Association between loneliness and depression, anxiety and anger during the COVID-19 pandemic: a nationwide population-based survey

Abstract

Objectives

This study aims to determine the association between loneliness and depression, anxiety and anger with a representative sample of the general population in Korea, which are the most prevalent mental health problems during the pandemic.

Design

Cross-sectional study.

Setting

National survey across all 17 provinces in South Korea between December 2021 and January 2022.

Participants

We conducted a national survey on 2699 participants aged 19–84 years using proportional stratified sampling. Using the UCLA Loneliness Scale and standardised questionnaires for depression (Patient Health Questionnaire-9), anxiety (Generalised Anxiety Disorder-7) and anger (Patient-Reported Outcomes Measurement Information System-Anger), we explored the prevalence and association of loneliness with these mental health outcomes.

Primary and secondary outcome measures

Primary outcomes included the prevalence and co-occurrence of depression, anxiety and anger across different levels of loneliness.

Results

Of total, 20.7% and 2.1% experienced moderately high and high levels of loneliness, respectively. Among participants with high levels of loneliness, 11.8%, 5.9% and 11.8% had depression, anxiety and anger, respectively, and 28.7% of them had depression, anxiety and anger together. The adjusted prevalence of depression was 0.2 (95% CI 0.0 to 0.5), 8.2 (95% CI 6.7 to 9.7), 31.3 (95% CI 27.4 to 35.3) and 63.5 (95% CI 50.1 to 76.8) for low, moderate, moderately high and high levels of loneliness, respectively. Similarly, increased adjusted prevalence of anxiety and anger was observed ㅈwith higher levels of loneliness.

Conclusions

Lonely people have a higher risk of depression, anxiety and anger. Identifying individuals who may be vulnerable to loneliness is important for early intervention.

STRENGTHS AND LIMITATIONS OF THIS STUDY.

• This study used a large, nationally representative sample encompassing all 17 provinces of South Korea, thereby enhancing generalisability.

• Validated instruments were used to measure loneliness, depression, anxiety and anger, supporting the reliability of findings.

• The cross-sectional design limits the ability to infer causality between loneliness and mental health outcomes.

• Self-reported, web-based data collection may introduce reporting biases.

• Potential selection bias may exist due to differences between survey participants and non-participants.

Introduction

Loneliness is a distressing feeling of isolation or insufficient meaningful connections characterised by an unmet need or disparity between an individual’s preferred and actual experiences.¹ It is becoming a major public health issue owing to its increasing prevalence and impact on people’s health.^{2 3} Loneliness could increase the risk of premature mortality by 26%,⁴ heart disease by 29%,⁵ stroke by 32%⁵ and dementia by 49%.⁶

Specifically, loneliness has a profound impact on mental health issues such as depression, anxiety and suicidality.7,9 A previous cross-sectional survey in Germany found that loneliness was associated with depression (OR 1.91, 95% CI 1.74 to 2.09), generalised anxiety (OR 1.21, 95% CI 1.09 to 1.34) and suicidal ideation (OR 1.35, 95% CI 1.19 to 1.44).⁸ Another study with 1700 Koreans from three metropolitan areas found that people with loneliness were more likely to have depression (OR 3.42, 95% CI 1.61 to 7.26), social phobia (OR 3.06, 95% CI 1.65 to 5.68) and suicidal thoughts (OR 4.21, 95% CI 2.21 to 8.02) than people without loneliness.¹⁰

According to a systematic review, more people experienced loneliness during the COVID-19 pandemic because of limited social interaction caused by social distancing and home quarantine.¹¹ Another systematic review that analysed the prevalence of mental health problems in China, Spain, Italy, Iran, the US, Turkey, Nepal and Denmark found that up to 80% of the general population experienced stress and up to 48.3%, 50.9% and 53.8% experienced symptoms of depression, anxiety and post-traumatic stress disorder, respectively, during the COVID-19 pandemic.¹² Considering the well-established link between loneliness and mental health, researchers have investigated the association between loneliness and mental health outcomes during the COVID-19 pandemic, but the results are inconsistent. Some studies have found a high correlation between an increased risk of loneliness and mental health problems.¹³ Other studies have found no association between loneliness and mental health during the pandemic.^{14 15} This may be due to different study settings and populations. Most studies have investigated older adults or adolescents who are more vulnerable to loneliness because of social isolation.13,16 In addition, most studies have evaluated single mental health problems, despite a high correlation between them. Therefore, we conducted a national survey with a representative sample of the general population in Korea to determine the association between loneliness and depression, anxiety and anger, which are the most prevalent mental health problems during the pandemic.

Methods

Study design and participants

We conducted a national survey using proportional stratified sampling across all 17 provinces in South Korea between December 2021 and January 2022. Eligible participants were adults aged 19–84 years residing in South Korea, capable of providing informed consent and without cognitive impairments. Owing to the COVID-19 pandemic, an online survey was conducted. However, considering digital health literacy, we conducted a face-to-face survey of people aged ≥65 years old. Because participants were selected using proportional stratified sampling from a registered online panel managed by an external survey agency, additional measures to prevent multiple participation in the web-based survey were not implemented. To minimise mode effects between the online and face-to-face surveys, both modes used identical questionnaires, and interviewers conducting face-to-face surveys received standardised training to ensure consistent administration and minimal intervention. For the analysis, we excluded participants with a self-reported history of psychiatric disease (n=222), resulting in a final analytic sample of 2477 participants. This exclusion allowed us to specifically examine associations between loneliness and undiagnosed or subclinical mental health problems in the general population. This study was approved by the Institutional Review Board of Samsung Medical Center (IRB no. SMC-2021-03-005). All the participants provided informed consent. The survey was conducted anonymously, and all data were stored securely to ensure participant confidentiality.

Measurements

Loneliness was measured using the UCLA Loneliness Scale (V.3), which has been validated in Korean.^{17 18} The Korean version of the UCLA Loneliness Scale has demonstrated satisfactory reliability (Cronbach’s α=0.94) and validity among Korean university students.¹⁸ Participants were asked about their subjective feelings of loneliness. This scale contains 20 items rated on a 4-point Likert scale ranging from 1 (never) to 4 (always). Items 1, 5, 6, 9, 10, 15, 16, 19 and 20 are reverse scored. Higher scores indicate higher levels of loneliness. The sum of scores of the 20 questions (total score) was categorised into four groups: scores of 20–34, 35–49, 50–64 and 64–80 were classified as low, moderate, moderately high and high levels of loneliness, respectively.¹⁹ Loneliness was defined as a moderately high or high level of loneliness.

We evaluated depression using the Patient Health Questionnaire-9 (PHQ-9),^{20 21} which contains nine items on a 5-point Likert scale ranging from 0 (never) to 4 (always). The Korean version of the PHQ-9 has shown good reliability (Cronbach’s α=0.79) and validity in the general Korean population.²¹ Depression was defined as a total score ≥5 points.²⁰ Anxiety was measured using the Generalised Anxiety Disorder-7 (GAD-7) scale, which contains nine items on a 4-point Likert scale ranging from 0 (not at all) to 3 (nearly every day).^{22 23} The Korean version of the GAD-7 exhibited high reliability (Cronbach’s α=0.88) and strong convergent validity among psychiatric outpatients in Korea.²³ Anxiety was defined as a total score ≥5 points.²² Anger was measured using the Patient-Reported Outcomes Measurement Information System (PROMIS)-Anger, which was developed to measure angry mood, negative social cognition and efforts to control anger in the past 7 days.^{24 25} The PROMIS-Anger contains 22 items rated on a 5-point Likert scale ranging from 1 (never) to 5 (always). A T-score metric from the assessment centre scored the questionnaire. The T-score metric rescaled the raw score into a standardised T-score with a mean of 50 and SD of 10. A higher T-score indicates a higher level of anger. The Korean version of the PROMIS-Anger scale demonstrated high reliability (Cronbach’s α=0.95) and robust construct validity.²⁵ In this study, anger was defined as a total T-score of ≥60 points.²⁴ All the measurements used in this study were validated in Korean.

Sociodemographic information, including age, gender, physical disability, marital status, education level, occupational status, annual household income and residential area, was collected from the study participants. Comorbidities, including cardiocerebrovascular diseases, cancer and COVID-19-related experiences, were also assessed. The survey items developed by the research team are provided in online supplemental table 1. Standardised instruments (UCLA Loneliness Scale, PHQ-9, GAD-7 and PROMIS-Anger) are cited but not reproduced due to copyright restrictions.

Statistical analysis

Analysis of variance and χ² tests were used for continuous and categorical variables according to loneliness levels, respectively. The adjusted prevalence and 95% CI for depression, anxiety and anger by levels of loneliness were calculated using logistic regression, adjusting for age and gender (table 1).²⁶ To identify factors related to loneliness, we conducted a logistic regression analysis, adjusting for age and gender (table 2). Other sociodemographic variables were not adjusted in the primary model due to their high correlation with age and gender (online supplemental figure 1). However, given the low risk of multicollinearity (all variance inflation factors <5), we conducted a sensitivity analysis using a fully adjusted model that included marital status, education, occupation, annual household income, residential area, physical disability, comorbidities and COVID-19 experience (online supplemental table 2). All statistical analyses were performed using R V.4.1.1 (Vienna, Austria; http://www.R-project.org/). P values were two-sided, and statistical significance was set at p<0.05.

Table 1. Adjusted prevalence with 95% CIs for mental health outcomes by loneliness level.

	Loneliness level
	Low	Moderate	Moderately high	High
Had at least 1 outcome	1.7 (0.7 to 2.8)	19.7 (17.6 to 21.8)	51.8 (47.5 to 56.1)	68.5 (55.6 to 81.5)
Depression	0.2 (0.0 to 0.5)	8.2 (6.7 to 9.7)	31.3 (27.4 to 35.3)	63.5 (50.1 to 76.8)
Anxiety	1.0 (0.2 to 1.9)	16.6 (14.6 to 18.6)	44.6 (40.3 to 48.9)	51.8 (38.0 to 65.5)
Anger	0.9 (0.1 to 1.7)	8.1 (6.6 to 9.6)	26.2 (22.5 to 29.9)	31.9 (19.6 to 44.2)
Had at least 2 outcomes	0.4 (0.0 to 0.8)	8.8 (7.3 to 10.4)	32.0 (28.0 to 35.9)	50.0 (36.4 to 63.6)
Depression+Anxiety	0.0 (0.0 to 0.0)	7.0 (5.6 to 8.4)	27.3 (23.5 to 31.1)	48.1 (34.5 to 61.7)
Depression+Anger	0.0 (0.0 to 0.0)	4.8 (3.6 to 6.0)	19.4 (16.0 to 22.7)	30.6 (18.3 to 42.8)
Anxiety+Anger	0.4 (0.0 to 0.8)	5.9 (4.6 to 7.1)	22.2 (18.7 to 25.7)	28.8 (16.8 to 40.7)
Had all 3 outcomes	0.0 (0.0 to 0.0)	4.4 (3.3 to 5.5)	18.4 (15.2 to 21.7)	28.7 (16.7 to 40.7)

Adjusted for age and gender.

Table 2. Factors associated with moderately high or high loneliness.

	Unadjusted OR(95% CI)	Age-gender adjusted OR*(95% CI)	Fully adjusted OR†(95% CI)
Age (years)	0.98 (0.98 to 0.99)	0.98 (0.98 to 0.99)	0.99 (0.98 to 0.99)
Young adult (18–29)	Reference	Reference	Reference
Adult (30–59)	1.03 (0.80 to 1.33)	1.03 (0.80 to 1.33)	1.26 (0.90 to 1.78)
Older adult (60–84)	0.49 (0.36 to 0.67)	0.49 (0.36 to 0.66)	0.58 (0.37 to 0.90)
Gender
Male	Reference	Reference	Reference
Female	1.10 (0.92 to 1.33)	1.13 (0.93 to 1.36)	1.07 (0.86 to 1.32)
Marital status
Married	Reference	Reference	Reference
Unmarried‡	1.67 (1.38 to 2.02)	1.38 (1.11 to 1.72)	1.50 (1.17 to 1.91)
Education
≤High school	Reference	Reference	Reference
≥College	1.47 (1.19 to 1.83)	1.10 (0.86 to 1.43)	1.20 (0.92 to 1.59)
Occupation
Employed§	Reference	Reference	Reference
Housewife	1.07 (0.82 to 1.40)	1.28 (0.94 to 1.73)	1.40 (1.02 to 1.92)
Students/soldiers	0.96 (0.66 to 1.38)	0.56 (0.37 to 0.83)	0.51 (0.33 to 0.77)
Unemployed¶	1.32 (0.98 to 1.76)	1.42 (1.05 to 1.90)	1.29 (0.93 to 1.78)
Annual household income
≤US$20 000	Reference	Reference	Reference
US$20 000–US$50 000	1.13 (0.87 to 1.47)	1.01 (0.78 to 1.33)	1.05 (0.79 to 1.42)
≥US$50 000	1.00 (0.77 to 1.31)	0.85 (0.65 to 1.12)	0.94 (0.68 to 1.29)
Residential area**
<1 00 000 people	Reference	Reference	Reference
1 00 000–3 00 000 people	1.32 (0.82 to 2.19)	1.21 (0.75 to 2.02)	1.13 (0.70 to 1.89)
>3 00 000 people	1.69 (1.09 to 2.74)	1.51 (0.96 to 2.45)	1.42 (0.91 to 2.33)
Physical disability	2.21 (1.32 to 3.62)	2.37 (1.41 to 3.92)	2.26 (1.31 to 3.84)
Comorbidity
Cardiovascular disease	0.79 (0.64 to 0.97)	1.06 (0.84 to 1.35)	1.06 (0.83 to 1.35)
Cerebrovascular disease	1.17 (0.56 to 2.26)	1.36 (0.65 to 2.65)	1.17 (0.53 to 2.39)
Cancer	0.80 (0.44 to 1.38)	0.92 (0.50 to 1.59)	0.78 (0.42 to 1.37)
COVID-19 experience††	1.46 (1.04 to 2.03)	1.32 (0.93 to 1.84)	1.29 (0.91 to 1.81)

^*Adjusted for age and gender.

^†Adjusted for age, gender, marital status, education, occupation, annual household income, residential area, physical disability, comorbidity and COVID-19 experience.

^‡Included separated, divorced and widowed.

^§Included self-employed, farm and livestock.

^¶Included unemployed, leave of absence and retired.

^**Classified into three groups according to population size and geographic location from Korean national statistics.

^††Included COVID-19 or quarantine experience.

Patient and public involvement

There was no patient or public involvement in the research design, conceptualisation, analysis, reporting or dissemination plans of this research.

Equity, diversity and inclusion statement

The authors conducted a study focusing on individuals who experience loneliness, a group often marginalised in previous research, including data regarding their mental health status and severity. Particularly, our research included the general Korean population, encompassing individuals experiencing loneliness with diverse genders, ages and demographics. However, since our study is limited to Koreans, further research with a diverse range of ethnicities is necessary for future generalisation.

Results

In the survey, all contacted individuals responded only once without multiple participation, and all variables were complete with no missing values. The mean age of the study participants was 49.0 years (SD, 16.7 years), and 50.5% (n=1251) of the participants were men (table 3). The mean loneliness score of the participants was 42.7 (SD 10.0), with 20.7% and 2.1% experiencing moderately high and high levels of loneliness, respectively. Among the 11 negatively worded (lonely) items, question 14, ‘How often do you feel isolated from others?’ had the lowest proportion of responses indicating ‘Always’ (2.7%) or ‘Sometimes’ (12.4%), totalling 15.1%. Among the nine positively worded (non-lonely) items, question 15, ‘How often do you feel you can find companionship when you want it?’ had the highest proportion of responses indicating ‘Rarely’ (31.3%) or ‘Never’ (29%), totaling 60.3% (figure 1). Participants with a high level of loneliness were more likely to be younger, female, unmarried, unemployed and physically disabled than those with a low level of loneliness (table 3).

Table 3. Demographical factors and comorbidities of the study population according to loneliness level (n=2477).

	Overall	Loneliness level				P value
		Low(n=582)	Moderate(n=1331)	Moderately high(n=513)	High(n=51)
Age, mean (SD)	49.0 (16.7)	50.5 (16.9)	49.8 (17.0)	45.9 (15.3)	42.0 (13.6)	<0.001
Young adult (18–29)	388 (15.7)	87 (22.4)	201 (51.8)	87 (22.4)	13 (3.4)	<0.001
Adult (30–59)	1365 (55.1)	291 (21.3)	715 (52.4)	325 (23.8)	34 (2.5)
Older adult (60–84)	724 (29.2)	204 (28.2)	415 (57.3)	101 (14.0)	4 (0.5)
Gender
Male	1251 (50.5)	302 (24.1)	675 (54.0)	254 (20.3)	20 (1.6)	0.35
Female	1226 (49.5)	280 (22.8)	656 (53.5)	259 (21.1)	32 (2.6)
Marital status						<0.001
Married	1532 (61.8)	407 (26.6)	830 (54.2)	278 (18.1)	17 (1.1)
Unmarried*	755 (30.5)	175 (18.5)	501 (53.0)	235 (24.9)	34 (3.6)
Education						0.002
≤ High school	745 (30.1)	171 (22.9)	438 (58.8)	122 (16.4)	14 (1.9)
≥ College	1732 (69.9)	411 (23.7)	893 (51.6)	391 (22.6)	37 (2.1)
Occupation						0.003
Employed†	1654 (66.8)	399 (24.1)	890 (53.8)	343 (20.8)	22 (1.3)
Housewife	369 (14.9)	82 (22.2)	201 (54.5)	75 (20.3)	11 (3.0)
Students/soldiers	182 (7.3)	54 (29.7)	89 (48.9)	33 (18.1)	6 (3.3)
Unemployed‡	272 (11.0)	47 (17.3)	151 (55.5)	62 (22.8)	12 (4.4)
Annual household income						0.02
≤US$20 000	475 (19.2)	95 (20.0)	276 (58.1)	95 (20.0)	9 (1.9)
US$20 000–US$50 000	980 (39.6)	210 (21.4)	534 (54.5)	213 (21.7)	23 (2.4)
≥US$50 000	1022 (41.3)	277 (27.1)	521 (51.0)	205 (20.0)	19 (1.9)
Residential area§						<0.001
<1 00 000 people	144 (5.8)	59 (41.0)	62 (43.0)	22 (15.3)	1 (0.7)
1 00 000–3 00 000 people	610 (24.6)	154 (25.2)	334 (54.8)	109 (17.9)	13 (2.1)
>3 00 000 people	1723 (69.6)	369 (21.4)	935 (54.3)	382 (22.2)	37 (2.1)
Physical disability						0.003§
Yes	67 (2.7)	8 (11.9)	33 (49.3)	22 (32.8)	4 (6.0)
No	2410 (97.3)	574 (23.8)	1298 (53.9)	491 (20.4)	47 (2.0)
Comorbidity
Cardiovascular disease	737 (29.8)	162 (22.0)	428 (58.1)	136 (18.4)	11 (1.5)	0.330
Cerebrovascular disease	43 (1.7)	6 (13.9)	26 (60.5)	11 (25.6)	0 (0.0)	0.330
Cancer	78 (3.1)	24 (30.8)	39 (50.0)	12 (15.4)	3 (3.8)	0.180§
COVID-19 infection						0.710§
Yes	42 (1.7)	12 (28.6)	20 (47.6)	9 (21.4)	1 (2.4)
No	2435 (98.3)	570 (23.4)	1311 (53.8)	504 (20.7)	50 (2.1)
Quarantine experience due to COVID-19 infection						0.050§
Yes	165 (6.7)	39 (23.6)	75 (45.5)	46 (27.9)	5 (3.0)
No	2312 (93.3)	543 (23.5)	1256 (54.3)	467 (20.2)	46 (2.0)

^*Included separated, divorced and widowed.

^†Included self-employed, farm and livestock.

^‡Included unemployed, leave of absence and retired.

^§Classified into three groups according to population size and geographic location from Korean national statistics.

†Fisher’s exact test.

Figure 1. Distribution of responses to UCLA loneliness questionnaire items.

Participants with a high level of loneliness were more likely to experience depression, anxiety and anger than those with a low level (figure 2 and online supplemental table 3). Among participants with high levels of loneliness, 11.8%, 5.9% and 11.8% had depression, anxiety and anger, respectively. Conversely, among participants with a low level of loneliness, none had depression or anxiety, and only 0.2% had severe anger (online supplemental table 3). The adjusted prevalence of depression was 0.2 (95% CI 0.0 to 0.5), 8.2 (95% CI 6.7 to 9.7), 31.3 (95% CI 27.4 to 35.3) and 63.5 (95% CI 50.1 to 76.8) for low, moderate, moderately high and high levels of loneliness, respectively. Similarly, increased adjusted prevalence of anxiety and anger was observed as severe loneliness (table 1).

Figure 2. Associations between loneliness and depression, anxiety and anger.

The adjusted prevalence of having at least one of depression, anxiety and anger was 68.5% (95% CI 55.6 to 81.5) in people with high levels of loneliness, while only 1.7% (95% CI 0.7 to 2.8) with low levels of loneliness had one of those mental health problems. In total, 28.7% of participants with high levels of loneliness had depression, anxiety and anger together (table 1).

Compared with people who were aged 18–29 years, the adjusted ORs (aORs) for feeling lonely in those aged 60–69 years and 70–79 years were 0.52 (95% CI 0.36 to 0.74) and 0.40 (95% CI 0.25 to 0.62), respectively. The aOR for feeling lonely comparing participants who were unmarried with those who were married was 1.38 (95% CI 1.11 to 1.72). Compared with participants who were employed, the aOR for feeling lonely in those who were students/soldiers was 0.56 (95% CI 0.37 to 0.83), while the aOR among those who were unemployed was 1.42 (95% CI 1.05 to 1.90). Particularly, the presence of a physical disability was strongly associated with feeling lonely compared with the lack of a physical disability (aOR 2.37, 95% CI 1.41 to 3.92) (table 2). In contrast, logistic regression analysis showed that having experienced COVID-19 was not significantly associated with increased feelings of loneliness (aOR 1.32, 95% CI: 0.93 to 1.84). These associations remained consistent in the fully adjusted model, indicating the robustness of our findings across different model specifications (table 2).

Discussion

Principal results

In this study, we found that 22.8% of Korean people experienced moderately high or high levels of loneliness during the COVID-19 pandemic. Younger people, those living alone, those unemployed and those with physical disabilities were more likely to experience high levels of loneliness. Moreover, high levels of loneliness were associated with depression, anxiety and anger. Almost 70% of the people with high levels of loneliness had at least one of these mental health problems, and approximately one-third of people with high levels of loneliness had depression, anxiety and anger together.

In our study, about one-fourth of the participants experienced loneliness. Similarly, in a UK survey conducted during the COVID-19 pandemic, approximately 7% and 28% of the respondents said that they felt lonely often or some of the time, respectively.²⁷ While a direct comparison is somewhat challenging owing to differences in measurement and cut-off scores of loneliness, the prevalence of loneliness in our study was slightly higher than that reported before the pandemic. For example, according to a 2018 survey, approximately 16% of the participants in South Korea reported feeling lonely.²⁸ This difference may be attributed to social distancing during the COVID-19 pandemic. During the pandemic, the number of intimate social interactions decreased because of self-quarantine. Although virtual meetings were held online, they might be relatively superficial ways of communicating compared with offline social interactions. Alternatively, this might be due to the rapid changes in culture and social environment in Korea. Recently, Korean society has been undergoing a weakening of the social support system and a deepening individualism owing to a super-aged society (19.2% over the age 65),²⁹ declining marriage rates (3.7 per 1000 people),³⁰ low birth rates (total fertility rate of 0.778)³⁰ and increasing divorce rates (1.8 per 1000 married individuals),³⁰ resulting in a rise of single-person households (34.5%)³¹ per the 2022 statistics. In Korean society, which is traditionally characterised by East Asian collectivist values that emphasise family and kinship networks, recent rapid social and structural changes may have led to confusion in personal and societal values, potentially exacerbating feelings of loneliness and social isolation.³² Those who do not have intimate supportive networks such as family or friends and rely on the support of volunteer services or social care could be placed at additional risk, along with those who are already lonely.³³

Our findings indicate that individuals reporting higher levels of loneliness were more likely to report symptoms of depression, anxiety and anger. Although we were unable to control for unmeasured factors that might influence mental health outcomes, such as social security and substance use,³⁴ the association remained statistically significant after adjusting for key measured confounders. This pattern is consistent with previous studies89 34,37 and supports the possibility that loneliness may be one of several contributing factors to poorer mental health outcomes during the pandemic. Given the persistent stressors and social restrictions during this period, loneliness may have interacted with other psychosocial factors to exacerbate emotional distress.¹³ Lonely people might perceive their social interactions negatively, which can elevate their stress levels and decrease their responsiveness to positive stimuli,³⁸ potentially contributing to depression. According to previous research on social media users, increased loneliness may reduce self-esteem, which can act as a mediator and ultimately exacerbate depression through psychosocial pathways.³⁹ Loneliness can also reduce social skills, sociability and levels of self-esteem and optimism and increase anxiety, anger, negativism and fear of negative evaluations.³⁷ Pathophysiological mechanisms, such as disruption of the limbic system, could explain the association between social isolation or loneliness and anxiety disorders.^{40 41} According to a previous study with 135 participants, loneliness can aggravate feelings of insecurity and rejection, thus leading people to feel anxious.⁴² Furthermore, previous studies suggest a potential bidirectional relationship between loneliness and mental health outcomes.^{9 35} In other words, worsening depression, anxiety or anger may also contribute to heightened feelings of loneliness. Therefore, future research is needed to better understand these complex interactions in order to inform effective psychological interventions.

In our study, 28% of the people with high levels of loneliness experienced depression, anxiety and anger together. This might be because people with one mental health problem are at high risk of developing other mental health problems.³³ Given the strong association between loneliness and mental health outcomes,89 34,37 it is important to identify early signs of depression, anxiety, anger and interventions to prevent complications such as the progression of severe major depressive disorder⁴³ and suicidal ideation⁴⁴ among people with high levels of loneliness. Alternatively, by assessing loneliness levels, we may be able to identify individuals with mental health problems. In fact, in our study, we excluded individuals with physician-diagnosed psychiatric diseases including depression and anxiety disorder. However, we found a fair number of people with depression, anxiety and anger problems, which were almost two times more prevalent among people with high levels of loneliness (online supplemental table 3). These findings suggest that screening for loneliness could be useful in identifying mental health concerns among individuals who may currently be under-recognised by formal psychiatric care systems. Given the persistent stigma surrounding psychiatric illness in Korean culture, individuals may be more likely to avoid seeking formal mental healthcare.⁴⁵ In this context, proactive screening for loneliness and timely intervention may be particularly critical to addressing hidden mental health needs.

In contrast to South Korea, where loneliness is still not yet widely recognised as a public health concern, countries like the UK and Japan have appointed Ministers of Loneliness—emphasising the need to address loneliness at the societal level, in line with our finding that loneliness is strongly associated with adverse mental health outcomes.^{46 47} Specifically, the UK’s Ministry of Loneliness conducts regular surveys on the general population to increase awareness of loneliness, screen high-risk populations and provide appropriate support.⁴⁸ Still, health professionals often do not prioritise loneliness as a significant health issue⁴⁹ and find it challenging to accurately detect lonely patients.⁵⁰ Public health campaigns and education are necessary to increase awareness of loneliness and its health consequences. Active interventions should be developed to reduce loneliness.^{35 51}

In our study, loneliness levels were highest in young adults (18~29 years) and lowest in older adults (60+years). This finding is consistent with those of the previous studies. Young adults may feel lonely because of a sense of emotional isolation, pressure, social comparison, experience of social media, transitions in education, relationships and employment.⁵² In our study, young adults were more likely to feel lonely because of a lack of companionship and feeling alone (online supplemental figure 2). These feelings may stem from the pressure they experience in relation to their expectations of contemporary social and working life, including marriage and employment. Additionally, social media seems to boost stress or loneliness as young adults compare themselves with others using images on social media, which often results in negative feelings. Future studies could explore whether offline community spaces for socially isolated young adults struggling with major life milestones, such as marriage and employment, effectively reduce their loneliness. Interestingly, students and soldiers were less likely to feel lonely than other young age groups. School and army helped young adults feel bonded with others and increase their social belonging. However, similar circumstances may lead to loneliness through different pathways. These intrapersonal processes may differ systematically across ages. The relationship between age and loneliness may be more complex than linear,^{10 53} requiring additional studies.

Among various sociodemographic factors, our study participants with a physical disability had the highest OR of experiencing moderately high or high levels of loneliness (OR 2.37, 95% CI 1.41 to 3.92). People with physical disabilities lack sufficient supportive resources to fulfil their daily needs and can easily experience social isolation and loneliness.⁵⁴ According to a cohort study conducted during the COVID-19 pandemic period involving 5820 individuals with physical disability aged 52 years and older in the UK, these individuals had 0.70 times less frequent contact using phone or video calls and 0.54 times less frequent contact using email or letters.⁵⁵ Additionally, their risk of experiencing loneliness was 1.52 times higher. Considering the increased risk of reduced social contact due to mobility constraints among people with physical disabilities during pandemics such as COVID-19, we need to explore strategies to improve their social connectedness through rehabilitative activities, for example, visiting medical care facilities or using digital health technologies such as remote telemedicine or chatbots.

Interestingly, in our study, the experiences related to COVID-19 were not associated with feelings of loneliness. This might be because social distancing policies during the pandemic influenced the entire population, not just those who were quarantined due to COVID-19. In other words, since social distancing measures were implemented universally, individuals without a history of COVID-19 infection also experienced reduced inperson interactions. This may help explain the lack of significant differences in loneliness between those with and without infection. Second, participants with COVID-19 experience may have preserved their subjective perceptions of social connectedness even when they were objectively isolated due to quarantine. In line with our study results, a longitudinal online survey of 1545 US participants from January to April 2020 reported no significant changes in the average level of loneliness (p>0.05) during the three assessment points during the COVID-19 pandemic.⁵⁶ Rather, participants reported increased support from others during the follow-up period (p<0.01). As almost everyone’s attention, including social media, was focused on the public health crisis of COVID-19 during the pandemic, they might have experienced being the centre of attention rather than being left out (isolated), even though they were physically quarantined. Additionally, individuals may have preserved their sense of social connectedness through remote communication technologies, such as social networking services, video conferencing or metaverse platforms, which became widely used during the pandemic, despite physical isolation.^{57 58}

During the COVID-19 pandemic, various online interventions were implemented.¹⁶ However, online interventions may not be effective owing to the potential confounding effects of social distancing and subsequent lockdowns during COVID-19. Additionally, most interventions in previous studies were for specific populations, such as older adults and children.13,16 Further research is required to investigate which interventions can provide social networks for these individuals to improve the social connectedness of weakened social networks across the general population.

Limitations and strengths

Our study had several limitations. First, the cross-sectional study design could not establish a confirmative causal relationship between loneliness levels and associated measurements, including mental disorders and sociodemographic factors. Thus, loneliness could be both a predisposing cause and symptomatic result of mental disorders such as depression, anxiety or anger,^{35 51} and our results also reflect the bidirectional relationship. To clarify these temporal and directional relationships, future studies employing longitudinal designs are warranted. Such approaches could identify whether loneliness precedes mental health and sociodemographic factors or vice versa, enabling better-targeted interventions and preventive measures. Second, although we recruited representative participants using a proportionate stratified sampling method covering all 17 provinces of Korea, a selection bias may have been present. People who participated in the survey differed from those who did not. Furthermore, because our survey was conducted primarily online, additional biases may have influenced our findings. Self-selection bias likely occurred, as individuals with greater internet literacy or more frequent online engagement were more likely to participate. Coverage bias, a common concern in online surveys due to limited internet access or lower digital literacy—particularly among older adults or residents of rural areas—was addressed by conducting face-to-face surveys for participants aged 65 years and older. Non-response bias might have arisen if certain groups were less inclined to complete the survey. Measurement bias could have been introduced due to the absence of interviewer guidance in the online survey, potentially leading to misinterpretation of questions. Third, excluding individuals with a psychiatric history may have led to an underestimation of the overall mental health burden in the general population, and future studies could consider including these populations to capture a broader spectrum of psychological distress. Finally, because we measured loneliness, depression, anxiety and anger online, people might have under-reported or over-reported their symptoms and feelings due to social desirability or privacy concerns associated with online reporting. However, we used valid measures to assess these outcomes, which have been used in several online surveys.

Conclusions

This study highlights the profound impact of loneliness on mental health during the COVID-19 pandemic. Our findings reveal that loneliness is not only a personal issue but also a significant public health concern, particularly in times of widespread social isolation. The strong association between high levels of loneliness and an increased prevalence of mental health issues such as depression, anxiety and anger underscores the necessity for targeted interventions. These interventions should focus on the general population and on specific vulnerable groups, including young adults, the unmarried, unemployed and those with physical disabilities. To address loneliness among vulnerable groups, community-based intervention strategies may be helpful.⁵⁹ For socially isolated young or unmarried individuals, the creation of offline community spaces—such as local hubs for hobby activities, peer support or job preparation programmes—could reduce stigma and encourage social engagement. For individuals with physical disabilities, integrative programmes combining physical rehabilitation and social interaction may foster a greater sense of connection.⁶⁰ Furthermore, for individuals experiencing severe symptoms of loneliness-related depression or anxiety, early intervention through evidence-based approaches such as cognitive behavioural therapy or pharmacological treatment may be warranted. Future research is needed to evaluate the feasibility and effectiveness of these proposed intervention strategies in reducing loneliness and improving mental health outcomes among high-risk populations.

Data availability statement

Data are available upon reasonable request.