School well-being in primary school children with chronic illness. A prospective cohort study

The ikidS cohort study

Detailed information on the ikidS Project and Cohort Study has been published elsewhere (Hoffmann et al., 2018; Urschitz et al., 2016). In short, a prospective, population-based cohort study was performed between September 2014 and July 2021 within the city of Mainz and the rural district of Mainz-Bingen. The study covered a geographical area of 704 square kilometres with a population of approximately 420,000.

In 2015, there were 79 public and private primary and special needs schools in this area, and 3,683 children were registered for their first year of school. The study population consisted of all children who had their preschool health examination within the study region between September 1st, 2014, and August 31st, 2015. Participants were enrolled prior to school entry on the day of their preschool health examination. This examination is a standardised, compulsory, state-wide health examination performed by public health physicians from the regional Department of Public Health of the Mainz-Bingen District.

The participating children as well as their parents and teachers were surveyed several times: in the final year of kindergarten (T0), 6 weeks before starting school (T1), 3 months after starting school (T2), and at the end of first, third, fourth and sixth grade (T3 to T6). The preschool health examination comprised a statutory parental questionnaire (T0; including information on parental education and immigrant status), a medical history and physical examination, and the administration of screening tests as well as tests of the students’ pre-academic skills. Results from the preschool health examination were provided to the investigators by the regional Department of Public Health (in personally identifiable form for study participants; in de-identified form for the total 2015 school enrolment cohort).

The cross-sectional surveys T1 to T6 of the ikidS Study collected information in the domains of general, physical and mental health, quality of life and social participation, school behaviour and performance (teachers’ questionnaire), family structure and socioeconomic factors, nutritional habits and leisure time activities, and assessed the use of health care services as well as special educational support services (teachers’ questionnaire).

Parental questionnaires covered different domains depending on the time points and comprised up to 20 pages and 200 items and were sent home per mail including pre-paid return envelopes. To increase the response rate, parents were sent up to three reminders to return the questionnaires. Whenever possible, we used accepted definitions and validated items and instruments from the German Health Interview and Examination Survey for Children and Adolescents (KiGGS) (Scheidt-Nave et al., 2007).

Out of the 3,683 students starting first grade in the 2015/2016 school year, 2003 children (54% of the population) were recruited. Until the end of first grade, when the children were surveyed for the present study, 168 participants (5% of the population) had withdrawn from the study, moved out of the study region, or had been deferred from school enrolment, resulting in 1,835 cohort participants (50% of the population). For the present study, children were excluded if (i) they suffered from an intellectual disability, (ii) did not participate in the children’s survey, or (iii) experienced comprehension difficulties during the children’s survey.

The ethics committee of the State Medical Association of Rhineland-Palatinate (file # 837.544.13 (9229-F)), the regional Supervisory School Authority, and the State Representative for Data Protection in Rhineland-Palatinate approved the study. Parents or legal guardians provided written informed consent for their children.

Physician diagnosis of a chronic illness

Information on chronic illness was obtained from the preschool health examination in the final year of kindergarten (T0) and parent surveys at T0–T3. During the preschool health examination, students are examined by school health staff and any medical, developmental, or behavioural disorders are recorded. We obtained the data for the 2015/2016 enrolment year from the Department of Public Health of the District of Mainz-Bingen.

In the parent surveys, parents were asked about the general health situation of their child, e.g., “In the past 12 months, has your child experienced any of the following diseases, developmental, or behavioural disorders diagnosed by a doctor or psychologist?” The following school-relevant diagnoses were considered: anaemia, attention-deficit/hyperactivity disorder, asthma, behavioural disorders, autism, depression, diabetes, emotional problems, epilepsy, preterm birth, cardiac conditions, hay fever, hearing impairment, short stature, foetal growth restriction, atopic dermatitis, thyroid disease, vision problems, speech anomalies, cancer, snoring and sleep disorder, overweight or obesity, and underweight.

Special health care needs

Parents were asked about their child’s health care needs using the validated German version of the Children with Special Health Care Needs (SHCN) screener (Bethell et al., 2002; McPherson et al., 1998; Schmidt & Thyen, 2008). The screener assesses five consequences of physical or mental health conditions: (i) need for prescription medications, (ii) need for medical, social, or educational care, (iii) functional limitations, (iv) need for specialized therapies, and (v) need for treatment due to psychological issues. Each area contains a primary filter question (e.g., “Does your child need or use more medical care, mental health or educational services than is usual for most children of the same age?”) and one or two supplementary questions. SHCN are present when at least one of the five filter questions and the corresponding supplementary questions were positively answered at one or more of the four time points (T0–T3).

Study groups

The main exposure of interest–chronic illness-was categorized into three levels:

1. current SHCN regardless of the presence of a physician diagnosis of a chronic illness (children with SHCN);

2. at least one physician diagnosis of a chronic illness but no current SHCN (children with a chronic condition but no SHCN);

3. no physician diagnosis of a chronic illness and no current SHCN (healthy children).

School well-being

The assessment of SWB was grounded in the Expectancy-Value Theory (Wigfield & Eccles, 2000) and the Self-Determination Theory (Deci & Ryan, 2008). Based on these theories, attitudes towards one’s own abilities, enjoyment of learning and social interaction were determined as relevant domains of SWB in the context of chronic illness (Schnick-Vollmer et al., 2020). These domains, in turn, are assessed by the Questionnaire for the Assessment of Emotional and Social School Experiences of First and Second Grade Primary School Children (German: “Fragebogen zur Erfassung emotionaler und sozialer Schulerfahrungen von Grundschulkindern erster und zweiter Klassen” (FEESS 1–2)) (Rauer & Schuck, 2004). The FEESS questionnaire was specifically designed for younger students as it requires very limited reading and writing skills.

We used three subscales from the FEESS: the cognitive-motivational subscale School-Related Self-Concept, the Social Integration subscale, which includes both socio-emotional and cognitive components, and the emotions subscale Joy of Learning. The School-Related Self-Concept subscale measures both the experience of competence and the expectation of success with which a student approaches a task. The Social Integration subscale captures the personal experience of social integration. The Joy of Learning subscale represents an emotional value component that arises from the pleasure of engaging in a task or from the resulting outcome.

These subscales also fitted to the framework model for the relationship between health behaviours, health status, education, and participation by Suhrcke & de Paz Nieves (2011), as well as the extensions by Dadaczynski (2012) and Urschitz et al. (2016). In these framework models, the factors “learning abilities”, “self-esteem”, and “discrimination” are defined as mediators in the relationship between health aspects and educational outcomes. These factors should be approximated by the three selected subscales.

The FEESS instrument was slightly modified from the original version: First, all items were phrased positively and addressed both boys and girls equally. Second, the original layout of the FEESS documentation sheets was changed. The animal pictures used for orientation purposes in the original version were replaced by pictures of fruits to avoid emotional reactions. Third, three different questionnaire versions were created based on the order of the item blocks to prevent fatigue effects. In addition, the response options “true” and “not true” were reworded to “yes” and “no”. The adapted instrument was validated and deemed sufficiently valid for both students with and without chronic illness (Schnick-Vollmer et al., 2020): the reliability of the adapted FEESS subscales (Cronbach’s alpha: School-Related Self-Concept: 0.84, Social Integration: 0.70, Joy of Learning: 0.83) ranged from satisfactory to good; the subscale correlations ranged from 0.58 (School-Related Self-Concept/Social Integration and Joy of Learning/Social Integration) to 0.72 (School-Related Self-Concept/Joy of Learning).

Students were asked about their SWB at the end of first grade (T3) during an in-classroom survey. The person administering the test read the statements out loud, using a neutral tone, and repeated them twice. Students responded “yes” or “no” to each item by checking the appropriate box on the questionnaire. If students required assistance, research assistants were available to support them. The study staff underwent training to keep the survey process consistent and to reduce bias.

Confounding variables

Child, family, and institutional variables that potentially confound the association between chronic illness and SWB were identified from the adapted framework model (Urschitz et al., 2016) based on the original work by Suhrcke & de Paz Nieves (2011) and a literature review. These variables were entered into a directed acyclic graph (Greenland, Pearl & Robins, 1999), and a minimally sufficient adjustment set for the analysis was identified: child’s gender (male/female), multiple birth (yes/no), breastfeeding duration (none/up to 6 months/6 months or more), chronic health conditions in the family (yes/no), immigrant status (child or at least one parent with foreign citizenship or born outside Germany, yes/no), socioeconomic status of the parents (ordinal) (22), attendance of the recommended paediatric check-up visits (yes/no), and school location (City of Mainz/District of Mainz-Bingen).

Statistical analysis

All descriptive statistics were performed on complete cases only and included an analysis of representativeness of the study participants and the analysis sample compared to the study population. Distributions of continuous variables were presented as means with standard deviations (SD); categorical variables were reported as absolute and relative frequencies.

The sum score for each FEESS subscale was calculated and converted into area-transformed T-values (mean and SD of the calibration sample of 50 and 10, respectively) as recommended by the developers of the questionnaire.

To examine the relationship between chronic illness and SWB at the end of first grade, a linear mixed regression model with chronic illness as the independent variable was established for each of the three FEESS subscales. Effect estimates (beta coefficients and standardized mean differences (SMD)) were adjusted for confounding variables.

Since the experiences of students within the same class are correlated, class was considered as a random-effects variable in the models (random intercept). The models included 168 school classes with an average of 8.9 participating children (SD = 3.1) per class.

Only observations with non-missing values for each FEESS subscale were included in the respective model. Missing values for the exposure and confounding variables were imputed using multivariate imputations by chained equations with 100 iterations (van Buuren & Groothuis-Oudshoorn, 2011).

All analyses were carried out with the statistical software package R (version 3.6.0, R Foundation for Statistical Computing, Vienna, Austria) (R Development Core Team, 2024).