COVID-19 Relates to Both PTSD and PTG in a Non-clinical Population, Why?

Abstract We assessed levels of post-traumatic stress disorder (PTSD) and post-traumatic growth (PTG) during the COVID-19 pandemic in the UK using an online questionnaire, in May and June 2020, during “lockdown.” Some 440 participants took part and 91.8% of the sample was female. Large numbers of participants had high levels of PTSD, with some 50.2% with “probable PTSD”. Yet 49.5% of participants also demonstrated high levels of PTG. Regression analysis revealed rumination to be key predictors for PTSD and PTG.


Introduction
COVID-19 emerged in the Wuhan Province of China in December 2019 and soon spread around the world. By March 2020, the World Health Organization (WHO) announced that the disease was in fact a global pandemic (Sominsky et al., 2020;Wind et al., 2020; World Health Organization Regional Office for Europe, 2020). Much of the world enforced strict hygiene protocols as well as social distancing, incorporated working from home and restricted travel (Wright et al., 2021). As of the 22nd of January 2022the disease has killed 5.6 million people worldwide (Wolrdometer, 2022) 1 , sparking reminders of the first SARS outbreak back in 2002 (Wang, Horby, et al., 2020). However, it was soon established that COVID-19 was more life-threatening than the first SARS outbreak. Therefore, on the 23rd of March 2020, the UK government announced the first national lockdown.
In terms of the research on COVID-19, interest soon arose about the mental health effects of the pandemic, focusing particularly on students and healthcare workers (HCWs). At the beginning of the pandemic, research into both students and HCWs found scores attributable to probable post-traumatic variables. Therefore, considering these variables is paramount in understanding the implications of the pandemic on members of the general public from within the UK.

Hypothesis 1
We predict that in a non-clinical sample during the early stages of the COVID-19 pandemic there will be evidence of PTSD.

Hypothesis 2
We predict there will be no evidence of PTG as the pandemic was only three to four months old at the time of the survey.
We also looked at the differences between "vulnerable" 2 individuals and non-vulnerable individuals and between keyworkers and non-keyworkers.

Measures
Impact of event scale-revised (IES-R; Weiss & Marmar, 1997) The IES-R scale, measures whether PTSD symptoms are experienced by individuals who have survived a potentially traumatic event(s), in this instance, the psychological impact of COVID-19. A previous study that assessed the psychometric properties of the IES-R found that it demonstrated both good reliability and validity when traumatized and non-traumatized individuals were compared (Rash et al., 2008). The IES-R contains a total of 22-items that are measured on a 5-point Likert scale, that ranges from 0 (not at all) to 4 (extremely). There is no specific cutoff score, however, scores over 33 are considered a cause for concern, as the higher the score, the higher the likelihood of PTSD (Creamer et al., 2003). The overall scale reliability in this study from Cronbach's a was .90.
Post traumatic growth inventory (PTGI; Tedeschi & Calhoun, 1996) The PTGI was used to measure whether individuals experienced growth following COVID-19 and contains 21 questions each answered on a 6-point Likert scale from 0 to 5. Considering what level each of the numbers represented, to comply with this research study, the wording was somewhat changed. Therefore, 0 which normally represented-"I did not experience this change as a result of my crisis" was altered to-"I did not experience this change as a result of COVID-19", with the same being done for 5-"A very great degree as a result of my crisis" was changed to "A very great degree as a result of COVID-19." These were the only two items that required any adaptation. As with many of the other scales, a higher score was consistent with experiencing PTG, with a cutoff score of 45 based on research into psychosis (Mazor et al., 2016). The PTGI overall demonstrated good validity and reliability, based on previous studies (Lenz et al., 2021;Shakespeare-Finch & Enders, 2008). The overall scale reliability in this study on Cronbach's a was .91.
The following questionnaires were also incorporated to allow for the completion of regression analyses to assess their predictive relationships with both PTSD and PTG. These measures were chosen as research has shown that these are important in the study of PTG (see Henson et al., 2021 for a review of PTG) and we wanted to further assess their relevance to PTSD. To assess these variables, we used: the Brief COPE scale (Carver, 1997); The Multidimensional Scale of Perceived Social Support (MSPSS; Zimet et al., 1988); The Connor-Davidson Resilience Scale-10 (CD-RISC-10; Connor & Davidson, 2003); The Life Orientation Test-Revised (LOT-R; Scheier et al., 1994), and The Event Related Rumination Inventory (ERRI; Cann et al., 2011). Reliability analysis of these questionnaires revealed Cronbach's a scores ranging from .76 to .95. The Centrality of Event Scale (CES; Berntsen & Rubin, 2006) was also included but owing to an administrative error some items were missed off the CES when creating the Qualtrics survey. The internal reliability of the shortened scale of 12 items was .90.

Participants
The final sample consisted of 440 participants, with age, gender, vulnerability, and keyworker status being accounted for (Table 1). In terms of racial background, the majority of the sample were White British (91.8%). (Vulnerability status. In Britain the National Health Service designated the entire population as "vulnerable" or "not vulnerable". While lockdown restrictions were imposed on the entire population, these were particularly stringent for the vulnerable groups, who were forbidden from leaving their houses for 10 weeks).

Procedure
Convenience sampling was used to collect data between May and June 2020, with participants completing the online questionnaire by a link posted on social media. A post was shared which provided a brief overview of the research and when participants clicked on the link, they were provided with the Participant Information Sheet explaining the nature of the research in more depth and that by completing the questionnaire and pressing submit, they consented to their data being analyzed. The questionnaire link was created using QualtricsXM. This cross-sectional research was conducted and distributed online via various social media platforms (i.e., Facebook, Twitter, LinkedIn) and was approved by the University of Bolton Ethics Committee.

Data analysis
Descriptive statistics and frequencies were calculated for both the IES-R and PTGI total scores and subscales. PTSD was divided into "probable PTSD" and "no PTSD" using the suggested cutoff scores. The same was performed for PTG with "high PTG" and "low PTG" based on the suggested cutoff scores. The vulnerable group was compared with the non-vulnerable group on all dependent variables (DVs), with 10 comparisons in total (IES-R score, MSPSS total score, CES score, LOT-R score, ERRI divided into intrusive and deliberate rumination, brief COPE divided into approach and avoidant coping, PTGI and finally, CD-RISC-10).
Keyworkers were compared with non-keyworkers. Finally, regression analysis was conducted into what variables best predicted post-traumatic stress and post-traumatic growth.

Results
Hypothesis 1: We predicted that in a non-clinical sample during the early stages of the COVID-19 pandemic there would be evidence of PTSD. As expected, the majority of our sample (N ¼ 221) displayed high levels of PTSD (50.2%) scoring 33 or above on the IES-R in line with the cut off score suggested by Creamer et al. (2003). This hypothesis is confirmed. Hypothesis 2: We predicted there will be no evidence of PTG. Contrary to our hypothesis this sample also experienced high levels of PTG, scores of 45 or above on the PTGI (N ¼ 218; 49.5%). This hypothesis is not supported (

Additional analyses
Vulnerable versus non-vulnerable participants As noted earlier, the UK National Health Services declared a number of individuals to be clinically vulnerable (the middle author being one). We might assume there might be some differences between this group and the general population. After completing a Bonferroni adjustment .05 Ä 10 ¼ .005, with only significant values of <.005 being considered. On 10 comparisons on all the study DVs, there were NO significant differences.

Keyworkers versus non-keyworkers
In the UK, keyworkers were those individuals who were on the frontline and worked in the health service or care sector. People who had to work and were not allowed or unable to work from home. Again, it might be expected that there might be some differences between them and the general population. On the same 10 comparisons, there was only one significant comparison. Following the completion of a Kruskal-Wallis test and with a Bonferroni adjustment of p<.005, only optimism out of the 10 dependent variables was found to have any significance (Kruskal-Wallis H ¼ 22.929, df ¼ 2, p ¼ .000). Participants of 65þ were found to be more optimistic (M ¼ 14.09, SD ¼ 4.52).

Male versus female
To establish comparisons regarding males and females, a further Kruskal-Wallis test was completed with the same Bonferroni adjustment (p<.005). On comparing the 10 dependent variables, NO significant comparisons were found.

Regression analysis for PTSD and PTG
To further understand the role of the nine variables on PTSD, a linear regression was completed using the enter method. PTSD was inputted as the Dependent Variable whilst the remaining nine variables were added as predictors. Multicollinearity diagnostics were performed, with the VIF statistic being below 10, suggesting no multicollinearity. The regression analysis revealed a significant model (F (9, 405) ¼ 41.47, p<.001) which explained 46.8% of the variance but only intrusive rumination and avoidance coping were significant predictors ( Table 3). The same analysis was completed to better understand PTG. Multicollinearity diagnostics were performed with the VIF statistic revealing no multicollinearity present. The analysis revealed a significant model (F (9, 405) ¼ 20.36, p<.001) which explained 29.6% of the variance but only centrality of event, deliberate rumination and social support were significant predictors (Table 4).

Discussion
We correctly predicted high levels of probable PTSD in the general population. We were wrong about PTG. Regression analysis highlighted the importance of rumination.
It was established that participants were experiencing symptoms relative to probable PTSD, which supports previous research into students and HCWS, as well as individuals infected with the disease Huang et al., 2020;Rossi et al., 2020;Sun et al., 2021;Tan et al., 2020;Tomaszek & Muchacka-Cymerman, 2020;Wang, Huang, et al., 2020), and studies relevant to the general population Liu et al., 2020;Wang, Pan, et al., 2020). Our findings  combined with these results demonstrate the severity of COVID-19 regarding mental health and that no matter what your background or whether you become infected or not, you could still experience PTSD symptoms or some form of psychological effect. Furthermore, research into frontline nurses found that they were experiencing traumatic stress, but did display signs of positive change, which they attributed to PTG but could not confirm this (Nowicki et al., 2020). Furthermore, Feingold et al. (2022) offer additional support for our findings as they too found high levels of PTG, but in a sample of frontline HCWs. Although our sample were from the general population, our findings can support their attribution as we found higher levels of PTG to be experienced, so it could be suggested that during the pandemic this relationship is not job-dependent and could in fact be something created by society as a whole. In contrast to our findings, research conducted on a Turkish population (Ikizer et al., 2021) found relationships to be present between PTSD and PTG suggesting cultural differences are important regarding experiences of the COVID-19 pandemic. However, considering the same study but in relation to rumination, our findings support this research, implying that cultural differences have no impact regarding levels of both PTSD and rumination. Accounting for the relationships found regarding avoidant coping these were as expected. Our findings regarding both approach and avoidant coping refute previous research which found both coping strategies predicting wellbeing but support the finding for avoidant coping and PTSD (Dawson & Golijani-Moghaddam, 2020). Even though wellbeing and growth are positive outcomes, it would appear that not the same coping strategies are beneficial. Other research into HCWs and non-HCWs found that active coping was mainly associated with having a positive effect on PTSD, whilst maladaptive coping was more negative (Ciułkowicz et al., 2021), which partially supports our research in a general population sample. It appeared that our participants utilized avoidant coping strategies, with expected results being found.
Considering PTG, during COVID-19 it was found that positive and negative coping strategies predicted growth (Fu et al., 2021), refuting our findings. The disparities present might be explained by the use of different scales for coping strategies. It could be suggested that as our sample displayed higher levels of PTG, they utilized deliberate rumination as we found associations between the two variables. This supports previous research into both students and general population samples that found relationships to be present between deliberate rumination and PTG (Shigemoto, 2022;Zeng et al., 2021), whilst intrusive rumination was found to be associated with PTSD (Ikizer et al., 2021), which was further confirmed in our study.
There are some limitations to this research, with the main weakness being the use of self-report measures, which do not allow for understanding around the participants' responses and how they truly felt during COVID-19. Therefore, qualitative analysis is required to allow for further knowledge around the psychological impact of COVID-19. This study also failed to account for previous trauma. Even though all the questions were directed to COVID-19, previous trauma could have influenced their responses. As we only used social media, the generalizability of this study could be questioned, as we never accounted for individuals with no Internet access. Although, the research participants were not from one specific location, which does help improve the generalizability of this study. The sampling obtained was disproportionately female, but this is often the case with online surveys.
It is apparent that research is being conducted on the psychological effects of the pandemic with a clear focus on PTSD, with more interest being shown around PTG. Most of the research is related to students and HCWs so future research should continue by focusing on general population samples regarding other positive psychology variables, with the potential creation of interventions. It would also be of importance for researchers and practitioners to focus on elements of positive psychology to help increase the levels of PTG by including such variables in training regimes and policies. His main research interests are positive psychology, happiness, alcohol addiction, bereavement, recovery from mental health problems and autoethnography. His enthusiasm for Psychology is greater now than it has ever been. One of his greatest pleasures is supervising undergraduate, Master's and PhD students and seeing them get their work published.

Notes
Gill Brown -Gill is a Teaching Professor and the head of Psychology at the University of Bolton. Prior to her academic career, she worked within the prison service, providing psychological services to offenders. She also has experience working within youth offending services and alongside forensic community mental health teams. Her PhD focused on the effectiveness of a joint crisis planning intervention for service users with severe mental illness, in reducing compulsory detention under the Mental Health Act. Current research interests span forensic and mental health topics, with a particular interest in disparities in mental health service needs and treatment provision for ethnic minority groups.

Data availability statement
The data that support the findings of this study are available from the corresponding author, [CLW], upon reasonable request.