Article Text

Original research
Mental health and well-being in the first year of the COVID-19 pandemic among different population subgroups: evidence from representative longitudinal data in Germany
  1. Philipp Jaschke1,
  2. Yuliya Kosyakova1,2,
  3. Coline Kuche3,
  4. Lena Walther4,
  5. Laura Goßner1,
  6. Jannes Jacobsen5,6,
  7. Thi Minh Tam Ta4,
  8. Eric Hahn4,
  9. Silke Hans3,
  10. Malek Bajbouj4
  1. 1Migration and International Labour Studies, Institute for Employment Research (IAB), Nuremberg, Germany
  2. 2Professorship of Migration Research, University of Bamberg, Bamberg, Germany
  3. 3Department of Sociology, Georg-August-Universität Göttingen, Göttingen, Germany
  4. 4Klinik für Psychiatrie und Psychotherapie, Charite Universitatsmedizin Berlin, Berlin, Germany
  5. 5Deutsches Zentrum für Integrations- und Migrationsforschung (DeZIM), Berlin, Germany
  6. 6Berlin Social Science Center (WZB), Berlin, Germany
  1. Correspondence to Philipp Jaschke; Philipp.Jaschke{at}iab.de

Abstract

Objectives To examine potential deteriorations in mental health and well-being in the first COVID-19 pandemic year compared with the previous decade focusing on the following vulnerable subgroups in Germany: women with minor children in the household, those living without a partner, younger and older adults, those in a precarious labour market situation, immigrants and refugees, and those with pre-existing physical or mental health risks.

Design Analyses of secondary longitudinal survey data using cluster-robust pooled ordinary least squares models.

Participants More than 20 000 individuals (aged 16+ years) in Germany.

Primary and secondary outcome measures Mental Component Summary Scale (MCS) of the 12-item Short-Form Health Survey measuring mental health-related quality of life, single item on life satisfaction (LS).

Results We find a decline in the average MCS in the 2020 survey that is not particularly striking in the overall time course, still resulting in a mean score below those of all preceding waves since 2010. We find no change in LS from 2019 to 2020 against the background of a general upward trend. Regarding vulnerability factors, only the results on age and parenthood are partially in line with our expectations. In 2020, LS declined among the youngest adults; MCS declined among mothers (and women and men without children) but not fathers. Unlike respective comparison groups, refugees, those unemployed before the pandemic and those with pre-existing mental health risks experienced no MCS declines in 2020, whereas persons living without a partner, the eldest, and those with pre-existing health risks exhibited continued increases in LS.

Conclusions There is no evidence for substantial breakdowns in mental health or subjective well-being in the first pandemic year in the German population or its subgroups, particularly when considering developments of the previous decade. Since the majority of hypothesised vulnerable groups to pandemic stressors showed more stable MCS and LS, our results warrant further study.

  • mental health
  • COVID-19
  • public health
  • quality of life

Data availability statement

Data are available on reasonable request. Data may be obtained from a third party and are not publicly available. The data analysed in this study is subject to the following licenses/restrictions: this article uses data from the SOEP-CORE (DOI: https://www.doi.org/10.5684/soep.core.v37eu), the IAB-BAMF-SOEP Survey of Refugees (DOI: 10.5684/soep.iab-bamf-soep-mig.2020) and the IAB-SOEP Migration Sample (DOI: 10.5684/soep.iab-soep-mig.2020). The German SOEP is an ongoing representative yearly panel survey of private households in Germany at the German Institute for Economic Research (DIW). The IAB-BAMF-SOEP Survey of Refugees is a representative longitudinal survey conducted jointly by the Institute for Employment Research (IAB), the research data center of the Federal German Office for Migration and Refugees (BAMF) and the German SOEP at the DIW. The IAB-SOEP Migration Sample is a longitudinal survey representative of the migrant population in Germany conducted jointly by the IAB in Nuremberg and the German SOEP at the DIW Berlin. Data access was provided via a Scientific Use File supplied by the Research Data Centre (FDZ) of the DIW. All documentation concerning both surveys, including questionnaires and data manuals, are made available by the DIW (https://www.diw.de/en/diw_01.c.603160.en/integrated_studies.html). Due to German Data Protection legislation, we cannot make the original data or the generated dataset available. Researchers can, however, apply for data access via the DIW. The computer codes for data preparation and analyses are available on request. This study design and analysis was not preregistered.

http://creativecommons.org/licenses/by-nc/4.0/

This is an open access article distributed in accordance with the Creative Commons Attribution Non Commercial (CC BY-NC 4.0) license, which permits others to distribute, remix, adapt, build upon this work non-commercially, and license their derivative works on different terms, provided the original work is properly cited, appropriate credit is given, any changes made indicated, and the use is non-commercial. See: http://creativecommons.org/licenses/by-nc/4.0/.

Statistics from Altmetric.com

Request Permissions

If you wish to reuse any or all of this article please use the link below which will take you to the Copyright Clearance Center’s RightsLink service. You will be able to get a quick price and instant permission to reuse the content in many different ways.

STRENGTHS AND LIMITATIONS OF THIS STUDY

  • The paper uses data from a large-scale longitudinal random sample of private households in Germany to uncover pandemic-related changes in the long-term trajectories of mental health and subjective well-being.

  • We employ analyses for two different indicators: the Mental Component Summary Scale capturing mental health-related quality of life as a mental health measure and life satisfaction as a measure of subjective well-being.

  • We consider heterogeneous trajectories of mental health and well-being by looking at several possible vulnerability factors, namely gender and children, living arrangements, age, precarious employment, migration status and pre-existing physical and mental health risks.

  • No data for the succeeding pandemic years after 2020 are available yet, limiting our possibilities to assess changes in mental health and well-being trajectories across vulnerability groups further into the pandemic.

Introduction

Since 2020 and for over 2 years, the COVID-19 pandemic has posed a threat to the physical health of populations globally and may have had direct and collateral effects on individuals’ mental health and well-being.1 2 In the first year of the pandemic, populations worldwide were exposed to major new stressors, such as the fear of infection with a poorly understood virus, severe illness or death, as well as the repercussions of measures to control the spread of the virus via social contact restrictions, as well as economic consequences.2 3 In Germany, the first case was reported at the end of January 2020, marking the start of phase zero with sporadic cases.4 Soon thereafter, the first wave of infections from March to May4 prompted a lockdown, including far-reaching contact restrictions in work and private settings, closures of schools, childcare facilities, non-essentials and gastronomy. After a summer with relatively few cases, a more severe second wave of infections and the beginning of the vaccination campaign followed between October 2020 and February 2021,4 with a partial lockdown in fall 2020 and a more far-reaching lockdown from December 20205 onwards. Owing to the pandemic and containment measures, declines in mental health and well-being in 2020 compared with previous years were widely expected.6 7 The early stages of the pandemic have been examined extensively regarding potential mental health declines in Germany8 and internationally.9–12 However, findings are mixed, and clear conclusions cannot be easily drawn.8 13

Despite the crisis’ global nature, some population groups are likely to have been disproportionally affected.14 The burden of the pandemic—both from the presence of the virus and measures for infection mitigation—may have been unequally distributed across different groups in the population throughout the pandemic. For instance, older individuals and those with pre-existing health conditions face a greater risk of falling severely ill with COVID-19,1 15 which likely led to an increased perception of threat and health anxiety, particularly in the first year of the pandemic, before vaccinations were developed. Parents,5 but especially mothers,16 17 may have been particularly burdened by new childcare demands during school and childcare facility closures. Younger adults experienced social contact and other restrictions during a transitional life stage.18 Older individuals and those living without a partner18 faced increased risks of isolation due to contact restrictions and social distancing.1 Individuals in precarious labour market situations, such as those unemployed or marginally employed, were more likely to endure personal repercussions from economic consequences of the pandemic.19–21 Migrants, particularly refugees (used here to refer to all persons who move to another country for humanitarian reasons, independent of their legal situation), may have faced increased risks due to restricted access to quality healthcare,22 poorer labour market attachment23 and greater risks of isolation when social networks in the country of residence are less established,24 25 and concerns for family and friends in their country of origin. Those with pre-existing mental health conditions may be particularly vulnerable to the potential mental health consequences of these various pandemic-related stressors.18

Indeed, existing research provides some evidence for associations between these vulnerability factors and poorer mental health in the first years of the pandemic in Germany or elsewhere: younger adults,8 26–28 older adults,29 parents,27 30 mothers,31 those with lower education levels,32 33 lower household income33 or those receiving welfare assistance,34 immigrants and refugees,25 35 36 individuals living alone,37 those with pre-existing mental8 38 and physical health conditions11 have been found to show an elevated risk of mental health deterioration. Fewer studies also examined life satisfaction, identifying corresponding declines in the overall German population39 40 and particularly for mothers.41 Yet, these studies looked only at shorter time spans (ie, one time point before the pandemic or retrospective measures).

In light of previous research, we identify three main research gaps. First, we argue that it is crucial to consider both measures of mental health and subjective well-being as outcomes because the pandemic may have different impacts on the latter, particularly on cognitive evaluative measures such as life satisfaction. Second, nationwide representative studies (exceptions include studies by Mauz et al and Sachser et al42 43) and, in particular, longitudinal studies (exceptions include studies by Hettich et al, Entringer and Kröger and Peters et al28 44 45) on mental health and well-being indicators prepandemic and peripandemic in Germany are still rare. Moreover, these and other existing studies consider only more recent prepandemic timespans or a single baseline period, not regarding longer prepandemic trends. In other words, previous studies have been limited in their capacity to distinguish differences in mental health and well-being between observation periods before and during the pandemic that may be part of longer ongoing trends or reflect a degree of fluctuation observed in previous years from pandemic-related changes. Third, migrant and especially refugee populations as potentially vulnerable groups have scarcely been addressed in the existing literature (exceptions include studies by Goßner et al and Entringer et al25 46).

The present study addresses these research gaps by examining the potential impact of the early phases of the pandemic on the development of the Mental Component Summary Scale (MCS) of the 12-item Short-Form Health Survey (SF-12), which assesses mental health-related quality of life,47 and life satisfaction, a cognitive component of subjective well-being,48 using nationwide, representative, longitudinal data from approximately 20 000 adults in Germany. We explore the heterogeneous impact of the pandemic by addressing a whole set of candidate vulnerability factors: being female with children under the age of 16 years in the household, living without a partner, being a younger or older adult, migration status, various less secure forms of employment and pre-existing physical and mental health risks. We hypothesised that these vulnerability factors are associated with declines in MCS and life satisfaction in the first pandemic year compared with the prepandemic period. We used data from the Socio-economic Panel (SOEP-Core) (17 611 individuals providing 75 266 person-year observations), IAB-SOEP Migration Sample (2018 individuals providing 6978 person-year observations) and IAB-BAMF-SOEP Survey of Refugees (2391 individuals providing 6014 person-year observations) (the number of observations refers to the analytical sample for the MCS and sociodemographic vulnerability factors), which encompass 6 survey waves for the MCS (between 2010 and 2020) and 11 survey waves for life satisfaction (2010–2020). The SOEP-Core study is a wide-ranging annual representative longitudinal study of private households, located at the German Institute for Economic Research (DIW). The IAB-SOEP Migration Sample is conducted jointly by the Institute for Employment Research (IAB) and the SOEP. The IAB-BAMF-SOEP Survey of Refugees is conducted jointly by the IAB, the research data center of the Federal German Office for Migration and Refugees (BAMF) and the SOEP. To analyse the developments in both indicators, we estimated weighted pooled ordinary least squares (OLS) regressions and assessed the pandemic impact by using interaction effects between vulnerability factors and survey year dummies.

Methods

Patient or public involvement

Patients and/or the public were not involved in the design, or conduct, or reporting, or dissemination plans of this research.

Study design and analytical sample

The data employed in the present study come from the German Socio-economic Panel (SOEP, V.37, EU version), which comprises the general SOEP-CORE population survey, 2010–2020,49 and two integrated studies covering the recent immigrants in Germany, that is, the IAB-SOEP Migration Sample, 2013–2020,50 and the IAB-BAMF-SOEP Survey of Refugees 2016–2020.51

The SOEP-CORE is a large-scale longitudinal representative study of private households in Germany, launched in 1984 and conducted annually.49 The target population of the IAB-SOEP Migration Sample was drawn from the register data of the Federal Employment Agency, the so-called Integrated Employment Biographies, in 2013.50 The study is representative of immigrants arriving in Germany since 1995 and descendants of immigrants born after 1976.52 53 As part of the SOEP-CORE study, the last immigrant refreshment sample before the IAB-SOEP Migration Sample dates back to 1995. The survey is based on a concept of households according to which every adult household member is interviewed. The IAB-BAMF-SOEP Sample of Refugees in Germany was launched in 2016, in the aftermath of the surge of refugee migration to Europe in 2015.51 The data were drawn from the Central Register of Foreigners54 and are representative of refugees who arrived in Germany between January 2013 and December 2016 (irrespective of their current legal status). It is based on the same household concept as that described for IAB-SOEP. By using appropriate sample weights, SOEP data allow us to make inferences for the population in Germany.

For analyses, we restricted the original data to the years 2010–2020 to minimise the lingering effects of the financial crisis in 2008/2009, which had a negative public health impact.55 56 Moreover, we considered 2010–2020 observations of respondents who participated in the survey year 2020, that is, pandemic survey year, and in at least one prepandemic survey year. Due to specifics of the fieldwork, SOEP-CORE interviews were collected between 10 January 2020 and 8 December 2020, interviews of the IAB-SOEP Migration Sample between 4 March 2020 and 12 August 2020, while the IAB-BAMF-SOEP Sample of Refugee interviews started on 24 August 2020 and were completed on 15 February 2021. We included data from respondents interviewed on 31 January or later (exclusion of four respondents), the day after WHO declared COVID-19 a ‘public health emergency of international concern’ and 3 days after the first case in Germany.57

These data restrictions resulted in a sample of 22 020 individuals for the analysis of mental health trajectories by sociodemographic vulnerability factors that have been interviewed up to 6 times (4.5 on average). While about 80% of interviews in our used SOEP-CORE and IAB-SOEP Migration Sample data were conducted by the end of May 2020, interviews as part of the IAB-BAMF-SOEP Survey of Refugees only started in August 2020, and 83% were conducted by the end of 2020. All analyses in this study are weighted with the sample weights provided with the survey data to compensate for distortions caused by over-represented or under-represented groups, and non-response.

Outcome variables

We consider mental health-related quality of life and subjective well-being as dependent variables. Our measure of mental health-related quality of life was the MCS from the SF-12, which includes six items capturing vitality (energy vs fatigue), social functioning, role limitations due to emotional problems and emotional well-being over the past 4 weeks on a 5-point scale.47 While the MCS has commonly been conceived of as assessing the mental health component of health-related quality of life,58 59 it has also shown to be suitable as a screening instrument for depression and anxiety disorders.60 61 We formed composite MCS-12 scores ranging from 0 to 100 normalised to the 2004 SOEP wave for comparability.62 Except the survey year 2017, when a new refreshment sample of >2000 refugees was added, MCS was collected in 2-year intervals from 2010 until 2020, providing one pandemic-time survey date (between 30 January 2020 and 15 February 2021).

Our measure of subjective well-being is a single item on global life satisfaction, a well-established 11-point scale ranging from 0 (completely dissatisfied with life) to 10 (completely satisfied with life). Life satisfaction is generally conceived as the cognitive component of subjective well-being.63 Life satisfaction was collected annually, yielding pandemic-time observations between 30 January 2020 and 15 February 2021.

Vulnerability factor variables

The vulnerability factors to be examined were measured using the following variables: gender and having children under the age of 16 years living in the same household were grouped in a variable gender and children with the following categories: (1) men without children (reference category), (2) men with children, (3) women without children and (4) women with children. To measure living arrangements, we contrasted living without a partner (including single, married or in a registered partnership living separately, widowed) with cohabiting with a partner (irrespective of marriage).

Age was grouped into the following five categories: (1) 16–25 years (reference category), (2) 26–45 years, (3) 46–59 years, (4) 60–74 years or (5) 75+ years. The age group of those aged 16–25 years represents young adults in education or early career. The second age group comprising persons aged between 26 and 45 years includes working adults. Those aged between 46 and 59 years represent a middle age group with a potentially more established and stable career path. Two last groups aged between 60 and 74 years and those aged 75 years and above are those most at risk of severe courses of disease due to COVID-19 infection.64 65

The variable for migration background is derived based on country of birth and legal status at arrival. The variable includes the following categories: (1) native-born ethnic majority (reference category), (2) immigrant (no refugee), (3) descendant of immigrant (no refugee), (4) refugee.

Employment status includes the following categories: (1) full-time, part-time employed or other (training/apprenticeship, sheltered workshop) (reference), (2) marginally employed (which is characterised by low absolute remuneration—for example, with a maximum of €450 in 2019 as defined by law—and short working hours), (3) self-employed, (4) unemployed (not working but job-seeking) and (5) inactive (retired or those not working but not job seeking).

An indicator for pre-existing (ie, prepandemic) mental health risks was coded to one for those who reported having received a diagnosis of depression or burnout at some point in their lives, and zero otherwise. An indicator for pre-existing physical health risks was coded to one for those who reported having been diagnosed with asthma, cardiopathy, cancer, stroke or hypertension at some point in their lives, and zero otherwise. Pre-existing conditions were coded such that they were carried forward from the first report and no conditions were carried backward.

To ensure the correct order of changes in time, we lagged the vulnerability factor values for employment status and health status from the last prepandemic survey wave to prevent reverse causality issues. We included corresponding dummy variables indicating missing information to capture item non-response in vulnerability factor variables.

Control variables

We account for potential confounders that may vary by vulnerability factors and simultaneously shape mental health-related quality of life and life satisfaction. Specifically, we control for the highest educational degree aggregated into: (1) lower secondary education (reference category), (2) secondary or short-cycle non-tertiary education, (3) bachelor’s degree or equivalent and (4) master’s degree or doctorate. We further control for square metres of living space per person (linear and squared) to account for the household’s availability of private space per person. Additionally, we control for district type in which the respondent resides categorised into (1) independent large city (reference category), (2) urban district, (3) rural district with some density and (4) sparsely populated rural district.66 We control for the average unemployment rate in the local labour market region in the interview month to account for the local economic situation. Since Germany’s counties and independent cities are connected by commuter linkages so that local labour markets extend beyond the boundaries of 401 administrative districts, we use the 141 functional local labour markets defined by Kosfeld and Werner.67 To control for temporal trends, we include survey year fixed effects and the calendar month of the interview. Note that in the analyses on life satisfaction, we include yearly dummies from 2010 (reference) to 2020. Since the MCS was surveyed in 2-year intervals, we consider 2-year dummies. Online supplemental appendix tables A1–A3 show descriptive statistics for the vulnerability factors and control variables.

Statistical analyses

All analyses were conducted in Stata V.17.0. Our analyses of the different vulnerability factors were in two parts: we separated analyses pertaining to structural factors and those pertaining to pre-existing health conditions. In the analyses pertaining to structural vulnerability factors, we applied pooled OLS models (with SEs clustered at person-level) regressing MCS score and life satisfaction on the vulnerability factor variables, one interaction term per model of each vulnerability factor variable by the survey year variable and control variables. Correspondingly, we calculated one model per vulnerability factor variable for each outcome variable to include only a single interaction term per model.

We analysed pre-existing health risks as vulnerability factors separately from the other factors because of their uniquely close relationship with the outcome variables. Another reason is reduced sample size, as information on pre-existing health risks is unavailable for most refugees. These pooled OLS models predicting MCS or life satisfaction included dummies for physical or mental health risks, one interaction term per model of physical or mental health risks by the survey year variable as well as all remaining structural vulnerability factors and further controls.

We calculated predictive margins for the outcome variables for each vulnerability factor subgroup from the regression results. Specifically, we estimated the following regressions separately for each of the five vulnerability factor groups V∈ {GENDCHILD, LIVARR, AGEGRP, MIG, EMPL}:

Embedded Image

where Y is either MCS or life satisfaction of individual i in region r in survey year t and bolt font indicates vectors. Concerning vulnerability factors, GENDCHILD relates to gender and children, LIVARR to living arrangements, AGEGRP to age group, MIG to migration status and EMPL to employment status. Embedded Image denotes individual control variables and Embedded Image regional control variables. Note that individual vulnerability factors vary in the prepandemic period, while for the postpandemic period, employment status and health status are fixed to the last observed prepandemic values. The main coefficient of interest θ refers to the interaction effects of the survey year dummies with the vulnerability factor v.

Results

Development of MCS and life satisfaction in the German population

We begin by examining the overall development of the MCS and life satisfaction in the German population in the recent decade. As shown in figure 1, in the pandemic year 2020, the average MCS significantly declined to a level below previous survey waves since 2010. We find a 0.7-point reduction in mean MCS (on the theoretical scale between 0 and 100) in the population in 2020 compared with the last prepandemic measure (2018). While this decline points to a possible pandemic-related impact on mental health-related quality of life, it is not a marked change in the overall time course and could be a continuation of declines seen since 2016. At the same time, we do not observe any change in the population average life satisfaction from 2019 to 2020. Considering the general upward trend of life satisfaction between 2010 and 2020, the absence of change may be linked to a pandemic-related attenuation. In the following, we examine whether the observed trends differ between population subgroups with different vulnerabilities to the pandemic.

Figure 1

Mental Component Summary Scale (MCS) and life satisfaction between 2010 and 2020. Source: Socio-economic Panel (SOEP)-CORE (V.37), IAB-SOEP Migration Sample 2013–2020, IAB-BAMF-SOEP Survey of Refugees 2016–2020, weighted.

Development of MCS by vulnerability factors

As shown in figure 2, most sociodemographic groups show a significant decline in estimated MCS from 2018 to 2020 (see online supplemental appendix table A4 for the corresponding regression coefficients): (1) women with and without children as well as men without children, (2) those cohabiting with a partner as well as those living without a partner, (3) all age groups, (4) the native-born ethnic majority as well as immigrants and their descendants (but not refugees) and (5) those in full-time/part-time employment, self-employed and those who are inactive (online supplemental appendix table A5 reports the main (non-interacted) effect of the dummy variable 1 (survey year=2020) for varying base categories in vulnerability factor variables). Likewise, following figure 3, (6) those with and without certain pre-existing physical health risks and (7) persons without certain pre-existing mental health risks show a significant decline in the estimated MCS (see online supplemental appendix table A6 for the corresponding regression coefficients). Yet, in the context of the overall trajectories since 2010, the MCS declines from 2018 to 2020 in the outlined groups are not of a remarkable magnitude or otherwise particularly striking.

Figure 2

Predictive margins for Mental Component Summary Scale (MCS) by sociodemographic vulnerability factors. Source: Socio-economic Panel (SOEP)-CORE (V.37), IAB-SOEP Migration Sample 2013–2020, IAB-BAMF-SOEP Survey of Refugees 2016–2020, weighted. Predictive margins partial out the main effects of vulnerability factor variables from other panels and all control variables as discussed above. Shaded areas denote the 95% CIs.

Figure 3

Predictive margins for Mental Component Summary Scale (MCS) by pre-existing health conditions. Source: Socio-economic Panel (SOEP)-Core (V.37), IAB-SOEP Migration Sample 2013–2020, IAB-BAMF-SOEP Survey of Refugees 2016–2020, weighted. Predictive margins partial out the main effects of sociodemographic vulnerability factor variables and all control variables as discussed above. Shaded areas denote the 95% CIs.

Five groups did not show estimated MCS declines from 2018 to 2020: (1) men with children, (2) the marginally employed and (3) unemployed group, (4) refugees and (5) those with prior mental health risks. For the marginally employed individuals, we observed declines as well, yet, these were statistically insignificant; however, it should be noted that the sample size in this group is small and limits statistical power. There is no evidence from the overall time trajectories that the absence of a decline in the other four groups may represent an attenuation of a previous upward trend. Even though these groups’ MCS may have improved without the pandemic, it can be ruled out that the pandemic led to a decline from previous levels. Note that individuals with prior mental health risks exhibit by far the lowest MCS of all groups throughout the observation period; the unemployed exhibit the second lowest levels of MCS.

Development of life satisfaction by vulnerability factors

In contrast to the results for the MCS, most sociodemographic groups do not show pronounced changes in estimated life satisfaction from 2019 to 2020 (see figure 4 and online supplemental tables A7 and A8). The only exceptions were decreases in life satisfaction among the youngest age group (aged 16–25 years), full-time/part-time employed, the self-employed and those without pre-existing physical health risks. Persons living without a partner, the two oldest age groups (aged 60–74 years and 75 years and over), those inactive in the labour market and the unemployed group exhibit significant increases in life satisfaction from 2019 to 2020. Pre-existing physical or mental risks are associated with a significant increase in life satisfaction from 2019 to 2020. In contrast, persons without pre-existing physical risks show some deterioration in the same period (figure 5 and online supplemental appendix table A9).

Figure 4

Predictive margins for life satisfaction by sociodemographic vulnerability factors. Source: Socio-economic Panel (SOEP)-CORE (V.37), IAB-SOEP Migration Sample 2013–2020, IAB-BAMF-SOEP Survey of Refugees 2016–2020, weighted. Predictive margins partial out the main effects of vulnerability factor variables from other panels and all control variables as discussed above. Shaded areas denote the 95% CIs.

Figure 5

Predictive margins for life satisfaction by pre-existing health conditions. Source: Socio-economic Panel (SOEP)-CORE (V.37), IAB-SOEP Migration Sample 2013–2020, IAB-BAMF-SOEP Survey of Refugees 2016–2020, weighted. Predictive margins partial out the main effects of sociodemographic vulnerability factor variables and all control variables as discussed above. Shaded areas denote the 95% CIs.

Looking at the overall time trajectories of estimated life satisfaction in figures 4 and 5, none of the significant increases or decreases from 2019 to 2020 appear particularly striking in the context of the observation period between 2010 and 2020. In most groups, life satisfaction has increased gradually over these 10 years. Thus, the absence of a change from 2019 to 2020 could represent pandemic-related attenuations, while significant increases may be the continuation of ongoing trends. Increases for persons without a partner, the oldest two age groups and persons with pre-existing physical or mental risks visually do not stand out against the trend of increasing life satisfaction before 2019. However, these increases are still unexpected, given that they are all found within hypothesised vulnerable groups and not their hypothesised non-vulnerable counterparts. The increases among inactive and unemployed persons appear slightly more marked in the time course. Concerning the observed significant declines, the overall time trends provide context as follows: in the youngest age group and the employed group, decreases are relativised by year-to-year fluctuations of similar magnitudes before 2019. However, the youngest age does reach a life satisfaction score estimate numerically below most previous estimates in the observation period. For the self-employed, it is quite striking that the decrease goes against a general trend of increase. The significant reduction in life satisfaction from 2019 to 2020 among persons without prior physical health risks also happens against an overall increasing trend; however, 2016–2017 saw an even slightly greater decline.

While life satisfaction results are largely inconclusive in light of longer time trends, it is noteworthy that despite the pandemic, several hypothesised vulnerable groups showed an increase in life satisfaction, and many groups exhibit no change, demonstrating the clear absence of a pandemic-related life satisfaction breakdown in most subgroups. Overall, only 4 out of 24 groups (youngest age group; full-time/part-time employed; self-employed; without pre-existing physical health risks) may have experienced pandemic-related declines in life satisfaction compared with the year before.

Discussion

Using panel data, we examined the development of mental health-related quality of life (measured using the MCS) and subjective well-being (indicated by life satisfaction) in >20 000 individuals in Germany before and during the pandemic (2010–2020). We found a decline in the MCS population average from 2018 to 2020. With regard to just the early phases of the pandemic, our results are in line with previous findings for Germany indicating increased symptoms of depression and anxiety between April 2020 and June 202028 36 compared with 2019 as well as April and May 2020 compared with a prepandemic baseline (2014–2019),45 but in contrast to a study finding declining symptoms in this period.42 While the mean MCS score in 2020 is below any mean score observed since 2010, this finding needs to be evaluated in light of previous trends suggesting a decline from 2016 to 2018. Further research is therefore needed to disentangle the potential effects of the pandemic from general time trends. Our results for life satisfaction also stress the importance of longitudinal analyses and the consideration of ongoing time trends. While our finding of no change in life satisfaction in 2020 supports resilience, we observed a rising trend in life satisfaction in our data before 2020. Hence, the absence of a decline from 2019 to 2020 may represent a pandemic-related attenuation.

Overall, our findings of declines in MCS scores and the absence of declines in life satisfaction demonstrate the importance of considering several indicators for a deeper understanding of developments in pandemic times. One caveat in comparing these two measures is that they differ regarding temporal specifications. With the MCS, respondents are asked to report on the last 4 weeks (‘How often in the last four weeks…’) while the assessment of life satisfaction is formulated more generally (translated from the German version: ‘How satisfied are you currently, all in all, with your life’). These temporal specifications may be particularly important in fast-moving times like the pandemic, especially given that most observations included in our analyses were collected in the very early phases of the pandemic, that is, possibly before any effects were experienced as longer-lasting or more far-reaching.

With regard to the examined candidate vulnerability factors, most of our findings did not match our expectations. Starting with the vulnerability factor of gender and children, MCS deteriorated in all considered subgroups, except for fathers, whereas none of these groups experienced life satisfaction deterioration. Our results, therefore, only partly conform to previous studies consistently reporting pandemic-related worsened mental health for women8 13 and mothers.31 However, our findings contradict our expectations to find declines for parents (compared with non-parents). However, the stable MCS levels among fathers conform to previous studies pointing to fathers’ increased family satisfaction after changing to short-time work.41 Spending more time with family, for example, through working from home policies and school and childcare facility closures, without bearing most of the childcare burden,5 may have dampened the negative effects of the pandemic for fathers. Overall, consideration of gender and parental status proved crucial in understanding potential pandemic impacts from the gender perspective.

Our analyses for the vulnerability indicator for living arrangements revealed declines in MCS in both groups and even a significant increase in life satisfaction among individuals living without a partner compared with no change among those cohabiting with a partner. Additional results provided in online supplemental appendix figure A1 show similar developments when comparing persons living alone with those living with others in the household (irrespective of partnership) based on the number of persons in the household. Taken together, these results only partly support our expectations. Both groups’ mental health declines may hint at different risks in both groups: on the one hand, decreased MCS of those living without a partner might reflect the challenges imposed by the pandemic. Social isolation and loneliness due to social distancing has been highlighted as a key concern for mental health during the pandemic,68 and being single has been previously linked to greater loneliness during the pandemic.37 On the other hand, decreased MCS of individuals cohabiting with a partner might be related to external stress due to the pandemic, such as autonomy-connection tensions.69 This external stress is associated with an increased risk of marital dissolution and challenges the partnership as such, exacerbating existing problems or creating new ones.70 71 Our findings of a potential attenuation of life satisfaction among those living with a partner but not among those living without a partner may also relate to these stressors. In sum, these results, and the challenge they pose for interpretation highlight the complexity of potential risk and resilience factors.

We found MCS declines from 2018 to 2020 for all age groups, contrary to our hypotheses of an increased risk for decline among the youngest and the eldest. For life satisfaction, we found increases from 2019 to 2020 for the older age groups (60–74 years and 75+ years), contrary to our hypotheses and decreases for those aged 16–25 years, in keeping with our hypotheses. These declines in life satisfaction and those in MCS of the youngest age group support the previous literature showing high psychological distress for young adults during the pandemic.26 27 Early adulthood functions as a critical period and includes necessary steps for interpersonal development like identity formation, separation from childhood family or significant decisions for education and career development.72 73 Social isolation measures and school closures drastically affected the daily life of young adults.

For the vulnerability factor immigration status, we found deteriorating MCS from 2018 to 2020 for all (including those of the native-born ethnic majority) except refugees and no change in life satisfaction from 2019 to 2020 for any of these groups. Our results for refugees are consistent with Entringer et al,46 who found no increase in refugees’ psychological distress from 2016 or 2019 to 2020. At this point, it remains unclear whether the results are due to changes in refugee population compositions, as those with poorer mental health were more likely to have left Germany.46 Our results for life satisfaction also contrast with the study by Goßner et al,25 who revealed negative effects for the specific event of the second nationwide lockdown in Germany on refugees’ life satisfaction. Hence, a closer look at specific time periods might yield different results. The absence of significant drops in refugees’ MCS in our analyses could be attributed to different underlying factors. First, refugees’ mental health may improve over time in the host country as postmigration stressors decrease and individuals become better adapted to the new environment,74 mitigating or eliminating any potential negative pandemic effects. Second, refugee populations may also be more resilient to stressors (such as the pandemic) because of their previous experiences and personal characteristics.75 It should also be noted that different field periods might limit comparability between groups by migration status. However, these differences do not affect the finding of an absence of pandemic-related declines in the refugee population.

Our findings for MCS and life satisfaction among employed (negative effects) and unemployed (positive or no effects) individuals resonate with the results for the UK showing increased mental distress for individuals employed before the pandemic and no such effect for those unemployed or inactive.27 However, we do not examine the possible impact of becoming unemployed during the pandemic, which has been associated with heightened depression symptoms in Germany.76 The negative trend for the self-employed group in both the MCS and life satisfaction is consistent with findings for the UK highlighting psychological distress among the self-employed during the pandemic.77 Financial worries seem to mediate the pandemic’s impact on mental distress,78 79 as the self-employed were more likely to expect income losses during the pandemic and were less likely to be considered in government assistance programmes. Contrary to our expectations, the marginally employed exhibited no MCS declines. However, the observation numbers for this group are small yielding low statistical power. Mixed results were also found for the inactive with decreasing MCS and increasing life satisfaction from the last prepandemic observation to 2020. We take from these results that the choice of measure for psychological well-being needs to be carefully considered and the use of multiple measures is critical to check the robustness of results.

Contrary to our expectations, there was no discernible difference in MCS trajectories between individuals with pre-existing physical health risks and those without in 2020. Patterns in life satisfaction were the direct opposite of our hypothesis, as the group with pre-existing physical health conditions displayed an increase between 2019 and 2020, while the other group experienced a decrease. Although the former group’s increase in life satisfaction could potentially be attributed to a pre-existing upward trend observed since 2016, the absence of any potential pandemic effects is still surprising, considering the heightened risk of severe COVID-19 cases15 in this group. A potential explanation for these results is that the pre-existing conditions were defined as self-reported past diagnoses, and some subset of individuals in this group had fully recovered (long) before the pandemic wave survey, and thus did not perceive a greater threat from COVID-19. Alternatively, these individuals’ overall well-being may have improved due to recovery or even just with time having passed since diagnosis, thereby masking potential deleterious effects of the pandemic.

For individuals with pre-existing mental health risks, we found no change in the MCS from 2018 to 2020 and an improvement in life satisfaction from 2019 to 2020. Given the previous developments over the years, neither finding stands out, making the pandemic impact on pre-existing mental health risks unclear. At the same time, we find a decline in the MCS and no change in life satisfaction for individuals without pre-existing mental health risks. While these results contrast our expectations and some previous literature,8 80 81 a systematic review and meta-analysis present comparable results revealing no evidence of a change in symptoms at the beginning of the pandemic among those with pre-existing mental health conditions, while overall increases in symptoms were found compared with prepandemic levels.11 The authors of this meta-analysis argue that this may be due to the positive impacts of lifestyle changes linked to transmission mitigation measures for this group as well as to regression to the mean effects, whereby recovery processes result in improvements in mental health outcomes over time in those with pre-existing conditions. Likewise, increased mental health problems at the beginning of the pandemic were noted among persons without a pre-existing clinical depression diagnosis compared with no change among persons with such diagnosis in the UK82 and even improvements in those with the most severe mental health disorder burden in the Netherlands.83

The results of the development of MCS during the pandemic raise the question of clinical relevance. Among all groups, the MCS declines from the last prepandemic observation by no more than about one point on the theoretical 100-point scale. Although this average decline is of small magnitude, some groups may experience effects that are of clinical relevance. To explore the clinical relevance of our findings, we dichotomised the MCS according to a threshold identified as clinically relevant in previous studies of the SF-12-based MCS. Specifically, the cut-off value of 45 has been found to have high predictive accuracy of depression and anxiety disorders, for example, in the Australian general population comparing the MCS-12 with physician diagnoses60 and in six European countries, where scores on the MCS-12 were compared with a WHO-issued method (CIDI 3.0) for determining mental disorders in in-depth interviews.61 The replications of our main results are shown in online supplemental tables A10 and A11 (regressions) and in online supplemental appendix figures A2 and A3 (corresponding predictive margins). Conforming to our main results, fathers, those living without a partner, refugees and persons with pre-existing mental health risks are less likely to fall below the threshold of 45 during the pandemic than their corresponding comparison groups if there were above this threshold before. For the comparison groups, the probability of falling below the clinically relevant threshold increases by four to five percentage points.

Limitations

We need to address some limitations. First, the distribution of interview modes in SOEP has changed during the pandemic due to legal contact restrictions and voluntary self-protection measures of respondents and interviewers.84 The survey methodology literature has found effects of survey mode, particularly when collecting health data,85 with face-to-face surveys leading to a higher response rate86 and fewer reports on mental health problems.87 In the 2020 survey, the switch to telephone interviews (CAPI-TEL) was above average among refugee respondents, while self-administered interviews were not offered to these groups. Among other respondents, the share of CAPI interviews decreased from 68% to 39% from 2019 to 2020 and self-administered interviews increased to just under one-third.84 Against this background, it cannot be ruled out that (a) changes in survey mode prepandemic versus peripandemic may affect time trends in the observed indicators and that (b) different adjustments in interview mode between refugees and non-refugees may distort the group comparisons in health indicators.

Second, our data are representative of adult individuals living in private households in Germany. Since healthy individuals are over-represented in such surveys,88 it is to be expected that individuals who are not encountered privately due to particularly severe courses of illness (eg, due to hospitalisation) are systematically under-represented in our sample. During the 2020 field period (January 2020 to February 2021), about 10% were hospitalised.4 However, with just over 2.2 million reported cases of infection by the end of January 2021,89 only a small share of the population had contact with the virus, even taking into account unreported cases that may be many times higher. Hence, we expect the bias due to such systematic wave drop-outs to be small and consider the pandemic effects we measured as representing conservative lower limits.

Third, since a substantial share of observations came from before phase I of the pandemic in Germany, and most of the rest of our data only captures the initial phase of the pandemic, these may have resulted in underestimation of any pandemic-related effects on mental health and subjective well-being. On the one hand, the first year of the pandemic may have been particularly stressful in some regard (lack of knowledge of the virus’ biology and health risk, first-time contact restrictions, etc). On the other hand, there is some empirical evidence for mental health declines, particularly in later stages, starting from late 2020.42 We are also unable to assess any potential changes within 2020 to examine, for example, the potential impacts of lockdown given that the overwhelming majority of our data in the pandemic survey wave was collected before the restrictions under the first lockdown were gradually relaxed.

Fourth, we cannot make a concluding statement regarding the resilience of the most vulnerable groups. They could reflect an individual psychological state resulting from prepandemic hardships, which taught them coping strategies, or reflect successful political measures implemented to alleviate hardships. It may also be that efforts to contain the spread of the virus even conferred some advantages to specific groups, for example, by decelerating life in different ways. In order to disentangle micro and macro effects in this respect, a quasi-natural experimental design between regions that implemented different anti-COVID-19 measures would be necessary.

Fifth, certain relevant vulnerability factors were beyond the scope of this study, primarily due to data limitations. For instance, while income or wealth levels have a strong correlation with the examined employment dimensions, they were not explicitly included in this study, despite their potential relevance to mental health and well-being during the pandemic. Furthermore, we were unable to assess the impact of pandemic-related changes to employment, such as remote working or job loss.

Conclusion

The present study adds to a growing literature on mental health and well-being development in the first pandemic year. It is among the rare studies that are nationally representative, longitudinal, include representative estimates from migrant and refugee populations and stands out for its 10-year prepandemic observation period. Yearly or biannual estimates this far back in time before the pandemic allow for a more comprehensive contextualisation and assessment of the significance of any potential pandemic-related changes.

From a broader perspective, our results reveal three points. First, the findings for pandemic impact must be contextualised into long-term developments. As we show, changes from the last prepandemic observation to 2020 were several times put into a different perspective when considering overall time trends. Second, our study shows pronounced differences between vulnerability groups, confirming that the pandemic did affect some subgroups disproportionally, and that vulnerability factors are worthy of consideration. Third, given the cases where we found different effects on mental health compared with life satisfaction, our study shows the importance of considering other measures to assess the psychological impact of stressful events such as the pandemic.

Our results indicate resilience in certain populations that were initially presumed to be most vulnerable to pandemic-induced stressors, highlighting the need for future research to delve into the underlying mechanisms. Putting aside the aforementioned limitations, our findings imply that vulnerability to novel stressors cannot be anticipated and that measures to protect mental health and well-being in crises may need to be more broadly targeted. Preventive measures to increase resilience could include the provision of educational material to improve mental health literacy, and information on developing self-help strategies, mindfulness skills and positive coping mechanisms.90 91 Alongside these preventive measures, it remains crucial to ensure the availability of psychological support for individuals who experienced lasting pandemic-related declines in mental health and well-being.

Data availability statement

Data are available on reasonable request. Data may be obtained from a third party and are not publicly available. The data analysed in this study is subject to the following licenses/restrictions: this article uses data from the SOEP-CORE (DOI: https://www.doi.org/10.5684/soep.core.v37eu), the IAB-BAMF-SOEP Survey of Refugees (DOI: 10.5684/soep.iab-bamf-soep-mig.2020) and the IAB-SOEP Migration Sample (DOI: 10.5684/soep.iab-soep-mig.2020). The German SOEP is an ongoing representative yearly panel survey of private households in Germany at the German Institute for Economic Research (DIW). The IAB-BAMF-SOEP Survey of Refugees is a representative longitudinal survey conducted jointly by the Institute for Employment Research (IAB), the research data center of the Federal German Office for Migration and Refugees (BAMF) and the German SOEP at the DIW. The IAB-SOEP Migration Sample is a longitudinal survey representative of the migrant population in Germany conducted jointly by the IAB in Nuremberg and the German SOEP at the DIW Berlin. Data access was provided via a Scientific Use File supplied by the Research Data Centre (FDZ) of the DIW. All documentation concerning both surveys, including questionnaires and data manuals, are made available by the DIW (https://www.diw.de/en/diw_01.c.603160.en/integrated_studies.html). Due to German Data Protection legislation, we cannot make the original data or the generated dataset available. Researchers can, however, apply for data access via the DIW. The computer codes for data preparation and analyses are available on request. This study design and analysis was not preregistered.

Ethics statements

Patient consent for publication

Ethics approval

This study involves human participants but was not approved by an Ethics Committee or Institutional Board. The study uses secondary survey data that is generally available for researchers upon registration with the research data center at the German Institute for Economic Research (DIW) in Berlin. Respondents are informed that their participation in the survey is voluntary and that they will not suffer any disadvantages if they do not participate. Before the data are delivered to researchers, appropriate measures are taken to ensure the anonymity of the respondents.

References

Supplementary materials

  • Supplementary Data

    This web only file has been produced by the BMJ Publishing Group from an electronic file supplied by the author(s) and has not been edited for content.

Footnotes

  • PJ, YK, CK and LW are joint first authors.

  • Twitter @Philipp_Jaschke

  • PJ, YK, CK and LW contributed equally.

  • Contributors PJ, YK, CK and LW conceived the study and planned the analyses with feedback from SH, JJ, EH, TMTT and MB. PJ and YK conducted the analyses. PJ, YK, CK, LW and LG wrote the manuscript. JJ, EH, MB and SH provided feedback on the manuscript. All authors read and approved the final manuscript. PJ acts as the guarantor.

  • Funding The authors have not declared a specific grant for this research from any funding agency in the public, commercial or not-for-profit sectors.

  • Competing interests None declared.

  • Patient and public involvement Patients and/or the public were not involved in the design, or conduct, or reporting, or dissemination plans of this research.

  • Provenance and peer review Not commissioned; externally peer reviewed.

  • Supplemental material This content has been supplied by the author(s). It has not been vetted by BMJ Publishing Group Limited (BMJ) and may not have been peer-reviewed. Any opinions or recommendations discussed are solely those of the author(s) and are not endorsed by BMJ. BMJ disclaims all liability and responsibility arising from any reliance placed on the content. Where the content includes any translated material, BMJ does not warrant the accuracy and reliability of the translations (including but not limited to local regulations, clinical guidelines, terminology, drug names and drug dosages), and is not responsible for any error and/or omissions arising from translation and adaptation or otherwise.