key: cord-1012401-iz697g8s
authors: Gouvernet, Brice; Bonierbale, Mireille
title: Bio-psychosocial study on the impact of the COVID-19 lockdown on depression and anxiety in a sample of 1753 French-speaking subjects
date: 2021-04-01
journal: Encephale
DOI: 10.1016/j.encep.2021.03.002
sha: ebde9862f67597a83e819e7ce732877f6e847ec2
doc_id: 1012401
cord_uid: iz697g8s

The aim of this article was to study the impact of the COVID19 lockdown on anxiety and depressive symptoms on the basis of responses to an online survey from 1753 French-speaking subjects, conducted between April 27 and May 11, 2020. Method: Using a biopsychosocial model, the effects of socio-demographic characteristics (age, gender at birth, socio-professional category, sexual orientation), lockdown conditions (material factors: urban density of the place of residence, surface area of the place of residence during lockdown), social characteristics: living with a partner during lockdown, presence of children during lockdown) and psychosocial history (attachment styles) on anxiety - evaluated on the GAD7 - and depression - evaluated on the MDI - were investigated. Ordinal regression analyses were conducted. Results: The rates of depression observed (moderate or severe depression: 22.5%) and anxiety (moderate or severe anxiety: 18.4%) were higher than usual but lower than what has been documented in other studies on the effects of lockdown. Women appeared to be more vulnerable than men (Anxiety: AOR = 1.647, CI95 = 1.647 - 2.530; Depression: AOR = 1.622, CI95 = 1.274 - 2.072). Bisexual individuals had an increased likelihood of anxiety symptoms (AOR = 1.962, CI95 = 1.544 - 2.490) and depression (AOR = 1.799, CI95 = 1.394 - 2.317). For homosexuals, only links with depression were observed (AOR = 1.757, CI95 = 1.039 - 2.906). People in a situation of economic vulnerability were more prone to anxiety disorders (e.g. people with no working activity: AOR = 1.791, CI95 = 1.147 - 2.790) or depression (e.g. people with no working activity: AOR = 2.581, CI95 = 1.633 - 4.057). Links with attachment styles were also found. Fearful subjects were particularly vulnerable (anxiety: AOR = 2.514, CI95 = 1.985 - 3.190; depression: AOR = 2.521, CI95 = 1.938 - 3.289), followed by subjects with an anxious attachment style (anxiety: AOR = 1.949, CI95 = 1.498 - 2.540; depression: AOR = 1.623, CI95 = 1.207 - 2.181). The impact of lockdown on avoiding subjects only concerned depression (AOR = 1.417, CI95 = 1.034 - 1.937). Being with a partner during lockdown appeared to have a protective effect against depression (AOR = .693, CI95=.555 - .866). Neither the presence of children, the surface area of the lockdown residence, nor the population density of the place of residence were associated with anxiety or depression. Conclusion: The impact of lockdown on mental health depends on a range of dimensions that need to be apprehended in order to tailor post-lockdown psychological and social support. Management based on a biopsychosocial approach should be favored.

= 1.633 -4.057). Links with attachment styles were also found. Fearful subjects were particularly vulnerable (anxiety: AOR = 2.514, CI95 = 1.985 - 3.190 ; depression: AOR = 2.521, CI95 = 1.938 -3.289), followed by subjects with an anxious attachment style (anxiety: AOR = 1.949, CI95 = 1.498 -2.540; depression: AOR = 1.623, CI95 = 1.207 -2.181). The impact of lockdown on avoiding subjects only concerned depression (AOR = 1.417, CI95 = 1.034 -1.937). Being with a partner during lockdown appeared to have a protective effect against depression (AOR = .693, CI95=.555 -.866). Neither the presence of children, the surface area of the lockdown residence, nor the population density of the place of residence were associated with anxiety or depression. Conclusion: The impact of lockdown on mental health depends on a range of dimensions that need to be apprehended in order to tailor post-lockdown psychological and social support. Management based on a biopsychosocial approach should be favored.

Keywords: Covid19 Lockdown; anxiety; depression; forms of attachment Résumé L'objet de cet article est d'étudier l'impact du confinement COVID19 sur les symptomatologies anxieuse et dépressive à partir des réponses à une enquête en ligne menée auprès de 1753 sujets francophones entre le 27 avril et le 11 mai 2020. Méthode : Partant d'un modèle biopsychosocial, l'effets des caractéristiques sociodémographiques (âge, état-civil à la naissance, CSP, orientation sexuelle), des modalités de confinement (matérielles : densité urbaine du lieu de résidence, superficie du lieu de confinement, sociales : confinement en couple, présence d'enfants pendant le confinement) et des antécédents psychosociaux (styles d'attachement) sur l'anxiété -évaluée par la GAD7 -et la dépression -évaluée par la MDI -sont investiguées. Des analyses de régression ordinale ont été conduites. Résultats : Les taux de dépression (dépression modérée ou sévère : 22,5 %) et d'anxiété (anxiété modérée ou sévère : 18,4 %) sont plus élevés que d'habitude mais inférieurs à ce qui a été documenté dans d'autres études sur les effets du confinement. Les femmes semblent être plus vulnérables que les hommes (Anxiété : AOR = 1,647, IC95 = 1,647 -2,530 ; Dépression : AOR = 1,622, IC95 = 1,274 -2,072). Les personnes bisexuelles ont une probabilité accrue de symptômes d'anxiété (AOR = 1,962, IC95 = 1,544 -2,490) et de dépression (AOR = 1,799, IC95 = 1,394 -2,317). Pour les homosexuels, seuls des liens avec la dépression ont été observés (AOR = 1,757, IC95 = 1,039 -2,906). Les personnes en situation de vulnérabilité économique sont plus sujettes aux troubles anxieux (par exemple les personnes sans activité professionnelle : AOR = 1,791, IC95 = 1,147 -2,790) ou à la dépression (par exemple les personnes sans activité professionnelle : AOR = 2,581, IC95 = 1,633 -4,057). Des liens avec les styles d'attachement sont également trouvés. Les sujets craintifs semblent particulièrement plus vulnérables (anxiété : AOR = 2,514, IC95 = 1,985 -3,190 ; dépression : AOR = 2,521, IC95 = 1,938 -3,289), suivis par les sujets ayant un style d'attachement anxieux (anxiété : AOR = 1,949, IC95 = 1,498 -2,540 ; dépression : AOR = 1,623, IC95 = 1,207 -2,181). L'impact du confinement chez les sujets évités ne concerne que la dépression (AOR = 1,417, IC95 = 1,034 -1,937). Le fait d'être en couple pendant le confinement semble avoir eu un effet protecteur sur la dépression (AOR = 0,693, IC95 = 0,555 -0,866). Ni la présence d'enfants, ni la taille du lieu d'enfermement, ni la densité de population du lieu de résidence ne sont associées à l'anxiété ou à la dépression. Conclusion : L'impact du confinement sur la santé mentale dépend d'une pluralité de dimensions qu'il convient de saisir pour adapter au mieux les accompagnements psychologiques et sociaux post-confinement. Une prise en charge axée sur une approche biopsychosociale serait à privilégier.

Page 3 of 18 J o u r n a l P r e -p r o o f

The first cases of SARS COV-2 infection (severe acute respiratory syndrome to due coronavirus-2) officially appeared in China in November 2019. The coronavirus-related illness (COVID-19) rapidly spread across the world, and at the end of January 2020, the World Health Organisation (WHO) declared a state of public health emergency on an international scale. Many countries had to adopt lockdown measures as a result of the exponential increase of COVID-19 cases. In France, the first lockdown period was started on March 17 th 2020 and ended on May 11 th 2020. A second lockdown period took place from October 30 th to December 1 st 2020. In this article, the exclusive focus is the first lockdown period. During that time, French people were required to stay at home and limit their social interactions. Educational institutions -primary schools, secondary schools and universitieswere closed to the physical presence of any students. Only basic-need shops were allowed to remain open. Daily leisure outings were only allowed within a one-kilometre radius of the home, either alone or with members of the same household for a maximum duration of one hour.

Lockdown linked to the coronavirus illness (COVID-19) had a considerable impact on the population's mental health [1] . Increases in the prevalence of mood disorders, anxiety disorders and sleep disorders were reported in the international literature [1] [2] [3] , compared to what is usually observed [4] [5] [6] [7] .

Different vulnerability factors were evidenced. Greater vulnerability was observed among women [8] [9] [10] [11] . Vulnerability among sexual minorities was also documented [12] [13] [14] [15] . The contextual characteristics of lockdown need to be taken into account to understand its impact [16] . The impact seemed to vary depending on the density of the geographical zone in which subjects were locked down [17] . Lockdown for a single person clearly does not have the same impact as lockdown with a partner [18, 19] , which may have a positive effect, but can also foster domestic violence [20] . The presence of children at home during lockdown may have created increased stress, thus making people more vulnerable [21] . More structural psychological dimensions also need to be considered to understand the variability of the lockdown effect on mental health [22] , and in particular adult forms of attachment (see [23, 24] ).

In order to contribute to research tackling the effects of lockdown on mental health, a bio-psychosocial approach was adopted for this study on the impact of the COVID-19 lockdown on depression and anxiety, based on Bruchon-Schweitzer's [25] model of health psychology. This study is based on a differential approach. The hypothesis underpinning this study is that the impact of lockdown on anxiety and depression depends on:

-Socio-demographic characteristics (age, gender, socio-professional category, sexual orientation); -Lockdown conditions: 

The present research is part of the more general context of an online survey carried out from April 27 th to May 11 th 2020 on the impact of the COVID-19 lockdown on sexuality. In the course of this survey, a multiplicity of dimensions and behaviours related to individual, relational, psychological and sexual functioning were collected, together with socio-demographic data and data relating to lockdown conditions. In this article, the focus was on responses concerning the emotional impact of lockdown, via cross-analyses of the responses concerning depression and anxiety, socio-demographic information and responses to questions dealing with forms of attachment. The survey was presented in a conditional format. Certain questions were thus only asked depending on responses given to previous questions. In particular, questions about forms of attachment were only asked to people who had reported having a partner at the start of the study (but not necessarily living together during lockdown). The analyses were therefore only carried out on a sub-sample of the overall study, which was made up of 1753 subjects who reported having a partner.

Participant recruitment was mainly carried out via the social media. The invitation to participate in the study, presented and entitled "Sexuality and lockdown", was relayed by journalists in the general public information media. The survey was created using Limesurvey free software. A secure university server was used to store the data. Participation in the study was anonymous and free of charge. Great attention was paid to make sure that no information likely to identify the subjects was requested.

Neither the IP addresses nor the cookies were saved. Likewise, times and dates of connection were not recorded to increase the confidentiality of the responses. This form presented the context of the research, its precise objectives, the mean time required to respond and the potential repercussions of the research. The subjects were required to acknowledge access to the detailed objectives of the study and the conditions to be able to take part in the study, before they could consent electronically. Acknowledgement of access to the detailed objectives and agreement to participate were both necessary to be granted access to the survey. The contact details of the main research investigators and the institutions they belonged to were presented to the participants. The consent form gave information on the possibility of withdrawing from the study at any moment. Furthermore, at the top of each page of the questionnaire, the subjects could tick a box if they wished to quit and erase their responses.

The overall research dealt with many different themes all related to the main research objective: sexuality and lockdown. Each theme was processed on a separate page. When a new theme was introduced, the subjects were informed and a brief general definition of the theme was provided.

Concerning the dimensions more specifically addressed in this article, the participants were told that their investigation enabled an assessment of the potential psychological impact of the lockdown situation through an assessment of their emotional state.

Socio-demographic data was collected with an ad hoc questionnaire. The participants' sexual orientation was inferred on the basis of a scale that asked the subjects to indicate with whom they had had sexual intercourse. At one end, subjects could say whether it was exclusively men and on the other end, they could say whether their partners were exclusively women. Between the two extremes, the subjects could indicate whether there were as many men as women, or preferentially men or preferentially women. By crossing the responses to these questions, and to the question relating to birth gender, the subjects' sexual orientation was inferred: when men reported sexual intercourse systematically with women, they were considered heterosexual, when they reported sexual intercourse systematically with men, they were considered homosexual. When partners were either men or women, the subjects were considered bisexual. A similar procedure was used to infer the sexual orientation of the women participating in the study.

The General Anxiety Disorder scale (GAD7) [26, 27] was used to study anxiety symptoms. Even though it was originally designed to screen for generalised anxiety disorder, the GAD7 scale also enables the presence of anxiety disorders in different forms to be assessed (panic attacks, post-traumatic stress syndrome, social anxiety disorder…). The scale comprises 7 items to which subjects respond on a 4point scale for the frequency with which they have encountered 7 potential problems in the last two weeks (0: not at all; 3: almost every day). The psychometric qualities of the GAD7 have been recently J o u r n a l P r e -p r o o f demonstrated [28] and were also found in the present study (α > .80). Categorisation of the level of anxiety was conducted on the basis of the following thresholds: GAD<5: no anxiety disorders; GAD<10: slight anxiety; GAD<15: moderate anxiety; GAD≥15: severe anxiety.

Depressive symptoms were collected using the Major Depression Inventory (MDI) developed by the WHO [29] . Twelve items compose this questionnaire. Subjects are asked to give the frequency of appearance of recent depressive symptoms (last 2 weeks) on a 6-point Likert scale (5: all the time; 0: never). The psychometric qualities reported in the literature [30] and in the present research (α > .80) are good. The following thresholds enabled subjects to be discriminated: no depression MDI<20; slight depression: MDI≤25; moderate depression: 25>MDI>31; severe depression: MDI≥31.

The perception of attachment to the partner was assessed using the Relationship structure questionnaire of experiences in close relationships -revised [31, 32] . This questionnaire is composed of 9 items presented in the form of a Likert scale which can be used to assess attachment to 4 potential attachment figures (mother, father, romantic partner and best friend). In the present research, attention was focused on romantic partnerships. The questionnaire provides a score that assesses the anxiety dimension of attachment and the avoidance dimension. The psychometric qualities found in the literature were good, as was also the case for the present study (α≥.70).

From the scores obtained for the anxiety and avoidance dimensions, the subjects were categorised according to Bartholomew and Horowitz's 4-group classification [33] . Subjects with anxiety and avoidance scores lower than the median were categorised as subjects with a secure attachment style. Subjects with an anxiety score higher than the median, but lower than the median for the avoidance score, were categorised as having an anxious style of attachment. Subjects with a score higher than the median for both attachment dimensions were considered as having a fearful style of attachment.

The data was processed with R Software (3.6.0). Ordinal regression analyses were conducted with anxiety or depression as predictive variables.

J o u r n a l P r e -p r o o f

The participant characteristics are presented in Table 1. The sample was relatively young and female for two-thirds. Executives and students were overrepresented, whereas farmers and manual workers were underrepresented. Most participants were living in moderate to high-density towns and in spacious living quarters. Two-thirds were living with a partner during lockdown and one third with children. The presence of anxiety symptoms concerned more than half of the subjects and 18.4% presented moderate to severe anxious symptoms. More than one subject out of five exhibited moderate to severe depressive symptoms. Almost a third of the subjects could be categorised as having a secure or fearful attachment style. One out of five presented an anxious or avoiding attachment style.

Ordinal regression analyses concerning anxiety are presented in Table 2 . There was greater vulnerability among women and people not working (unemployed or students). Vulnerability linked to sexual orientation was found, but for bisexual people specifically. With regard to attachment styles, only subjects with avoiding and fearful styles seemed to present an increased risk of anxiety symptoms.

Results for the regression analyses concerning depression are presented in Table 3 . Increased vulnerability to depression was found among women. Four socio-professional categories presented a greater probability of depression than executives. Homosexual and bisexual people exhibited a risk of depressive disorders that was at least 1.75 times greater than for heterosexual people. Subjects with an insecure attachment style showed greater vulnerability than subjects with a secure attachment style.

The results of the present study are in line with the international literature on the subject of the impact of lockdown on populations' mental health, despite the fact that the rates of depression and anxiety in this sample were lower than those documented in other studies [3] ; the rates are however non-negligible compared to data usually observed [4] [5] [6] .

The present study is also a reminder that the health crisis did not have the same effects for everyone, which encourages the adoption of a bio-psychosocial and differential perspective, both to theorise the impact of lockdown and to suggest care tailored to the specificities of each individual. As J o u r n a l P r e -p r o o f suggested in other studies, women seemed more vulnerable towards the situation than men [8] [9] [10] [11] .

While this result has frequently been reported in studies on depression and anxiety [4] [5] [6] , it is important to bear in mind the specificity of the lockdown setting for women: increased economic insecurity, domestic violence, difficulties in accessing bare necessities or seeing gynaecologists, for instance.

Compared to executives and higher intellectual professions, students, people having difficulty finding employment or lower category salaried employees seemed more affected by lockdown. As executives and higher intellectual professionals belong to economically favoured socio-professional categories, this result is in line with what has been observed for the role played by economic vulnerability in understanding the impact of lockdown [10] and more broadly the impact of economic determinants on mental health [7] . This is not the first study to have underlined how very vulnerable people belonging to sexual minorities are. Existing research has however focused on the vulnerability of the LGBT community as a whole (for instance: [15] ). Nevertheless, in the present study, bisexual people were found to be more vulnerable than homosexual people, whether for the level of significance of the results or the amplitude of the effects (AOR). This result is a reminder of the need for responsiveness, accompaniment and prevention adapted to the specificities of the issues encountered [16] to lockdown also need to be considered (for instance: [15] ). The present research does not however enable these explanatory factors to be better understood among the subjects in this study. Further investigation is therefore necessary to determine their respective weights.

Some studies have investigated the role played by the different attachment styles in the psychological and social adaptation to lockdown [22] , but to our knowledge, none was conducted on a French-speaking sample. In line with the literature on adult attachment [23, 24] , the present results underline how vulnerable people with insecure attachment styles have been, with a probability 1.4 and 2.5 times greater of having been affected by the health crisis than subjects with a secure attachment style. The vulnerability varies with the type of insecure attachment, with fearful subjects and those with an anxious form of attachment to their partner seeming to be the most vulnerable.

Despite what was expected, contextual factors had little impact on depression and anxiety levels.

Living with a partner during lockdown was found to be beneficial, as depressive affects were probably minimised by social support, perceived to be of good quality (see Bruchon-Schweitzer, 2002) . However, neither the presence of children, nor the urban density of the place of residence appeared as factors, unlike reports in the literature [17, 21] . These differences could be due to cultural disparities -to our knowledge, studies in France have not examined these aspects. They could also be the result of socio-demographic differences between the different populations studied.

Similarly, they could be linked to differences in lockdown conditions across countries. Indeed, one study that showed links between urban density and lockdown impact was a Chinese study [17] .

However, urban density in China is hardly comparable to French urban densities, and the lockdown measures that were put into place by the Chinese authorities were not identical to the first French lockdown. Therefore, even if data from the international literature provides substantial resources to understand the effects of the health crisis on psychological functioning, the present results underline the need to take cultural/socio-demographic contexts into account if the lockdown effects on mental health are to be apprehended in a specific population/culture/society. Furthermore, because (1) the absence of significance of the results does not mean the absence of effects, but the inability to conclude on the existence of an effect, and because (2) anxiety and depression could enhance risks of child abuse [21] , long-term attention to the various impacts of lockdown is still necessary.

The main limitation encountered in the present study was its cross-sectional design, which made it difficult to evidence causal links between lockdown conditions, symptoms and socio-demographic variables. Links between socio-economic vulnerability and mental health have been extensively documented in the scientific literature and international reports outside lockdown settings [7] .

Similarly, the deleterious impact of insecure attachment styles has been demonstrated many times [23, 24] . It is therefore possible that the links observed between these different dimensions preexisted before the health crisis. What is more fundamental is to determine whether the impact of lockdown, or more globally the COVID-19 health crisis, has led to the emergence of psychopathological issues of a new nature -qualitatively different -or whether this health crisis reactivated or aggravated pre-existing forms of vulnerability. The difference in this case would be more quantitative. Beyond the theoretical reflection, the stakes are also in terms of care and accompaniment: whilst quantitative changes could encourage increased efforts to cater for the most vulnerable people on the basis of usual practice, qualitative changes could encourage reflexion on new modes of support and care.

It would have been interesting to take into consideration the effect of the approach of the end date of lockdown on the subjects' psychological functioning. It is indeed possible that the expectation of coming out of lockdown and a pandemic recession had an impact on anxiety and depression scores, or even that the effect of expecting to come out of lockdown had a different impact depending on the population's demographic characteristics. However, for confidentiality reasons and to ensure the strictest anonymity, this information was not recorded in the present study.

As with many studies carried out during lockdown, data collected remotely could have entailed a selection bias among subjects: only those with access to Internet and at ease with IT took part.

Examination of socio-professional categories, the characteristics of the place of residence, and anxiety and depression levels showed that the subjects in the present study belonged to a population group that was less vulnerable than the average. Because of lockdown, it is probable that the subjects completed the survey from their place of residence. However, a great majority were living with a partner during lockdown. These constraints, linked the situation of lockdown, did not as a result enable access to the most vulnerable subjects -for instance those for whom the relationship with their partner was a source of conflict or violence.

In this non-typical sample, which is likely to have been preserved from the most negative aspects of lockdown, the repercussions on mental health were nevertheless non-negligible. It is also possible that these repercussions were under-estimated as a result of the characteristics of the sample. The impact of lockdown on mental health depends on numerous dimensions that need to be understood to improve and tailor post-lockdown psychological and social care. Care centred on a biopsychosocial approach certainly needs to be encouraged. It could also be improved by the integration of specificities linked to biological gender, addressing them by taking norms and gender stereotypes into account. This approach could explore sexual orientations to determine the potential impact of discrimination experienced by subjects. Economic vulnerability should also be taken into consideration and an investigation into relationships with partners is needed. On this last point, the consideration of attachment styles towards partners could contribute to a better understanding of how lockdown is experienced.

There is no conflict of interest. 

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

Effects of Covid-19 lockdown on mental health and sleep disturbances in Italy

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

Global prevalence of anxiety disorders: a systematic review and meta-regression

Prévalence de la dépression majeure en France en population générale et en populations spécifiques de 2000 à 2018: une revue systématique de la littérature

Prevalence of depression in the community from 30 countries between 1994 and 2014

Élargir l'accès aux soins pour lutter contre les troubles mentaux, neurologiques et liés à l'utilisation de substances psychoactives

The indirect impact of COVID-19 on women

Period Poverty in a Pandemic: Harnessing Law to Achieve Menstrual Equity

Implications of the COVID-19 Pandemic for Gender Equity Issues at Work

Policy Brief: The Impact of C OV I D -19 on Women

The Untold Side of COVID-19": struggle and perspectives of the sexual minorities

Sexual and gender minority health vulnerabilities during the COVID-19 health crisis

COVID-19: The suffering and resilience of LGBT persons must be visible and inform the actions of States

Physical distancing due to COVID-19 disrupts sexual behaviors among gay and bisexual men in Australia: implications for trends in HIV and other sexually transmissible infections

COVID-19, HIV and key populations: cross-cutting issues and the need for population-specific responses

Urban-rural disparities in mental health problems related to COVID-19 in China

Making love in the time of corona-considering relationships in lockdown

How Is the COVID-19 Pandemic Affecting Our Sexualities? An Overview of the Current Media Narratives and Research Hypotheses

The impact of the Covid-19 pandemic in the precipitation of intimate partner violence

Stress and parenting during the global COVID-19 pandemic

Affective temperament, attachment style, and the psychological impact of the COVID-19 outbreak: an early report on the Italian general population

An attachment perspective on psychopathology

Attachment in adulthood: Structure, dynamics, and change

Psychologie de la santé. Modèles, concepts et méthodes. Dunod, éditeur. Paris

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

Rapid detection of generalized anxiety disorder and major depression in epilepsy: validation of the GAD-7 as a complementary tool to the NDDI-E in a French sample

Psychometric properties of the GAD-7 in a heterogeneous psychiatric sample

The sensitivity and specificity of the Major Depression Inventory, using the Present State Examination as the index of diagnostic validity

Psychometric evaluation of the Major Depression Inventory (MDI) as depression severity scale using the LEAD (Longitudinal Expert Assessment of All Data) as index of validity

The experiences in close relationships-Relationship Structures Questionnaire: A method for assessing attachment orientations across relationships

Questionnaire sur les structures relationnelles: adaptation et validation de la version française du Experiences in Close Relationships-Relationship Structures Questionnaire

Attachment styles among young adults: a test of a four-category model