Depression affects >264 million people worldwide and was the third major contributor of years lost to disability in 2017 [1]. Yet, approximately 50% of the people living in high-income countries and at least 75% of the people living in low- and middle-income countries [2,3] have neither been diagnosed nor receive treatment. Untreated depression is associated with increased morbidity and mortality, including death by suicide, poverty and unemployment at an individual level, and significant increase in health expenditure at health system level because of increased health care use and decreased workforce productivity of the affected individuals [3,4]. Although the reasons for this wide treatment gap are multifarious, suboptimal understanding of depression and mental health disorders in the general population is considered an important contributor to low use of mental health services. Studies assessing the general population’s mental health literacy through the use of standardized clinical vignettes have shown poor recognition of common mental health disorders [5-7].

Mental health literacy is defined as “the knowledge and beliefs about mental disorders, which aid their recognition, management or prevention” [8]. It encompasses the acquisition of factual knowledge of mental health disorders as well as development of competencies and beliefs enabling prevention, early recognition of symptoms, help seeking, self-help, and provision of first aid to others [5]. Evidence-based clinical guidelines for management of depression indicate that psychoeducation is pivotal in terms of disease management [9-12] and necessary to expand the shared-decision model of care in mental health services [13].

Educational topics focused on depression may include information related to its natural history, symptoms and signs, treatment options such as pharmacotherapy and a range of nonpharmacological treatments and possible side effects, as well as information related to prognosis and effective self-management interventions [9]. Psychoeducational interventions have been shown to increase mental health literacy and, most importantly, to yield small but significant reductions in symptoms of depression and mental distress [14,15], hence offering simple, inexpensive, and readily available tools for symptom management [14].

There are other notable benefits of psychoeducation, particularly stigma reduction [16-18] and increased help-seeking behavior [6,19-21]. Mental health stigma remains a significant barrier to the use of mental health services because it affects the access and quality of health care provision for people living with mental health disorders and depression [22,23]. Social stigma or public stigma are linked to discrimination, avoidance, and inadequate treatment [24], whereas self-stigma or internalized stigma may erode self-esteem and self-efficacy or lead to anger or indifference [24], which may in turn hamper help seeking [25]. Although multilevel coordinated and sustained efforts are needed to mitigate the prevailing mental health stigma, simple interventions such as the use of personal narratives have shown promise in fostering depression awareness [26,27] and more positive attitudes among the public [26,28,29].

Since the early 2000s, the search for health information has shifted increasingly to the web [30], and this trend has further intensified with the advent of smartphones [31] and mobile apps. A recent systematic review on the provision of medical education using smartphones reported that approximately two-thirds of the reviewed interventions were effective in improving patients’ knowledge and clinical outcomes [32]. The Apple App Store and Google Play, which are the major sources of apps worldwide [33], currently include >10,000 mental health apps [34] that offer a wide range of functionalities comprising education, screening, and self-management programs for a wide range of disorders [35,36]. However, most of the published apps have not been assessed in clinical trials and are not evidence based [34,37].

Given the increasing popularity of health apps, a number of studies evaluating direct-to-consumer apps have been published in recent years, including the general characteristics of the apps [38], key features and functionalities of highly downloaded [39] or highly rated apps [40], techniques commonly used in face-to-face psychotherapy [41], or adherence of self-guided app interventions to evidence-based clinical guidelines [42]. These studies include a variety of assessment methodologies, from researcher-developed checklists [38,41,42] used in specific projects to the use of standardized assessment tools such as the Mobile App Rating Scale [36,43] or the Organization for the Review of Care and Health Applications–24 Question Assessment [44], focusing on the usability and technical aspects of the apps. In contrast, the health-related content of apps is seldom included in assessment checklists.

Existing mental health apps often include educational components, either as its sole functionality or within a depression or mental health management app. The trustworthiness, adherence to evidence, and depth of information offered by these apps is unclear because, to date, no reviews offering an in-depth assessment of the content of education on depression modules provided by mental health apps has been published.

Therefore, this study aims to systematically evaluate the adherence to evidence-based clinical guidelines of the information on depression provided by mental health and depression apps available in the Apple App Store and Google Play.

The research team designed the assessment criteria to evaluate the clinical and technical features of apps. The criteria included the following three domains:

The Apple App Store and Google Play were systematically searched using 42matters, a proprietary app database [54], on July 8, 2020, using the terms depression, depressive, depressed, mood disorder, sadness, and melancholia. The search was limited to four app store categories: education, health and fitness, lifestyle, and medical. Eligible apps were required to conform to the criteria presented in Textbox 1.

The app selection process is presented as a flowchart (Figure 1) [55]. Before screening for eligibility, we excluded all Android apps with <1000 downloads as reported in the 42matters search output. As iOS does not report the number of app downloads, the iOS versions of the excluded Android apps were also excluded on the assumption that the app would have a similar number of downloads in both app stores. The remaining apps underwent a 2-step selection process that consisted of (1) screening the app name and app store description from the 42 matters search output and (2) downloading and screening for eligibility all apps included in step 1. Working in parallel, 2 pairs of investigators (LM and Goh Jun Wei; LM and Matthew Teo Siu Yan) independently completed the app selection process. Disagreements were resolved through discussion.

Apps were assessed by 2 assessors working in parallel using an iPhone 7 (iOS version 14.0.1) and a Sony XPERIA XZs (Android version 8.0.0) smartphone. If apps were available on both platforms, as per the 42matters search output, we assessed both versions and counted each version as an individual app. We used descriptive statistics to analyze the data. We compiled and tabulated the results and reported them as a narrative synthesis.

We performed a subgroup analysis of Android apps to assess whether the number of app downloads was associated with the educational content or the quality or number of features offered by the apps. Apps were categorized into two groups: apps downloaded 1000-10,000 times and apps downloaded >10,000 times. The assessment included selected items from all 3 sections of the assessment. The data were tabulated and compared using a significance test for categorical variables: the chi-square test was used if each category contained >10 variables and a 2-tailed Fisher exact test was used if any of the categories in the contingency table contained <10 variables. Statistical significance was set at P<.05. Statistical analyses were performed in RStudio (R version 4.0.3). iOS apps were not included in this analysis because the Apple App Store does not include this information.

Table 1 presents a summary of the characteristics of the included apps. Of the 58 apps included in this analysis, 29 (50%) were Android apps and 29 (50%) were iOS apps; 21% (12/58) of the apps were available on both platforms, and 60% (35/58) of the apps belonged to the health and fitness app store category. Of the 29 Android apps, 3 (10%) had been downloaded >1 million times [56-58]. Of the 58 included apps, 15 (26%) offered only education and information modules, whereas 37 (64%) offered education modules along with other mental health or depression management features. A simple interface that did not allow for user feedback or customization was offered by 41% (12/29) of the Android apps and 31% (9/29) of the iOS apps. More than 10% of the apps offered education modules targeted to specific user groups.

All assessed apps worked as intended and were generally easy to use. Data entry from users was not requested by 31% (9/29) of the iOS apps and 41% (12/29) of the Android apps (Table 3).

Of the 58 included apps, 13 (22%) available on both platforms did not include a privacy policy (6/29, 21%, Android apps and 7/29, 24%, iOS apps). These apps offered passive information that did not require data entry by users, except for 3% (1/29) of the Android apps and 7% (2/29) of the iOS apps, which requested the user’s email address to set up a password.

Of the 58 apps, 30 (52%) included references or the author’s signature or shared links to reputable websites for the information provided. Half of the Android apps and a quarter of the iOS apps did not declare developers’ affiliations, and 41% (24/58) of the apps included a disclaimer that the information provided did not replace a health care provider’s advice.

Of the 58 apps, 10 (17%) contained advertisements. These apps were free to download and use, except for an Android app that offered in-app purchases. The advertisements included in 75% (6/8) of the Android apps filled the screen and disrupted the use of the app, whereas 25% (2/8) of the apps included banner advertisements that allowed users to continue using the app. All advertisements were generally unrelated to the app content, promoting other Google Play apps, video games, beauty products, education support centers, and health care services such as dental services. Of these 8 Android apps, 1 (13%) promoted psychological services. In contrast, 100% (2/2) of the iOS apps offered banner advertisements that allowed for continued use of the app. The advertisements were not related to the app content.

Our subgroup analysis assessing the association between the Android apps’ popularity (as indexed by the number of app downloads) and the breadth and depth of the educational topics covered showed no statistically significant difference between apps downloaded <10,000 times and apps downloaded >10,000 times. Nonetheless, apps downloaded 1000-10,000 times provided non–evidence-based information more often than apps downloaded >10,000 times (7/15, 47% vs 3/14, 21%, respectively); however, this difference was not statistically significant. Of the 7 apps providing non–evidence-based information and downloaded <10,000 times, 2 (29%) belonged to the medical category in the app store. Of these 2 apps, 1 (50%) claimed that adhering to the Law of Attraction improve depression outcomes. Table 4 summarizes the analysis.

To our knowledge, this is the first systematic assessment of the information on depression in existing mental health and depression apps. Our findings suggest that the information included in these apps is often limited and not aligned with evidence. Most of the assessed apps offered brief, factual information on symptoms and treatment options, with only a few apps addressing all aspects of disease presentation and management or including information on recovery or providing personal narratives of people living with depression. The inclusion of personal accounts of people living with, and recovering from, a mental health disorder has been associated with increased understanding of the mental health disorder and recovery process, validation of the personal experience, and reduction of stigma [28,60].

Patient education is one of the pillars of effective depression management resulting in better treatment compliance and improved outcomes [9,11], particularly if they are complemented with personal narratives and positive messages emphasizing recovery [51]. Adequate health literacy is associated with tolerance and acceptance of people living with a mental disorder [16], increased help-seeking behavior [16], and adherence to treatment [61,62], and it is a prerequisite to engage in shared-decision models of care [63]. Yet, 25% of the potentially eligible apps, including some of the most popular mental health apps with >1 million downloads each, were excluded in the second round of screening because they did not provide any information on depression.

Access to mental health care is often inadequate, particularly in low- and middle-income countries [3] and among people of lower socioeconomic status as well as those living in remote areas in high-income countries [64,65].

Mental health apps may play an important role in improving access to mental health care, supported by an increasing expansion of mobile network coverage to remote areas worldwide [31]. However, our assessment shows that apps currently available in public app marketplaces do not offer a holistic, evidence-based self-management program to meet these health needs. For example, less than half of the assessed apps include a screening test for depression and only a subset of these apps offer follow-up recommendations to users potentially living with depression. Of the 58 assessed apps, 12 (21%) offered non–evidence-based information, including developers’ or columnists’ personal views on depression, and suggested scientifically unproved treatments. Of these 12 apps, 2 (17%) were classified as medical in the app store. Lack of evidence-based information in health apps is a recurrent theme in most systematic assessments performed by our group [45,46,48] and others [42,66,67], revealing that current app development and publication processes overseen by app stores might not be suitable for health apps. We previously highlighted the lack of governance and quality assurance of the health app industry [48], which we believe constrains the further development of health apps as genuine tools to support access to mental health care. Therefore, it is imperative to develop a multifaceted approach involving the research community, commercial app developers, app store managers, and official regulatory bodies to define development and publication regulatory frameworks for health apps.

Our subgroup assessment of Android apps suggests an association between the number of downloads and the overall quality of the apps. In the analysis, apps downloaded >10,000 times were more trustworthy, included more evidence-based content, presented better compliance with data privacy and security, presented references to endorse claims, and incorporated health care providers into the app development team more often than apps downloaded <10,000 times. Excluding app credibility, the differences were not statistically significant, probably because of the small size of the sample.

In this study, we used an established systematic assessment approach, which is based on the rigorous systematic review and which we have applied to various health domains [46-48] over the years. This methodology included the use of a commercial database to search for eligible apps, which provides a wider geographical scope for the search, and the development of exhaustive assessment criteria using renowned evidence-based clinical guidelines and our center’s criteria for technical and quality assurance of apps.

There are limitations to our research as well. The search strategy included depression-related terms and omitted education-based terminology; thus, we may have omitted some relevant apps. However, the inclusion of such terms would have greatly increased the overall number of retrieved apps and would have also led to retrieval of many irrelevant apps because the database used for app retrieval does not allow the use of search strings; instead, it retrieves individual results for each search term. We restricted our search to apps for mental health, excluding apps offering medical or general health advice that may have presented summaries on depression presentation or management. Our assessment was limited to apps downloaded at least 1000 times, which may have omitted newly launched apps that have yet to reach the minimum required number of downloads. We also restricted our search to apps in English, potentially excluding relevant apps in languages other than English.

Information in mental health and depression apps is often brief and incomplete, with 1 in 5 apps providing non–evidence-based information. Given the unmet needs and stigma associated with depression, it is imperative that apps seize the opportunity to offer quality, evidence-based education and point users to relevant resources. A multistakeholder consensus on a more stringent development and publication process for mental health apps is essential.