This is an open-access article distributed under the terms of the Creative Commons Attribution License (https://creativecommons.org/licenses/by/4.0/), which permits unrestricted use, distribution, and reproduction in any medium, provided the original work, first published in JMIR Public Health and Surveillance, is properly cited. The complete bibliographic information, a link to the original publication on https://publichealth.jmir.org, as well as this copyright and license information must be included.
Social media has become a ubiquitous part of daily life during the COVID-19 pandemic isolation. However, the role of social media use in depression and suicidal ideation of the general public remains unclear. Related empirical studies were limited and reported inconsistent findings. Little is known about the potential underlying mechanisms that may illustrate the relationship between social media use and depression and suicidal ideation during the COVID-19 pandemic.
This study tested the mediation effects of social loneliness and posttraumatic stress disorder (PTSD) symptoms on the relationship between social media use and depressive symptoms and suicidal ideation, as well as the moderation effect of age on the mediation models.
We administered a population-based random telephone survey in May and June 2020, when infection control measures were being vigorously implemented in Hong Kong. A total of 1070 adults (658 social media users and 412 nonusers) completed the survey. Structural equation modeling (SEM) and multigroup SEM were conducted to test the mediation and moderation effects.
The weighted prevalence of probable depression was 11.6%; 1.6% had suicidal ideation in the past 2 weeks. Both moderated mediation models of depressive symptoms (
Social media may be a “double-edged sword” for psychosocial well-being during the COVID-19 pandemic, and its roles vary across age groups. The mediators identified in this study can be addressed by psychological interventions to prevent severe mental health problems during and after the COVID-19 pandemic.
Unprecedented control measures, such as lockdown, quarantine, social distancing, and home confinement, have been implemented to contain the spread of COVID-19, an infectious disease caused by a coronavirus that was newly discovered in 2019 [
Reduced social loneliness and increased trauma-related stress (eg, posttraumatic stress disorder [PTSD] symptoms) may serve as two important psychosocial mechanisms that explain the relationship between social media use and depression and suicidal ideation. Social loneliness refers to the situation where an individual has a smaller number of relationships, stemming from the absence of a broader group of contacts or an engaging social network, such as friends, colleagues, and neighbors. Meanwhile, social media has been demonstrated to play an important role in forming and maintaining social networks and social capital, which may counter social loneliness [
On the other hand, intensified trauma-related stress and PTSD symptoms may explain the positive association between social media use and mental health (eg, Chao et al [
Theoretically, the proposed mediation effects of social loneliness and PTSD symptoms can be supported by the conservation of resources (COR) theory [
Furthermore, an increasing number of older adults have been using the internet and social media. In Hong Kong, people aged 45 years or above have caught up rapidly with their social media participation rate (ie, 78% in 2018) [
In this study, we randomly recruited both social media users and nonusers to create a representative sample of the Hong Kong population; we examined whether and how social media use is associated with depressive symptoms and suicidal ideation through two psychosocial processes: social loneliness and PTSD symptoms. We hypothesized the following:
Social media use would be negatively associated with social loneliness; in turn, loneliness would be positively associated with depressive symptoms and suicidal ideation.
Social media use would be positively associated with PTSD symptoms; in turn, PTSD symptoms would be positively associated with depressive symptoms and suicidal ideation.
In addition, we also tested whether these mediation effects would be constant among younger and older people.
We administered a population-based, random telephone survey between May 14 and June 4, 2020, when infection control measures (eg, social distancing, business restrictions, and border control) were being vigorously implemented in Hong Kong. Participant inclusion criteria included the following: (1) Chinese speaking, (2) 18 years old or above, and (3) Hong Kong resident (ie, holder of a Hong Kong identification card). The telephone interviews were conducted between 6 PM and 10 PM in order to avoid undersampling working individuals. The interviewers were well trained and had at least 6 months of interviewing experience. They were supervised on site by a senior project coordinator. Telephone numbers were randomly drawn from the latest residential telephone directory by a random phone number generator program. Telephone numbers were selected randomly from an updated landline telephone directory as seed numbers. Another three sets of numbers were then generated using the randomization of the last two digits to recruit unlisted numbers. Eligible household members whose day and month of birth was closest to the survey date were invited to join the study. Two follow-up calls were conducted for unanswered calls before a telephone number was considered invalid. Verbal informed consent was obtained from the participants. The anonymous interview took 10 to 15 minutes. No incentive was given to the participants. Of the 1882 eligible participants identified and invited, 1070 completed the interviews, resulting in a modest response rate (56.9%).
The study was approved by the Survey and Behavioural Research Ethics Committee of the corresponding author’s affiliated university, the Chinese University of Hong Kong (reference No. SBRE-19-645). The study followed the ethical standards of the responsible committee on human experimentation, institutional and national, and of the Helsinki Declaration of 1975, as revised in 2000.
The participants were asked whether they had used an SNS in the past 12 months, such as Facebook, Twitter, WhatsApp, or WeChat, which are platforms for communicating with one another [
The 8-item Posttraumatic Stress Disorder scale (PTSD-8) [
The 3-item social loneliness subscale of the De Jong Gierveld Loneliness Scale [
The 10-item Center for Epidemiologic Studies Depression Scale (CESD-10) assessed depressive symptoms during the past week [
Item 9 of the 9-item Patient Health Questionnaire (PHQ-9) [
The participants were also asked to report their sociodemographic information, including sex, age, current marital status, educational level, income, health status, and mandatory quarantine status (ie, whether one had been subjected to compulsory quarantine at designated places—home, hotel, or other accommodation—under government order for COVID-19 infection control).
Descriptive statistics were computed for both background and psychological variables. Age-standardized weighted prevalence of probable depression was calculated by the direct method and the age distribution for the 2020 census population. Simple logistic regression analyses were conducted to test the associations between background, independent, and mediation variables and probable depression and suicidal ideation. Odds ratios and 95% CIs were reported. Structural equation modeling (SEM) was conducted to test the proposed mediation models of depressive symptoms and suicidal ideation. For the variables of PTSD symptoms and depressive symptoms, indicators were created by the item parceling method. Since the two scales are unidimensional, the random method of combining items was used to create item parcels. For the latent factors of social loneliness, all three individual items of the scale were used as indicators. The observed variable of suicidal ideation was created by using item 9 of the PHQ-9. Goodness of fit was tested by using the chi-square test, the comparative fit index (CFI), the nonnormed fit index (NNFI), and the root mean square error of approximation (RMSEA). Standardized regression coefficients (β) and 95% CIs were reported. Bootstrapping based on 5000 bootstrap samples was performed to test for indirect effects. A statistically significant indirect effect would be observed when the CI did not include zero. Multigroup SEM analyses were conducted to test the moderation effect of age on the mediation models. The age of retirement of most people in Hong Kong ranges from 55 to 65 years [
The background characteristics of the participants are presented in
Associations between the background or mediator variables and probable depression and suicidal ideation are presented in
Background characteristics of the participants recruited from the adult population in Hong Kong during the COVID-19 pandemic.
Background characteristic | All participants (N=1070) | Younger adults |
Older adults |
|||
|
.01 | |||||
|
Male | 346 (32.3) | 137 (37.3) | 204 (29.8) |
|
|
|
Female | 724 (67.7) | 230 (62.7) | 480 (70.2) |
|
|
|
N/Ab | |||||
|
18-35 | 115 (10.7) | N/A | N/A |
|
|
|
36-55 | 252 (23.6) | N/A | N/A |
|
|
|
56-65 | 301 (28.1) | N/A | N/A |
|
|
|
>65 | 383 (35.8) | N/A | N/A |
|
|
|
Refused to answer | 19 (1.8) | N/A | N/A |
|
|
|
<.001 | |||||
|
Single | 201 (18.8) | 148 (40.3) | 50 (7.3) |
|
|
|
Cohabiting or married | 745 (69.6) | 209 (56.9) | 528 (77.2) |
|
|
|
Separated, divorced, or widowed | 107 (10.0) | 5 (1.4) | 101 (14.8) |
|
|
|
Refused to answer or missing value | 17 (1.6) | 5 (1.4) | 5 (0.7) |
|
|
|
<.001 | |||||
|
Primary school or below | 355 (33.2) | 7 (1.9) | 345 (50.4) |
|
|
|
Secondary school | 384 (35.9) | 139 (37.9) | 239 (34.9) |
|
|
|
College or above | 294 (27.5) | 211 (57.5) | 81 (11.8) |
|
|
|
Refused to answer | 37 (3.5) | 10 (2.7) | 19 (2.8) |
|
|
|
<.001 | |||||
|
≤20,000 | 627 (58.6) | 92 (25.1) | 526 (76.9) |
|
|
|
20,001-30,000 | 124 (11.6) | 77 (21.0) | 47 (6.9) |
|
|
|
30,001-50,000 | 103 (9.6) | 64 (17.4) | 39 (5.7) |
|
|
|
>50,000 | 84 (7.9) | 62 (16.9) | 22 (3.2) |
|
|
|
Refused to answer or missing value | 132 (12.3) | 72 (19.6) | 50 (7.3) |
|
|
|
<.001 | |||||
|
No | 716 (66.9) | 334 (91.0) | 369 (53.9) |
|
|
|
Yes | 354 (33.1) | 33 (9.0) | 315 (46.1) |
|
|
|
.25 | |||||
|
No | 1045 (97.7) | 361 (98.4) | 665 (97.2) |
|
|
|
Yes | 25 (2.3) | 6 (1.6) | 19 (2.8) |
|
|
|
.64 | |||||
|
No | 1050 (98.1) | 361 (98.4) | 670 (98.0) |
|
|
|
Yes | 20 (1.9) | 6 (1.6) | 14 (2.0) |
|
|
|
.02 | |||||
|
No | 1055 (98.6) | 358 (97.5) | 679 (99.3) |
|
|
|
Yes | 15 (1.4) | 9 (2.5) | 5 (0.7) |
|
|
|
<.001 | |||||
|
No | 412 (38.5) | 25 (6.8) | 379 (55.4) |
|
|
|
Yes | 658 (61.5) | 342 (93.2) | 305 (44.6) |
|
|
|
<.001 | |||||
|
0.0-2.0 | 232 (35.3) | 89 (26.0) | 137 (44.9) |
|
|
|
2.5-4.0 | 261 (39.7) | 133 (38.9) | 126 (41.3) |
|
|
|
4.5-6.0 | 110 (16.7) | 77 (22.5) | 30 (9.8) |
|
|
|
>6.0 | 55 (8.4) | 43 (12.6) | 12 (3.9) |
|
|
Posttraumatic stress disorder symptoms, mean (SD)f | 4.2 (3.8) | 4.7 (3.6) | 4.0 (3.9) | .008 | ||
Social loneliness, mean (SD)g | 4.4 (1.7) | 4.4 (1.8) | 4.3 (1.7) | .76 |
aBased on chi-square tests or independent-samples
bN/A: not applicable; the number of participants in each age group was reported for the total sample only (hence, the
cA currency exchange rate of HK $1=US $0.1287 is applicable.
dChronic diseases included hypertension, diabetes, cancer, etc.
eMental health problems included depression, anxiety, insomnia, etc.
fThe 8-item Posttraumatic Stress Disorder scale (PTSD-8) was used to assess posttraumatic stress responses and symptoms in the past month. Summed scores range from 0 to 23; higher summed scores indicate greater symptoms of PTSD.
gThe 3-item social loneliness subscale of the De Jong Gierveld Loneliness Scale was used to assess social loneliness. Summed scores range from 3 to 9; higher scores suggest higher levels of loneliness.
Associations between the background or mediator variables and depressive symptoms and suicidal ideation among the adult population in Hong Kong during the COVID-19 pandemic (N=1070).
Variable | Probable depression |
Suicidal ideation |
|||
|
|||||
|
Male | 1b |
|
1 |
|
Female | 1.09 (0.71-1.67) | .69 | 0.68 (0.26-1.80) | .44 | |
|
|||||
|
18-35 | 1 |
|
1 |
|
|
36-55 | 0.90 (0.49-1.66) | .73 | 0.68 (0.19-2.45) | .55 |
|
56-65 | 0.55 (0.29-1.04) | .07 | 0.37 (0.09-1.52) | .17 |
|
>65 | 0.39 (0.21-0.75) | .004 | 0.22 (0.05-0.99) | .049 |
|
|||||
|
Single | 1 |
|
1 |
|
|
Cohabiting or married | 0.61 (0.39-0.96) | .03 | 0.53 (0.18-1.58) | .26 |
|
Separated, divorced, or widowed | 0.46 (0.20-1.04) | .06 | 0.75 (0.14-3.91) | .73 |
|
|||||
|
Primary school or below | 1 |
|
1 |
|
|
Secondary school | 1.41 (0.83-2.39) | .21 | 2.18 (0.56-8.49) | .26 |
|
College or above | 2.45 (1.46-4.09) | .001 | 2.86 (0.73-11.17) | .13 |
|
|||||
|
≤20,000 | 1 |
|
1 |
|
|
20,001-30,000 | 2.41 (1.38-4.18) | .002 | 1.45 (0.30-7.07) | .64 |
|
30,001-50,000 | 1.56 (0.80-3.04) | .20 | 2.66 (0.68-10.45) | .16 |
|
>50,000 | 2.78 (1.50-5.15) | .001 | 4.43(1.27-15.46) | .02 |
|
|||||
|
No | 1 |
|
1 |
|
|
Yes | 0.84 (0.55-1.29) | .43 | 0.84 (0.29-2.41) | .75 |
|
|||||
|
No | 1 |
|
1 |
|
|
Yes | 12.44 (5.50-28.15) | <.001 | 15.12 (4.55-50.26) | <.001 |
|
|||||
|
No | 1 |
|
1 |
|
|
Yes | 40.21 (13.18-122.72) | <.001 | 28.83 (9.03-92.08) | <.001 |
|
|||||
|
No | 1 |
|
1 |
|
|
Yes | 2.19 (0.61-7.89) | .23 | 10.67 (2.21-51.45) | .003 |
|
|||||
|
No | 1 |
|
1 |
|
|
Yes | 1.98 (1.27-3.10) | .003 | 2.96 (0.85-10.38) | .09 |
|
|||||
|
0.0-2.0 | 1 |
|
1 |
|
|
2.5-4.0 | 1.29 (0.72-2.29) | .39 | 0.29 (0.08-1.08) | .06 |
|
4.5-6.0 | 2.12 (1.10-4.08) | .02 | 0.23 (0.03-1.82) | .16 |
|
>6.0 | 2.12 (0.94-4.79) | .07 | 0.46 (0.06-3.70) | .46 |
Posttraumatic stress disorder symptoms | 1.38 (1.31-1.46) | <.001 | 1.21 (1.11-1.31) | <.001 | |
Social loneliness | 1.17 (1.05-1.30) | .004 | 1.20 (0.93-1.54) | .16 |
aOR: odds ratio; based on logistic regression analyses.
bVariable items with a value of 1 are the reference items.
cA currency exchange rate of HK $1=US $0.1287 is applicable.
dChronic diseases included hypertension, diabetes, cancer, etc.
eMental health problems included depression, anxiety, insomnia, etc.
Both the measurement model (
The proposed mediation model of depressive symptoms with standardized regression coefficients (β) among younger adults in Hong Kong during the COVID-19 pandemic (n=367).
The proposed mediation model of depressive symptoms with standardized regression coefficients (β) among older adults in Hong Kong during the COVID-19 pandemic (n=684).
Both the measurement model (
The proposed mediation model of suicidal ideation with standardized regression coefficients (β) among younger adults in Hong Kong during the COVID-19 pandemic (n=367).
The proposed mediation model of suicidal ideation with standardized regression coefficients (β) among older adults in Hong Kong during the COVID-19 pandemic (n=684).
This population-based study investigated the prevalence of probable depression and suicidal ideation in Hong Kong adults during the COVID-19 pandemic. Furthermore, the study tested a complex relationship between social media use and mental health (ie, depressive symptoms and suicidal ideation). Findings suggest that this relationship could be explained by the mediation effects of PTSD symptoms and social loneliness and moderated by age. Specifically, social media use was indirectly and positively associated with depressive symptoms and suicidal ideation through PTSD symptoms in both younger and older adults. In addition, social media use was directly and indirectly associated with depressive symptoms through social loneliness in older adults.
The weighted prevalence of probable depression was higher than that reported in pre–COVID-19 research conducted in 2007 (11.6% versus 8.6%) [
Furthermore, this study brings novel information to the field about the underlying mechanisms of the relationships between social media use and depression and suicidal ideation. The proposed mediation models based on the COR theory were well supported by the acceptable model fit. This is the first study that applied this theory to understand the roles of social media use in the context of the COVID-19 pandemic. From the COR perspectives, people strive to develop, maintain, or restore important resources, such as social relationships, well-being, and a low state of stress, and a loss of these resources can, in turn, lead to mental health problems [
On the other hand, social media use might indirectly reduce depression because it can provide opportunities to maintain and enhance interpersonal resources (eg, reduced social loneliness) by using SNSs, such as Facebook, WeChat, and WhatsApp, during the COVID-19 pandemic isolation. This result provides preliminary empirical evidence for the assertion of recently published commentaries [
Unexpectedly, social loneliness was not a significant mediator between social media use and suicidal ideation because it was not significantly associated with suicidal ideation. Inconsistently, previous studies found that loneliness and social connection were significant interpersonal factors of suicidal ideation [
The positive mediation effect of PTSD symptoms among younger and older people and the negative mediation effect of social loneliness among older people suggest that social media use may have both beneficial and harmful effects on mental health during the COVID-19 pandemic, and that age plays a significant role. Notably, the mediation effect of PTSD symptoms was larger than that of social loneliness. This is consistent with the principle of the COR theory, in that resource loss is disproportionately more salient than resource gain [
Such results have important practical and political implications. First, the high prevalence of mental health problems during the COVID-19 pandemic is a significant public health concern, and high-risk groups (eg, younger people) need particular attention from health care service providers. From a public health perspective, there are effective mental health interventions (eg, cognitive behavioral therapy and mindfulness-based interventions) available, which can be delivered via the internet during the COVID-19 pandemic isolation. Second, since the COVID-19 pandemic might persist and a digital lifestyle could become inevitable, it is important to understand the psychological mechanisms that may explain how digital technology users and nonusers may be different in psychosocial status and mental health. Our studied psychosocial mediators can be modified by interventions and can be used to guide prevention programs for mental health problems. For example, health education programs and public health strategies are recommended to enhance awareness of digital literacy, strategic social media use, and potential harms of social media use in the general public to reduce their trauma-related stress. A large-scale online relaxation training program is also feasible to help the general public manage their trauma-related stress [
This study has several limitations. First, it was cross-sectional in nature. It is plausible that people with the greatest concerns and depressive symptoms may be more likely to seek out media coverage of the event. Longitudinal studies to monitor the trajectories of social media use and psychological responses are warranted. Second, we recruited the participants via landline telephone numbers, and this sampling method might exclude those without landline telephones or those who were not at home during the survey period (eg, younger adults who are more likely to use mobile phones and less likely to have landline telephones). Thus, this sampling method might also have influenced the representativeness of the sample. Third, this study only focused on the use of SNSs that people use to build social networks or social relationships with other people. We did not investigate the content and functions of the SNSs in this study, or those of other types of SNSs, that may cause different psychological responses to the COVID-19 pandemic [
The findings suggest that social media may be a “double-edged sword” for psychosocial well-being during the COVID-19 pandemic and its roles vary across age groups. The mediators identified in this study should be further validated through qualitative inquiry and longitudinal cohort studies and can be addressed by psychological interventions to prevent severe mental health problems.
10-item Center for Epidemiologic Studies Depression Scale
comparative fit index
conservation of resources
Diagnostic and Statistical Manual of Mental Disorders, Fourth Edition
nonnormed fit index
9-item Patient Health Questionnaire
posttraumatic stress disorder
8-item Posttraumatic Stress Disorder scale
root mean square error of approximation
structural equation modeling
social networking site
World Health Organization
XY conceived the research questions, conducted the statistical analysis, drafted the manuscript, and supervised the project’s implementation. XY, BY, and SW designed this study, assembled the team of collaborators, and gave comments regarding the intellectual content of the manuscript. All authors assisted in questionnaire design, data collection, and data interpretation, and gave comments regarding the intellectual content of the manuscript.
None declared.