Stef van Buuren’s research while affiliated with Utrecht University and other places

Objective Extremes of prepregnancy maternal BMI increase neonatal mortality and morbidity at term. They also increase the risk of extremely preterm (EP, i.e., <27 weeks' gestational age) births. However, the association between maternal BMI and outcomes for EP babies is poorly understood. Methods We used a cross‐country design, bringing together the following three population‐based, prospective, national EP birth cohorts: EXPRESS (Sweden, 2004–2007); EPICure 2 (UK, 2006); and EPIPAGE 2 (France, 2011). We included all singleton births at 22 to 26 weeks' gestational age with a live fetus at maternal hospital admission. Our exposure was maternal prepregnancy BMI, i.e., underweight, reference, overweight, or obesity. Odds ratios (OR) for survival without severe neonatal morbidity to hospital discharge according to maternal BMI were calculated using logistic regression. Results A total of 1396 babies were born to mothers in the reference group, 140 to those with underweight, 719 to those with overweight, 556 to those with obesity, and 445 to those with missing BMI information. There was no difference in survival without major neonatal morbidity (reference, 22%; underweight, 26%, OR, 1.31, 95% CI: 0.82–2.08; overweight, 23%, OR, 1.00, 95% CI: 0.77–1.29; obesity, 19%, OR, 0.94, 95% CI: 0.70–1.25). Conclusions No associations were seen between maternal BMI and outcomes for EP babies.

Multiple imputation is a popular method for handling missing data, with fully conditional specification (FCS) being one of the predominant imputation approaches for multivariable missingness. Unbiased estimation with standard implementations of multiple imputation depends on assumptions concerning the missingness mechanism (e.g. that data are "missing at random"). The plausibility of these assumptions can only be assessed using subject-matter knowledge, and not data alone. It is therefore important to perform sensitivity analyses to explore the robustness of results to violations of these assumptions (e.g. if the data are in fact "missing not at random"). In this tutorial, we provide a roadmap for conducting sensitivity analysis using the Not at Random Fully Conditional Specification (NARFCS) procedure for multivariate imputation. Using a case study from the Longitudinal Study of Australian Children, we work through the steps involved, from assessing the need to perform the sensitivity analysis, and specifying the NARFCS models and sensitivity parameters, through to implementing NARFCS using FCS procedures in R and Stata.

The lack of a valid and interpretable score to track early child development over time is a primary reason for neglecting child development in policymaking. Many instruments exist, but there is no accepted method for comparing their scores across different ages, samples, and instruments. This paper aims (1) to enhance the Development Score (D-score), a unidimensional scale for early child development, to compare measurements across ages, samples, and instruments, (2) to develop a conversion key that enables the transformation of measurements obtained from existing instruments into a D-score, and (3) to investigate two new measures designed to optimize the quantification of the D-score. Study 1 gathered data from 51 sources in 32 countries among 66,075 children using 18 instruments with 2,211 items. Subject matter experts used the output of the Study-1 true score equating model to create the Global Scales for Early Development Short Form (GSED SF) and Long Form (GSED LF). Study 2 collected additional data on the GSED LF and GSED SF in three countries among 4,374 children. The Study-2 model enables the conversion of measurements into a D-score for 20 different instruments. We propose the D-score as a unifying evaluation unit to reduce fragmentation, simplify measurement, and enhance comparability.

Background The proliferation of instruments that define instrument-specific metrics impedes progress in comparative assessment across populations. This paper explores a method to extract a common metric from related but different instruments and transform the original measurements into scores with a standard unit of measurement. Methods Existing data from four assessment instruments of child development, collected from three different samples of children, were used to create “equate clusters” of items that measure the same behaviour in (slightly) different ways. A probability model was formulated to identify best items and groups to serve as anchors linking the instruments, assuming that items in an anchoring or “active” equate cluster are psychometrically equivalent. Quantification and inspection of item characteristic curves were used to resolve which equate clusters should be active. We simulated the impact of various analytic choices. Results Simulation confirmed the feasibility of creating a common metric from data collected with different instruments from respondent samples with different abilities. The method performed as expected in an application in early childhood development. Conclusions The use of equate clusters is an intuitive and flexible way to establish a common metric across instruments and facilitates the transformation of measurements obtained to a standardized scale. Standardizing instrument scores to a common metric allows for population-level comparisons on a global scale.

Background: Neurodevelopmental trajectories of preterm children may have changed due to changes in care and in society. We aimed to compare neurodevelopmental trajectories in early and moderately late preterm children, measured using the Developmental (D)-score, in two cohorts born 15 years apart. Methods: We included early preterm and moderately late preterm children from two Dutch cohorts (LOLLIPOP, 2002-2003 and ePREM, 2016-2017). ePREM counterparts were matched to LOLLIPOP participants by gestational age and sex. D-score trajectories were summarized by a multilevel model with random intercepts and random slopes, and multigroup analyses were used to test if the intercepts and slopes differed across cohorts. Results: We included 1686 preterm children (1071 moderately late preterm, 615 early preterm) from LOLLIPOP, and matched these with 1686 ePREM counterparts. The neurodevelopmental trajectories of the two cohorts were mostly similar. For early preterm children, we found no statistically significant differences. For moderately late preterm children, both the intercept (43.0 vs. 42.3, p < 0.001) and slope (23.5 vs. 23.9, p = 0.002) showed some, but only clinically minor, differences. Conclusion: Developmental trajectories, measured using the D-score, in the first four years of life are comparable and stable across a period of 15 years for both early and moderately late preterm children. Impact: Neurodevelopmental trajectories are similar for early and moderately late preterm children born 15 years apart and thus seem quite stable in time. The validated Developmental score visualizes these trajectories based on developmental milestone attainment Because of its stability over time, the Developmental score trajectory may aid clinicians in neurodevelopmental assessment of preterm children as this simplifies monitoring and interpretation, similar to a growth chart.

Twenty-seven scientists met for the annual Auxological conference held at Aschau, Germany, to particularly discuss the interaction between social factors and human growth, and to highlight several topics of general interest to the regulation of human growth. Humans are social mammals. Humans show and share personal interests and needs, and are able to strategically adjust size according to social position, with love and hope being prime factors in the regulation of growth. In contrast to Western societies, where body size has been shown to be an important predictor of socioeconomic status, egalitarian societies without formalized hierarchy and material wealth-dependent social status do not appear to similarly integrate body size and social network. Social network structures can be modeled by Monte Carlo simulation. Modeling dominance hierarchies suggests that winner-loser effects play a pivotal role in robust self-organization that transcends the specifics of the individual. Further improvements of the St. Nicolas House analysis using re-sampling/bootstrap techniques yielded encouraging results for exploring dense networks of interacting variables. Customized pediatric growth references, and approaches towards a Digital Rare Disease Growth Chart Library were presented. First attempts with a mobile phone application were presented to investigate the associations between maternal pre-pregnancy overweight, gestational weight gain, and the child’s future motor development. Clinical contributions included growth patterns of individuals with Silver-Russell syndrome, and treatment burden in children with growth hormone deficiency. Contributions on sports highlighted the fallacy inherent in disregarding the biological maturation status when interpreting physical performance outcomes. The meeting explored the complex influence of nutrition and lifestyle on menarcheal age of Lithuanian girls and emphasized regional trends in height of Austrian recruits. Examples of the psychosocial stress caused by the forced migration of modern Kyrgyz children and Polish children after World War II were presented, as well as the effects of nutritional stress during and after World War I. The session concluded with a discussion of recent trends in gun violence affecting children and adolescents in the United States, and aspects of life history theory using the example of "Borderline Personality Disorder." The features of this disorder are consistent with the notion that it reflects a "fast" life history strategy, with higher levels of allostatic load, higher levels of aggression, and greater exposure to both childhood adversity and chronic stress. The results were discussed in light of evolutionary guided research. In all contributions presented here, written informed consent was obtained from all participants in accordance with institutional Human investigation committee guidelines in accordance with the Declaration of Helsinki amended October 2013, after information about the procedures used.