Emotions and behaviours of child and adolescent psychiatric patients during the COVID-19 pandemic

Background Previous pandemics have had negative effects on mental health, but there are few data on children and adolescents who were receiving ongoing psychiatric treatment. Aims To study changes in emotions and clinical state, and their predictors, during the COVID-19 pandemic in France. Method We administered (by interview) the baseline Youth Self-Report version of the CoRonavIruS Health Impact Survey v0.3 (CRISIS, French translation) to 123 adolescent patients and the Parent/Caregiver version to evaluate 99 child patients before and during the first ‘lockdown’. For 139 of these patients who received ongoing treatment in our centre, treating physicians retrospectively completed longitudinal global ratings for five time periods, masked to CRISIS ratings. Results The main outcome measure was the sum of eight mood state items, which formed a single factor in each age group. Overall, this score improved for each age group during the first lockdown. Clinician ratings modestly supported this result in patients without intellectual disability or autism spectrum disorder. Improvement of mood states was significantly associated with perceived improvement in family relationships in both age groups. Conclusions Consistent with previous studies of clinical cohorts, our patients had diverse responses during the pandemic. Several factors may have contributed to the finding of improvement in some individuals during the first lockdown, including the degree of family support or conflict, stress reduction owing to isolation, limitations of the outcome measures and/or possible selection bias. Ongoing treatment may have had a protective effect. Clinically, during crises additional support may be needed by families who experience increased conflict or who care for children with intellectual disability.

Supplementary Tables 1-9 Emotions and behaviors of child and adolescent psychiatric patients during the COVID-19 pandemic C Laurent-Levinson et al.For individual items, "rev" indicates that the directionality of the rating was reversed so that a high rating would indicate changes that would typically be viewed as more adverse, e.g., for Change in Family Relations, in the questionnaire a rating of 5 denotes "much better" and 1 denotes "much worse.These values were subtracted from 6 to create the reversed score (1 denotes much better 5 denotes much worse).For the Factor 2 score for Adolescents, loadings have been reversed here and in Figure 2     Shown are the Mean, SE and SD of clinician CGI-S ratings for 5 time periods --PAST (>3 months before COVID); the 3 months BEFORE confinement; CONF1 (first confinement); DECONF (first deconfinement); and CONF2 (second confinement) --for the entire cohort (N=139), and for subgroups with Intellectual Deficiency and/or Autism Spectrum Disorder (N=35) or without either diagnosis (N=104), as illustrated in Figure 4.
The last 3 columns summarize the LMM analysis of CGI-S ratings using R package lme4.The model included Subjects (random effect, intercept); and Time, Diagnosis and Time x Diagnosis (fixed effects) (see Methods for details).Significance was evaluated by likelihood ratio tests.There was a significant interaction between time and diagnostic group (p=0,026): CGI-S differed across time points in those without ID or ASD (p=3.20E-9) but not in those with ID or ASD (p=0.62).For patients without ID or ASD, * indicates mean CGI-S values that were significantly different than the previous time period (see details in B below).As shown in (A) above, LMM analysis showed significant change of CGI-S score over time, but only in patients without ID or ASD.Here we show results of post hoc LMM analysis of differences in CGI-S scores between adjacent time periods.Consistent with the primary analyses of Mood8 scores in both Adolescents and Children, they show an improvement, on average, during CONF1.This was followed by a modest additional improvement during DECONF and then a modest worsening during CONF2 (back to the CONF1 level).

B. Pairwise comparisons of adjacent time periods (subgroup without ID or ASD, N=104)
(left panel) to clarify that worsening of Change in Family Relations was correlated with worsening (positive change) of Mood8-Change scores, as shown in the Figure.This was true in Children as well.

Table S2 : Life Changes factor analysis (children) Variable 1 Relationships 2 Stress 3 Insecurity 4 External Change in family relations (rev) 0.806
Shown are the results of Principal Components Analysis of 12 Life Changes items in each cohort, after Varimax rotation, retaining factors with Eigenvalue>1.Total variance explained is 61.9% in adolescents and 58.2% in children.

Table S5 : MOOD8 scores in relation to confinement and diagnosis (adolescents)
Shown are mean MOOD8 scores and SD before and during the first COVID-19 confinement.The first row shows Before and During scores for All participants and the results of a paired t-test demonstrating a significant change (lower MOOD8 during confinement).The remaining rows report data for participants without (Dx-) and with (Dx+) five frequent ICD-10 diagnoses and results of t-tests comparing those subgroups.The Change score is CONF1-Before, so that a negative Change score reflects less low mood and worry during CONF1.Anxiety includes Separation Anxiety Disorder, Performance Anxiety Disorder, Social Phobia, Obsessive-Compulsive Disorder, or Other Anxiety Disorder.MDD = Major Depressive Disorder.ADHD = Attention Deficit Hyperactivity Disorder.ASD = Autism Spectrum Disorder (these were mostly higher-functioning ASD, without comorbid ID).ID = Intellectual Disability or Autism Spectrum Disorder.ID = intellectual Disability.

-Mean SD Dx+ Mean SD T df P
See legend forTable S5.ASD was not analyzed separately for children, because out of 22 ASD and 31 ID patients, 19 had both diagnoses.