Social Media Use and Mental Health During the COVID-19 Pandemic: A Meta-Analysis of 14 Cross-Sectional Studies

Background: Public isolated due to the early quarantine regarding coronavirus disease 2019 (COVID-19) increasingly used more social media platforms. Contradictory claims regarding the effect of social media use on mental health needs to be resolved. We aimed to summarise the association between the time spent on social media platform during the COVID-19 quarantine and mental health outcomes (i.e., anxiety and depression). Methods: We searched 327 studies from the PubMed, Embase, and Cochrane Library databases. These studies evaluated the association between use of social media platform and mental health outcomes, including anxiety and depression, ascertained by various screening tools. We summarised the �nal 14 studies in terms of the odds ratios (ORs). Data analyses included a random-effect model and an assessment of inter-study heterogeneity. This meta-analysis review was registered in PROSPERO (https://www.crd.york.ac.uk/PROSPERO/, registration No CRD42021260223


Introduction
In the rst half-year of the virus pandemic, more than 100 million people were con rmed with coronavirus disease 2019 (COVID- 19), and approximately 500 thousand have died [1].Despite the widespread vaccination worldwide, the coronavirus outbreak is not expected to dampen due to the continuous emergence of new viral strains and di culty in effective quarantine interventions.As a result of strong quarantine measures, private meetings, gatherings, and physical contact with intimate relatives have been reduced [2].Prolonged social distancing and loss of intimate interpersonal contact increase feelings of frustration, boredom, anxiety, and potentially depression [3].Social media is reportedly considered as a channel of communication that could relieve negative aspects of isolation through helping people escape negative emotions [4], projecting their personality as they desire, and evoking the impression of gaining back some control [5].
Social media may be helpful for relieving anxiety and depression by providing information regarding the pandemic [6,7].However, prolonged use of social media by the isolated could be a double-edged sword that can adversely affect mental health due to sustained exposure to excessive information and misinformation [8][9][10].Due to con icting evidence and views regarding the effect of social media platform on the mental health, the recommendation for the use of social media in the era of COVID-19 has been questioned.Therefore, we conducted a meta-analysis to con rm the contradictory effects of social media platform on anxiety and depression based on studies reporting an association between the use of social media and mental health outcomes (i.e., anxiety and depression) on the pandemic setting.
Every article was independently evaluated by two researchers (YJJ and YRL), and a third researcher (SJJ) mediated the nal selection in case of differences in opinion.Study data were extracted by two independent researchers (YRL and YJJ).Studies were subdivided into categories according to the summary estimate of effect sizes (odds ratio [OR], beta estimate from multiple linear regression [β], and correlation coe cient [Pearson's r]).

Inclusion and exclusion criteria
Studies included data regarding the time spent on social media during the COVID-19 pandemic and various mental health symptoms (e.g., anxiety, depression, psychological distress, fear, panic, posttraumatic stress symptoms, rumination, vicarious traumatisation, life satisfaction, information avoidance, sense of inadequacy, and perceived societal risk).
The initial literature search was intended to include all types of observational study designs (cohort, casecontrol, and cross-sectional studies).The included studies were all cross-sectional studies.Case reports, letters, comments, and narrative reviews without quantitative results were not included.Studies were excluded when they examined traditional social media (television and radio).Studies without clear explanations of variable measurements were excluded.We included studies that investigated the association between time spent on social media and mental health outcomes (e.g., anxiety and depression) (see also Supplementary Material 1).While nal included studies presented their results in either OR, β, and/or Pearson's r (Supplementary material 1), the pooled effect size of this meta-analysis was mainly presented in an odds ratio.

Exposure variables
The nal 14 studies measured the amount of time spent on social media, which was either categorical or continuous variables (see Supplementary Material 1).It was measured based on the response to an item in the questionnaire: "How often were you exposed to social media?[categorical]" and "How long (in hours) were you exposed to social media?[continuous]" The exposure was measured in different wordings as follows: "Less" vs. "Frequently," "Less" vs. "Often", "less than 1 hour" vs. "2 hours or more," or "less than 3 hours" vs. "3 hours or more."To calculate the overall effect, we operationally rede ne these individually measured exposure levels (e.g., "Less" and "Few" were considered the same as "less than 2 hours;" "less than 1 hour," "Frequently," and "Often" were treated the same as "2 hours or more" and "3 hours or more").

Statistical analysis
Each included study reported summary statistics in different ways.Six studies that dealt with anxiety symptoms and seven with depression (Supplementary Material 1-1-1, 1-2-1) reported ORs and their 95% con dence intervals (CIs) (n = 9,579 and n = 13,241 for anxiety and depressive symptoms, respectively).Three studies each on anxiety and depression (Supplementary Material 1-1-2, 1-2-2) reported their ndings in β (n = 2,376 and n = 2,574 for anxiety and depression, respectively).The remaining studies reported Pearson's r (n = 2,483 for anxiety).We presented our pooled effect sizes of the meta-analysis in odds ratio and its 95% CIs (Figure 2), and other pooled statistics such as β and Pearson's r were presented in Supplementary material 2.
We used each estimate in the nal adjusted model.All statistical analyses and visualisations were performed with the "meta," "metaphor," and "dmeter" package of R version 3.6.3(https://cran.r-project.org/),using a random-effect model [16][17][18].We calculated the pooled effect sizes, Cochrane's Q, and I 2 to assess heterogeneity.The pooled effect sizes, CIs, and prediction intervals were calculated by estimating the pooled effect and CIs using the Hartung-Knapp-Sidik-Jonkman method, which is known as the one of the most conservative methods [19].The degree of heterogeneity was categorised as low, moderate, or high with threshold values of 25%, 50%, and 75%, respectively [20].Possible causes of heterogeneity among study results were explored by statistical methods such as in uential analysis, the Baujat plot, leave-one-out analysis, and Graphic Display of Heterogeneity analysis [21].In addition, publication bias was assessed using funnel plots, Egger's tests, and the trim-and-ll method [22].

Quality assessment
Quality assessment was conducted by two independent researchers, a psychiatrist (SHK) and an epidemiologist (YRL), using the Risk of Bias Assessment Tool for Nonrandomized Studies (RoBANS), which can assess cross-sectional studies [23].RoBANS has been validated with moderate reliability and good validity.RoBANS applies to cross-sectional studies and comprises six items: participant selection, confounding, exposure measurement, blinding of outcome assessments, missing outcomes, and selective reporting of outcomes.Each item is measured as having a "high risk of bias," "low risk of bias," or "uncertain."For example, based on "participant selection," each researcher marked an article as having a "high risk of bias" if, for example, the patient de nitions of depression were generated by self-reported data.In cross-sectional studies, misclassi cation cases due to an unreliable self-contained questionnaire for categorizing depressive patients were rated as "high risk."For the qualitative assessment, we classi ed a study as "low quality" if there were two or more "high risk of bias" grades.The study was rated as "high quality" only if the evaluation of both raters was congruent.For sensitivity analysis, we repeated our analysis including only "high quality" studies and compared it with the pooled estimates of overall results.

Ethical approval
No ethical approval and patient consent are required since this study data is based on published literature.This meta-analysis review was registered with PROSPERO (https://www.crd.york.ac.uk/PROSPERO/, registration No CRD42021260223, 15 June 2021).

Included and excluded studies
We selected 346 studies from the database search (288 from PubMed, 34 from Embase, and 24 from the Cochrane Library).After removing 19 duplicate publications, 327 studies were included for the title and full-text review.Non-original studies and those conducted with irrelevant subjects (n = 218) were excluded.Another 95 studies were excluded nally due to inconsistent study estimates.Each of the nal 14 studies measured multiple mental health outcome variables (i.e., anxiety and depression), and pooled effect sized were calculated for each outcome variable.Figure 1 shows the number of included publications for each mental health outcome.All included studies were cross-sectional studies.The pooled effect size was presented in odds ratio.
Of studies ascertaining anxiety, six studies used the GAD, three used the DASS-21, and the rest were measured by PHQ-9 and GHQ-28.Of the 9 papers regarding depression, ve studies used PHQ-9 (cut-off: 10+), three used DASS-21, and the rest used Chinese version of WHO-5.Of the seven papers on psychological distress, DASS-21 was used in four papers for measuring stress (see also Supplementary Material 1 for more details).

Publication bias
Publication bias was assessed by funnel plot analysis and Egger's test (Supplementary Material 4-1).Funnel-plot analyses revealed symmetrical results (Supplementary Material 4-2).In addition, all results of the Egger test were statistically insigni cant, indicating improbable publication bias.After applying the trim-and-ll method, the funnel plot revealed no asymmetry (Supplementary Material 5), indicating no signi cant publication bias.

Discussion
Our study aimed to present a comprehensive direction of relevance by analysing studies investigating the association between time spent on social media during the COVID-19 pandemic and symptoms of mental health indices (e.g., anxiety, depressive, and psychological distress) among the public.By summarising the different estimates (ORs, β, and Pearson's r) presented in the cross-sectional studies, we found that time spent on social media was associated with an increased likelihood of symptoms of anxiety and depression.However, no signi cant association was seen with psychological distress.
Our results are consistent with previous systematic reviews and meta-analysis performed before the COVID-19 pandemic.A systematic literature review before the COVID-19 outbreak (2019) found that the time spent by adolescents on social media was associated with depression, anxiety, and psychological distress [24].A meta-analysis of 11 studies (2017) also reported a weak association between social media use and depressive symptoms in children [25].A meta-analysis of 23 studies (2018) reported signi cant correlation between social media use and psychological distress [26].Likewise, our study also observed a similar trend of a negative effect of social media on mental health outcomes in the COVID-19 pandemic.However, the estimates of inter-study heterogeneity of these meta-analysis were relatively high (meta-analysis of 11 studies: I 2 = 92.4%;meta-analysis of 23 studies: I 2 = 62.00% for anxiety, I 2 = 80.58% for depression) compared to our analysis, which implies relative homogeneity of our study population and reliable results.
Despite the bene ts of social media, unveri ed information is freely disseminated on social media, while opinions are perceived as facts.There has been a stream of news regarding the pandemic, creating a sense of urgency and anxiety.Repeated exposure to this stream may affect the construct of external reality and may lead to a delusion-like experience, which has been linked to anxiety and social media overuse [27,28].
Additionally, discrimination and stigma related to COVID-19 on social media can make people fearful of infection and exacerbate depression and anxiety [29].Fear of COVID-19 may be compounded by coexisting depression and anxiety disorders [30].Due to the high accessibility of social media and the ease of socialisation in a controlled setting, individuals with underlying depression may be more drawn to social media interactions rather than face-to-face ones, more so in the pandemic era [31].
In interpreting the ndings of this study, several limitations should be considered.First, all the studies included were cross-sectional design.The possibility of a reverse causal relationship cannot be ruled out.
The COVID-19 pandemic has been around for less than a year and longitudinal studies might not be under the publication process at the time of study conducted.This may explain to some extent why all included studies were cross-sectional studies.Further studies with longitudinal data are warranted.Second, our results do not represent the general population since most of the studies recruited participants through a web-based survey, which may have had a selection bias.Lastly, some of the analysis showed a relatively high inter-study heterogeneity (range: I 2 = 0.00%-80.53%).To identify the cause of heterogeneity, we summarised the results of the statistical approaches (in uential analysis, Baujat plot, leave-one-out analysis, and GOSH analysis [21]) in Supplementary Material 6 and 7.
Despite these limitations, our study exhibits a number of strengths; as far as we understand, our study is the rst meta-analysis to examine the relationship between social media use and mental health outcomes during the COVID-19 pandemic, to validate the results by various veri cation methods such as trim-and-ll methods, in uential analysis, and heterogeneity analysis.In addition, various sensitivity analyses such as quality assessment were also attempted.
Our analysis demonstrates that excessive time spent on social media is associated with increased anxiety and depressive symptoms in the era of COVID-19.While social media may be considered as an alternative for alleviating some stress caused by the pandemic, our ndings suggest that excessive use of social media can be detrimental for mental health.A longitudinal study to determine the true effect of social media is required.

c
Beta value calculated by linear regression analysis indicates the change in score over time in hours.d Correlation analysis Β, Beta value; CI, Con dence Interval; DASS, Subscale scores of Depression, Anxiety, and Stress Scale; GAD, Generalized Anxiety Disorder; GHQ-28, General Health Questionnaire-28; PHQ, Patient Health Questionnaire; SAS-Self-Rating Anxiety Scale * Signi cant results are in bold text.

Table 1 .
Association between social media use and anxiety a and depression b