure S19: Forest plot of all-cause discontinuation:
Study name Statistics for each study Risk ratio and 95% CI
Risk Lower Upper ratio limit limit Z-Value p-Value
Attux 2013 0,975 0,483 1,971 -0,070 0,944Bartels 2013 0,985 0,525 1,849 -0,047 0,963Bartels 2015 0,722 0,413 1,263 -1,142 0,254Battaglia 2013 0,611 0,124 3,000 -0,607 0,544Beebe 2015 5,000 0,289 86,430 1,107 0,268Brar 2005 1,451 0,701 3,005 1,002 0,316Cordes 2014 1,100 0,794 1,522 0,571 0,568Daumit 2013 1,429 0,464 4,399 0,622 0,534Evans 2005 0,414 0,181 0,945 -2,094 0,036Forsberg 2008 0,708 0,242 2,071 -0,630 0,529Gaughran 2017 1,812 1,047 3,137 2,124 0,034Gillhoff 2010 0,923 0,061 13,953 -0,058 0,954Goldberg 2013 1,620 0,953 2,755 1,781 0,075Green 2015 1,207 0,620 2,351 0,553 0,580Holt 2018 1,238 0,811 1,889 0,989 0,322Iglecias-Garcia 20102,667 0,126 56,626 0,629 0,529Khazahl 2007 2,903 0,635 13,268 1,375 0,169Kilbourne 2012 2,000 0,544 7,352 1,044 0,297Kilbourne 2013 1,034 0,669 1,599 0,152 0,879Kilbourne 2016 0,886 0,524 1,498 -0,452 0,652Kwon 2006 5,000 0,709 35,285 1,614 0,106Lee 2014 1,667 0,381 7,288 0,679 0,497Lovell 2014 0,675 0,229 1,990 -0,713 0,476masa-font 1,447 0,830 2,523 1,301 0,193Mauri 2008 0,857 0,362 2,028 -0,351 0,726McKibbin 2006 1,333 0,324 5,486 0,399 0,690Ratliff 2014 1,000 0,072 13,868 0,000 1,000Scheewe 2013 0,873 0,464 1,645 -0,419 0,675Scocco 2006 0,200 0,011 3,705 -1,081 0,280Skrinar 2005 1,500 0,528 4,258 0,762 0,446Speyer 2016 0,919 0,512 1,651 -0,282 0,778Weber 2006 3,636 0,201 65,861 0,874 0,382Wu 2008 0,969 0,211 4,439 -0,041 0,967Wu 2007 0,143 0,008 2,644 -1,307 0,191Osborn 2018 1,051 0,573 1,929 0,161 0,872
1,103 0,973 1,249 1,533 0,1250,01 0,1 1 10 100
Fav ours interv ention Fav ours controls
Relative risk of discontinuation
Meta Analysis
Figure S20: Forest plot of deaths:
40
Statistics for each study Risk ratio and 95% CI
Risk Lower Upper ratio limit limit Z-Value p-Value
Speyer 2016 0,343 0,036 3,258 -0,932 0,351Green 2015 0,954 0,061 15,005 -0,033 0,973Daumit 2013 0,685 0,116 4,041 -0,417 0,677Holt 2018 6,933 0,360 133,375 1,283 0,199Osborn 2018 0,370 0,039 3,519 -0,865 0,387
0,722 0,260 2,006 -0,624 0,532
0,01 0,1 1 10 100Fav ours interv ention Fav ours controls
Relative risk of death
Meta Analysis
Figure S21: Forest plot of somatic hospitalisations:
41
Study name Statistics for each study Risk ratio and 95% CI
Risk Lower Upper ratio limit limit Z-Value p-Value
Speyer 2016 0,761 0,425 1,360 -0,923 0,356Green 2015 0,298 0,114 0,777 -2,476 0,013Daumit 2013 1,273 0,752 2,155 0,898 0,369Holt 2018 1,170 0,537 2,552 0,396 0,692Osborn 2018 0,476 0,125 1,807 -1,091 0,275
0,785 0,479 1,286 -0,962 0,336
0,01 0,1 1 10 100Fav ours interv entions Fav ours controls
Relative risk for somatic hospitalisation
Meta Analysis
Figure S22: Forest plot of psychiatric hospitalisations:
42
Study name Statistics for each study Risk ratio and 95% CI
Risk Lower Upper ratio limit limit Z-Value p-Value
Speyer 2016 0,787 0,500 1,239 -1,035 0,300Green 2015 0,954 0,470 1,936 -0,130 0,896Daumit 2013 0,720 0,433 1,196 -1,270 0,204Holt 2018 1,340 0,738 2,433 0,961 0,336Osborn 2018 0,303 0,086 1,065 -1,862 0,063
0,838 0,607 1,156 -1,079 0,281
0,01 0,1 1 10 100Fav ours controls Fav ours interv entios
Relative risk for psychiatric hospitalisations
Meta Analysis
43