Turning to Religion During COVID-19 (Part I): A Systematic Review, Meta-analysis and Meta-regression of Studies on the Relationship Between Religious Coping and Mental Health Throughout COVID-19

The COVID-19 pandemic and the many associated socio-economic changes constitute a stressful event that required adaptation to new, dynamic, and often threatening conditions. According to the literature, coping strategies are one of the factors that determine a person’s degree of adaptation to stressful situations. A systematic review and meta-analysis was performed on the relationship between religious coping and selected indicators of mental health. Due to the large amount of data, this work has been divided into two parts: this first part discusses positive mental health indicators, while the second discusses negative mental health indicators (Pankowski & Wytrychiewicz-Pankowska, 2023). A systematic review of PubMed, Science Direct, the Cochrane Library, Google Scholar, the Database of Abstracts of Reviews of Effects, and Google Scholar databases was carried out. In addition to the synthesis of information obtained from the research, a meta-analysis of correlation was also performed to determine the strengths of the relationships between the analysed variables, and selected moderators were assessed using meta-regression. Quality of life, well-being, satisfaction with life, happiness, and post-traumatic growth were the positive mental health indicators considered. Meta-analyses indicated a statistically significant relationship between positive religious coping and flourishing (well-being) with overall correlation values of 0.35 [0.30; 0.40]. Further calculations also indicated a relationship between negative religious coping and flourishing − 0.25 [− 0.34; − 0.15]. Data synthesis shows associations between religious coping and such indicators as satisfaction with life and post-traumatic growth, but these issues require further investigation.


Introduction
The COVID-19 pandemic was an event that had a massive effect on the world around us. It affected practically all spheres of life: physical health (Daher et al., 2020), the economy (Ashraf, 2020), education (Marinoni et al., 2020), work (Kramer et al., 2020), mobility (Abu-Rayash et al., 2020), and mental health (MH; Bourmistrova et al., 2022;Samji et al., 2022). Isolation from loved ones (both voluntary for fear of their health and due to quarantine), rising prices and/or scarcity of consumer goods, paralysis of health care systems, economic uncertainty, including layoffs and bankruptcy, remote work/education without physical contact with other people, and many other factors contributed to reduced quality of life and greater levels of depressive symptoms and anxiety (Brenner et al., 2020;Rumas et al., 2021). The reality created by the pandemic made it necessary for many people to reorganize their day-to-day lives (Cancello et al., 2020). It should be noted that the COVID-19 situation was a new one for the vast majority of people-recent pandemics did not reach such a scale and did not last as long. In addition to its indirect impact (changes and limitations resulting from the pandemic), COVID-19 also directly influenced the functioning of people who became infected. In addition to long COVID (Crook et al., 2021), serious health problems (e.g. ECMO treatment or other systemic complications, see: Gribensk et al., 2022;Shanbehzadeh et al., 2021) could lead to the emergence and worsening of mental health difficulties.
There is no doubt that the pandemic was a major stressor that, to some extent, affected the vast majority of people in the world. According to the assumptions of the theoretical models used in research on stress, strategies for coping with stress are one of the factors that regulates the consequences of a given stressful situation. Lazarus and Folkman (1984) described coping as different types of deliberate efforts (cognitive, emotional, and behavioural) aimed at resolving a given stressful situation through either addressing the problem itself or the resultant emotions (including avoidance). In a situation that is assessed as stressful (cognitive appraisal process), actions are taken to restore the previous balance. An individual may use a variety of coping strategies depending on, among other things, their availability or subjective assessment of their effectiveness in a given situation or based on previous experience (Lazarus et al., 1984). One strategy used in stressful situations is religious coping (RC), which involves the use of various types of religious practices. Religious coping can, for example, help one to reinterpret a stressor (e.g. "it is God's will"), or provide emotional support, both from God and from religious communities. RC can be divided into positive religious coping (pRC) and negative religious coping (nRC). The former refers primarily to a sense of connectedness with a transcendent force and a secure relationship with a caring God, while the latter refers to feeling abandoned or punished by God as well as interpersonal religious discontent (Pargament et al., 2011).
The coping process ends when a given situation no longer causes stress (which does not necessarily mean that the stressor has been eliminated), which is largely dependent on subjective appraisal; a given stressor may also be successfully ignored, so that it no longer functions as a source of stress (Lazarus et al., 1984). The effects 1 3 of the coping process can be operationalized in many ways, but most often the effectiveness of a given strategy is determined by its impact on MH. Commonly used indicators include, for example, quality of life, level of anxiety, or the severity of depressive symptoms (De Ridder et al., 2008). The issue of the negative impact of the pandemic on mental health has been the subject of much analysis during the pandemic. Numerous studies have indicated, inter alia, high prevalences of depression, anxiety (Salari et al., 2020), PTSD (Cénat et al., 2021), sleep problems (Jahrami et al., 2021), and other mental health issues. However, in addition to negative mental health indicators, many studies have also focused on positive aspects, such as life satisfaction and flourishing. An important issue here is the fact that a high intensity of negative MH indicators (such as anxiety or depression) does not necessarily mean a low level of positive indicators, such as life satisfaction or happiness (see: Seligman, 2008); therefore, these two issues should be analysed separately, taking into account the broadest possible repertoire of variables that may be affected by stressful situations. These studies sought to identify the factors responsible for high levels of negative indicators, as well as protective factors that could buffer the negative effects of the pandemic. Therefore, we decided to conduct a systematic review of the literature on the relationship between religious coping and mental health indicators, in particular, focusing on both negative determinants of mental health, such as stress, depression, and anxiety, as well as positive determinants, such as post-traumatic growth, well-being (WB), and quality of life (QoL). Due to, inter alia, the fact that studies were conducted all over the world, the observed relationships between RC and MH indicators differed between studies. Therefore, we attempted to identify factors that may play significant roles in determining the strength of these relationships by performing a meta-regression of relevant studies.

Search Strategy
This review was pre-registered at OSF (Open Science Framework) below the link: osf.io/54ygr (https:// doi. org/ 10. 17605/ OSF. IO/ GMNFV) and after the registration DP and KWP independently conducted systematic literature searches using the PRISMA protocol (Moher et al., 2009). The reviewers searched PubMed, Science Direct, the Cochrane Library, Google Scholar, and the Database of Abstracts of Reviews of Effects using the keywords: "religious coping OR religion coping OR spiritual coping AND COVID". In the review, it was decided to also include articles from the "grey zone" (Google Scholar) only when both authors, having carefully read the text, had no doubts about the methodological soundness and quality of the article (see Appendix 1). Each article was independently assessed by both authors.
The search was limited to the English language, but no limitation was placed on the ethnicity of the human participants. The reference lists of relevant articles were also reviewed for other relevant articles.

Selection Criteria
Based on the presence of the searched-for or synonymous terms, 394 articles were selected for further analysis. The process for selecting articles is shown in Fig. 1. Ultimately, the following criteria were adopted for the systematic review: (1) the study must refer to RC; (2) the study must describe a relationship between RC and mental health; (3) the methods used in the study must have a described reliability; (4) it is clearly stated in the title or abstract that the research relates to the pandemic (or resultant difficulties, such as lockdown or distance learning). In our work, we focused on the mental health indicators operationalized as (a) presence/intensity of psychopathological symptoms such as PTSD, anxiety or depressive symptoms; (b) the level of functioning of a given person in various areas of life-defined as quality of life, and (c) positive indicators of adaptation, such as well-being, satisfaction with life or post-traumatic growth. Pre-prints, conference reports, and dissertations were not included in the review. We assumed that due to methodological issues, in the case of cross-sectional and longitudinal studies, the tools used to evaluate RC should have proven reliability. The review included only articles in which the instructions in the tool related to coping with stressstudies on, for example, the level of religiosity or the frequency of religious practices were excluded.
Furthermore, for the meta-analyses and meta-regressions, additional criteria were adopted: the analysis must have examined at least 3 studies that used the same method of assessing RC and MH and the articles must give the value of the correlation coefficient.
Two researchers (DP and KWP) independently reviewed 394 texts, identifying those that should be included in the analyses based on these criteria. In cases of

Included
Reports sought for retrieval (n = 149) Reports not retrieved (n = 0) Fig. 1 Flow diagram of studies identified, excluded and included in the systematic review disagreement, consensus was reached after discussion between the 2 reviewers. The reference lists were reviewed to identify other studies related to the topic.

Data Extraction
DP and KWP independently extracted the data for each study. The variables of interest for the systematic review were: year of publication, country of origin, sample size, basic sociodemographic data (sex, age, population), methods of assessing RC and MH, and the main findings of the study. For the meta-analysis, the strengths of the relationships between the two variables were extracted. Only the correlation coefficients were taken into account-when the relationships were reported using the beta value, the results were not included in the meta-analysis.
As potential moderators, we extracted: the percentage of women in the sample, the mean age in the study group, the proportion of people professing a given religion, the percentage of people declaring themselves to be believers, the groups examined (e.g. the general population or medics), the percentage of people who were infected with COVID-19 at the time of the study, the percentage of people who had been infected with COVID-19, the percentage of people in quarantine, the percentage of people who reported somatic symptoms (e.g. fever), the percentage of people with higher education, the percentage of people in a stable relationship (married), the percentage of people living in rural areas, the percentage of people living alone, and the percentage of people with chronic diseases. Due to differences in the introduction and nature of restrictions in different countries (e.g. local lockdowns), we decided to not examine as potential moderators the time when the research was conducted and the nature of the restrictions in each country.
The method for recording the data was agreed upon before coding the results. Inter-rater reliability was satisfactory; in cases of disagreement, consensus was reached after a discussion between the two reviewers.

Quality Assessment
To evaluate the quality of the selected studies, we used an adapted version of the Newcastle-Ottawa cohort scale for cross-sectional studies (Modesti et al., 2016), which takes into consideration the selection of samples, comparability of subgroups, and exposure. Using this scale, we scored each study independently. Inter-rater compatibility was satisfactory, and scoring differences were reconciled through discussion (Appendix 1).

Statistical Analysis
Meta-analysis and meta-regression were conducted in the R Studio platform (R Studio, Boston, MA, USA) using the R software environment; the "metacor" library was used to calculate correlations (Schwarzer, 2007). The meta-analysis used the inverse variance method with restricted maximum-likelihood estimator for tau 2 , Q-Profile method for confidence interval of tau 2 and tau. Fisher's Z transformation of correlations was used. Overall rates and 95% confidence intervals were calculated using a random effect model. Funnel plots were used to determine the possibility of publication bias (Appendixes 2). The heterogeneity of effects across studies was estimated using the Q test and quantified using the I 2 statistic. Meta-analyses were performed with random effects models. Any p values less than 0.05 were considered significant.

Results of the Systematic Review
The search using the methodology described above yielded 59 studies that fulfilled all selection criteria for the systematic review. Altogether, 68,824 participants participated in these studies. The mean age of the participants ranged from 20.8 (Masha'al et al., 2022) to 76.3 (Willey et al., 2022) years. For details on the number of surveys and the number of respondents from each country, see Figs. 2 and 3, respectively, and Appendix 3.
For the sake of the readability of the results, we decided to analyse each aspect of mental health separately. It should be noted that a large part of the research focused on several MH indicators, so the numbers of studies and participants do not add up to the numbers given previously in the text. In the following subsections, selected indicators of MH are analysed in detail.

Fig. 2 Choropleth map of numbers of studies by country
Due to the large amount of collected data, we decided to divide the article into two parts. In the first, positive MH indicators were analysed: quality of life, wellbeing, satisfaction with life (SWL), levels of happiness, and post-traumatic growth (PTG).
The second part of the review focuses on negative MH indicators: severity of depressive symptoms, anxiety, stress levels, symptoms of peri-and post-traumatic stress, and general negative MH indicators.

Quality of Life
The systematic literature review identified 5 studies in which 3363 participants took part. The research was conducted from April 2020 to March 2021. A large variety of tools were used to assess RC, making it impossible to carry out a meta-analysis of the collected results. QoL was examined in over half of the studies using the WHO Quality of Life scale (WHOQOL-BREF). The exact results are shown in Table 1.

Well-being
The systematic literature review identified 10 studies addressing the relationship between RC and WB, with a total of 17,469 participants. The research was conducted from February 2020 to December 2021. The Brief-COPE was used to evaluate RC in half of the studies (n = 5), while WB was assessed using various tools, including the PERMA Profiler (n = 4) and Flourishing Scale (n = 3). Most studies (except two) showed a relationship between RC and WB. The exact results are shown in Table 2.  *In studies that did not specify the exact dates on which data were collected (months only), the beginning (1) and end (30) of the month were used as the starting and ending points. nRC-negative religious coping; pRC-positive religious coping Journal of Religion and Health (2023) 62:510-543 *In studies that did not specify the exact dates on which data were collected (months only), the beginning (1) and end (30) of the month were used as the starting and ending points

Meta-analysis
The analysis of the studies included in the review only allowed for a meta-analysis of the relationship between flourishing and RC assessed with the Brief-RCOPE. The meta-analysis of the relationship between level of flourishing and nRC (Brief-RCOPE) was conducted on 3 studies identified by the literature search as meeting the inclusion criteria. Their results were pooled to give a correlation of − 0.25 [− 0.34; − 0.15] suggesting a statistically significant negative relationship between these variables (Z = − 4.91; p < 0.001). (Fig. 4). Statistically significant heterogeneity was observed between studies (Q = 9.62; p < 0.05). The estimated amount of total heterogeneity was Tau 2 = 0.006 and I 2 = 79%.
Due to the heterogeneity of the results, potential moderators were also analysed in more detail: the percentage of women, relationship status (percentage of married people), and level of education (percentage of people with higher education). Unfortunately, due to deficiencies in the reported data, it was not possible to include more moderators. The tests for moderators showed that the percentage of women (QM (1) = 3.58; p = 0.058) and the percentage of married people (QM (1) = 3.21; p = 0.073) were at the statistical trend level. However, after taking into account the above moderators, residual heterogeneity decreased to an insignificant level: Q = 1.62 for percentage of women and Q = 1.73 for relationship status. In turn, higher education (QM (1) = 0.01; non-significant) was a statistically insignificant moderator.
The next step was a meta-analysis of the relationship between pRC and levels of flourishing. The meta-analysis conducted for the relationship between flourishing and pRC (Brief-RCOPE) also included 3 studies. Studies identified in the literature search as meeting the inclusion criteria were pooled to give an overall correlation of 0.35 [0.30; 0.40] suggesting a statistically significant positive relationship (Z = − 3.18; p < 0.01). (Fig. 5). Statistically significant heterogeneity was not observed between studies (Q = 3.27; p > 0.05). The estimated amount of total heterogeneity was Tau 2 = 0.0007 and I 2 = 38.8%.

Satisfaction with Life and Happiness
In the next step, the relationships of RC with SWL and happiness were analysed. The literature review identified 2 studies related to the relationship between RC and Fig. 4 Negative religious coping and flourishing: forest plot SWL and 1 related to the relationship between RC and happiness. The total number of participants in these 3 studies was 1789, and the research was conducted from April 2020 to March 2021. The Brief-RCOPE was used to evaluate RC in 2 of 3 studies, while happiness and SWL were analysed with different methods, which unfortunately made it impossible to carry out a meta-analysis of the collected data. All studies found relationships between RC and SWL and happiness. The exact results are shown in Table 3.

Post-traumatic Growth
The relationship between PTG and RC was analysed in 5 studies, in which 22,165 participants took part. The research was conducted from April 2020 to May 2021. The Brief-COPE was used to evaluate RC in all studies, while PTG was assessed using the Post-Traumatic Growth Inventory and Post-Traumatic Growth Index. Each of the studies showed a positive relationship between RC and PTG. The exact results are shown in Table 4. Unfortunately, due to the lack of data on the correlation coefficients, it was not possible to perform a meta-analysis.

Discussion
The first part of this systemic review focused on research on the correlation between religious coping (RC) and positive mental health (MH) indicators. The analysis of the literature allowed us to identify the following studies that met the inclusion criteria: 5 studies on the relationship between RC and quality of life (QoL), 10 studies on the relationship between RC and well-being (WB), 2 studies related to the relationship between RC and satisfaction with life, 1 related to the relationship between RC and happiness, and 5 studies on the relationship between RC and post-traumatic growth (PTG). It was only possible to conduct a meta-analysis on the relationship between negative religious coping (nRC) and flourishing, which found a value of r = − 0.25 [− 0.34; − 0.15]; however, the results were heterogeneous. The percentage of women in the studies and the percentage of married people used in the analysis as moderators were on the verge of statistical tendency. On the other hand, for  The synthesis of data from studies covering the relationship between RC and QoL indicated that the vast majority of studies conducted in this area were crosssectional. Only the study by Schamblaw et al. (2021) had a follow-up after 1 month. Data from cross-sectional studies were collected from different populations (including students, people with chronic illnesses, and GPs) in different countries and with the use of different questionnaires; in most cases, these were tools assessing pRC/ nRC, and two used the Brief-COPE. It should be noted that in most of the crosssectional studies, a relationship was observed between QoL and RC, while the data from the longitudinal study did not find a relationship between RC and QoL. Data from a longitudinal study by Danhauer et al. (2009) on women with breast cancer indicate that coping strategies predict QoL to a limited extent, while QoL may predict coping strategies equally or to a greater extent. In turn, the studies by Pargament et al. (2004) and Trevino et al. (2010) indicate a weak association between QoL and pRC/nRC. It should be noted that the coping process is dynamic, and both the strategies used and the extent to which they are used may change over time depending on the effects achieved. This is important not only for the analysis of possible relationships or the impact of a given strategy, but also due to the content of the items used in the questionnaires to assess pRC/nRC. The content of items in these questionnaires may measure the effects of coping with a given stressor to a greater extent than the strategy used; for example "Felt punished by God for my lack of devotion" or "Wondered what I did for God to punish me" (Brief-RCOPE) indicate a failure to cope with a given stressor-that is, they seem to reflect the transaction effect (Lazarus & Folkman, 1984) to a greater extent than the methods used to reduce the level of stress or regulate emotions. At the same time, it should be noted that the analysis did indicate such a relationship, but the data from the longitudinal study may indicate that the choice of this strategy was more dependent on QoL; however, this requires further additional investigation due to limited data.
Studies assessing the relationship between RC and WB were also conducted around the world in different populations throughout the entire period of the pandemic. Most of these studies were cross-sectional, except for one longitudinal study (Davis et al., 2021). The cross-sectional studies mostly indicated the presence of this relationship, while longitudinal studies indicated no such relationship. As in the case of QoL, we should wonder to what extent the degree of RC use may be determined by the current state of the participant. The collected data also allowed for a meta-analysis of the results of 3 studies on the relationship between flourishing and p/nRC. However, these results should be approached with caution due to the limited data included in the analysis. In the case of nRC, statistically significant heterogeneity was observed, which could not be explained with the use of moderators: the percentage of women and married people reached the level of statistical trend, while the level of education was statistically insignificant. It should be noted that a very limited number of variables were used as moderators due to uncontrolled data or deficiencies in reporting data describing the studied group, which could be further analysed as a potential source of variability in the research results. However, all three studies showed a negative relationship. A meta-analysis of studies assessing the relationship between pRC and flourishing showed a weak positive relationship, with the results being homogeneous. In conclusion, the collected data indicate the existence of a relationship between WB and RC, and the nature of this relationship requires more analysis-longitudinal data do not indicate that RC determines changes in WB. Meta-analyses confirm a positive relationship between pRC and flourishing irrespective of sociodemographic factors, and the strength of the correlation between nRC and flourishing seems to depend on factors that could not be determined due to gaps in the reported data.
Next, satisfaction with life (SWL) and happiness were analysed. Cross-sectional studies covering this topic were conducted in various countries, most of them (n = 2) on the student population. The synthesis of the results allows us to draw conclusions about the relationship between these constructs: both neutral RC and pRC were positively associated with SWL, while nRC was negatively associated with levels of happiness and SWL. It should be noted that different assessment tools were used to assess both SWL and RC. The lack of longitudinal studies prevents conclusions being drawn about the direction of this relationship and its possible impact.
The last of the positive MH indicators included in this review was post-traumatic growth (PTG). The review of the studies allowed us to identify 5 cross-sectional studies conducted in different countries on very diverse populations. All studies assessed RC using the Brief-COPE, but meta-analysis could not be performed due to deficiencies in the reported correlation coefficients. All studies clearly indicated a positive relationship between PTG and RC. Unfortunately, it was not possible to identify longitudinal studies that would allow the assessment of the impact or direction of this relationship. Earlier longitudinal studies performed in clinical trials indicated no relationship (Scrignaro et al., 2011) or a negative relationship between RC and PTG (Rzeszutek et al., 2017). Studies conducted on a population of natural disaster survivors indicate the beneficial effects of pRC (Arkin, 2022;Chan et al., 2013). It would be valuable to further explore this relationship: whether it depends on an objective/subjective assessment of a given event, the nature of its causes, or whether such differences may result from the methodology used. It should be noted that in the case of the first two examples, the Brief-COPE was used, in which the RC items are formulated in a neutral way, and in the case of hurricane victims, RC was assessed in terms of p/nRC.
To sum up, the coping process usually involves a whole range of different types of strategies based on, inter alia, what resources are available (e.g. social support), how they can be used, as well as the expected effectiveness of their use. For almost all people, the pandemic was a novel situation where previous strategies did not necessarily have the expected results, which could have been an additional source of stress. The vast majority of studies included in the review suggested either no or very weak relationships between positive MH indicators and RC. The exception seems to be PTG, for which cross-sectional studies showed a beneficial effect. Single longitudinal studies indicated no influence of RC on the analysed dependent variables (QoL, WB), but these results should certainly be replicated. It is also worth noting that most analyses were based on the variable-centred approach, focusing on the percentage of variation in dependent variables that was explained by coping strategies; but it is worth remembering that they occur in particular combinations. It would also be worth considering the analysis of subgroups that differ in their degree of use of specific strategies with the use of methods allowing for the identification of such profiles (k-means clustering; LPA, LCA).

Study Limitations
The cross-sectional nature of most of the analysed studies prevents inferences about the direction of the relationship between the variables included in the review. The use of this strategy results primarily from the highly burdensome nature of the stressor, which may be largely related to the observation of artificial positive correlations between RC and worse mental health indicators in cross-sectional studies. These types of results may also lead to overlooking the positive aspects that RC may have on mental health in the long term.
The most important limitation of this review is the small number of studies (n = 3) included in the calculations; the results of the meta-analyses should therefore be interpreted with caution. Another serious limitation is the heterogeneity obtained in one of the meta-analyses, which could not be explained based on the available moderators. Another factor influencing the quality of the obtained results is directly related to this: unfortunately, in many cases, the basic data describing the sample were not reported, so these variables could not be used as moderators in meta-regression. Correlation matrices between the analysed variables were not reported in many studies, which limited the possibilities for further data analysis. Authors should keep in mind that sample description and data reporting should allow for both replication and further analyses: the results obtained in a given study may depend on many factors related to, for example, the selection of the sample or the time and place they were carried out; only a further collective analysis of the results of many studies will allow us to draw more precise conclusions.

Conclusions
The collected data suggest that there is a relationship between RC and positive MH indicators. Data from longitudinal studies in turn suggest that RC is not associated with levels of positive MH indicators. Meta-analyses indicate that the relationship between pRC and flourishing is less sensitive to differences in sample structure, and the association between nRC and flourishing may be partially dependent on some sociodemographic indicators; however, this requires further analysis. Due to the shortcomings in reporting both analyses and data on the sample structure, the results should be interpreted with caution. (2023)