A confirmatory factor analysis of a cognitive model of COVID-19 related anxiety and depression

The mental health impact of the COVID-19 pandemic has been significant, with many regions across the globe reporting significant increases in anxiety, depression, trauma, and insomnia. This study aims to validate a potential cognitive model of maintenance factors of COVID-19 related distress by examining psychological predictors of distress, and their goodness-of-fit as a coherent model. Participants from the general population (n = 555) were recruited using a cross-sectional on-line survey design, assessing Demographic factors, Anxiety, Depression, Loneliness, COVID-19 related distress, Trauma Cognitions related to COVID-19, Rumination, Safety Behaviours, Personality Factors, and Mental Effort related to COVID-19. A series of stepwise linear regressions found that components of the model were significant and accounted for a large percentage of variance when examining Covid-19 related distress (R2 = 0.447 Covid Stress Scale), Anxiety (R2 = 0.536 DASS-Anxiety Subscale) and Depression (R2 = 0.596 Depression DASS-subscale). In a confirmatory factor analysis, Loneliness, Post-Traumatic Cognitions about Self, Post-Traumatic Cognitions about the World, Emotional Stability, and Mental Effort related to COVID-19 loaded onto a single factor. The final model showed adequate fit (CFI = 0.990, TLI = 0.983, RMSEA = 0.053 (0.027–0.080), GFI = 0.986, SRMR = 0.0216, χ2 = 23.087, p = .006). The results highlight the importance of cognitive factors, such as post-traumatic cognitions, rumination, and mental effort in maintaining COVID-19 related distress.


Introduction
The COVID-19 pandemic has led to close to four million deaths (World Health Organization, 2021). In addition, the virus has had a profound social and economic impact worldwide, leading to increased unemployment rates, racial discrimination, socioeconomic disparities, and mental health disorders (Kola et al., 2021;Wu et al., 2021).
The mental health impact of the COVID-19 pandemic has been significant, with many regions across the globe reporting significant increases in anxiety, depression, distress and insomnia (Salari et al., 2020;Wu et al., 2021). The restrictions associated with the COVID-19 pandemic, particularly mandatory lockdowns, have been linked to increased loneliness, isolation, post-traumatic stress and depression, with prolonged lockdown duration being associated with higher levels of distress (Brooks et al., 2020;Gong, Cui, Xue, Lu, & Liu, 2021).
Meta-analyses of risk factors for COVID-19 related distress have reported a range of demographic factors: female gender, younger age (21-40), higher level of education, having a close relative or friend who has COVID-19, pre-existing mental and physical health conditions, low socioeconomic status and risk of infection with COVID-19 (Santabárbara et al., 2021;Wang, Kala, & Jafar, 2020;Wu et al., 2021). Studies carried out on samples from the UK and Ireland reported similar risk factors (Hyland et al., 2020;. In addition, repeated close exposure to the virus, such as working in a healthcare setting or having been placed under medical quarantine, has been linked to more severe increases in symptoms of COVID-19 related distress (First, Shin, Ranjit, & Houston, 2020;. A recent study by Faustino, Vasco, Delgado, Farinha-Fernandes, and Guerreiro (2022) highlighted the negative effect that early maladaptive schemas could have on mental health within the context of COVID-19. The study found maladaptive schemas such as mistrustfulness and vulnerability to harm were strongly associated with increased levels of COVID-19 related anxiety and distress.
Typically, cognitive models of distress have highlighted that a person's beliefs and related behaviours can create and maintain a sense of threat that maintains a person's anxiety at times of relative safety (e.g. Beck & Haigh, 2014;Salkovskis, 1996), a key question remains whether the existing evidence-base of psychological treatments applies in a situation of unique, realistic, and ongoing health, economic, and social threat. Cognitive models have addressed various factors that contribute to the maintenance of distress. These factors include selective attention, misappraisal of physical symptoms, avoidance behaviours, overoverestimation of the possible risk of infection or death, body scanning, and hyper-vigilance (e.g. Ehlers & Clark, 2000;Rachman, 2012;Salkovskis & Warwick, 2001). A recent study by Mohammadkhani, Akbari, Shahbahrami, Seydavi, and Kolubinski (2022) examined the role of Cognitive Attentional Syndrome in relation to , which highlighted that dysfunctional beliefs and ruminations about the pandemic were significant predictors of COVID-19 related anxiety. Furthermore, increases in attentional control were linked to lower levels of rumination and, in turn, lower levels of COVID-19 related distress (Guo, Yang, Elhai, & McKay, 2021). This highlights the role that rumination and selective attention play in the formation and maintenance of COVID-19 related distress. These recent studies highlight that traditional cognitive factors are potentially useful and appropriate targets for treatment in the unique context of the COVID-19 pandemic.
A second question raised by the pandemic is how best to conceptualise the psychological impact on the general population. Studies have argued for the appropriateness of understanding the population-level impact through the conceptualisations of health anxiety, depression, OCD, and trauma Dennis, Radnitz, & Wheaton, 2021;Karatzias et al., 2020). Recent research has argued that a broader trauma-based conceptualisation of the pandemic (rather than specifically PTSD) could be applicable (Shevlin, Hyland, & Karatzias, 2020;Yuan et al., 2021).
This study examines potential maintenance factors of COVID-19 related distress in a sample from the general population. The proposed model is a conceptualisation; the model proposes the following potential maintenance factors as predictors of COVID-19 related distress: Negative Emotion (Anxiety, Depression, Loneliness), Physical Anxiety Symptoms, Post-Traumatic Cognitions related to COVID-19, Safety Behaviours, Rumination and Mental Effort related to COVID-19. This paper utilises a general population sample. Similar research has previously used samples with diagnosed mental health difficulties (Calkins, Berman, & Wilhelm, 2013;Hofmann, 2007). However, given the universal nature of the impact of the COVID-19 pandemic, the widespread psychological impact within the general population (Kola et al., 2021;Wu et al., 2021) and the multiple psychological disorders related to the COVID-19 pandemic (Salari et al., 2020;Wu et al., 2021), it seemed particularly important to validate a model within a sample of the general population rather than a clinical subgroup.
These proposed maintenance factors of the Cognitive Model of COVID-19 related distress (Aswad & Gaynor, 2021) are drawn from general cognitive therapy literature (David, Cristea, & Hofmann, 2018) through the author's clinical contact with individuals experiencing COVID-related distress in the early months of the pandemic, and specific data emerging from research on the psychological impact of the COVID-19 pandemic. Specifically, typical negative emotions associated with living through the pandemic have been identified as anxiety, including physical anxiety sensations, depression and loneliness (Ettman et al., 2020;Groarke et al., 2020). Traumatic and peri-traumatic responses have been common population-wide reactions (Forte, Favieri, Tambelli, & Casagrande, 2020;Liu, Zhang, Wong, Hyun, & Hahm, 2020;Silver, 2020). Likewise, rumination has been identified as a common psychological response to the pandemic (Satici, Saricali, Satici, & Griffiths, 2020;Zhou, MacGeorge, & Myrick, 2020). Intrusive rumination has previously been found to mediate the relationship between traumatic experiences and PTSD (Wozniak, Caudle, Harding, Vieselmeyer, & Mezulis, 2019). Of particular interest but also particularly difficult to measure during heightened governmental safety measures is the role of safety behaviours in maintaining distress during COVID-19 (Knowles & Olatunji, 2021;Schimmenti, Billieux, & Starcevic, 2020). Although not usually discussed in cognitive models, mental effort and cognitive load be related to anxiety (Calvo & Eysenck, 1992). Mental effort is particularly relevant to the COVID-19 pandemic as the constant influx of new information and exposure to media created a condition of high cognitive load and increased feelings of worry and fear about COVID-19 (Sasaki, Kuroda, Tsuno, & Kawakami, 2020). People's intrinsic cognitive load has increased due to the constantly changing environment, new restrictions and high levels of information related to COVID-19 (Rathore & Farooq, 2020). Similarly, although not typically included in a cognitive model, it was noticeable that many of the negative emotional impacts of COVID-19 have been moderated by personality factors, in particular, Emotional Stability (Asselmann, Borghans, Montizaan, & Seegers, 2020;Gubler, Makowski, Troche, & Schlegel, 2021) and thus personality factors have been added to the proposed model. The proposed conceptualisation model is in line with classic cognitive models of distress but also grounded in the emerging science of psychological responses to  This study aims to examine potential cognitive predictor variables of COVID-19 related distress and their validity as a single model with strong goodness-of-fit.

Design
A cross-sectional quantitative survey was carried out to measure potential maintenance factors of COVID-19 related distress in a convenience sample of the general population recruited via social media.

Participants
An a priori power analysis was conducted using G*Power version 3.1.9.7 (Faul, Erdfelder, Lang, & Buchner, 2007) for sample size estimation. The estimated effect size for this study was 0.3, which is considered a small effect size under Cohen's (1988) criteria. With a significance criterion of α = 0.05, power = 0.95, and df = 9, the minimum sample size needed with this effect size is N = 263 for confirmatory factor analysis.
Participants were recruited using convenience sampling via social media platforms (N = 555). The survey was advertised through major social media platforms. To participate in the survey, participants were required to be over the age of 18 and had to be able to understand and complete the survey in English. Responses were collected using Qualtrics. Participants were not compensated for participating. Demographic-related data were collected, including gender, age, nationality, employment status, living environment, number of children in the household, number of adults in the household, whether the respondent had an underlying health condition, whether the respondent had a household member with a health condition, loss of income, whether a person had experienced a COVID-19 infection, whether the respondent had been a close contact of a person with COVID-19 (Table 1), the respondent's self-perceived risk from COVID-19.
The following measures were used. An adapted version of the COVID-19 Stress Scale (CSS; Taylor et al., 2020) was used to measure levels of COVID-19 related stress. The Covid Stress Scale included a total of 24 items across three of the five original subscales of the CSS: (1) danger and contamination fears, (2) compulsive checking and reassurance seeking, and (3) traumatic stress symptoms about COVID-19. The subscales measuring xenophobia and socioeconomic consequences were not included. Participants were asked to what extent they felt worries related to COVID-19 over the last six months on a scale of 1 (Not at all) to 5 (Extremely). The internal reliability of the CSS was high (α = 0.929). There are no clinical cut-off scores for the CSS.
The UCLA Loneliness Scale (UCLA, Russell, Peplau, & Cutrona, 1980) was used to measure loneliness. The UCLA contained a total of 8 items which were assessed on a 4 Point Likert scale ranging from 1 to 4. Higher scores on the UCLA indicated higher levels of negative constructs of loneliness. The Internal reliability of the UCLA was measured as high (α = 0.886).
An adapted version of the Post Traumatic Cognitions Inventory (PTCI; Foa, Ehlers, Clark, Tolin, & Orsillo, 1999) was used to measure post-traumatic thinking concerning the COVID-19 pandemic. The PTCI included a total of 28 items across 2 subscales: Negative Cognitions about self (21 items) and Negative Cognitions about the World (7 items). Participants were asked to evaluate a series of statements on a scale of 1 (Totally disagree) to 7 (Totally Agree). The Self-Blame subscale was excluded due to the universal nature of the pandemic. The participants were instructed to use their experience of the pandemic as the reference experience. Higher scores on the PTCI indicated higher levels of posttraumatic thinking regarding the COVID-19 pandemic. The PTCI, as measure of cognition, does not have clinical cut-offs. However, the original validation paper described successfully differentiating between three groups: non-trauma (mdn = 45.5; SD = 34.76), trauma without PTSD (mdn = 49.0; SD = 23.52), and trauma with PTSD (mdn = 133.0; SD = 44.17), based on the Total and subscale scores. The internal reliability of the adapted version of the PTCI was measured as high (α = 0.964) and across both subscales (Self (α = 0.963) and World (α = 0.894)). Safety Behaviours were measured by an adapted version of the Safety Behaviours Scale (Freeman, Garety, & Kuipers, 2001). However, the Safety Behaviour Scale was excluded from the final analysis due to poor internal reliability and a low response rate (α = 0.568).
The Ruminative Response Scale (RRS, Treynor, Gonzalez, & Nolen-Hoeksema, 2003) was used to measure rumination. The RRS was measured on a 4-point Likert scale ranging from 1 (Never) to 4 (Always) to measure the frequency of ruminating thoughts or behaviours. The Internal Reliability of the RRS was measured as high (α = 0.890).
This study used an adapted version of the Rating Scale of Mental Effort (RSME; Zijlstra & Doorn, 1985) to measure Mental effort related to COVID-19. Participants were asked how much mental effort thinking about COVID-19 costs them daily. A slider ranging from 0 (absolutely no effort) to 100 (extreme effort) was used to record participant responses. Higher scores on this scale indicated higher levels of Mental Effort related to COVID-19. The reliability of the original scale showed a strong correlation between test and retest scores (r = 0.96).
The Ten-Item Personality Inventory (TIPI, Gosling, Rentfrow, & Swann, 2003) was used to measure personality dimensions as described in the 5 Factor Model of Personality. There were 10 items distributed evenly across the 5 subscales: Agreeableness, Extraversion, Openness to Experiences, Emotional Stability and Conscientiousness. Each item consisted of either two desirable or undesirable descriptions of a trait.
Participants were asked to rank on a scale of 1 (strongly disagree) to 7 (strongly agree) how much they believed each pair of descriptions applied to themselves. The internal reliability of the subscales of the TIPI ranged from low to high: Agreeableness (α = 0.374), Extraversion (α = 0.681), Openness to Experiences (α = 0.522), Emotional Stability (α = 0.738) and Conscientiousness (α = 0.480).

Procedure
Data were collected from December 2020 to February 2021. This was the longest "lockdown" period in the Republic of Ireland, where a large percentage of the sample was drawn. Convenience sampling was used to recruit participants across various social media platforms such as Facebook, Instagram, and Twitter. Participants were given an informed consent sheet describing the nature and content of the survey and were asked to confirm that they were 18 years or older to continue with the survey. Participants were then asked to complete a series of demographic-related questions before completing the subsequent scales. After completing the scale portion of the survey, participants were given a debrief sheet that included links to relevant support services.

Data analysis
Data analyses were carried out using IBM SPSS version 26. Missing data analyses were carried out to remove incomplete responses under a 75 % completion cut-off point; Reliability and normality analyses were carried out. Log 10 transformations were carried out on any skewed data. Demographic variables were recoded into dummy variables to allow for further analyses. Independent t-tests were carried out to assess demographic variables and COVID-19 related distress. Stepwise linear regressions examined demographic and psychological predictors of COVID-19 related distress. A confirmatory factor analysis was carried out using IBM AMOS to examine the factor loading of the primary variables to emerge from the regression analyses.

Ethics
Ethics approval was granted by Human Research Ethics Committee University College Dublin Research Ethics Committee (HS-20-55-Gaynor).

Preliminary results
Four hundred eighty-eight participants were removed from the dataset as they failed to provide consent or complete 1-2 % of the total survey, below the agreed threshold of 75 % completion. The remaining sample included 555 participants. Forced response was used on all scales. Closing the survey before completion was deemed a withdrawal from the study, resulting in a nearly complete data set for the remaining participants.
The UCLA Loneliness Scale was recoded to range from 1 (Never) to 4 (Often) to reflect the original scale design. Reliability analyses were conducted to assess the internal reliability of the CSS, DASS, UCLA, PTCI, Safety Behaviours, RRS and TIPI, and all relevant subscales. All scales bar the Safety Behaviour Scale and two subscales of the TIPI showed high reliability above 0.68; The Safety Behaviours Scale was excluded from further analysis. The TIPI was retained due to the adequate reliability of the other subscales (see Supplementary Table 1).
Normality tests were carried out across all scales. Normality was assessed based on fitting a skewness (− 1/+1) and kurtosis (− 3/+3) range and scores on the Kolmogorov-Smirnoff test of normality in line with standard approaches (Durbin, 1970;Tabachnick & Fidell, 2019). Based on these tests, log 10 transformations were carried out on all scales and subscales bar the RSME, the Stress subscale of the DASS, the Extraversion subscale of the TIPI, the agreeableness subscale of the TIPI, the emotional stability subscale of the TIPI and the negative cognitions about the world subscale of the PTCI. Post log 10 transformation, only the CSS successfully passed the Kolmogorov Smirnoff test (Kolmogorov Smirnoff = 0.031, df = 555, p = .200). For all other scales, normality was met based on the acceptable range for skewness and kurtosis (see Supplementary Table 2).

Demographics
All demographic variables were recoded into binary variables. A series of independent t-tests were conducted to compare levels of COVID-19 related distress as measured by the CSS. The results of these independent t-tests highlighted that female gender, younger age (under 35), living with family, change in employment, having children, loss of income, having been infected with COVID-19, having an underlying health condition, and living with a person with an underlying health condition showed higher levels of COVID-19 related distress. Following a Bonferroni correction, only female gender, change in employment status and living with a person with an underlying health condition showed significantly higher levels of COVID-19 related distress. There were no differences in nationality, number of adults in the house and close contact with an infected person (please see Table 2).
Levels of Traumatic Cognitions measured by the PTCI were assessed based on the group means listed by Foa et al. (1999). Forty-seven per cent of participants displayed traumatic cognitions associated with a trauma diagnosis on the PTCI World subscale. Furthermore, 55.8 % of participants scored within the trauma category on the PTCI Self subscale. This highlights that significant levels of trauma can be found within the sample based on the cut-offs described in Foa et al. (1999).
Anxiety and depression were assessed based on the clinical cut-off points described for the DASS-21 (Lovibond & Lovibond, 1995). Fortyseven per cent of the sample were not depressed, 13.6 % reported mild levels, 16.8 % moderate levels, 9.2 % severe levels, and 13.2 % extreme levels. Sixty-two per cent of the sample did not report significant anxiety; 12.5 % reported mild levels; 7.4 % moderate levels; 5.7 % severe levels; and 12 % extreme levels.

Stepwise linear regression
Prior to further analysis, assumptions of linearity, normality, independence of observations and homoscedasticity were checked using scatterplots and histograms. Due to possible multicollinearity, the stress subscale of the DASS was removed from the analysis. Three stepwise regressions were carried out with the following dependent variables: Covid Stress (CSS), Anxiety (DASS-subscale) and Depression (DASSsubscale). All demographic variables and the following scales were included in each regression: UCLA, PTCI Self, PTCI World, RRS, RSME and all subscales of the TIPI (Table 3). The results showed that when the dependent variable was defined as Covid Stress (CSS), the significant predictors were: Negative Cognitions about the World (PTCI), Negative Cognitions about Self (PTCI), Mental Effort related to COVID-19 (RSME) and Emotional Stability (TIPI) (F (5,549) = 90.738, p < .000, R 2 = 0.447).

Confirmatory factor analysis
For this paper, a single-factor confirmatory factor analysis was conducted (see Fig. 2). The following variables loaded onto a single factor: Loneliness (UCLA), Negative Cognitions about the World (PTCI), Negative Cognitions about Self (PTCI), Rumination (RRS), Mental Effort related to COVID-19 (RSME) and Emotional Stability (TIPI). Assumptions of multivariate normality were assessed based on kurtosis values (kurtosis = − 0.290), and an absolute cut-off point of 5 was used; as such, this model satisfies the assumption of multivariate normality. The univariate normality of all scales was assessed prior to this analysis (see preliminary results section). Mahalanobis distance was used to screen for outliers; no potential outlier was identified in the search when a criterion of p < .001 was applied. The DASS subscales of Anxiety and Depression were not included in the final model due to possible misspecification with the post-traumatic cognitions inventory subscales (1.96 cut-off criteria). Overall, the model showed a good fit to the data (see Table 4) based on the full range of model fit indices, including CFI, TLI, RMSEA, GFI and SRMR. All factor loadings were above 0.5 and showed significance at an alpha level of p < .001. Negative Cognitions about Self showed the highest factor loading (0.90). The lowest loading was Mental Effort (RSME) (0.53) (see Fig. 2). The model showed poor fit on the χ 2 ; however, the overall model fit was assessed based on the fit indices below as the χ 2 can be highly influenced by sample size (Alavi et al., 2020) (see Table 4).

Discussion
This study aimed to examine potential cognitive maintenance factors of COVID-19 and their coherence as a single model. The cognitive behavioural model of COVID-19 related distress (Fig. 1) proposed by Aswad and Gaynor (2021) included Negative Emotions (Loneliness), Physical Anxiety Symptoms, Post-traumatic Cognitions related to COVID-19, Safety Behaviours, Rumination and Mental Effort related to COVID-19. Three regression analyses measuring distress across the domains of anxiety, depression and COVID stress were carried out. The regression analyses produced significant models accounting for large percentages of variance, indicating that Emotions (Loneliness), Trauma Cognitions related to COVID-19 (World and Self Subscales), Rumination, Mental Effort related to COVID-19, Personality Factors (Emotional Stability) and demographic factors (Living Environment, Underlying Health Condition and Number of Children) were significant predictors of a variety of measured forms of distress.
A confirmatory factor analysis was carried out, showing that Loneliness (UCLA), Trauma Cognitions related to COVID-19 (PTCI World and Self subscales), Rumination, Mental Effort related to COVID-19 and Emotional Stability (TIPI) successfully mapped onto a single factor (see Fig. 2). This highlights the importance of cognitive factors in the maintenance of COVID-19 related distress. These findings mirror the findings of Mohammadkhani et al. (2022), highlighting that dysfunctional beliefs and ruminations about the pandemic were significant predictors of COVID-19 related anxiety. Research carried out by Table 3 Stepwise regressions examining different aspects of distress: Covid Stress (CSS), Anxiety (DASS-subscale) and Depression (DASS-subscale).   (Aswad, Gaynor, in preparation). Note: Variables identified by square boxes represent causal factors as identified through meta-analyses. Variable identified by ovals represent maintenance factors, tested in this paper.  Faustino et al. (2022) further highlighted the role of negative cognitions as predictors of COVID-19 related distress. Furthermore, decreases in attentional control were linked to higher levels of rumination and, in turn, higher levels of COVID-19 related distress (Guo et al., 2021). This highlights the role that rumination and selective attention play in the formation and maintenance of COVID-19 related distress.
In comparison, demographic factors, which in previous studies such as Salari et al. (2020) and Shevlin, McBride, et al. (2020) were strong predictors of COVID-related distress, were excluded by the regression analyses when psychological factors were included in the models. It is interesting to note that demographic factors were the only significant predictors of anxiety, particularly the presence of an underlying health condition, living environment and the number of children. All three domains were predicted by personality factors such as Emotional Stability and Conscientiousness. Therefore, it can be suggested that while such demographic variables are related to increased anxiety levels, models also need to include intrapersonal psychological factors in predicting COVID-related distress.
This paper aimed to address a secondary hypothesis highlighting that Post-Traumatic Cognitions related to COVID-19 were a maintenance factor of COVID-related distress. Horesh and Brown (2020) and others (Kira et al., 2020) have argued that responses to the pandemic mirror symptoms of collective trauma, including hyper-vigilance, avoidance behaviours, negative cognitions, negative effect changes, fear of death or illness and fear of the future. While there has been some argument about whether a traditional categorisation of PTSD is applicable in the context of COVID-19 (van Overmeire, 2020), Shevlin, Hyland, & Karatzias (2020) have argued that the COVID-19 pandemic meets the criteria for a traumatic event and that a Type III trauma conceptualisation of the pandemic is the most applicable understanding. This study's results support this conceptualisation that living through a pandemic is for the general population experiencing a continuous, collective trauma, similar to living through a period of terrorism or an apartheid regime. Preliminary results carried out as part of this study indicated that based on cut-off points described in Foa et al. (1999), 47.9 % of the sample fell into the trauma category based on scores on the negative cognitions about the world subscale of the PTCI.
Furthermore, 55.8 % of participants fell into the trauma category based on scores on the negative cognitions about the self subscale. This highlights significant levels of continuous traumatic cognitions present within the sample. This further supports the secondary hypothesis that traumatic cognitions associated with COVID-19 represent an important maintenance factor of covid distress.
The results of the regression analyses have highlighted those posttraumatic cognitions related to COVID-19 were common in this general population sample and were significant predictors of all three measured domains of distress. Both domains of the Post-Traumatic Cognitions Inventory successfully and strongly mapped onto the factor of COVID-19 related distress.
Notably, the t-test and the regression analysis highlighted that neither contact with the virus nor close contact with an infected person were significant predictors of COVID-related distress. This is in line with previous research on collective trauma, such as the events of 9/11, where those experiencing trauma symptoms following the incident did not have any direct contact with or experience the event (Marshall et al., 2007). This is of particular interest as studies on previous pandemics such as SARS primarily examined continuous trauma-related cognitions in those who had close contacts with the virus, such as frontline workers and those who contracted SARS (Hong et al., 2009). The cognitive nature of the identified maintaining factors reinforces the importance of meaning-making rather than direct exposure to maintaining distress. As such, the role of frequent news broadcasts with updates on the virus and government safety guidelines may heavily influence beliefs about the threat even in populations that were not directly exposed, as proximity and intensity of the threat were highlighted daily (Xiong et al., 2020).
There were a variety of limitations regarding this study. The results of this study were based on data obtained at a single time point; in the context of COVID-19, this marks a significant limitation as findings obtained in one stage of the pandemic may be different from later stages. Therefore, it is important to note that other factors that may mitigate COVID-19 distress, such as vaccine availability and increased freedom of movement, were not assessed as part of this study. As this model is a conceptualisation of a cognitive maintenance model of Covid-19 related distress, it is important to note that the data were collected at a single time point. A longitudinal study would be required to validate the proposed model further. This study's results also reflect the cultural context in which they were obtained, particularly relevant in a global pandemic with a wide range of different governmental responses. The data used in this study was obtained using convenience sampling, which may limit the findings' generalisability. The final sample predominantly reflected a European cultural context, with a relatively large portion self-identifying as female. Many of the short forms of the measures were used to create a survey manageable in an appropriate timeframe. This highlights the need to replicate the study with larger, more representative samples using more in-depth assessments of various factors.
Furthermore, due to the low reliability and low rate of responses to the Safety Behaviours Questionnaire, the behavioural element commonly cited as a maintaining factor of distress was excluded from all analyses. Safety Behaviours form an important aspect of COVID-19 distress. However, due to the complexity of the pandemic, many outside factors, such as government mandates, may significantly affect the domain of what can be considered abnormal avoidant behaviour. Further research should consider the impact of behavioural variables such as safety behaviours to increase the model's accuracy.
The model proposed in this study addresses a crucial service applicability issue, as previous research has heavily investigated the causal factor of distress. However, there needs to be more focus on maintenance factors. This model highlights that a cognitive behavioural approach toward treating these maintenance factors is an appropriate intervention, especially focusing on trauma-related cognitions. A CBTbased intervention has been developed and tested based on this model. The paper outlining this intervention is currently in preprint (Aswad & Gaynor, 2021). This intervention focused on addressing the core factors described in the current model. A self-help manual based on the examined factors was also distributed at the time of writing . Future research should utilise the model outlined in this study as a framework for a CBT-based treatment.
In conclusion, the results emphasise that a cognitive model, including domains such as Loneliness, Post-Traumatic Cognitions related to COVID-19, Rumination, Mental Effort related to COVID-19 and Emotional Stability, may be an accurate representation of the psychological maintenance factors of COVID-19 related distress. The final model illustrates the vital role those post-traumatic cognitions related to COVID-19 play in the maintenance of distress in the context of the COVID-19 pandemic and suggests that a conceptualisation of continuous, collective trauma may be applicable within this context. The final model highlights the role that cognition-based factors play in maintaining distress.

Ethics approval
The questionnaire and methodology used in this study was approved by the Human Research Ethics Committee of University College Dublin (HS-20-55-Gaynor).

Consent to participate
Informed Consent was obtained from all individual participants in this study. No information that could be used to identify a participant was used in this study.