Introduction

Depression is one of the main contributors to global burden of disease and has been a major public health concern for decades [1]. People with depression have a reduced expectancy of life, frequent physical comorbidities, and increased risk for suicide [2, 3]. Depression is still largely undertreated. To the high rates of non-response to conventional treatments, we have to add the shortcoming of new advances in the treatments of depression and the lack of access to existing ones. The World Mental Health surveys from 23 countries show that only 16.5% of patients with depression receive a minimally adequate treatment [4].

Mobile health may provide an exit to this stalled situation. The application of mobile phone technology to healthcare settings—usually known as m-health—is increasing in popularity. Above three quarters of mental health patients own a smartphone, and around 90% declare they use mobile apps regularly [5]. When asked about their opinion on m-health, most patients show a positive response, with three quarters declaring their interest in using a mobile application for improving their mental health [5]. Thanks to the widespread ownership of smartphones and the endless diversity of applications they can host, m-health represents a promising new approach to the management of mental disorders [6,7,8].

Although this technology does not intend to replace human interaction, it may have some advantages compared with traditional treatments [9•]. m-health eliminates physical barriers, provides 24/7 access, and is relatively inexpensive [5, 10]. Additionally, patients may feel more comfortable discussing their symptoms through interactive software than face-to-face [10].

However, there appears to be two different rhythms in the development and implementation of m-health. Most health apps are commercially based, and while the market is growing exponentially, research cannot keep up the pace: the number of studies evaluating these apps is much lower than the number of apps available [11]. Also, there are few standards regarding the label, approval in clinical use, and methods of reporting and scoring these apps [12].

Choosing one application among such immense variety poses a challenge to mental health patients [13••]. Without guidance, apps for depression are at risk of devaluing their purpose and even becoming iatrogenic instead of therapeutic. For instance, if inadequately informed, patients may be led to believe a certain mobile intervention is sufficient and not seek the professional help they need.

In this review, we summarize the latest evidence about mobile applications for the management of depressive disorders. We sought to determine whether current m-health technologies for depression are ready for implementation in the clinical practice, and to identify the next steps to advance in the field.

Landscape of the Market

The lack of proper standards and reporting guidelines gets in the way when trying to define and characterize an app for depression. There are many apps in the market that claim to provide a treatment for depression, but not all of them deliver evidence-based interventions. This lack of regulation also makes it difficult to establish the exact number of apps for depression. While the total number of mental health applications is estimated to be over a thousand [13••], systematic searches of online marketplaces focusing on depression and applying selection criteria reveal around 278–310 depression apps [14,15,16].

Most apps for depression are commercially driven, while only a few are created by research teams and healthcare facilities. In a 2015 study exploring the market of apps for depression, only 5.3% of developers came from medical, academic, or research settings, while most of them were commercially oriented companies (29.5%), or did not reveal their affiliation (65.3%) [14]. Other aspects of the description, such as the type of treatment, or the evidence on which the app is based on, are often obscure or even misleading [14]. Similarly, Kumar et al. [16] found that only 9% of interactive apps for depression defined their scope. Moreover, some of the descriptions found in these apps could be misleading, as many make claims that users might be led to believe are medical and evidence-based, when in fact none of them has been approved by the FDA [17].

Contrary to what one might believe, being developed by a health or research institution does not guarantee adherence to evidence-based principles. Conversely, using evidence-based principles does not guarantee user satisfaction, or the popularity of downloads [15].

Regarding the availability of applications, the most common marketplaces to find apps for depression are Google Play Store and iPhone Apps store, which contain the highest number of apps for depression [14]. An important factor to take into account is that mobile apps developed by institutions are often not open for downloading on an e-market, but rather are provided for testing only for participants in the study.

Types of Apps and Characteristics

Applications for the treatment of depression are a heterogeneous group. Applications can be categorized by the features they offer, target population, technological aspects, developer, and theoretical framework.

Features

Regarding their features, text, audio, and video files are the most common means to deliver mobile-based interventions, mainly as a form of psychoeducation. Another frequent feature is logs that allow the user to register and monitor their progress in certain areas. Apps aiming for more complex forms of interaction incorporate live chats with different kinds of professionals, or even volunteers, in a similar approach to suicide hotlines. Some apps also offer in-built games to increase engagement, a concept known as gamification.

Types of Intervention

Therapeutic interventions for depression are numerous and varied. Apps that claim to provide a treatment for depression do so by interventions as varied as hypnosis, brainwave entrainment, music therapy, spiritual/faith-based, positive affirmation, breathing techniques, and yoga. However, the most common type of intervention is psychotherapy in its varied forms, and among these, cognitive behavioral therapy (CBT) is the most frequent.

CBT is an evidence-based psychotherapy that has proven effective in the treatment of depression. Although it is typically delivered face-to-face, there have also been successful attempts to administer it by e-health [18, 19], including Web-based interventions, and more recently mobile apps. There are two basic features that act as criteria for CBT-based apps: a monitoring function—some kind of log, so that patients can register their daily status, including mood, thoughts, and/or physical sensations, and keep track of their progress; and a psychoeducational module through either text, video, or audio files destined to increase self-awareness and knowledge about one’s own condition. CBT-based mobile applications offer behavioral and cognitive techniques, and usually include a function to monitor users’ status (including their cognitions, emotions, behaviors, and/or physical sensations), along with a psychoeducational component.

Shen et al. described 10 apps available in the market that delivered evidence-based CBT for depression, while Hughet identified 12 apps consistent with a CBT framework [14, 15]. Among the CBT-based apps with the highest number of downloads are Depression CBT Self-Help Guide, and The Mood Tools–Depression Aid, both of which have a user rating above 4 out of 5 (average ratings were 4.2 and 4.3 out of 5, respectively) [15]. The highest rating was for The Depression Cure, available for iOS, which received an average rating of 4.5 out of 5 [15].

As for the CBT-based apps that have been empirically tested, we have the MoodHacker, which showed a significant reduction in depressive symptoms in a 2016 randomized clinical trial (RCT) that tested the application in a sample of 300 adults with clinically diagnosed depression [20].

In contrast, most CBT-based apps lack empirically tested studies. In fact, of the 12 apps identified by Huguet et al. available in the market, none of them had been tested for effectiveness [15].

There are also apps that are not specifically designed for the treatment of depression, but that can significantly help in the treatment by tackling some of the core symptoms of depression. This is the case with apps for the treatment of insomnia or the prevention of suicidal behavior.

Insomnia and other sleep disorders are a frequent feature of depression and have been associated with higher recurrence and resistance to antidepressant treatment [21]. Among the many tools for the management of sleep disturbances, we have the Sleepcare app, which tackles insomnia through CBT techniques and which has shown promising results after it was tested in community-dwelling adults [22].

Another of the potential uses of m-health technologies is suicide risk assessment and prevention. Mobile applications have been used in clinical settings to reduce the risk of re-attempt. Berrouiguet et al. [23] followed up suicide attempters using a short messaging service (SMS), while Nuji et al. [24] designed a smartphone-based platform called the CASPAR (Continuous Assessment for Suicide Prevention And Research) that allowed patients to monitor their progress.

A transversal intervention can also be provided by medication and appointment reminders [25]. Improving treatment adherence and healthcare attendance is crucial, since lack of compliance is one of the most studied factors increasing treatment resistance [26].

One way to optimize the effect of m-health is to design applications for special populations, thus increasing the personalization of treatments. Some developers have focused on the particularities of certain groups, such as pregnant women [27•], the LGBT community [28], or children and adolescents [29]. The latter could benefit greatly from m-health interventions, since they are more proficient in the use of technology, and also present frequent problems of seeking professional health [30].

Effectiveness

Although studies evaluating the effectiveness of apps for depression are still far fewer than the applications in the e-market, an increasing number of RCTs are becoming available.

In 2017, Firth et al. performed the first meta-analysis of RCTs evaluating apps for depression. They found 18 studies testing 22 different m-health interventions, performed in either clinical or non-clinical populations. Pooled results from a total of 3414 participants showed that mobile applications reduced depressive symptoms to a significant greater extent than control interventions (p < 0.001). However, this clinical response was only observed in people with up to moderate depressive symptoms, but not in patients clinically diagnosed with mood disorders or anxiety. Authors also explored what features were associated with greater effectiveness, and they found that apps containing in-person interventions or cognitive training modules were significantly less effective [13••].

Since this meta-analysis, other clinical trials evaluating the effectiveness of mobile applications for depression have been conducted. Hur et al. evaluated the mobile application Todac-Todac in 34 participants with depression. The Todac-Todac is a CBT-based app designed to reduce negative beliefs in depressed people. The app was evaluated against a daily mood monitor that served as a control. After 3 weeks, both symptoms of depression and dysfunctional beliefs were reduced significantly more in the intervention group than in the control group [31•].

Pratap et al. evaluated the effectiveness of three apps for depression: Project EVO, which delivered a CBT intervention; iPST, which delivered a different psychotherapeutic intervention; and Health tips, which worked as a treatment monitor. These apps were effective for people with moderate symptoms of depression, and cognitive training was the most effective intervention [32•].

Table 1 summarizes the main findings regarding effectiveness.

Table 1 Main findings on effectiveness
Full size table

Feasibility

There appears to be an inverse correlation between length of use and effectiveness [13••]. Consequently, if we want to achieve long-term clinical responses, attrition is a crucial battlefront. Attrition has two components: dropout and non-usage [46]. Even if participants complete the study follow-up, using the app tends to decrease over time. For instance, in the RCT performed by Pratap et al., participation halved after a month of follow-up. By that time, passively collected data were twice as much as actively collected information [32•].

Evaluation of feasibility is sometimes clouded by the use of incentives, which is controversial. In order to increase engagement and decrease dropouts, some studies reward the participants who complete the study. These incentives are usually in the form of shopping vouchers. This is the case of the study performed by Pratap et al. which rewarded those users that completed the 12-week assessment with Amazon gift vouchers worth 75$ [32•]. Bluewatch users were also rewarded with a shopping voucher, this time worth $20 [47]. Studies that use incentives are not assessing feasibility in real-world conditions, which limits the validity of their assumptions.

Regarding user satisfaction, those who have used the apps are generally happy with them, demonstrated by the fact that the average rating for depression apps in 2015 was 3.5 out of 5 stars [14]. A survey performed on US veterans attending a mental health center revealed that the features that patients found more attractive were increasing physical exercise, improvement of sleep, cognitive restructuring, and behavioral activation [48]. In contrast, the main reasons reported for not using a mental health app were lack of evidence of their effectiveness, privacy issues, and inability to find the right app [48].

Table 2 summarizes the main findings regarding feasibility.

Table 2 Main findings on feasibility
Full size table

Barriers to the Adoption of m-health

Results from qualitative studies and surveys done to both patients and healthcare professionals can give us an idea of the issues yet to be overcome for the adoption of m-health. Although patients are open to using mobile health applications, interest differs from actual use. A 2019 survey performed on veterans with either depression or anxiety disorders revealed that while 73.1% of them were interested in mental health applications, only 10.7% actually used one of them [48]. Similarly, a study exploring the use of m-health technologies in patients with post-traumatic stress disorder (PTSD) showed that only 10% of them used a health app, even if over half of them had manifested their interest in this kind of interventions [51].

Difficulty in finding the right app is another major problem. The market of m-health grows at a much faster pace than the research destined to validate these interventions. Moreover, m-health—or e-health for that matter—is not integrated in public health systems, which can raise suspicion among users and deprive them of the possibility of asking their healthcare provider what app should they use, if any. Without counsel, users interested in e-health are left with the apps description available on the online store. Shen et al. [14] found that most of these descriptions were insufficient and many of them did not provide information of the developer’s affiliation, or the theoretical framework on which the intervention was based.

The non-integration of m-health into public health systems also implies that many may ignore the existence of mental health care apps, or distrust them as they are not covered by their health insurance [51]. In this regard, the UK’s National Health System (NHS) has made an effort in starting to consider e-health as a therapeutic tool integrated in healthcare coverage [52].

Suspicion about privacy violation is another reason for patients’ reluctance [50, 53]. Concerns about privacy are reported as a barrier to the use of apps by several studies [50]. Lack of privacy policies is a frequent issue in apps. Users care about privacy and one of the features that they value is that sensitive information, such as logs, is password protected [15].

Barriers related to the mental disorder itself also come into play. A study showed that veterans without PTSD were more interested in engaging in mental healthcare applications than their peers with PTSD, even though the latter group demanded more mental health services [54].

Healthcare professionals can also slow down the adoption of m-health in the clinical practice. There are several concerns among the medical community regarding the reliability of m-health interventions. The number of studies evaluating these apps is markedly lower than the number of apps itself [14], and most mental healthcare applications have not been explored in terms of tolerability and effectiveness. Training doctors, nurses, and other healthcare professionals in the use of mobile health technology is a crucial step to take for the implementation of this technology in the clinical practice [55].

Conclusions

Are depression apps ready to work? This is the question that arises after reviewing the (limited) evidence available on the matter. We have a number of apps that have proven effective in the treatment of depression relatively free of bias. Albeit small, this number would be sufficient to start implementing this advancement in the clinical practice. After all, only one antidepressive agent is necessary to start treating depressive patients with antidepressants. m-health presents the further advantage of the lack of physical side effects—although they can present other risks, such as psychological dependency. Why then is this technology not being used in clinical practice? There are several factors that can get in the way of using a treatment that are independent of the treatment itself. Apps for depression may be ready, but we are not.

The adoption of m-health interventions in the clinical practice depends on many factors. We can establish different levels of action, from regulatory bodies to the app users themselves, as well as developers and researchers—who are sometimes the same—and healthcare professionals. Tackling the barriers observed at certain levels cannot be done if some issues have not been solved previously in the levels above. For example, we have seen that users are often overwhelmed by the number of options available in the digital health market, and that they have difficulties in choosing the application that best suits their needs [14]. To solve this problem, the role of healthcare providers is essential: if they have sufficient understanding of digital mental health, they will be in a position to guide their patients to the best intervention. However, it is difficult for health professionals to do this if they have not received the necessary training, which in turn requires sufficient funding to organize refresher courses and workshops [56].

Similarly, developers need to improve apps’ graphics and interface in order to make the software more attractive to patients [57, 58]. Quality of mobile software is increasing rapidly, but its uses are mainly commercially oriented. Without proper investment, there is a risk that the most useful application of this technology—that of serving public health—will be left using outdated software compared with its commercially oriented counterparts.

Figure 1 illustrates the different actions that can be taken at each level to overcome the barriers to the adoption of m-health in the clinical practice. This flow of influence goes in the opposite direction too: if authorities are not pressured from below, actions are unlikely to be taken. Users can also influence the actions and attitudes of healthcare professionals. In this regard, the interest and positive reception of psychiatric patients toward m-health should be a wake-up call [48].

Fig. 1
figure 1

Levels of action for reducing barriers to the adoption of m-health

Full size image

The negative attitude of some professionals towards this technology is not unfounded but can close doors to further advancement in the right direction. Developers also need to integrate the information coming from users and professionals in order to improve application designs and interfaces, and optimize them for use in clinical practice.

Another point of interest is the development of unified standards about the definition and reporting of mobile apps. The lack of standards in creating, evaluating, and reporting health apps is one of the main obstacles to the implementation of these interventions in clinical practice. In this regard, regulatory bodies have created a model for the assessment of health apps called the Digital Health Precertification Program (Pre-Cert Program), which is a crucial step toward the regularization of these interventions, and of e-health in general. This model aims to be based on real-world data, that is, performance in real conditions, the actual settings at which the apps will be used [59]. Other efforts toward the standardization of m-health include the system proposed by Chan et al. [60] to evaluate mobile apps based on three areas: usefulness, usability, and infrastructure.