Stefan Kuhle

Objective: Children with special healthcare needs (SHCN) due to a chronic health condition perform more poorly at school compared with their classmates. We aimed to estimate the effects of past, current, transient, emerging and persistent SHCN on school performance in primary school children.Methods: Data from the German population-based prospective cohort study ikidS were used. The children withSHCN screener was administered before school entry (T1) and at the end of first (T2) and third grade (T3). Grades for German, maths and science (range: 1 (Very Good) to 6 (Failure)) were obtained at the end of third grade (age 8–9 years), and an average grade was calculated. Associations between the timing of SHCN and average grade were estimated by mixed linear regression models adjusted for potential confounding variables.Results: 751 children were included, and 21% had ever SHCN. Children with ever SHCN had poorer school performance than children with never SHCN (adjusted mean difference in average grade [95% CI]: 0.17 [0.06; 0.28]). SHCN in the third year were associated with a poorer average grade (0.29 [0.16; 0.41]) compared with healthy children. Only emerging (0.31 [0.15; 0.48]) and persistent (0.25 [0.07; 0.43]) SHCN were associated with average grade.Conclusions: This study demonstrates the negative effect of current, emerging and persistent SHCN on academic performance in primary school children. Consequently, students should be regularly assessed for SHCN during school age. Timely interventions may help reduce the adverse effects of chronic health conditions on academic achievements in childhood.

Background: Children with chronic illness perform poorer at school, and school well-being (SWB) may mediate this association. We investigated the association between chronic illness and three domains of SWB in children in first grade. Methods: Data from a German population-based prospective cohort study were used. Children with chronic illness were identified via their preschool health examination and follow-up parent surveys during first grade. Children were grouped as either (i) having current special health care needs (SHCN), (ii) having at least one physician diagnosis of a chronic illness but no current SHCN, or (iii) being healthy. SWB was assessed at the end of first grade and measured by the Questionnaire for the Assessment of Emotional and Social School Experiences of First and Second Grade Primary School Children. Based on SWB theory and previous frameworks, the following subscales were used: School-Related Self-Concept, Social Integration, and Joy of Learning. The sum score for each subscale was converted into area-transformed T-values (mean 50 and standard deviation 10). Associations between chronic illness groups and SWB subscales were investigated by multivariable linear regression models. Effect estimates were adjusted for potential confounding variables and standardized mean differences (SMD) were calculated. Results: Of the 1,490 children included, 15% had current SHCN and 37% had a physician diagnosis of a chronic illness but no current SHCN. Compared to healthy children, children with SHCN had lower scores for the School-Related Self-Concept and the Joy of Learning subscale (SMD −0.18 for both) but not for the Social Integration subscale. In contrast, children with a chronic condition but no SHCN had lower scores only for the Social Integration subscale (SMD −0.12). Conclusions: Primary school students with a chronic illness with or without SHCN have lower SWB in some domains compared to their healthy peers. SWB may be a mediator in the association between chronic illness and poor school performance.

The objectives of this study were to examine the total effect of grandmaternal (G0) prepregnancy body mass index (BMI) on infant (G2) birthweight z score and to quantify the mediation role of maternal (G1) prepregnancy BMI. Data were extracted from the Nova Scotia 3G Multigenerational Cohort. The association between G0 prepregnancy BMI and G2 birthweight z score and the mediated effect by G1 prepregnancy BMI were estimated using g-computation with adjustment for confounders identified using a directed acyclic graph and accounting for intermediate confounding. A total of 20 822 G1-G2 dyads from 18 450 G0 participants were included. Relative to G0 normal weight, G0 underweight decreased mean G2 birthweight z score (−0.11; 95% CI, −0.20 to −0.030), whereas G0 overweight and obesity increased mean G2 birthweight z score (0.091 [95% CI, 0.034-0.15] and 0.22 [95% CI, 0.11-0.33], respectively). G1 prepregnancy BMI partly mediated the association, with the largest effect size observed for G0 obesity (0.11; 95% CI, 0.080-0.14). Estimates of the direct effect were close to the null. In conclusion, grandmaternal prepregnancy BMI was associated with infant birthweight z score. Maternal prepregnancy BMI partly mediated the association, suggesting that factors related to BMI may play an important role in the transmission of weight across the maternal line.

This paper critically examines ‘kitchen sink regression’, a practice characterised by the manual or automated selection of variables for a multivariable regression model based on p values or model-based information criteria. We highlight the pitfalls of this method, using examples from perinatal/neonatal medicine, and propose more robust alternatives. The concept of directed acyclic graphs (DAGs) is introduced as a tool for describing and analysing causal relationships. We highlight five key issues with ‘kitchen sink regression’: (1) the disregard for the directionality of variable relationships, (2) the lack of a meaningful causal interpretation of effect estimates from these models, (3) the inflated alpha error rate due to multiple testing, (4) the risk of overfitting and model instability and (5) the disregard for content expertise in model building. We advocate for the use of DAGs to guide variable selection for models that aim to examine associations between a putative risk factor and an outcome and emphasise the need for a more thoughtful and informed use of regression models in medical research.

Background: Antenatal corticosteroids (ACS) are widely prescribed to improve outcomes following preterm birth. Significant knowledge gaps surround their safety, long-term effects, optimal timing and dosage. Almost half of women given ACS give birth outside the “therapeutic window” and have not delivered over 7 days later. Overtreatment with ACS is a concern, as evidence accumulates of risks of unnecessary ACS exposure. Methods: The Consortium for the Study of Pregnancy Treatments (Co-OPT) was established to address research questions surrounding safety of medications in pregnancy. We created an international birth cohort containing information on ACS exposure and pregnancy and neonatal outcomes by combining data from four national/provincial birth registers and one hospital database, and follow-up through linked population-level data from death registers and electronic health records. Results and discussion: The Co-OPT ACS cohort contains 2.28 million pregnancies and babies, born in Finland, Iceland, Israel, Canada and Scotland, between 1990 and 2019. Births from 22 to 45 weeks’ gestation were included; 92.9% were at term (≥ 37 completed weeks). 3.6% of babies were exposed to ACS (67.0% and 77.9% of singleton and multiple births before 34 weeks, respectively). Rates of ACS exposure increased across the study period. Of all ACS-exposed babies, 26.8% were born at term. Longitudinal childhood data were available for 1.64 million live births. Follow-up includes diagnoses of a range of physical and mental disorders from the Finnish Hospital Register, diagnoses of mental, behavioural, and neurodevelopmental disorders from the Icelandic Patient Registers, and preschool reviews from the Scottish Child Health Surveillance Programme. The Co-OPT ACS cohort is the largest international birth cohort to date with data on ACS exposure and maternal, perinatal and childhood outcomes. Its large scale will enable assessment of important rare outcomes such as perinatal mortality, and comprehensive evaluation of the short- and long-term safety and efficacy of ACS.

Objective: To determine risk factors for mechanical (noninfectious) complications in peripherally inserted central catheters (PICCs) in children.Design: Retrospective cohort study.Setting: Pediatric tertiary-care center in Nova Scotia, Canada.Patients: Pediatric patients with a first PICC insertion.Methods: All PICCs inserted between January 2001 until 2016 were included. Age-stratified (neonates vs non-neonates) Fine–Grey competing risk proportional hazard models were used to model the association between each putative risk factor and the time to mechanical complication or removal of the PICC for reasons not related to a mechanical complication. Models were adjusted for confounding variables identified through directed acyclic graphs.Results: Of 3,205 patients with PICCs, 706 had mechanical complications (22% or 14 events/1000 device days). For both neonates and older children, disease group, lumen count, and prior leak were all associated with mechanical complications in the adjusted proportional hazards model. Access vein and prior infection were also associated with mechanical complications for neonates, and age group was associated with mechanical complications among non-neonates.Conclusions: We have identified several risk factors for mechanical complications in patients with PICCs that will help improve best practices for PICC insertion and care.

Objective: To examine the relationship between reported prenatal cannabis use and neonatal and maternal outcomes and whether the legalisation of cannabis in Canada affected the rates of reported use or the association with maternal and neonatal outcomes.Design: Population‐based retrospective cohort study.Setting: Routinely collected data in a real‐world setting.Population: All women in the Canadian province of Nova Scotia with singleton births between 1 January 2004 and 30 June 2021.Methods: The association between cannabis use and maternal and neonatal outcomes was examined using generalised linear models with inverse probability weighting.Main outcome measures: Maternal and neonatal outcomes in the peripartum and postpartum period.Results: Rates of reported cannabis use in pregnancy increased from 1.3% to 7.5% over the study period with no appreciable change in slope after legalisation in 2018. Infants of mothers reporting cannabis use in pregnancy were more likely to have major anomalies and a 5‐minute Apgar score ≤7, require neonatal intensive care unit admission, and had lower birthweight, head circumference and birth length than infants of mothers not reporting cannabis use. These associations did not differ before and after legalisation.Conclusions: Reported cannabis use during pregnancy is associated with early postnatal complications and reduced fetal growth, even after taking into account a range of confounding factors. Rates of reported cannabis use during pregnancy increased over the past 5 years in Nova Scotia with no apparent additional effect of legalisation.