Psychosocial predictors of distress in East and West Germans during the COVID-19 pandemic

Abstract

Background

Psychological (meaning in life, science attitude, internal locus of control, religiosity), and social factors (social support, cohesion) can counteract stressor-related distress. We investigated these factors’ links with peri-pandemic distress (depression, anxiety, intrusions) and whether they weakened the impact of being affected by the COVID-19 pandemic. We compared prior East and West Germans on predictors and distress to investigate if their different backgrounds created lasting differences.

Methods

A population-representative German sample aged 45 to 70 (N = 380) in terms of age, sex, and school education completed online questionnaires in May–July 2020 and June–July 2021. We examined the predictive relations with correlation, forward inclusion regression, and moderation analyses.

Results

Social support predicted lower distress, also prospectively. Meaning in life predicted lower distress cross-sectionally. Religiosity predicted greater distress. Life meaning and social support partly weakened the link between being affected by the pandemic and distress, religiosity and science attitude strengthened this link. The only significant East/West difference was in religiosity, which was higher in the West.

Conclusion

Social resources appeared particularly important in adjusting to the pandemic. The identified predictors may inform interventions. East and West Germans’ similarity might indicate that their post-war separation did not create lasting differences in the investigated factors.

Keywords:

• COVID-19 pandemic
• distress
• social support
• meaning in life
• Germany (East/ West)

Introduction

The COVID-19 pandemic created stressors that negatively affected daily life for many (Horesh & Brown, 2020). Social distancing, quarantining, lockdowns, and pandemic-related restrictions changed life in Germany and around the world (Meintrup et al., 2022). Consequently, the prevalence of self-reported psychological distress such as depression and anxiety rose (Bäuerle et al., 2020), resulting in high point prevalence estimates (for an international meta-analysis, see Salari et al., 2020), and an increase in clinical diagnoses of anxiety disorders (in Germany; Jacob et al., 2021). Also, prevalence rates of posttraumatic stress disorder (PTSD) with hallmark symptoms such as intrusions were higher than usual (e.g. in March 2020; Shevlin et al., 2020). Identifying factors that counteract distress during the pandemic may inform intervention design and implementation and help identify at-risk groups. Therefore, the aim of the current study was to investigate the roles of several candidate predictors of peri-pandemic distress.

Factors which may counteract peri-pandemic psychological distress

Psychological and social factors may counteract distress such as depression, anxiety, and PTSD (e.g. in Blackburn & Owens, 2015; Onyedire et al., 2017) and stave off aversive effects of stressor exposure, as demonstrated by a moderation of the association between stressor severity (a distress predictor, for a meta-analysis on PTSD see Brewin et al., 2000) and distress (e.g. in Howell et al., 2021; Krause, 2007). Moreover, distress predictors may differ between populations (Koltko-Rivera, 2004). Therefore, we aimed to investigate two sub-populations of Germany, prior East and West Germany, and predominantly selected predictors which may differ between these groups. In particular, the current study examines the relative importance of meaning in life, an internal locus of control, a positive attitude towards science, religiosity, and a supportive social environment as candidate predictors of lower peri-pandemic distress. These factors have been shown to be related with less distress in prior stressor contexts but have not (jointly) been investigated during the COVID-19 pandemic. Modeling these predictors together offers insights above and beyond testing their links with distress individually because it investigates the relative importance of these factors. Except for life meaning, we predicted that these factors differ between East and West.

Previous COVID-19 pandemic studies in Germany (Beutel et al., 2021; Liu et al., 2021; Meyer et al., 2021) and other countries such as the US and Canada (Prout et al., 2020; Taylor et al., 2020) found that demographic factors such as older age, male sex, and fewer issues with socioeconomic and housing factors correlated with lower distress including depression and anxiety. Also psychological factors have been found to be correlated with lower distress, including believing less in the dangers of COVID-19 (Taylor et al., 2020), fearing COVID-19 infections less (Meyer et al., 2021), feeling less lonely (Liu et al., 2021), as well as lower somatization, and more adaptive coping (Prout et al., 2020).

This study adds to the literature on stressor-related, and particularly pandemic-related, distress in several ways: (1) by investigating the relative importance of several candidate distress predictors, the current study may inform intervention choice and design; (2) by comparing the predictors between two cultural sub-populations, East and West Germans, it may inform about cultural differences in distress predictors and thereby inform culturally sensitive intervention choice and design; (3) by following up prior findings regarding life meaning by exploring the role of life meaning subcomponents, this study adds to the literature on life meaning and may ultimately inform meaning-related interventions; and (4) by controlling for baseline distress in prospective models, stronger inferences about links between predictors and outcomes are warranted than in cross-sectional studies.

Meaning in life

Life meaning, including the subcomponents of perceiving (1) comprehension/coherence of one’s life, (2) purpose and associated goals and direction, and (3) significance of one’s existence (e.g. George & Park, 2016; Martela & Steger, 2016), may counteract distress. Comprehension, which has been posed to originate from clear meaning frameworks including goals and beliefs (George & Park, 2016) may lower distress by reducing distress-fostering uncertainty (e.g. Ostafin et al., 2022). The uncertainty model is supported by recent research showing indirect effects of the relation between life meaning and depression and anxiety through fear of uncertainty (Ostafin et al., 2022). Moreover, targeting purpose-related constructs (e.g. goals, motivation) has been shown to be beneficial in treating depression (e.g. Lejuez et al., 2001) and may counteract peri-pandemic distress, for example by preventing behavioral disengagement. As the pandemic introduced great uncertainty (e.g. Koffman et al., 2020) and may have fostered behavioral disengagement (e.g. through restrictions including social distancing and curfew), life meaning may counteract peri-pandemic distress.

Cross-sectional studies have found that life meaning is commonly associated with lower distress including anxiety and depression (for a meta-analysis, see Winger et al., 2016), as well as posttraumatic stress disorder (PTSD) symptoms (Blackburn & Owens, 2015; Fischer et al., 2020) in stressor-exposed individuals. The inverse relation between life meaning and distress has also been found during the COVID-19 pandemic (Schnell & Krampe, 2020). These cross-sectional findings have been extended with research showing that life meaning prospectively predicted lower distress in stressor contexts such as natural disasters (a flood) and the COVID-19 pandemic (Ostafin & Proulx, 2020, Studies 2 and 3; Seidel et al., 2023).

In addition to bivariate relations between life meaning and distress, life meaning has also demonstrated a moderator effect, weakening the link between stressors and distress. For example, one study showed a moderation effect between trauma exposure and depression in elderly individuals (Krause, 2007). However, a veteran study did not find a moderation effect between combat exposure and PTSD (Blackburn & Owens, 2015). These inconsistent results might be due to sample size differences (i.e. smaller in veteran study), given that detecting moderation effects requires high statistical power (Murphy & Russell, 2017).

Internal locus of control

Additionally, internal locus of control (LOC), the generalized belief of having control over one’s life (Rotter, 1966), may counteract distress. Reasons for this include that LOC may enable individuals to assume responsibility for their fate, which may foster adaptive behaviors (e.g. illness prevention, and efforts to improve functioning; Strickland, 1978). During the pandemic, LOC may counteract distress, e.g. when it enables individuals to care for their mental and physical health.

Empirical evidence supports the idea that LOC is inversely related with distress severity, including depression (see meta-analysis by Presson & Benassi, 1996). Moreover, LOC may moderate the aversive effects of stressors (for a review, see Lefcourt, 2013).

Positive attitude towards science

Another factor that may lower peri-pandemic distress is a positive attitude towards science including trusting that science can solve problems and foster development (Schnell, 2015). Because the COVID-19 pandemic posed a scientific challenge regarding the treatment and prevention of the new virus, trusting that the scientific community will provide solutions and guidance may foster optimism regarding issues such as the manageability of the pandemic, thereby lowering distress. Another pandemic study supported the role of optimism, a potential working mechanism for science attitude on distress during the pandemic, in distress. Here, optimism correlated with lower depression, anxiety, and distress, and moderated the link between fear of COVID-19 and distress and depression in Dutch-speaking adults (Vos et al., 2021).

Although we did not find any research on the relation between positive science attitudes and distress, related findings have shown that trust in institutions such as the government and the health care system coincided with greater well-being and lower distress in European adults during the pandemic (S. Lee, 2022) and predicted a weaker distress (e.g. depression, anxiety) increase after the COVID-19 outbreak in Germany (Bäuerle et al., 2020).

Religiosity

Also religiosity may reduce pandemic-related distress. Reasons for this include that religiosity may foster life meaning (e.g. Garssen et al., 2021; Schnell, 2009), thereby engaging the meaning-related mechanisms listed above. Further, religiosity offers adaptive coping options such as redefining the stressor as benevolent and potentially beneficial or seeking support from their religious community and/or God (Ano & Vasconcelles, 2005; Pargament et al., 2000). For example, trusting that God will alleviate COVID-19 infections might lower the distress linked with the illness. Previous research has shown that adaptive religious coping correlated with lower distress (for a meta-analysis, see Ano & Vasconcelles, 2005) and moderated the relation between stressors and distress (Park et al., 1990).

Empirically supporting its beneficial role in distress, previous research demonstrated negative associations between religiosity and distress (e.g. depressive and anxious symptoms, for meta-analyses, see Hackney & Sanders, 2003; Smith et al., 2003), also in longitudinal research predominantly from the USA (for a meta-analysis, see Garssen et al., 2021). However, the link between religiosity and distress was insignificant in German-speaking countries (for a meta-analysis, see Hodapp & Zwingmann, 2019), potentially due to cultural differences or the inclusion of more studies on maladaptive religious coping. Additionally, religiosity has been shown to weaken the association between life stress and depression (for a meta-analysis, see Smith et al., 2003), and poor physical health and depression (Pirutinsky et al., 2011).

Supportive social environment

Additionally, a supportive social environment may lower pandemic-related distress. Specifically, social support, the provision of resources of a social network to aid an individual’s coping with stress (Cohen, 2004), and neighbourhood social cohesion, the sense of attachment to a neighbourhood and positive interactions with neighbours (Wilkinson, 2007) may counteract distress. These factors may lower distress by fostering adaptive, distress-reducing (Kato, 2015) and stressor-effect buffering (Jesus et al., 2016) coping behaviors (Cohen, 2004). Additionally, social factors may foster life meaning (Cohen, 2004; Schnell, 2009). During the pandemic, these social factors may have lowered distress when resulting in practical and emotional support with difficult experiences, e.g. illness and quarantine.

In previous stressor research, these social factors were associated with lower distress. Social support was related with lower PTSD (for meta-analyses, see Brewin et al., 2000; Ozer et al., 2003), depression, and anxiety symptoms (for a meta-analysis, see Guilaran et al., 2018), including during the COVID-19 pandemic (Ahrens et al., 2021). Neighbourhood cohesion has also been shown to be inversely associated with depression (Erdem et al., 2016), also prospectively (for a meta-analysis, see Baranyi et al., 2020).

Moreover, social support has been shown to weaken the link between acculturation stress with psychological distress (J.-S. Lee et al., 2004) and between life adversities and PTSD (Howell et al., 2021). Similarly, social cohesion weakened the association between socioeconomic stressors and distress (Erdem et al., 2016).

Two sides of Germany: the prior East and the prior West

Cultural and societal backgrounds may affect meaning frameworks including beliefs, expectations, and goals and may hence influence the interpretation of and coping with stressors (Koltko-Rivera, 2004). Therefore, populations may differently endorse some of the above-mentioned factors which might, in turn, relate to potential distress differences. Since Germany had been separated into East and West Germany between the end of WWII and the reunification in 1989/1990, one may speak of two German sub-populations. Hence, it stands to question whether cultural differences remain which may affect coping with the pandemic.

First, prior East Germany has undergone significant secularization and distancing from religion. Suggested reasons for this include top-down discouragement from the socialist government (e.g. Peperkamp & Rajtar, 2010). While also prior West Germany has become increasingly secular, differences remain. For example, 69% of West Germans reported that they belong to a church compared to only 27% of East Germans, and similar differences have been shown regarding religious practices (Meulemann, 2009).

Moreover, a positive attitude towards science and the inclusion of scientific knowledge and viewpoints in the education system were parts of the Soviet agenda in Eastern Germany (Schmidt-Lux, 2010). However, published research examining East/West differences in attitudes toward science appears to be lacking. One study found that interest in science and research was comparable between East and West (Wissenschaft im Dialog, 2021). However, this study also included younger participants (above age 13), who did not experience the separated Germany.

Additionally, prior research found that East Germans reported lower LOC than West Germans (Friehe et al., 2015), potentially because external forces such as strong ruling institutions more heavily influenced East Germans’ lives growing up. Early environmental factors have been shown to shape (LOC) beliefs (e.g. Pedron et al., 2021).

The socialist culture of East Germany valued and propagated social cohesion and community-focused behaviors, for example through educational agendas and youth organizations. Related, East German post-reunification nostalgia commonly entails social aspects (e.g. Neller, 2006) such as a sense of togetherness and neighbourhood connections across social groups. In contrast, West German culture has been described as more individualistic (e.g. Kirkcaldy et al., 1999). Hence, East Germans may report greater social support and social cohesion than West Germans. One study supported this idea (Kirkcaldy et al., 1999), though another study indicated the opposite (Dragolov et al., 2016).

Hypotheses

Based on these conceptualizations, we hypothesized that during the COVID-19 pandemic, life meaning, LOC, a positive science attitude, religiosity, and social support and cohesion would relate with lower distress (i.e. depression, anxiety, intrusion symptoms) (H1a) cross-sectionally and (H1b) prospectively. Moreover, we hypothesized that these factors would moderate the associations between being negatively affected by the pandemic and distress, both cross-sectionally (H2a) and prospectively (H2b). We also hypothesized that East Germans would report lower religiosity and LOC, a more positive science attitude, and a stronger social environment than West Germans (H3). Last, we planned to explore whether distress predictors mediated the link between origin (East/West) and distress. These hypotheses are included in the preregistration no. 42656 at aspredicted.org.

Methods

Participants and assessment timepoints

From May to July 2020, N = 630 (n_{(East G.)} = 286; n_{(West G.)} = 344) individuals completed the 2020 assessment. This sample is considered population representative as it was stratified by age, gender, and school education according to census data (statista) per subsample. Participants’ average age was 57 years (range: 45–70) and 51% of participants were female. Regarding school education, roughly equal parts completed the three different versions of German secondary education (32.2% ‘Hauptschule’/equivalent/did not complete school, 34.8% ‘Realschule’/equivalent, 33.0% A-levels/equivalent). In June and July 2021, N = 380 participants (n_{(East G.)} = 153; n_{(West G.)} = 227) completed the 2021 assessment. N = 250 dropped out. With this sample size, the power to detect small prospective effects (f² = .02) is .87. Both assessments took place during periods with lower infections rates, fewer pandemic restrictions compared to the preceding months, and less overall pandemic-related burden (Meintrup et al., 2022).

Materials

2020 Assessment

Distress.

Participants reported depressive symptoms on the Patient Health Questionnaire—9 (Kroenke et al., 2001; current sample alpha = .91). They indicated symptom (e.g. ‘Little interest or pleasure in doing things’) frequency over the last two weeks from ‘Not at all’ (0) to ‘Nearly every day’ (3) on nine items. Sum scores of at least 10 are assumed to indicate clinically relevant symptoms (Kroenke et al., 2001).

Anxiety was assessed with the General Anxiety Disorder Questionnaire (Spitzer et al., 2006; current sample alpha = .92). Participants reported the frequency of seven symptoms (e.g. ‘Feeling nervous, anxious, or on edge’) over the two weeks prior from ‘Not at all’ (0) to ‘Nearly every day’ (3). Scores of ten and above are thought to indicate clinically relevant anxiety (Spitzer et al., 2006).

Participants reported pandemic-related intrusions on five items from the PTSD Checklist 5 (Krüger-Gottschalk et al., 2017; Weathers et al., 2013; current sample alpha = .91) that were adapted to reflect peri-pandemic experiences. They indicated how much, e.g. ‘Repeated, disturbing, and unwanted memories of a specific experience related to the corona pandemic’ had bothered them over the last month from ‘Not at all’ (0) to ‘Extremely’ (4).

Predictors of distress.

Participants reported their meaning in life on a translated version of the Multidimensional Existential Meaning Scale (George & Park, 2017; current sample alpha = .92), which assesses the three meaning subcomponents with five items each. Participants indicated their agreement with, e.g. ‘My life makes sense’ (comprehension), ‘I have aims in my life that are worth striving for’ (purpose), ‘I am certain that my life is of importance’ (mattering) from ‘Very strongly disagree’ (1) to ‘Very strongly agree’ (7).

LOC was assessed with three items from a brief version of Levenson’s Locus of Control scale (Sapp & Harrod, 1993; current sample alpha = .71). Participants indicated their agreement with e.g. ‘My life is determined by my own actions’ from 1 (‘Don’t agree at all’) to 6 (‘Completely agree’).

Science Attitude was assessed with the four-item subscale of the Dimensions of Secularity inventory (Schnell, 2015; current sample alpha = .87), including items such as ‘Science offers solutions for all our problems’ that were rated from 0 (‘Don’t agree at all’) to 5 (‘Completely agree’).

Religiosity was assessed with the three-item subscale of the Sources of Meaning and Meaning in Life Questionnaire (Schnell, 2009; current sample alpha = .93). Response options to statements such as ‘Religion plays an important role in my life’ ranged from ‘Don’t agree at all’ (0) to ‘Completely agree’ (5).

Participants indicated their agreement (‘Don’t agree at all’ (1); ‘Completely agree’ (5)) with six social support statements (e.g. ‘I know a very close person whose help I can always count on’) on the Brief Perceived Social Support Questionnaire (Lin et al., 2019; current sample alpha = .90). Social cohesion was assessed with the subscales about neighbourhood attachment and neighbour interactions from the Neighbourhood Cohesion Instrument (Wilkinson, 2007; current sample alpha = .95) that were translated. Participants answered, e.g. ‘A feeling of fellowship runs deep between me and others in this community’ from ‘Completely agree’ (1) to ‘Don’t agree at all’ (5). Due to the inverted anchors, item scores were recoded.

Being Negatively Affected by the Pandemic was assessed with one item developed for this study, ‘To what extent did the coronavirus pandemic generally negatively affect your life’? rated from ‘Not at all’ (1) to ‘Very much’ (7). To complement this information, participants also indicated 1) to what extent the pandemic threatened their livelihood (e.g. job security, financial situation), how (2) controllable and (3) harmful they found the pandemic on the same scale, and (4) their average daily time spent consuming COVID-19-related content in 30-minute increments (‘0—30 min’ (1); ‘above 150 min’ (6).

2021 Assessment

In the 2021 assessment, the same measures of depression (Kroenke et al., 2001; current sample alpha = .91; n = 376), anxiety (Spitzer et al., 2006; current sample alpha = .92; n = 377), intrusions (Krüger-Gottschalk et al., 2017; Weathers et al., 2013; current sample alpha = .93; n = 376), life meaning (George & Park, 2017; current sample alpha = .94; n = 375), science attitude (Schnell, 2015; current sample alpha = .84; n = 377), social support (over last month; Lin et al., 2019; current sample alpha = .86; n = 378), and being affected by the pandemic were used as in 2020.

Procedure

The Ethical Committee Psychology of the University of Groningen approved this study. At both assessments, participants gave electronic informed consent and received a financial compensation (1€ and 1.25€, respectively). respondi, a German panel company, invited participants—individuals aged 45 to 70 currently living in the same part of Germany (i.e. prior East, prior West) they lived in their childhood. This age group was chosen because participants lived a significant portion of their lives in the separated Germany. Both online assessments through Qualtrics included further measures not relevant for this manuscript.

Statistical analyses

We utilized SPSS Statistics for Windows, Version 26 for all analyses. We obtained descriptive statistics for key variables and checked model assumptions by inspecting plots (Ernst & Albers, 2017). For all analyses, we conducted separate tests per outcome (depression, anxiety, intrusions). We tested Hypotheses 1a and b with correlations and linear regression analyses with forward selection (cut-off p-value = .05) to identify the relatively most important factors. To test the moderation hypotheses (H2a, H2b), we standardized the model variables. In longitudinal analyses (H1b, H2b), we controlled for the respective 2020 distress outcome. Since age was correlated with distress, we repeated the regression analyses with age as a covariate. This did not change the significant effects. Therefore, the results without covariates are reported. We tested mean differences between East and West with independent samples t-tests. Since East and West did not significantly differ in distress, we omitted the planned mediation analyses. We opted to reduce the family-wise error rate (1) by using a Bonferroni-Holmes correction (using alphas of .008, .01, .0125, .017, .025, and .05 to test the correlations in H1 between the predictors and each distress outcome) and (2) by limiting the number of tested regression coefficients through hierarchical forward inclusion of predictors.

Missing data and model assumptions

As the dropout between assessments was substantial (n = 250; reasons: n = 110 de-enrolled from the panel; n = 98 did not sign up again; n = 42 failed a quality control question (Please click ‘5’)), we did not impute missing cases (Jakobsen et al., 2017). In a logistic regression, study continuation was independent of 2020 distress, predictors, and gender (p-values ranged from .074 to .738). However, continuation was related with origin (Exp(B) = .58; p = .002; continued: East = 53%; West = 66%), school education (Exp(B) = 1.91; p < .001; M_(cont.) = 2.17; SD = 0.77; M_{(not cont.)} = 1.77; SD = 0.80), and age (Exp(B) = .967; p = .005; M_(cont.) = 56.36; SD = 6.96; M_{(not cont.)} = 58.04; SD = 8.08). No data were missing in 2020. In 2021, drop-out-unrelated missing data (< 2% per outcome) was independent of 2020 distress and predictors, and likely completely at random (Little’s MCAR test: Chi² = 2.586; df = 6; p = .859). Thus, this missingness is considered minimal and complete cases are analyzed (Jakobsen et al., 2017). We consider all data valid and did not detect response patterns. The final sample sizes ranged from N = 374 to N = 377 for longitudinal analyses and N = 629 to N = 630 for cross-sectional analyses with 2020 data. Regarding model assumptions, mixed results emerged. For depression and anxiety, we found no substantial deviations from the assumptions. For intrusions, plots indicated potential assumption violations. Hence, inferences should be drawn cautiously (Ernst & Albers, 2017).

Results

Descriptive statistics

For an overview of the descriptive statistics as well as the correlations between the variables, please see Table 1. Across assessments, participants indicated being moderately negatively affected by the pandemic (response range: 1–7). In 2020, they reported low to moderate livelihood threat (M = 2.65; SD = 1.87) and found the pandemic moderately controllable (M = 4.13; SD = 1.64), harmful (M = 4.43; SD = 1.83) and personally relevant (M = 4.10; SD = 1.63). They consumed 30—60 minutes of pandemic-related content daily (Mdn = 2; M = 2.04; SD = 1.30). Except for controllability (r = −0.15), these measures correlated positively with ‘being affected by the pandemic’ (rs from .30 to .49; all ps < .001). Clinically relevant levels of depression (2020: 15.6%; 2021: 14.9%) and anxiety (2020: 8.9%; 2021: 10.9%) exceeded prior German estimates based on the same measures (5.6% depression; Kocalevent et al., 2013; 5%–6% anxiety; Hinz et al., 2017; Löwe et al., 2008). In 2020, only 14.6% reported intrusions (i.e. not zero), while this rate increased to 41.0% in 2021. Depression and anxiety severity did not significantly differ between assessments, but intrusion severity increased (Table S1). While a positive science attitude and social support decreased between assessments, life meaning did not change and participants felt more affected by the pandemic over time (Table S1). While the levels of most predictors were comparable to previous studies, this sample indicated more positive science attitudes and lower social support compared to other, pre-pandemic German samples (Lin et al., 2019; Schnell, 2015). Similar to prior studies (e.g. Taylor et al., 2020), peri-pandemic distress was inversely associated with age (rs ranged from −0.09 for intrusions in 2020 to −0.21 for depression and anxiety in 2021; except for the smallest correlation (p = .027), p-values were ≤ .001). Unlike in some previous studies (e.g. Beutel et al., 2021), distress was not significantly related to gender (rs ranged from .02 for intrusions in 2021 to .09 for anxiety in 2021; ps ranged from .073 to .678 indicating insignificantly higher distress in women).

Table 1. Descriptive statistics of and correlations among key study variables.

Variables M (SD)	1	2	3	4	5	6	7	8	9	10	11	12	13	14	15	16
2020
1. Life meaning 69.84 (15.69)
2. LOC 14.34 (2.53)	.46**
3. Science attitude 11.67 (4.75)	.22**	.21**
4. Religiosity 3.62 (4.63)	.15**	−.01	−.12**
5. Social support 23.76 (5.64)	.49**	.38**	.24**	.02
6. Social cohesion 50.70 (12.23)	.38**	.24**	.20**	.08	.57**
7. Affected 3.17 (1.74)	−.06	−.06	.04	.11**	−.07	−.05
8. Depression 4.86 (5.32)	−.41**	−.25**	−.15**	.11**	−.38**	−.22**	.38**
9. Anxiety 3.83 (4.43)	−.36**	−.21**	−.15**	.12**	−.34**	−.20**	.38**	.84**
10. Intrusions 0.55 (1.88)	−.10**	−.06	.01	.17**	−.16**	−.04	.34**	.55**	.60**
2021
11. Life meaning 71.06 (16.96)	.77**	.41**	.27**	.13*	.47**	.31**	−.03	−.40**	−.33**	−.07
12. Science attitude 11.26 (4.66)	.25**	.23**	.59**	−.15*	.21**	.18**	−.09	−.11*	−.12*	.03	.24**
13. Social support 21.76 (5.84)	.42**	.36**	.13**	−.02	.68**	.50**	−.11*	−.29**	−.26**	−.11*	.44**	.23**
14. Affected 3.47 (1.74)	−.06	−.12*	.05	.12*	−.09	−.04	.60**	.33**	.32**	.31**	−.09	−.07	−.10*
15. Depression 4.48 (5.04)	−.34**	−.23**	−.06	.17**	−.34**	−.14**	.35**	.69**	.66**	.45**	−.42**	−.05	−.34**	.41**
16. Anxiety 3.57 (4.39)	−.32**	−.23**	−.13*	.20**	−.31**	−.12*	.35**	.68**	.76**	.48**	−.40**	−.13*	−.29**	.37**	.84**
17. Intrusions 1.67 (3.26)	−.03	−.09	.05	.20**	−.18**	−.03	.40**	.50**	.55**	.55**	−.05	.02	−.19**	.40**	.63**	.68**

Note: In 2020, N’s ranged from 629 to 630. In 2021, N’s ranged from 374 to 377. * p < .05. ** p < .01. Italicized coefficients were statistically significant when applying the Bonferroni-Holmes correction.

Hypothesis 1a: Cross-sectional predictors of distress

At both assessments, most candidate predictors were inversely correlated with the distress outcomes (Table 1, see italicized coefficients for significant effects when applying the Bonferroni-Holmes correction). Life meaning and social support were most strongly related with lower distress. Religiosity was related with greater distress.

In forward inclusion models, life meaning and social support predicted lower depression and anxiety while religiosity predicted greater symptoms (Table 2). Moreover, social support predicted lower intrusions, and religiosity and science attitude greater intrusions. Since life meaning predicted depression and anxiety, we repeated these analyses with the meaning components as separate predictors. Comprehension emerged as the first-included predictor for both distress outcomes (β_(depression) = −0.47; p < .001; β_(anxiety) = −0.42; p < .001). Purpose and mattering had no significant effects.

Table 2. Prediction of distress severity in 2020 and 2021: results of hierarchical regression analyses using forward inclusion.

	Depression 2020			Anxiety 2020			Intrusions 2020
Step	Added variable	beta; p	R²	Added variable	beta; p	R²	Added variable	beta; p	R²
1			.17			.13			.03
	Life meaning	−.41; <.001		Life meaning	−.36; <.001		Religiosity	.17; <.001
2			.21			.16			.05
	Life meaning	−.29; <.001		Life meaning	−.26; <.001		Religiosity	.17; <.001
	Social support	−.24; <.001		Social support	−.21; <.001		Social support	−.17; <.001
3			.23			.19			.06
	Life meaning	−.32; <.001		Life meaning	−.29; <.001		Religiosity	.18; <.001
	Social support	−.23; <.001		Social support	−.20; <.001		Social support	−.19; <.001
	Religiosity	.16; <.001		Religiosity	.17; <.001		Scientism	.08; .049
	Depression 2021			Anxiety 2021			Intrusions 2021
Step	Added variable	beta; p	R²	Added variable	beta; p	R²	Added variable	beta; p	R²
1			.48			.57			.30
	Depression 2020	.69; <.001		Anxiety 2020	.76; <.001		Intrusions 2020	.55; <.001
2						.58			.30
				Anxiety 2020	.74; <.001		Intrusions 2020	.53; <.001
				LOC	−.09; .011		Religiosity	.09; .040
3						.58			.31
				Anxiety 2020	.73; <.001		Intrusions 2020	.51; <.001
				LOC	−.09; .008		Religiosity	.09; .033
				Religiosity	.07; .045		Social support	−.09; .040

Note: N = 629 for 2020 outcomes. N = 376 for depression and intrusions in 2021. N = 377 for anxiety in 2021.

R² refers to adjusted R². Step 1 for the outcomes in 2021 was pre-set.

Hypothesis 1b: Prospective predictors of distress

Several 2020 candidate factors were inversely related with 2021 distress (Table 1, see italicized coefficients for significant effects when applying the Bonferroni-Holmes correction). The inverse relations with life meaning and social support were the strongest. In forward inclusion models controlling for the respective 2020 outcomes, none of the factors prospectively predicted depression and only LOC and religiosity predicted anxiety, negatively, and positively, respectively (Table 2). Moreover, religiosity predicted greater intrusions and social support predicted lower intrusions in forward inclusion models controlling for 2020 intrusions. Similar results emerged when modeling the factors individually and controlling for baseline distress (Table S3).

Exploratory follow-up analysis

One potential reason for the null findings in the prospective analyses (controlling for baseline distress) is that 2020 distress accounted for much variance of 2021 distress (rs ranged from .55 to .76; Table 2), as distress remained relatively stable over time (see Descriptive Statistics; Tables S1, S2).

Hypothesis 2a: Cross-sectional moderation models

Life meaning and religiosity moderated the link between being negatively affected by the pandemic and depression (Table 3). Specifically, life meaning weakened the link and religiosity strengthened it. Religiosity strengthened the association between being affected and anxiety. Moreover, religiosity and science attitude strengthened the link between being affected and intrusions.

Table 3. Output from multiple linear regression analyses including all predictors and interaction effects.

	Outcome 2020			Outcome 2021
	Depression	Anxiety	Intrusions	Depression	Anxiety	Intrusions
Predictor	beta; p	beta; p	beta; p	beta; p	beta; p	beta; p
Life meaning	−.30; <.001	−.27; <.001	−.06; .156	−.07; .170	−.08; .072	.06; .229
LOC	−.02; .539	.00; .994	−.01; .772	−.04; .330	−.09; .02	−.10; .035
Scientism	−.03; .461	−.04; .253	.08; .031	.05; .183	−.00; .938	.07; .101
Religiosity	.10; .002	.10; .003	.13; <.001	.08; .044	.08; .034	.08; .074
Social support	−.22; <.001	−.19; <.001	−.17; <.001	−.10; .060	.01; .831	−.10; .077
Social cohesion	.04; .378	.04; .373	.07; .121	.08; .089	.05; .184	.04; .405
Affected	.33; <.001	.34; <.001	.31; <.001	.08; .065	.07; .073	.22; <.001
2020 symptoms				.60; <.001	.67; <.001	.40; <.001
Life meaning*A	−.09; .022	−.03; .435	−.04; .420	−.02; .650	−.03; .495	.05; .346
LOC*A	.06; .108	.02; .704	−.03; .424	−.04; .319	.04; .352	.00; .973
Scientism*A	.05; .163	.02; .692	.14; .001	.11; .013	.12; .002	.16; <.001
Religiosity*A	.10; .004	.10; .003	.22; <.001	−.00; .959	.04; .275	.11; .016
Social support*A	−.08; .080	−.08; .095	−.07; .160	−.02; .667	−.10; .021	−.20; <.001
Social cohesion*A	.03; .431	.05; .304	.04; .366	.04; .329	.01; .832	−.02; .672
R² full model	.35	.31	.21	.49	.60	.41

Note: N = 629 for 2020 outcomes. N = 376 for depression and intrusions in 2021. N = 377 for anxiety in 2021.

2020 symptoms stands for the respective symptom severity in 2020. *A indicates interaction effects with being negatively affected by the pandemic.

R² refers to adjusted R².

We executed separate exploratory moderation analyses with the meaning components and depression as the outcome to increase the insights into the life meaning effects. Results showed that comprehension (beta = −0.11; p = .006) and purpose (beta = −0.09; p = .018) were significant moderators, but mattering was not (beta = −0.02; p = .653).

Hypothesis 2b: Prospective moderation models

Science attitude strengthened the link between being affected by the pandemic and depression (for an overview of the results, Table 3). For anxiety, social support weakened the link with being affected and science attitude strengthened it. Social support weakened the link between being affected and intrusions. Science attitude and religiosity strengthened it.

To elucidate the unexpected science attitude effect, we explored whether it was linked with potentially distressing pandemic-related factors which may result in current findings. Indeed, science attitude coincided with greater media exposure (r = .10; p = .011), a distress predictor (Bendau et al., 2021), and finding the pandemic personally relevant (r = .15; p < .001), but not with finding the pandemic harmful (r = .04; p = .286).

Hypothesis 3: Differences between prior East and prior West Germany

East and West only differed significantly on religiosity (t(620.29) = 5.20; p < .001; d = .41; West: M = 4.47; SD = 4.71; East: M = 2.60; SD = 4.32). For other predictors, p-values ranged from .202 to .752. The distress variables did not differ between East and West (p-values ranged from .179 to .914). Therefore, no mediation analyses were executed.

Discussion

This study investigated the individual roles and relative importance of candidate predictors of psychological distress during the COVID-19 pandemic in two ways: (1) cross-sectionally after the first lockdown in Germany (May–July 2020) and (2) prospectively, with 2020 factors predicting 2021 post-lockdown distress. These predictors included the psychological factors of meaning in life, internal locus of control, a positive attitude towards science, and religiosity, and the social factors of social support and social cohesion.

Most candidate factors had bivariate inverse correlations with the distress outcomes of depression, anxiety, and intrusion severity, cross-sectionally and prospectively. However, only a few factors predicted lower distress prospectively above and beyond 2020 distress. Social support and life meaning most strongly predicted lower distress and partly weakened (i.e. moderated) the link between being negatively affected by the pandemic and distress. Conversely, religiosity predicted greater distress and partly strengthened (i.e. moderated) the link between being affected by the pandemic and distress (Table 4). East and West Germans did not differ on distress but on religiosity.

Table 4. Schematic overview of significant relationships.

	Meaning			LOC			Scientism			Religiosity			Social support			Social cohesion
Outcome	r	b	ME	r	b	ME	r	b	ME	r	b	ME	r	b	ME	r	b	ME
Depression 2020	−	−	−	−			−			+	+	+	−	−		−
Depression 2021	−			−					+	+			−	−		−
Anxiety 2020	−	−		−			−			+	+	+	−	−		−
Anxiety 2021	−			−	−		−		+	+	+		−	−	−	−
Intrusions 2020								+	+	+	+	+	−	−
Intrusions 2021									+	+	+	+	−	−	−

Note: rs indicate Pearson correlations. bs indicate regression coefficients. ME’s indicate moderation effects.

− indicates a significant negative effect, blank cells indicate no significant effect, + indicates a significant positive effect.

Social support and life meaning as relatively robust predictors of psychological distress

Social support

In the current study, social support was the factor most consistently related with lower distress (Table 4). In addition to being correlated with all distress outcomes cross-sectionally and prospectively, social support was also a significant predictor for all outcomes cross-sectionally and for intrusions prospectively. Moreover, social support weakened (i.e. moderated) the relation between being affected by the pandemic and 2021 intrusions and anxiety.

In prospective analyses, more moderation effects and fewer bivariate relations with social support emerged compared to the cross-sectional analyses. This may indicate that while social support was generally beneficial after the first lockdown, it was more strongly related with distress for individuals more strongly affected by the pandemic a year later, possibly because they were in greater need of support. Moreover, social support prospectively predicted less intrusions, a core symptom of PTSD. This might indicate that social support counteracted distress related to potentially traumatic events (e.g. death of a loved one; for a meta-analysis showing social support as a PTSD predictor, see Ozer et al., 2003) rather than related to the diffuse pandemic stressors (e.g. social distancing).

That social support was the most consistent predictor of lower distress might be because it helps to foster effective coping (Cohen, 2004). This finding adds to the literature pointing to the importance of social support in stressor contexts, with direct (e.g. Brewin et al., 2000; Guilaran et al., 2018; Ozer et al., 2003), and stressor-buffering effects (e.g. Howell et al., 2021) and extends prior findings by replicating beneficial effects during the pandemic and demonstrating its relative importance compared to other potential distress-reducing factors. Since social support also predicted lower distress earlier during the pandemic (Ahrens et al., 2021), this indicates that social support may have benefited the continued adaptation to the pandemic.

Meaning in life

Meaning in life showed inverse correlations with distress cross-sectionally (as in these COVID-19 studies: Schnell & Krampe, 2020; Seidel et al., 2023) and prospectively, but did not prospectively predict less distress when controlling for baseline distress (Table 4). Also, life meaning weakened (i.e. moderated) the link between being affected by the pandemic and depression cross-sectionally (Table 3). That life meaning did not have prospective effects above and beyond baseline distress contrasts with previous findings (e.g. Disabato et al., 2017; Krause, 2007). One potential reason for the discrepancy is related to the sample. Our participants showed limited variability in terms of both life meaning and distress over time. This may be related to their age group (as also in Nivard et al., 2015), leaving limited variance to be explained. Moreover, the studies demonstrating prospective effects had larger sample sizes (e.g. N = 797 in Disabato et al., 2017; N = 1093 in Krause, 2007) and hence greater statistical power (.98 and .99, respectively, for a small effect, f² = .02). This also applies for a peri-pandemic study which showed prospective effects (Schnell & Krampe, 2022; N = 431). In the current study, statistical power may have been too low to detect smaller effects (e.g. post-hoc power to detect the prospective effect on depression was .36).

That life meaning was related with lower distress in cross-sectional regression models might be because life meaning reduced uncertainty (Hirsh et al., 2012) and hence the distress associated with it (Ostafin et al., 2022). Supporting this idea, the meaning component of comprehension, which has been conceptualized as the opposite of uncertainty (Martela & Steger, 2016), was the strongest predictor out of the meaning components in exploratory, secondary analyses. However, cross-sectional results do not warrant causal inferences and the variables might be related in other ways (e.g. distress lowering meaning).

Religiosity

Against our prediction and contrasting with prior findings (e.g. Hackney & Sanders, 2003; Hodapp & Zwingmann, 2019), religiosity was consistently related with greater distress and predicted greater distress in joint cross-sectional and prospective models (Table 4). Religiosity also exacerbated the effect of being affected by the pandemic on all outcomes cross-sectionally and on intrusions prospectively. These unexpected findings might be related with the current religiosity measure. This measure has not been used widely, and when grouped with spirituality items into ‘vertical transcendence’, the measure was not related with depression and anxiety (Schnell, 2009). However, since the measure addresses deriving strength from faith and finding religion and praying important, which resemble the ‘personal devotion’ aspect of religion, we expected it to be inversely related with distress (for a meta-analysis, see Hackney & Sanders, 2003). Another potential reason for the positive relations with distress is that religiosity might have fostered maladaptive religious coping (e.g. interpreting the pandemic as Divine punishment; for a meta-analysis on differential religiosity effects depending on the derived coping mechanism, see Ano & Vasconcelles, 2005). In line with this proposition, research with German cancer patients showed that higher age coincided with more maladaptive religious coping, which, in turn, predicted greater anxiety (Zwingmann et al., 2008). Hence, also in the current, partly older sample, maladaptive religious coping might have fostered distress.

Less robust predictors of psychological distress

A positive attitude towards science

Contrary to our prediction, science attitude was only partly related with lower distress and predicted greater distress in joint models. Specifically, science attitude predicted greater intrusions cross-sectionally and strengthened (i.e. moderated) the effects of being affected by the pandemic on distress, especially prospectively (Table 4).

Potential reasons that science attitude predicted greater distress include that individuals with greater trust and interest in science consumed more pandemic-related content, a factor linked with distress (Bendau et al., 2021). Further, scientific-minded individuals may have been more aware of the threats of the COVID-19 outbreak which might have fostered distress. In the context of traumatic events, threat awareness has been linked with greater PTSD severity (for a meta-analysis, see Ozer et al., 2003). However, although science attitude was linked to finding the pandemic personally relevant, it was not linked to finding the pandemic harmful, drawing this reasoning into question.

Internal locus of control (LOC)

LOC, the generalized belief to have control over one’s life, though cross-sectionally and prospectively related with lower distress, only significantly predicted 2021 anxiety in joint models (Table 4). While LOC may counteract distress by enabling individuals to take responsibility for their fate (Strickland, 1978), it might be that a contextual factor related to the pandemic, uncontrollability, hindered potential effects. In an experimental study, LOC only had a beneficial effect on a psychophysiological stress index when participants believed that the stressor was influenceable (Bollini et al., 2004). During the pandemic, many aspects of life were influenced by external factors (e.g. lockdown-related restrictions), limiting personal control. And indeed, participants found the pandemic only moderately controllable, which might have weakened LOC effects and resulted in current results.

Social cohesion

Although social cohesion was, as shown in prior research (e.g. Baranyi et al., 2020; Erdem et al., 2016), associated with less depression and anxiety cross-sectionally and prospectively, it was not a significant predictor in joint models, nor a moderator in the relation between being affected and distress (Table 4). A potential reason for these insignificant effects is that social cohesion may have a positive effect through individual-level social support. That is, it may aid adaptation because it entails having a supportive environment offering help and comfort when needed (also noted in Erdem et al., 2016). In the current sample, social cohesion and social support were moderately correlated, which when modeled together, might have rendered social cohesion, as the more distal factor, insignificant (as shown in a previous study; Mulvaney-Day et al., 2007). Moreover, the links between distress and social cohesion were weaker than distress links with social support, potentially because social support is more closely linked with distress since it aims to support coping with stressors and challenges (Cohen, 2004) while social cohesion might not be as directly stressor-related.

Similarities rather than differences between prior East and prior West Germany

Given the differential childhood environments of participants who grew up in prior East Germany and prior West Germany, we compared how these two subsamples responded to the pandemic. There were no significant differences in distress severity. Moreover, we expected differences in religiosity, LOC, and social factors. However, the regions only differed on religiosity (i.e. higher in the West; as in Meulemann, 2009). A potential reason that prior research showed differences in LOC (Friehe et al., 2015) and social cohesion (Dragolov et al., 2016; Kirkcaldy et al., 1999) whereas the current study did not is that East and West Germany might have become more similar over time. How and if cultures can change (e.g. through intercultural contact) has been discussed elsewhere (e.g. Kaasa & Minkov, 2020). However, the LOC and cohesion measures used in the prior studies were broader, encompassing constructs beyond our conceptualizations of these factors (e.g. LOC measure included external LOC and achievement; Friehe et al., 2015; cohesion measures included accepting diversity and trust in institutions; Dragolov et al., 2016; Kirkcaldy et al., 1999). Such operationalization differences impede comparison with current findings since the additional constructs could have influenced the effects.

Peri-pandemic distress was limited in Germans aged 45 to 75

Noteworthily, although depression and anxiety prevalence rates exceeded pre-pandemic estimates, they were lower than other pandemic estimates (e.g. Salari et al., 2020; Shevlin et al., 2020). Moreover, most participants reported having no intrusions. Since intrusions are linked to PTSD and traumatic experiences (Weathers et al., 2013), intrusion absence, together with the rather low scores on ‘being negatively affected by the pandemic’ might indicate that the pandemic did not, on average, represent a major stressor for this group. Moreover, age was related with lower distress. Current findings align with research showing that older adults were less affected by the COVID-19 pandemic than younger ones (e.g. Shevlin et al., 2020). That might be because older individuals’ distress levels tend to be more stable over time (e.g. Nivard et al., 2015) and because pandemic restrictions might have negatively affected older individuals’ lives less.

Strengths and limitations

The current study had several strengths, including the prospective design controlling for baseline distress, which offers insights into the processes of continued adaptation to the pandemic. Moreover, the sample was population-representative, allowing cautious generalization to the studied population. However, the differential dropout renders prospective results less representative. The current study also had several limitations. For example, that for intrusions, model assumptions (e.g. normality) might have been violated draws intrusion findings into question (Ernst & Albers, 2017). Another limitation is that since moderation effect sizes tend to be small, our study might have had too low statistical power to detect effects (Murphy & Russell, 2017). Last, executing several analyses may have increased the family-wise Type I error rate, even though we aimed to reduce this issue by applying a Bonferroni-Holmes correction in correlation analyses and by using a hierarchical regression model limiting the number of tested effects.

Recommendations for future research

The current study identified factors linked with pandemic-related distress. Especially the links with social support, meaning in life, and religiosity should be examined in future stressor research, particularly in longitudinal and experimental settings to investigate their causality and reliability. Future research on stressor-buffering effects should include large samples to maximize statistical power, especially in samples whose symptoms are likely relatively stable, such as older age groups.

Conclusion

During the COVID-19 pandemic, psychological distress was slightly elevated in Germans aged 45 to 70. Of the candidate predictors, social support was most consistently linked with lower distress, indicating that social resources might be particularly important in adjusting to the pandemic. Meaning in life was also related with lower distress, indicating that it might have played a limited protective role against distress. In contrast to our hypotheses, religiosity predicted greater distress, suggesting that religious practices and beliefs might have been maladaptive in this group and context. Further, East and West Germany showed more similarity than expected regarding the assessed candidate distress predictors. This study adds to the literature on adaptation to the COVID-19 pandemic by demonstrating the relative importance of psychosocial distress predictors, particularly social support and religiosity. The here-identified predictors of distress should be studied further and might be considered for future interventions.

Disclosure statement

No potential conflict of interest was reported by the authors.

Additional information

Funding

This study was supported by the UG BSS Internet Research Fund.