Along with poor mental health, people with severe mental illness (SMI), such as bipolar disorder, schizophrenia, and other psychotic disorders, show elevated risks of engaging in adverse health behaviours (Carney, Cotter, Bradshaw, Firth, & Yung, 2016; Firth et al., 2019). For example, in comparison with the general population people with SMI are more likely to smoke cigarettes (Prochaska, Das, & Young-Wolff, 2017), are less physically active, and have higher daily calorie and sodium intake (Teasdale et al., 2019; Vancampfort et al., 2017). This may be partly attributable to the psychotropic medications used to treat SMI, as antipsychotics have been found to increase appetite, delay satiety signalling, and cause sedation (Mazereel, Detraux, Vancampfort, Van Winkel, & De Hert, 2020). Finding novel ways to promote healthy lifestyles in SMI is crucial for reducing morbidity and mortality (Firth et al., 2019), with increasing evidence to suggest this could also improve mental health outcomes (Firth et al., 2020; Pape, Adriaanse, Kol, van Straten, & van Meijel, 2022).
Health behaviour change (HBC) interventions include a broad range of psychological techniques, and target modifiable health behaviours such as diet, physical activity, smoking, sleep, substance or alcohol use, and medication adherence. Traditional face-to-face HBC, while ideal in many respects, interventions are resource intensive (Bennett & Glasgow, 2009) and can be impacted by the capability and capacity of the person delivering the intervention. Interest in online HBC (web-based and smartphone) has grown rapidly in popularity (Arigo et al., 2019), given their potential to improve access to HBC for people with SMI, without relying on costly face-to-face interventions (Young et al., 2017). Previously, there have been concerns that people with SMI may experience socio-economic barriers – such as unstable housing, low income, and unemployment – which may limit their access to the internet and online interventions (Borzekowski et al., 2009). Encouragingly however, smartphone and internet use is increasing among those with SMI (Firth et al., 2016; Thomas, Foley, Lindblom, & Lee, 2017; Trefflich, Kalckreuth, Mergl, & Rummel-Kluge, 2015).
While previous reviews have focused on the feasibility and acceptability of digital interventions generally for symptom management and relapse prevention in SMI (Naslund, Marsch, McHugo, & Bartels, 2015b), there is still limited understanding of how digital HBC could work to improve outcomes specifically in this population. Therefore, this review aimed to systematically identify and evaluate the current evidence around the feasibility, acceptability, and effectiveness of digital HBC for not only physical health, but also broader behavioural and psychological well-being outcomes, in people with SMI.
Specifically, this review addressed the following research questions: (i) are digital approaches towards delivering HBC feasible and acceptable for people with SMI?; (ii) can digital HBC for people with SMI change health-related behaviour?; (iii) can digital HBC for people with SMI improve health outcomes?; and (iv) what specific intervention components and strategies influence user engagement with digital HBC interventions in people with SMI?
The Preferred Reporting Items for Systematic Reviews and Meta-Analyses (PRISMA) checklist for reporting systematic reviews (Moher, Liberati, Tetzlaff, Altman, & PRISMA Group, 2009) guided this review, which was pre-registered on the online review protocol database, PROSPERO (CRD42021261267).
A systematic literature search was conducted in January 2022 using the following databases: Cochrane Central Register of Controlled Trials; Health Technology Assessment; AMED (Allied and Complementary Medicine); APA PsycInfo; Embase; and MEDLINE®, using the following keyword search algorithm: [psychosis OR psychotic OR schizophr* OR severe mental OR serious mental OR bipolar] AND [Behaviour change OR Behavior change OR behavioural change OR behavioral change OR Lifestyle OR Healthy Living OR Health Behaviour OR Health Behavior OR physical activity OR exercise OR smok* OR tobacco OR sexual health OR Sleep OR Alcoho* OR diet* OR Sedentary OR substance abuse OR weightloss OR weight loss OR obes*] AND [online or web-based or app-based or Internet or e-Health or mhealth or smartphone or mobile phone or iphone or android or wearable or digital].
Searches were restricted to publication in English language in peer-reviewed journals and all articles were included regardless of publication date. Reference and citation list searches were also conducted to search for additional studies, alongside a basic search of Google Scholar and the Journal of Medical Internet Research (JMIR).
English language articles were included. Randomised controlled trials (RCTs), non-RCTs, pilot studies, feasibility studies, quasi-experimental studies, and qualitative studies examining the feasibility, acceptability, or effectiveness and other outcomes of a digital HBC, delivered online via computer smartphone apps, social media and/or ‘wearable’ formats, among people with SMI were eligible.
For the purpose of this review, ‘SMI populations’ included any groups of individuals (of any age) diagnosed and/or receiving treatment for bipolar or psychotic disorders. Studies of non-entirely SMI samples were included, if either (i) where data pertaining to the SMI sub-sample were reported separately, or, (ii) where the overall sample contained over two-thirds of individuals diagnosed/treated for SMI.
Studies that reported changes in health behaviours relevant to physical health and overall well-being (such as smoking, substance use, sleeping, diet, physical activity, and sexual behaviours) as primary or secondary outcomes were included. Studies will be eligible for inclusion if they deliver a behavioural change intervention fully online, or where the digital technology forms a well-defined and central part of a multi-component intervention. Interventions in which the HBC aspect only made up a tangential or minor part of the intervention, or where relevant technological aspects were limited to text messages, emails or phone calls, were excluded.
Initial screening of titles and abstracts was conducted by one reviewer (C.S.). Two reviewers (C.S. and G.M.), who were blind to each other's review, independently reviewed all full-text articles meeting the inclusion criteria (interrater agreement: 83%). A third reviewer (J.F.) resolved any disagreements between the two reviewers.
Data were independently extracted by two reviewers (C.S. and G.M.), using a pre-determined data extraction form specifically designed for this review. The extraction form collected the following data: (i) study information (sample size, mean age of participants, diagnostic information, and study design); (ii) intervention features (intervention platform, app/programme name, trial/feasibility details, regularity of instructed use, digital intervention summary, any additional intervention components, and details of the control condition); and (iii) effects on behaviour and health outcomes (changes in behaviour, changes in physical and/or mental health before and after interventions).
Given there are no established standards for assessing feasibility, acceptability, and usability (Greenhalgh et al., 2017; Jacob, Sezgin, Sanchez-Vazquez, & Ivory, 2022), measures were chosen from validated scales [e.g. the System Usability Scale (SUS) (Hyzy et al., 2022)] and in line with prior research (Balaskas, Doherty, Schueller, & Cox, 2021). Feasibility measures included recruitment rates, attrition to study, reasons for refusal or ineligibility, and adherence to intervention. Acceptability measures included usage data (e.g. duration of use, modules completed, etc.) and user perspectives from interviews and quantitative questionnaires (e.g. regarding relevance of content and readability). Usability measures included the SUS, task scores, and interview comments regarding design, layout, and/or other aspects of the user interface. Finally, behavioural outcomes included intervention effect sizes and/or changes in target behaviour, depression, and/or anxiety.
Due to variations in design, intervention approaches, and primary outcome measures it was not appropriate to conduct a meta-analysis and therefore quantitative findings were synthesised narratively (Liberati et al., 2009). For qualitative studies, themes were identified and the detailed analytical narratives reported within the text were summarised, following the principles of thematic synthesis (Thomas & Harden, 2008). Mixed methods studies contributed separately to both types of synthesis. The data in each section were different and were therefore not double counted.
Overall, 2196 results were identified from the databases. Figure 1 presents a PRISMA flowchart of study selection procedures. After removing duplicates (N = 262), 1934 titles and abstracts were screened. Following initial screening, 1827 articles were removed, leaving 107 full texts to be reviewed. A further 76 were then excluded, leaving 31 papers. Five additional articles were found through separate literature searches of reference lists and Google Scholar, yielding 36 papers in total.
Figure 1.
PRISMA flowchart of study selection.
[Figure omitted. See PDF]
Tables 1–3 summarise the characteristics of selected studies, including descriptions of participants and interventions. There were 14 pilot RCTs, 14 non-RCTs, three qualitative studies, and five mixed-method studies. The majority (N = 35) of studies were conducted in the United States, with only one study conducted in the UK, Australia, Portugal, Netherlands, and Canada. Fourteen of the studies consisted of mainly White participants. Sample sizes ranged from 5 to 2570 participants. Schizophrenia was the most common diagnosis in smoking and physical activity studies, whereas bipolar disorder was the most common diagnosis for ‘other behaviours’ (Tables 1–3).
Table 1. Descriptive characteristics for the studies on digital interventions in SMI for smoking
Authors, year |
Study design |
Target sample, N | Diagnosis (%) | Mean age ( |
Intervention mode | Number phases |
Control condition | Details about intervention |
---|---|---|---|---|---|---|---|---|
Browne et al. (2021) |
Pilot RCT, 16 weeks | ⩾18 years old with ICD-10 diagnosis of an SMI and smoke ⩾5 cigarettes daily. |
SZ 24% |
LTQ 46.1 ( |
Digital [compared two apps; Learn to Quit (LTQ) v. QuitGuide] | Phase: n/a |
None | LTQ: 28 modules that provide knowledge, skills, and recommendations for smoking cessation designed for persons with SMI. |
Brunette et al. (2011) |
Quasi-experimental design |
Adult smokers with an SMI who are receiving care in supported housing or psychiatric rehabilitation. |
SZ 34.1% | I: 47 ( |
Digital (website electronic decision support system) | Phase: n/a |
Waitlist | Electronic decision support system (30–90 min): videorecorder narrator, who identifies as a former smoker with SMI, guides users through the programme and encourages them to quit smoking. |
Brunette et al. (2012) |
Usability and qualitative study |
Adult smokers receiving treatment for an SMI. |
SZ 75% |
49.9 (8.9) years old |
Digital (four websites) |
Phase: n/a |
None | n/a assessed whether four smoking cessation websites met usability guidelines |
Brunette et al. (2016) |
Usability study |
Adult smokers with psychotic illnesses who are interested in quitting in the next month. |
SZ spectrum 91% | 49 (10) years old |
Digital (website prototype Let's Talk About Quitting Smoking) | Phase: 2 |
None | Website was adapted for people with SMI.
|
Brunette et al. (2018) |
Pilot RCT |
Daily smokers aged 18–30 with schizophrenia. |
SZ spectrum 43.2% |
All 24.2 (3.6) years old |
Digital (website either LTAS or NCI patient education handout presented on a laptop) | Phase: n/a |
None | LTAS: tailored for those with an SMI. Focuses on financial and social impacts of smoking and quitting. |
Brunette et al. (2019) |
Pilot |
⩾17 years old daily smokers, who had a diagnosis of SMI and was an outpatient who want to quit in the next month. |
SZ spectrum 60% | 46 (13.6) years old |
Digital (website LTAS) | Phase: n/a |
None | Practised accessing website, once able to use they were provided with a laptop and jetpack for wireless internet and asked to use website 5× a week. |
Brunette et al. (2020) |
RCT |
Adult daily smokers with schizophrenia spectrum disorders. Smokers were excluded if they had recently (past month) used evidence-based smoking cessation treatment. |
SZ spectrum 100% | 45.91 (11.3) years old |
Digital (website either LTAS or NCI education handout presented on a laptop) | Phase: n/a |
None | LTAS: tailored for those with an SMI. Focuses on financial and social impacts of smoking and quitting. 3 modules assessment/feedback, quitting intentions, and education about smoking cessation treatments |
Ferron et al. (2011) |
Qualitative study |
Daily smokers with mental health condition. |
SMI 100% | Age: NR |
Digital (website does not state name) | Phases: 3 |
None | Not included. |
Gowarty et al. (2021a) |
Usability study |
18–35 receiving treatment for SMI and a smartphone user. |
Psychotic disorder 41% | 29 (4) years old |
Digital (app assigned to either QuitGuide or quitSTART app) | Phase: n/a |
None | Both apps have many similar features (such as set notifications based on time or location), but their design and content differ.
|
Gowarty et al. (2021b) |
Usability study |
18–35 receiving treatment for SMI and a smartphone user. |
Psychotic disorder 41% | 29 (4) years old |
Digital (compared two apps, QuitGuide v. quitSTART) | Phase: n/a |
None | Both apps have many similar features (such as set notifications based on time or location), but their design and content differ.
|
Heffner et al. (2018) |
RCT |
Daily smokers with a mental health condition. |
BP 8.6% |
46.2 (13.3) years old |
Digital (website compared WebQuit Plus v. Smokefree.gov) | Phase: n/a |
None | WebQuit Plus: the website is ACT based and has four parts that help users: (1) make a quit plan, (2) develop awareness of smoking triggers, (3) develop acceptance-based coping skills to handle triggers, and (4) identify and engage personal values and self-compassion to support long-term abstinence. |
Heffner et al. (2020) |
Pilot RCT |
Bipolar I or II disorder. |
BP 88% | 49.0 years old (10.8) |
Digital (website compared WebQuit Plus v. Smokefree.gov) | Phases: n/a |
None | WebQuit Plus: the website is ACT based and has four parts that help users: (1) make a quit plan, (2) develop awareness of smoking triggers, (3) develop acceptance-based coping skills to handle triggers, and (4) identify and engage personal values and self-compassion to support long-term abstinence. |
Klein et al. (2019) |
Qualitative study |
Adults with SMI who had attempted to quit smoking in the last 12 months or ex-smokers. |
SZ 75% |
Median: 47.5 (range 31–53) years old |
App (Kick.it) | Stages: 2 |
None | Four core features are:
|
Medenblik et al. (2020) |
Pilot RCT |
Adults aged between 18 and 70 years old, who smoke at least 10 cigarettes per day and have smoked for ⩾1 year. |
SZ 44% |
48.2 (9.9) years old |
Blended: in-person and phone counselling and apps (mcm readings uploaded on an app) | Phase: n/a |
Intensive treatment comparison, which consisted of: 5 CBT counselling sessions and pharmacotherapy. | iCOMMIT consists of: |
Vilardaga et al. (2016) |
Usability study |
Adult with SMI. |
SZ 40% |
51.2 (4.27) years old |
Digital (QuitPal app) | Phase: not stated |
QuitPal | |
Vilardaga et al. (2018) |
Mixed methods |
Daily smokers recruited from outpatient MH clinic. |
SMI 100% | 44 (7.5) years old |
Digital (apps LTQ v. QuitGuide) | Phases: 3 AB crossover intervention, B-phase training designs and bi-phasic, AB single-case design |
None | LTQ: developed for those with an SMI. The app has 28 modules (14 lessons and 14 skills) that provide knowledge, skills, and recommendations for smoking cessation. |
Vilardaga et al. (2019) |
Case studies with crossover AB interventions |
Adults with SMI. |
SZ 14.2% |
45 (9.5) years old |
Digital (compared two apps LTQ v. QuitGuide) | Phases: total 7 |
None | LTQ: developed for those with an SMI. The app has 28 modules (14 lessons and 14 skills) that provide knowledge, skills, and recommendations for smoking cessation. |
Vilardaga et al. (2020) |
Pilot RCT |
Adults with SMI and stable housing, who smoke >5 cigarettes a day. |
RMD 27% |
LTQ 46.1 years old (11.3) |
Digital (LTQ app v. QuitGuide app) | Phase: n/a |
None | LTQ: developed for those with an SMI. The app has 28 modules (14 lessons and 14 skills) that provide knowledge, skills, and recommendations for smoking cessation. |
Wilson et al. (2019) |
Successive cohort design |
Aged 18–70, smoke ⩾10 cigarettes daily and smoking for ⩾1 year, and met criteria for schizophrenia, schizoaffective disorder, or another psychotic disorder. |
SZ 46.7% |
47.8 years old (11.0; 26–63) |
Blended: digital (Stay quit coach app and mcm readings uploaded on an app) and in-person and telephone counselling | Phase: two cohorts |
None | iCOMMIT consists of: |
BD, bipolar disorder; CBT, cognitive behavioural therapy; FTDN, Fagerström test for nicotine dependence; LTAS, Lets Talk About Smoking; LTQ, Learn To Quit; MDD, major depressive disorder; MH, mental health; NA, not applicable; NCI, National Cancer Institute; NR, not reported; NRT, nicotine replacement therapy; NS, not significant; PNTS, prefer not to say; PNOS, psychosis not otherwise specified; PPA, point prevalence abstinence; QG, quit guide; RCT, randomised controlled trial; RMD, recurrent major depression;
Table 2. Descriptive characteristics for the studies on digital interventions in SMI for physical activity
Authors, year |
Study design |
Target sample, N | Diagnosis | Mean age ( |
Intervention mode | Intervention duration number phases | Control condition | Intervention components |
---|---|---|---|---|---|---|---|---|
Aschbrenner et al. (2015) |
Mixed-methods repeated measure design |
Adults with an SMI diagnosis and a BMI ⩾25 |
SZ 20% |
46.6 years old (8.7; 30–57) |
Blended – digital elements: wearable and app | 24 weeks in 3 phases: |
None | In-person: |
Aschbrenner et al. (2016a) |
Pilot RCT |
Adults with SMI and BMI ⩾30 |
SZ 22% |
48.8 years old (11.9) |
Blended – digital elements: wearable, app, and social media | 24 weeks |
None | In-person: |
Aschbrenner et al. (2016b) |
Feasibility study |
⩽21 years old with an SMI and BMI ⩾30 |
SZ 27% |
48.8 years old (10.6; 21–58) |
Blended – digital elements: text reminders, wearable, app, and social media | 24 weeks | None | In-person: |
Aschbrenner et al. (2021) |
RCT |
18–35 years old with an SMI and BMI ⩾25 |
SZ or SZA 40% |
28.4 (4.5) years old |
PeerFIT: |
PeerFIT: 12 months |
None | PeerFIT |
Campos et al. (2015) |
Feasibility study and quasi-experimental trial |
Adults diagnosed with schizophrenia |
SZ: 100% | I: 39.77 years old (9.2) |
Digital: Exergame (Microsoft Kinect online) | 8 weeks | TAU | Exergame, Microsoft Kinect: |
Looijmans et al. (2019) |
RCT |
SMI patients with at least one metabolic risk factor |
Psychotic disorder 57.6% |
46.1 (10.8) years old |
Blended – digital element (webtool) | 12 months, two phases |
TAU | 26 fortnightly meetings between MH nurse and service user, using the web tool. |
Macias et al. (2015) |
RCT |
Individuals with an SMI diagnosis and own a smartphones |
SZ 40% |
Range 22–61 years old |
Digital (WellWave app) | 4 weeks | None | Digital: app which prompts, records, and tracks walking (the date and time of each walk, duration in minutes, step count, and speed) |
Muralidharan et al. (2018) |
RCT |
Veterans with SMI and (BMI) >30 or BMI >28 with weight gain of ⩾10 lbs in the last 3 months |
100% SMI | WebMOVE: 54.7 (8.9) years old |
Digital (website; WebMOVE) v. |
6 months | TAU | MOVE-SMI: |
Naslund et al. (2015a, 2015b) |
Feasibility study |
Adults with an SMI and who are overweight/obese |
SZ 30% |
47.7 (9.0; 30–58) years old |
Blended – digital elements (wearable and tracking app) | Duration: wore wearable for 80–133 days depending on date of enrolment | None | In-person: |
Naslund et al. (2016) |
Feasibility study |
Adults with an SMI and have a BMI ⩾30 |
SZ 27% |
48.2 (11.2) years old |
Blended – digital elements (wearable, tracking app, and social media) | 6 months | None | In-person at a community mental health centre: |
Naslund et al. (2018) |
Feasibility study |
⩾21 years old with an SMI and a BMI ⩾30 |
SZ 20% |
49.2 (11.8) years old |
Blended – digital elements (wearable, tracking app, and social media) | 6 months | None | In-person delivered at community mental health centre: |
Olmos-Ochoa et al. (2019) |
Mixed method (RCT + Qual) |
Veterans with SMI and BMI >30 or BMI >28 with weight gain of ⩾10 lbs in the last 3 months |
100% SMI | WebMOVE: 54.7 (8.9) years old |
WeBMOVE: digital (website, calls, and pedometer) MOVE-SMI: in-person | Interviewed after 6 months on intervention | TAU | WebMOVE – Digital: |
Sylvia et al. (2021) |
Comparative effectiveness study |
18–65 years old with a self-reported lifetime depression, and increased CVD risk. |
100% self-report of lifetime depression | 43.5 (11.5) years old |
CBT: digital (website and wearable) |
8 weeks | None | CBT – Digital: |
Young et al. (2017) |
RCT |
Adults currently prescribed antipsychotic medication and a BMI >30 or BMI >28 with weight gain of ⩾10 lbs in the last 3 months. |
SMI 100% | WebMOVE: 55.5 (9.2) years old |
WebMOVE: digital (website + weekly telephone calls) |
6 months | TAU (educational handout on the benefits of weight loss) | WebMOVE* – Digital: |
BD, bipolar disorder; BMI, body mass index; C, control; CBT, cognitive behavioural therapy; CVD, cardiovascular disorder; I, intervention; lbs, pounds; MCBT, mindfulness cognitive behavioural therapy; MH, mental health; mHealth, mobile health; PA, physical activity; RCT, randomised controlled trial;
Table 3. Descriptive characteristics for the studies on digital interventions in SMI for others
Authors, year |
Study design |
Study focus | Target sample, N | Diagnosis (%) | Mean age ( |
Intervention mode | Number phases |
Control condition/comparator | Detail about intervention |
---|---|---|---|---|---|---|---|---|---|
Hammond et al. (2020) |
RCT |
Promoting seeking treatment for substance use after discharge | Self-reported substance use in last 30 days and a diagnosis of SUD and a psychiatric illness. |
MDD 42% |
C: 39.8 years old (8.1) |
Digital (website) | Duration not stated but completed while an inpatient at specific, monitored times (up to 3 × 1 h sessions a week) | Control: TAU | 65 interactive modules on substance use-related topics (e.g. drug refusal skills training effective, HIV and AIDS, and problem solving). |
Melamed et al. (2022) Canada | Single-arm, uncontrolled study |
Perceived benefit of changing smoking habits, physical activity, and nutrition |
Aged 16–29, diagnosed with a psychotic disorder within the past 5 years. |
SZ 13% |
HI: 22.7 years old (3.2) |
HI: digital (website and online calls) |
12 weeks | None | HI: |
Taylor et al. (2022) |
Feasibility study |
Sleep | Aged 16–65 with a schizophrenia-spectrum diagnosis and experiencing sleep difficulties for the least 4 weeks |
MDD 42% |
35.57 years old (10.88; 22–57) |
Blended – digital element (ExpiWell app) | 6 weeks | Control: None | In-person: prior to the digital intervention, participants met with the research therapist to |
BD, bipolar disorder; e-platform, electronic platform; FU, follow-up; HI, high intensity; IQR, inter-quartile range; LI, low intensity; MDD, major depressive disorder; NS, non-significant; OMI, other mental illness;
The HBC used included those delivered entirely digitally and those using ‘multi-component’ approaches (i.e. digital and in-person aspects). Tables 4–6 present study outcomes and interventions. Nineteen studies focused on smoking as the primary behavioural outcome (Tables 1 and 4). Fourteen studies focused on physical activity, weight loss, and cardio-metabolic health (Tables 2 and 5) and three papers focused on ‘other behaviours’ (Table 6), specifically sleep (Taylor, Bradley, & Cella, 2022), substance use (Hammond, Antoine, Stitzer, & Strain, 2020), and invoking changes in the perceived benefit of changing health behaviours, rather than the behaviour itself (Melamed et al., 2022). No HBC targeted sexual health.
Table 4. Key outcomes and findings from studies of digital interventions in SMI for smoking
Authors, year | Recruitment | Attrition | Adherence to intervention | Usage data | Usability (SUS) task scores | User perspectives and feedback (questionnaires) | User feedback (interview) | Behavioural outcomes | Physical and mental health outcomes |
---|---|---|---|---|---|---|---|---|---|
Browne et al. (2021) | NR | NR | NR | • Number of days interacted were similar between groups (LTQ: 34.1 days v. QuitGuide: 32.0 days)
|
NR | NR | NR | Number of cigarettes smoked were reduced in both groups |
NR |
Brunette et al. (2011) | 43/48 approached (90%) were interested
|
0% lost at 2 month FU | 100% | NR | NR | NR | The electronic decision system was comprehensible, easy to use. | 2 month FU: participants who used the decision support system were significantly more likely to have engaged in ⩾1 smoking cessation behaviour (67% v. 35%):
|
NR |
Brunette et al. (2012) | Not reported | 0% | NR | NR | 0 websites were usable for minimal computer users (<5 times) |
NR | NR | NR | NR |
Brunette et al. (2016) | NR | NR | Phase 1: ⩾75% of users were highly satisfied with modules 2 and 4–8.
|
Phase 2:
|
Phase 1:
|
Phase 2 (self-reported at 3 weeks FU):
|
NR | ||
Brunette et al. (2018) | 89 consented and assessed for eligibility 81 (91%) were eligible and enrolled 7 (8%) screen fails 1 (1%; control) dropout | 11.1% at 3 month FU | 100%
|
LTAS: on average 58 (±22) min on intervention |
NR | A higher % of users rated LTAS at least good compared to users of NCI website (83.4% v. 71.4%)
|
NR | LTAS participants significantly more likely to have quit smoking (~25% LTAS; ~9% NCI; ~4% control)
|
No adverse events reported |
Brunette et al. (2019) | NR | 5% attrition at assessment and 2 month follow-up | 100% used intervention, 15% access intervention on phone 85% accessed intervention on computer 11/20 (55%) used NRT | Average sessions accessed 23.6 ( |
NR | Participants were highly satisfied with the Website. 87% of the sessions were rated ⩾3 (4 = ‘I liked it very much’). | Positive feedback was received regarding the amount and content of information and ease of use |
25% reduced their smoking |
NR |
Brunette et al. (2020) | 184 consented 162/184 (88%) were randomised to interventions 11/184 (6%) were ineligible 1 moved 1 lost to FU 1 unable to continue 8 (4%)declined (reasons not stated) | 9.9% (n = 16) at 3 months |
100% | NR | LTAS usability and satisfaction mean summary index scores were significantly higher compared with those assigned to NCI | NR | Most participants (95.38% of LTAS users and 83.1% of NCI education users) reported they were ⩽least satisfied |
Biologically verified abstinence at 6 months:
|
No adverse events were reported during the use of the interventions. |
Ferron et al. (2011) | 6/89 (7.9%) referred did not respond 7/89 (7.9%) was not interested 4/89 (4.5%) DNA research visit 2/89 (2.2%) ineligible due to their reading ability 71/ 89 (89.8%) consented | NR | NR | NR | NR | NR | Barriers to using the first iteration included:
|
NR | NR |
Gowarty et al. (2021a) | 19% (n = 19/98) screened consented
|
0% at visits 1 and 2 | NR | Usage: quitSTART was used on average for more days over 2 weeks [10.8 ( |
Usability (SUS): |
NR | Task completion: QuitGuide completion rates were high at both visits and similar between diagnosis groups |
77% reported attempting quit/reduce smoking during the 2-week trial period. The % of people who attempted to quit/reduce smoking similar for each app (78% QuitGuide v. 75% quitSTART users). 2 quitSTART participants no longer smoked at visit 2 (confirmed with breath carbon monoxide <7 ppm) | NR |
Gowarty et al. (2021b) | 98 screened, 19% (n = 19) screened consented 35% (n = 34) were ineligible 28% (n = 27) declined, no reason 7% refused due to work/childcare commitments 1% moving 10% (n = 10) no phone 2/19 who consented were ineligible due to carbon monoxide levels 17% (n = 17) included in the study | Attrition: 0% at visits 1 and 2 | 100% (17) completed both visits 1 and 2 Usage data available for 88% (15) | Days used: quitSTART was used on average for more days over 2 weeks [10.8 ( |
SUS scores: |
Ease of use |
QuitGuide was easy to use | Similar quitting rates for both groups (78% quitGuide and 75% quitSTART) | NR |
Heffner et al. (2018) | NR | NR | NR | ACT arm, the BD group (M = 13.5, |
NR | NR | NR | 6-month PPA (NS between groups): total 20%; no MHC 22%; BD 14%; AD 18%, 12-month PPA (NS between groups): total 22%; no MHC 25%; BD 16%; AD 18% | NR |
Heffner et al. (2020) | 51/119 (42.9%) enrolled over 24 months 22/119 (18.5%) not eligible 5/119 (4.2%) not interested 11/119 (9.2%) unable to attend 8/119 (6.7%) missed the window 7/119 (5.8%) don't smoke enough 15/119 (12.6%) other reasons | 14/51 (27%) end of treatment 10/51 (20%) at 1-month FU | NR | Days logged in nearly twice as high for Smokefree.gov |
NR | NR | NR | A greater % of WebQuit plus (12%) participants had 7-day PPA at end of treatment compared to Smokefree.gov (8%). 7 day PPA at 1 month FU was 8% for both groups | Depression and mania score slightly improved (NS) for Smokefree.gov |
Klein et al. (2019) | NR | 0% | NR | NR | NR | 83% were enthusiastic about engaging in a social network and liked chatrooms to connect to others. | Personalising app to users' psychosocial needs (i.e. stigma and social isolation) |
NR | NR |
Medenblik et al. (2020) | 35/42 (83.3%) screened were enrolled | NR | Only 2.9% (1) used NRT offered, as they were reluctant to add another prescription to medication regime | NR | NR | NR | NR | iCOMMIT: 9.5% self-reported abstinence at 7 and 30 day FU. |
NR |
Vilardaga et al. (2016) | 5/10 (50%) participated 30% were Ineligible 20% lost at contact | 0% | 100% participants tracked their cigarettes on a daily basis | NR | Average SUS score was 65.5 ( |
NR | • Incremental rewards better than setting larger saving goals. • More focus on the process of cutting down rather than quitting. • QuitPal served as an awareness tool, which needed finer-grained cigarette tracking. • The need for interactive and motivating features. • Multiple layers of app made it difficult for users to interact. | NR | NR |
Vilardaga et al. (2018) | NR | 0% | 100% | NR | SUS learnability (mean = 60) |
NR | • Homescreen confusing • Liked the gamification (quizzes and rewards badges) • Liked the cartoons and storytelling • Liked Acceptance and commitment therapy • Liked simplicity of app • Need for further technical support | NR | NR |
Vilardaga et al. (2019) | 37% (14/38) screened positive for 24% (9/38) were eligible 18% (7/38) completed the study | 22.2% (2//9) 1 due to hospitalisation |
NR | Mean days across participants |
(P1 and P2): LTQ's usability scores were above the standard cut-off (i.e. SUS = 68) in both cases , with QG slightly outperforming LTQ in P1, but largely underperforming LTQ in P2. | Low levels of UE for both participants with each of the app's tracking features (e.g. cigarette use, mood, craving), with a larger average for the QG app. | Bi-phasic AB single-case studies (P5, P6, and P7):
|
NR | NR |
Vilardaga et al. (2020) | 498 potential candidates 11% (55/498) refused 25.7% (128/498) ineligible 31.1% (155/498) unconfirmed 2% (12/498) unsure 14.1% (70/498) lost before randomisation 1% (2) self-withdraw before randomised 18.5% (92/498) in-person screening 12.7% (63/498) randomised | 4 week FU 8% (5/63); 8 week FU 14% (9/63); 12 week FU 9% (6/63); 16 week FU 3% (2/64) | NR |
|
NR | NR | NR | NS difference between 2 groups abstinence (biochemically verified) at FU. |
Small reductions in symptom severity, anxiety, and depression for all. |
Wilson et al. (2019) | NR | Cohort 1: 28.6% (n = 2) withdrew prior to study 20% (1/5) attrition at 3 month FU Cohort 2: 2/13 (15.4%) withdrew prior to study 12.5% (1/8) attrition at 3 month FU | Cohort 1: 40% (n = 2) had high adherence to both therapy sessions and carbon monoxide readings |
NR | NR | NR |
|
Cohort 1:
|
NR |
AEs, adverse events; CPD, cigarettes per day; LTAS, Lets Talk About Smoking; LTQ, learn to quit; MH, mental health; NCI, National Cancer Institute; NS, NOT SIGNIFICANT; PNTS, prefer not to say; PPA, point prevalence abstinence; QG, quit guide; RCT, randomised controlled trial; SAEs, serious adverse events;
Table 5. Key outcomes and findings from studies of digital interventions in SMI for physical activity
Authors, year | Recruitment | Attrition | Adherence to intervention | Usage data | Usability (SUS) | User perspectives and feedback (questionnaires) | User feedback (interview) | Behavioural outcomes | Physical and mental health outcomes |
---|---|---|---|---|---|---|---|---|---|
Aschbrenner et al. (2015) | 10/24 (42%) participated
|
1/10 (10%) due to unrelated health concerns. | Phase 1 – training 100%
|
NR | NR | Participants:
|
|
Stake holders interviews:
|
|
Aschbrenner et al. (2016a) | 13/29 (45%) participated
|
2/13 (15%) at 6 months |
|
NR | NR | NR | NR | NR |
|
Aschbrenner et al. (2016b) | 13/29 (45%) participated
|
15% (N = 2) at 6 months |
|
NR |
|
|
NR |
|
|
Aschbrenner et al. (2021) | NR |
|
PeerFIT:
|
NR | NR | NR | NR | NR | 6 months FU:
|
Campos et al. (2015) | 8/46 not eligible |
3/32 (9.4%) all from intervention group | 69.3% completed the intervention within the proposed timescale. | Average level of games completed was of 45.54 ± 12.18 |
NR | 92% rated the exergames intervention and interactive. |
NR | NR | No significant multivariate effects for combined physical activity and motor function outcomes |
Looijmans et al. (2019) | I: Consented: 140 (82.4%) of the 170 recruited |
I: 43.4% (n = 57) at 6 months; 33.1% (n = 43) at 12 months |
108/140 (77%) completed at ⩾1 lifestyle behaviour screening and made lifestyle plans with lifestyle goals. |
NR | NR | NR | NR | NR | Waist circumference change (after adjusting for antipsychotic medication) between intervention and control were −0.15 cm (NS) and −1.03 cm (NS) after 6 and 12 months, respectively |
Macias et al. (2015) | 11 meet eligibility criteria | 18.2% (2/11) |
98% responded to personalised text messages |
NR |
|
100% of participants were satisfied with the app overall. | Favourite features were text message conversations with peer staff, and peer staff testimonial videos |
|
Slight improvement in physical health self-ratings from pre-test to post-test |
Muralidharan et al. (2018) | NR | WebMOVE: 25% (n = 23) at 3 months; 17.4% (n = 16) at 6 months |
WebMOVE: 18% (n = 17) did not attend |
NR | WebMOVE participants completed slightly more modules. |
NR | Walking was used as the primary source of exercise in WebMOVE and MOVE-SMI. |
WebMOVE: |
NR |
Naslund et al. (2015a, 2015b) | NR | 1/10 (10%) due to mental health | Ten wore the devices for a mean of 89% ( |
NR | NR | Participants reported high satisfaction | (i) Wearable were easy to use |
NR | NR |
Naslund et al. (2016) | Recruitment: 13 were recruited |
23% (n = 3) due to medical or personal reasons | 82% of participants used their Fitbit Zips for over 80% of the days enrolled.
|
NR | Fitbits were worn for an average of 84.7% ( |
Fitbit easy to use, highly satisfied with the device. |
Feedback included: |
NR | NR |
Naslund et al. (2018) | 32 enrolled |
1/32 (3%) stopped due to substance use; 1/32 (3%) pregnancy; 1/32 (3%) hospitalisation; 2/32 (6%) not interested; 2/32 (6%) lost at FU | 76% (n = 19) joined the Facebook group. |
NR | Overall there were 208 interactions with an average of 13.0 ( |
NR | NR | NR | Those who interacted in the Facebook group had clinically significant reduction in cardiovascular risk (⩾5% weight loss) or improved fitness (Facebook group 19.1 ± 20.5; not in group 3.9 ± 6.7) |
Olmos-Ochoa et al. (2019) | 277/1429 (19%) screened were randomised to a group | NR | WebMOVE 35% (about 10.5 sessions) sessions were completed |
NR | NR | NR | Participants were satisfied with their intervention |
NR | NR |
Sylvia et al. (2021) | Contacted (n = 5500) (MoodNetwork) and 4.1% emailed back. |
NR | CBT arm: |
NR | NR | NR | NR | NR | NR |
Young et al. (2017) | 1429 were screened for eligibility
|
Total: 39/276 (14.1%) |
100% completion of intervention component | NR | WebMOVE:
|
NR | WebMOVE: Overall, the intervention was well received. |
NR | WebMOVE change in BMI 34.9 ± 0.43 to 34.1 ± 0.43 |
BMI, body mass index; C, control; CBT, cognitive behavioural therapy; CVD, cardiovascular disorder; I, intervention; lbs, pounds; MCBT, mindfulness cognitive behavioural therapy; MH, mental health; mHealth, mobile health; PA, physical activity; RCT, randomised controlled trial;
Table 6. Key outcomes and findings from studies of digital interventions in SMI for others
Authors, year | Recruitment | Attrition | Adherence to intervention | Usage data | Usability (SUS) | User perspectives and feedback (questionnaires) | User feedback (interview) | Behavioural outcomes | Physical and mental health outcomes |
---|---|---|---|---|---|---|---|---|---|
Hammond et al. (2020) | 95/213 (44.6%) enrolled |
Baseline:
|
TAU + TES: 58% completed >2 modules | On average 5.5 ( |
NR | TES group reported significantly more usefulness for OMI (TAU: M = 7.9 v. |
NR | Similar rates of enrolment between two groups. |
NR |
Melamed et al. (2022) | 192/510 (37.6%) were eligible |
Condition: 4 (5.7%) lost (2 from each)
|
HI group:
|
A>% of HI participants used the e-platform for ⩾10 days compared to those in LI (52% v. 21%) |
NR | NR | NR | No changes in health behaviours over time | NR |
Taylor et al. (2022) | 21/60 (35%) referrals were screened |
2 lost (1.4%)
|
|
|
NR | Acceptability criteria were not met for two participants (1 due to disruption to lives and 1 due to usual daily activities):
|
|
|
|
e-platform, electronic platform; FU, follow-up; HI, high intensity; IQR, inter-quartile range; LI, low intensity; OMI, other mental illness; NR, not reported; NS, not significant;
Results of the included studies were synthesised in their respective classes of health behaviours, namely: (i) smoking; (ii) physical activity, weight loss, and cardio-metabolic health; and (iii) other behaviours. For each HBC class, feasibility, acceptability, usability, and impacts on behaviour/outcomes were summarised.
Eight studies delivered HBC through smartphone apps and nine used web-based interventions. The remaining two were multi-component interventions (Table 1).
Feasibility. Recruitment rates ranged from 13% to 91% across studies (Table 4). Overall, attrition was low, ranging from 0% to 23%; common reasons for dropout included hospitalisation and loss of interest. Adherence to digital interventions was generally high, ranging from 43% to 100% (Table 4).
Acceptability. Apps developed for those with an SMI had greater engagement when compared with apps for the general population (Browne, Halverson, & Vilardaga, 2021; Vilardaga et al., 2019, 2020). One website, Lets Talk About Smoking (LTAS), was developed specifically for individuals with SMI, providing interactive tailored smoking cessation advice. This scored more highly on patient satisfaction when compared to users of static National Cancer Institute (NCI) patient education handout, which was developed for the general population (Brunette et al., 2018).
Participants reported links between symptom severity and smoking, and saw benefits in tracking smoking alongside their mental health (Klein, Lawn, Tsourtos, & van Agteren, 2019). Real-time support, such as a person or distraction task, was deemed essential to help with cravings (Klein et al., 2019).
Usability. People with SMI viewed easy navigation and engaging content and design as preferable, or even essential, for digital HBC (Brunette et al., 2011; Klein et al., 2019; Vilardaga et al., 2016, 2018). Issues were reported with readability, difficulty using support chatrooms, and navigation for certain websites or apps, particularly if pages had multiple functions. Difficulties simultaneously filming and uploading carbon monoxide readings were also reported (Wilson et al., 2019). In Brunette et al.'s (2012) study, four websites developed for the general population were difficult to use among people with SMI who had less experience using computers. Promisingly all participants learnt to use the website developed specifically for SMI populations, regardless of experience, with minimal training (one to three sessions) (Brunette, Ferron, Gottlieb, Devitt, & Rotondi, 2016).
Several apps developed for the general population (QuitGuide, quitSTART, QuitPal) scored below acceptable standards on the ‘System Usability Scale’ in some studies. In Vilardaga et al.'s (2019) study, both QuitGuide and the ‘Learn to Quit’ (LTQ) app – developed for people with SMI – met usability cut-offs as rated by two (all) participants. It is worth noting that when detecting usability problems, studies using samples as small as five can be deemed acceptable (Lewis, 1994).
Behaviour and health outcomes. All 13 studies, which evaluated digital HBCs' impact on smoking behaviours, found self-reported smoking reductions (Table 4). Five studies confirmed smoking abstinence through biochemical verification (Table 4). Unpromisingly, in one intervention, which took a multi-component approach, self-reported 7-day point prevalence abstinence decreased from 38–40% (from both cohorts) to 9.4% in the pilot RCT (Medenblik et al., 2020; Wilson et al., 2019).
Notably, Brunette et al. (2018) found, after 3 months, greater percentage of LTAS participants had biologically verified abstinence, compared to the NCI education group. Additionally another app developed for people with SMI (LTQ) was more effective in promoting smoking cessation, with those assigned to the QuitGuide app (developed for the general population) making more quit attempts and subsequently more relapses (Browne et al., 2021; Vilardaga et al., 2020).
Only two studies measured mental health outcomes. Heffner et al. (2020) reported a potential improvement in depression and mania scores, while Vilardaga et al. (2020) demonstrated a reduction in negative symptoms and a small non-significant reduction in depression, anxiety, and symptom severity across both groups.
One study delivered HBC through a smartphone app (WellWave), two studies used a web-based intervention and one study used an app with an associated wearable device. Seven studies used a multi-component approach (Table 2). Three studies compared web-based interventions with in-person interventions (Muralidharan et al., 2018; Olmos-Ochoa et al., 2019; Young et al., 2017).
Feasibility. Recruitment rates varied across studies, with studies recruiting from veteran centres reporting lower rates (19% participated and 58% were ineligible) than studies recruiting from mental healthcare services (42–45% participated and 28% were ineligible; Table 5). The highest recruitment rates were observed in studies conducted through outpatient clinics with subsequent participation not involving additional in-person sessions (Campos et al., 2015; Looijmans, Jörg, Bruggeman, Schoevers, & Corpeleijn, 2019).
Where reported, overall retention was high, with 75–90% of participant completing follow-up measures at the final time point, which ranged from 1 to 6 months (Table 5). Retention rates dropped after 12 months, to around 33% (Aschbrenner et al., 2021; Looijmans et al., 2019). Reasons for dropout included health concerns, hospitalisation, and competing time commitment.
Levels of adherence were generally high, particularly with digital components of the studies (Table 5). In one study of 32 participants, the in-person exercise sessions achieved only a 28% attendance rate, while 100% and 76% used the provided Fitbit and private Facebook group, respectively (Aschbrenner, Naslund, Shevenell, Kinney, & Bartels, 2016a). Similar findings were reported in Aschbrenner et al.'s (2021) study, with 70% of PeerFit participants attending at least one in-person exercise session, while 97% of BEAT participants attended at least one online coaching session (Aschbrenner et al., 2021).
Acceptability. Usage of digital interventions was also generally high (Table 5); in Muralidharan et al. (2018), Olmos-Ochoa et al. (2019), and Young et al. (2017), more modules were completed by those in the digital v. the in-person arm.
Feedback indicated that peer interaction, particularly interacting with peer coaches and learning about others experiences, seemed to be a popular component of interventions among patients. Conversely, the main barriers to use were physical limitations and pain and, when attending in-person sessions, time constraints and travel burden (Aschbrenner et al., 2015; Muralidharan et al., 2018; Olmos-Ochoa et al., 2019). Some participants attending in-person sessions also found it difficult to engage with new people (Olmos-Ochoa et al., 2019; Young et al., 2017). Other less commonly reported barriers included concerns about their environment and safety (for in-person interventions), financial barriers, control over food preparation, and lack of support from others (Olmos-Ochoa et al., 2019). Concerning wearables, participants found them helpful for setting goals, motivation, and useful for self-monitoring (Aschbrenner et al., 2015; Naslund, Aschbrenner, Barre, & Bartels, 2015a; Naslund, Aschbrenner, & Bartels, 2016). Some participants did experience frustration due to forgetting to wear pedometers (Young et al., 2017) and the cost of wearables was identified as a barrier in one study (Naslund et al., 2015a, 2015b).
Usability. None of the studies measured usability using formally validated scales, complicating evaluation. Some participants did comment that they found wearables easy to use (Naslund et al., 2015a, 2015b). Notably some participants did experience technical issues when using equipment (Young et al., 2017) or logging into digital interfaces for the first time. Peer coaches were noted as helpful in combatting such issues. Of note, in an intervention that involved participants playing physically active video games, 69% of participants completed the intervention using Kinect, although 85% reported would not have done so without technical support (Campos et al., 2015). Thus suggesting that without support this is not acceptable for those with an SMI and technical support would be required for real-world implementation in mental health settings (Campos et al., 2015).
Behaviour and health outcomes. Nine studies assessed the impact of the digital interventions on physical activity and/or weight loss (Table 5). Five studies showed at least some promising results, with two studies in particular reporting participants lost at least 5% of their body weight and clinically significant reductions in cardiovascular risk (⩾5% weight loss or improved fitness) (Aschbrenner et al., 2016a, 2021). Additionally Aschbrenner et al. (2016a, 2016b) reported 17% of participants showed clinical significant improvements in cardiovascular fitness. Two interventions lead to increases in physical activity (Macias et al., 2015; Muralidharan et al., 2018). Only one study, which used a web tool designed to help patients set goals, monitor their progress, and receive feedback via a mental health nurse (Looijmans et al., 2019) found no significant reductions in body mass index (BMI) or waist circumference at 6/12-month follow-ups.
Papers investigating a digital intervention called ‘WebMOVE’ (Muralidharan et al., 2018; Young et al., 2017) reported more weight loss than the in-person comparator intervention (MOVE-SMI). Both provided pedometers and access to peer coaches and comprised of the same educational content, differing only in delivery mode. However, in another study, both individual mHealth coaching and in-person HBC were similarly effective; with both groups achieving clinically significant weight loss and reduction in cardiovascular risk at 6 and 12 months (Aschbrenner et al., 2021).
Only one study looked at the effect of digital HBC on mental health (Campos et al., 2015) and found slight, non-significant improvements in these domains.
Two studies delivered HBC through web-based interventions, with one promoting the treatment for substance use disorder (SUD) (Hammond et al., 2020) and the other changing attitudes towards health behaviours as a route to behavioural changes (Melamed et al., 2022). One study targeted sleep, which used an app (Taylor et al., 2022).
Feasibility, acceptability, effectiveness, and outcomes of other interventions. Across the three studies, recruitment appeared to be challenging, with issues around screening and ineligibility (Table 6). Retention for the primary endpoint of theses interventions was excellent, ranging from 93% to 97%, though longer term follow-up (24 weeks) dropped to 40% in one study (Melamed et al., 2022).
Adherence to interventions varied across the three studies (58–100%). Adherence was highest for the sleep intervention. Overall participants had positive experiences, but many felt the 6-week intervention was not long enough and needed more variety of content and games (Taylor et al., 2022).
The app-based sleep intervention had a large effect on behaviour (sleep) and a small-to-medium effect on mental health (Taylor et al., 2022). The attitude-focused intervention led to positive changes in individual attitudes but did not ultimately change behaviours (Melamed et al., 2022).
The SUD intervention was rated highly across several measures, including acceptability (Hammond et al., 2020). At the end of the web intervention period, similar rates of participants had enrolled in SUD treatment, at 30 days post discharge, as that observed under treatment-as-usual conditions.
This paper reviewed 36 studies and systematically identified 29 digital HBC (with overlap of components for some of the physical activity interventions) for people with SMI. Feasibility, acceptability, and outcomes of interventions were evaluated and intervention components and strategies which were preferred by people with SMI were identified. Overall, 70% of the studies established support for the acceptability and/or feasibility of digital behavioural change interventions. However, themes around the need for human support for both digital literacy/navigation and engagement were common across all clinical targets. Given the pilot nature of studies and the heterogeneous outcomes, it is not possible to determine an effect size estimate, but current evidence shows that these interventions do have the potential to change health behaviours.
Across the studies reviewed, there was a relatively consistent result that digital interventions to change behaviours are both feasible and acceptable for use among people with SMI. This is an important finding, due to the large health disparity among this group and insufficient resources in mental healthcare settings to provide lifestyle interventions in mental healthcare settings (Firth et al., 2019). Despite concerns about smartphone use as a main barrier to digital interventions in this population, the majority of participants with SMI reported digital interventions were easy to use and several studies even reported participants completed additional modules or sessions voluntarily (Aschbrenner, Naslund, Shevenell, Mueser, & Bartels, 2016b; Aschbrenner et al., 2016a; Brunette et al., 2016, 2018). It is important to note some participants did struggle with accessibility, internet access, and/or needed additional support, in particular for those with limited experience using technology (Campos et al., 2015; Ferron et al., 2011; Naslund et al., 2016; Olmos-Ochoa et al., 2019; Taylor et al., 2022; Vilardaga et al., 2016, 2019; Young et al., 2017). Promisingly those with little experience using digital platforms could use them after assistance from peer coaches (Olmos-Ochoa et al., 2019; Young et al., 2017) or training sessions (Brunette et al., 2016). Therefore, to reduce the digital divide in future, it would be crucial to have human support available in mental healthcare settings to facilitate use, such as digital navigators (Sylvia et al., 2021; Wisniewski, Gorrindo, Rauseo-Ricupero, Hilty, & Torous, 2020; Wisniewski & Torous, 2020).
It is important to mention participants recruited from veteran centres and inpatient settings had high rates of ineligible participants, which may limit the generalisability of the results to other patient populations and settings. Further, some of the digital interventions (such as Microsoft, iCOMMIT, and LTQ) are not publicly readily available to use. Two interventions financially compensated participants for ongoing engagement, which may not be sustainable in real-world healthcare services (Linardon & Fuller-Tyszkiewicz, 2020).
Compared to in-person interventions, digital HBC had benefits such as greater adherence, lower resource intensity, and the potential for non-clinical staff to deliver them (Aschbrenner et al., 2021). Further, the outcomes/changes in behaviour from digital interventions seemed similar to in-person interventions of the same content (Muralidharan et al., 2018; Olmos-Ochoa et al., 2019; Young et al., 2017). Such findings are promising given the lack of capacity in mental healthcare services for in-person HBC (Ayerbe et al., 2018; Bailey et al., 2019). However, future work is required to compare the effectiveness of delivering an intervention digitally v. non-digitally to people with SMI.
Therefore, digital HBC are poised to play a crucial role in the near future. Digital interventions can also increase engagement and overcome socioeconomic and barrier issues reported by participants regarding the in-person elements of the multi-component interventions.
Peer/social support – offline and online – was perceived positively among many of the physical activity interventions (Aschbrenner et al., 2015; Macias et al., 2015; Muralidharan et al., 2018; Young et al., 2017) and, from the interviews, social support was a strongly desired element for smoking cessation apps (Gowarty, Aschbrenner, & Brunette, 2021a; Gowarty et al., 2021b; Klein et al., 2019). Also, design features and content that made platforms more interactive, usable, and tailored to those with SMI enhanced engagement (Aschbrenner et al., 2016b; Browne et al., 2021; Brunette et al., 2016, 2018, 2020; Klein et al., 2019; Naslund et al., 2015a, 2015b; Taylor et al., 2022; Vilardaga et al., 2020).
With regards to behavioural change techniques, it appears that setting goals and reviewing progress may not be enough to change behaviour for people with SMI. Setting diet and physical activity goals, behavioural monitoring, and receiving feedback from health professionals failed to reduce BMI or waist circumference at 6/12-month follow-ups in one study (Looijmans et al., 2019). In contrast, interventions that involved exercise sessions, information about preparing healthy meals, wearables, provided rewards/trophies or had social support, led to weight loss for the majority of participants (Aschbrenner et al., 2016a, 2016b, 2021; Muralidharan et al., 2018; Naslund, Aschbrenner, Marsch, McHugo, & Bartels, 2018; Young et al., 2017). Previous research has shown that demonstrating exercises at home yielded large impacts on physical activity in low-income groups (Bull et al., 2018). Further research would be required to determine the feasibility and acceptability of digital home workouts in people with SMI.
People with SMI appear more amenable to HBC tailored to consider their needs. This review highlighted examples where digital HBC developed for those with SMI were found to have superior outcomes, including higher rates of smoking abstinence and/or greater reduction in cigarettes smoked (Browne et al., 2021; Brunette et al., 2018), fewer relapses (Vilardaga et al., 2020), and enhanced usability (Brunette et al., 2020). For example, with smoking interventions, tailoring could mean normalising relapses and integrating their mental health symptomology, while with physical activity interventions considering the physical limitations of people with SMI could be important (Aschbrenner et al., 2015; Klein et al., 2019; Muralidharan et al., 2018; Olmos-Ochoa et al., 2019).
A strength of this study was the comprehensive nature of the methods, which applied a systematic approach and broad search terms to capturing digital HBC for people with SMI, including various study designs. Although only one reviewer was responsible for an initial screening at title and abstract stage, this was done only to remove the obviously ineligible articles swiftly (i.e. those in which no part of the title or abstract indicated relevance to this review). Any study with an indication of eligibility from title/abstract content was subject to full-text screening, conducted by two reviewers. Given the auxiliary search methods conducted alongside the main search, we are confident this review captures the relevant published literature on this nascent but growing topic. However, a key limitation is that due to the preliminary nature of most studies conducted so far (which were largely focused on feasibility, or pilot studies with small-sample sizes consisting of mostly, if not all, Caucasian participants), the research may be too nascent at present to draw any definitive conclusions on the effectiveness of digital approaches for health promotion in SMI. Additionally due to the short-term follow-up of most studies (<4 months), the degree of engagement with digital interventions over longer durations is unknown.
While this review was able to summarise the acceptability/feasibility of digital HBC from a range of different metrics in SMI samples, a further limitation is that many of the included studies were conducted in the United States, which may limit generalisability to other healthcare systems. Furthermore, most studies only recruited participants who had the ability and/or interest in using digital technologies, making it hard to determine the actual feasibility of such approaches, across the entire clinical populations of those treated for SMI (i.e. beyond those individuals who are eligible and willing to join the reviewed studies, to begin with).
Current results suggest digital HBC overall are acceptable and useful for people with SMI, but some individuals may need extra support with technology. The effectiveness of these interventions has yet to be fully established. Nonetheless, there are many provisional findings of digital technologies can result in positive HBC among people with SMI, and even better engagement when compared with some in-person intervention components. To ensure accessibility and usability, the design process of digital interventions should aim to involve people with SMI throughout. Future research should also examine the cost-effectiveness and implementation of digital HBC for promoting health behaviours into real-world clinical settings and healthcare systems for people with SMI.
Corresponding author: Joseph Firth; Email: [email protected]
You have requested "on-the-fly" machine translation of selected content from our databases. This functionality is provided solely for your convenience and is in no way intended to replace human translation. Show full disclaimer
Neither ProQuest nor its licensors make any representations or warranties with respect to the translations. The translations are automatically generated "AS IS" and "AS AVAILABLE" and are not retained in our systems. PROQUEST AND ITS LICENSORS SPECIFICALLY DISCLAIM ANY AND ALL EXPRESS OR IMPLIED WARRANTIES, INCLUDING WITHOUT LIMITATION, ANY WARRANTIES FOR AVAILABILITY, ACCURACY, TIMELINESS, COMPLETENESS, NON-INFRINGMENT, MERCHANTABILITY OR FITNESS FOR A PARTICULAR PURPOSE. Your use of the translations is subject to all use restrictions contained in your Electronic Products License Agreement and by using the translation functionality you agree to forgo any and all claims against ProQuest or its licensors for your use of the translation functionality and any output derived there from. Hide full disclaimer
Copyright © The Author(s), 2023. Published by Cambridge University Press. This work is licensed under the Creative Commons Attribution License This is an Open Access article, distributed under the terms of the Creative Commons Attribution licence (http://creativecommons.org/licenses/by/4.0/), which permits unrestricted re-use, distribution and reproduction, provided the original article is properly cited. (the “License”). Notwithstanding the ProQuest Terms and Conditions, you may use this content in accordance with the terms of the License.
Abstract
The use of digital technologies as a method of delivering health behaviour change (HBC) interventions is rapidly increasing across the general population. However, the role in severe mental illness (SMI) remains overlooked. In this study, we aimed to systematically identify and evaluate all of the existing evidence around digital HBC interventions in people with an SMI. A systematic search of online electronic databases was conducted. Data on adherence, feasibility, and outcomes of studies on digital HBC interventions in SMI were extracted. Our combined search identified 2196 titles and abstracts, of which 1934 remained after removing duplicates. Full-text screening was performed for 107 articles, leaving 36 studies to be included. From these, 14 focused on physical activity and/or cardio-metabolic health, 19 focused on smoking cessation, and three concerned other health behaviours. The outcomes measured varied considerably across studies. Although over 90% of studies measuring behavioural changes reported positive changes in behaviour/attitudes, there were too few studies collecting data on mental health to determine effects on psychiatric outcomes. Digital HBC interventions are acceptable to people with an SMI, and could present a promising option for addressing behavioural health in these populations. Feedback indicated that additional human support may be useful for promoting adherence/engagement, and the content of such interventions may benefit from more tailoring to specific needs. While the literature does not yet allow for conclusions regarding efficacy for mental health, the available evidence to date does support their potential to change behaviour across various domains.
You have requested "on-the-fly" machine translation of selected content from our databases. This functionality is provided solely for your convenience and is in no way intended to replace human translation. Show full disclaimer
Neither ProQuest nor its licensors make any representations or warranties with respect to the translations. The translations are automatically generated "AS IS" and "AS AVAILABLE" and are not retained in our systems. PROQUEST AND ITS LICENSORS SPECIFICALLY DISCLAIM ANY AND ALL EXPRESS OR IMPLIED WARRANTIES, INCLUDING WITHOUT LIMITATION, ANY WARRANTIES FOR AVAILABILITY, ACCURACY, TIMELINESS, COMPLETENESS, NON-INFRINGMENT, MERCHANTABILITY OR FITNESS FOR A PARTICULAR PURPOSE. Your use of the translations is subject to all use restrictions contained in your Electronic Products License Agreement and by using the translation functionality you agree to forgo any and all claims against ProQuest or its licensors for your use of the translation functionality and any output derived there from. Hide full disclaimer
Details



1 Division of Psychology and Mental Health, The University of Manchester, Manchester Academic Health Science Centre, Manchester M13 9PL, UK
2 School of Population Health, University of New South Wales, Randwick, NSW 2052, Australia; Discipline of Psychiatry and Mental Health, University of New South Wales, Randwick, NSW 2052, Australia
3 Department of Psychiatry, Harvard Medical School, Boston, MA, USA; Department of Psychiatry, Massachusetts General/Mclean Hospital, Boston, MA, USA
4 Department of Child and Adolescent Psychiatry, NYU Grossman School of Medicine, New York, NY, USA
5 Hospital del Mar Research Institute, Institut de Salut Mental, Hospital del Mar, Barcelona, Spain; Centro de Investigacion Biomedica en Red de Salud Mental (CIBERSAM), Spain; Department of Psychiatry, the Donald and Barbara Zucker School of Medicine at Hofstra/Northwell, New York, USA
6 Department of Psychiatry, Institute for Behavioral Science, Feinstein Institutes for Medical Research, Manhasset, NY, USA; Zucker School of Medicine at Northwell/Hofstra, New York, NY, USA; Department of Psychiatry, Beth Israel Deaconess Hospital, Boston, MA, USA
7 Division of Psychology and Mental Health, The University of Manchester, Manchester Academic Health Science Centre, Manchester M13 9PL, UK; Greater Manchester Mental Health NHS Foundation Trust, Manchester Academic Health Science Centre, Manchester M13 9PL, UK