The use of eHealth to promote physical activity in people

Achieving adequate amounts of physical activity (PA) Background: confers important physical and mental health benefits. Despite this, people with mental health conditions often do not meet recommended levels of PA. eHealth, the delivery of health information through internet and mobile technologies, is an emerging concept in healthcare which presents opportunities to improve PA. The aim of this systematic review is to describe the use of eHealth to increase or monitor PA levels in people with mental health conditions. Databases searched included OVID Medline, EMBASE, Methods: PsychInfo and Web of Science using a combination of key-words and medical subject headings. Articles were included if they described an eHealth technology designed to improve or monitor PA in people with mental health conditions. Two reviewers screened articles. Articles included in the qualitative synthesis were screened for risk of bias using the Cochrane Risk of Bias Tool for experimental studies and Downs and Black Checklist for non-experimental studies. Seven studies met the eligibility criteria. A variety of eHealth Results: platforms designed to promote or monitor PA were described in these studies; web-based (n=4), web and mobile application (n=3) and e-mail-based (n=1), one study used both a web-based and mobile application. Three studies reported eHealth interventions significantly increased PA levels, however it is unclear if eHealth interventions are superior at promoting PA compared to conventional interventions. Four studies reported that higher levels of PA, measured using eHealth, were associated with better mental health profiles. eHealth interventions may be an innovative low-cost method Conclusion: to increase PA levels which may have knock-on effects on mental health outcomes. Although some of the included studies in this review 1 1 1 2


Introduction
Physical activity (PA) is associated with a number of healthrelated benefits such as improved cardiovascular health, bone strength, and a reduced risk of developing chronic conditions such as colorectal and breast cancers, cardiovascular disease, and Type II diabetes 1,2 . In addition, the benefits of PA among people with mental health conditions extend beyond physical health benefits and include improved mood and sleep, reduced stress, and enhanced self-esteem 3,4 . Despite the numerous physical and mental benefits of PA, insufficient levels are prevalent among people with mental health conditions 5,6 . The low levels of PA among this population and potential mental and physical health gains make a strong case to explore innovative and effective ways to improve PA levels.
eHealth is a relatively new concept in healthcare which may present unique opportunities to improve PA levels. eHealth is an umbrella term including 'the transfer of health resources and health care by electronic means, including, but not limited to the delivery of health information through the internet and mobile technologies' 7,8 . The implementation of internet technology in health-care provides a number of benefits such as convenience for users, easy storage of large amounts of information, ease of updating information, and ability to provide personalized feedback 9 . eHealth interventions have been extensively studied in a number of populations ranging from cancer survivors to community dwelling adults [10][11][12][13][14] . Systematic review evidence has consistently supported the effectiveness of eHealth interventions to increase PA levels.
eHealth based interventions may be well suited to improve PA levels among people with mental health conditions. Internetbased interventions have previously addressed several barriers common to traditional PA based interventions including overcoming geographical restrictions and combating a lack of human resources 15 . These advantageous features are some of the reasons the National Institute for Clinical Excellence (NICE) has identified computerised cognitive behavioural therapy as part of an approach to improving standard care of people with depression 16 . In addition, these features may be applicable to help promote PA in people with mental health conditions. Furthermore, people with mental health conditions are reported willing to use eHealth for health-related reasons. A study of 100 people with mental health conditions at a psychiatric outpatient facility reported that 72% of people owned a smartphone and 67% were eager to use a smartphone application to track their condition 17 . Therefore, eHealth interventions may potentially be a useful platform to monitor and increase PA levels in people with mental health conditions.
To our knowledge, no systematic review has synthesised the literature in the field of eHealth and PA for people with mental health conditions. To address this gap, the aim of this systematic review was to describe the use of eHealth to increase or monitor PA levels in people with mental health conditions. Secondary objectives of this review included (i) To investigate the effectiveness of eHealth interventions as a stand-alone or multimodal intervention to promote PA in people with a mental health condition (ii) To explore the extent to which eHealth technologies are used to measure PA among people with mental health conditions (iii) To report associations between PA measured using eHealth devices and mental health outcomes.

Study design
This systematic review was conducted to identify eHealth technologies with a primary or secondary aim to promote or monitor PA in people with mental health conditions. The "Preferred Reporting Items for Systematic Reviews and Meta-Analyses (PRISMA)" 18 and the criteria outlined in "A Measurement Tool to Assess Systematic Reviews (AMSTAR) checklist" 19 guidelines were followed in drafting this review (PRISMA checklist in Supplementary File 1). The protocol outlining the planned search strategy and method of analysis for this review was registered online and is available on PROSPERO, a registry of systematic reviews (CRD42017068834). It was originally planned to include intervention based studies only, however due to the relative paucity of available trials due to the emerging nature of this research field, a pragmatic decision was taken to broaden the objectives of the review. Therefore, studies that used eHealth technology to monitor PA among people with mental health conditions were also included.

Eligibility criteria
Experimental studies and observational studies, with or without controls, were eligible for inclusion if they evaluated an eHealth-based technology to promote or monitor PA, (internet and mobile technologies) delivered to participants with mental health conditions which included PA as a primary or secondary outcome measure. As per Ritterband et al. (2006) 2 we included eHealth research into the use of web-based and mobile health technologies to measure, track or encourage increases in PA levels among people with mental health conditions. Mental health conditions were characterized as some combination of abnormal thoughts, emotions and relationships with others 20 , which included but was not limited to; depression, bipolar disorder, anxiety disorders and schizophrenia, spectrum disorders. Single eHealth interventions or multi-modal interventions in conjunction with eHealth were included. Studies were excluded if only telephone calls, short message service (SMS) or conference calls were used. Authors of relevant abstracts or conference presentations were contacted to obtain a full-text article or detailed methodology and data set. Abstracts and conference presentations without an accompanying full-text article were excluded due to lack of a detailed methodology

Amendments from Version 2
In this version of the article, an additional column has been added to Table 1. This column presents to the reader if interventions consisted of a standalone eHealth based physical activity intervention, or if the eHealth physical activity intervention was a component of a multimodal intervention. In addition, some small grammar changes have been made.

REVISED
and potential for high risk of bias. Review articles, case studies and letters to the editor were also excluded.
PA is a complex multi-dimensional construct measured through objective (e.g. indirect calorimetry, accelerometers, pedometers) or self-report methods (e.g. questionnaire, log) 21 . Domains of PA can be considered on a continuum from light activity (e.g. slow walking) through to moderate level activity (e.g. brisk walking) and vigorous activity (e.g. jogging). Sedentary behaviour consists of low levels of activity, similar to resting (e.g. sitting or lying down) 22 . There are many different ways of quantifying PA. We included the following methods of quantifying PA, but not limited to the following; MET-minutes.week -1 , minutes in light, moderate and/or vigorous PA per week, and meeting/not meeting PA guidelines (150 minutes per week of moderate/vigorous activity) 23 . All methods of measuring PA were included e.g. self-report, objective or direct measures.

Data sources & search strategy
An experienced medical librarian was consulted and a comprehensive search strategy was developed with all keywords and subject headings included (DM). The search strategy consisted of a search of four electronic databases: OVID Medline, EMBASE, PsychInfo, and Web of Science. Search terms included keywords and medical subject headings adapted for each database. These related to three categories: 1) the condition (e.g. 'mental health' 'depression', 'bipolar disorder', 'schizophrenia' and 'anxiety disorder'), 2) technology (e.g.'teleHealth', 'telerehabilitation', 'mobile health', 'Mhealth', 'eHealth', 'e-health', 'mobile technology', 'smartphone'), and 3) PA (e.g. 'exercise', 'physical activity', 'exercise therapy', 'physiotherapy'). There was no limit placed on the year published as it was believed that the search strategy would produce only articles published within the last ten years, due to the relatively novel nature of this technology. Databases were searched until August 2017. The bibliographies of all included studies were examined to identify further studies. The search strategy is available in Supplementary File 2.
Selection of eligible studies Two researchers (JM and GK), independently screened titles and abstracts to identify studies that met the eligibility criteria. Any disagreements between researchers were discussed and if a consensus could not be reached a third researcher (JB) intervened. All full-texts were retrieved and examined in detail to assess for inclusion in this review.

Risk of bias and classification of intervention type within studies
Two researchers (JM and GK) independently appraised the risk of bias of included studies; any disagreements were resolved through discussion. The Downs and Black checklist was used to assess the risk of bias of all included observational studies 24 . This checklist contains 27 items, with a maximum possible score of 32 points. The final score is variable as some items of the checklist may not be applicable and can be excluded. In addition, the Cochrane Collaboration's tool 25 was used to assess risk of bias for each RCT. Risk of bias was assessed in the following six areas; sequence generation (randomisation); allocation concealment; blinding of participants, personnel and investigator; incomplete data (e.g. losses to follow-up, intention-to treat analysis); selective outcome reporting; and other possible sources of bias.

Data extraction & analysis
Data were extracted by two researchers (JM and GK) independently onto standardised data extraction forms. Any disagreements were discussed, if a consensus could not be reached, a third member of the research team (JB) arbitrated. The standardised data extraction form was piloted on two randomly selected studies and modified accordingly. Data were extracted using the following headings: methods, allocation, blinding, duration, design, setting, participants, diagnosis, age, sex, inclusion criteria, exclusion criteria, intervention, control group, primary outcomes, secondary outcomes, results in PA outcomes, results in secondary outcomes.

Study selection & design
The PRISMA flow diagram outlines study selection ( Figure 1). A total of 2,994 articles were retrieved and 191 duplicates were removed. Following title and abstract screening, 2,728 articles were excluded leaving 75 full-text articles to be screened. Six abstracts were excluded as following contact with abstract authors, no full-texts could be obtained. Ultimately, seven articles were included in this review.
Types of studies were mixed, including RCTs (n=3) and observational studies (n=4). Table 1 describes the methodological features of included studies. Included studies were varied in design, likely reflecting this emerging research field. Three studies compared an eHealth intervention to a control group. The remaining studies (n=4) used an eHealth intervention to measure PA in participants with mental health conditions. Mobile technologies such as smartphones and the Fitbit were used to measure PA levels and predict clinical signs and symptoms of mental health conditions such as mood. The length of interventions ranged from 9 days to 12 months 26,27 , with the majority of studies not assessing PA post-study completion. Only one study assessed maintenance at 6 months post-baseline 28 .
A quantitative synthesis of included data were planned, but was deemed inappropriate due to the heterogeneity of study design, participants, interventions and outcomes. Consequently, a qualitative synthesis of study interventions and results was completed. A number of sub-group analyses were planned, including comparing self-report and objectively measured PA and intervention focus such as smart phone applications vs. web-based interventions. Due to insufficient data in included studies these comparisons could not be completed.

Participant characteristics
Participant characteristics are also summarised in Table 1. A total of 811 participants were recruited with 102 dropping out. Ultimately, 709 participants were analysed across seven studies.
A total of 101 participants analysed had depression. There were 487 participants with psychological distress, identified with a score of ≥ 16 using the Kessler-10 screening tool. The remaining mental health conditions included; schizophrenia or schizophrenia spectrum disorders (n=69) and bipolar disorders (n=22). One study did not report the specific diagnoses of mental health conditions included 29 .
eHealth interventions and control treatments A variety of eHealth platforms designed to increase PA were described in these studies; web-based (n=4), web and mobile application (n=3) and e-mail-based (n=1), one study used both a web-based and mobile application 26 . A breakdown of each technological intervention is detailed in Table 2 below.
eHealth interventions included internet delivered cognitive behavioural therapy (CBT) 30 . An internet-based PA intervention and an internet-based therapist delivered self-help programme 28,29 . Control treatments included standard care 29 , waiting list care 28 and an active control group 30 . Participants in the active control group underwent a 12-week online programme that delivers health information on topics including nutrition, stroke, PA, medicines in the home, blood pressure and cholesterol, and heart health.
All experimental studies (n=3) reported eHealth interventions significantly increased PA levels from baseline, however, it is unclear if eHealth interventions are superior to traditional mental health services at increasing PA. Glozier and colleagues reported a greater proportion of participants with psychological distress (n=487) engaging in the recommended levels of PA (≥150 mins a week) who performed internet based cognitive behavioural therapy (ICBT) compared to the online active control group (67% in ICBT vs 61% in control group, Odds Ratio: 1.91, 95% CI: 1.01-3.61). In contrast two studies reported there were no significant differences in PA levels between eHealth interventions and control treatments. Mailey and colleagues noted an increase in PA in both the internet delivered PA intervention and standard care control group. However, there was a larger increase in mean PA in the intervention   .1)) in 36 participants with depression. In addition, sedentary behaviour reduced from baseline to 12 weeks, however, this did not reach statistical significance (Table 5).

Risk of bias of included studies
Risk of bias of all included studies is noted in Table 4 & Table 5. The Cochrane Collaboration's tool 25 was used to evaluate the risk of bias of the three included RCTs. The Downs and Black checklist assessed the risk of bias of the remaining observational studies (n=4). Individual risk of bias for all of the included studies is included as Supplementary File 3.

Discussion
This systematic review comprehensively searched and evaluated the effect of eHealth interventions on PA levels in participants with a range of mental health conditions. Overall, eHealth interventions appear to be beneficial at promoting PA, although consistent increases in PA were not demonstrated across    Table 6.
Glozier and colleagues noted a greater proportion of participants achieved the recommended levels of PA (≥150 mins a week) in favour of the e-health intervention compared to the control group (67% in ICBT vs 61% in control group, Odds Ratio: 1.91, 95% CI: 1.01-3.61). The risk of bias for this study as measured using the Cochrane collaboration tool was relatively low in a number of domains. In contrast, two studies comparing eHealth interventions to control treatments reported no significant differences between the intervention and control arms in terms of PA, however both of these studies were rated as unclear risk of bias in a number of domains 28,29 . In addition, both of these studies had much smaller sample sizes compared to the study by Glozier and colleagues which was notably much larger in size (n=487) compared to other studies included in this review. It should be noted however that Glozier and colleagues employed a subjective measure of PA, the IPAQ, compared to the more reliable objective measure of PA, the Actigraph accelerometer used by Mailey and colleagues 34 . Experimental studies in this review varied in the type of behavior change theory supporting the eHealth intervention. Glozier and colleagues employed an internet-delivered CBT approach in people with psychological distress and was compared to an online active control group. This online programme, Health-Watch, consisted of 12 weeks of information on topics such as PA and nutrition. In addition, Ström and colleagues performed a similar experimental study comparing internet-delivered CBT compared to a wait-list control group. It was not possible to individually assess or estimate whether it was the method of delivery or behavioural change theory supporting the intervention or a combination of these two elements which resulted in any observed changes.
eHealth technologies are rapid and constantly evolving through continuous software and hardware updates that regularly outpace medical research. The RCT is widely regarded the gold-standard of experimental research, however the mean duration from enrolment to publication is 5.5 years 39 . eHealth technologies are likely to become obsolete within this time-frame. A call has been made for medical research to evolve and adapt to maintain pace with developments in eHealth 40 . The Continuous Evaluation of Evolving Behavioural Intervention Technologies (CEEBIT) methodological framework has been proposed as an alternative to the conventional RCT design 41 . It is statistically powered to continuously evaluate eHealth applications throughout the study duration while accounting for updates to the application. Therefore, future eHealth interventions should consider using this novel methodological framework specific to the ever evolving eHealth technologies.
Further research is required to make a judgement of the ability of eHealth interventions to increase PA in people with mental health conditions. A recent systematic review showed that drop-out rates from exercise trials in people with depression are lower when delivered by a health professional with specific training in exercise prescription 42 . The need for qualified personnel to supervise PA programmes for people with schizophrenia was also echoed in a review by Vancampfort (2016) 43 . Drop-out rates of the PA arm of randomised controlled trials in people with schizophrenia was reported to be 26.7% 43 . Amalgamated drop-out rates for the current review show a lower drop-out rate of 12.5% but this may be reflective of the mixed mental health population with the majority having mild-moderate depression. It is not known whether the remotely delivered nature of eHealth interventions may result in less or more efficacious outcomes than traditionally delivered programmes. Head-to-head comparisons between these intervention mediums are necessary to elucidate the relative benefits of each.
Previous reviews in other clinical populations such as cancer survivors have reported that the initial results of eHealth technologies to increase PA in the cancer rehabilitation setting are promising 44 . However, similar to this review, weaknesses in methodological quality and uses of subjective measures of PA limit the interpretability of these findings.
Mental health conditions and CVD are inextricably linked as there is a high prevalence of CVD in people with mental health conditions due to a number of behavioural and lifestyle factors that confer increased CVD risk 45 , and similarly people with CVD have a high prevalence of mental health disorders 46 . Therefore, evaluation of the ability of eHealth interventions to ameliorate CVD risk is an important consideration, but this was beyond the scope of this review. Future reviews should explore this topic.
Perhaps due to this nascent field of research, the methodological quality of the included studies is low. This review has a number of suggestions to improve the methodological quality of studies examining eHealth interventions and PA participation among mental health populations. Future studies should use objective measures of PA, including but not limited to pedometers, accelerometers and wearable technology.
In addition, eHealth interventions should adhere to improved reporting of interventions, to ensure that such interventions can be repeated. Follow-up times in this review have varied from 9 days-12 months, with the majority of studies not recording PA levels in the maintenance phase. Therefore, the long-term implications of eHealth technologies to increase PA in a mental health population should be explored.

Limitations
There are several notable limitations to this review. Firstly, due to the relatively new nature of eHealth technologies to promote PA among people with mental health conditions, the number of studies included was relatively low (n=7). Secondly, six studies were excluded as only abstract proceedings were available. In each case the authors were contacted to ascertain if further information pertaining to these studies could be supplied however, no further data was supplied and these studies were subsequently excluded from this review. Although this significantly reduced the number of articles a lack of a detailed methodology may have increased bias if these studies were included. Thirdly, eligibility criteria in the study by Mailey and colleagues was unclear it was reported that participants with mental health disorders were recruited, however the criteria used to classify mental health disorders was not specified. Therefore, it is unclear the exact type of mental health disorders in this study population. In addition, Glozier and colleagues reported recruiting participants with mild-moderate depression. They used Kessler-10 screening tool to screen for depression, however it is a global measure of distress encompassing questions about both anxiety and depression. A possible further limitation is the distinction we have made between observational and interventional studies, as plausibly, if PA is monitored, this may in itself influence PA behaviour blurring the distinction between these two types of studies.
The extent of behavioural change as a result of monitoring PA using eHealth is not known at this time and warrants further investigation. Finally, both observational and interventional studies were included in this review which resulted in strong heterogeneity which precluded the ability to quantitatively analyse results.
Conclusion eHealth interventions appear beneficial at promoting PA and improving mental health symptoms for people with mental health conditions. Even though some of the included studies in this review demonstrated promising results, methodological restrictions and potential biases from using subjective measures of PA limit the interpretability of these results. Currently, it is unclear if eHealth interventions are superior compared to traditional interventions methods to increase PA. Larger well-designed studies are needed to extensively evaluate the true potential of this medium.

Data availability
All data underlying the results are available as part of the article and no additional source data are required.

Competing interests
No competing interests were disclosed.
The funders had no role in study design, data collection and analysis, decision to publish, or preparation of the manuscript.

Methods
In the 'eligibility criteria' section, was a full-text version of a study an inclusion criterion? Please provide full word for SMS In the data analysis section, suggest changing the word 'abstraction' to 'extraction' In Figure 1: please add in the box for number of duplicates removed in line with PRISMA flowchart.

Results
The results section is quite confusing to read, generally because of the way the information is presented in the different tables. The presentation of the results in the different tables could be clearer. I suggest merging the content of Tables 1-3 into 2 tables. Please also check the abbreviations in the tables are explained in the legends e.g. CVD, PAR-Q not explained currently. Review text in each table thoroughly to ensure information is presented more succinctly In table 2, in the column (sex baseline), suggest changing to 'Gender'. I am unclear why gender is reported for non-missing data and why the word 'baseline' is there. In relation to the Mailey study, if the mental health condition was not reported, how did you determine it was eligible for inclusion. In relation to Table 3: it would be more useful to provide information on risk of bias for individual items, as overall risk of bias is somewhat meaningless. Issues such as randomisation, baseline comparability, intention-to-treat analysis and drop-out rates are important to report to allow the reader to interpret results in the context of study methodology. Table 4: Row on Belwinkel study: results presented in the last column are confusing. I do not think that the between-patient analysis and between-patient analysis is relevant. The results should focus on pre-post intervention results. Row on Naslund study: in the last column, were these results related to post 6-month lifestyle programme? Row on Shin study 2016-please review wording in the first column to ensure clarity. Row on Strom study, please provide the units of measure for IPAQ in the last column In the paragraph on 'Participant Characteristics' is there a typing error here '709 participants 7 were In the paragraph on 'Participant Characteristics' is there a typing error here '709 participants 7 were analysed'. Please review this sentence. Pg 8, 3 paragraph, please clarify that 'significance' relates to 'statistical significance' Pg 8, 5 paragraph change 'there was no significant differences' to 'that no significant differences' In the same paragraph, what does the 'd' refer to in relation to the results presented-should this be p?

Discussion
The discussion section is well written and raises some interesting points. Please add a section outlining the limitations of this review Are the rationale for, and objectives of, the Systematic Review clearly stated? Yes

Are the conclusions drawn adequately supported by the results presented in the review? Partly
No competing interests were disclosed.

Competing Interests:
I confirm that I have read this submission and believe that I have an appropriate level of expertise to confirm that it is of an acceptable scientific standard, however I have significant reservations, as outlined above.
benefits of physical activity and describing eHealth. Further information on the benefits of using eHealth for people with mental health conditions, over other modes of service delivery, would strengthen the rationale for this review.
Appropriate methods were used to address the objective of the review. In particular, study eligibility was assessed independently by two reviewers, data were extracted by two reviewers using a standardised data extraction form, and risk of bias was assessed. The methods and results of this review are strengthened by the use of the PRISMA checklist and by registering the protocol in advance of conducting the review.
However, the review would be improved if there was better alignment between the objective, methods and results. There is discrepancy between the objective of the review, the types of study included in the review, and the results. The description of the types of study included in the review could be clearer. Study designs are often considered as "intervention" studies and "observational" studies. An intervention study may be randomised or non-randomised. Critically, the inclusion of studies that include an intervention but are termed observational studies seems contradictory. This may be improved if the use of eHealth is considered an exposure and studies were therefore classified as cohort studies, but if this is the case it could be explained more clearly. The lack of clarity over study design is compounded by the description of study design in Table 1. For example, Beiwinkel et al. is reported as a pilot observational study, but the type of observational design is not reported (e.g. cohort, case-control, cross-sectional), while this detail is provide for Naslund (i.e. "observational cohort study"). Perhaps of more importance is that Kerr et al. is reported as a pilot study but it is not clear if it is an intervention study or observational study.
The authors identify only a small number of studies examining the effect of eHealth on physical activity in people with mental health conditions (n=7), and only three of these are randomised controlled trials (n=3). The results of the randomised controlled trials provide inconsistent evidence that eHealth results in an increase in physical activity. However, it is difficult to determine if the content of the intervention (e.g. CBT) or the fact that it's delivered electronically is responsible for the change in the PA. It would help the reader to understand the causal pathway if the content of the comparator was described when reporting the results of the study. For example, if the control group received the same intervention (e.g. CBT) as the intervention group but delivered in person rather than electronically it is more plausible than any change in PA is due to eHealth rather than the content of the intervention. If there is no difference in PA between groups eHealth may offer additional benefits such as being more feasible and cost-effective to deliver.
It is not clear, based on the objective and eligibility criteria, why studies that used an eHealth intervention to measure physical activity were included. If monitoring physical activity was considered an intervention, and physical activity was measured pre-and post-introduction of PA monitoring then these studies should be considered intervention studies, and the outcome reported should be change in PA. Monitoring PA is a behaviour-change technique, and as such is very likely an intervention. The authors do report change in PA following PA monitoring for some studies but not all.
The results include findings relating to associations between physical activity/change in physical activity and symptoms and weight loss. While this is of interest it goes beyond the objectives of the review. If an objective of the review is to explore associations between PA and other variables in people with mental health conditions it should be stated under the study objective and the eligibility criteria should be amended accordingly.
Further information on how each domain on the risk of bias tools were assessed and how the overall risk of bias was determined for RCTs should be provided to allow the reader to understand the quality of the of bias was determined for RCTs should be provided to allow the reader to understand the quality of the studies and justify areas for future research. It is stated in the discussion that Glozier et al. used a self-report measure of physical activity and this may introduce bias. This should be captured under detection bias when assessing risk of bias. However, despite this, Glozier et al. is rated as low risk of bias. It would be beneficial to understand how the authors determined the study was at low risk of bias despite the outcome assessor not being blinded (as a self-report measure was used).
The authors provide a summary of the findings of the review and discuss a number of pertinent points including the potential to introduce bias by using self-report measure of PA and the need for studies to directly compare eHealth with physical activity interventions delivered in person. The discussion suggests that eHealth is feasible. However, examining the feasibility of eHealth is not a specific objective and the presentation of results do not support this conclusion. If this is of interest, and studies examine feasibility, this should be an objective and results reported accordingly.
The authors reach a sound conclusion based on the findings of the review, acknowledging the limitations with previous studies. However, given that Glozier et al. was a large study (n=562) and it was rated low risk of bias, the statement that larger well designed studies are needed is not justified. Specific information about the limitations of Glozier et al., which a future trial needs to address, should be provided.
Further points for consideration: To improve consistency throughout the review, the title and objective should align (i.e. the title states to "promote physical activity in patients with mental health conditions" and the objective states "to increase physical activity in individuals with mental health conditions".
The results section of the abstract is misleading and suggests different conclusions than those presented in the main text. Although three studies reported that eHealth increased PA, the change in PA over time is not of interest when a control group is included, as is the case in these studies. Where a control group is included the effect of interest is whether a larger change in PA is observed in the intervention group in comparison to the control group.
A definition of eHealth could be provided under "eligibility criteria" to further explain decisions regarding exclusion of studies. The description of eHealth in the introduction, as "..the transfer of health resources and health care by electronic means, including but not limited to the delivery of health information through the internet and mobile technologies" suggests that interventions that use only telephone calls, SMS or conference calls, are eHealth. These studies were however excluded.
A definition of mental health conditions and/or examples of mental health conditions of interest included in the section on eligibility criteria would be beneficial in order to describe the participants of interest.
Under eligibility criteria it states that "we included the following methods of measuring PA…". These, however, are ways of quantifying rather than measuring PA. As the sentence after this correctly states, methods of measuring PA are self-report measures e.g. questionnaires, and objective measures, e.g. accelerometers, indirect calorimetry.
For clarity, it would be helpful if the authors stated if data were extracted independently by two researchers.
The authors describe where they were not able to conduct the review as planned e.g. quantitative synthesis. However, it would be helpful for the reader to have a section summarising the differences Reviewer Report This systematic review explores the available evidence to support ehealth technology (utilising internet and mobile technologies) as a stand-alone intervention or as part of a multimodal intervention to increase physical activity in individuals with mental health conditions. The introduction addresses the importance of Physical Activity (PA) in general and mental health populations and supports the use of eHealth technology, as an interventional means to affect positive change in mental health. It highlights that no systematic review to date has considered the utility of eHealth technology to promote PA for mental health conditions/ in mental health populations and as such this review makes a valuable contribution to the literature in the area. The methodology outlined conforms to best practice guidelines in systematic review (PRISMA) and the review was registered with the international prospective register of systematic reviews (PROSPERO).
As behavioural intervention technologies are a relatively new and evolving area, the authors of the review opted to include both experimental and observational studies in their review and data synthesis. The reviewers believe that this was the correct choice and commend the authors in taking this approach. However this brings its own challenges when reporting and summarising the available evidence in a clear and accessible format for readers of this review. These difficulties are further compounded when the interventions delivered include multi-modal packages underpinned by different theoretical frameworks for behaviour change, different delivery platforms and heterogeneous populations with a variety of mental health conditions. Greater clarity on how the authors are distinguishing between mobile technologies which monitor activity levels versus those which are utilised as an intervention designed to improve PA. For example the Shin et al., 2016 study included in this review, employs a Fitbit to measure activity levels in individuals with chronic schizophrenia and examines the association between this measure and psychopathological profiles. Table 5 which summarises the included e-health interventions, outlines that feedback of PA levels to participants was not reported in the Shin et al. study, thus ruling out the device as an ehealth intervention for promoting PA. While the results of this study are narratively reported in the review synthesis, it is not clear whether this was a pre stated objective in the methodology.
To address these issues, the reviewers suggest that the title and objectives should reflect broader objectives e.g. The use of eHealth to promote or monitor physical activity in individuals with mental health conditions: a systematic review In addition, given the complexity in study types and interventions considered in this review, the reviewers recommend the authors state their objectives as clinically relevant questions and that the results reported follow this outline and conclude with a best evidence synthesis summary for each question.
recommend the authors state their objectives as clinically relevant questions and that the results reported follow this outline and conclude with a best evidence synthesis summary for each question. e.g. Can eHealth technology as a stand-alone or component of a multimodal intervention improve PA in individuals with a mental health condition? Do eHealth interventions designed to promote PA improve general or mental health profiles in individuals with a mental health condition? Are eHealth captured PA levels, of individuals with a mental health condition, associated with severity of clinical signs and symptoms of the mental health condition?
Results of the review can then be narratively synthesised and a best evidence synthesis summary provided. An example of what is suggested as summary is provided below but should be in table format:

Question:
Can ehealth technology as a stand-alone or component of a multimodal intervention improve PA in individuals with a mental health condition? Best available evidence Evidence from one high quality RCT supports the use of ehealth technology as part of a multimodal intervention in individuals with psychological distress and concomitant CVD in increasing the likelihood of achieving PA guidelines for adults (Odds Ratio 1.91, 95%CI: 1.01-3.61) Question: Do ehealth interventions designed to promote physical activity improve general or mental health profiles in individuals with a mental health condition? Best available evidence.....
Other comments the peer reviews raise for consideration are summarised below:

Title:
The reviewers recommend that the authors change the wording to patients with mental health conditions and continue this throughout the text. individuals with mental health conditions

Abstract:
In the results, it is not correct to say that four studies reported that higher levels of PA resulted in improvements in mental health outcomes. We suggest this should read as ... four studies reported that higher levels of PA are associated with better mental or general health profiles.

Manuscript:
In the data extraction and analysis section, the authors provide results with reference to Methods: heterogeneity in study design, participants etc. These should be reported in the results section. In the methods section only the proposed analysis should be stated. Similarly with respect to subgroup analyses, only the proposed analyses should be reported in the methodology section. Results of the data presented from the review that prohibited this analysis should not be presented in this section.
The reviewers recommend collapsing a number of the study summary and quality review tables Tables: and include the reference number with the study identified to allow the reader to better consider multiple study aspects together and reference back and forth more easily to the narrative summary provided.
In the results section of table 4 the reviewers advise the following: Beiwinel et al., remove the between patient and within patient analysis labels as these are not relevant to the analysis and results provided 22 the analysis and results provided Kerr et al., please report were within group changes statistically significant and provide p values Mailey et al., please report the results of the main effect analysis. Naslund et al., please detail whether within group change scores over the intervention period were reported and their significance Sin et al., please provide a reference correlation coefficient (r or rho) used in the analysis conducted and comment on the strength of the association identified (moderate to strong) before addressing whether these were significant associations Strom et al., please provide the results of the within and between group analysis conducted

Results:
The peer reviewers request further clarity with respect to the Glozier study in the participant characteristics section of the text. Here the Kessler-10 screening tool for psychological distress was used to identify individuals with CVD and a mental health condition identified as a K-10 >=16. As this tool screens for both symptoms of depression and anxiety, please justify the inclusion of this group under the summary of included participants with a depressive disorder.
In the risk of bias section in the text, please provide a summary of the quality of the observation studies.
Results are difficult to interpret from the table provided. Please indicate whether the results generated indicate high or low quality in the included studies. This detail becomes more relevant when readers consider the best available evidence, in the absence of RCTs in the area.

Discussion:
The discussion section considers the findings of this review well and with broad consideration of the contemporary literature. As the primary finding of this review lies in a population of individuals with high psychological distress and concomitant CVD, the discussion would benefit from considering what is known with respect to eHealth technologies for promoting PA in CVD. Similarly where the association between higher levels of PA and clinical signs and symptoms is discussed, it would be worth exploring the broader PA in mental health literature to strengthen discussion in this area. Here the peer reviewers reiterate concerns in stating that increased PA levels are associated with improved clinical signs and symptoms. This is only possible where the studies have looked at the linear association between changes in clinical signs and symptoms and changes in PA levels and we believe the studies cited have not clearly addressed this.
In conclusion this systematic review was conducted with scientific rigour and makes a valuable contribution to the scientific literature in the area. The authors, as a result of this review, provide valuable recommendations for future research with respect to PA measurement and intervention reporting that will enable better standardisation and data synthesis at a future date.

Are sufficient details of the methods and analysis provided to allow replication by others? Yes
Is the statistical analysis and its interpretation appropriate?