key: cord-0728165-se1wmdt4
authors: Keller, Amélie; Groot, Jonathan; Matta, Joane; Bu, Feifei; El Aarbaoui, Tarik; Melchior, Maria; Fancourt, Daisy; Zins, Marie; Goldberg, Marcel; Nybo Andersen, Anne-Marie; Rod, Naja H.; Strandberg-Larsen, Katrine; Varga, Tibor V.
title: Housing environment and mental health of Europeans during the COVID-19 pandemic: a cross-country comparison
date: 2022-04-04
journal: Sci Rep
DOI: 10.1038/s41598-022-09316-4
sha: fc14d85ae88828cce43a7c307eb2279a53e905cd
doc_id: 728165
cord_uid: se1wmdt4

Many studies have investigated the impact of the COVID-19 pandemic on mental health. Throughout the pandemic, time spent at home increased to a great extent due to restrictive measures. Here we set out to investigate the relationship between housing conditions and the mental health of populations across European countries. We analyzed survey data collected during spring 2020 from 69,136 individuals from four cohorts from Denmark, France, and the UK. The investigated housing conditions included household density, composition, and crowding, access to outdoor facilities, dwelling type, and urbanicity. The outcomes were loneliness, anxiety, and life satisfaction. Logistic regression models were used, and results were pooled using random-effects meta-analysis. In the meta-analysis, living alone was associated with higher levels of loneliness (OR = 3.08, 95% CI 1.87–5.07), and lower life satisfaction (OR = 1.27, 95% CI 1.05–0.55), compared to living with others. Not having access to an outdoor space and household crowding were suggestively associated with worse outcomes. Living in crowded households, living alone, or lacking access to outdoor facilities may be particularly important in contributing to poor mental health during a lockdown. Addressing the observed fundamental issues related to housing conditions within society will likely have positive effects in reducing social inequalities, as well as improving preparedness for future pandemics.

Participating cohorts. Four cohorts were included in this study: The Danish National Birth Cohort (DNBC, Denmark; n = 21,889) 27 , Constances (France, n = 28,171) 28 , TEMPO (France, n = 424) 29 20 . The participating cohorts have collected data on housing environment and mental health outcomes and are part of the international COVID-MINDS network [https:// www. covid minds. org/]. The participating cohorts, relevant ethical considerations, and a sample size flowchart for all cohorts, are presented in Supplementary Text S1. The four cohorts were divided into the following age and gender categories: (1) Young people 18-25 years; (2) Women above 25 years; (3) Men above 25 years. Young people were considered a separate category, as young adults between ages 18 and 25 have been consistently reported to be especially vulnerable and susceptible to symptoms of poor mental health during a lockdown 30 . This category was not further stratified by gender due to sample size considerations. While some cohort surveys provided options for self-reported gender other than men or women (e.g. other or prefer not to say), the resulting groups were too small to perform analysis on. The UCL COVID-19 cohort included all three populations, while only young people and women > 25 years were included in the DNBC, and only women and men > 25 years were included in the TEMPO and Constances cohorts. All four cohorts are longitudinal; however, for this analysis, only data from the Spring 2020 lockdown period were used (March 30-April 5, 2020, for the DNBC, April 6-June 15 for Constances, April 6-April 12, for TEMPO, and April 6-April 12 for UCL COVID-19) (Fig. 1 ). The combined sample size for our analyses was 69,136 individuals across the four cohorts.

Housing conditions. Access to outdoor facilities such as a garden, balcony, or yard was dichotomized as having access to any outdoor space (Yes or No) . Household density was defined as the dwelling size (m 2 floor area) divided by the number of persons living in the dwelling. The average floor area per person in the EU is 42.6 m 2 per person 31 ; hence the variable was dichotomized as < 43 m 2 or ≥ 43 m 2 per person. Household crowding 32, 33 was adapted from the Eurostat criteria 34 with alterations due to data constraints. A minimal number of rooms for a given household was calculated as follows: (i) one room for the household; (ii) one room per couple in the household; (iii) one room for each adult single person; (iv) ½ room for each child. Household crowding was categorized as reference (minimal or minimal + 1), crowded (< minimal), and under-occupied (> minimal + 1). Based on previous literature 33 , household composition 35 was categorized as households with children, adult-only www.nature.com/scientificreports/ households, and single-person households. Dwelling type was categorized as apartment versus house. In the UCL COVID-19 cohort, urbanicity was coded as urban (cities), semi-urban (large/small towns), and rural (village, hamlets, or other isolated dwellings) based on self-reported postal codes and EU standards of area of living 36 .

In the Constances cohort, urbanicity was categorized as rural versus urban. Some of the exposures were ascertained differently across cohorts (e.g. see urbanicity definitions above); data harmonization was performed by re-categorizing survey responses to optimize the comparability of data from the four datasets. The UCLA short three-item T-ILS version loneliness scale was used in the DNBC, TEMPO, and UCL COVID-19 to assess the degree of loneliness of participants. The total score ranges from 3 to 9. Scores were dichotomized, and severe loneliness was defined as scores 7-9 37 . In Constances, loneliness was measured by the question "How often have you felt lonely during the past week?", collected on a 1-4 Likert scale from "Rarely or None of the Time (Less than 1 Day)" to "Most or All of the Time (5-7 Days)". Scores were dichotomized, and severe loneliness was defined as score = 4. Anxiety was measured using the GAD-7 scale 38 , ranging 0-21, in Constances and UCL COVID-19. Scores were dichotomized, and severe anxiety was defined as scores 15-21. In the DNBC, the SCL-ANX4 39 was used. Answers to five questions are collected on a Likert scales ranging 0-4 and are then dichotomized to 0 (not at all) or 1 (a little-extremely) and summed for each subscale. Thus, the summed score for anxiety ranges from 0-4. Severe anxiety was defined as ≥ 3 as suggested by the authors of the CMDQ 39, 40 .

The ONS Life Satisfaction scale was used to measure life satisfaction in the UCL COVID-19 cohort, while the Cantril Self-Anchoring Striving Scale was used in the rest of the cohorts. Both scales range between 0 (not at all satisfied with life) and 10 (completely satisfied with life). Among young people, the scale was categorized as low (0-5), medium (6) (7) (8) , and high (9-10) life satisfaction according to cut-offs used in previous literature 41, 42 . Among women and men > 25 years, the scale was categorized as low (0-4), medium (5) (6) , and high (7) (8) (9) (10) life satisfaction according to cut-offs used in previous surveys 43 . In line with our other analysis where severe loneliness and anxiety were used as outcomes, we dichotomized life satisfaction as low vs. medium or high life satisfaction. Thus, for all three outcomes, odds ratios above 1 infer that the exposure is associated with a poorer outcome (higher loneliness and anxiety, and lower life satisfaction).

Statistical analysis. Statistical analysis was performed using STATA for the DNBC and UCL COVID-19, SAS for Constances, and R for TEMPO.

Weighting was performed using the entropy balancing method 44 for UCL COVID-19, and marginal calibration weighting 45 for Constances to achieve more representative population-based samples. No weighting was performed for DNBC and TEMPO (Supplementary Text S1).

To describe the cohorts, means and standard deviations were used for normally distributed continuous variables, and medians and interquartile ranges (IQR) were used for non-normally distributed continuous variables. Frequencies and percentages were used for categorical variables.

Outcomes were used as binary variables, and binary logistic regression analyses were performed for each exposure and outcome combination within each cohort. Odds ratios (OR) and 95% confidence intervals (CI) are reported. Missing data (19.7%) was imputed using proc mi (SAS) in the Constances cohort. In all other cohorts, complete case analysis was performed, excluding individuals with information missing on any covariate (19% for DNBC, 43.5% for TEMPO, and 53% for UCL COVID-19). The analyses were performed separately for young people, women > 25 years, and men > 25 years. Confounders were identified using directed acyclic graphs (DAGs) and previous literature. Four models were tested:

(i) Unadjusted model: testing the association between exposure (access to outdoor facilities, household density, crowding, and composition, dwelling type, urbanicity) and outcome (high levels of loneliness, anxiety, and low levels of life satisfaction); (ii) Model 1: Unadjusted model + age + educational attainment + sex (in the models for young people). Results from the cohorts for each exposure-outcome combination using Model 3 were pooled using randomeffects meta-analysis (metafor package in R), as heterogeneity was expected between the cohorts due to geographical differences, varying governmental restrictions, lockdown severities, and varying exposure and outcome ascertainment. Subgroup analyses by gender and age categories were undertaken. Subgroups analyses by country were also evaluated to investigate cross-country differences.

Ethics statement. Ethics 

Descriptive statistics of the four cohorts can be found in Supplementary Table S1 . All association results from the cohorts using the unadjusted model, Model 1, Model 2, and Model 3 (the fully adjusted model) are presented in Supplementary Table S2, Supplementary Table S3, Supplementary Table S4, and Supplementary Table S5 (Fig. 5) . While the association between living alone and loneliness was the association of the greatest magnitude that we observed in this study, no associations were detected between living alone and anxiety (OR = 0.86, 95% CI 0.59-1.25). Living with children was associated with higher levels of anxiety amongst women Association between access to outdoor space (yes vs. no) and high levels of anxiety (N = 68,712). The cohort-specific estimates and 95% confidence intervals were obtained using binary logistic regression models. Age, educational attainment, gender, previously diagnosed chronic disease status and presence of previously diagnosed mental illness were used a covariates. In addition, the models included mutual adjustment for the other housing exposures. 95% CI 95% confidence interval, OR odds ratio. N = 18,652) . The cohort-specific estimates and 95% confidence intervals were obtained using binary logistic regression models. Age, educational attainment, gender, previously diagnosed chronic disease status and presence of previously diagnosed mental illness were used a covariates. In addition, the models included mutual adjustment for the other housing exposures. 95% CI 95% confidence interval, OR odds ratio. The cohort-specific estimates and 95% confidence intervals were obtained using binary logistic regression models. Age, educational attainment, gender, previously diagnosed chronic disease status and presence of previously diagnosed mental illness were used a covariates. In addition, the models included mutual adjustment for the other housing exposures. 95% CI 95% confidence interval, OR odds ratio. 

We show results from an analysis of ~ 70,000 individuals from three European countries with available housing environment and mental health data during the first lockdown of the COVID-19 pandemic during spring 2020.

As previous research shows that Europeans spend most of their life indoors 13 , which was the case to an even greater extent during lockdowns with curfews and stay at home recommendations, we set out to quantify the associations between various facets of the housing environment and the mental health and wellbeing of Europeans during the first severe lockdown of the pandemic.

Our key findings are as follows: (1) Living alone was consistently related to worse mental health during lockdown; (2) living in dwellings offering more rooms for a given household was related to better mental health, albeit with some differences by country and age groups; (3) having access to outdoor space appeared to be associated N = 40,965) . The cohort-specific estimates and 95% confidence intervals were obtained using binary logistic regression models. Age, educational attainment, gender, previously diagnosed chronic disease status and presence of previously diagnosed mental illness were used a covariates. In addition, the models included mutual adjustment for the other housing exposures. 95% CI 95% confidence interval, OR odds ratio. www.nature.com/scientificreports/ with lower levels of anxiety; (4) urbanicity and dwelling types were not associated with mental health status; (5) women's mental health seemed to be more strongly associated with household composition than men's. Our results regarding poorer mental health, especially loneliness, amongst people living alone during the lockdown are consistent with other studies 20, 46 . As suggested in the literature, lockdown situations heighten already existing risk factors of loneliness, such as living alone 20 . Therefore, as both social isolation and loneliness have been shown to be associated with increased mortality, impaired quality of life, and wellbeing 47 , people living alone might benefit from community support during lockdowns. With the rise in the proportion of people living alone across age-groups 48 , we also recommend piloting screening methods to identify the most vulnerable individuals in communities (e.g. those living alone) for the prioritization of additional support.

Our results indicate that living in under-occupied dwellings is associated with better mental health outcomes: lower levels of loneliness and anxiety, and higher life satisfaction. Another finding is that living in less dense environments (having access to more square meters per person) is associated with worse outcomes amongst adults > 25, but better outcomes in young adults < 25. While these findings seem counterintuitive, they appear to be largely driven by results from the French Constances cohort, where (1) both measures of household density and crowding were simultaneously available, so we were able to capture the difference between these determinants; (2) both low household density (vs. high household density) and living in more crowded dwellings (vs. ideal levels) were associated with worse outcomes, e.g. higher levels of loneliness. As Constances was the only cohort where both household density and crowding were measured, we were unable to verify these findings in other cohorts. However, as emphasized above, overall results from the meta-analyses show that living in under-occupied dwellings is generally associated with better mental health. These results are supported by a 2018 systematic review on household crowding 49 , which concluded that crowding may be associated with poor mental health. During a lockdown, it has been reported that people living in overcrowded dwellings were more likely to report poor mental health than people living in non-overcrowded housing 50 .

Our results suggest that lacking access to outdoor spaces is associated with higher levels of anxiety, but not with loneliness or quality of life. Evidence from previous studies suggests that access to outdoor green and blue spaces is beneficial for mental health 51, 52 . During lockdowns, several studies also reported that access to natural green spaces was associated with better mental health 53, 54 . Although having access to natural green spaces might be most beneficial for mental health 55 , this study suggests that having access to any type of outdoor spaces (e.g. balcony, garden, or yard) compared to none, may lower levels of anxiety during a lockdown. Consistently, results from our previous study among young people in Denmark have shown that individuals without access to outdoor spaces experienced greater decline in well-being during the COVID-19 pandemic 21 . Therefore, housing and urban design strategies should ensure access to outdoor spaces for all 56 .

In this study, no differences in mental health indicators were found by urbanicity or dwelling type. The relationship between urbanicity and mental health pre-pandemic is contradictory, with some reports showing a clear benefit of rural environments 57 , but some showing no difference between urban and rural environments 58 . However, recent evidence pointed towards a higher prevalence of poor mental health indicators, such as anxiety disorders and PTSD, in urban environments 59 . In line with this review, recent cross-cohort analyses from the UK reported that urban residents had a higher risk of being lonely both before and during the pandemic 20 . Similarly, it was reported in a Scottish cross-sectional study that urban residents were more psychologically distressed during the COVID-19 pandemic compared to rural residents 60 . The discrepancies between our results and the UK studies might have arisen due to differences in urbanicity classification. It may also be that our definition was too broad to be sensitive to certain aspects of urbanicity that would truly matter during lockdowns, such as the urban physical (e.g. noise pollution) and social (e.g. social cohesion) environments 59 . In line with these considerations, we expect a greater proportion of dwellings with direct access to outdoor space to be rural or semi-urban, thus, mutual adjustment for these factors is expected to result in attenuated estimates relative to urbanicity, which we do observe for some of our results.

Previous studies have shown that women, young people, and those with young children experienced the greatest increase in mental distress during the COVID-19 pandemic 4, 61 . In this study, women living in households with children reported higher levels of anxiety compared to men and younger adults living with children. This result might be explained by a salient aspect of the COVID-19 crisis, which involved large-scale closures of daycare centers and schools, resulting in children staying at home and needing to be cared for and home-schooled 62 . Previous studies have shown that parents experienced high levels of stress, anxiety and depressive symptoms due to homeschooling during COVID-19 lockdowns [63] [64] [65] . As, globally, women are still responsible for most of the unpaid care work, such as raising children 66 , we recommend the prioritization of the provision of child care, especially for young children, as well as flexible working hours and paid sick leave for parents during lockdowns.

The collected data make it possible to compare associations between housing environment exposures and mental health outcomes across three European countries. The Oxford COVID-19: Government Response Tracker was developed to rank a range of governmental policies in response to the pandemic on a score from 0 to 100 according to their stringency. According to this stringency metric, France had the most stringent governmental response with a score of 88 out of 100 compared to 80 in the UK and 72 in Denmark three weeks into the respective national lockdowns in the spring of 2020 67 . France and the UK likely had higher scores as both countries implemented curfews during this period, while more freedom of movement was allowed in Denmark. Thus, we expected that associations between certain characteristics of the household environment and negative mental health outcomes would be starker in France and the UK compared to Denmark. Our findings appear to be partly in line with this expectation, as a range of associations were most apparent in France and were likely driving some of the overall findings. The lack of differences in some of the examined associations can probably be attributed to the fact that the range of stringency levels in the three countries were still relatively close to each other; future comparisons with countries implementing less strict governmental responses-e.g. Sweden 68 -are warranted to explore the extent to which governmental stringency is relevant to the examined associations. www.nature.com/scientificreports/ The importance of our results is threefold. First, our study can inform policy makers, authorities, and other stakeholders about the associations between various facets of the housing environment and mental health indicators in a period when citizens are encouraged or mandated to stay at home. Second, this study demonstrates that some living conditions (e.g. living alone, crowded households, lacking access to outdoor spaces) may be especially important to certain aspects of mental health during a lockdown; the identification of these factors might help to develop interventions, novel community-based support mechanisms and screening procedures to help those most affected by lockdowns. Third, while many of the reported inequalities have been observed before, it is likely that the COVID-19 pandemic and its lockdowns exacerbated some of these already existing associations. In the long term, this study might inform policy-makers and urban planners on these aspects with potential indications of specific requirements for different residents.

Important strengths of this study are its large sample size, the utilization of four cohorts from three European countries, and the availability of detailed harmonized information of housing environments, related factors and mental health outcomes during the first lockdown of the COVID-19 pandemic. Our sample includes individuals across a wide age range and from three European countries, enabling cross-country as well as sex-and age-specific comparisons.

The most important limitations of our study are the self-reported data on housing conditions and mental health outcomes, and that we did not have access to pre-pandemic data to undertake comparisons. Self-reported data are prone to various biases, and it is possible that these biases impacted our analysis. As complete-case analysis was used in most cohorts, missing data on exposures, outcomes or covariates might have also impacted our findings. This might be especially important regarding the mental health outcomes, where we expect those with the worst mental health to decide not to respond or underreport certain exposures; such biases would most likely attenuate findings. Another key limitation is that this study only included data from three Western and Northern European countries, and thus, our findings might not generalize to other settings. We consider crosscountry comparisons suggestive and explorative; as results generally emanate from a single large cohort for all participating countries, it is hard to deduce whether larger, more comprehensive, and nationally representative samples would have produced different results. While weighting strategies were utilized in Constances and UCL COVID-19, where this was possible, TEMPO and DNBC were not designed to be representative of the general population. Last, in addition to our limitations regarding geographical generalizability, the cross-sectional nature of the data also limits our ability to generalize over time. Our results are based on self-reported data, which were collected at the beginning of the COVID-19 pandemic (spring 2020), at a time when most individuals likely experienced dynamic changes in their mental health in response to changing biological, societal, and political realities 6, 30 .

In conclusion, we show that while urbanicity and household dwelling type was not associated with mental health, the other examined housing conditions, namely household composition, access to outdoor facilities, and household crowding and density, were associated with various aspects of mental health in three European countries during the first strict lockdowns of the COVID-19 pandemic, in spring 2020. Living alone, especially, was associated with severe loneliness and lower life satisfaction, but not with anxiety, while living with children was associated with higher levels of anxiety amongst women. These results pinpoint population groups that might need targeted interventions to ameliorate the negative mental health impact of the COVID-19 pandemic and its associated lockdowns.

Due to ethical reasons, investigators are not able to share individual level data. Summary data are shared in the Supplemental Material. www.nature.com/scientificreports/

Prevalence of stress, anxiety, depression among the general population during the COVID-19 pandemic: A systematic review and meta-analysis

2019-nCoV epidemic: Address mental health care to empower society

The consequences of the COVID-19 pandemic on mental health and implications for clinical practice

Mental health before and during the COVID-19 pandemic: A longitudinal probability sample survey of the UK population

Mental health of French students during the Covid-19 pandemic

Standing together -at a distance": Documenting changes in mental-health indicators in Denmark during the COVID-19 pandemic

Time trends in mental health indicators during the initial 16 months of the COVID-19 pandemic in Denmark

Trajectories of anxiety and depressive symptoms during enforced isolation due to COVID-19 in England: A longitudinal observational study

Risk of stress/depression and functional impairment in Denmark immediately following a COVID-19 shutdown

Depression and anxiety symptoms in pregnant women in Denmark during COVID-19

Comparison of mental health symptoms prior to and during COVID-19: Evidence from a living systematic review and meta-analysis

Estimating the effects of non-pharmaceutical interventions on COVID-19 in Europe

Indoor Air Pollution: New EU Research Reveals Higher Risks than Previously Thought

Housing and health inequalities: A synthesis of systematic reviews of interventions aimed at different pathways linking housing and health

Housing and health: Time again for public health action

Protective elements of mental health status during the COVID-19 outbreak in the Portuguese population

Psychological impact of the COVID-19 outbreak on students in a French region severely affected by the disease: Results of the PIMS-CoV 19 study

Who maintains good mental health in a locked-down country? A French nationwide online survey of 11,391 participants

Covid-19 lockdown: Housing built environment's effects on mental health

Who is lonely in lockdown? Cross-cohort analyses of predictors of loneliness before and during the COVID-19 pandemic

Impact of housing conditions on changes in youth's mental health following the initial national COVID-19 lockdown: A cohort study

Traditional reviews, meta-analyses and pooled analyses in epidemiology

Methods for pooled analyses of epidemiologic studies

Comparison of effect estimates from a meta-analysis of summary data from published studies and from a metaanalysis using individual patient data for ovarian cancer studies

Meta-analysis using individual participant data: One-stage and two-stage approaches, and why they may differ

The Strengthening the Reporting of Observational Studies in Epidemiology (STROBE) statement: Guidelines for reporting observational studies

The Danish National Birth Cohort-its background, structure and aim

The French CONSTANCES population-based cohort: design, inclusion and follow-up

Tobacco, alcohol, cannabis and other illegal drug use among young adults: The socioeconomic context

Loneliness, worries, anxiety, and precautionary behaviours in response to the COVID-19 pandemic: A longitudinal analysis of 200,000 Western and Northern Europeans

Housing Space per Person

Definitions of crowding and the effects of crowding on health

The effect of household crowding and composition on health in an Inuit cohort in Greenland

Is Your Home

A short scale for measuring loneliness in large surveys: Results from two population-based studies

A brief measure for assessing generalized anxiety disorder: the GAD-7

A brief case-finding questionnaire for common mental disorders: the CMDQ

Screening instruments for anxiety and depression in patients with irritable bowel syndrome are ambiguous

Trends in high life satisfaction among adolescents in five Nordic countries 2002-2014

What does the Cantril Ladder measure in adolescence?

What makes for a better life? The determinants of subjective well-being in OECD countries-Evidence from the Gallup World Poll

Entropy balancing for causal effects: A multivariate reweighting method to produce balanced samples in observational studies

Calibration estimators in survey sampling

Loneliness in the UK during the COVID-19 pandemic: Cross-sectional results from the COVID-19 Psychological Wellbeing Study

Social isolation, loneliness, and all-cause mortality in older men and women

Pathways into living alone in mid-life: Diversity and policy implications

WHO Housing and health guidelines: Web annex A: Report of the systematic review on the effect of household crowding on health

National Housing Federation. Housing Issues During Lockdown: Health, Space and Overcrowding

Long-term exposure to residential green and blue spaces and anxiety and depression in adults: A cross-sectional study

Momentary mood response to natural outdoor environments in four European cities

Contact with blue-green spaces during the COVID-19 pandemic lockdown beneficial for mental health

Does greenery experienced indoors and outdoors provide an escape and support mental health during the COVID-19 quarantine?

Effect of nature exposure on perceived and physiologic stress: A systematic review. Complement

Built and natural environment planning principles for promoting health: An umbrella review

Cities and mental health

Are mental disorders more common in urban than rural areas of the United States

Current evidence on urbanicity and the impact of neighbourhoods on anxiety and stress-related disorders

Are rurality, area deprivation, access to outside space, and green space associated with mental health during the covid-19 pandemic? A cross sectional study (charis-e)

The effect of age, gender, income, work, and physical activity on mental health during coronavirus disease (COVID-19) lockdown in Austria

The Impact of COVID-19 on Gender Equality

Parental experiences of homeschooling during the COVID-19 pandemic: Differences between seven European countries and between children with and without mental health conditions

Child and family outcomes following pandemics: A systematic review and recommendations on COVID-19 policies

Families in the COVID-19 pandemic: parental stress, parent mental health and the occurrence of adverse childhood experiences-Results of a representative survey in Germany

The COVID-19 pandemic has increased the care burden of women and families

Oxford COVID-19 government response tracker

Nordic responses to Covid-19: Governance and policy measures in the early phases of the pandemic

Funding This study was made possible by a grant from the RealDania Foundation (PRJ-2019-00020 'Indoor environment and child health…'). The DNBC was established with a significant grant from the Danish National Research Foundation. Additional support was obtained from the Danish Regional Committees, the Pharmacy Foundation, the Egmont Foundation, the March of Dimes Birth Defects Foundation, the Health Foundation and other minor grants. Follow-up of mothers and children has been supported by the Danish Medical Research Council (SSVF 0646

the French Inter-departmental Mission for the Fight against Drugs and Drug Addiction (MILDeCA); the French Institute of Cancer

and the Pfizer Foundation. The CONSTANCES COVID-19 Study was funded by: ANR (Agence Nationale de la Recherche

Infrastructure nationale en Biologie et Santé" and benefits from a grant from the French National Agency for Research (ANR-11-INBS-0002). CONSTANCES is also partly funded by Merck Sharp & Dohme (MSD), AstraZeneca, Lundbeck and L'Oréal. None of these funding sources had any role in the design of the study, collection and analysis of data or decision to publish. The Covid-19 Social Study was funded by the Nuffield Foundation [WEL/FR-000022583], but the views expressed are those of the authors and not necessarily the Foundation. The study was also supported by the MARCH Mental Health Network funded by the Cross-Disciplinary Mental Health Network Plus initiative supported by UK Research and Innovation

The authors thank the INSERM-Versailles Saint Quentin en Yvelines University "Population-based Epidemiologic Cohorts Unit" (Cohortes épidémiologiques en population) which designed and manages the Constances Cohort Study. They also thank the National Health Insurance Fund ("Caisse nationale d'assurance maladie des travailleurs salaries," CNAMTS) and its Health Screening Centres ("Centres d' examens de santé"), which are collecting a large part of the data, as well as the National Old-Age Insurance Fund (Caisse nationale d'assurance vieillesse) for its contribution to the constitution of the cohort, ClinSearch, Asqualab and Eurocell, which are conducting the data quality control.

The authors declare no competing interests.

Supplementary Information The online version contains supplementary material available at https:// doi. org/ 10. 1038/ s41598-022-09316-4.Correspondence and requests for materials should be addressed to T.V.V.Reprints and permissions information is available at www.nature.com/reprints.Publisher's note Springer Nature remains neutral with regard to jurisdictional claims in published maps and institutional affiliations.Open Access This article is licensed under a Creative Commons Attribution 4.0 International License, which permits use, sharing, adaptation, distribution and reproduction in any medium or format, as long as you give appropriate credit to the original author(s) and the source, provide a link to the Creative Commons licence, and indicate if changes were made. The images or other third party material in this article are included in the article's Creative Commons licence, unless indicated otherwise in a credit line to the material. If material is not included in the article's Creative Commons licence and your intended use is not permitted by statutory regulation or exceeds the permitted use, you will need to obtain permission directly from the copyright holder. To view a copy of this licence, visit http:// creat iveco mmons. org/ licen ses/ by/4. 0/.