INTRODUCTION
The coronavirus disease of 2019 (COVID-19) pandemic situation is a stressor because it evokes excessive and continuous distress in people and thus impacts society as a whole [1]. While stress responses are physiological and psychological reactions to deal with emergency situations or crisis, which is quite normal under everyday conditions, excessive stress responses can result in people exaggerating the seriousness of the pandemic. Adverse effects from these stressors are largely associated with diverse mental health and behavioral issues such as depression, anxiety, psychosomatic preoccupations, insomnia, substance use, and domestic violence [2], as well as maladaptive behaviors, emotional distress, and defensive responses [3]. Particularly relevant stressors include exposures to infected sources, infected family members, loss of loved ones, physical distancing, and quarantine, which are critical triggers that elevate people’s stress levels [2]. Prolonged confinement or quarantine, insufficient supplies, difficulty securing medical care and medications, as well as secondary adversities such as economic/financial loss are also critical stressors during the pandemic [4]. In particular, a few studies have emphasized the long-term effects of people’s responses to infectious diseases on mental health outcomes. As one of the responses to infectious diseases, fear of COVID-19 was associated with people’s mental health outcomes such as depression and anxiety [1,4,5]. However, few studies have longitudinally observed the relationship between fear of COVID-19 and mental health issues. Thus, this study particularly focused on this relationship from a long-term perspective.
Since the COVID-19 outbreak, findings from many crosssectional studies suggested that elevated psychological mental health problems stemming from the COVID-19 pandemic have become a serious worldwide concern. Several studies have been published regarding various mental health issues such as stress and fear [1,4], anxiety [6-18], depression [4,5,7,11-13,16,17,19], loneliness [1], and suicidal ideation [9,20-22]. Specifically, depression has become one of the most concerning mental health issues during the COVID-19 pandemic. For example, Ettman et al. [19] reported that about 27.8% of 1,441 adults in the U.S. were at risk of depression. In contrast, prior to the COVID-19 outbreak, only 2.4% experienced moderate depressive symptoms, according to the survey conducted in 2019 [16]. The risk of depression among 3,094 adults in the U.K. was 31.6% after the COVID-19 outbreak [11]. Findings from both the U.S. and U.K. indicated that approximately one out of three people would require professional help.
In the early stage of the pandemic, Wang et al. [18] studied 1,210 Chinese adults and reported that 16.5% of the respondents showed moderate-to-severe depressive symptoms, and women tended to show higher depressive levels than men. In a sample of 1,212 people in Sweden, 30.0% of the respondents reported significant levels of depression and poor self-rated overall health [23]. In Greece, about 22.8% of 3,209 respondents reported to be at high risk of depression during COVID-19 [5]. In the Philippines, 16.9% of 1,879 people from the general population showed moderate-to-severe depressive symptoms [24]. In Saudi Arabia, in a sample of 1,160 respondents of the general population, 28.3% reported moderate-to-severe depressive symptoms [25].
Like other countries, people in South Korea have also experienced mental health issues during the COVID-19 pandemic. Compared to the findings from the Korean Community Health Survey (KCHS) in 2018, people’s depression levels have increased approximately 2.4 times after the COVID-19 outbreak in 2020. Based on a national sample of 2,063 adults (aged 19 to 70), the average Patient Health Quesionnaire-9 (PHQ-9) score was 5.52 (standard deviation [SD]=5.49) in December 2020, with females showing higher levels of depression than men. Additionally, 20.0% of the respondents were at high risk of depression, which is about 4.86 times higher than prior to the COVID-19 outbreak [26]. Although the prevalence of individuals at high risk of depression in South Korea was lower than those of the U.K. (31.6%) [11], the U.S. (27.8%) [19], and Greece (22.8%) [5], it was higher than other East-Asian countries such as Japan (17.3%) [17] and China (14.8%) [12].
A limitation of the abovementioned cross-sectional studies is that they were conducted in the early stages of the pandemic, except for the studies in South Korea, the U.K., Japan, and Spain. Therefore, there is limited information regarding changes across time during the development of the COVID-19 pandemic. From previous experiences pertaining to epidemics, such as with severe acute respiratory syndrome (SARS), Middle East respiratory syndrome (MERS), and a novel influenza A (H1N1), it has been noted that in some cases, the psychological effects from infectious diseases can deteriorate over periods of months and years post outbreak [27-29]. Hence, it is necessary to conduct prospective studies that examine the psychological impacts over time. A longitudinal study including a national sample of 3,077 adults in the U.K. showed that over the course of 2.25 months from baseline to follow-up (wave 1 from March 31 to April 9; wave 2 from April 10 to 27; and wave 3 from April 28 to May 11, 2020) [20], suicidal ideation increased while depressive symptoms and levels of loneliness remained as before without any change. Additionally, females, young adults (18–29 years), and those with pre-existing mental health issues showed serious mental health outcomes.
Another study conducted in Japan reported that, among 2,078 respondents there was a significant increase in the people experiencing severe psychological distress (SPD) from 9.3% to 11.3% between the early phase of the study period (February 25–27, 2020) and community-transmission phase (follow-up during April 1–6, 2020) [30]. In particular, people in the lowest income group and those with respiratory diseases were approximately twice as likely to develop SPD than their counterparts [30]. Another longitudinal study (n=4,724) in Spain found that symptomatic scores of depression increased from 44.1% to 46.4% between March and April in 2020 [31].
During the pandemic, mental health professionals need to address current and emergent mental health issues, to create services plans to meet people’s needs, and to effectively and timely deliver the adequate services. Therefore, researchers play a key role in finding and disseminating important knowledge related to the psychological impact of COVID-19. For this reason, this study investigated the changes in depressive levels over the course of 9 months and identified predictors of people’s depressive levels. In particular, this study focused on the influence of people’s COVID-19 infection fear on their depressive levels while considering the COVID-19 pandemic.
METHODS
Study sample and data collection
Four cross-sectional surveysa were conducted in 2020. A total of 6,142 adults (aged 19 to 70) were randomly recruited by using a cluster sampling. The different samples participated in self-administrated online surveys in March (n=1,014), May (n=1,002), September (n=2,063), and December (n=2,063) 2020.b These four periods of data collection enabled us to compare the periodic changes in people’s depressive levels and their COVID-19 infection fear since the COVID-19 outbreak. All four surveys were approved by the Institutional Review Board (IRB) of Kangwon National University in South Korea (KWNUIRB-2020-03-004-001; KWNUIRB-2020-03-004-003).
Measures
The Korean version of PHQ-9 was used to measure the sample’s level of depression [32-34]. The PHQ-9 including nine items was administered as a self-report measure using a 4-point scale (0=not at all to 3=nearly every day). Higher scores indicated higher depressive levels; the total PHQ-9 scores over 10 indicated moderate or severe depression (cut-off ≥10) [32-34]. The total scores of PHQ-9 were used for statistical analysis (ranged 0 to 27). The PHQ-9 in this study showed a high internal consistency at the four survey points (overall Cronbach’s α=0.905).
The covariates were comprised of three blocks to explain the variance of PHQ-9: 1) survey points, 2) demographic variables, and 3) COVID-19 infection fear. As categorical variables, the survey points were assigned by the month of each survey, such as March (time 1; reference), May (time 2), September (time 3), and December (time 4) in 2020. Five demographic variables included gender, age, employment, family types, and regions. Gender was defined as a binary variable: male and female. Age was gauged as years at the point of each survey. Unemployment was coded as 1; others were coded as 0. The sample’s family type was categorized into four attributes: 1) living alone (reference), 2) living with only a partner, 3) living with a partner and child, and 4) other family/co-residence composition. Regions were divided into three attributes: 1) Seoul Capital area (reference), 2) other metropolitan (urban) areas, and 3) provinces (Table 1).
The “COVID-19 infection fear” scale was originally developed and validated by a group of multidisciplinary mental health specialists and researchers (i.e., psychiatry, social welfare, clinical psychology, and nursing) for measuring fear that was directly related to the COVID-19 situation and in consideration of the Korean culture and situations, which distinguished the construct validity from other COVID-19 fear-related measures. Prior to the surveys, the face and content validity of COVID-19 infection fear scale were obtained through several reviews and agreements of 13 mental health specialists and researchers from the multidisciplinary field. The COVID-19 infection fear scale consisted of nine itemsc including: I am afraid that 1) “I may get infected by coronavirus,” 2) “My family members might get infected by coronavirus,” 3) “I may be infected with coronavirus and pass it onto my family members or others,” 4) “I may be infected with coronavirus and will harm others, such as my colleagues at work or school shutdown,” 5) “If I and/or my family get infected, we will be separated/quarantined for treatment,” 6) “I may not be able to receive proper medical treatment,” 7) “I will be stigmatized as a confirmed person of coronavirus infection,” 8) “My community will be stigmatized,” and 9) “I may lose a job or have economic difficulties.” Each item was answered using a 4-point rating (0=very disagree, 1=disagree, 2=agree, and 3=very agree). Higher scores indicated higher levels of fear on COVID-19 infection. The average scores of the COVID-9 infection fear scale were used in the statistical analysis (ranged 0 to 3).d Although the COVID-19 infection fear scale was urgently developed at the early stage of the pandemic, it showed high internal consistency at all survey points (overall Cronbach’s α=0.928) (Table 2).
Statistical analysis
Descriptive analysis was conducted to explore the variability and distribution of collected data; a one-way analysis of variance (ANOVA) was performed to compare the changes in people’s depressive levels (i.e., PHQ-9) and their COVID-19 infection fear. The internal consistency of PHQ-9 and COVID-19 infection fear scale (i.e., Cronbach’s α) was also tested. In addition, the bivariate correlations between COVID-infection fear and PHQ-9 were tested by times. Finally, multiple regression models were built to identify the predictors of people’s depressive levels.
RESULTS
Demographic variables
Table 1 shows the sample’s characteristics. Nearly half of the respondents were females (49.27%, n=3,026) and the average age was 44.52 years old (SD=13.43). About 8.60% of the sample (n=528) were unemployed at the time of survey. 13.20% of the sample (n=811) lived alone; 13.32% (n=818) were married without a child; and 44.38% (n=2,726) lived with a partner and at least one child. Geographically, 32.61% of the sample (n=2,003) resided in Seoul Capital area; 36.86% (n=2,264) lived in other metropolitan (urban) areas; and 30.53% (n=1,875) lived in provinces. There were no significant differences of demographic attributes across four survey points.
Trends of PHQ-9 and COVID-19 infection fear
Table 2 shows the time-series trends of people’s depressive levels and COVID-19 infection fear in 2020. Overall, the mean of total PHQ-9 score was 5.50 (SD=5.51). The means of PHQ-9 by survey points were continuously changed throughout 2020. Specifically, the means of PHQ-9 in March (time 1), May (time 2), September (time 3), and December (time 4) were 5.10 (SD=5.26), 5.12 (SD=5.44), 5.86 (SD=5.66), and 5.52 (SD=5.49), respectively. The overall trends of PHQ-9 gradually increased from March (time 1) to December (time 4) (F[3, 6138]=6.282, p<0.001). As a result of the post hoc tests, the means of PHQ-9 increased by 0.76 between March (time 1) and September (time 3) (p<0.001) and elevated by 0.74 between May (time 2) and September (time 3) (p<0.001). In addition, the prevalence of depression (total PHQ-9 scores ≥10) significantly increased by times (χ2[3]=11.079, p=0.011). The prevalence of depression in March and December were 17.5% (n=177) and 20.0% (n=412), respectively.
Similar to the trends of PHQ-9, COVID-19 infection fear has increased since the COVID-19 outbreak (F[3, 6138]= 16.531, p<0.001). The overall means of average COVID-19 infection fear score was 1.74 (SD=0.71) in 2020. Although the means of COVID-19 infection fear slightly decreased by 0.14 (p<0.001) between March (time 1) and May (time 2), they increased again after May (time 2). Between May (time 2) and September (time 3), they increased by 0.17 (p<0.001). Compared to May (time 2), COVID-19 infection fear in December (time 4) increased by 0.18 (p<0.001). Mostly, both PHQ-9 and COVID-19 infection fear gradually increased over 9 months after the COVID-19 outbreak. In particular, they significantly increased after May, 2020 (time 2).
On the other hand, COVID-19 infection fear and PHQ-9 were positively correlated regardless of the survey points (r=0.424, p<0.001). The correlations between COVID-19 infection fear and PHQ-9 were to become stronger from March (time 1; r=0.389, p<0.001) to December (time 4; r=0.416, p< 0.001). In particular, the strength of correlations rapidly increased between March (time 1; r=0.389, p<0.001) and May (time 2; r=0.451, p<0.001).
Multiple regression results of PHQ-9
Table 3 presents the predictors of the PHQ-9, especially in the context of the prolonged pandemic. Model 1, only including the four survey points as the covariates, shows significant changes of PHQ-9 during the pandemic (F[3, 6138]=6.282, p<0.001; R2=0.003). Compared to March (time 1), PHQ-9 in September (time 3) and December (time 4) increased by 0.757 (b=0.757, t=3.588, β=0.065, p<0.001) and 0.425 (b=0.425, t= 2.015, β=0.044, p=0.044), respectively. Like the results of oneway ANOVA (Table 2), these results showed a gradual increase in people’s depressive symptoms as the pandemic persisted.
In Model 2, the demographic variables and the COVID-19 infection fear were added to Model 1 (F[12, 6129]=129.734, p<0.001). The inputted variables in Model 2 explained approximately 20.3% of the variance of PHQ-9 (R2=0.203). Similar to Model 1, there was a significant increase in PHQ-9 between March (time 1) and September (time 3) in 2020, after controlling for the demographic variables and COVID-19 infection fear (b=0.575, t=3.002, β=0.049, p=0.003).
As well, the PHQ-9 was significantly associated with the demographic variables such as gender, age, employment, and family type. Specifically, females showed 0.724 higher scores of PHQ-9 than males (b=0.724, t=5.684, β=0.066, p<0.001). As the sample’s age increased by one year, the PHQ-9 scores decreased by 0.027 (b=-0.027, t=-5.062, β=-0.067, p<0.001). Unemployed people showed 1.184 higher PHQ-9 scores than their counterparts (b=1.184, t=5.212, β=0.060, p<0.001). In addition, people living alone showed higher PHQ-9 scores than those in any other family types. For example, people who lived only with a partner showed 1.126 lower scores of the PHQ-9 than those who lived alone (b=-1.126, t=-4.487, β=-0.069, p<0.001). Additionally, people who lived with a partner and a child(ren) showed 1.189 lower scores than those who lived alone (b=-1.189, t=-5.858, β=-0.107, p<0.001). Finally, the level of COVID-19 infection fear was more strongly associated with PHQ-9 than other covariates, such as the survey points and demographic variables. As the sample’s average scores of COVID-19 infection fear increased by one-point, total PHQ-9 scores increased by 3.232 (b=3.232, t=35.790, β= 0.416, p<0.001). These results suggested the influence of people’s fear of infectious diseases on depressive levels in the context of the ongoing pandemic.
DISCUSSION
Among various issues related the COVID-19 pandemic, this study focused on the changes of people’s depressive levels through analyzing four consecutive surveys that were collected in 2020. To summarize study findings on changes in depressive levels and COVID-19 infection fear among Korean adults over 9 months after the COVID-19 outbreak in 2020, people’s depressive levels were more severe among women than men. We may speculate that given the high composition of female employees in South Korea, higher depressive levels and greater fear of infection may be more prominent among individuals who may need to be compatible with occupation as well as caregiver roles. Further, individuals with economic hardships including unemployment were at higher risk for continuously depressive levels. In the context of a pandemic, people’s COVID-19 infection fear had a significant influence on their depressive level. These findings are consistent with the recent studies on psychological effect of the COVID-19 [4,7,11,19].
Since the beginning of the COVID-19, we have experienced the second and third wave during the 2.5 years of the prolonged pandemic to this moment. Specifically, there was a gradual increase of people’s depressive levels between March and December in 2020 based on the analyses of time-series trends in the depressive levels using periodically collected data during the COVID-19 pandemic. Demographically, being a female, younger in age, and being unemployed were associated with people’s depressive levels in the ongoing pandemic. In addition, people who lived alone were comparably more vulnerable than any other population, as they reported significantly higher PHQ-9 scores than those who lived with someone. Presumably, this may be because, if infected, they may lack social support to deal with their illness and related difficulties, not to mention experiencing fear related to appropriate or timely access to the healthcare system. Given that the prevalence of depression is increasing over time, this may indicate that the high-risk group ratio is also on the rise. Therefore, it is necessary to plan a depression prevention program for targeted groups and to strengthen preemptive psychological support by local governments, centering on regional trauma centers.
Lastly, our findings suggest that people’s fear of infectious diseases was crucial to addressing mental health issues in the context of a long-term pandemic, as COVID-19 infection fear was more strongly associated with people’s depressive levels than other predictors (e.g., survey points and demographic characteristics). To reduce and preemptively prevent people’s depression, approaches aimed at lowering the fear of infection in the context of the COVID-19 pandemic should be carefully considered. Furthermore, efforts at the national level are necessary to improve the knowledge related to COVID-19 and health promotion practices. When an infectious disease spreads, the public experiences confusion and fear, resulting in a strong demand for information from the media [35]. To prevent the spread of fear and anxiety about the disease due to exaggerated reports and unverified articles in the media, reliable information should be delivered in a timely manner. Most importantly, access to mental health services should be secured particularly for individuals who present greater vulnerabilities due to socioeconomic factors that may affect their mental health.
Despite the meaningful findings, our study includes a few limitations that should be considered in future studies. The COVID-19 infection fear scale, which was the critical predictor of PHQ-9 in this study and was developed by our multidisciplinary mental health professionals, is awaiting publication for testing its measurement validity. Additionally, because this study used a trend data that involved recruited different samples at four survey points within a year, it was impossible to test the actual changes at the individual level. Fortunately, we began building a panel data since December, 2020 (time 4). Thus, we expect to estimate more accurately the changes in individuals’ depressive levels with a longer time window by using advanced statistical analyses (e.g., fixed effects model, latent growth model, etc.).
