Depressive and anxiety disorders are among the leading causes of global health burden, with negative impacts on wellbeing, quality of life, physical health, and social, family, and work functioning (Ballenger, 2000). However, only around forty percent of men with current depression or anxiety receive any mental health care (i.e., medication or psychotherapy) from a healthcare practitioner (Blumberg et al., 2015; Terhaag et al., 2020). A variety of factors may help to explain this. Men with depression or anxiety may present as angry or irritable, or engage in risky behaviours rather than reporting better-recognised symptoms such as sadness or worry. These atypical symptoms may not be recognised by men themselves or by healthcare practitioners as indicators of a common mental health issue (Fisher et al., 2021; Rice et al., 2015). Holding traditional masculine beliefs (e.g., stoicism and self-reliance) may also negatively influence help-seeking (Ogeuji & Okoloba, 2022; Seidler et al., 2016), as men may also be less likely to engage with assistance due to conflicts between these norms and key aspects of many therapies (e.g., self-disclosure and vulnerability; Oliffe et al., 2019). Some men feel they do not have the skills needed to speak to a healthcare provider (Milner et al., 2019), and men’s often low mental health literacy may also reduce their likelihood of accessing assistance (Oliffe et al., 2016). Cultural norms, expectations, and stigma related to masculinity, mental health, and help-seeking (Lindinger-Sternart, 2015), and systemic obstacles including differential access to health care, high costs, long wait lists, appointments clashing with other responsibilities, and lack of local services have also been shown to influence men’s mental health care access (Swami et al., 2020).

The ways in which men often prefer to seek mental health assistance appear to sidestep or reduce many of these identified barriers. Men are less likely than women to endorse seeking formal support for depression (Wendt & Shafer, 2016) and to get mental health information from a doctor, but are more likely to seek information about depression on the internet (Beyondblue, 2016). While men tend to consult doctors when a health issue is more advanced and serious, rather than in the earlier stages or for preventive care (Mursa et al., 2022), they have also been found to actively engage in health self-monitoring and informal help-seeking before seeing a doctor (Smith et al., 2008). Given these preferences for privacy, autonomy, self-monitoring, and initial informal help-seeking, e-mental health (eMH) approaches ("mental health services and information delivered or enhanced through the internet and related technologies"; Christensen et al., 2002) have received attention for their potential to provide mental health care in a way that suits men’s needs and preferences. EMH may be delivered through a website or smartphone or tablet application (‘app’), and can comprise fully self-guided interventions or incorporate assistance via other modalities (e.g., online peer discussion, emailed support from a clinician). Psychotherapeutic eMH programs are commonly based on evidence-based treatments such as cognitive-behavioural therapy (CBT; Cuijpers et al., 2010). Strengths of eMH include the often-ubiquitous nature of smart devices, ability to access assistance without delay, anonymously, and when convenient, affordability, promotion of real-time self-monitoring, and potential for intervention customisation to the individual user (Borghouts et al., 2021; Christensen & Hickie, 2010; Lal & Adair, 2014). Conversely, potential difficulties include a need for technical proficiency and internet/device access, privacy concerns, usage challenges for individuals with health limitations (including mental health-related ones), and suggestions that anonymity may reduce engagement (Borghouts et al., 2021; Lal & Adair, 2014).

In mixed sex/gender samples, eMH interventions appear effective. An umbrella review (systematic review of reviews) of data from nine meta-analytic reviews of 166 studies on internet-delivered CBT for adults showed moderate to large positive effects on panic disorder, social anxiety disorder, generalised anxiety, posttraumatic stress, and major depression (Andersson et al., 2019). However, outcomes may vary according to intervention, trial, and participant characteristics (Firth et al., 2017a, 2017b; Firth et al., 2017a, 2017b; Karyotaki et al., 2018; Renfrew et al., 2020) and for users who are men, may also be influenced by whether an intervention is ‘gender sensitive’ (that is, whether it has been designed for the specific needs, preferences, and realities that men consider important; World Health Organisation, 2007). To date, the effects of using eMH on men’s mental health, and factors influencing any effects, have not been systematically examined.

Objectives

This review aims to (1) synthesise the characteristics of eMH interventions for depression or anxiety that have been trialled and evaluated by men, and (2) synthesise and meta-analyse the effects of eMH interventions on depression and anxiety outcomes in men, including examination of the effects of participant, intervention, and study characteristics.

Methods

A systematic review and meta-analysis were conducted based on an a priori-registered PROSPERO protocol (CRD42020212039).

Eligibility criteria

We sought peer-reviewed, published, English language studies that presented full-text, primary data on the effectiveness of a psychotherapeutic eMH intervention targeting depression or anxiety in men. Studies must have utilised a full- or quasi-experimental study design to detect the effect of the intervention, and must have presented quantitative outcome data on the effect of the intervention.

Population

We sought data on individuals identified as male or a man (as per their reported sex or gender), aged 18 + years, who had participated in studies investigating the effects of one or more eMH interventions for depression or anxiety. Where a study included men aged both 18 + and < 18 years of age, we included that study only if (a) the mean participant men age was 18 + years and (b) less than one-third of men in the sample were aged < 18 years. We excluded studies that included men < 18 years if they did not fit these criteria or where we could not make a reasonable determination due to missing age data. Studies including participants of multiple sexes or genders (e.g., men and women) were eligible if sex- or gender-disaggregated data for male/men participants were reported separately and in sufficient detail to analyse.

We included studies comprising (a) non-clinical populations (not recruited based on a specific health or wellbeing issue) and (b) populations with high prevalence physical comorbidities or harmful health behaviours such as obesity, Type 2 diabetes mellitus, smoking, or physical inactivity. We sought to exclude medical or psychosocial conditions that did not fit these criteria (exclusions based on this criterion included samples comprising individuals with multiple sclerosis, traumatic brain injury, dementia, and schizophrenia). Studies of participants with both included and excluded conditions were eligible if data for participants with only included conditions could be extracted separately.

Interventions

We sought quantitative data on the effects of psychotherapeutic eMH interventions primarily administered via an app or website. The intervention had to be specifically designed to prevent or treat a depressive or anxiety disorder or a key symptom of depression or anxiety. Depressive and anxiety disorders were classified in accordance with the Depressive and Anxiety Disorders chapters of the Diagnostic and Statistical Manual of Mental Disorders-5 (DSM-5; American Psychiatric Association, 2013). In accordance with these chapters, includable disorders included major depressive disorder, persistent depressive disorder (dysthymia), generalised anxiety disorder, specific phobia, panic disorder, social anxiety disorder (social phobia), agoraphobia, and unspecified anxiety and depressive disorders. We also included interventions that aimed to prevent or treat a key symptom of an anxiety or depressive disorder, rather than an actual disorder. As per the DSM-5 diagnostic criteria for depressive and anxiety disorders, for depression, we included interventions targeting depressed or low mood, while for anxiety disorders, we included those that targeted anxiety, worry, or fear (American Psychiatric Association, 2013). We excluded interventions designed to prevent or treatment any disorders or symptoms not fitting these criteria.

Psychotherapeutic eMH interventions are designed to “create positive cognitive, behavioural, and emotional change” (Barak et al., 2009). Their content is generally informed by evidence-based psychotherapy and presented in structured, interactive modules. They may incorporate psychoeducation, symptom and goals monitoring, provide the user with feedback, and take the user through steps to make cognitive and/or behavioural changes (Barak et al., 2009). We included studies that examined any intervention fitting this definition. We excluded those that examined an intervention that (a) was primarily about exercise, neuropsychological interventions for improving cognition/memory/attention, or that was otherwise non-psychotherapeutically focused, or (b) included only psychoeducation/was not interactive, or (c) was not primarily self-administered, or (d) used a lab-based trial.

Comparison

Studies with any or no comparator were eligible. A comparison group not subjected to the psychotherapeutic eMH intervention was not required.

Outcomes

Outcomes included characteristics of the psychotherapeutic eMH interventions for depression or anxiety that had been trialled and evaluated in men, including features of the conducted studies and their participant men. Additional outcomes were the effects of psychotherapeutic eMH interventions on depression and anxiety outcomes, including the impacts of participant, intervention, and characteristics on those effects.

Data sources and search strategy

We followed the Preferred Reporting Items for Systematic Reviews and Meta-Analyses (PRISMA) guidelines (Moher et al., 2009) in the conduct and reporting of this study (Supplementary Table S1). We conducted systematic searches for peer-reviewed, English language, published literature in six electronic databases (Cochrane Library, PubMed, PsycINFO, CINAHL, Web of Science, and Embase). Our search strategy was developed based on the PICO criteria previously presented, adapted for each database with input from an experienced research librarian (Supplementary Table S2). Database-specific indexing, thesaurus terms, and free text expressions were used. With online mental health programs beginning in the early 2000s (Andersson, 2018), we searched for studies published between 1 January 2000 and the search date (8 October 2020). We also manually searched reference lists of articles meeting inclusion criteria and relevant systematic reviews and conducted citation searching on eligible studies to identify potentially eligible studies not captured in the initial searches.

Screening and full text review

Initial search results were merged, and duplicates were automatically removed and then manually checked in Endnote X9 (2013, Clarivate, Philadelphia). The de-duplicated results were then uploaded to Covidence (Veritas Health Innovation, Melbourne) for title and abstract screening, full text review, data extraction, and quality assessment. MJO screened all abstracts and full-text papers. Inter-rater agreement with co-screeners across 10% of total titles and abstracts was 94% (GW) and 95% (DT), with disagreements resolved by discussion. Agreement with MO across 5% of full-text papers was unanimous.

Data extraction

A data extraction form was developed and piloted on two included studies with no changes required. Data extracted from each article included (i) study design and characteristics, (ii) intervention characteristics, (iii) participant characteristics, and (iv) depression and anxiety outcomes. MJO extracted the data, which was checked by HG, with discrepancies resolved by discussion. We attempted to contact authors from four studies to clarify extracted data; two responded with the requested information (Fogarty et al., 2017; Geraedts et al., 2014). The other two studies were excluded because we did not have data necessary for their inclusion and did not receive a response to our requests for this data.

Quality assessment

Study quality was assessed using the QualSyst checklist for quantitative studies (Kmet et al., 2004). This checklist comprises 14 items related to study reporting and method (e.g., “Some estimate of variance is reported for the main results?”), with possible ratings of Yes (2), Partial (1), No (0), and where relevant, Not Applicable. A summary score is calculated by dividing the total by the possible total. We pre-specified a “relatively liberal” (Kmet et al., 2004) minimum summary score of 0.55 for inclusion. Quality assessments were conducted independently by MJO and HG, with disagreements resolved by discussion.

Data synthesis and meta-analysis

Extracted data were narratively synthesised before meta-analyses were conducted in Comprehensive Meta-Analysis (CMA; Biostat Inc., Engelwood, NJ). Data were input into CMA as published or as provided by study authors. Where a study reported eligible data on two measures for the same condition (e.g., two depression measures), we used the primary measure if specified, or the most common measure across included studies if not.

We used Hedges’ g as the effect size estimate to enable correction for smaller samples. Interpretation of effect sizes was based on guides of 0.2, 0.5, and 0.8 representing small, medium, and large effect sizes, respectively (Cohen, 1988). As studies did not routinely report the correlation between their pre-post outcomes, we imputed a conservative estimate of r = 0.70. Random effects models were used throughout to account for the likelihood that effects may have differed between studies due to intervention, study, sampling, or other differences. Effects were interpreted as statistically significant where p < 0.05.

We pre-planned to analyse depression and anxiety outcome data separately and followed Cochrane guidance that “two studies is a sufficient number to perform a meta-analysis, provided that those two studies can be meaningfully pooled and provided their results are sufficiently ‘similar’” (Ryan & Cochrane Consumers and Communication Review Group, 2016). As such, we could conduct meta-analyses only of pre-post data from depression treatment studies (Nstudies = 2). The small number of included studies and/or their variance meant that we could not conduct planned analyses of (a) the overall effects of interventions on depression and anxiety, (b) between-group RCT data (e.g., intervention vs. control), (c) the effects of prevention interventions, and (d) the impacts of other participant, study, and intervention characteristics on outcomes.

Between-study heterogeneity was evaluated using I2, which estimates the percentage of variation across studies due to heterogeneity, and Cochran’s Q, which evaluates variation across studies versus within-subjects in individual studies. A significant Q or an I2 of ≥ 75% indicated considerable between-study heterogeneity. Given the low number of studies in our analysis, we did not inspect funnel plots or associated tests of significance to assess asymmetry. Instead, potential for publication bias was assessed by calculating Orwin’s fail-safe N (Nfs; Orwin, 1983) for each effect. Nfs determines how many missing studies would be needed to bring the overall effect to below the lowest possible level of practical importance (set at g < 0.2 here). Nfs was considered acceptable where it was greater than the number of studies in that effect size analysis (Nfs > Nstudies). We considered a finding to be meaningful if the effect size (g) was ≥ 0.2 (at least a small effect), the finding was statistically significant (p < 0.05), and the fail-safe N was acceptable (Nfs > Nstudies).

Reporting of participant gender/sex

Sex (the biological and physiological characteristics of males, females, and intersex persons) and gender (the socially-constructed roles, behaviours, and norms of men, women, and individuals identifying as other genders; World Health Organisation, 2022) are distinct concepts, and both are important determinants of health and wellbeing (Heidari et al., 2016). However, these concepts are not always easily separable, and combining data from participants described as men and male adds complexities for respectful, precise reporting. We have attempted to follow Sex and Gender Equity in Research (SAGER) guidance (Heidari et al., 2016) that data should be disaggregated by sex and gender wherever possible. As such, when referring to individual primary sources, we have endeavoured to use the participant descriptor (male/males or man/men) reported by the original researchers. However, for ease of reading, where we cite multiple studies and in our synthesis, analysis, and discussion we have used gender descriptors (man/men) regardless of whether participants were described as men or males in primary data.

Results

Study selection

Searches identified a total of 14,330 records, of which 4772 were duplicates. A total of 9557 non-duplicate references were screened at title and abstract level, with 9077 excluded at this point. Of the 482 reports retrieved for full-text assessment, 475 were excluded; seven eligible articles progressed to quality assessment (Fig. 1).

Fig. 1
figure 1

PRISMA 2020 Flow Chart

Quality assessments

Quality scores for the seven evaluated studies were high (Table 1; Supplementary Table S3). With a lowest total of 0.85 (Botella et al., 2016), no studies were excluded due to a sub-threshold total and all seven studies were included. The most common potential biases were group selection methods or information sources not being described or appropriate (Botella et al., 2016; Rice et al., 2020) and studies not having adequate controls for confounding (Davidson et al., 2020; Deady et al., 2018; Fogarty et al., 2017; Geraedts et al., 2014; Nolte et al., 2021; Rice et al., 2020). Nine quality criteria, including use of appropriate study designs and analytic methods, reporting an estimate of variance, and conclusions that were supported by the results, were fulfilled by all included studies (Fig. 2).

Table 1 Study Characteristics
Fig. 2
figure 2

Aggregated Study Quality Assessment Item Ratings

Study characteristics

Table 1 details the characteristics of each included study. All were published from 2014 onwards and conducted in high income, ‘Western’ countries: Australia (Davidson et al., 2020; Deady et al., 2018; Fogarty et al., 2017; Rice et al., 2020), Spain (Botella et al., 2016), Netherlands (Geraedts et al., 2014), and Germany (Nolte et al., 2021).

Three studies were RCTs (Botella et al., 2016; Geraedts et al., 2014; Nolte et al., 2021) and four used single arm pre-post designs (Davidson et al., 2020; Deady et al., 2018; Fogarty et al., 2017; Rice et al., 2020). Five targeted depression (Botella et al., 2016; Deady et al., 2018; Fogarty et al., 2017; Geraedts et al., 2014; Nolte et al., 2021), one targeted depression and anxiety (Davidson et al., 2020), and the other targeted social anxiety (Rice et al., 2020). Three focused on work populations: unemployed men (Botella et al., 2016), employees in male-dominated industries (Deady et al., 2018), and taxi drivers (Davidson et al., 2020). Three targeted people with particular levels of symptoms: men with at least mild depression (Fogarty et al., 2017), individuals with mild to moderate depressive symptoms (Nolte et al., 2021), and young people with social anxiety (Rice et al., 2020). One targeted both workplace and symptom levels, investigating current employees with depression (Geraedts et al., 2014). Two studies were centred on prevention (Botella et al., 2016; Davidson et al., 2020), while five focused on treatment/symptom reduction (Deady et al., 2018; Fogarty et al., 2017; Geraedts et al., 2014; Nolte et al., 2021; Rice et al., 2020). Participant recruitment occurred in clinical (Nolte et al., 2021; Rice et al., 2020), community (Botella et al., 2016; Davidson et al., 2020; Deady et al., 2018; Fogarty et al., 2017; Nolte et al., 2021), and workplace settings (Davidson et al., 2020; Deady et al., 2018; Geraedts et al., 2014).

EMH intervention characteristics

The characteristics of each eMH intervention are reported in Table 2. The interventions’ psychotherapeutic components were almost exclusively centred on CBT or cognitive therapy (CT) techniques but varied widely in their content, length, and format. For example, Deady et al.’s (2018) psychotherapeutic component was one short task per day for 30 days. Geraedts et al. (2014) used six weekly sequential sessions, and Davidson et al. (2020) comprised 31 short activities for use in any order. In Nolte et al. (2021), users completed 12 weekly 10–60 min sequential sessions organised as a simulated interactive dialogue. Fogarty et al.’s (2017) intervention comprised three personalised weekly sequential interactive sessions. In Rice et al. (2020), the therapeutic content was delivered via bespoke interactive online comics. Additional aspects of some interventions included between-session homework tasks (Geraedts et al., 2014; Rice et al., 2020), mood or activity self-assessment/monitoring (Botella et al., 2016; Davidson et al., 2020; Deady et al., 2018; Fogarty et al., 2017), details of further supports (Davidson et al., 2020; Deady et al., 2018), and the ability to receive reminders via SMS or email (Fogarty et al., 2017; Geraedts et al., 2014). Total time allowed to use or complete the interventions ranged from four (Davidson et al., 2020; Fogarty et al., 2017) to twelve weeks (Nolte et al., 2021; Rice et al., 2020).

Table 2 Characteristics of the E-Mental Health Interventions

Three interventions incorporated human interaction, such as coaches providing written feedback on homework or moderator-facilitated peer social networking (Geraedts et al., 2014; Nolte et al., 2021; Rice et al., 2020). Nolte et al. (2021) provided feedback on program use and the ability to contact a clinician only to participants with moderate depressive symptoms (n = 308/509).

Just one intervention was classified as gender sensitive, having been designed for the needs, preferences, realities, and/or roles that men consider important. This intervention, Man Central, was developed “for men with depression, based on the results of two previous phases of research exploring men’s use of and preferred positive strategies to prevention and manage symptoms of depression” (Fogarty et al., 2017). Two interventions (HeadGear (Deady et al., 2018) and Driving to Health (Davidson et al., 2020)) were developed for people working in ‘male-dominated’ industries but not specifically for men in those industries. The Rice et al. (2020) intervention was not designed specifically for men, but the researchers incorporated gender-sensitive strategies aiming to reduce their attrition.

Characteristics of participant men

Participant characteristics and outcomes data are reported in Table 3. A total of 552 men made up the study intervention groups (39 (Geraedts et al., 2014) to 159 (Nolte et al., 2021) per study). Four samples (Botella et al., 2016; Davidson et al., 2020; Deady et al., 2018; Fogarty et al., 2017) comprised only men; in the others, they made up 31% (Nolte et al., 2021) to 48% (Rice et al., 2020) of the intervention groups. Participant men’s ages were reported in five studies (Botella et al., 2016; Davidson et al., 2020; Deady et al., 2018; Fogarty et al., 2017; Rice et al., 2020), with means ranging from 20 (Rice et al., 2020) to 40 years (Fogarty et al., 2017). The majority of the studies investigated participant men aged in their 30 s or 40 s (Botella et al., 2016; Davidson et al., 2020; Deady et al., 2018; Fogarty et al., 2017).

Table 3 Characteristics and E-Mental Health Intervention Outcomes Data of Participant Men

Outcomes

Four studies did not assess participants beyond post-intervention (Davidson et al., 2020; Deady et al., 2018; Fogarty et al., 2017; Rice et al., 2020) and the longest follow-up was one year post-intervention in one study (Geraedts et al., 2014). No study reported the specific number of days post-intervention at which participants had completed measures. Rates of post-intervention questionnaire completion varied widely, from 35% (Deady et al., 2018) to 100% (Botella et al., 2016).

All studies used patient self-report symptom questionnaires to measure the depression/anxiety outcome(s) of interest. Most common was the Patient Health Questionnaire-9 (PHQ-9), which was used in three studies (Deady et al., 2018; Fogarty et al., 2017; Nolte et al., 2021). All studies reported symptom levels, rather than whether or not a participant fulfilled diagnostic criteria for the outcome condition. A total of six studies (four treatment studies Deady et al., 2018; Fogarty et al., 2017; Nolte et al., 2021; Rice et al., 2020) and two prevention studies (Botella et al., 2016; Davidson et al., 2020)) reported pre- to post-intervention changes in these outcomes for men. All four treatment studies demonstrated post-intervention symptom reductions: three in depression (Deady et al., 2018; Fogarty et al., 2017; Nolte et al., 2021) and one in social anxiety (Rice et al., 2020). Conversely, both prevention studies reported no post-intervention reductions: Botella et al. (2016) reported no change in depression symptoms at 3-month follow-up despite medium/large positive effects (d = 0.59–0.77), and Davidson et al. (2020) found no change in depression symptoms and an increase in generalised anxiety symptoms. In between-group analyses, all three RCTs reported no detectable difference between intervention and control participant men’s depression symptoms at post-intervention (Botella et al., 2016; Nolte et al., 2021) or over one year (Geraedts et al., 2014).

Four studies reported men’s satisfaction and perceptions of the intervention’s usefulness (Botella et al., 2016; Davidson et al., 2020; Deady et al., 2018; Fogarty et al., 2017). Their average ratings of satisfaction and usefulness appeared largely positive in all studies, though significant variation in questions used to assess these made between-studies comparisons problematic. Within studies, participants’ ratings varied widely depending on the particular outcome. For example, in Fogarty et al. (2017), 80% said the intervention was convenient to use, 69% said it was easy to understand, 55% said it had taught them skills to handle future problems, and 47% said it led to symptom improvements (Table 3).

Meta-analysis

Meta-analysis of the two depression treatment studies (Deady et al., 2018; Nolte et al., 2021; pre-intervention N = 193, post-intervention N = 192) showed a significant, medium-sized, positive effect on depression symptoms (g = 0.64, 95%CI: 0.29, 0.99, p < 0.005, Nfs = 4.4), which met the criteria for a meaningful finding. Heterogeneity was low (I2 = 48.87; Q = 1.96, p = 0.16).

Discussion

We believe this to be the first systematic review and meta-analysis of the effects of psychotherapeutic eMH interventions on men’s depression and anxiety. Seven eligible studies were included. Study quality was high. Five eMH interventions targeted depression, one targeted depression and generalised anxiety, and one targeted social anxiety. Though their psychotherapeutic components almost exclusively centred around CBT or CT, there was little overlap in the interventions’ length, format, and inclusion of human interaction. Just one intervention was gender sensitive. Most participant men were aged in their 30 s or 40 s. All studies used non-gendered, self-assessed symptom questionnaires to measure the reviewed outcomes and few assessed participants beyond immediately post-intervention. Men appeared generally satisfied with the interventions and rated them as useful. All three RCTs (two treatment studies, one prevention study) reported no post-intervention difference between intervention and control group participant men on depression outcomes (no RCTs addressed anxiety). In studies reporting pre-post (intra-individual) intervention group data, all four treatment studies reported pre- to post-intervention improvements in either depression symptoms (Nstudies = 3) or social anxiety symptoms (Nstudies = 1), while both prevention studies (Nstudies = 2) reported no positive post-intervention symptom changes. These findings were supported by our meta-analysis of two studies, which showed a positive, medium-sized effect (g = 0.64, p < 0.05) of treatment interventions on men’s depression symptoms in pre-post data.

Our review did not include any studies comparing eMH to clinician-based therapy. However, eMH interventions are not intended to compete with or decrease men’s use of clinician-facilitated face-to-face or telehealth treatments. On the contrary, eMH may provide a complementary path to help, especially for those who may not otherwise access care (Christensen & Hickie, 2010). For example, eMH may act as an introduction to therapy, increasing men’s confidence to access clinician-based assistance at a later date. It may also facilitate access to care where clinician assistance is delayed, unwanted, stigmatised, or inaccessible, or allow men to monitor and better understand their symptoms before deciding whether to see a doctor. Our review provides preliminary evidence that psychotherapeutic eMH interventions can reduce the negative symptoms of common mental health issues in men under two conditions: (1) where the intervention aims to treat existing depression symptoms, and (2) in intra-individual pre- to post-intervention effects. The outcomes of the included RCTs in our review suggest that psychotherapeutic eMH interventions for men may not be significantly more effective than inactive/care as usual control conditions; however, two of the included RCTs that did not find significant between-group effects did report larger positive effects in their intervention groups than their controls, despite both studies suffering from low power and small samples (Botella et al., 2016; Nolte et al., 2021) and one investigating participants with minimal (normal or mild) baseline depressive symptoms (Botella et al., 2016). In the third RCT (Geraedts et al., 2014), both intervention and control participants had improved at follow-up. We are yet to understand how psychological factors, including expectancy effects and the benefits of attention, may influence intervention outcomes in this field. Research is also needed into ‘digital placebo effects’, wherein engaging with an online intervention leads to psychological benefits regardless of the intervention content (Firth et al., 2017a, 2017b). Unsurprisingly, interventions to treat existing symptoms showed more positive effects than those aiming to prevent them, given that the two prevention studies (Botella et al., 2016; Davidson et al., 2020) did not aim to, or tried not to, recruit participants with significant symptoms and were not aiming to reduce symptoms.

Due to the limited and variant data, we could not conduct further analyses on the influence of participant, intervention, and study factors on depression and anxiety outcomes. Previous mixed-sex/gender meta-analyses of eMH interventions have found greater benefits for mild to moderate rather than severe depression, and larger effects when compared to inactive rather than active controls in trials for depression (Firth et al., 2017a, 2017b) and anxiety (Firth et al., 2017a, 2017b). In a further individual patient data meta-analysis of internet interventions for adult depression (Karyotaki et al., 2018), older and native-born participants were more likely than younger and ethnic minority participants to respond positively to treatment. Evidence is also mixed as to whether incorporating human support (e.g., clinician feedback on homework tasks) into eMH interventions influences outcomes (Renfrew et al., 2020). Assessment of the effects of factors such as age, symptom severity, intervention content, format, duration, and inclusion of human feedback or interaction will be important inclusions in subsequent studies. With most included studies reporting only short-term follow-up, further research is also needed to investigate longer-term effects.

Previous reviews into mental health promotion interventions for men indicate that those with a gender-sensitive approach are more likely to be successful (Robertson & Baker, 2016; Seaton et al., 2017). With only one gender-sensitive intervention included in this review, further research into gender sensitive eMH interventions and comparisons of gender sensitive and non-gendered approaches will be important. Research is needed regarding methods of disseminating or promoting psychotherapeutic eMH interventions to men (Christensen et al., 2006). Within-group studies of variation in men’s needs and experiences related to eMH interventions are also needed and should be considered opportunities to optimise and personalise interventions for varied populations of men (Firth et al., 2017a, 2017b).

The psychotherapeutic eMH interventions in our review were almost exclusively centred around CBT or CT strategies but were diverse in length, content, format, and focus. As the men’s eMH evidence base grows, it will be preferable to analyse results from multiple trials of individual eMH interventions rather than pooling multiple diverse interventions. Future studies should also consider employing gender-sensitive mental health assessments when evaluating psychotherapeutic eMH interventions for men. While two studies in our review (Fogarty et al., 2017; Rice et al., 2020) did just that, we did not analyse data from those gender-sensitive measures as they were not the primary outcomes of interest. It is unclear whether the primary measures used by the included studies were able to fully capture men’s mental health symptoms and any effects the intervention had on these outcomes.

Strengths and limitations

A major limitation of our review was that we could not include 177 studies that met all other inclusion criteria (i.e., included men as participants, investigated the effects of a psychotherapeutic eMH intervention for depression or anxiety, etc.), but did not present sex- or gender-disaggregated data. Failure to report sex and gender-based analyses is common in many disciplines (Heidari et al., 2016) and appears similarly poor in eMH intervention evaluations. Lack of attention to sex/gender in eMH is problematic because it can influence intervention efficacy, preferences, experiences, and adherence (Karyotaki et al., 2018; Wang et al., 2020). We echo previous recommendations (Wang et al., 2020) that sex and gender be considered throughout future eMH intervention design, dissemination, and evaluation and that sex- and gender-disaggregated data be routinely collected, analysed, and reported as per the SAGER guidelines (Heidari et al., 2016). An individual patient data meta-analysis of data from participant men in the seven studies in this review and the aforementioned 177 excluded studies may represent a valuable step in this field, allowing investigation of outcomes and subanalyses that could not be conducted here due to the limited and variant available data.

Our review had a number of strengths. We cast a wide net to ensure all potentially relevant papers were included, conducting rigorous, comprehensive searches in six databases. The review followed an a priori registered protocol and PRISMA guidelines. While recognising that RCTs are the gold standard for evaluations, we elected to include any study design that quantitatively investigated the effects of interventions. As the first review of this topic, this allowed us to conduct a broad analysis and compare evidence from studies using varying designs. While our inclusion of only depression- and anxiety-targeted interventions was deliberate to facilitate a focused review targeting the two most common, frequently comorbid categories of mental health issues (Yang et al., 2021), it likely resulted in the exclusion of studies on more positively-focused, strengths-based psychotherapeutic eMH interventions (e.g., those aiming to improve wellbeing or resilience) that may also impact anxiety and depression and appeal to men.

Subsequent investigations will benefit from using randomised between-groups designs, evaluation of longer-term effects, and investigation into the impacts of expectancy effects, age, baseline symptom severity, intervention content, and human interaction on outcomes. Further, with all studies fitting our inclusion criteria having been conducted in high income, ‘Western’ countries and the majority focusing on men aged in their 30 s and 40 s, there is a need for further research involving more diverse populations of men. Investigation into the effects of eMH on a broader range of mental health and wellbeing outcomes in men is also needed.

Conclusion

The findings of our review indicate that men find psychotherapeutic eMH interventions useful and satisfactory, and that their use may lead to improvements in men’s depression symptoms. Development of future eMH interventions for men should emphasise men’s specific needs, desires, and circumstances throughout their design, dissemination, and evaluation, with particular consideration to developing gender-sensitive interventions, using gender-sensitive outcome measures, and reporting gender/sex-disaggregated data.