INTRODUCTION
The coronavirus disease-2019 (COVID-19) pandemic has precipitated alterations in the daily lives of individuals worldwide. The initial stages of the pandemic, following the first confirmed case of COVID-19 in Korea on January 20, 2020, strained the foundational aspects of the healthcare system. The surge in confirmed cases led to a shortage of hospital beds, resulting in instances where infected individuals, awaiting care, succumbed to the virus at home [1]. In response, the government established “living treatment centers” for asymptomatic and mild patients, initiated drive-thru screening clinics, and implemented social distancing measures for quarantine purposes [2]. The daily count of confirmed cases fluctuated, reaching its peak in March 2022. On August 31, 2023, the cumulative number of confirmed coronavirus cases was recorded at 34,571,873 and the cumulative number of deaths was 35,934. The Korean government responded by lowering the crisis alert level due to the spread of COVID-19 from “Serious” to “Alert” as of June 1, 2023 [3].
In the initial phases of the COVID-19 pandemic, uncertainty about the virus’s characteristics and transmission routes elevated fear and anxiety. The enforcement of social distancing measures led to social isolation, triggering various psychological challenges, including depression and anxiety. In contrast to past infectious diseases such as swine flu and Middle East Respiratory Syndrome (MERS), the absence of a vaccine or specific treatment for COVID-19 heightened concerns about infection, negatively impacting various aspects of life and exacerbating psychological difficulties [4,5]. While previous infectious diseases prompted discussions about the mental health of confirmed patients, their families, and directly involved medical staff, COVID-19 induced widespread anxiety and subsequent depression among the general public [6]. This fostered a prevailing belief that “anyone can be contagious and infectious” [7].
Numerous endeavors have been undertaken to detect the fear, anxiety, and stress experienced by individuals in the context of the COVID-19 situation. Some studies have utilized pre-existing scales adapted for infectious disease scenarios such as COVID-19, while others have employed newly developed scales specifically tailored for this purpose. For instance, the Pandemic-related Perceived Stress Scale (PSS-10) for COVID-19 [8] comprises 10 questions adapted from the existing PSS-10 [9] to align with the unique challenges posed by the pandemic. This scale assesses difficulties in coping and maintaining a sense of control in the face of the COVID-19 situation. The COVID Stress Scale [10,11] comprises 36 questions and is divided into six subscales: danger, socio-economic consequences, xenophobia, contamination, traumatic stress, and compulsive checking. The Fear of COVID-19 Scale [12-14] consists of seven questions designed to measure psychological and physical symptoms of fear associated with the COVID-19 situation. The Coronavirus Anxiety Scale [15] assesses anxiety symptoms, including dizziness, nausea, sleep problems, paralysis, and loss of interest in eating, associated with the impact of the coronavirus situation. Additionally, the COVID-19 Peritraumatic Distress Index [16] was developed to examine the overall psychological distress experienced by the general public. It measures the frequency of anxiety, depression, specific phobias, cognitive changes, avoidance and compulsive behaviors, physical symptoms, and the loss of measured social functioning. The scales gauge fear, stress, as well as psychological and physical symptoms prevalent in the context of the coronavirus situation. Each scale possesses distinct constructs and measurement ranges tailored to capture the varied facets of the psychological impact associated with the pandemic.
In an infectious disease situation, emotional responses such as fear, anxiety, and disgust experienced by individuals are influenced by social and cultural factors. The Korean government’s response to COVID-19, known as K-Quarantine, involves strategies to curb the spread by tracking the movements of confirmed patients through epidemiological investigations and conducting rapid polymerase chain reaction tests. However, concerns was raised regarding the disclosure of personal information of confirmed patients [15]. Furthermore, Korean society has traditionally exhibited a predominant collectivist culture, emphasizing the avoidance of harm to others and the community [17]. In the context of an infectious disease situation, fear extends beyond individual concerns about infection to encompass fears experienced within the framework of interpersonal relationships. For instance, Koreans not only harbor concerns about personal infection with COVID-19 but also experience fears of causing inconvenience to their families, others, or the community due to their infection, as well as a notable fear of being stigmatized as a confirmed case. While various COVID-19-related scales have been developed, there is currently no scale specifically measuring the fear of COVID-19 infection within a collectivist culture such as Korea. Hence, this study aimed to develop and validate a COVID-19 fear scale that reflects the collective cultural context.
In the scale development and validation process, factor analysis based on correlations between individual items is mainly used, but it is difficult to confirm information on whether individual items and response categories function properly for the purpose of the test through factor analysis alone [18]. Recently, there has been a surge in research focusing on the development and validation of scales based on Item Response Theory [19] In line with this trend, this study aimed to assess the unidimensionality of the scale and analyze the fit and response scale of individual items using the Rasch measurement model [20], based on individual item responses. Additionally, we sought to validate the COVID-19 Infection Fear Scale (CIFS) by analyzing differential item functioning based on gender and age, and conducting correlation analyses with the frequency of anxiety, depression, and coping strategy usage.
METHODS
Participants
A total of 1,002 adults, who were registered as panel members with research company, Korea Research, participated in an online survey and completed a series of self-report questionnaire. The inclusion criteria were adults aged between 19 and 70 who voluntarily agreed to participate in the study. Exclusion criteria were not specified. The recruitment process for this study, conducted in May 2020, involved the utilization of a cluster sampling technique to randomly select 1,004 adults within the same age range. The sampling procedure aimed to achieve a sampling error of ±3.1% with a 95% confidence level. All participants provided written informed consent by ticking a box before filling out their response through an online survey platform. Table 1 presents an overview of the general characteristics of the survey participants (n=1,002). The sample comprised 512 males (51.1%) and 490 females (48.9%). The mean age of the participants was 44.31 years, with a standard deviation (SD) of 13.62. Analysis of the age distribution between sexes indicated no statistically significant difference (males: 43.97 years [SD=13.741], females: 44.66 years [SD=13.499], t[1000]=-0.804, p=0.421). Regarding occupational status, 578 respondents (57.7%) reported employment in sectors such as agriculture, self-employment, production, sales, and office work. Additionally, the sample included 157 housewives (15.7%), 82 students (8.2%), 72 individuals classified as “others” (7.2%), and 113 unemployed individuals (11.3%). Random assignment of participants divided the survey sample into two groups, with sample 1 employed for exploratory factor analysis (EFA) and sample 2 utilized for confirmatory factor analysis (CFA). No statistically significant differences were observed between the two samples in terms of sex (χ2[1]=0.399, ns), age groups (χ2[4]=2.332, ns), or occupation (χ2[8]=10.864, ns). The present study was approved by the Ethics Committee of Kangwon National University (IRB approval number: KWNUIRB-2020-03-004-008).
Measures
COVID-19 Infection Fear Scale
Based on previous literature on infection diseases such as SARS and MERS, as well as current news related to the ongoing COVID-19 pandemic, items describing COVID-19 infection fear were gathered. The aim was to develop a comprehensive list of items that accurately reflect individual’s perceptions of both the fear of infection itself and fear of negative outcomes following infection. In a multidisciplinary research team consisting of experts in fields such as clinical psychology, social work, and psychiatry, we categorized the fears that the public is currently experiencing regarding COVID-19 into two distinctive domains: 1) the fear of infection, and 2) the fear of negative outcomes post-infection. Initially, 7 items were developed, with two additional items added later to enhance the scale’s comprehensiveness. Finally, 9 items were finally developed for each specific domain, ensuring content validity. Participants were asked to rate each item on a 4-point Likert scale (0=“not at all,” 1=“not much,” 2=“somewhat,” 3=“very much”), where high scores indicate a greater level of fear regarding COVID-19 infection and its consequence. Supplementary Table 1 describes the domains and each item of the CIFS.
Generalized Anxiety Disorder-7 item scale
The Generalized Anxiety Disorder-7 (GAD-7), developed by Spitzer et al. [21], was used to assess anxiety levels of the participants. Each of the 7 items was rated on a 4-point scale (0=not at all to 3=nearly every day), with higher scores indicating a great level of anxiety. Cronbach’s α for this study was 0.922.
Patient Health Questionnaire-9
The Patient Health Questionnaire-9 (PHQ-9), developed by Kroenke and Spitzer [22], was utilized to measure depression. This measure consisted of 9 items. Participants were asked to rate on a 4-point scale (0=not at all to 3=nearly every day). The higher scores reflected a great level of depression (Cronbach’s α=909).
Coping strategy frequency for stress situations, including infectious diseases
The research team developed a scale to assess how frequently individuals employ coping strategies in response to stress situations, including infectious diseases such as COVID-19. This scale consists of a total of 14 items, which encompass 7 items related to coping strategies for general stress situations (e.g., maintaining a balanced diet & sufficient sleep; engaging in light exercise or physical activity; spending quality time with family) and 7 items tailored to coping strategies for infectious disease situations (e.g., seeking COVID-19-related information; adhering to preventive measure like maskwearing & hand hygiene; practicing self-restraint & physical distancing when going out; staying at home for 3-4 days when feeling unwell). Participants rated the frequency of utilizing each coping strategies on a scale (0=no used at all, 1=used occasionally, 2=used frequently, 3=used very frequently), with higher scores indicating a more frequent application of the respective coping strategies. The scale demonstrated good internal consistency with a reliability coefficient of 0.787 in this study.
Statistical analysis
Basis statistical analysis including demographic characteristics was conducted using SPSS 27.0 (IBM Corp., Armonk, NY, USA). To conduct exploratory and CFA, participants were randomly divided into two sets (set 1 for exploratory & set 2 for CFA). In an exploratory analysis, the appropriate number of factors was determined using the minimum residual extraction method and oblimin rotation. Parallel analysis was conducted with the number of factors fixed at 1 and 2, followed by EFA using the Principal Axis Factoring extraction method and oblimin rotation to confirm model fit. Parallel analysis and factor analysis fit were conducted using jamovi [23].
A CFA was conducted using structural equation modeling to confirm the two subfactors of the COVID-19 fear scale, namely, fear of infection (Factor 1) and fear of negative outcomes post-infection (Factor 2). If the initial model fit was unsatisfactory, modification indices (MI) were examined and the final model was selected using the correlation between measurement errors as a free parameter. Model fit was verified using χ2 test, root mean square error of approximation (RMSEA), comparative fit index (CFI), and Tucker Lewis index (TLI) [24]. Andrich [25] Rating Scale Model [26], a polytomous model among the Rasch models, was applied to analyze the fit and response scale of individual questions using jMetrik [27]. Examination included category probability curves for response categories, item characteristic curves and 95% confidence intervals, and internal and external fit indices for individual questions. Both internal and external fit indices, sensitive to abnormal item response patterns, were considered, ensuring they had an expected value of 1 [28]. The generally accepted fit index range is 0.5 to 1.5. Values less than 0.5 suggest overfitting, prompting a check for redundant questions. Values exceeding 1.50 may indicate misfit, and removal of the question could be considered [29]. The Generalized Mantel-Haenszel test was used to analyze the discrimination function questions based on sex and age. For age, the analysis was divided into two groups using the average age as a reference. A Pearson correlation analysis was conducted to verify the construct validity of the COVID-19 fear scale. This involved examining the relationships between the GAD-7, the frequency of suicidal thoughts (item 9), and coping strategies of the PHQ-9 and PhQ-9.
RESULTS
Descriptive analysis
Table 2 presents descriptive statistics and item analysis of the COVID-19 fear scale. Modified item-total correlations ranged from 0.655 to 0.798 and internal consistency reliability was 0.933 for the entire scale. For the fear of infection subscale, modified item-total correlations ranged from 0.773 to 0.855, and internal consistency reliability was 0.918. For the fear of negative consequences after infection subscale, modified item-total correlations ranged from 0.663 to 0.792, and internal consistency reliability was 0.888.
Exploratory analysis
The Kaiser-Meyer-Olkin measure demonstrated a sampling adequacy value of 0.916, while Bartlett’s test of sphericity yielded a chi-value of 3,281 (df=36, p<0.001). These results affirm the suitability of the dataset for subsequent EFA. Parallel analysis indicated that two factors are suitable. The results of the exploratory analysis are presented in Table 2, with the number of factors fixed at 1 and 2, respectively. In the onefactor model, the explained amount of the total variance was 59.6%, and the factor loading of each item ranged from 0.679 (item 6) to 0.811(item 4). In the two-factor model, factor 1 explained 34.2% of the total variance and factor 2 explained 33.7%, increasing the total cumulative explained amount to 67.8%. For factor 1, questions 5, 6, 7, 8, and 9 showed factor loading values above 0.30, and for factor 2, questions 5, 6, 7, 8, and 9 showed factor loading values above 0.30. Item 5, “if I and my family become infected, I am afraid that I will be quarantined for treatment.” Spans for both factor 1 and 2, but the factor loading values of factor 2 is higher and semantically indicates a negative effect after infection. As it was judged to be closer to fear of the outcome after infection, it was decided that it was appropriate to include it in factor 2. The RMSEA of factor 2 was better than those of factor 1, thought it did not reach a satisfactory level. Nevertheless, considering content validity and interpretability by experts at the time of item development, the two-factor model was considered more appropriate. Please refer to Table 3 for the detailed results of the exploratory analysis for the CIFS.
Confirmatory analysis
Utilizing the second sample (set 2), CFA was conducted with the two subscales of the CIFS, based on the results of EFA detailed above. The fit chi-square of the initial model was significant (χ2=171.350, df=26, p<0.001) and the model fit indices such as CFI and TLI showed a good-of-fit statistics. However, RMSEA was not satisfying level. The researchers’ professional judgement was further explored to interpret a realistic and interpretable structure of the model; 1) The MI, which is the chi-square that decreases when the correlation between fixed measurement errors is estimated with a free parameter, 2) although the upper-level factors are different, correlation between 4 measurement errors was allowed based on expert judgement on related lower-level factors [30]. As a results, the empirical model fit indices were greatly improved and showed good values. Comparing the initial model and the modified model revealed significant enhancements in the goodness-of-fit statistics (χ2[22]=81.087). Specifically, CFI and TLI surpassed (Table 4 and Figure 1).
Item response theory analysis
Item fit
Through a comparison of expected probability values predicted by the Rasch measurement model and actual respondent response values, the validity of individual items was assessed. In Rasch modeling, two essential fit indices, infit mean square (internal fit) and outfit mean square (external fit), are employed. These indices evaluate how well individual items align with respondents’ ability levels. Outfit index is sensitive to response patterns that include extremely easy or difficult items relative to the respondents’ ability level, while the infit index, considering the impact of a minimal number of aberrant item responses, places emphasis on the fit of item with the respondent’s ability level [28]. Ideal fit indices fall within the 0.5 to 1.5 range, with values around 1 indicating a good alignment between observed and expected values [29]. Infit values of fear of infection subscales ranged from 0.83 to 1.06, outfit values ranged from 0.76 to 1.03, confirming all values fell within an appropriate range. Infit values of fear of negative consequences after infection subscale ranged from 0.78 to 1.24, outfit values ranged from 0.78 to 1.23, confirming all values were within the acceptable range (Table 5).
Reliability
Upon analyzing the scale, the reliability at both the item and persons levels indicated good values (all>0.8). The separation index suggests that the variability of the items within each subscale (6.29 and 5.75) is very good (above 3.0), reflecting the diversity of the items. While the value of separation index in person is at the value of 2.27 and 2.32 confirmed good person separation, effectively distinguishing between individuals. These findings collectively affirm the robustness and reliability of both subscales at the scale level (Table 6).
Construct validity
The Pearson correlations were calculated to investigate construct validity between the scale used in the present study. The results showed that the total scores of the CIFS were significantly associated with the total scores of subscales of the CIFS, anxiety, depression, suicide ideation as measured by the PHQ-9 (item 9), and coping strategies for infection disease situations. Subscale 1 (fear of infection) and 2 (fear of negative consequences after infection) showed similar results, confirming good construct validity (Table 7).
DISCUSSION
The main objective of the present study was to develop and validate the CIFS. By the recognition of the vital role that cultural norms and values play in shaping psychological assessments, this study aimed to encapsulate the fears related to COVID-19 infection within the family and other significant social environments such as neighborhoods and workplaces. These settings are particularly important in collectivist societies like Korea, where communal well-being significantly influences individual psychological well-being, behavior and attitudes. The main findings are as follows; Firstly, the exploratory analysis confirmed two-factor model for the CIFS. Secondary, two factors model was further supported through CFA, which highlighted the scale’s robust psychometric properties. In an item-response theory analysis, both infit and outfit values showed fell within an acceptable range, indicating a strong fit. Moreover, the significant correlations between the CIFS, the mental health measures-specifically, the PHQ-9, the GAD-7, and suicidal ideation as measured by item 9 of the PHQ-9 demonstrated the scale’s commendable construct validity. Our findings provide strong evidence for the reliability and validity of the CIFS.
EFA and CFA confirmed a two-factor model, aligning with our theoretical distinction between fear of infection and fear of negative outcomes post-infection. The initial CFA model showed significant improvement in fit indices after modifica-tion, highlighting the importance of expert judgement in refining model. Importantly, the Rasch analysis indicated that all items fit well within the model parameters, with infit and outfit indices falling within the acceptable range of 0.5 to 1.5. This suggests that each item is appropriately aligned with the underlying construct of fear being measured.
The CIFS demonstrated high internal consistency for both subscales (fear of infection and fear of negative consequences post-infection) with Cronbach’s alpha of 0.918 and 0.888, respectively, and 0.933 for the overall scale. Additionally, the item and person separation indices were satisfactory, indicating that the scale reliably distinguishes between different levels of fear among respondents. The results showing the significant correlations between the CIFS and related constructs such anxiety (GAD-7), depression (PHQ-9), suicidal ideation (PHQ-9, item 9), and coping strategies confirm the scale’s construct validity. These correlations indicate that higher levels of fear as measured by the CIFS are associated with higher levels of anxiety, depression, and the use of coping strategies, which is consistent with existing literature on the psychological impact of pandemics. These results of the current study align with and extend the findings of previous research on fear and psychological distress during pandemics. For example, Ahorsu et al. [12] developed the Fear of COVID-19 Scale (FCV-19S) and validated the relationship between COVID-19 scale (FCV-19S), which similarly showed high internal consistency and validated the relationship between COVID-19 fear and psychological distress. Their study demonstrated that higher levels of fear of the FCV-19S were significantly related to increased anxiety and depression.
Similarly, the COVID-19 Stress Scale (CSS) developed by Taylor et al. [11] identified multiple dimensions of stress related to the pandemic, including fear of danger and contamination, which align with our fear of infection scale. Notably, our study extends this by specifically addressing the correlation coefficient between the fear of negative consequences post-infection and psychological distress (anxiety, depression, suicidal thoughts) of subscale 2 was slightly higher than that of subscale 1 (fear of infection). This indicates that fear of negative consequences is more closely related to psychological distress, underscoring its significance of considering the aftermath of infection, such as quarantine and social stigma, which may be particularly salient in collectivist cultures like Korea.
Additionally, fear of COVID-19 infection showed significant positive correlation with the frequency of using coping strategies related to infectious diseases situations, such as mask-wearing and physical distancing. Our results are in line with the previous study that developed the Pandemic-related PSS-10, which emphasized coping difficulties [8]. This finding suggests that an appropriate level of fear can lead to preventive behaviors, which are crucial during a pandemic.
The FCV-19S measures general and individual-level of emotional and physical responses to the fear of virus, while the CSS focuses on a broad range of pandemic-related stressors. The CIFS is unique in its dual focus on one’s perception on infection itself and its social consequences, making it a valuable tool for understanding inner fear in cultures where the impact of family and community are paramount. In particular, the CIFS addresses concerns about how one’s infection might affect others, a key factor in cultures, like Korea. This makes the CIFS more sensitive to detecting fears that are highly relevant in collectivistic societies but may not be as well captured by scales like FCV-19S, CSS, and PSS-10.
This scale can be a valuable tool for mental health professionals and researcher in identifying not only fear of infection but also its direct social impacts through the quarantine, stigma, and long-term social isolation. Understanding the specific fears related to infection and post-infection consequences can inform targeted interventions and support mechanisms to mitigate these fears and improve mental health outcomes. Furthermore, the significant associations with anxiety, depression, and coping strategies highlight the interconnectedness of these psychological constructs. Interventions designed to reduce fear may also positively impact other areas of mental health, such as reducing anxiety and depression and promoting effective coping strategies.
Despite the findings, several limitations should be noted. The study’s cross-sectional design limits the ability to draw causal inferences. Longitudinal studies are needed to exam how fear of COVID-19 evolves over time and its long-term psychological impact. Additionally, while the sample was diverse, future research should aim to validate the CIFS in different cultural contexts to enhance its generalizability. Future research should also explore the impact of specific demographic factors, such as socioeconomic status and occupational risk, on the fear of COVID-19. Understanding these nuances can further refine the CIFS and enhance its applicability across various populations. One of the limitations of this study is that test-retest reliability was not assessed due to practical and urgent constraints posed by the evolving nature of the pandemic. Future research should aim to investigate test-retest reliability of the CIFS to future establish its stability over time. Finally, in the early stage of the COVID-19 pandemic, a COVID-19-specific infection anxiety scale had not yet been developed, so an existing infection anxiety scale was modified for use. Similarly, if another pandemic occurs in the future, it would be beneficial to propose methods for the rapid development and validation of tools to measure infection-related anxiety. One potential approach is to adapt existing scales, such as the CIFS, by adjusting disease-specific references. This flexibility would allow the scale to be relevant in assessing fear and anxiety in the context of new pandemics. Additionally, utilizing online platforms for data collection could speed up creating reliable and valid measurement tools.
In conclusion, the CIFS is a reliable and valid tool for measuring fear related to COVID-19 infection and its consequences. The scale’s development and validation underscore the importance of culturally sensitive measures in understanding and addressing the psychological impact of pandemics. The CIFS has significant potential for guiding clinical assessments and informing public health strategies to support mental health during ongoing and future public health crisis.
PDF Links
PubReader
ePub Link
XML Download
Full text via DOI
Download Citation
Supplement
Print
