Social networking sites (SNSs) are Web-based platforms on which individuals connect with other users to generate and maintain social connections [1]. Considerable disagreement exists as to associations that SNS use may have with depression and anxiety [2,3]. On the one hand, SNSs may protect from mental illness, as they support and enable social interaction and connection [1,4], and allow users to reflect aspects of their identity and express emotion that may be relevant to their lived experience [5]. On the other hand, there are many opportunities for miscommunications and mismanaged expectations, and maladaptive tendencies can be exaggerated, leaving individuals feeling a greater sense of isolation [2,6]. As a whole, the SNS environment may be just as complex as face-to-face interactions. As SNS membership continues to rise [7], it is becoming increasingly important to address the possible benefits and detriments the use of SNSs may have on mental health.

Affective disorders such as depression and anxiety have been shown to have bidirectional interactions with the social environment that influence the path of illness onset and maintenance [8]. Depression and anxiety have an approximate prevalence of 4.7% and 7.3%, respectively, in the global population [9,10]. These disorders have high levels of comorbidity [11] and impact the quality of social relationships [12,13]. Depression and anxiety may be implicated in determining the size and structure of an individual’s social network [12], the quality of interactions within these networks, and how effectively social capital may be leveraged or developed to provide an individual with social support [8,14].

The social characteristics (both qualitative and structural) affected by depression or anxiety are also relevant to one’s sense of well-being. Current mental health theories suggest that the presence of well-being is not the same as the absence of mental illness; a complete model of mental health requires not just the absence of psychopathology, but also a focus on positive indices of functioning such as subjective well-being [15]. This is particularly pertinent when exploring how the social environment may affect an individual, as such environments may simultaneously confer a number of benefits to the individual and exaggerate deficits [16-18].

Social aspects of the Internet have been argued to augment social relationships and support mental health. SNSs in particular connect us to friends, family, colleagues, strangers, and celebrities and can help users to maintain and make new friendships, express thoughts and feelings, and express identity [1,4,19]. The primary social functions that SNSs perform may augment the benefits of engaging in face-to-face interaction by extending the reach and accessibility of our social networks [20]. Indeed, SNS use is associated with lower levels of loneliness and greater feelings of belonging (social connectedness), social capital, and actual and perceived access to social support and is generally associated with higher levels of life satisfaction and self-esteem [6,21-26].

As a whole, the positive social components of SNS use suggest a protective role against depression and anxiety. For instance, higher levels of self-esteem and life satisfaction may aid in attenuating depressive symptoms [27]. Kraut et al [28] found that frequent general Internet use did not increase depression over time, and, in a second study, communication activities on the Internet were shown to be associated with lower levels of depressive symptoms [29]. Computer-mediated communication (CMC; eg, email, instant messaging) allows users to express and interpret emotion in a similar way to face-to-face interaction [17]. CMC may therefore be beneficial for emotion regulation as has been demonstrated for offline forms of written emotional expression [30,31].

However, for individuals with depression or anxiety, the interpretation and frequent exposure to this emotion may have a negative impact [13]. SNS use may increase an individual’s exposure to negative social interactions (eg, cyberbullying), which may negatively impact mood and mental health [2]. For example, negative interaction quality was associated with decreases in self-esteem and life satisfaction [32]. Even passive exposure to the language used in SNS posts has been shown to influence the emotive language subsequently expressed by the receiving SNS user, where positive or negative emotions are argued to transfer via contagion [33-35]. As SNSs explicitly support a number of social features, the relationships and interactions between the user, their emotional experience, and Web-based technology are likely to be complex and may even accentuate differences between those who are doing well in life and those who are struggling.

Cognitive and social factors frequently emerge as both moderators and mediators of the relationships between offline social interactions or events and depression [36-38] and might also occur in Web-based environments. For instance, self-esteem mediates the pathway between relationship interactions and depressive symptoms [39], but it might also moderate how a person uses and is affected by the SNS. Rumination, a response style where an individual maintains a passive and repetitive focus on their distress [40], is one mechanism linking stressful life events and the development or maintenance of depression [41], and the SNS environment provides opportunity for a person to both internally ruminate on bad events and have an entire social network further accentuate shortcomings. Social support has additionally been shown to moderate relationships between stress and depression, with greater levels of social support acting as a buffer to depressive symptoms [42]. This is pertinent to SNSs as they present a potential intervention opportunity for developing and strengthening supportive social networks for vulnerable individuals.

Since the advent of SNSs, a number of articles have been published examining the relationship between SNS use and depression and anxiety. The interaction between SNSs and our mental health and well-being is clearly varied and complex. The objective of this paper was to provide a systematic review of literature examining SNSs and their relationship with depression and anxiety. It also considers links with well-being, as well as potential mediators and moderators to these relationships.

Figure 1 summarizes the search strategy and article selection. A multidatabase search identified studies conducted between January 2005 and June 2016. The databases included were PsycINFO, MEDLINE (Ovid), Scopus, IEEE Xplore, CINAHL (Cumulative Index to Nursing and Allied Health Literature), Education Resources Information Center, Social Sciences Citation Index, and Communication and Mass Media Complete. The inclusion of conference papers accessed through IEEE Xplore was intended to capture the research within the computer sciences and engineering fields that may have been relevant to the psychological literature.

Search terms were selected in order to comprehensively capture the various ways mental health, mental illness, subjective well-being, and SNSs have been defined and explored in the existing literature.

SNSs were defined as conceptualized by Ellison and Boyd [1] as sites that are a Web-based communication platform with 3 distinct characteristics: (1) user profiles are unique and created through user-provided content and content provided by other users, (2) the network connections between individuals are visible and can be navigated through by other users, and (3) individuals can broadcast content and consume and interact with content contributed by others in a continuous stream of information. Prototypical examples of SNSs include Facebook, Twitter, Myspace, and Instagram.

For mental health, search terms specifically focused on depression and anxiety, as well as overall well-being (eg, subjective well-being, psychological well-being, wellness; see Figure 1 for full list of search terms).

Studies were included if they had a primary focus on SNS use as a behavior. As such, studies that referred to SNSs as a recruitment method only or used SNSs as a means for intervention delivery were excluded.

Articles were included if they provided results addressing anxiety or depression directly and were excluded if they were only referred to in the context of general psychological distress (or similar). As the primary focus of the review was on depression and anxiety, not the broader well-being construct, articles addressing well-being were only included if they also included specific reference to anxiety or depression.

The search was limited to articles published after 2005 to capture research on the prototypical examples of SNSs that include the basic features of modern networks. Studies that had a primary focus on the Internet, chat rooms, or online support forums were also excluded; although they may contain some of the features of SNSs, differences in the function they perform for users may exist [19].

Additionally, articles were restricted to English language, peer-reviewed journal or conference proceedings, and quantitative or mixed methodologies. Gray literature, commentary and editorial, qualitative research, literature reviews, and descriptive case studies were excluded.

Two raters (the first author and a trained research assistant) reviewed all abstracts returned from the literature search and selected abstracts for full-text reading based on the inclusion and exclusion criteria. All articles that included measurement of depression, anxiety, or well-being were retained. The selected full-text articles were downloaded and reviewed by the first and third authors.

To provide some preliminary evaluation of the strength of the research, three risk of bias indicators were adapted from the Cochrane bias tool (Cochrane Handbook for Systematic Reviews of Interventions [43]), which classifies methodology that may limit replicability or generalizability. Studies were rated to indicate whether the study (1) included psychometrically reliable and valid measures, (2) used an external measurement criterion for mental health, and (3) provided description of the sample demographics including some SNS activity statistics (eg, number of friends and/or use frequency). These were rated by the first and third authors from “0=No bias,” “1=Unclear risk of bias,” and “2=High risk of bias” and were summed to create a final score between 0 and 6. A linear weighted kappa statistic for interrater reliability (.78, SE=.06) indicated that there was very good agreement in applying the bias criteria. Consensus was reached on all ratings. Articles with a rating of 3 or above were excluded [44-52], resulting in the final set of 70 studies, as presented in Multimedia Appendix 1.

From each article, the year of study, population of interest, type of SNS, and variables used (anxiety, depression, well-being) were noted, along with whether or not any formal mediators or moderators of these relationships were indicated. Information was then qualitatively synthesized to identify common themes.

Figure 2 indicates the number of articles addressing SNSs, depression and anxiety, and well-being from 2005 through 2016, based on the 302 full-text articles initially reviewed. There were considerably more articles addressing well-being alone than articles only addressing depression and anxiety. Only 15 articles included both positive and negative aspects of mental health. This review includes the 70 articles that include depression or anxiety only or depression or anxiety and well-being.

A total of 22 studies addressed potential moderators or mediators in SNSs’ relationship with depression or anxiety (see Multimedia Appendix 1). Most articles obtained a bias rating of 0 to 1. Ratings of 1 or above were primarily due to the limited focus on reporting SNS activity statistics, such as the number of friends or average frequency of use, which help characterize the average SNS user in each sample. Facebook was the most commonly explored SNS followed by the measurement of SNS use as a general category (ie, no specific platform explored). The majority of studies examined young adults (late teens or early 20s).

Across the 70 articles, several general themes were apparent: frequency of use, size and structure of the SNS, language features and observable SNS activities, self-disclosure and expression, quality of interactions, social support, social connectivity, social comparison, addictive and problematic behaviors, and physiological associations. Findings are summarized in Multimedia Appendix 2 and are described below, with particular attention to moderators and possible mechanisms involved in the associations. As some articles were relevant to multiple themes, these articles appear in multiple sections. Studies that included well-being are also highlighted.

This systematic review examined associations between SNS use and anxiety and depression. Across 70 studies reviewed, a number of positive and negative correlates have been suggested, as well as moderators and mechanisms of these associations. On the basis of this review, it is likely that there are differing engagement and interactional styles on SNSs for users high in social anxiety and depression. These may be driven or defined by both symptoms and motives to compensate for needs that are not met face-to-face. Negative interactions, frequent social comparison, and SNS addiction or problematic use are related to higher levels of depression and anxiety. Furthermore, cognitive response styles such as rumination or brooding may exacerbate the negative interactions between SNS use, depression, or anxiety for some individuals.

While these potential risks exist for mental health, it is also clear that SNSs can provide considerable benefits to their users. Positive quality interactions, social support, and social connectedness most consistently related to lower levels of depression and anxiety. Social support and connectedness derived from SNS use may be uniquely beneficial to individuals with social anxiety who are unable to access these resources face-to-face. However, especially for those with depression, some evidence suggests that there is a discrepancy between the perceptions of interaction quality and social support and the actual content of their SNS communications, which may attenuate the potential positive impacts of SNS use.

Across a number of studies, observable SNS features such as language use and expressions of identity on user profiles have been demonstrated to provide insight into the depression and anxiety status of the SNS user. With continuing research these characteristics may be a useful tool for monitoring mental health. The content and quality of interactions on SNSs may provide the clearest candidates for monitoring depression and anxiety and may be potential intervention targets for improving mental health and well-being through engaging with SNSs.

Across studies, social aspects, including feelings of social support, social connectedness, and positive interaction quality, emerged as protective factors for SNS users. The SNS network structure itself may play an important role in supporting mental health, in that some platforms may better provide social resources to individuals with depression. Indeed, more integrated social networks on SNSs were associated with lower levels of depression [89]. Studies suggest that social support and social connectedness derived from SNSs are constructs distinct from general social support or connectedness [111,113]. SNSs may therefore be contributing additional benefit to their users by creating another domain in which individuals can access, or have greater perceived access to, social support, especially with individuals for whom face-to-face interaction is difficult [123-125]. The broad and visibly articulated social context on SNSs may contribute to the feeling of social connectedness derived from SNSs and its association with better mental health outcomes [126]. As such, SNSs may provide an environment where those already high in social skills and resources are benefiting from their cumulative sources of social support (“rich-get-richer”; see [28]) as well as augmenting social support access for those who have difficulties engaging face-to-face [111,123-125].

Consistent with offline research, the perception of social support appears to be more important than actual support [126-128]. Findings demonstrated that perceived social support was greater in those with lower depression scores and that perceived communication competence may contribute to this relationship [79,93]. Greater perceived positive interaction quality and greater reciprocity in interactions are also indicative of lower depression and anxiety. Similarly, Valkenburg et al [32] demonstrated higher levels of life satisfaction and self-esteem for those who frequently reported positive peer experiences on SNSs. However, aspects of the individual that drive depressive feelings and social anxiety, greater use of negative language, and cognitive aspects such as social comparison and rumination, can prevent the user from perceiving support that is actually there [93], further contributing to depressive or anxious symptoms.

The valence of posts on SNSs may both reflect and impact depression and anxiety. Individuals scoring higher on depression scales in the reviewed studies generally expressed more negative affect on SNSs and were more likely to perceive negative interactions. The way individuals interpret emotional and social content on SNSs may place depression as antecedent to maladaptive SNS use, which may, in turn, maintain depressive symptoms. For individuals who are already depressed, ambiguous interactions are often interpreted as negative [13,129], which may attenuate the potential benefits available through SNS use.

Evidence suggests that frequent positive expressions are associated with better mental health, and frequent negative expressions are associated with depression and poorer life satisfaction [67,91,96]. While therapeutic writing can provide some benefits in reducing distress and improving well-being [30,31], online writing may serve a different function, with Web-based expressions reflecting the lived experience of the individual (eg, [91,130-132]), rather than providing a therapeutic outlet. Indeed, relative increases in posting frequency were shown to be associated with greater depressive symptoms [84]. For others, the presence of social anxiety may hinder the use of posting functions for emotional disclosure on SNSs [59], which may decrease access to potential social interaction [98]. As emotional content can be effectively communicated on the Web [133], SNSs represent another space in which positive and negative interactions can be enacted and may provide key behavioral insights into the mental health and well-being of a SNS user. Alternatively, increases in self-expression on SNSs may be more beneficial to well-being domains (such as connectedness, social support, and life satisfaction) but may not have an impact on depression or anxiety. A direct comparison of these relationships has not been conducted, and might be an area to investigate in the future.

The prominent risk factors for depression and anxiety that emerged from this review included frequent SNS social comparison, negative perceived interaction quality, addictive or problematic SNS use, and rumination (or brooding). These factors represent cognitive and interactional styles that have well-established associations with depression and anxiety but may be enhanced by the enduring nature of social content on SNSs. Although the total frequency of SNS use does not appear to be directly related to either depression or anxiety, there are different moderating and mediating factors [68,73,77,78] and patterns in the functions of SNS use by individuals with higher depression or anxiety that may contribute to or exacerbate symptoms [69,74-76,78].

One of the risk factors for depression and an individual’s interaction with SNSs was rumination. Greater rumination is frequently associated with higher ratings of depression and also impacts well-being by maintaining a focus on negative affect [134,135]. Rumination is a likely mechanism for the relationship between negative interactions with SNSs and depression based on its role in SNS negative emotional expression [67] and social comparison [115]. There is considerable potential for SNSs to amplify and assist ruminative processes by exposing SNS users to a constant stream of rich social information that can be selectively reflected on as permanent content on a user’s profile [54,115].

Similar to depression, the cognitive risk factors for social anxiety include social comparison (via brooding) and the perception of frequent negative interactions. However, the pathway to and importance of these risk factors may differ from depression. In contrast to those with depression, those high in social anxiety mainly use SNSs for passive browsing and private communication, not for content production [75]. The passive uses of SNSs may place individuals at greater risk of more frequent social comparison, which may have negative mental health effects [114]. This differs from the relative benefit of content production on SNSs for an individual with social anxiety, as posts are often rated as being more appreciated by friends in the network [98], which may have a flow-on effect to the perception of SNS-derived social support [111] and may even reflect more positive interactions with peers [103].

The reduced social cues on SNSs may be attractive to individuals with social anxiety, as has previously been suggested in the general Internet literature [124]. However, the need to compensate for a lack of belonging and social reassurance in face-to-face interactions, in conjunction with lower self-regulation, may drive problematic SNS use for individuals with social anxiety [65,104,106,117]. Similarly, these motives may also contribute to individuals with social anxiety generating more content on their profile pages than others [57], and for those highest in social anxiety it may contribute to a higher frequency of SNS use [69]. On the whole, there appear to be a number of well-being benefits to using SNSs for individuals high in social anxiety that cannot be gained in face-to-face interactions; however, the pattern of SNS use may negatively affect other domains.

The frequency of SNS use as a whole suggested no clear association with depression and anxiety. Longitudinal research suggests that depression and anxiety remain stable in the context of how frequently a user engages with SNSs [54,56,61,63,77] and the function of use holds clearer associations with depression and anxiety [75]. This is consistent with the literature examining general Internet use where total frequency of use is often not a predictor of depression, particularly when examining the social features of the Internet [28,125]. For example, when examining different functions on the Internet, Morgan and Cotten [29] showed that more hours spent using the Internet for social activities (IM’ing, chat rooms) are associated with decreased levels of depression and that informational uses and gaming are associated with increases in depression.

While total SNS use may not affect psychopathology, it may be related to subjective well-being. This was illustrated in the study by Kross et al [63], in which more frequent SNS use was related to experiencing more negative affect and reducing life satisfaction. As frequent experience of negative affect may contribute to the onset and maintenance of depression, it is likely that a pathway to poorer mental health outcomes exists via the impact SNS use has on the frequency of experiencing positive and negative emotions [54,63,67]. Additionally, other SNS features and cognitive processes (eg, network size, structure, and composition, tendency to ruminate, frequent social comparison) may be more informative in describing the impact frequent SNS use has on mental health.

In contrast with the literature examining social network size and structure offline [12,136], SNS friendship network size, on the whole, was not associated with depression or anxiety. However, some evidence has shown distinct network structure differences between individuals with depression and those without in terms of the interconnection between friends within a network [84]. Individuals with depression or anxiety have previously been shown to have more impoverished social networks, and changes in mental health are often associated with changes in an individual’s social network [12,137]. Impoverished social networks are often a risk factor for depression and anxiety by reducing access to “buffering” social support and increasing feelings of isolation [138-140]. They may also result from poor-quality social interactions, often typical of depression and anxiety [137].

The absence of a clear association between depression or anxiety and the number of friends on SNSs may be explained by one of the major differences between the offline and online social networks; that is, the way friendships are maintained over time. As SNSs do not necessitate direct social interaction to maintain the status of “friendship,” many users may not actively redefine their networks [141]. It is likely that the social pruning and the dissolution of social ties associated with mental illnesses such as depression and anxiety may not be visible on SNSs. Social pruning does occur for many SNS users (eg, 63% of American SNS users endorsed that they had removed friends from the “friends” list; [141]), but how comprehensively this behavior is performed remains unknown. Therefore, change in mental health status for SNS users may not be as accurately detected by a decreased social network size online as it may be when observing offline networks. Other metrics, such as communication output and reciprocity, may be more informative in describing the social network changes associated with depression and anxiety. For instance, De Choudhury et al [91] demonstrated that the volume of tweets and the associated replies were reduced in Twitter users with depression compared with those without.

As with any study, there are both strengths and limitations of this review. We included a basic criterion for bias that focused on evaluating the methodology of studies, which considered whether papers included (1) the use of psychometrically reliable and valid measures; (2) an external measurement criterion for mental health; and (3) description of sample demographics that included basic SNS user activity statistics. Only 9 studies were excluded for bias, suggesting that there is relative strength in defining the variables of interest in this field. However, a greater focus on defining the SNS characteristics of the sample is required.

The review attempted to characterize the research in terms of the populations and specific SNSs that have been studied. Studies have focused rather narrowly on the young adult population. While these individuals tend to represent the highest membership category of SNSs, recent estimates have suggested that SNS use is becoming more evenly represented across the life span, with more than 50% of older Internet users (65+ years) now also using SNSs [7]. This is an important consideration for future research as the social connection that may be gained through SNSs may provide more benefit for older users as quality of the interactions, particularly through language use, may vary significantly over the life span [142].

Despite the systematic approach to this review, the identified themes are not exhaustive. Other themes such as the differences between SNS users and nonusers and SNS use motives may have been extracted and more explicitly discussed. The discussion of results was limited to the depression or anxiety context and did not discuss findings outside this scope. Well-being, which clearly is becoming a growing area of interest (Figure 1), was only included if there was also a focus on depression or anxiety. Future studies might extend to other aspects of mental illness and wellness.

Finally, although we identified some moderating characteristics, few studies have considered individual differences such as gender and personality and their interaction with SNS variables. Future studies might give greater attention to how characteristics of users impact the identified factors.

The results of this systematic review have revealed considerable support for the importance of examining the content and quality of the interactions a user has with SNSs. As such, the language used in interactions on SNSs could become a target of interest, particularly as it has been shown to be sensitive in identifying individuals with depression [91,92,94,143]. Further research should also focus on the interplay between the network structure components and dynamic interactions observable on SNSs. The SNS friend structure could be instrumental in defining the type and efficiency with which social resources may be accessed on SNSs. Examining network structure in concert with the quality of interactions, characteristics such as perceived social support, and mental health could provide rich explanations for why some people benefit from SNS use and others are placed at risk, echoing the detailed social network research that has occurred offline (eg, [12]).

Only a few studies in this review utilized SNS-derived data to answer their research questions. The majority focused on the use of self-report survey and relied on participant estimates of their SNS behaviors, which may have introduced considerable retrospective bias. This bias was addressed to some extent by including ESMs that more accurately sample a participant’s lived experience [144]. The studies directly observing SNS behaviors indicate that the mental health status of SNS users may be at least partly derived from their patterns of use, language expression, and profile information. These findings provide more weight to the potential of using computational science techniques within psychological research, particularly in characterizing well-being in large community samples [33-35,145,146], as well as predicting personality [147]; see also [148]. In reference to depression and anxiety, SNS data hold huge potential for early identification and time-sensitive monitoring of symptoms [143]. SNS data should be leveraged in future research as a part of ESMs to provide real-time, unobtrusive accounts of social behavior in a natural setting.

This systematic review examined the recent research on associations between SNSs and depression and anxiety. It examined findings in association with the suggested mediators and moderators and the links made with well-being. With more than 50% of adults using multiple SNSs [7], they permeate many aspects of daily life. For many, SNSs represent a way to socially connect with others. However, for others, SNSs may encourage and perpetuate maladaptive tendencies. SNSs maintain and reflect the complexities of the offline social environment and the risks and benefits it may pose to mental health. SNSs represent a novel, unobtrusive, real-time way to observe and leverage mental health and well-being information in a natural setting, with the ultimate potential to positively influence mental health.