INTRODUCTION
The coronavirus disease 2019 (COVID-19) pandemic had an unprecedented level of impact worldwide, from post-acute eISSN 1976-3026 OPEN ACCESS sequelae and mortality to social confinement in daily life and psychological distress. Studies have reported a possible surge in mental health problems among the general population and not just high-risk populations, such as individuals with preexisting psychiatric conditions [1]. As the COVID-19 pandemic is associated with an increase in psychiatric morbidity, including depression and suicidal behavior [2], emphasis on the need for mental health management in the post-COVID-19 era has been growing. Previous studies have mainly investigated depression, anxiety, post-traumatic stress disorder (PTSD), and suicidal ideation in relation to the negative impact of the COVID-19 pandemic on mental health [3-7].
According to a report published by the Organization for Economic Cooperation and Development, the prevalence of depression, in particular, has increased by 2-3 fold compared to the pre-pandemic levels. The prevalence of depression was the highest in the Republic of Korea (36.8%), much higher than that of Sweden (30.0%), Mexico (27.6%), Australia (27.6%), the United States (23.5%), Greece (22.8%), Austria (21.0%), and Belgium (20.0%) [8]. According to the result of the COVID-19 National Mental Health Survey released by the Ministry of Health and Welfare (MOHW) of Korea, the group at risk of depression (a score of 10 points or higher out of 27 points) accounted for 18.95% in the December 2022 survey. Although the trend of a dramatic increase in depression has gradually subsided over time since the initiation of the COVID-19 National Survey, the percentage is greater by more than 5-fold compared to that in 2019 (3.24%), indicating overall deterioration of mental health among the general population in Korea [9].
After being hit by the COVID-19 infection, many people reported experiencing fear of contracting the disease, fear of a financial crisis, fear or disgust toward those who may have been infected, compulsive checking, and seeking reassurance about pandemic-related threats. Additionally, symptoms of PTSD were shown to exacerbate their depressive symptoms [10-12]. Studies have reported on mental health risks according to different demographic and socioeconomic characteristics such as age, gender, employment status, and financial situations, and the proportion of the high-risk group for depression was reportedly higher among single-person households, young people, people with low socioeconomic status, and the unemployed [8,13,14]. In addition, social distancing caused significant disruption to the existing patterns of social interactions, contributing to growing feelings of isolation among individuals, increasing depression. Moreover, the perceived impact of COVID-19 on daily life was positively associated with the severity of depression [15-17].
Most of the existing studies on depressive symptoms, one of the key indicators of mental health during the COVID-19 pandemic, examined the trend of high-risk groups according to the total scores measured using depression screening instruments as part of a mental health survey or identified causal variables (independent variables) of depression.
The present study conducted latent profile analysis (LPA) to identify the types of depression that require priority intervention in a pandemic and to understand the characteristics of depressive symptoms that better reflect the patterns of reallife depression. LPA assumes that the total population is composed of heterogeneous subpopulations and classifies the sample by type based on a variety of individual behavioral characteristics, thereby enabling the understanding of individual behaviors from a person-centered perspective [18,19]. Based on this approach of categorization or grouping, LPA allows the derivation of a high-risk group and identification of the characteristics of a target group that requires priority intervention; furthermore, it enables interventions customized to change the identified characteristics of the target group [20].
In addition, by adopting the methodology of LPA with utility, this study aimed to examine the antecedents of different types of depressive symptoms and the differences in post-traumatic growth (PTG) in the identified latent profiles. Although traumatic events have numerous negative physical and psychological consequences, people who have undergone trauma not only report negative experiences such as PTSD but also positive changes, referred to as “post-traumatic growth [21,22]" Thus, examining how the degree of PTG varies by the type of depressive symptoms and exploring the methods of recovery according to the identified types of depression would be imperative for experiencing positivity during COVID-19 and overcoming difficulties arising from the pandemic.
This study analyzed the data collected from the 5th survey of the COVID-19 National Mental Health Survey conducted by the Korean Society for Traumatic Stress Studies (KSTSS) commissioned by the MOHW of Korea. The period of March 2021, when the 5th survey was conducted, corresponded to the third wave of COVID-19, and with the successive emergence of major SARS-CoV-2 variants (alpha variant in December 2020, beta and gamma variants in January 2021, and delta variant in April 2021, confirmed in Korea). COVID-19 spread nationwide starting from the Seoul metropolitan area. Transmission and reports of confirmed cases increased in environments close to the daily living of the public, such as long-term care facilities, medical institutions, saunas, sports and leisure facilities, workplaces, and family and friend gatherings [23]. An indepth understanding of the impact of the pandemic on the mental health of the adult population in Korea could be gained through measuring depressive symptoms and the associated variables during a period of emergence of SARS-CoV-2 variants and their nationwide spread. The discovery of the variants led to unusually high level of threats related to the fear of infection, aversion, and rising disgust sensitivity toward others with a high risk of infection, and psychosocial distress related to the prolonged pandemic.
In this study, latent profiles of depressive symptoms were derived from the items of the Patient Health Questionnaire (PHQ-9) among the depression assessment scale. Additionally, the characteristics and predictors of each latent profile were investigated. Furthermore, differences in PTG according to the identified latent profiles were examined. The results of this study are expected to serve as useful evidence and data for developing intervention measures and policy proposals for situations during a pandemic.
METHODS
Participant
The participants of this study were 2,110 adults aged 19-71 years residing in Korea, who voluntarily participated in the 5th COVID-19 National Mental Health Survey24 conducted by the KSTSS upon commission from the MOHW in the first quarter of 2021.
Instruments
Depression
For assessing depressive symptoms, PHQ-9 was used. It was developed by Spitzer et al. [25] to assess the mental health status of individuals in primary health care centers. Han et al. [26] translated the scale into Korean, adapting it for the Korean population, and verified its reliability and validity. When administering PHQ-9, participants are asked to rate the duration that they have been bothered by the problems described in each item, in their recent episode of depression on a 4-point scale as follows: “0=Not at all,” “1=Several days,” “2=More than a week,” “3=Nearly every day.” The total score is 27, and the interpretation of each range of the total score is as follows: 0-4: normal level (minimal depression), 5-9: mild depression, 10- 19: moderate depression, and 20-27: severe depression. The cut-off score is 10. Higher scores indicate more severe depressive symptoms. The value of Cronbach’s α for testing the scale’s reliability was 0.920 in this study.
COVID-19 stress
COVID-19-related stress was measured using items developed for the Korean National Mental Health Survey conducted by the KSTSS under the commission of the MOHW. These items were specifically designed to capture pandemicrelated psychosocial stressors in the general population. This instrument consists of 15 items on stress that can be experienced following COVID-19. According to the extent of stress experience, the participants indicate their scores for each item on a 4-point scale as follows: “Never experienced=0,” “Experienced a little=1,” “Experienced much=2,” “Experienced very much=3.” Higher scores indicate a higher level of stress caused by COVID-19. Cronbach’s α for the scale’s reliability was 0.916 in this study.
Post-traumatic growth
For the assessment of PTG, the Korean version [27] of the Posttraumatic Growth Inventory (PGI) [28] was used. It consists of 16 items categorized into four factors as follows: changes of self-perception, increase of depth in interpersonal relationships, finding of new possibilities, and increase in spiritual/religious interest. Participants indicate their scores on a 6-point scale according to their experience of changes in each item of the PGI following a traumatic, stressful life event as follows: “0=never experienced,” “1=a very small degree of experience,” “2=a small degree of experience,” “3=a moderate degree of experience,” “4=a great degree of experience,” and “5=a very great degree of experience.” Higher scores indicate a higher level of PTG. Cronbach α for the scale was 0.955 in this study.
Data collection and ethical considerations
This study was conducted upon approval by the Kangwon National University Institutional Review Board (KWNUIRB-2020-03-004-007). Before data collection, the purpose and intent of the study were explained to the participants, and their voluntary consent was obtained. Data were collected through an online survey platform from March 29 to April 12, 2021.
Statistical analysis
For data analysis, SPSS 21.0 and Mplus 8.0 were used. First, SPSS 21.0 was used for analyzing the descriptive statistics of the variables and the reliability of the scales used in this study.
Second, LPA was performed using Mplus 8.0 to identify subpopulations according to the nine items in PHQ-9. LPA is designed to identify the types of latent subpopulations from the collected data by grouping individuals with similar patterns of responses into one group [19,29]. LPA enables optimization of the classification between latent profiles, accurate estimation of parameter values, and control of measurement errors by using latent variables. In addition, LPA has an advantage over cluster analysis in that it can reduce errors arising from the subjective judgment of researchers because it allows the optimal latent profile model to be determined using model fit indices [18,30].
To determine the number of latent profiles related to depressive symptoms, the information criteria used were Akaike’s information criterion (AIC), Bayesian information criterion (BIC), and sample-size adjusted BIC (SABIC). The smaller the value of these three indices, the more fit the model [31]. For comparative testing of the models, an adjusted likelihood ratio test (Lo-Mendell-Rubin likelihood ratio test: LMR-LRT) and parametric bootstrapped likelihood ratio test (BLRT) were conducted [32]. Classification accuracy was measured through entropy and posterior membership probabilities. Entropy has a value between 0 and 1, and a higher value indicates high accuracy in classification. An entropy value of 0.8 or higher indicates accurate classification [33]. Posterior membership probability is generally considered to be relatively accurate when the diagonal values of the matrix are 0.7 or above [34]. The above statistical analysis indices were synthesized with comprehensive considerations and by accounting for the interpretability, the optimal number of latent profiles was finally determined.
After classifying the latent profiles for depressive symptoms, a 3-step approach was adopted to analyze the impact of predictors and the difference in PTG according to the classified latent profiles [35]. The 3-step approach is a method of controlling any possible impact on grouping into classes or profiles when analyses are performed by including the independent and dependent variables in the model along with the latent classes/profiles. The method has the advantage of preventing the latent classes or profiles from being altered by the independent and dependent variables [36].
Study model
The study model developed and proposed in this study is illustrated in Figure 1. First, latent profiles of depressive symptoms were classified based on the participants’ responses to the questionnaire. Next, the significance of predictors assumed to affect the latent profiles of depressive symptoms was evaluated. Finally, the difference in the degree of PTG was tested for each latent profile.
RESULTS
Results of descriptive statistics
The descriptive statistics of the variables used in this study are outlined in Table 1. Of the 2,110 participants, 1,065 (50.5%) were male and 1,045 (49.5%) were female, and the average age of the participants was 44.9 years. The average score for depressive symptoms was 0.63, the average score for COVID-19 stress was 1.09, and the average score for PTG was 2.74.
Determination of the number of latent profiles
No clear methodology has been established for determining the optimal number of latent profiles in LPA, and the researchers usually make a judgment by comprehensively accounting for various figures, analyses, and interpretations related to the model fit [19,32]. The commonly applied procedure involves increasing the number of latent profiles one by one and comparing the models; this procedure was implemented in this study. When comparing between candidate models, optimal loglikelihood estimation was also performed for each model. The result of the LPA is presented in Table 2. When the number of latent profiles was set to 4, a group with a classification probability of 1 was obtained, indicating that the data in this study were not sufficient for the use of the 4-latent profile model. Thus, the 2-latent profile model and 3-latent profile model were comparatively analyzed. Furthermore, the information criteria values, AIC, BIC, and SABIC, were all lower when the number of latent profiles was 3 than when the number of latent profiles was 2, indicating that the 3-latent profile model provided a better fit. In addition, in the calculation of LMRLRT and BLRT, which compared the 2-latent profile model and 3-latent profile model, both methods indicated that the 3-latent profile model was more suitable, with statistical significance at the 0.1% level.
When entropy calculation was performed, the 2-latent profile model was closer to 1 than the 3-latent profile model, but when referring to the existing literature [37] entropy is a less preferred measure for determining the number of latent profiles or classes compared to other measures. Furthermore, as no particular problem was encountered with the classification probability of the 3-latent profile model, the 3-latent profile model was selected as the final model of choice in this study, even considering the analysis and interpretation to be described later.
Characteristics by the latent profiles
The estimated means of the latent profiles for the depressive symptoms of the participants is presented in Table 3, and the distribution is illustrated in Figure 2. When comparing the three groups, the first group (59.9% of the total respondents) was named the “overall low-level group” because it showed low scores in all PHQ items. The second group (29.8% of the total respondents) was named the “lethargy and physical symptoms group” because its scores for the following items were higher than those of other items: feeling tired or having little energy (1.593), trouble falling or staying asleep, or sleeping too much (1.405), little interest or pleasure in doing things (1.367), poor appetite or overeating (1.324), and feeling down, depressed, or hopeless (1.198).
The third group (10.3% of the total respondents) was named the “overall high-level group,” because it showed higher scores than those of the other groups in most of the PHQ items.
Testing the predictors of the latent profiles of depressive symptoms
The effects of predictors that were assumed to have an impact on the final model with the three latent profiles were tested. The results of analyzing variables with a statistically significant impact on the latent profiles among the variables of sex, age, and COVID-19 stress are presented in Table 4.
When comparing the latent profiles, female participants, younger participants, and participants with experience of physical and mental health problems were more likely to be classified into the lethargy and physical symptoms groups than into the overall low-level group. In addition, the participants with experience of the indifference of the society/community to the loss and damage, conflict with family members, conflict and distrust with neighbors, and fear of personal information disclosure were more likely to be classified into the lethargy and physical symptoms group than into the overall low-level group.
Younger participants and participants with experience of physical and mental health problems and those with experience of indifference of the society/community to the loss and damage were more likely to be classified into the overall highlevel group than into the overall low-level group. Furthermore, participants facing difficulty in returning to work following the loss of jobs or leave of absence while having less difficulty in academic or work performance due to environmental changes and lower stress levels from the trend of media coverage were more likely to be classified into the overall high-level group than into the overall low-level group. Participants with more experience of physical and mental health problems, less difficulty in academic or work performance due to environmental changes, more experience of conflict with family members, and more experience of rows over liability or legal disputes were more likely to be classified into the overall highlevel group than in the lethargy and physical symptoms group.
Testing the difference in PTG according to the latent profiles of depressive symptoms
The results of the examination of significant differences in PTG by latent profiles classified according to depressive symptoms are summarized in Table 5.
A significant overall difference in PTG across the three latent profiles was observed (χ²=31.42, p<0.001). Bonferronicorrected post-hoc comparisons revealed that the overall high-level group showed significantly lower PTG than both the lethargy and physical symptoms group and the overall low-level group. That is, the degree of PTG was lower with increasing severity of depressive symptoms.
DISCUSSION
In this study, data from a survey conducted during a period of heightened psychosocial stress due to the emergence of SARS-CoV-2 variants and their nationwide spread were used. LPA was conducted to classify the types of depressive symptoms among the general adult population in Korea and to examine the relationship between the latent types, predictors (antecedents), and outcomes. The main findings of this study are as follows.
First, when the depressive symptoms of the general adult population were grouped, the following latent profiles were obtained: “overall low-level group” (59.9%), “lethargy and physical symptoms group” (29.8%), and “overall high-level group” (10.3%). Although not all participants exceeded the clinical cutoff for depressive symptoms, the identified latent symptom patterns are considered to meaningfully reflect the heterogeneity of depressive symptom experiences in the general adult population. Among the three groups, the overall low-level group showed low scores in all items of depressive symptoms. The lethargy and physical symptoms group showed relatively high scores for the items of feeling tired or having little energy; trouble falling or staying asleep or sleeping too much; little interest or pleasure in doing things; poor appetite or overeating; and feeling down, depressed, or hopeless. The overall high-level group showed higher scores than the overall low-level group or the lethargy and physical symptoms group for most of the items of depressive symptoms, confirming that this group requires priority intervention. The distinct contribution of this study compared to previous studies is that the types of latent subpopulations were identified from the collected data by employing LPA for depressive symptoms, one of the key indicators of mental health related to the COVID-19 pandemic. In particular, this study revealed that, although not categorized as a high depression risk group, almost 30% of the general adult population showed various patterns of depressive symptoms, such as a hopeless and lethargic state, loss of interest, fatigue, and problems related to sleep and diet. This result is consistent with the finding reported in previous studies [10,13,38] that depressive symptoms, such as the sense of lack of energy all over the body or feeling depressed, down, and hopeless for no reason, may persist and mental health problems may aggravate during a period when the economic and social impact of a prolonged pandemic takes a toll on the general population. And this pattern still reflects a clinically meaningful form of subclinical depressive symptoms such as loss of energy, anhedonia, and sleep problems. Previous studies [39-41] have demonstrated that such subthreshold or subclinical depressive symptoms constitute an early manifestation of major depression and are associated with a substantially increased risk of subsequently developing major depressive disorder, along with significant functional impairment.
Second, the results of examining the effects of predictors on the latent profiles of depressive symptoms were as follows. Women were more likely to be classified into the lethargy and physical symptoms group than into the overall low-level group. This supports the results of previous studies [42-44] that COVID-19-related depression has more impact on women and points to the need for psychological interventions that can alleviate lethargy, loss of interest, hopelessness, poor appetite or overeating, and sleep problems in women. Next, when analyzing by age, people aged 19-29 years compared to those in their 50s were more likely to be classified into the overall high-level group than into the overall low-level group. People aged 19- 29 years compared to those in their 60s were also more likely to be classified into the lethargy and physical symptoms group or the overall high-level group than into the overall low-level group. These results are consistent with the reports from previous studies [17,42,45], where younger age was found to be positively associated with the experience of depressive symptoms, pointing to the necessity of interventions for the prevention of the overall worsening of depressive symptoms and interventions for lethargy and physical symptoms among young adults.
Due to the impact of COVID-19 stress, the experience of physical and mental health problems of individuals was positively associated with the likelihood of being classified into the lethargy and physical symptoms group or the overall highlevel group compared to the overall low-level group. Individuals with experience of physical and mental health problems were also more likely to be classified into the overall high-level group than into the lethargy and physical symptoms group. It can be inferred that increased experience of physical and mental health problems is associated with increasing levels of depressive symptoms [2,10]. To facilitate effective coping for physical and mental health problems that may occur with the prolonged pandemic, problems such as limitation in the use of medical institutions and disruption of public agency services should be resolved, thereby enabling access to appropriate treatment for individual problems or conditions. Next, the experience of the indifference of the society/community to the loss and damage and the experience of difficulty in returning to work following the loss of jobs or leave of absence were positively associated with the likelihood of classification into the overall high-level group. This result supports the finding of previous studies [46,47] that financial difficulties caused by the pandemic had an impact on the deterioration of mental health. Therefore, in addition to providing financial aid at the governmental level, means of psychological support are needed for low-income groups, especially those with markedly reduced income due to the pandemic. Next, difficulty in academic or work performance due to environmental changes (work-fromhome arrangement, online classes, etc.) and stress from the trend of media coverage were positively associated with the likelihood of classification into the overall low-level group than into the overall high-level group. One possible explanation for this counterintuitive finding is that individuals who previously experienced difficulties with in-person academic or occupational activities—potentially related to depressive symptoms— may have perceived the shift to remote or non-faceto-face environments during the pandemic as relatively less burdensome. For such individuals, reductions in social interaction and performance-related demands may have temporarily alleviated functional strain, resulting in fewer reported academic or occupational difficulties despite elevated depressive symptom profiles. This interpretation is consistent with prior longitudinal studies [48,49] showing heterogeneous psychological responses to pandemic-related environmental changes, including evidence that depressive symptoms declined over time or returned to baseline as individuals adapted to reduced social demands. The interpretation of stress related to media coverage warrants a more nuanced consideration. Although excessive or uncontrolled media exposure may exacerbate psychological distress, actively engaging with media and social media platforms can also enhance perceived information control, emotional support, and collective coping, thereby buffering depressive symptoms. Consistent with this perspective, prior studies [50-52] have demonstrated that media use during the COVID-19 pandemic may exert both detrimental and protective effects on mental health, depending on usage patterns and coping-related functions. Next, participants with experience of conflict with family members were more likely to be classified into the lethargy and physical symptoms group or overall high-level group rather than the overall low-level group, and more likely into the overall high-level group than into the lethargy and physical symptoms group. Thus, when family members spend longer time together at home due to environmental changes such as work-from-home arrangements or online classes and tightened control of social distancing policy, conflicts experienced within the family increase, and all items of the depressive symptoms are likely to be scored high. Consequently, these individuals may become a high depression risk group or experience lethargy and physical symptoms. This result supports the findings of previous studies [53-55] that under situations of COVID-19 transmission, the changes in family life and relationships had an impact on perceived psychosocial stress, suggesting the need for active implementation of family education, family counseling, and couples counseling to improve and build strong and healthy family relationships. Next, the participants with experience of conflict and distrust with neighbors and experience of disputes over liability or legal disputes were found more likely to show lethargy and physical symptoms and found more likely to have aggravated overall depressive symptoms.
The success of K-Quarantine can be attributable to the Korean government’s agile responses, built on a sound system of democracy and civic responsibility. However, with the prolonged continuation of COVID-19 quarantine measures, the so-called empathy fatigue phenomenon, where the public becomes increasingly fatigued and loses the ability to form empathy, has worsened, and various forms of social conflicts such as problems of inequality and interpersonal relationships have occurred [56]. It will take considerable time and capacity building for society to develop enough resilience to resolve the social problems and conflicts caused by COVID-19, and to this end, the development of a social support system that enables the creation and consolidation of community solidarity is necessary.
Third, when the differences in PTG were analyzed by the latent profiles of depression, the overall high-level group showed significantly lower PTG than the other two groups (the overall low-level group or the lethargy and physical symptoms group). This result is, in conflict with the report of previous studies [57-59] that the general population who reported depression and anxiety caused by a high level of COVID-19 stress experienced PTG over time. PTG was found to be low in the overall highlevel group among the latent profiles of depressive symptoms. Therefore, focusing on enhancing resilience and capacity building through psychological acceptance-based interventions is necessary, which can facilitate changes in self-perception, increase in depth of interpersonal relationships, and finding new possibilities.
The limitations of this study and suggestions for future research are as follows. First, it is difficult to conclude that the participants in this study constitute a group with clinically significant depressive symptoms. Depressive symptoms were assessed using the PHQ-9, which is a self-report screening measure rather than a diagnostic interview. Accordingly, caution is warranted when interpreting the overall findings, particularly with respect to clinical implications. Second, the data used in this study are cross-sectional data, which have limitations in terms of identifying the long-term characteristics of individuals, unlike the methodology of panel surveys. Since the clinical conditions related to the mental health of adults, such as depressive symptoms, can change depending on external environments or events, collecting data over a long period may allow for latent transition analysis or growth mixture modeling to be adopted for longitudinal analysis to clarify how depressive symptoms change over time and the factors affecting this change. Because certain COVID-19-related stress items may overlap conceptually with depressive symptoms, the direction of the observed associations should be interpreted with caution. Furthermore, the cross-sectional design of the study precludes confirmation of temporal precedence. Third, although the COVID-19 stress items were developed for a national survey, their psychometric properties have not been fully validated. Therefore, the interpretation of results related to COVID-19-related stress should be made with caution.
Despite these limitations, this study has made significant contributions in that unlike existing methodologies or approaches, it grouped the depressive symptoms of the general adult population in Korea during the COVID-19 pandemic into latent profiles and presented directions for interventions for depression by adopting a person-centered latent profile model.
PDF Links
PubReader
ePub Link
XML Download
Full text via DOI
Download Citation
Print
