Physical activity and screen time in children and adolescents born very preterm: a cross-sectional study

Introduction

Advances in the field of neonatology have led to both an increase in infants born very preterm (<32 weeks gestational age) and an improved likelihood that these infants will survive the neonatal period [1, 2]. As these children grow through adolescence, concerns move beyond physical health to lifestyle, wellbeing and mental health that may all be compromised, can persist into adulthood and affect activities of daily living, school performance and quality of life [3-6].

Systematic reviews and meta-analyses reveal that those born very preterm or with very low birthweight generally exhibit reduced physical exercise capacity and significantly lower physical activity levels compared to their full-term peers [7-12]. Regular moderate-to-vigorous physical activity is emphasised as vital for improving exercise capacity, and also enhancing social interactions and quality of life [13]. In addition, data from longitudinal studies indicate that increased physical activity in early childhood correlates with better lung function during the growing years [14, 15].

Physical activity and sedentary behaviour including screen time are key lifestyle factors identified by the World Health Organization (WHO) that are associated with various measures of physical and mental health and quality of life [16] . High screen time, an increasing issue especially in adolescents, has been shown to be associated with health harms in general in healthy children and adolescents [17], as well as with cognitive function and behaviour in preterm-born children [18]. More generally, physical inactivity or sedentary time can have negative effects on health outcomes, whether or not physical activity recommendations are being fulfilled [19-21].

In 2021, we conducted a cross-sectional survey, FLiP (“Frühgeborenen Lungen Projekt” / Premature Infant Lung Project), of children and adolescents, born before 32 weeks of gestational age as well as their full-term siblings, primarily to assess respiratory symptoms and treatment burden in this population [22]. Children and their family were also asked about health-related quality of life [23] and lifestyle behaviours as these factors may be associated with or influence pulmonary health in our population. In the present analysis, we aimed to explore the association of very preterm birth with physical activity and screen time. Specifically, we aimed 1) to compare the levels of physical activity and screen time between the very preterm-born and their siblings, 2) to assess the association of respiratory symptoms and chronic conditions with physical activity and screen time, and 3) to identify correlates of physical activity and screen time among the very preterm-born.

Materials and methods

The cross-sectional survey FLiP recruited children born at less than 32 weeks gestation between January 2006 and December 2019, in the greater Zurich area, Switzerland. All participants were part of the Swiss Neonatal Network & Follow-Up Group (SwissNeoNet), a nationwide registry of very preterm children [24]. Of 1720 potentially eligible children, parents of 1401 children with valid postal addresses were invited by mail (May – December 2021) to complete an online survey for their preterm child as well as for one full-term sibling (at least 37 weeks gestation) aged 1 to 18 years and nearest in age to the very preterm-born child (hereafter “controls”). Families who had not completed the survey within two weeks received a reminder call or, if the phone number was not available, a second invitation letter. They could complete the survey online or on paper, in German, English, French or Italian. The present analysis included those participants who were aged at least 5 years. No other exclusion criteria were used in this analysis. The study was approved by the Ethics Committee of the Canton of Zurich, Switzerland (2020–02396). Filling out the survey was considered as providing consent. The FLiP study was powered to assess the prevalence of respiratory symptoms among children born <32 weeks gestation [22].

For external comparison with the general population, we also considered two other cohorts of children and adolescents in Switzerland. The longitudinal cohort study SOPHYA [25] (“Swiss children’s Objectively measured PHYsical Activity”) examined physical activity in healthy 6- to 14-year-olds in Switzerland from 2013 to 2016. SOPHYA assessed physical activity using accelerometer data, and presented summary data in their final report (table 3.1) [26], from which we extracted minutes per day in moderate-to-vigorous physical activity by 2-year age categories. The Ciao Corona study [27], a cohort study of healthy children and adolescents aged 6–17 in Switzerland during the COVID-19 pandemic, assessed physical activity and screen time via questionnaire [28] at various time points from June 2020 to July 2022 (the questions are available in the supplementary material). We extracted the questionnaires from September 2021 to correspond with the timeframe of the FLiP survey, and summarised physical activity and screen time according to the age groups available for SOPHYA.

The outcomes in this analysis were physical activity and screen time. Both were assessed in hours per day, and dichotomised according to whether they met WHO recommendations (at least 1 hour of physical activity per day or less than 2 hours of screen time per day) [16]. The exact wording of the questions regarding physical activity and screen time are presented in the appendix. Additional data collected included participants’ sex and age (later stratified as 5–10 years vs ≥11 years to correspond with age categories used in lifestyle recommendations), gestational age (in weeks, range: 24–31 weeks), socioeconomic status, family unit (family identifier grouping children in the same household), chronic health conditions, presence of respiratory symptoms due to exertion or in daily activities, participation in music lessons, participation in scouts, participation in sports and need for various types of therapy. Socioeconomic status was determined as the sum of both parents’ education level (range from 2, highest education, to 12, lowest education; see appendix). Chronic health conditions were categorised as heart/lung, motor or other (categorisations are provided in the appendix). To assess whether respiratory symptoms affected daily life, parents were asked several questions related to whether their child had cough or wheezing due to physical exertion in the last 12 months or whether cough or wheezing restricted their daily activities.

Demographic variables were summarised as median (range) or n (%). Outcomes were compared between very preterm-born children and their full-term siblings using linear or logistic mixed models, including age group and sex as well as family unit as a random effect. Coefficients (or odds ratios) and corresponding 95% confidence intervals (CI) were interpreted according to their possible relevance, rather than using p-values [29, 30]. To explore other potential correlates of physical activity and screen time among very preterm-born children, we used conditional inference trees [31, 32] estimated by binary recursive partitioning, using variables that we preselected for consideration in the model. To handle missing predictor values, the conditional inference trees used up to 3 surrogate splits [31]. The algorithm stopped if no split with $\alpha$ <0.05 could be constructed or if a subgroup had fewer than 25 participants. The statistical analysis was performed using R version 4.6.0 (2026-04-24). Mixed models were fitted using the lmerTest package [33] and classification trees were fitted using the ctree function from partykit [34]. Descriptive tables were produced with gtsummary [35].

Results

We examined data from 482 very preterm-born children and 149 of their full-term siblings aged 5 years or older (table 1). Among the very preterm-born subjects, 22% had any chronic health condition, including 9% with chronic motor conditions, 9% with chronic heart or lung conditions and 8% with other chronic health conditions. In comparison, 7% of the full-term siblings reported any chronic health condition. Median physical activity was 0.7 hours per day, regardless of subgroup (table 2, figure 1), though somewhat higher at 0.9 hours per day among control siblings younger than 11 years old. Median screen time was 1.1 hours per day among very preterm-born participants compared with 1.0 hour per day among their siblings (table 2, figure 1), although more screen time was observed among adolescents (median 2 hours per day among very preterm-born, and 2.2 hours in siblings) than among those younger than 11 years (0.9 in very preterm-born, 0.6 in siblings).

Neither physical activity or screen time had meaningful differences between very preterm-born children and their full-term siblings, regardless of age group. Comparing children meeting the WHO recommendations for physical activity and screen time, we found no differences between preterm-born children and their siblings (table 2). Comparing very preterm and control siblings after adjusting for age group, sex and family unit, average physical activity in preterm children was the same as that of their control siblings (mean difference: 0.0063 hours, 95% CI: -0.10–-0.12) and screen time was 0.13 hours lower among preterm children (95% CI: -0.32–0.05). Further stratified analysis by gestational age, sex, heart or lung conditions, or motor conditions revealed similar levels of both physical activity and screen time in very preterm-born children and their siblings (tables S1, S2, S3 and S4). We also compared physical activity among very preterm-born children, by chronic motor conditions or respiratory symptoms, adjusting for age and sex, and observed only small differences between these subgroups. Those with chronic motor (heart or lung) conditions had slightly less physical activity than those without such conditions (chronic motor conditions mean difference: -0.21 hours, 95% CI: -0.49–0.07, table S5; chronic heart or lung conditions mean difference: -0.14 hours, 95% CI: -0.12–0.39, table S6), while those with respiratory symptoms had 0.07 hours less physical activity than those without respiratory symptoms (95% CI: -0.26–0.11, table S7).

Comparing physical activity and screen time from FLiP with those of two other comparator cohorts from Switzerland, children and adolescents from the SOPHYA and Ciao Corona studies indicated higher levels of physical activity (e.g. among 10–11 year olds, 74% of those in Ciao Corona met physical activity requirements, compared with 65% in SOPHYA, but only 29% of very preterm-born children in our sample and 39% of their siblings), with smaller differences in adolescents (e.g. among 14–16 year olds, 60% of those in Ciao Corona met physical activity requirements, compared with 22% in SOPHYA, but only 23% of very preterm-born children in our sample and 31% of their siblings) (figure 2, and supplementary table S8). No differences among the cohorts were observed with respect to screen time.

Finally, none of the variables presented in table S9 were identified as significant predictors of physical activity. For screen time, age group was the most important predictor, followed by socioeconomic status, with those aged 11 years or over having an average of 2.3 hours of screen time, compared with 0.9 hours for those aged under 11 of high socioeconomic status and 1.7 hours for those aged under 11 of low socioeconomic status (figure S1).

Discussion

We observed no difference in physical activity or screen time between very preterm-born children and their full-term siblings, even after accounting for age group, sex and family unit. But both groups had equally low adherence to WHO guidelines for physical activity, and adolescents especially were unlikely to meet guidelines for screen time. The large variability in physical activity and screen time in very preterm-born children and adolescents was not explained by the presence of respiratory symptoms or chronic conditions. In a classification tree analysis, screen time was primarily positively correlated with age and negatively with socioeconomic status. No correlates of physical activity were identified in our analysis.

Our results are in line with similar studies comparing preterm-born children and adolescents to their peers or siblings. It has previously been reported that difference in physical activity between preterm-born children and their peers is small; however, even these small differences in physical activity could be related to reduced exercise capacity or fixed or exercise-induced bronchoconstriction in the preterm [8]. A study of 11- and 15-year-old schoolchildren found only minimal differences in physical activity and sedentary time associated with gestational age [36], and a study of 9- to 11-year-old children found no difference in participation in sports activities, in hours spent playing outside or in sedentary activities between preterm-born children and their peers [37]. More generally, recent studies of physical activity or screen time in preterm-born children and adolescents show much heterogeneity (tables S10 and S11), with some studies showing differences in physical activity of preterm-born children compared to their peers, while others showed little difference. There were however few studies pertaining to screen time in this population (table S11).

Nevertheless, our sample showed lower physical activity than recommended by the WHO [16]. We therefore examined other comparative data on lifestyle in children and adolescents in Switzerland. Data from the SOPHYA study [26] indicated higher levels of physical activity in children than we have observed here in younger children, though the differences were smaller in adolescents (figure 3). SOPHYA however used accelerometry data to measure physical activity, while FLiP used questionnaires. Additionally, data from the Ciao Corona study, a cohort study of healthy children and adolescents in Switzerland during the COVID-19 pandemic, indicated higher adherence to screen time recommendations among children 5–9 years of age, even before the pandemic [28]. Using lifestyle questions from Ciao Corona taken in September 2021 to correspond to the timeframe of the FLiP survey, it appeared that they had much higher rates of fulfilling physical activity recommendations than among those in the FLiP cohort, regardless of preterm birth status. There appears, however, to be no difference between the FLiP and Ciao Corona cohorts with regards to screen time.

While the very preterm-born children in our sample had lower physical activity than in these similar studies, so did their full-term born siblings. Self-reporting of physical activity likely led to the same bias in both the very preterm-born group and in their full-term siblings, as both were filled out by the same parents. Differences in physical activity between these three cohorts are at least partly explained by different measurement methods, different wording of questions or possibly different timeframes. Differences between FLiP and the other cohorts may partially be explained by the COVID-19 pandemic-related shielding of children with bronchopulmonary dysplasia or other conditions, despite no clear signal for higher risks, and this shielding may well have included their siblings. Nevertheless, Switzerland had only a short period of school closures – from March to May 2020 – and by 2021 children and adolescents had few pandemic restrictions. While the COVID-19 pandemic may have led to increased screen time, there was no restriction on physical activity by 2021.

If the lower rates of meeting physical activity recommendations we have observed in the FLiP cohort are accurate, it may be that the repercussions of premature birth for family life beyond infancy and throughout childhood [38, 39] include siblings as well [40]. Such consequences could potentially extend to physical activity and screen time for the family as a whole. Even those previously preterm-born children without respiratory and chronic disease are highly inactive and have low fitness [41, 42]. The restriction of physical activity might be explained by overprotection, fear and/or ignorance of parents, teachers, health practitioners and social self-isolation [43]. If so, an intervention by health professionals clarifying the misconceptions related to the safety and benefits of physical activity would benefit these families including also the siblings. Alternatively, reduced exercise capacity in preterm-born children [41, 42, 44-46] could impact physical activity in the family setting.

Our analysis has several strengths. The analysis used a relatively large registry of very preterm-born children in Switzerland and included full-term siblings as a control group. Physical activity and screen time were considered both as continuous outcomes in hours per day and whether they met WHO recommendations. Family unit was included in the analysis, accounting for a number of family-specific potential confounders that may or may not have been measured (e.g. socioeconomic status, parents’ physical activity, school policies, access to outdoor recreation areas, etc), as well as age group and sex. Nonetheless, there may be additional unmeasured confounding due to factors such as access to recreation. Unmeasured factors such as parental physical activity and school policies have been accounted for by adjusting for family unit. Further, Swiss schools place great emphasis on physical activity and being outdoors, in that, for example, field trips in the forest or school-wide excursions with walking between 3 and 12 kilometres, are common, even in primary school. To account for differing levels of prematurity, we stratified the analysis by gestational age. We considered a broad range of possible correlates, including respiratory symptoms, bronchopulmonary dysplasia and various chronic conditions, in a classification tree analysis, which to our knowledge has not previously been performed in this population.

Key limitations are that longitudinal data on lifestyle were not available, and not all very preterm-born participants had control siblings. No causal conclusions can be made on this cross-sectional data. We did not have information on other potentially important variables, for example, on access to recreational facilities [47] or on parents’ screen time [48], which could have provided useful information in the conditional inference tree analysis. There were also not many full-term children with chronic health conditions in the FLiP sample, which would have allowed us to further explore the associations between very preterm birth, chronic health conditions and lifestyle. Like other studies of very preterm-born children, our sample may have had selection bias [49, 50].