Determinants of health-related quality of life in healthy children and adolescents during the COVID-19 pandemic: Results from a prospective longitudinal cohort study

The data for this analysis come from the school-based longitudinal cohort study Ciao Corona [22], in which 55 randomly selected schools (primary school grades 1–6 and secondary school grades 7–9, ages 6–17 years) in the canton of Zurich, the largest canton in Switzerland of approximately 1.5 million inhabitants (18% of the total Swiss population), took part. Subjects (or their parents) were also asked to fill out a baseline questionnaire at the time of their first antibody test and to complete follow-up questionnaires on a periodic basis (July 2020, January 2021, March 2021, September 2021, and July 2022). The analysis set of this study included children and adolescents who had KINDL total scores in June 2022 and at least one questionnaire filled out earlier during the pandemic.

The study was conducted in accordance with the Declaration of Helsinki and approved by the ethical committee of the canton of Zurich (2020-01336), and the study design has been published elsewhere [22] (ClinicalTrials.gov identifier: NCT04448717). All participants provided written informed consent before being enrolled in the study. Results relating to lifestyle behaviors [23] and HRQOL [24] have been reported previously.

The primary outcome was HRQOL in June 2022, assessed using the KINDL questionnaire, filled out either by primary school students and parents together, or by students in secondary school on their own. KINDL is a reliable and valid measure of HRQOL in children and adolescents [25, 26], on a scale from 0 (worst) to 100 (best). It has 6 subscales, each from 0 to 100: physical, emotional, self-esteem, family, friends, and school. Several slightly different versions of KINDL are available, of which the parent version for children 7–17 years old was used [27]. Age group was determined by the highest grade level achieved during the study period. Children who were in sixth grade or lower were in the primary school group (having never gone to secondary school), while those who were in at least seventh grade by 2022 were in the secondary school group, even if they were still in primary school in 2020.

Possible determinants of HRQOL were pulled from questionnaires and categorized into three categories: biological, psychological, and social determinants, as has previously been described as a model for HRQOL [7, 8, 28, 29] (Fig. 1). Biological variables included sex (male, female, other), BMI, PA, screen time (ST), sleep duration, the presence of chronic health conditions, and symptoms possibly compatible with post-COVID-19 condition (also known as long COVID). BMI was calculated according to weight and height and compared with the standard Swiss population [30] to derive z-scores. BMI was then categorized as overweight if its z-score was 1 or higher. PA, ST, and sleep were recorded in hours per week, which were then compared with World Health Organization recommendations [31] (\(\ge 1\) h/day of PA, \(\le 2\) h/day of ST, and 9–11 h/night of sleep for 6–13 year olds or 8–10 h/night for 14–16 year olds). Chronic health conditions included asthma, celiac disease, neurodermatitis, type I diabetes, inflammatory bowel disease, hypertension, attention deficit hyperactivity disorder, epilepsy, joint disorders, and depression/anxiety. Possible post-COVID-19 condition [32, 33] was identified if participants reported any number of symptoms lasting 3 months or longer that might be related to a COVID-19 infection in seropositive participants.

Psychological variables included sadness, anxiousness, and stress that were taken from Health-Behaviour in School-Aged Children questionnaires [34]. For sadness, children were asked how often they felt sad or depressed in the last 6 months (daily, multiple times per week, once per week, once per month, seldom or never). For anxiousness, they were asked how often in the last 6 months they felt scared or anxious (same responses as for depression). For stress, parents were asked how they would assess the level of stress in the child’s life on a scale from 1 (no stress) to 6 (extreme stress).

Social variables collected were parents’ nationality (at least 1 Swiss parent vs other), parents’ highest education level (at least one with college prep high school or university vs compulsory high school or professional school or lower), the presence of household financial difficulties (yes/no), and change of parents’ working situation (reduction in or loss of work, vs no change), or change or loss of employment due to parents’ health [35].

Parents of participants were asked by email to fill out, with their child, a baseline questionnaire online at the time of their first serological test (June/July 2020, Oct/Nov 2020, Mar 2021, Nov/Dec 2021, June 2022). For the HRQOL, sadness, anxiousness, stress, and physical activity questions, parents were specifically asked to answer them together with their child. Thereafter, they were invited to fill out an online follow-up questionnaire at semi-regular intervals (Sept/Oct 2020, Jan 2021, Mar 2021, Sept 2021, Dec 2021, June 2022). While the baseline questionnaires asked a number of demographic details (e.g., parents’ nationality and education levels), there was significant overlap in content between the baseline and follow-up questionnaires. For data analysis, timepoints were grouped by year: 2019 (retrospective questions in the June 2020 questionnaire relating to the prepandemic period), 2020, 2021, and 2022. HRQOL was taken from the June 2022 questionnaire, while possible determinants were taken from 2019 to 2021. If multiple questionnaires existed for a subject in the same year with the same question, the mean or most frequent response was taken. The variables post-COVID-19 condition, household financial difficulties, or change in parent’s employment situation were coded “yes” if there were any relevant responses in the given year, and otherwise “no”. Sleep, PA, and ST were summarized as the proportion of responses meeting recommendations in the given year. Further details on the questionnaires used in Ciao Corona are found in Online Resource 1.

KINDL scores were summarized as median [interquartile range (IQR)]. We used conditional inference trees [36, 37] estimated by binary recursive partitioning to identify possible determinants of HRQOL in children and adolescents. This statistical method identifies subgroups where stratifying by possible predictor variable produces a statistically significant difference in the outcome variable. For further details, see Online Resource 2. This procedure was repeated for the KINDL total score as well as for each of its subscales and stratified by age group (primary vs secondary school). As a sensitivity analysis, multiple imputation using chained equations was used to impute missing covariates [38] (m = 100), and then, recursive partitioning was used to identify significant predictors of HRQOL in each of the imputed datasets. We then counted how often each variable was included in the model selection procedure, with more important variables appearing more often than variables which are not determinants. All analysis was performed in R (R version 4.3.2 (2023–10-31)) using the packages partykit [36, 39] and mice [40].