Customize your JAMA Network experience by selecting one or more topics from the list below.
Identify all potential conflicts of interest that might be relevant to your comment.
Conflicts of interest comprise financial interests, activities, and relationships within the past 3 years including but not limited to employment, affiliation, grants or funding, consultancies, honoraria or payment, speaker's bureaus, stock ownership or options, expert testimony, royalties, donation of medical equipment, or patents planned, pending, or issued.
Err on the side of full disclosure.
If you have no conflicts of interest, check "No potential conflicts of interest" in the box below. The information will be posted with your response.
Not all submitted comments are published. Please see our commenting policy for details.
Perlis RH, Green J, Simonson M, et al. Association Between Social Media Use and Self-reported Symptoms of Depression in US Adults. JAMA Netw Open. 2021;4(11):e2136113. doi:10.1001/jamanetworkopen.2021.36113
Is social media use by adults associated with subsequent increases in depressive symptoms?
In this survey study, 5395 individuals with minimal depressive symptoms on initial survey who reported use of Snapchat, Facebook, or TikTok were more likely to report increased levels of depressive symptoms on a later survey.
These results suggest that certain social media use preceded worsening of depressive symptoms.
Some studies suggest that social media use is associated with risk for depression, particularly among children and young adults.
To characterize the association between self-reported use of individual social media platforms and worsening of depressive symptoms among adults.
Design, Setting, and Participants
This survey study included data from 13 waves of a nonprobability internet survey conducted approximately monthly between May 2020 and May 2021 among individuals aged 18 years and older in the US. Data were analyzed in July and August 2021.
Main Outcomes and Measures
Logistic regression was applied without reweighting, with a 5 point or greater increase in 9-item Patient Health Questionnaire (PHQ-9) score as outcome and participant sociodemographic features, baseline PHQ-9, and use of each social media platform as independent variables.
In total, 5395 of 8045 individuals (67.1%) with a PHQ-9 score below 5 on initial survey completed a second PHQ-9. These respondents had a mean (SD) age of 55.8 (15.2) years; 3546 respondents (65.7%) identified as female; 329 respondents (6.1%) were Asian, 570 (10.6%) Black, 256 (4.7%) Hispanic, 4118 (76.3%) White, and 122 (2.3%) American Indian or Alaska Native, Pacific Islander or Native Hawaiian, or other. Among eligible respondents, 482 (8.9%) reported 5 points or greater worsening of PHQ-9 score at second survey. In fully adjusted models for increase in symptoms, the largest adjusted odds ratio (aOR) associated with social media use was observed for Snapchat (aOR, 1.53; 95% CI, 1.19-1.96), Facebook (aOR, 1.42; 95% CI, 1.10-1.81), and TikTok (aOR, 1.39; 95% CI, 1.03-1.87).
Conclusions and Relevance
Among survey respondents who did not report depressive symptoms initially, social media use was associated with greater likelihood of subsequent increase in depressive symptoms after adjustment for sociodemographic features and news sources. These data cannot elucidate the nature of this association, but suggest the need for further study to understand how social media use may factor into depression among adults.
Social media use has been associated with diminished well-being and greater levels of anxiety and depression, predominantly in cross-sectional studies among adolescents1 or young adults,2 although concern has been raised that reporting bias may result from individuals with greater depressive symptoms overreporting social media use.3 A small number of short-term longitudinal studies provide further support for this association—for example, among 82 young adults sampled 2 weeks apart.4,5
These results have 2 notable gaps that raise questions. First, is the observed cross-sectional relationship also apparent in longitudinal studies, as suggested by a randomized trial of Facebook discontinuation?6 Second, does this risk apply to older consumers of social media? To investigate these questions, we used data from multiple waves of an ongoing 50-state US survey.
Data were drawn from 13 waves of a nonprobability internet survey deployed via a multipanel commercial vendor approximately every month between May 2020 and May 2021 in individuals aged 18 years and older. The study was approved by the institutional review board of Harvard University, and participants provided written informed consent. Reporting followed the American Association for Public Opinion Research (AAPOR) reporting guideline.
The survey sampling strategy incorporated quotas for sex, age, and race and ethnicity within each state; attention checks and open-ended answers were used to filter out unreliable respondents. Race and ethnicity was self-reported from 5 census categories to confirm representativeness of the US population. Participants also completed the 9-item Patient Health Questionnaire (PHQ-9).7 They were asked, “Do you ever use any of the following social media sites or apps?”; we focused a priori on Facebook, Instagram, LinkedIn, Pinterest, TikTok, Twitter, Snapchat, and YouTube. They were further asked to identify any sources of COVID-19 related news they consumed in the past 24 hours (ie, cable television, network television, or news website), which we used as a proxy for news sources more generally, while examining web-based vs television-based news separately; number of social supports available “to talk to if you had a problem, felt sad, or depressed”; and face-to-face meetings with nonhousehold members in the prior 24 hours. As some survey modules were randomly assigned to a subset of participants in each wave to diminish overall survey length, 1646 returning participants completed news source questions, and 4808 completed social supports questions (Table 1).
The study cohort included participants who completed the PHQ-9 in at least 2 survey waves with a PHQ-9 total score less than 5 (ie, less than mild depression) at the index survey. We applied logistic regression for participants with a 5 or more point increase in PHQ-9 score (ie, the threshold for clinical significance7) as outcome, and for sociodemographic features, baseline PHQ-9, and use of each social media platform as independent variables using glm in R version 3.6 (R Project for Statistical Computing) without reweighting. The threshold for statistical significance was set at 2-sided P < .05.
Overall, 5395 of 8045 individuals (67.1%) with a PHQ-9 score below 5 on initial survey completed a subsequent PHQ-9 (Table 1). These respondents had mean (SD) age of 55.8 (15.2) years; 3546 respondents (65.7%) identified as female; 329 respondents (6.1%) were Asian, 570 (10.6%) Black, 256 (4.7%) Hispanic, 4118 (76.3%) White, and 122 (2.3%) American Indian or Alaska Native, Pacific Islander or Native Hawaiian, or other. Mean (SD) PHQ-9 score at initial survey for this group of subsequent responders was 1.29 (1.43). At first follow-up survey, 482 respondents (8.9%) experienced an increase in PHQ-9 score of 5 points or greater.
In adjusted regression models, Snapchat, Facebook, and TikTok use at first survey were significantly associated with greater risk of increase in self-reported depressive symptoms, with adjusted odds ratios (aOR) of 1.53 (95% CI, 1.19-1.96), 1.42 (95% CI, 1.10-1.81), and 1.39 (95% CI, 1.03-1.87), respectively (Table 2). Incorporating terms for television or internet news consumed in the past 24 hours, number of social supports, and number of daily face-to-face interactions on the initial survey did not meaningfully change these associations with the notable exception of Snapchat, where aORs were diminished from 1.53 (95% CI, 1.19-1.96) to 1.12 (95% CI, 0.70-1.81) with the inclusion of terms for news source (Table 2).
In logistic regression models for increase in depressive symptoms, we further identified significant interactions of platform use with age group for Facebook, TikTok, and Snapchat (Table 3). For TikTok and Snapchat, use was associated with depressive symptoms among those ages 35 years or older but not among those younger than age 35 years (eg, Snapchat: aOR, 1.96; 95% CI, 1.44-2.65 vs aOR, 1.17; 95% CI, 0.78-1.77). For Facebook, the opposite pattern was observed; use was associated with depressive symptoms among those younger than 35 years, but not among those aged 35 years and older (aOR, 2.60; 95% CI, 1.46-1.62 vs aOR, 1.12; 95% CI, 0.85-1.48).
In this analysis of survey data, we found that some forms of social media use—in particular, Snapchat, Facebook, and YouTube—were associated with greater levels of self-reported depressive symptoms on a subsequent survey. Notably, with the exception of Snapchat, this association was not explained in regression models by use of other news sources, suggesting it is not accounted for by differential media consumption more broadly. Likewise, the association was not meaningfully changed by number of social supports or face-to-face social interactions at baseline, suggesting it is not mediated by reduction in social interactions more broadly.
While we cannot test causation, as in a prior study of Facebook discontinuation,6 our design allowed us to look at incident self-reported depression following social media use. Our results extend prior observations of elevated depressive symptoms in adolescents1 or young adults2 associated with social media use in cross-sectional studies. They are also consistent with 2 prior short-term longitudinal studies.4,5 Beyond the ability to look longitudinally, this study also extends prior studies by examining a broader age range, with a mean age of 56 years. Our results suggest that the associations previously observed with depressive symptoms are not limited to young adults; indeed, while effect sizes for Facebook are greater among younger adults, effect sizes for TikTok and Snapchat are greater among those age 35 years or older.
This study had several limitations. Results are limited by the inability to control for all potential confounding, lack of dose-response data, and inability to measure nature of social media use, which has been suggested may moderate impact.8 Notably, social media use may simply be a marker of underlying vulnerability to depression. On the other hand, one investigation found that viewing one’s own Facebook profile was associated with increased physiological stress response.9 The extent to which these findings generalize beyond the COVID-19 pandemic period also remains to be determined; it may be that the effects of social media use are specific to content viewed, for example. A further caveat is that only a (nonrandom) subset of participants return for a second survey, which could yield a biased sample with less generalizability. However, as the surveys are not labeled as COVID-19 specific, the likelihood is lower that these results reflect a selection bias in which individuals with greater interest in, or who had been more affected by, COVID-19 preferentially responded.
In this survey study of US adults, we identified associations between type of social media use at initial survey and greater levels of depressive symptoms on a subsequent survey. Our findings complement those using cross-sectional designs2 or small longitudinal cohorts4,5 and extend them to older adults, indicating the need for further investigation of the relationship between social media use and mental health.
Accepted for Publication: September 28, 2021.
Published: November 23, 2021. doi:10.1001/jamanetworkopen.2021.36113
Open Access: This is an open access article distributed under the terms of the CC-BY License. © 2021 Perlis RH et al. JAMA Network Open.
Corresponding Author: Roy H. Perlis, MD, MSc, Massachusetts General Hospital, 185 Cambridge St, 6th Floor, Boston, MA 02114 (firstname.lastname@example.org).
Author Contributions: Dr Perlis had full access to all of the data in the study and takes responsibility for the integrity of the data and the accuracy of the data analysis.
Concept and design: Perlis, Druckman, Della Volpe.
Acquisition, analysis, or interpretation of data: All authors.
Drafting of the manuscript: Perlis, Della Volpe.
Critical revision of the manuscript for important intellectual content: Perlis, Green, Simonson, Ognyanova, Santillana, Lin, Quintana, Chwe, Druckman, Lazer, Baum.
Statistical analysis: Perlis, Green, Santillana, Della Volpe.
Obtained funding: Ognyanova, Druckman, Lazer, Baum.
Administrative, technical, or material support: Perlis, Simonson, Lin, Quintana, Chwe, Druckman, Lazer.
Supervision: Perlis, Quintana, Lazer.
Conflict of Interest Disclosures: Dr Perlis reported receiving consulting fees from Belle Artificial Intelligence, Burrage Capital, Genomind, RID Ventures, and Takeda; he reported holding equity in Psy Therapeutics. Dr Ognyanova reported grants from the National Science Foundation during the conduct of the study. No other disclosures were reported.
Funding/Support: This study was supported by the National Institute of Mental Health (grant No. R01MH116270; Dr Perlis).
Role of the Funder/Sponsor: The funder had no role in the design and conduct of the study; collection, management, analysis, and interpretation of the data; preparation, review, or approval of the manuscript; and decision to submit the manuscript for publication.
Disclaimer: Dr Perlis is an associate editor for JAMA Network Open but was not involved in the editorial review or decision process for this manuscript.