Psychiatric and non-psychiatric population vulnerabilities in time of a crisis: the unsuspected aggression factor

COVID-19 studies have reported increased anxiety and depression levels in the general population across multiple countries from Asia to Europe [14-16]. Some research showed increased psychological discomfort [17] and even increased psychiatric symptoms among the general population, with, for example, 67% experiencing post-traumatic stress disorder (PTSD) and 19% suicidal ideation [18].

Fernandez et al. [19] looked at participants’ risk profiles, finding that, second to sociodemographic factors, preexisting psychiatric issues were associated with risk of psychological distress. In 2020, Iasevoli et al. [20] found that patients with serious mental illness were more likely to experience high pandemic-related stress and had a higher risk of experiencing more severe anxiety and depressive symptoms, suggesting that healthy controls might experience less psychological distress. Indeed, when measuring caregivers’ mean scores on the depression, perceived stress, and general anxiety disorder scales versus controls, Iasevoli et al. found a lower depression score and comparable stress and general anxiety disorder scores. The mean care givers self report score corresponded to mild burden.

Several researchers concluded that, second to demographic predictors, mental disorder history increases the risk effect of all mental health state indicators [21, 22]. To our knowledge, little research has assessed the anxiety and depression levels of mentally ill persons during COVID-19 in relation to aggression levels or impulsivity scores.

An increased number of psychiatric care consultations was predicted in 2020 by psychiatrists [23], more so for patients with prior psychiatric diagnoses [24]. Many researchers recommended undisrupted care [25-27] as the vulnerable mentally ill were especially at risk of developing increased anxiety [28, 29]. A history of psychiatric illness was associated with experiencing anxiety and anger four to six months after release from isolation measures [30]. Jeong et al. showed that we could prevent mental health problems by providing support to individuals with vulnerable mental health, providing accurate information and appropriate supplies and accommodations during a health crisis. Very rapidly, experts recommended telehealth solutions to prevent some of the worst effects of mental health crises [31]. At first, the Chinese, Singaporean, and Australian governments noted the psychological side effects of COVID-19 that needed to be considered. The Lancet [28] alerted readers that the psychiatric population “might experience worsening symptoms”, whereas the general population “might develop new mental health problems, especially depression, anxiety, and post-traumatic stress (all factors associated with increased suicide risk)”.

Patients with identified psychiatric disorders (e.g., affective disorder, schizophrenia, and addictive disorders) reported heightened stress levels and half of them reported experiencing critical stress. A quarter of patients with affective disorders reported increased difficulties sleeping and increased irritability [29].

In a nonclinical population, decision-making impulsivity was thought likely to increase in several life domains, such as diet, high-risk behavior, and habits (e.g., smoking and legal and illegal substance use) [32]. People also experienced psychological conflict between the urge to feel safe and the desire for a pleasurable life, which resulted in maladaptive behaviors [33].

Some elements from the 2020’s literature filled us with impulsivity or aggressive outbursts fear that could be displayed by young adults or members of known impulsive populations [33-37]. Nivette et al. [33] feared that young adults displaying low trust and “antisocial potential” with previous low rule acceptance, low shame, and poor self-control (i.e., delinquents) would oppose or comply less with governmental measures. Populations with “dark personality” traits were thought likely to knowingly expose others to risks [34]. This vulnerability to impulsive behaviors also applied to the young adult population renowned for its proneness to gaming addiction. Verizon, an online game provider, noted a 70–75% increase in online gaming activity coinciding with initial stay-at-home directives [35, 38] and an increase in the number of active users to 20 million [39, 40].

Considering the adverse effects on depression and anxiety levels expected from the COVID-19 lockdowns and our initial observations as field clinicians, we hypothesized that participants with a history of psychiatric care would be more vulnerable than the general population and experience heightened psychological issues compared with the non-psychiatric population. Therefore, we chose to use questions about the context (i.e., sense of loneliness, impact of isolation on relational and emotional life, number of cohabitants during quarantine, and number of people the respondent was in contact with every week), clinical scales, and other scales found in the literature to confirm this hypothesis, assess the situation more precisely, and better understand the causes and consequences.

Our hypothesis is that the psychiatric population would experience more anxiety/depression, suicidal risk, aggression, and impulsivity than would the non-psychiatric population during the COVID-19 crisis.

We recruited 383 respondents to an online questionnaire using Google Forms during France’s second month of lockdown. The overall sample contained 110 students, four workers, 64 self-employed, 42 employees, 57 people with white-collar professions, 11 with non-specific status, 39 unemployed or on medical leave, 16 retirees, and 39 workman position. The gender ratio was 286 female, 95 male, and two nongendered participants. We recruited part of the sample from the general population, another part from private outpatient clinical care, and the last part from the Nîmes University health service. The participants completed clinical scales and supplied demographic data (20–30 min). The exclusion criteria for both groups were known neurological disease and developmental disability. All participants were proficient in the French language, had normal or corrected-to-normal vision, and were naive as to the purpose of the study. According to ethical provisions of the World Medical Association Code of Ethics (Helsinki Declaration) for experiments involving human subjects, participants gave online informed consent to participate in this experiment.

The Beck Hopelessness Scale (BHS) from 1974 [41] was translated into French by Cottraux et al. in 1985 [42]. This scale is intended to evaluate pessimism and cognitive beliefs about the future, indirectly capturing suicidal intentions. Its items elicit binary true/false responses, with a total score ranging from 0 to 20. The Cronbach’s alpha of the instrument was 0.72.

The Suicide Behaviors Questionnaire – Revised (SBQ-R), formulated by Osman et al. in 2001 [43], assesses suicidal behaviors. SBQ-R is one of the few tools asking about anticipated future suicidal thoughts or actions as well as past and present ones; it includes items about lifetime suicidal ideation, plans to commit suicide, and actual attempts. Shakeri et al. [44] later reformulated it for a psychiatric population. A total score of 7 and higher in members of the general population or of 8 and higher in patients with psychiatric disorders indicates a significant risk of suicidal behavior. We used the validated French version of the instrument by Potard et al. [45]. The Cronbach’s alpha of the SBQ-R items was 0.80.

The Hospital Anxiety and Depression Scale (HADS) was created and validated by Zigmund and Snaith in 1983 and has been widely used in both general health care and psychiatric research [46]. The HADS identifies the presence of anxiety disorders and depression. It is divided into anxiety (HADS-A) and depression (HADS-D) subscales [47], having cut-off scores of 9 and 11, respectively; the Cronbach’s alpha was 0.67 for anxiety and 0.79 for depression.

The Aggression Questionnaire (AQ12), in its 2009 French translation by Genoud and Zimmermann [48], contains 12 items assessing the dimensions of aggression, but can be used to yield a single score. The questionnaires use a Likert scale ranging from 1 (“Not at all like me”) to 6 (“Completely like me”) and had a Cronbach’s alpha of 0.80.

Impulsive Behavior Scale – Short version (UPPS-S) was translated into French and validated by Billieux et al. in 2012 [49]. It consists of a self-report scale with 20 items assessing four factors of impulsivity: urgency (negative and positive), lack of premeditation, lack of perseverance, and sensation seeking. Positive urgency assesses impulsivity due to positive emotion, whereas negative urgency assesses impulsivity due to negative emotion. The respective Cronbach’s alphas indicated good consistency: negative urgency alpha = 0.78, positive urgency alpha = 0.70, lack of premeditation alpha = 0.79, lack of perseverance alpha = 0.84, and sensation seeking alpha = 0.83.

We formulated a questionnaire to collect demographic data regarding age, gender, and sociodemographic category. It also included the following questions: a) Have you ever been treated for psychological problems (by a psychologist or psychiatrist)? Response: Yes/No; b) Are you currently receiving professional psychological/psychiatric care (e.g., psychiatrist, psychologist, or psychological care structure such as daycare/hospital/ association)? Response: Yes/No; c) Do you feel more isolated/lonely at the moment (since confinement)? Response ranging from: No, not at all = 1 to Yes, very often = 5; d) Since lockdown, with how many people have you had regular contact (at least twice a week, either face to face or via telephone or video-conference)?; e) How many people did you live with during confinement?; f) What has been the overall impact of confinement on your relationship life? Response ranging from: Negative = 1 to Positive = 5; g) What has been the overall impact of the lockdown on your emotional life? Response ranging from: Negative = 1 to Positive = 5.

Our first defined group (Gp1) consisted of participants naive to psychiatric care (n = 201, 70% women, 29% men, 1% non-binary), the second group (Gp2) consisted of participants with past experience of such care (n = 114, 78% women, 21% men, 1% non-binary), and the third group (Gp3) consisted of current outpatient care recipients (n = 68, 79% women, 19% men, 1% non-binary). The nonclinical and clinical populations were recruited via social media and information about the research delivered to current patients of several clinical psychologists volunteering to recruit participants within their own networks. These three groups were created based on the questionnaire responses, allowing us to distinguish participants who had never had any experience of psychological/psychiatric care, participants who had receivedpsychiatric care at some point in life, but were not currently receiving it, and participants who reported having had experience of psychological/psychiatric care and were currently being supported by such treatment.

We used SPSS 2.0 software for descriptive analysis, correlation analysis, and regression observation, whereas additional analyses were performed using R. For observational purposes, we ran several sample analyses. Demographic measure means are presented in Table 1 and clinical measure means are presented in Table 2.

We ran t-test analyses comparing Gp1 and Gp2, finding significant differences (t [313]) in SBQ-R (p = ***), HADS-A (p = *), negative urgency (p = *), BHS (p = *), and AQ12 (p = **). For the comparison of Gp1 with Gp3, we found significant differences (t [267]) in age (p = *), sense of loneliness (p = *), number of persons one is isolated with (p = *), lack of perseverance (p = ***), BHS (p = ***), HADS-D (p = ***), HADS-A (p = ***), SBQ-R (p = ***), and AQ12 (p = **). For the Gp2 and Gp3 comparison, there were significant t-test results for SBQ-R (p = ***), sense of loneliness (p = *), HADS-D (p = *), HADS-A (p = *), negative urgency (p = *), and lack of perseverance (p = *). All results are summarized in Table 3 and Fig. 1.

For further insight, we compared the present scores with scores from the literature and clinical thresholds, as presented in Table 4. These results led us to further analyze the correlation within these groups.

For Gp1, representing participants who had never received psychological care, we ran a regression analysis to see which factors predicted HADS-A. Four dimensions predicted anxiety levels, i.e., BHS (p = 0.035, β = 0.159), AQ12 (p = 0.008, β = 0.310), HADS-D (p = 0.035, β = 0.360), and positive urgency (p = 0.012, β = 0.178), with R² = 0.500 and F = 15.209.

For HADS-D, there were four predictive dimensions, i.e., relational life (p = 0.037, β = –0.151), BHS(p = 0.015, β = 0.187), HADS-A (p = 0.000, β = 0.340), and AQ12 (p = 0.030, β = 0.165), with R² = 0.487 and F = 16.148.

For AQ12, there were four predictive factors, i.e., HADS-D (p = 0.015, β = 0.173), SBQ-R (p = 0.000, β = 0.245), HADS-A (p = 0.001, β = 0.249), and age (p = 0.007, β = –0.192), with R² = 0.45 and F = 12.790.

For patients with past therapy (Gp2): HADS-A was predicted by AQ12 (p = 0.019, β = 0.221), BHS (p = 0.041, β = 0.213), and number of cohabitants (p = 0.012, β = 0.230), with R² = 0.313 and F = 8.289; HADS-D was predicted by BHS (p = 0.013, β = 0.277) and sense of loneliness (p = 0.004, β = –0.284), with R² = 0.337 and F = 7.532; and AQ12 was predicted by only HADS-A (p = 0.017, β = 0.256) and sensation seeking (p = 0.005, β = 0.276), with R² = 0.273 and F = 4.773.

For Gp3, representing patients currently receiving therapy, HADS-A was predicted by HADS-D (p = 0.002, β = 0.437) and AQ12 (p = 0.001, β = 0.363), with R² = 0.637 and F = 11.776, whereas HADS-D was predicted by HADS-A (p = 0.002, β = 0.430) and BHS (p = 0.003, β = 0.370), with R² = 0.658 and F = 0.9.628.

For a graphical overview of the three regression arrays depending on the group, see Figs. 2, 3 and 4. The regression tables are available as Supplementary Material 2 (Tables D, E, and F).

We did not control for any precise psychopathological diagnoses, as this was an online questionnaire and we decided to protect medical privacy. The second limitation comes from collecting our data during the second month of lockdown. Third, we lack baseline data and therefore insight into the evolution of the respondents’ scores from the pre-pandemic period, and comparison with previous scores from the literature cannot be considered sufficient to draw any conclusions. Finally, as the gender ratio in all groups was above 70% women, the results would be difficult to generalize to more evenly mixed samples..