Design
This study is a cross-sectional convenience sample from a cohort study in children with a chronic condition (8-19 years old) and a convenience population sample of healthy controls (11-17 years old).
Children with a chronic condition
Data for children with a chronic condition (number (N)=291) were obtained via the ongoing Patient Reported Outcomes in Children and children with Chronic/life-threatening conditions and Tailored InterVentions in a digital Environment cohort (PROactive cohort) [12]. Detailed information about the PROactive cohort study can be found elsewhere [12], [13]. The PROactive study was classified by the institutional review board (Medical Ethics Review Committee of the University Medical Center Utrecht, The Netherlands), to be exempt from the Medical Research Involving Humans Act (WMO), case number METC 16/707-C, and adhered to all local laws and the Declaration of Helsinki.
Healthy controls
Reference data for healthy controls were obtained via the Health Behavior in School-aged Children (HBSC) study [14]. HBSC is a cross-sectional, school-based survey with a focus on health and wellbeing that is conducted every four years in over 40 countries among nationally representative samples of children. The present study made use of the Dutch data collected in 2017 (N=8306) [14]. More detailed information about the methodology of the HBSC study can be found in the study protocol [15].
All children with a chronic condition and the same age as children in the HBSC sample, were matched to healthy controls using a 1:4 ratio and Propensity Score Matching with Optimal Matching [16]. Matching was done based on age and sex, as these are important factors influencing sleep and were variables that were collected in both convenient samples [17], [18]. Matching was considered successful if age and sex did not differ significantly between children with a chronic condition and their healthy peers, as determined by independent samples T-tests.
Study Sample
Children with a chronic condition
Participants were included as part of the PROactive cohort, for which data collection started in December 2016 and is still ongoing [12]. Inclusion criteria are: children aged 8 to 19 years at time of assessment, with the following diagnoses: cystic fibrosis (CF), chronic kidney disease (CKD), congenital heart disease (CHD) or (auto-)immune disease (AID).All participants were patients at the Wilhelmina Children’s Hospital, Utrecht, the Netherlands. As determined by the treating physician, all participants were in stable stage of their chronic condition (assessment took place at least one year after diagnosis) [12].
Children with medically unexplained symptoms (MUS) formed a distinct group of participants from the PROactive cohort. Children with MUS were included, analyzed and reported separately in order to be able to distinguish between sleep in children with a chronic condition with or without an identified pathophysiological cause. Inclusion criteria for children with MUS were: children aged 8-19 years with MUS and pain or fatigue as their main complaint.
Exclusion criteria for all children with a chronic condition, including MUS, were: cognitive functioning below the level of an eight-year-old child, the inability to understand or read Dutch, the inability to fill out online questionnaires.
Children aged 8-11 years old were assisted by their parents, children 12 years and older filled out the questionnaires individually. All children were asked to fill out a questionnaire about sleep once. Items from this questionnaire were based on HBSC-items about sleep, and thus the same questions as healthy controls filled out [19].
Healthy controls
Sleep related data for healthy controls (11-17 years) were collected from the HBSC database [14]. Children filled out a questionnaire about sleep quantity and self-reported sleep quality under supervision of a trained research assistant [15]. Because some of these children might have underlying conditions that influence the results, healthy controls who reported that they had a long-term (>3 months) mental or physical condition or disability were excluded from the analysis. This was done to ensure that the health controls did not include children with a chronic condition.
Outcome measures
Self-reported sleep quantity
Self-reported sleep quantity was measured by ‘sleep duration’ in hours as calculated by ‘the moment you closed your eyes and started sleeping’ minus ‘the moment you woke up’. This information was derived from HBSC sub-questions, according to the HBSC protocol [15].
For each age, sleep quantity was compared to the recommended hours of sleep according to the Consensus statement of the American Academy of Sleep Medicine [7].
Self-reported sleep quality
Self-reported sleep quality was measured by means of the Groningen Sleep Quality Scale, with an acceptable Cronbach’s alpha of 0.71 [19], [20]. Items were: ‘I feel that I slept poorly’, ‘It took me more than half an hour to fall asleep, ‘I feel that I didn’t get enough sleep’, ‘After I woke up, I had trouble falling asleep again’ and ‘I felt rested after waking up in the morning’. Questions were answered using a 5-point liker scale (1=never, 2=almost never, 3=occasionally 4=often 5=(almost) always). The latter item, ‘I felt rested after waking up in the morning’, required recoding in order to be scaled in the same direction as the other items (i.e. a higher score indicates worse self-reported sleep quality). Finally, the average score of the five items was calculated. A self-reported sleep quality score of >3.5 was considered poor sleep quality [14].
All sleep quantity- and quality-related items were asked about the previous week (Sunday to Thursday). Since sleep rhythms are known to be irregular in children and may differ too much from the child’s typical sleep rhythm during the week, only weekdays were included in the analysis [21], [22].
Data Analysis
All statistical analyses were performed using SPSS for Windows, Version 27.0 (IBM Corp. Released 2017. IBM SPSS Statistics. Armonk, NY: IBM Corp.). To determine the normality of the data, Shapiro-Wilk tests were used. Normally distributed data were shown as mean ± standard deviation (SD), frequencies as number (N), %, and non-parametric data as median± interquartile range (IQR). A p-value of less than 0.05 was considered statistically significant. To identify the missing data pattern, the Little’s Missing Completely at Random (MCAR) test was used. Patients who had non-random missing data for sleep quantity and quality variables were excluded from the analysis.
We first examined all completed self-reported sleep quantity and quality solely in the PROactive dataset. Data of children with a chronic condition were divided into primary school (up to 12 years) and secondary school children (13-19 years). This cut-off value was chosen because HBSC uses the same cut-off age value [15].
Descriptives were used to compare the sleep quantity of children with a chronic condition with the Consensus statement of the American Academy of Sleep Medicine [7]. One-way ANOVA with Tukey’s post-hoc testing was used to analyze differences in sleep quantity and quality between diagnosis groups.
Secondly, children with a chronic condition were matched to healthy controls. Analyses were performed between children with a chronic condition other than MUS, children with MUS, and healthy controls. One-way ANOVA with Tukey’s post-hoc testing was used to analyze differences in sleep quantity and quality between children with a chronic condition, MUS and healthy controls. Descriptives were used to present sleep quality as well as calculation of the percentage with poor/good sleep quality (using the definition mentioned above) for each group. Differences for subgroups (diagnosis groups, age and sex), were analyzed using one-way ANOVA with Tukey’s post-hoc testing.