The Unexpected Decline in Feelings of Depression among Adults Ages 50 and Older in 11 European Countries amid the COVID-19 Pandemic

Abstract

Findings on the mental health impact of the first wave of the coronavirus disease 2019 (COVID-19) pandemic in Europe are mixed and lack a comparative and longitudinal perspective. The authors used the Survey of Health, Ageing, and Retirement in Europe and fixed-effects regressions to estimate within-individual change in the probability to report feelings of depression between 2005 and 2017 and directly following the first wave of the COVID-19 pandemic in 11 European countries for adults ages 50 and older. The authors found an unprecedented decline in feelings of depression between 2017 and 2020 in all countries that was larger than any previous observed change. The probability to report feelings of depression decreased by 14.5 percentage points on average, ranging from 7 to 19 percentage points in Spain and Switzerland, respectively. Moreover, there were no systematic within-country differences by socioeconomic characteristics, chronic health conditions, virus exposure, or change in activities. These findings challenge conventional wisdom about the mental health impact of the COVID-19 pandemic.

Keywords

COVID-19 mental health comparative longitudinal

The first wave of the coronavirus disease 2019 (COVID-19) pandemic forced most European countries to implement wide-reaching restrictions on private and public life. Scholars immediately voiced their concern, suggesting that heightened economic insecurity and social isolation stemming from nonpharmaceutical interventions could lead to a sharp increase in mental health issues (Holmes et al. 2020). Research on the relationship between the COVID-19 pandemic and numerous aspects of mental health, such as psychological distress, feelings of depression and anxiety, and general subjective well-being, has grown rapidly. By the end of 2020, systematic reviews counted more than 2,000 studies in the field (Aknin et al. 2021; Prati and Mancini 2021). The plethora of research results has shown that the impact of the COVID-19 pandemic on subjective well-being is more complex than originally expected.

The majority of studies assume that the pandemic and governments’ containment measures have a negative effect on mental health (e.g., Fiorillo and Gorwood 2020). For example, government-imposed lockdowns as well as quarantine, social distancing, and self-isolation measures reduced social interactions and increased loneliness, both of which are associated with depression (Killgore et al. 2020). Moreover, concerns about one’s own health or the health of loved ones as well as a feeling of uncertainty about the current situation and the future may engender fear and increase feelings of depression. Those feelings of fear and uncertainty may be reinforced by the rapid spread of disinformation, especially on internet platforms. Both of these arguments linking the pandemic to decreased mental health—reduced social interactions and feelings of worry and uncertainty—may be heightened among older adults ages 50 and older.¹ Older adults are more likely than younger or middle-aged adults to become severely ill and be hospitalized following a COVID-19 infection. Therefore, it is likely that older adults did not increase their social interactions after lockdowns were lifted and that the elderly are more concerned about health issues.

In contrast, some studies have argued that the pandemic is associated with increased subjective well-being and mental health, which has been coined the “eye of the hurricane” paradox (Recchi et al. 2020). Perceptions of subjective well-being and mental health may be relative or positional. Individuals comparing themselves with those suffering the most during the pandemic may declare higher levels of well-being than they would have otherwise. In addition, numerous studies have documented astounding resilience and prosocial behavior in the aftermath of natural disasters and other dramatic events, resulting in psychological gains from adversity (Mancini 2019; Quarantelli 1985; Uchida, Takahashi, and Kawahara 2014). Older adults may also be more likely to experience gains in mental health following the pandemic. For example, retired adults were protected from labor market and income uncertainty, which may lead them to perceive a higher degree of well-being relative to adults still active in the labor market.

In this study, we concentrate on adults ages 50 and older, who are medically more vulnerable to COVID-19 than younger segments of the population. Within the group of older adults, there are subgroups that may be more or less vulnerable to the mental health consequences of the pandemic. For example, older adults become increasingly susceptible to the virus with age. Those without a partner, because of never partnering, divorce, or widowhood, may be at an especially heightened risk of social isolation. Older adults who have not yet entered retirement may be put under stress by employment and income uncertainty. Finally, those who either tested positive for COVID-19 or know someone who tested positive may suffer under the effects of social stigma associated with proximity to infection.

In addition to subgroup variation among older adults, there are likely cross-national differences across European countries. Not only were there large country differences in the timing and incidence of COVID-19 during the first wave, but countries also differed in terms of the strictness of measures taken to limit the spread of the virus (Aidukaite et al. 2021; Béland et al. 2021; Cantillon, Seeleib-Kaiser, and Veen 2021; Greve et al. 2021; Moreira et al. 2021). For example, the first wave of the COVID-19 pandemic was especially pronounced in Italy, Spain, and France. In contrast, countries in Central Europe and Eastern Europe, such as Germany and Poland, were left relatively untouched by the first wave. Therefore, if the pandemic does affect mental health negatively, then the association would be expected to be strongest in countries such as Italy and Spain compared with countries such as Germany and Poland. In addition, prepandemic differences in national labor market and social policy institutions increased the ability of certain countries to limit the economic impact of the pandemic for both individuals and society (e.g., Esping-Andersen 1990; Ferragina and Seeleib-Kaiser 2011; Ferragina, Seeleib-Kaiser, and Tomlinson 2013).

The use of cross-sectional and nonharmonized data sources limits the ability of researchers to assess whether cross-national differences are substantive or a methodological artifact. Sources of methodological differences include the measurement of mental health, target population, survey design, and analytical approach. For example, the foci and measurement instruments of studies range from anxiety, sadness, and depression to inclinations to self-harm, insomnia, and suicide. Most of these are measured through ad hoc questionnaire items or established psychological scales. The target populations of these studies vary across both contexts and selected groups, such as specific occupational or age groups, to local communities and nationally representative samples. The results of numerous studies are based on cross-sectional surveys, most of which rely on convenience and unsystematic sampling (Prati and Mancini 2021). Repeated cross-sectional surveys have enabled researchers to compare the prevalence of mental health issues in pre- and postpandemic periods found in different samples. Other cross-sectional surveys fielded during the pandemic have asked respondents to assess the impact of the COVID-19 on their own mental health. However, the latter cross-sectional approach likely exacerbates the impact of the pandemic because of social desirability bias and lack of a prepandemic baseline estimate of mental health.

Repeated cross-sectional designs have their merits in estimating change in population health across time (Yee and Niemeier 1996). For example, repeated sampling accounts for population change (e.g., compositional differences in the target population). However, longitudinal studies, despite problems with panel attrition and target population definition, are needed to assess within-individual change across time. This is especially important if compositional change masks changes in the underlying population. Some studies have adopted longitudinal designs, taking as a starting point the early months of the pandemic (Fancourt, Steptoe, and Bu 2021; Varga et al. 2021). Varga et al. (2021), for example, tracked respondents’ self-assessed COVID-19-related anxiety in four European countries (the United Kingdom, Denmark, the Netherlands, and France) and found large similarities and a common trend of slight decline in anxiety from March to July 2020. However, an accurate estimate of how the pandemic influenced subjective well-being should ideally cover a longer longitudinal window of observation and include comparable information from the same respondents before and during the pandemic.²

A few studies have used longitudinal data sources based on representative samples from single countries. The results of these studies indicate considerable within- and cross-national heterogeneity. In the United Kingdom, an increase in psychological distress from 2017 to April 2020 and a substantial decrease from April to June 2020 were found (Niedzwiedz et al. 2021). Yet the level of distress remains above prepandemic levels (Daly, Sutin, and Robinson forthcoming). Another study showed that anxiety rose in comparison with prepandemic levels, but not depression (Kwong et al. 2020). Still another showed that young respondents drove the overall increase in depression levels in the United Kingdom (Pierce et al. 2020). Evidence from the Netherlands shows stability in anxiety and depression in the overall population (van der Velden et al. 2020) and among a subsample of individuals diagnosed with depression prior to the pandemic (Pan et al. 2021). A Swedish panel study did not reveal significant changes in indicators of well-being and even demonstrated an increase in self-rated health (Kivi, Hansson, and Bjälkebring 2021). Similarly, a longitudinal study in France found evidence for a rise in average levels of subjective well-being compared with the period from 2017 to 2019 (Recchi et al. 2020). In sum, a lack of comparability among studies hinders substantive insight on how cross-national differences in the exposure and government response might translate into country variation in the relationship between the COVID-19 pandemic and mental health.

We contribute to this debate, providing a unique longitudinal and comparative assessment of the pandemic impact on subjective feelings of depression—one of the most studied outcomes to assess people’s mental health—in a segment of the population that has a higher risk for severe infection and social isolation. Using seven waves of the Survey of Health, Ageing, and Retirement in Europe (SHARE) and individual fixed-effects regression models, we investigate within-individual changes in feelings of depression across 15 years and 11 countries on a sample of older adults.

Our approach has at least four advantages compared with previous literature in the field. First, our long observation period allows us to assess the amount of change in depression levels at seven points in time before the pandemic (2005, 2007, 2009, 2011, 2013, 2015, and 2017) and in the months directly following the first wave of the COVID-19 pandemic in Europe (May to August 2020). Second, we are able to use a comparable measure of mental health to investigate the existence of common or dissimilar trends in depression levels across 11 European countries (Sweden, Denmark, Germany, France, Belgium, Switzerland, the Netherlands, Italy, Spain, the Czech Republic, and Poland). The countries in our study differ both in terms of prepandemic economic performance and institutional arrangements but also in their exposure and response to the pandemic. Third, we provide a stress-test measurement of depression trends with a focus on an epidemiologically vulnerable population. Specifically, our target population of men and women ages 50 and older compose a segment of the population in which we would expect to find a growth in feelings of depression due to the pandemic. Fourth, we are able to assess heterogeneity in pandemic’s effect on mental health across a wide array of socioeconomic characteristics as well as individuals’ exposure to the virus, their vulnerability, and the impact of the pandemic on their everyday lives.

Data and Methods

Sample Selection

We use the first, second, fourth, fifth, sixth, and seventh waves of SHARE (release 7.1.0) in combination with the SHARE COVID-19 special release of the eighth wave (release 0.0.1 beta) (Börsch-Supan et al. 2013). SHARE is a cross-national household panel study that collects information on health, socioeconomic status, and social and family networks of individuals ages 50 and older on a biennial basis.

The first wave of SHARE was collected in 2004 with representative samples of individuals in 11 European countries and Israel. The third wave of SHARE, also called SHARELIFE, consisted of a life history module rather than the regular questionnaire from the first and second waves. In the life history modules of SHARELIFE, retrospective information on respondents’ work, family, residential, and health histories as well as information about childhood was collected. The fourth, fifth, and sixth waves returned to the regular questionnaires from the first two waves that collected various prospective information and saw the addition of numerous countries. The seventh wave, collected in 2017, was a dual wave: respondents and countries who did not participate in the third wave (SHARELIFE) were given the retrospective life history questionnaires rather than the regular prospective interview. Only respondents who had participated in the third wave were given the regular questionnaire. The SHARE COVID-19 dataset collected between June and August 2020, includes information on 27 countries.

We restrict our country sample to those that participated in the third wave, because all respondents from countries that entered SHARE afterward received retrospective questionnaire. Therefore, data on mental health are missing from all respondents in those countries in 2017, which makes it impossible for us to compare indicators collected in 2017 and 2020. The 11 remaining countries include Sweden, Denmark, Germany, France, Belgium, the Netherlands, Switzerland, Spain, Italy, the Czech Republic, and Poland. Note that we exclude Greece, a country that participated in the third wave, because we lack information from waves 4 and 5.

Dependent Variable

We rely on an indicator of subjective depression that was collected in all regular waves.³ In the first wave, respondents were asked, “Has there been a time or times in your life when you suffered from symptoms of depression which lasted at least two weeks?” In the follow-up regular waves and the COVID-19 waves, the question was modified: “In the last month, have you been sad or depressed?” Valid and nonmissing responses to the question are “yes” and “no.”

Stratifying Variables

We include a wide range of variables to estimate differences in the change in depression between 2017 and 2020. First, we analyzed differences for a set of socioeconomic indicators: gender, whether respondents are older or younger than 70 years, whether respondents are active in the labor market, and whether respondents live with or without a partner in the household. In addition, we analyzed four sets of variables. All of the following binary indicators in our analyses are zero during regular waves and only vary between zero and one in the special COVID-19 wave.

Our main variable for the first group is whether respondents reported to have been diagnosed with a major illness or health condition since the last interview, which would have been in 2019 for the majority of respondents. Respondents who answered in the affirmative are then asked whether this was a hip fracture, diabetes or high blood sugar, high blood pressure or hypertension, a heart attack including myocardial infarction or coronary thrombosis or any other heart problem including congestive heart failure, chronic lung disease such as chronic bronchitis or emphysema, cancer or malignant tumor including leukemia or lymphoma but excluding minor skin cancers, or any other illness or health condition. We include the variable in binary form in the presence of a chronic illness. In addition, we used respondents’ answers to the question of whether they had forgone medical treatment because of fear of becoming infected by the coronavirus.

Our main variable for the second group is whether respondents experienced a reduction in labor market attachment when the coronavirus broke out, including those who became unemployed, were laid off, or had to close their business or reduce their weekly work hours including overtime. Respondents who were not active in the labor market prior to the outbreak were coded zero together with those who experienced no reduction in labor market attachment. We also included an indicator for whether respondents reported less monthly income since the pandemic compared with prior to the outbreak, as well as a subjective indicator for how well households can “make ends meet,” dichotomized to either easy or difficult.

The main variable for our third category is whether respondents said that they know someone with symptoms or tested positive for COVID-19. Specifically, respondents were asked whether they or anyone close to them experienced symptoms that they would attribute to COVID-19, such as cough, fever, or difficulty breathing. In addition, they were asked whether they or anyone close to them had been tested for the coronavirus with a positive result, meaning that the person had COVID-19. We additionally analyzed these indicators—symptoms and positive test results—separately. We pooled the two indicators to increase the sample size but also because roughly 700 reported not knowing anyone with symptoms but knowing someone who had tested positive.

The fourth category of our main variable is whether respondents reported going for a walk less often. This is one of several indicators based on the question of whether respondents had left their homes since the outbreak. Those who reported not having left their homes were additionally asked whether they went shopping less often, whether they met five or more people outside home less often, and whether they visited family less often. We analyzed all these indicators separately.

Analytical Strategy

We use time-demeaned individual fixed-effects models to estimate average within-individual change in depression across waves estimated separately by country. Specifically, for each country we estimate

y_{i t} - {\bar{y}}_{i} = β_{0} + \sum_{j = 0}^{J} β_{1_{j}} (W_{j_{i t}} - {\bar{W}}_{j_{i}}) + (\in_{i t} - {\bar{\in}}_{i})

(1)

where β_1j represents the average within-individual change for waves 1, 2, 4, 5, 6, and 8 (2005, 2007, 2011, 2013, 2015, and 2020 respectively) in self-assessed depression relative to wave 7 (2017). The estimates for waves prior to 2017 allow us to assess the degree of cross-temporal change in subjective depression between 2005 and 2017, while the estimate for 2020, displayed in equations 2 to 4 as β_1,8, represents our estimate for the impact of the coronavirus pandemic and its accompanying phenomena until the summer of 2020 relative to 2017.

Once we established our baseline estimate for change between 2017 and 2020, we aimed to assess whether there is systematic variation by the indicators discussed previously. We therefore interacted each indicator, represented by X, with the wave indicators as follows:

\begin{array}{l} y_{i t} - {\bar{y}}_{i} = β_{0} + \sum_{j = 0}^{J} β_{1_{j}} (W_{j_{i t}} - {\bar{W}}_{j_{i}}) \\ + β_{2} (W_{8_{i}} - {\bar{W}}_{j_{i}}) (X_{i t} - {\bar{X}}_{i t}) + (\in_{i t} - {\bar{\in}}_{i}) \end{array}

(2)

As our stratifying indicators are constantly zero in the regular waves, we receive only one estimate, denoted β₃, that indicates how the change between the 2017 and 2020 wave differs for respondents with a value of one rather than zero on the binary indicator. We use cross-sectional weights to correct for differential sample probabilities and generate representative country sample at each wave.⁴ Regressions adjusted for sociodemographic factors and other individual characteristics retrieve substantively similar results. We therefore report estimates from more parsimonious models below. A description of the sample and missing values on the depression indicator are presented in Tables B1 and B2 in the online supplement.⁵

Results

Change in Feelings of Depression across Countries

First, we assess how feelings of depression changed from 2005 to 2017 and whether there is evidence that the first wave of the COVID-19 pandemic in 2020 affected individual feelings of depression. Figure 1A, shows within-individual change in the probability of expressing feelings of depression in the prepandemic years 2005, 2007, 2009, 2011, 2013, and 2015 and the pandemic year 2020 compared with 2017 in all 11 countries (see Table A1 in the online supplement).

Figure 1.

Change in the probability of reporting feelings of depression by wave and country.

We find that changes in feelings of depression prior to the pandemic year of 2020 are minor across countries and are likely attributable to country- and time-specific factors. However, we find a substantial decline in feelings of depression between 2017 and 2020 in all countries (Figure 1). This decrease is considerably larger than any previous observed change in levels of reported depression since 2005. In Sweden, Denmark, and France—despite different epidemiological incidence rates and strategies for dealing with the first wave of the pandemic—subjective depression declined by roughly 15 percentage points between 2017 and 2020, while there was no statistically significant change in the previous 12-year period.⁶ Likewise, in Germany, Belgium, and Switzerland, we witness a similar decline of subjective depression levels. The largest decrease among these countries was found in Switzerland, where subjective depression decreased by almost 19 percentage points.

The Czech Republic and Poland are more similar to the rest of our sample, with declines in subjective depression levels of roughly 13 and 11 percentage points, respectively, although these two countries were less affected by the first wave of COVID-19. In Spain and Italy, we observe the smallest decline in subjective depression across our country sample between 2017 and 2020: 7 and 9 points, respectively. This could indicate that the large death toll of the pandemic and strict lockdown measures hindered the reduction of subjective depression that we observed elsewhere in Europe. Nonetheless, our results indicate that such strict measures did not reverse the general trend toward lower levels of subjective feelings of depression.

Heterogeneity of Change in Feelings of Depression across Countries

What might explain the decrease in feelings of depression following the first wave of the COVID-19 pandemic across all study countries? Moreover, what might lead to the overall similarity in the change in feelings of depression in the midst of the pandemic? One possibility is that our average estimates mask important heterogeneity. Feelings of depression may vary more markedly in certain socioeconomic groups. For example, increased subjective well-being during the first lockdown in France was found mostly among socioeconomically advantaged and employed individuals (Recchi et al. 2020).

The estimated averages of within-individual change between 2017 and 2020 by gender, age, household composition, and employment show that there is little heterogeneity with regard to socioeconomic and demographic characteristics (see Figure 2 and Tables A2a–A2d in the online supplement). For example, we find no statistically significant differences between men and women or between employed and nonemployed persons. Stated otherwise, the decrease in feelings of depression in the summer of 2020 compared with 2017 does not differ substantively by gender or employment status within countries. Age seems to be an exception. In Sweden, the likelihood that adults between the ages of 50 and 69 reported feelings of depression decreased by 22 percentage points between 2017 and 2020, compared with a nonsignificant decrease of 7 percentage points for persons ages 70 and older. We observe a similar trend in France and Poland, although not statistically significant. In Spain, the probability that persons without a partner in the household report feelings of sadness and depression declines by 17 percentage points, compared with virtually no change for respondents with a partner in the household.

Figure 2.

Change in the probability of reporting feelings of depression in 2020 relative to 2017 by country and (a) gender, (b) age, (c) household composition, and (d) employment status.

The estimated average of within-individual change between 2017 and 2020 by (1) chronic illness, (2) reduced labor market attachment, (3) knowing someone with symptoms of COVID-19 or a positive test, and (4) going for a walk less often show that there is little heterogeneity with regard to socioeconomic and demographic characteristics (see Figure 3 and Tables A3a–A3d in the online supplement). In all countries, we found a general trend that respondents recently diagnosed with major health-impairing conditions are more likely to report feelings of depression compared with healthy respondents; moreover, they tend to experience a less marked decline in depression during the pandemic. However, these differences are statistically significant in three countries only: Italy, France, and Switzerland.

Figure 3.

Change in the probability of reporting feelings of depression in 2020 relative to 2017 by country and (a) chronic illness, (b) reduced labor market attachment, (c) knowing someone with symptoms of COVID-19 or a positive test result, and (d) going for a walk less often.

We find few differences on the basis of our other indicators, with the exception of change in labor market attachment in Spain: the probability of reporting feelings of depression increased by 29 percentage points for respondents who experienced losing employment or reduced working hours. This could reflect the particularly tight conditions of the labor market in Spain, where the probabilities to exit unemployment are comparatively more meagre than elsewhere. In all other countries, except for Italy, those with reduced labor force participation reported lower levels of depression, much like their counterparts. Interestingly, this trend can be found with other indicators as well, such as change in income and self-reported ability to make ends meet (see Figures S2 and S3).

When looking closer at the reaction of those persons who had exposure, whether themselves or through others, to the virus, some interesting differences emerge. In Italy, respondents who themselves had symptoms of COVID-19 or tested positive (or know people who did) are less likely to report decreases in feelings of depression. In Germany, we find the opposite, which could be accounted for by its mild first-wave experience, which was even characterized by negative excess mortality (Stang et al. 2020). Finally, we assessed heterogeneity by numerous indicators for sociability, such as visiting with family less often, and physical activity, such as going for walks less often. We find no statistically significant differences across similar indicators (see Figures S6–S8).

Limitations

It is important to note that the data we used were collected in the summer of 2020, after most European countries had lifted lockdowns and other nonpharmaceutical interventions affecting private and public life. Therefore, our results could reflect relief felt by many Europeans following three months of extensive restrictions (see Varga et al. 2021). We were able to assess whether the date of interview increased the change in feelings of depression (see Figure S9). However, we found evidence only in Spain that the decrease in feelings of depression was larger in July and August compared with May and June. Another important caveat about the data involves potential mode effects. With the exception of the Netherlands, all prepandemic interviews were collected face to face, but the 2020 interviews were conducted over the phone. The decrease in feelings of depression could be due to increased desirability bias. However, analyses using the Dutch mode experiment indicate that if our results are biased because of mode effects, we are likely underestimating rather than inflating the decrease in feelings of depression (see Figure S10).

Discussion

Research has shown that the impact of the COVID-19 pandemic on subjective well-being and mental health is more complex than originally expected. Previous findings stress heterogeneity across and within countries in individuals’ self-assessments of how the pandemic affected their well-being. However, methodological differences limit the ability of researchers to assess whether variation across and within countries is substantive or methodological in nature. We assess within-individual change in feelings of depression before (2005, 2007, 2011, 2013, 2015, and 2017) and after the first wave of the COVID-19 pandemic (May to August 2020) on a population of men and women ages 50 and older in 11 European countries. This methodological approach facilitates cross-national comparison and allows us to benchmark change in feelings of depression associated with the pandemic relative to prepandemic change. Despite differences in country exposure and response to the pandemic, we observe an unprecedented decrease in the probability to report feelings of depression across all countries without systematic within-country differences.

Although existing research on the mental health implications of the first wave of the COVID-19 pandemic does not provide a consistent picture, when pre- and postpandemic outbreak longitudinal studies are taken into account, the prevailing evidence points more to stability or even improvement rather than a worsening of psychological well-being. Although other studies have assessed the impact of the pandemic on mental health from a longitudinal perspective with representative single-country samples (see Prati and Mancini 2021 for a review), our study is the first to include both longitudinal measures and a harmonized cross-national research design, relying on a preexisting and large panel study. Its findings of a systematic decrease in feelings of depression following the first wave of the COVID-19 pandemic suggest that older adults across 11 European countries were highly resilient in the face of an unparalleled health crisis. Future research in this area should go beyond indicators of mental health to more encompassing measures of social functioning, which would allow researchers to assess whether the pandemic has changed older adults’ ability for self-care.

It was beyond the scope of our study to adjudicate the mechanisms behind the decline in feelings of depression across 11 European countries. This decline proves true across most social groups and countries. We provide three possible interpretations of these surprising findings based on psychological and sociological theory. First, from a rational-choice perspective, we suggest that self-rated depression is positional in character and relative to the perceived average in society. When the world outside is in a crisis, individuals may consider themselves less depressed. Psychologists have also found that “downward social comparison” (i.e., with individuals and groups faring worse) is particularly likely in situations of crisis (Schwarz and Strack 1999). Second, depression may be buffered by a generalized state of “collective effervescence” in the face of adverse societal shocks (e.g. wars, recessions, or epidemics) (Durkheim 1897). Third, nonpharmaceutical interventions may have triggered a “deceleration” of everyday life—for example, fewer time-pressed obligations—with beneficial mental health outcomes (Rosa 2013).

Our results are based on a set of comparatively affluent European societies, which share well-established public health and welfare institutions. Comparative research covering countries with different levels of economic development and social protection may well reveal that these macro-institutional factors are key in explaining our counterintuitive findings. Our study is not able to predict how the second and third waves of the COVID-19 pandemic in Europe have influenced older adults’ mental health. It is possible that the long-term restrictions put in place in the fall of 2020 extending at least until the spring of 2021 had negative effects on mental health. However, our comparative assessment of how feelings of depression changed across 11 countries for a potentially highly exposed group does not suggest that such a bleak scenario is inevitable, as long as there is some path dependency in subjective well-being. In the immediate aftermath of a sudden and unpredictable health crisis, the psychological resilience of older adults living in established welfare societies proved solid and pervasive.

Supplemental Material

sj-docx-1-srd-10.1177_23780231211032741 – Supplemental material for The Unexpected Decline in Feelings of Depression among Adults Ages 50 and Older in 11 European Countries amid the COVID-19 Pandemic

Supplemental material, sj-docx-1-srd-10.1177_23780231211032741 for The Unexpected Decline in Feelings of Depression among Adults Ages 50 and Older in 11 European Countries amid the COVID-19 Pandemic by Zachary Van Winkle, Emanuele Ferragina and Ettore Recchi in Socius

Footnotes

Data and Materials Availability

The data used in this study, SHARE, may be used exclusively for scientific research, in accordance with European Union and national data protection laws and SHARE conditions of use, free of charge following individual registration at http://www.share-project.org/data-access/user-registration.html. A replication package can be found at .

Supplemental Material

Supplemental material for this article is available online.

ORCID iD

Zachary Van Winkle

1

Note that the term older adult is generally applied to persons ages 65 and older. For brevity, we use older adult to describe our target population of persons ages 50 and older, especially when distinguishing this group from younger and middle-aged adults.

2

Taking into account other natural disasters, there are no more than a dozen studies that estimated within-individual psychological changes before and after the disaster outbreak ().

3

Depression is 1 of 12 items (depressed mood, pessimism, suicidality, guilt, sleep, interest, irritability, appetite, fatigue, concentration, enjoyment, and tearfulness) collected in regular SHARE interviews that constitute the composite Euro-D depression scale for the frequency and severity of depression. Of these 12 items, only depression and trouble sleeping were asked about in the special COVID-19 wave. We chose to analyze depression because it is the most straightforward indicator of mental health.

4

Longitudinal weights for the COVID-19 survey data, which would additionally address selective panel attrition, were not available for calibration at the time the analyses were performed. However, the weighting scheme used accounts for unit nonresponse and attrition errors (see ).

5

After assessing the source and extent of missingness, we opted not to use imputation techniques for two reasons. First, as can be seen in , the amount of missingness is relatively low in waves 1 to 7. Second, our analyses tend to include a maximum of three variables: the dependent variable depression, the independent variable survey wave, and in further analyses a stratification variable. The variable wave is per definition never missing, and missingness on the stratifying variable tends to be similar with missingness of depression because of unit nonresponse. Therefore, we gain little in terms of statistical power or removed bias through imputation techniques, while adding potential biases due to imputation, especially for the dependent variable.

6

We report coefficients in terms of average marginal effects, which can be interpreted as effect sizes as a statistical measure (e.g., Olivier, May, and Bell 2017). In terms of relative effect size or social significance (Bernardi, Chakhaia, and Leopold 2017), we argue that they are considerable for two reasons. First, when benchmarking our estimated effect on previous periods, change between 2017 and 2020 in most countries is at least twice as large as any change previously observed. Second, change between 2017 and 2020 represents one sixth to half of the level of depression in 2017 (see ).

Author Biographies

Zachary Van Winkle is an assistant professor of sociology at Sciences Po, Observatoire sociologique du changement, and an associate member of Nuffield College at the University of Oxford. He has also held positions at the Institute for Social Sciences, Humboldt University Berlin, and the Department of Sociology, University of Oxford. His research focuses on the interplay between family demography and social inequalities from a life course and comparative perspective. In 2021, he is principal investigator of the ANR Résilience COVID-19 project WIDOW19, which seeks to assess the economic consequences of COVID-19-induced widowhood in Europe, as well as the Trilateral Initiatives in Emerging Regions project “A Holistic Approach to Social & Life Course Change in China.” He has published in sociological and demographic journals such as Demography, Social Forces, the European Journal of Population, the Journal of European Social Policy, and Demographic Research.

Emanuele Ferragina is an associate professor of sociology at Sciences Po Paris. Prior to Sciences Po, he was a lecturer at the University of Oxford, where he also received his DPhil. He is interested in the political economy of the welfare state, family policy, and social capital. He has published peer-reviewed articles in political economy (New Political Economy, Review of International Political Economy), social policy (Journal of European Social Policy, Social Policy & Administration, Social Policy & Society, Social Politics), sociology (Research in Social Stratification and Mobility, International Journal of Comparative Sociology, Stato e Mercato, The Tocqueville Review) and political science (Political Studies Review, Policy & Politics) journals.

Ettore Recchi is a professor of sociology at Sciences Po Paris, where he is a member of the Observatoire sociologique du changement and director of the MA and PhD programs in sociology. He is also a part-time professor at the Migration Policy Centre of the European University Institute in Florence and Fellow of Institut Convergences Migrations in Paris. He is a methodologically versatile sociologist especially committed to comparative research; his last book is Everyday Europe: Social Transnationalism in an Unsettled Continent (Policy Press, 2019). In 2021, he is coordinating two research projects: Coping with COVID, on the impact of the pandemic on social life in France, and The Airport Factor, on the effect of air travel on the spread of COVID-19 globally.

References

Aidukaite

Jolanta

Saxonberg

Steven

Szelewa

Dorota

Szikra

Dorottya

. 2021. “Social Policy in the Face of a Global Pandemic: Policy Responses to the COVID-19 Crisis in Central and Eastern Europe.” Social Policy & Administration 55(2):358–73.

Aknin

Lara

De Neve

Jan-Emmanuel

Dunn

Elizabeth

Fancourt

Daisy

Goldberg

Elkhonon

Helliwell

John

Jones

Sarah P.

, et al. 2021. “Mental Health during the First Year of the COVID-19 Pandemic: A Review and Recommendations for Moving Forward.” PsyArXiv. Retrieved July 6, 2021. https://psyarxiv.com/zw93g/.

Beaglehole

Ben

Mulder

Roger T.

Frampton

Chris M.

Boden

Joseph M.

Newton-Howes

Giles

Bell

Caroline J.

2018. “Psychological Distress and Psychiatric Disorder after Natural Disasters: Systematic Review and Meta-Analysis.” British Journal of Psychiatry 213(6):716–22.

Béland

Daniel

Cantillon

Bea

Hick

Rod

Moreira

Amílcar

. 2021. “Social Policy in the Face of a Global Pandemic: Policy Responses to the COVID-19 Crisis.” Social Policy & Administration 55(2):249–60.

Bernardi

Fabrizio

Chakhaia

Lela

Leopold

Liliya

. 2017. “‘Sing Me a Song with Social Significance’: The (Mis)Use of Statistical Significance Testing in European Sociological Research.” European Sociological Review 33(1):1–15.

Börsch-Supan

Axel

Brandt

Martina

Hunkler

Christian

Kneip

Thorsten

Korbmacher

Julie

Malter

Frederic

Schaan

Barbara

, et al. 2013. “Data Resource Profile: The Survey of Health, Ageing and Retirement in Europe (SHARE).” International Journal of Epidemiology 42(4):992–1001.

Cantillon

Bea

Seeleib-Kaiser

Martin

Veen

Romke van der

. 2021. “The COVID-19 Crisis and Policy Responses by Continental European Welfare States.” Social Policy & Administration 55(2):326–38.

Daly

Michael

Sutin

Angelina R.

Robinson

Eric

. Forthcoming. “Longitudinal Changes in Mental Health and the COVID-19 Pandemic: Evidence from the UK Household Longitudinal Study.” Psychological Medicine.

Durkheim

Emile

. 1897. Le Suicide: Étude de Sociologie. Paris: F. Alcan.

10.

Esping-Andersen

Gøsta

. 1990. The Three Worlds of Welfare Capitalism. Cambridge, UK: Polity Press.

11.

Fancourt

Daisy

Steptoe

Andrew

Feifei

. 2021. “Trajectories of Anxiety and Depressive Symptoms during Enforced Isolation Due to COVID-19 in England: A Longitudinal Observational Study.” The Lancet Psychiatry 8(2):141–49.

12.

Ferragina

Emanuele

Seeleib-Kaiser

Martin

. 2011. “Thematic Review: Welfare Regime Debate: Past, Present, Futures?” Policy & Politics 39(4):583–611.

13.

Ferragina

Emanuele

Seeleib-Kaiser

Martin

Tomlinson

Mark

. 2013. “Unemployment Protection and Family Policy at the Turn of the 21st Century: A Dynamic Approach to Welfare Regime Theory.” Social Policy & Administration 47(7):783–805.

14.

Fiorillo

Andrea

Gorwood

Philip

. 2020. “The Consequences of the COVID-19 Pandemic on Mental Health and Implications for Clinical Practice.” European Psychiatry 63(1):e32.

15.

Greve

Bent

Blomquist

Paula

Hvinden

Bjørn

van Gerven

Minna

. 2021. “Nordic Welfare States—Still Standing or Changed by the COVID-19 Crisis?” Social Policy & Administration 55(2):295–311.

16.

Holmes

Emily A.

O’Connor

Rory C.

Perry

V.Hugh

Tracey

Irene

Wessely

Simon

Arseneault

Louise

Ballard

Clive

, et al. 2020. “Multidisciplinary Research Priorities for the COVID-19 Pandemic: A Call for Action for Mental Health Science.” The Lancet Psychiatry 7(6):547–60.

17.

Killgore

William D. S.

Cloonan

Sara A.

Taylor

Emily C.

Dailey

Natalie S.

2020. “Loneliness: A Signature Mental Health Concern in the Era of COVID-19.” Psychiatry Research 290:113117.

18.

Kivi

Marie

Hansson

Isabelle

Bjälkebring

Pär

. 2021. “Up and About: Older Adults’ Well-Being during the COVID-19 Pandemic in a Swedish Longitudinal Study.” Journals of Gerontology: Series B 76(2):e4–e9.

19.

Kwong

Alex S.F.

Pearson

Rebecca M.

Adams

Mark J.

Northstone

Kate

Tilling

Kate

Smith

Daniel

Fawns-Ritchie

Chloe

, et al. 2020. “Mental Health during the COVID-19 Pandemic in Two Longitudinal UK Population Cohorts.” medRxiv. Retrieved July 6, 2021. https://www.medrxiv.org/content/10.1101/2020.06.16.20133116v1.

20.

Mancini

Anthony D.

2019. “When Acute Adversity Improves Psychological Health: A Social–Contextual Framework.” Psychological Review 126(4):486–505.

21.

Moreira

Amílcar

Léon

Margarita

Moscarola

Flavia Coda

Roumpakis

Antonios

. 2021. “In the Eye of the Storm . . . Again! Social Policy Responses to COVID-19 in Southern Europe.” Social Policy & Administration 55(2):339–57.

22.

Niedzwiedz

Claire L.

Green

Michael James

Benzeval

Michaela

Campbell

Desmond

Craig

Peter

Demou

Evangelia

Leyland

Alastair

, et al. 2021. “Mental Health and Health Behaviours before and during the Initial Phase of the COVID-19 Lockdown: Longitudinal Analyses of the UK Household Longitudinal Study.” Journal of Epidemiology and Community Health 75(3):224–31.

23.

Olivier

Jake

May

Warren L.

Bell

Melanie L.

2017. “Relative Effect Sizes for Measures of Risk.” Communications in Statistics—Theory and Methods 46(14):6774–81.

24.

Pan

Kuan-Yu

Kok

Almar A. L.

Eikelenboom

Merijn

Horsfall

Melany

Jörg

Frederike

Luteijn

Rob A.

Rhebergen

Didi

, et al. 2021. “The Mental Health Impact of the COVID-19 Pandemic on People with and without Depressive, Anxiety, or Obsessive-Compulsive Disorders: A Longitudinal Study of Three Dutch Case-Control Cohorts.” The Lancet Psychiatry 8(2):121–29.

25.

Pierce

Matthias

Hope

Holly

Ford

Tamsin

Hatch

Stephani

Hotopf

Matthew

John

Ann

Kontopantelis

Evangelos

, et al. 2020. “Mental Health before and during the COVID-19 Pandemic: A Longitudinal Probability Sample Survey of the UK Population.” The Lancet Psychiatry 7(10):883–92.

26.

Prati

Gabriele

Mancini

Anthony

. 2021. “The Psychological Impact of COVID-19 Pandemic Lockdowns: A Review and Meta-Analysis of Longitudinal Studies and Natural Experiments.” Psychological Medicine 51(2):201–11.

27.

Quarantelli

Enrico L.

1985. “An Assessment of Conflicting Views on Mental Health: The Consequences of Traumatic Events.” Pp. 173–80 in Trauma and Its Wake, Volume I: The Study and Treatment of Post-traumatic Stress Disorder, edited by Figley

Charles R.

New York: Routledge.

28.

Recchi

Ettore

Ferragina

Emanuele

Helmeid

Emily

Pauly

Stefan

Safi

Mirna

Sauger

Nicolas

Schradie

Jen

. 2020. “The ‘Eye of the Hurricane’ Paradox: An Unexpected and Unequal Rise of Well-Being during the Covid-19 Lockdown in France.” Research in Social Stratification and Mobility 68:100508.

29.

Rosa

Hartmut

. 2013. Social Acceleration: A New Theory of Modernity. New York: Columbia University Press.

30.

Schwarz

Norbert

Strack

Fritz

. 1999. “Reports of Subjective Well-Being: Judgmental Processes and Their Methodological Implications.” Pp. 61–84 in Well-Being: The Foundations of Hedonic Psychology, edited by Kahneman

Diener

Schwarz

New York: Russell Sage.

31.

Stang

Andreas

Standl

Fabian

Kowall

Bernd

Brune

Bastian

Böttcher

Juliane

Brinkmann

Marcus

Dittmer

Ulf

, et al. 2020. “Excess Mortality Due to COVID-19 in Germany.” Journal of Infection 81(5):797–801.

32.

Stuck

Stephanie

Zuber

Sabrina

Franzese

Fabio

Gruber

Stefan

Birkenbach

Tim

Plfueger

Senta-Melissa

Atzendorf

Josefine

. 2020. “SHARE COVID-19 Survey—Release Guide.” MEA. Retrieved July 6, 2021. http://www.share-project.org/fileadmin/pdf_documentation/SHARE_COVID19_Survey1_Release_Guide.pdf.

33.

Uchida

Yukiko

Takahashi

Yoshiaki

Kawahara

Kentaro

. 2014. “Changes in Hedonic and Eudaimonic Well-Being after a Severe Nationwide Disaster: The Case of the Great East Japan Earthquake.” Journal of Happiness Studies 15(1):207–21.

34.

van der Velden

Peter G.

Contino

Carlo

Das

Marcel

van Loon

Peter

Bosmans

Mark W. G.

2020. “Anxiety and Depression Symptoms, and Lack of Emotional Support among the General Population before and during the COVID-19 Pandemic: A Prospective National Study on Prevalence and Risk Factors.” Journal of Affective Disorders 277:540–48.

35.

Varga

Tibor V.

Feifei

Dissing

Agnete S.

Elsenburg

Leonie K.

Herranz Bustamante

Joel J.

Matta

Joane

van Zon

Sander K.R.

, et al. 2021. “Loneliness, Worries, Anxiety, and Precautionary Behaviours in Response to the COVID-19 Pandemic: A Longitudinal Analysis of 200,000 Western and Northern Europeans.” The Lancet Regional Health – Europe 2:100020.

36.

Yee

Julie

Niemeier

Debbie

. 1996. “Advantages and Disadvantages: Longitudinal vs. Repeated Cross-Section Surveys.” Project Battelle 94-16, FHWA, HPM-40. Retrieved July 6, 2021. https://rosap.ntl.bts.gov/view/dot/13793/dot_13793_DS1.pdf.

Supplementary Material

Please find the following supplemental material available below.

For Open Access articles published under a Creative Commons License, all supplemental material carries the same license as the article it is associated with.

For non-Open Access articles published, all supplemental material carries a non-exclusive license, and permission requests for re-use of supplemental material or any part of supplemental material shall be sent directly to the copyright owner as specified in the copyright notice associated with the article.

0.00 MB

1.13 MB