The main finding of this study is that the impact of the COVID-19 restrictions increased parental stress, which in turn caused poorer mental health in children, particularly for externalizing problems and parent-child conflict. In addition, we found that a family resiliency factor, the ability to solve problems within the family, mitigates the negative effects of parental stress on parent-child conflict. Before drawing practical conclusions from these findings, we will comment on some general aspects of the results.
The main cause of the deterioration in children's mental health was the difficulty in coping with COVID-19 policies, rather than the fear of COVID-19 itself. However, the fear of COVID-19 did impact parents’ mental health: fear was associated with increased general stress, worsened anxiety, and depression in parents. Although fear of COVID-19 impacted parental mental health, this effect did not translate into worsened child mental health and parent-child conflict, primarily because fear of COVID-19 had only a minor effect on parental stress. This is in contrast with the impact of COVID-19, which had a considerate effect on parental stress. Parental stress was the main mediator of the effect of COVID-19 on children’s mental health and parent-child conflict. Therefore, a deterioration in parental mental health only had a negative effect on children's mental health if parental stress was affected. This implies that when designing interventions, the first step is to clearly define which specific mediators to focus on, based on the targeted outcome.
Parental stress was a systematic mediator in the significant pathways from COVID-19 impact to child behavioural difficulties. Also, when more than one mediator was involved, the first step was always general stress as measured by the DASS. When DASS anxiety and depression were part of the pathway, the indirect effects of the pathways were reduced and sometimes even reversed, but still small. This observation is consistent with results obtained in the general population. In Singapore, parental stress mediated the effect from the perceived impact of COVID-19 to harsh parenting and decreased parent-child closeness [6]. Parental stress also mediated the effect from parents' perceived distress to worsening of SDQ hyperactivity/inattention, conduct, and emotional problem scores [24]. This suggests that the same universal approaches could be used for all families to mitigate the impact of the COVID-19 crisis, although they may have to be adapted for families and children with severe mental health challenges.
We controlled several covariates in our analyses to evaluate the effects of the impact of COVID-19. The associations found for these covariates turned out to be consistent with the literature while clarifying some aspects of the model. For children, internalized problems were associated with older age and externalized problems with younger age, a pattern that is also found in the general population [25]. Increased parent-child conflict for older children may be explained by the greater difficulty adolescents have in complying with parental rules and restrictions. For parents, the time since the declaration of the state of emergency was associated with an increase in general stress, but not in parental stress. The persistence of the crisis and the uncertainty about the return to normality had cumulative effects on general stress. In the year following March 17, 2020, Ontario experienced three periods of school closures separated by periods of reopening and the 2020 summer vacation. The alternating closures and openings may have had a cumulative effect on overall stress, while parental stress was more stable over time—perhaps because the COVID-19 restrictions were relatively stable over this period. Finally, parental anxiety was lower in those with higher income, reflecting a well-recognized social inequality throughout the COVID-19 crisis [26].
But the clinically significant result is that the ability to solve problems at the family level moderates the effect of parental stress on parent-child conflict. This moderating effect is specific to parental stress and is not significant when considering the indirect route through general stress. It is significant after controlling for level of parental psychopathology, family income, time since the onset of the emergency, and children's age and gender. An intervention focused on strengthening problem-solving skills at the family level could be an effective prevention strategy to limit parent-child conflict, especially for parents with high levels of parental stress. This discovery is consistent with a meta-analysis that found that greater use of an individual-level problem solving strategy was associated with lower levels of externalizing symptoms [27]. In addition, the relationship between parent-adolescent conflict and externalizing and internalizing problems in adolescents was found to be moderated by the type of conflict resolution. Adolescents with a high level of problem solving skills had better adjustment than those with a low level [28].
This study has several limitations. First, because mediation involves a process that takes place over time, it is best captured by longitudinal data. Examining this mediation through cross-sectional data, however, is justified in our study by the fact that these relationships are nearly simultaneous, unfolding rapidly in repeated clashes. COVID policies, such as closing schools and mandating remote work had an immediate effect on parental stress (e.g., finding a new balance between parenting and work responsibilities), which could quickly exacerbate parent-child conflict (parent's role in distance learning, restricting contact with friends). In addition, mediation on cross-sectional data requires that the temporal ordering of the variable being examined be correct. There is no doubt that the impact of COVID is a causal factor, and the objective of the research was to estimate how this impact may affect parental stress and in turn child behavior. In addition, in a follow-up study, we employed a randomized intervention to manipulate the effect of the mediator on the outcome, a design that could more clearly show the causal implication of the mediating pathway.
Second, we did not stratify by diagnosis or broad categories of diagnosis for the following reasons. First, as the children were followed by different mental health professionals, a formal mental health diagnosis was not always made, and when diagnoses were available there were often multiple. But the main reason is that our hypothesis was that parental stress was a mediator of the impact of COVID-19 in this high-risk sample, as in the general population, irrespective of diagnosis. However, it is possible that some specific mental conditions are more, or less, sensitive to the effect of parental stress.
Third, we measured the impact of COVID-19 in parents, but not directly in children. This was not possible because of the very large age range, the clinical heterogeneity of the sample, and the fact that COVID-19 restrictions did not allow us to assess the children in person. The direct effect of the COVID-19 impact on child mental health can be understood through the modeling and reinforcement of parents’ response on their child’s behavior. However, some direct and child-specific aspects (such as limited relationships with peers, or changes in mental health care) were not measured, but most likely did have an impact on their mental health.
This study also has some strengths. We examined a wide range of variables that can impact behavioural difficulties in children, including both general and parental stress in a relatively large sample aged 4 to 18 years old. Most importantly, we were able to study a very vulnerable pediatric population that suffered more from COVID-19 restrictions than the general population and was difficult to engage in research because families were overburdened. We were able to propose a family intervention, which we then tested in a randomized trial by delivering a virtual, abbreviated version of the Collaborative Problem Solving [29] approach to parenting, the results of which will be presented separately.