Psychological state of a sample of patients with mood disorders during the first French COVID-19 lockdown

Since the beginning of the COVID-19 pandemic, evidence shows the negative psychological impact of lockdown measures in the general population. It is also important to identify predictors of psychological distress in vulnerable people, particularly patients with history of depressive episodes (the most prevalent psychiatric disorder), in order to adapt mental health strategies for future lockdown measures. This study aim was to (1) compare in 69 healthy controls (HC) and 346 patients with a major depressive episode in the two previous years (PP) self-reported psychological symptoms (depression, anxiety, insomnia, suicidal ideation, traumatic stress, anger) and living conditions during the first national French lockdown, and (2) identify predictors of significant psychological distress in PP. The levels of psychological symptoms were very low in HC compared with PP, independently of the living conditions. Half of PP had no psychiatric contact during the lockdown. Loneliness and boredom were independent predictors of depression, anxiety and insomnia, whereas daily physical activity was a protective factor. Virtual contacts protected against suicidal ideation. Our results highlight the need of specific strategies to target loneliness and boredom and to improve care access, including telepsychiatry. Longitudinal studies must investigate the COVID-19 pandemic psychological impact in clinical samples.


Discussion
This is the first study investigating psychological distress and its predictors during the first French national lockdown in a homogeneous sample of patients with a previous diagnosis of major depressive episode by a psychiatrist. This is different from most of the previous studies on COVID-19-related lockdowns that included  15 . Our analysis suggests that the presence of psychological symptoms was not related to living conditions and changes in habits during the lockdown because these variables were comparable between HC and PP. The limited access to care or healthcare interruption could have contributed to the higher risk of acute mental symptoms during the COVID-19 pandemic. Half of participants in the PP group did not have any contact with a psychiatric service since the lockdown initiation, for not specified reasons. In China, approximately 22% of patients reported a pandemic-related interruption of psychiatric care 19 . Moreover, in our study, changes in psychotropic use was a common predictor of all psychological outcomes. However, this change mostly consisted in an increase of psychotropic intake, which may be a consequence rather than a cause of the acute mental symptoms. Health professionals must find new methods to look after patients, and pro-active non-intrusive links with the healthcare system should be offered to people with pre-existing poor mental health. It is essential to provide continued psychiatric intervention through telepsychiatry, and to strengthen the patients social support via community mental health services.
It has been hypothesized that the pandemic might increase substance use in an attempt to cope with negative feelings 20 . In our sample, PP reported more frequently use of tobacco and cannabis than HC, who consumed more often alcohol. Approximately one third of substance users reported increased consumption. Similarly, in Spain, Solé et al. 14 found that psychiatric patients increased the use of tobacco, but not of alcohol and cannabis, compared with controls. Moreover, a regular web-based survey of a representative sample of the French general population carried out by Santé Publique France ("Public Heath France") to monitor health behaviors and mental health during the pandemics 21 reported that one third of subjects increased tobacco consumption and 10% of subjects increased alcohol consumption associated with an initial increase in depression and anxiety levels after the lockdown implementation. Our analysis also found that alcohol use was predictive of anxious symptoms.
In our sample, lifetime anxious disorder was the only psychiatric disorder that predicted psychological distress. Indeed, people with anxiety-related disorders have been the most affected by the COVID-19 pandemic, with greater fears about several consequences 18 . In our study, loneliness and boredom were two common and independent predictors of depression and anxiety. Loneliness may be particularly prevalent and devastating during a pandemic, due to social distancing measures. At the onset of COVID-19, there has been concern about the effect of increased isolation on loneliness and other mental health conditions. A recent randomized controlled trial showed that a layperson-delivered, empathy-oriented short telephone call program reduces loneliness, depression, and anxiety and improves the general mental health of participants within 4 weeks 22 . Sense of isolation, confinement, reduced social and physical contacts can frequently cause boredom 23 . Boredom, a state that relates to low arousal with dissatisfaction due to perceived monotony and repetition 24 , is considered as one of the most relevant stressors in individuals who experienced isolation during the pandemic 25 . Perceived stress may influence emotional distress through boredom proneness 26 . People who are quarantined should be advised to stave off boredom and provided with practical advice on coping and stress management techniques 3 , such as mindfulness training 27 or engaging in creative behaviors 28 . People should also be encouraged to regularly practice a physical activity that reduces boredom and the feeling of time slowing down during a lockdown 29 . In agreement, in our study, daily physical activity was a protective factor against depression, anxiety and insomnia. Besides its effect on boredom, regular physical activity might also reduce anxiety and depression 30 . The World Health Organization recommends 150 min of moderate intensity or 75 min of vigorous intensity physical activity per week, or a combination of both, during self-quarantine 31 . Our result might have been biased because only less anxious and depressed subjects were able to practice regular physical activity.
Another interesting result of our study is that virtual contacts were the only protective factor against suicidal ideation. This strengthens the need to propose individual-level interventions to reduce loneliness, enhance social support, and increase opportunities for social interactions. Psychological counselling telephone helplines and online consultations played a significant role in maintaining the citizens' good mental health in China 32 . Telepsychiatry emergency services or hotlines should be made available to patients with intense suicidal ideation. In France, a brief contact intervention, including telephone calls (VIGILANS program), has shown its efficiency for reducing suicide reattempt 33 . The impact of such intervention deserves to be studied. Telemedicine provides new opportunities to address the patients' mental health needs by creating disease awareness and improving treatment adherence 34,35 .
The present results must be interpreted in the light of some limitations. First, our findings may not be generalizable to all patients with depression. Participation in this study was on a voluntary basis and sampling was carried out online when strict lockdown measures were in place, thus excluding patients who did not have internet access. Second, the cross-sectional approach does not allow demonstrating a causality between selfperceived psychological status and lockdown. Future longitudinal studies should assess the consequences of the COVID-19 pandemic in psychiatric patients. Third, the current clinical characteristics were self-reported, but the lifetime psychopathology was diagnosed by a clinician before the lockdown.
To conclude, improved access to telepsychiatry services, home delivery of psychotropic medications, online psychiatric first-aid resources, and infectious disease outbreak preparedness play a pivotal role in minimizing the severity of psychiatric symptoms experienced by psychiatric patients. Our results might contribute to the www.nature.com/scientificreports/ development of specific strategies for mental health care by identifying potential targets of assessment and care in psychiatric patients, beyond the usual risk factors, such as loneliness and boredom. Mental health preparedness and anticipation of future outbreaks will lead to an increased awareness of the needs of psychiatric patients and of the contingency plans to be put in place.

Material and methods
Design and participants. This observational study was carried out the Department of Psychiatric Emergency and Acute Care of the Academic hospital of Montpellier, France. The adult participants have been previously included in research projects or followed at our department between March 15, 2019 and March 15, 2020. This allowed us to constitute two groups: HC and PP (out-and in-patients with history of depressive episode). Exclusion criteria were: absence of a previous psychopathology assessment using a validated psychometric tool (Mini-International Neuropsychiatric Interview, MINI; Structured Clinical Interview for the DSM-IV Axis I disorders, SCID-1; or Diagnostic Interview for Genetic Studies, DIGS), and refusal to participate. In total, 69 HC without any history of psychiatric disorder and 346 patients with history of depressive episode within the last 2 years according to the DSM-IV criteria (PP) accepted to participate in this on-line survey. Lifetime psychopathology was previously assessed by a trained psychiatrist or psychologist using the MINI or DIGS.
The study protocol was registered in the Clinical Trials Registry (ClinicalTrials.gov NCT04374643) and was approved by the Institutional Review Board of Montpellier Academic Hospital (IRB-MTP_2020_12_202000421 (30/03/2020) and IRB-MTP_2020_12_202000436 (08/04/2020). All experimental methods were carried out in accordance with the ethical guidelines determined by the National Ministry of Health, Labour and Welfare and the Declaration of Helsinki. All participants provided written informed consent before entering the study.
Procedure. During the French lockdown (March 17-May 11, 2020) an e-mail was sent to participants with an anonymization number and an information note about the study aim. In another e-mail, they received a computerized form to complete in which they needed to add their anonymization number.
Outcomes. The study focused on the prevalence of six outcomes in the last 15 days: depression, anxiety, anger, insomnia, traumatic stress, and suicidal ideation. Depression symptoms and the presence of suicidal ideation were assessed with the PHQ-9 36 . A PHQ-9 score ≥ 10 has been associated with major depressive disorder (88% sensitivity and 88% specificity) 36,37 . The presence of suicidal ideation was assessed with the corresponding item of the PHQ-9 > 0 which evaluates the frequency of passive thoughts of death or self-injury within the last two weeks 38 , anxiety symptoms were measured with the GAD-7 38 . A GAD-7 score ≥ 10 corresponds to moderate to severe generalized anxiety disorder (89% sensitivity and 82% specificity) 39 . French versions of PHQ-9 and GAD-7 were freely downloadable on the patient Health Questionnaire website (www. phqsc reene rs. com). Anger state was assessed using the STAXI-state 40,41 . Distress resulting from a traumatic life event was evaluated with the 22-item IES-R40 using a cut-off of 21 42 , and insomnia with the ISI 43,44 using a cut-off of 14 for clinical insomnia 45 . Moreover, a variable was created to describe psychological distress based on the presence of severe self-reported symptoms, i.e. presence of suicidal thoughts (suicidal item of the PHQ-9 > 0) or high score for at least one scale (PHQ-9 score > 9; GAD-7 score > 9; IES-R score > 21; ISI score > 14; STAXI-state above the median).
Regarding the factors associated with mental health outcomes and potentially related to the lockdown, sociodemographic data (gender, marital status, professional status), living conditions during the lockdown (living alone or not, home characteristics), habits during the lockdown [frequency of virtual (i.e. telephone and video) and written contacts (i.e. text messages, mails), feelings of loneliness and boredom, frequency of physical activity and going out from home, and their corresponding changes compared to usual life], COVID-19-related data (infectious status, fear of infection for the subject and relatives), access to psychiatric care, substance consumption (tobacco, alcohol, analgesics and psychotropic drugs) were collected and analyzed.
Statistical analyses. The characteristics (psychological outcomes, socio-demographic variables, conditions of living and habits during the lockdown, COVID data) in the two groups (HC and PP) were described using medians (minimum-maximum) and numbers (frequency, percentages) for quantitative and qualitative variables, respectively. The significance of between-group differences was evaluated using the Kruskal-Wallis test, Fisher's exact test, or Chi square tests. Cliff 's delta effect size and Odds Ratio were computed with their 95% confidence intervals for respectively quantitative and qualitative variables. P values were corrected for multiple testing using false discovery rate correction (FDR) 46 .
To manage the high number of variables evaluated as potential risk factors for psychological outcomes and their possible collinearity, a two-step variable selection was carried out for each outcome: (1) logistic regression was used to evaluate the crude relationships between the outcome and the potential risk factors from which only variables with p value < 0.15 were kept for step 2; (2) the retained variables were included in a multivariate logistic model and selected using the stepwise Akaike Information Criterion method (involving both forward and backward approaches) to keep only informative variables. Socio-demographic variables and history of psychiatric disorder were systematically kept in the model when analyzing the whole sample (HC and PP subjects). The same procedure was used also for PP alone. Variables describing habit changes were excluded from these analyses because they partly determined habits during the lockdown. Odds ratio (OR) and 95% confidence intervals (CI) were estimated for the best selected models. Significance of the associations might be estimated from the 95% CI. All analyses were performed with the R 4.0.3 software (R Core Team 2018, https:// www.R-proje ct. org/).