The chemicals in question are ubiquitous in modern life. Phthalates, which make plastics flexible, appear in food packaging, pharmaceuticals, dietary supplements, and personal care products. Phenols, including bisphenol A, bisphenol S, parabens, benzophenone-3, and triclosan, are found in everything from water bottles and receipt paper to cosmetics and antibacterial soaps. Because these substances are metabolized and excreted within hours to days, they are classified as non-persistent, yet the near-constant stream of exposure means biomonitoring studies routinely detect them in more than 90 percent of pregnant women in the United States. That combination of ubiquity and biological transience is precisely what makes studying their health effects so challenging.

The scientific rationale for concern is grounded in developmental biology. During embryonic and fetal development, organogenesis and tissue differentiation are exquisitely sensitive to hormonal signals, and endocrine-disrupting chemicals can interfere with the hormonal regulation of metabolism and energy balance. Phthalates, for example, can activate nuclear hormone receptors, particularly peroxisome proliferator-activated receptors, which govern adipogenesis, lipid metabolism, and energy storage. Activation of PPAR-gamma in the developing fetus may promote adipocyte differentiation and alter the production of adipokines such as leptin and adiponectin, potentially predisposing offspring to altered adiposity trajectories after birth. Some phthalates also exhibit anti-androgenic and weakly estrogenic activity, disrupting hormone signaling in utero, and previous research has linked prenatal phthalate exposure to altered steroid hormone profiles in humans as well as adverse metabolic outcomes in animal models.

To interrogate these questions at scale, the research team, led by Dorothy Nakiwala of the University of Colorado Anschutz Medical Campus, drew on 586 mother-child pairs from three ECHO cohort sites: the PROTECT Study in Puerto Rico, which recruited between 2010 and 2018; the Healthy Start Study in Colorado, recruited from 2009 to 2014; and the Illinois Kids Development Study, recruited from 2013 to 2018. All births occurred between 2009 and 2020. Maternal urine samples collected during pregnancy were shipped frozen to the Centers for Disease Control and Prevention, where concentrations of the phenol and phthalate metabolites were quantified using established laboratory methods. Children’s weight and height were measured primarily during clinic visits, and body mass index was converted to sex- and age-specific z-scores using the CDC 2000 growth charts, with assessments occurring between ages two and five at a mean age of 3.96 years.

The methodological architecture of the study is where it distinguishes itself from much of the prior literature. Rather than relying on a single statistical approach, the researchers triangulated their findings using covariate-adjusted single-pollutant linear regression alongside two complementary mixture methods: Bayesian Weighted Quantile Sum regression and Bayesian Kernel Machine Regression. The BWQS approach collapses multiple correlated exposures into a single weighted index under the assumption of a monotonic dose-response relationship, while BKMR uses a kernel function to flexibly model non-linear and interactive exposure-response patterns without imposing such assumptions. Models were adjusted for child sex and age, maternal age at delivery, pre-pregnancy BMI, parity, maternal race, maternal education, and study site, with covariate selection guided by a directed acyclic graph. Exposures were modeled both as continuous, log-transformed variables and as tertile categories to allow detection of threshold effects.

The headline result was null. When the entire 23-chemical mixture was evaluated, neither BWQS nor BKMR identified a statistically significant association with childhood BMI z-scores. The BWQS posterior mean estimate for a one-tertile increase in the mixture index was −0.06, with a credible interval spanning zero, and sex-stratified estimates of −0.24 for boys and 0.03 for girls were similarly inconclusive. The BKMR overall risk function, comparing joint increases in all exposures from the 25th to the 75th percentile against the 25th percentile, showed a non-significant, non-linear relationship with wide uncertainty intervals. There was also no evidence of interactions among pollutants, meaning the effect of any single chemical did not change appreciably as levels of the others rose.

The single-pollutant analyses, however, told a subtler story. When exposures were treated as continuous variables, no significant associations emerged. But when chemicals were categorized into tertiles, non-monotonic patterns appeared. Mono-benzyl phthalate, a metabolite of benzyl butyl phthalate, showed an inverse association: children whose mothers fell in the highest exposure tertile had BMI z-scores 0.33 lower than those in the lowest tertile, with a 95 percent confidence interval of −0.60 to −0.06. Mono-ethyl phthalate, a metabolite of diethyl phthalate, showed the opposite tendency, with the second tertile associated with a BMI z-score 0.27 higher than the first. Most strikingly, among boys only, mono-n-butyl phthalate in the second tertile was associated with a BMI z-score 0.45 higher than the first tertile, a non-monotonic pattern in which the third tertile showed no significant elevation. Among girls, the mono-ethyl phthalate pattern mirrored the pooled results but did not reach statistical significance.

The researchers are careful to contextualize these effect sizes. A shift of 0.1 to 0.2 standard deviations in BMI z-score may not be clinically meaningful for an individual child, but at the population level, even modest average shifts can increase the proportion of children classified as overweight or obese. Given that prenatal exposure to phenols and phthalates is essentially universal, small average effects could translate into a meaningful public health burden. The inverse association with mono-benzyl phthalate, meanwhile, aligns with at least one prior study, and one hypothesis holds that certain phthalates may impair growth by selectively disrupting muscle development rather than promoting fat accumulation, which could explain why phthalate exposure sometimes correlates with lower rather than higher BMI. The positive associations with mono-ethyl phthalate and mono-n-butyl phthalate, both low molecular weight phthalates found in pharmaceuticals, supplements, and personal care products, echo earlier reports linking prenatal mono-ethyl phthalate exposure to higher BMI trajectories from ages two to fourteen.

The consistency of the null mixture findings with previous research is itself informative. Studies using BKMR by Güil-Oumrait and colleagues, Berger and colleagues, and Ouidir and colleagues similarly reported no significant associations between prenatal phenol-phthalate mixtures and BMI z-scores at ages ten, five, and three, respectively, while cumulative-exposure studies using weighted quantile sum approaches by Svensson and Montazeri and their teams also found null results. One plausible explanation lies in exposure misclassification: because these chemicals clear the body so rapidly, spot urine samples may not adequately represent exposure across an entire pregnancy, and random misclassification typically biases results toward the null. The authors also note that BMI, their primary outcome, reflects overall body size without distinguishing fat from lean mass, potentially masking specific adiposity pathways, and that residual confounding by dietary behaviors associated with packaged and processed food consumption cannot be excluded.

Looking forward, the study’s authors argue that future research should prioritize repeated biomonitoring throughout pregnancy to better characterize exposures that fluctuate substantially over time, along with adequate statistical power to evaluate sex-specific effects, longitudinal follow-up, and direct measures of body composition. They also emphasize that postnatal and concurrent childhood exposures deserve attention, particularly given the short biological half-lives of these compounds. For now, the message is one of calibrated caution: the strongest evidence to date finds no cumulative effect of prenatal phenol-phthalate mixtures on early childhood BMI, but suggestive non-monotonic and sex-specific signals for individual chemicals underscore that the obesogen hypothesis, and the chemistry of everyday products that underpins it, remains a question science has not yet closed.

Subject of Research: Prenatal exposure to endocrine-disrupting phenols and phthalates and early childhood body mass index

Article Title: Prenatal exposure to nonpersistent endocrine disruptors and early childhood BMI: single pollutant and mixture analyses from the ECHO Program

Article References: Nakiwala, D., Perng, W., Barrett, E. S., Niu, Z., Alshawabkeh, A. N., Meeker, J. D., Dabelea, D., Starling, A. P., for the ECHO Cohort Consortium, Smith, P. B., Newby, L. K., Adair, L., Jacobson, L. P., Catellier, D., McGrath, M., Douglas, C., Duggal, P., Knapp, E., Kress, A., … Smith, L. M. (2026). Prenatal exposure to nonpersistent endocrine disruptors and early childhood BMI: single pollutant and mixture analyses from the ECHO Program. Pediatric Research. https://doi.org/10.1038/s41390-026-05466-7

Image Credits: AI Generated

DOI: 10.1038/s41390-026-05466-7

Keywords: endocrine-disrupting chemicals, phthalates, phenols, prenatal exposure, childhood obesity, BMI z-score, ECHO Program, chemical mixtures, BKMR, BWQS, developmental programming, biomonitoring

The study, published in JAMA Network Open, evaluated a community-based behavioral intervention called H2GO!, which was designed to reduce children’s consumption of sugar-sweetened beverages and encourage them to drink more water. Over a 12-month follow-up period, children who participated in the intervention showed a significantly greater reduction in their body mass index z-score, or zBMI, compared with peers at comparison sites. The zBMI measure is a standard tool in pediatric research because it tracks BMI in growing children while accounting for age and sex, allowing researchers to assess whether a child’s weight status is improving or worsening relative to growth expectations rather than relying on raw BMI numbers alone.

Why focus on sugary drinks? Sugar-sweetened beverages, including regular sodas, fruit-flavored drinks, sports drinks, and energy drinks, are the largest source of added sugars in children’s diets. These liquid calories carry a particular metabolic liability: they do not produce the same sensation of fullness as solid food, so children who consume them tend not to compensate by eating less at their next meal. The result is an excess of total daily calories that can drive weight gain over time and raise the risk of developing type 2 diabetes, cardiovascular disease, and dental decay. Because beverage consumption patterns are established during formative life stages, and because intake remains stubbornly high among children and adolescents, sugary drinks represent an attractive and concrete target for prevention.

The intervention was delivered at Boys & Girls Club sites across Massachusetts, an organization that serves more than four million school-aged youth nationwide through academic, social, and leadership programming at roughly 5,500 locations. Trained club staff led 12 in-person group sessions with 437 children ages 8 to 13, each participating alongside a parent or caregiver. The six-week program, conducted in multiple waves between 2020 and 2025, blended interactive education with youth engagement: children took part in taste tests and sugar measurement demonstrations that made the hidden sugar content of common drinks visible and tangible, created print, music, and video projects explaining why drinking water mattered to them personally, and completed take-home activities with their parents to reinforce the message outside the club.

The study population reflected the communities that bear a disproportionate share of the childhood obesity burden. More than two-thirds of participating children qualified for free or reduced-price lunch, and more than half identified as Hispanic or Black. These communities are more likely to face structural barriers to maintaining a healthy weight, including economic constraints, neighborhood and environmental conditions that limit access to affordable healthy foods and safe opportunities for physical activity, and unequal access to preventive healthcare. The research team deliberately designed H2GO! to work within a trusted youth-based setting where children already develop everyday habits, rather than expecting families to seek out clinical services or navigate unfamiliar systems.

That design philosophy grew out of the team’s formative work with families and community partners. Corresponding author Dr. Monica Wang, associate professor of community health sciences at Boston University School of Public Health, who co-led the study with Selenne Alatorre, a senior research project manager at BUSPH at the time of the study, explained that asking families to change many dietary behaviors at once can quickly become overwhelming, particularly for those facing competing time and financial pressures. It is also difficult for a child, or an adult, to simply choose differently when sugary drinks are everywhere in the environment. The team focused instead on one simple, concrete change that did not require a costly or complicated overhaul of family life: drink fewer sugary beverages and more water.

The results at the 12-month follow-up were striking. At the start of the study, children consumed an average of 12 ounces of sugar-sweetened beverages and 24 ounces of water per day. By the end of the follow-up period, children who received the H2GO! intervention reported reducing their daily sugary drink intake by an average of 3.6 ounces while increasing daily water consumption by an average of 6.6 ounces. In exploratory analyses, nearly 30 percent of participating children shifted to a lower BMI category over the 12 months, compared with 18 percent of children at comparison sites. Among children who began the study with overweight, 44.4 percent of those in the intervention moved into the healthy-weight category, compared with 31.6 percent at comparison sites, and fewer progressed to obesity, at 5.6 percent versus 21.1 percent. Among children who began with obesity, 47.5 percent of those in H2GO! moved to a lower BMI category, compared with 36.7 percent at comparison sites.

Dr. Wang described one of the most striking findings as the ability to shift participants’ BMI in a healthier direction at a time when national childhood obesity trends were moving the opposite way, a signal that well-designed, community-based behavioral interventions can make a meaningful difference even against a difficult broader backdrop. The results also reaffirm that improving children’s health happens far beyond the pediatrician’s office. Community organizations, she noted, are critical partners in prevention because they are where children spend time, build relationships, and learn from one another. Embedding evidence-based health programming in settings children already trust may be one of the most practical routes to reaching families who face the greatest barriers to clinical care.

The trial builds on an earlier H2GO! pilot study conducted over six months at two Boys & Girls Club sites from 2016 to 2017, which also produced lower zBMI scores and reduced sugary drink intake among participants. The latest multi-site randomized controlled trial expanded the approach across the Commonwealth of Massachusetts, demonstrating that the community-tailored model is both effective and scalable. The intervention culminated in a community event where children showcased the work they had created and the lessons they had learned, reinforcing ownership of the health messages they had crafted themselves. The study was funded by the National Institute of Diabetes and Digestive and Kidney Diseases and coauthored by researchers at Boston University School of Public Health, Harvard T.H. Chan School of Public Health, Boston Medical Center, the University of Connecticut School of Social Work, the Massachusetts Alliance of Boys & Girls Clubs, and the University of Massachusetts T.H. Chan School of Medicine.

The researchers are careful to note the limitations of the work. Measurement differences and reliance on self-reported sugary beverage and water intake mean that more research is needed to understand whether additional factors may have contributed to the observed changes in children’s BMI trajectories. Future studies, the team says, should explore how this type of intervention could be delivered in other community settings and applied to other dietary behaviors, such as the consumption of ultra-processed foods. Still, the central message is an encouraging one for public health: large-scale policy and environmental changes remain essential, but communities do not have to wait for them to make progress. By giving children and families practical tools to make healthier choices within the environments they live in now, programs like H2GO! demonstrate that acknowledging structural barriers and strengthening the agency that families and communities already have can go hand in hand—and that sometimes the most powerful intervention is also the simplest one.

Subject of Research: A community-based behavioral intervention reducing sugar-sweetened beverage consumption and body mass index in school-aged children

Article Title: Afterschool-based intervention helps steer children toward healthier BMI scores

Article References: Afterschool-based intervention helps steer children toward healthier BMI scores. (n.d.). Original publication

Image Credits: AI Generated

DOI: Not provided

Keywords: childhood obesity, sugar-sweetened beverages, BMI z-score, randomized controlled trial, afterschool programs, Boys & Girls Clubs, water consumption, behavioral intervention, public health, health equity, pediatric nutrition, JAMA Network Open

The findings come from First Foods New Zealand, an observational study led by Alice M. Cox of the University of Otago and colleagues, published in the International Journal of Obesity. What sets the work apart is its methodological ambition. Rather than relying on a single measure of eating speed, the team triangulated three distinct approaches: two multiple-pass 24-hour diet recalls capturing 4,405 individual eating occasions, the ‘slowness in eating’ subscale of the Children’s Eating Behaviour Questionnaire completed by caregivers, and, for a subsample of 205 Dunedin families, a GoPro video recording of a real home mealtime. That combination of subjective and objective measurement is rare in infant nutrition research, where most previous studies have depended entirely on parental questionnaires that are vulnerable to recall and reporting bias.

The technical details of the video analysis reveal how demanding this kind of measurement is. Researchers defined the start and end of each eating occasion as the first and last time any complementary food passed through the infant’s lips, then subtracted any ‘interruption to the food supply’ longer than 30 seconds, periods when the baby had no food within arm’s reach and was not actively being offered any. Videos were discarded if filming began after eating had started, if the meal appeared to finish early, or if the infant did not eat. The result was 176 analysable mealtime recordings, most filmed in the evening, in which caregivers reported that the occasion was typical for their child.

Across the diet recalls, infants averaged 3.6 eating occasions per day, spent about 19 minutes eating at each one, and consumed roughly 1,322 kilojoules daily from complementary foods, eating at an average speed of 22 kilojoules per minute. The associations with total intake were striking. Each 10 kilojoule per minute increase in eating speed was linked to an extra 253 kilojoules consumed per day, and each additional 10 minutes of eating duration was associated with 198 more kilojoules daily. Because the average eating speed in the sample was itself only 22 kilojoules per minute, a 10 kilojoule per minute difference represents roughly half of typical eating speed, making the effect size far from trivial. Caregiver perceptions told the same story: each one-unit increase on the ‘slowness in eating’ scale corresponded to 125 fewer kilojoules per day.

Perhaps the most surprising result, however, is what did not appear. Despite the clear link between eating speed and energy intake, eating speed showed no significant relationship with infant weight status as measured by BMI z-scores calculated against World Health Organization growth standards. Only one measure, eating duration from the diet recalls, was associated with BMI, and that weakly: each extra 10 minutes of eating was linked to a slightly lower BMI z-score. The authors caution that several explanations are plausible. Without measures of energy expenditure, it is impossible to know whether faster eating produced genuinely excessive intake, and at 7 to 10 months these infants had only recently been introduced to solid food, so any weight consequences may simply not have had time to emerge. The evidence in older children is also mixed, and nearly all of it, including this study, is cross-sectional, which rules out any claim of causation.

The study also tested a contemporary parental anxiety: the squeeze pouch. Critics, including the German Society for Pediatrics and Adolescent Medicine, have warned that pouches, with their smooth purées and easy-access nozzles, might train babies to eat rapidly. The New Zealand data offer little support for that fear in this age group. Infants defined as frequent pouch users, those eating from a pouch five or more times a week, did not eat faster than other babies on any measure. The only significant difference was shorter meal duration on the video measure, about three minutes less, and even that did not translate into faster eating speed. Notably, only around 5 percent of infants in the study actually sucked food directly from the pouch nozzle, so the findings say little about that specific mode of consumption, which the authors suggest should be the focus of future research.

Baby-led weaning told a much clearer story. Infants classified as following ‘full’ baby-led weaning, meaning they mostly or always fed themselves rather than being spoon-fed, spent nearly four minutes longer per eating occasion and ate about 3 kilojoules per minute more slowly than traditionally spoon-fed infants according to the diet recalls, with similar patterns in the video data. On the caregiver questionnaire, the difference was moderate in size, a standardised difference of 0.6 standard deviations. This is the first published evidence on eating speed and baby-led weaning in infants this young, and it aligns with a retrospective study of slightly older infants, though two studies in toddlers aged one to four found no lasting differences, possibly because of recall bias in how feeding method was measured.

Milk-feeding status mattered too. Breastfed infants, who made up two-thirds of the sample, ate complementary foods more slowly than formula-fed infants on every measure: about 3 kilojoules per minute slower in the recalls, 9 kilojoules per minute slower on video, with 3.4 minutes longer eating duration and higher ‘slowness in eating’ scores. One plausible mechanism is overlap between feeding styles, since baby-led weaning was more prevalent among breastfed infants in this cohort, and previous research has consistently linked breastfeeding with self-feeding approaches. The study could not determine whether breastfed babies were fed milk immediately before meals, which might also influence how eagerly they attacked solid food.

The appetite psychology underneath these behaviours is equally revealing. Faster eating, measured objectively, was correlated with lower satiety responsiveness and lower food fussiness, both food-avoidance traits, and with higher food responsiveness and higher enjoyment of food, both food-approach traits. In other words, the babies who ate fastest were also the ones most driven by food and least sensitive to fullness cues, a combination that in older children predicts higher adiposity risk. The correlations were statistically robust but individually weak, each below 0.3 standard deviations, and eating duration showed fewer and less consistent associations with these traits, hinting that speed and duration, though related, are genuinely distinct behaviours that may follow different developmental paths.

The study’s strengths, a large, ethnically and socioeconomically diverse sample, careful dietary assessment using the Multiple Source Method to estimate usual intake, standardised anthropometry, and naturalistic video measurement, make it a landmark first description of eating speed during the earliest phase of complementary feeding. Its limitations are equally clear: a cross-sectional design that cannot establish direction, a video subsample that was younger and less ethnically diverse than the full cohort, and a single filmed meal that may not represent every baby’s typical behaviour, though most caregivers rated the occasion as usual and the video estimates closely matched the recall data. What the research ultimately delivers is a provocative early snapshot. The seeds of eating pace, and its links to appetite and intake, appear to be sown in the first months of solid food, shaped in part by whether a baby is handed a spoon or handed the meal itself.

Subject of Research: Eating speed and its associations with energy intake, weight status, breastfeeding, baby food pouch use and baby-led weaning in infants aged 7 to 10 months

Article Title: Eating speed in complementary fed infants: associations with energy intake, weight status, breastfeeding, baby food pouch use and baby-led weaning

Article References: Cox, A. M., Heath, A.-L. M., Haszard, J. J., Conlon, C. A., Beck, K. L., von Hurst, P. R., Te Morenga, L. A., Daniels, L., Jones, E. A., Katiforis, I., Brown, K. J., Rowan, M., Casale, M., McLean, N. H., Bruckner, B. R., Jupiterwala, R. M., & Taylor, R. W. (2026). Eating speed in complementary fed infants: associations with energy intake, weight status, breastfeeding, baby food pouch use and baby-led weaning. International Journal of Obesity. https://doi.org/10.1038/s41366-026-02234-9

Image Credits: AI Generated

DOI: 10.1038/s41366-026-02234-9

Keywords: infant nutrition, eating speed, complementary feeding, baby-led weaning, baby food pouches, breastfeeding, energy intake, childhood obesity, eating behaviour, satiety responsiveness, BMI z-score, First Foods New Zealand

The research team, led by Nu Hoang and Thin Nguyen of Deakin University’s Applied Artificial Intelligence Institute, together with collaborators from the university’s Global Centre for Preventive Health and Nutrition, set out to solve a problem that has long undermined evaluations of community interventions. Traditional statistical approaches, such as linear regression, demand careful selection of covariates to ensure valid causal inference. Choosing the wrong variables, either omitting a genuine confounder or adjusting for an inappropriate one, can bias estimates of a program’s effect. Even with the right covariates, misspecifying the functional form of the model can still produce misleading conclusions. The team’s answer was not a new algorithm but a principled, end-to-end workflow that chains together causal discovery, formal identification through do-calculus, double machine learning, and interpretable subgroup discovery.

The first stage of the pipeline is causal discovery, the task of inferring cause-and-effect relationships directly from data rather than assuming them from prior theory. The researchers employed a hybrid causal discovery model designed for mixed-type data, meaning it can handle both continuous measurements, such as body mass index, and categorical variables, such as sex. The model works in two phases: a randomized conditional independence test builds the skeleton of the causal graph, and a cross-validation-based scoring function then orients and refines the edges. Because community intervention data are inherently longitudinal, the team adapted the model with temporal constraints, learning a graph over baseline variables first and then forcing edges to flow forward in time. This two-step design proved its worth in validation: it achieved significantly higher log-likelihood on held-out data than a one-step alternative, with a p-value below 0.0001.

The resulting causal graph is a map of how children’s characteristics and behaviors interconnect. Baseline age stood out as the most influential upstream variable, with the highest mean out-degree of 13.4 across bootstrap resamples, linking to physical activity, sedentary behavior, active transport, fruit intake, sweet beverage consumption, takeaway food consumption, and the outcome itself: the change in body mass index z-score over time. Baseline zBMI and sex were the next most recurrent upstream variables. Critically, the confounding pathways from sex, baseline age, and baseline zBMI to program participation appeared with a bootstrap frequency of 1.0, meaning they were perfectly stable across repeated resampling of the data. This stability gave the researchers confidence that the adjustment set identified from the graph could reliably strip away spurious correlations.

With the causal structure established, the team turned to double machine learning, a technique from econometrics that allows flexible machine learning models to be used in causal estimation without sacrificing statistical consistency. The method models both the outcome and the treatment assignment as functions of the covariates, then exploits two tricks to avoid the biases that normally plague machine learning estimators. Cross-fitting splits the data into partitions so that models are never evaluated on the data used to train them, guarding against overfitting bias. Orthogonalization, rooted in the classical Frisch-Waugh-Lovell theorem, regresses out nuisance functions and estimates the causal parameter from residuals, delivering a root-N consistent estimator free of regularization bias. Here, LightGBM models served as the nuisance learners, with hyperparameters tuned by fivefold cross-validated grid search.

The headline result of the estimation stage is a picture of profound heterogeneity. The average treatment effect across all children was a modest reduction of 0.02 in zBMI, with a 95 percent confidence interval spanning from minus 0.09 to plus 0.05, a range that crosses zero. But the individualized estimates told a far richer story, ranging from minus 0.25 to plus 0.15. Roughly 18.7 percent of children had confidence intervals entirely below zero, indicating a genuine benefit, while 8.7 percent had intervals entirely above zero, suggesting the program may have been counterproductive for them. The remaining 72.6 percent had intervals that included zero, reflecting the substantial uncertainty inherent in individual-level causal estimates. The researchers are careful to stress that these individualized figures describe broad patterns of heterogeneity rather than decision-grade predictions for any single child.

To probe what drives this variation, the team first examined linear correlations between the estimated effects and baseline variables. Age showed the strongest association, with a Pearson coefficient of 0.59, while no other baseline variable correlated significantly. Because linear correlation can miss nonlinear structure, the researchers then trained a shallow decision tree to classify children into positive-effect, negative-effect, and no-effect groups based on nine baseline characteristics. The tree partitioned the data using just two variables: age and weekly takeaway food consumption. Children under 8.9 years old formed the most robustly benefited subgroup, with a dominant-class probability of 0.93 and a mean individual treatment effect of minus 0.096, more than four times the pooled average. Children aged 8.9 to 13.6 who ate takeaway food less than once per week also benefited, with a mean effect of minus 0.038. In contrast, adolescents above 15.2 years showed a clearly negative response, with a mean effect of plus 0.018.

The robustness of these findings was tested from multiple angles. Bootstrap validation of the decision tree across 100 resampled datasets confirmed that age was selected as a splitting variable in every single run, while takeaway food consumption appeared in 40 percent of runs and no other variable appeared at all. Refutation tests bolstered the causal estimate itself: a placebo treatment test, which replaces the real treatment with random values, yielded an effect of minus 0.0038 that was statistically indistinguishable from zero, exactly as expected if the original estimate reflects a genuine causal relationship. Subset validation produced an effect of minus 0.016, not significantly different from the original. An E-value sensitivity analysis indicated that an unmeasured confounder would need to be associated with both treatment and outcome by a risk ratio of at least 1.23 to explain away the average effect, though the subgroup effect for younger children is substantially larger and more resilient.

The authors are candid about the limitations of their work. Communities self-selected into the intervention programs, so unmeasured factors such as local political support or socioeconomic resources could confound the results, potentially overstating benefits. Spillover effects, in which children in comparison communities are indirectly exposed to intervention activities, could bias estimates toward the null. The pooled average effect is small and its confidence interval crosses zero, so the researchers emphasize that the study’s main contribution lies in demonstrating how heterogeneity can be identified and characterized, not in claiming a large overall benefit. The individualized estimates are explicitly framed as exploratory and potentially model-dependent, useful for revealing broad subgroup structure rather than for clinical decision-making at the level of a single child.

Even with those caveats, the implications are considerable. If community-based obesity prevention is genuinely most effective for younger children and least effective for older adolescents, then policymakers have a concrete, actionable signal: interventions may need age-sensitive redesign, with different strategies for teenagers than for primary-school children. The interaction analysis adds nuance, finding that the combination of age and takeaway food consumption drives heterogeneity beyond age alone, and that physically active children who eat fewer takeaway meals benefit more. Beyond obesity, the framework is explicitly generalizable. The authors argue that the same workflow, causal discovery to build the graph, do-calculus to identify the adjustment set, double machine learning to estimate effects, and tree-based subgroup analysis to interpret them, could be applied to any complex community intervention where average effects conceal the individuals the program actually helps. In an era when public health budgets are finite and one-size-fits-all programs increasingly look inadequate, that may be the study’s most viral-worthy message: the average is a lie, and the tools to see past it now exist.

Subject of Research: Causal machine learning analysis of heterogeneous treatment effects in community-based childhood obesity prevention interventions

Article Title: Causal machine learning for understanding heterogeneous effects of childhood obesity prevention

Article References: Hoang, N., Nguyen, T., Duong, B., Nichols, M., Brown, V., Backholer, K., Allender, S., & Nguyen, T. (2026). Causal machine learning for understanding heterogeneous effects of childhood obesity prevention. International Journal of Data Science and Analytics, 22(1), Article 326. https://doi.org/10.1007/s41060-026-01156-z

Image Credits: AI Generated

DOI: 10.1007/s41060-026-01156-z

Keywords: causal machine learning, causal discovery, causal inference, childhood obesity, community-based interventions, heterogeneous treatment effects, double machine learning, BMI z-score, public health, age heterogeneity, subgroup analysis, preventive health

The research team set out to address a persistent gap in nutritional science. While numerous studies in adults have connected ultra-processed food intake with overweight and obesity, robust longitudinal evidence in children and adolescents has remained limited. Most existing pediatric studies rely on short follow-up periods or single measurements, making it difficult to establish whether diet drives weight change or whether body weight shapes dietary behavior. The Dutch researchers designed their analysis to capture the dynamic relationship between diet and growth over the crucial transition from early childhood into adolescence, using repeated measurements of height and weight collected by trained nurses rather than self-reported values.

The study drew on two complementary Dutch cohorts. The first, GECKO Drenthe, is a birth cohort coordinated through the Groningen Expert Center for Kids with Obesity, in which 1,091 children, evenly split between boys and girls, had their diets assessed at age three using a validated food frequency questionnaire completed with parental assistance. The second, Lifelines, is a large population-based cohort in which 2,970 children aged eight to twelve, half of them male, provided dietary information through the same validated instrument. In both cohorts, food items were categorized according to the NOVA classification system, the most widely used framework for distinguishing ultra-processed products, which are industrial formulations typically made from refined substances and additives with little resemblance to whole foods.

The scale of ultra-processed food consumption among these Dutch children was striking. In the GECKO cohort, the median intake at age three was 744 grams per day, accounting for roughly 52 percent of total food intake by weight. In Lifelines, children aged eight to twelve consumed a median of 908 grams per day, representing about half of everything they ate. These figures mean that the average Dutch child in these cohorts obtained the majority of their daily food, by weight, from industrially manufactured products, a pattern consistent with broader trends across high-income countries where ultra-processed items dominate children’s diets.

The central finding emerged from the younger cohort. In GECKO Drenthe, children in the highest quartile of ultra-processed food consumption at age three showed a significantly greater increase in their age- and sex-standardized body mass index, known as BMI z-score, between age three and age ten or eleven compared with children in the lowest quartile. The regression coefficient was 0.22, with a 95 percent confidence interval of 0.07 to 0.37, indicating a statistically robust association. BMI z-scores are a standard pediatric metric that expresses a child’s body mass index relative to peers of the same age and sex, allowing researchers to track whether a child’s weight trajectory is drifting upward relative to the population norm over time.

Perhaps the most intriguing result, however, came from the cross-lagged panel models, a statistical technique designed to probe the temporal direction of relationships between variables measured at multiple time points. In the GECKO cohort, these analyses indicated a directional association in which early ultra-processed food intake predicted later BMI z-score, rather than the reverse. This is a critical distinction in observational nutrition research, because heavier children are sometimes assumed to simply eat more of everything, including processed products. The cross-lagged evidence suggests that, at least in early childhood, the arrow of causality points from diet to weight, with high ultra-processed food consumption preceding and predicting subsequent weight gain rather than resulting from it.

The picture in older children was more complicated. In the Lifelines cohort, higher ultra-processed food intake at ages eight to twelve was actually associated with lower BMI z-scores measured between ages nine and seventeen. Compared with children in the lowest consumption quartile, those in the second quartile had a coefficient of minus 0.12, those in the third quartile minus 0.20, and those in the highest quartile minus 0.13, all with confidence intervals excluding zero. The authors themselves characterize this inverse association as less consistent, and they caution against interpreting it as evidence that ultra-processed foods protect against weight gain in adolescence. Reverse causation, residual confounding by socioeconomic status or physical activity, and the coarser dietary assessment possible in older children may all contribute to the unexpected pattern.

Several biological and behavioral mechanisms could explain why early childhood appears especially vulnerable to the effects of ultra-processed foods. During the first years of life, the body undergoes rapid growth and metabolic programming, and adiposity trajectories established in this period tend to track into later childhood and adulthood. Ultra-processed products are typically energy-dense, high in added sugars, refined starches and unhealthy fats, and low in fiber and micronutrients, a combination that can promote passive overconsumption because they are palatable, convenient and often marketed in large portions. Early exposure may also shape taste preferences and eating behaviors that persist for years, embedding a dietary pattern that continuously favors excess energy intake. The developing hypothalamic regulation of appetite and satiety may likewise be influenced by the nutrient profile of the early diet, though the authors emphasize that further longitudinal research across developmental stages is needed to confirm these pathways.

The study’s strengths lie in its prospective design, its use of two independent cohorts, objectively measured height and weight collected by trained nurses, and a validated food frequency questionnaire applied consistently across age groups. Standardized BMI z-scores and the International Obesity Task Force criteria for overweight and obesity provided internationally comparable endpoints. Nevertheless, the researchers acknowledge limitations inherent to observational nutrition studies. Food frequency questionnaires depend on parental reporting for young children and self-report for older ones, both of which are subject to measurement error and social desirability bias. The NOVA classification, while influential, groups heterogeneous products together, and residual confounding by family socioeconomic circumstances, physical activity and other dietary components cannot be fully excluded. The inverse association observed in Lifelines in particular underscores the need for caution before drawing firm conclusions about adolescents.

The implications for public health are nonetheless significant. If early childhood is indeed a sensitive period for the obesogenic effects of ultra-processed foods, interventions aimed at reducing these products in toddler and preschool diets could yield disproportionate long-term benefits, potentially preventing the upward drift in body mass index that foreshadows adolescent and adult obesity. The findings add momentum to international efforts, including front-of-pack labeling schemes, marketing restrictions targeting young children and dietary guidelines that explicitly recommend limiting ultra-processed products. As ultra-processed foods continue to expand into low- and middle-income countries, understanding which developmental windows matter most will be essential for designing effective prevention strategies. This Dutch study provides some of the strongest longitudinal evidence yet that the earliest years of life may be exactly when dietary quality matters most for lifelong weight trajectories.

Subject of Research: Longitudinal associations between ultra-processed food consumption and body mass index development from childhood to adolescence

Article Title: Ultra-processed food consumption and body mass index development from childhood to adolescence in Dutch cohorts

Article References: Yang, J., Navis, G., & Corpeleijn, E. (2026). Ultra-processed food consumption and body mass index development from childhood to adolescence in Dutch cohorts. BMC Medicine. https://doi.org/10.1186/s12916-026-05270-4

Image Credits: AI Generated

DOI: 10.1186/s12916-026-05270-4

Keywords: ultra-processed foods, childhood obesity, BMI z-score, NOVA classification, GECKO Drenthe cohort, Lifelines cohort, pediatric nutrition, longitudinal study, weight gain, cross-lagged analysis, public health, Netherlands

The findings carry substantial public health significance. Chronic pain is increasingly recognized as a global burden, with recent analyses of the Global Burden of Diseases study projecting steep rises in chronic pain prevalence through 2035. Meanwhile, child and adolescent overweight and obesity continue to climb worldwide, with forecasts suggesting that a substantial share of children and adolescents will be affected by mid-century. If elevated or rapidly increasing body fat during childhood predisposes young people to persistent, multi-site pain, then the two epidemics may be intertwined from the very start of life, with implications for how clinicians and policymakers approach prevention.

The study drew on Generation XXI, a population-based birth cohort assembled in Porto, Portugal, and re-evaluated at ages 4, 7, 10, and 13. A total of 4,563 participants had the data needed for analysis. Rather than relying on a single snapshot of body weight, the researchers modeled each child’s individual adiposity trajectory across the four follow-up waves, extracting patterns from mixed-effects polynomial trend models. Two complementary adiposity parameters were examined: fat mass percentage, measured through bioelectric impedance, and body mass index z-scores standardized for age and sex.

Pain was assessed at age 13 using the Luebeck Pain Questionnaire, a validated instrument for pediatric pain research. The investigators classified pain in two clinically meaningful ways: chronic pain, defined as pain lasting longer than three months, and multisite pain, defined as pain reported at two or more body sites. Because multisite pain in adolescence is considered an early marker of risk for widespread pain conditions in adulthood, the distinction matters for anyone thinking about long-term trajectories of musculoskeletal health.

The statistical approach was deliberately forward-looking. Instead of simply comparing children who were heavier at any one moment with their peers, the team tested whether the shape of each child’s adiposity curve from ages 4 to 13 predicted pain outcomes at 13. Linear and quadratic terms captured both the overall level and the acceleration of fat gain. Logistic regression models were adjusted for pubertal development, assessed using Tanner staging, and for participation in programmed physical activity, both of which could plausibly confound or modify the adiposity-pain relationship during this developmental window.

The headline result was strikingly sex-specific. Among girls, quadratic fat mass trajectories, characterized by accelerated fat mass gain over childhood, were associated with significantly increased odds of both chronic pain and multisite pain at age 13. The odds ratio for chronic pain was 1.37, with a 95 percent confidence interval of 1.12 to 1.69, and for multisite pain it was 1.27, with a confidence interval of 1.05 to 1.53. In plain terms, girls whose body fat accelerated more steeply than expected across the decade from age 4 were meaningfully more likely to report persistent pain and pain in multiple body sites as they entered adolescence.

Complementary cross-lagged analyses reinforced the temporal direction of the association. These models test whether adiposity at one wave predicts pain at the next, and vice versa, while accounting for the stability of each measure over time. Among girls, the results suggested that adiposity at age 10 was associated with pain reports at age 13, supporting the interpretation that higher adiposity precedes and predicts subsequent pain rather than simply emerging as a consequence of it. This bidirectional framing is important because pain can plausibly drive weight gain through reduced activity and comfort eating, and the cross-lagged design helps disentangle which direction dominates at this age.

Among boys, the picture was far less conclusive. Only suggestive associations emerged, particularly for quadratic body mass index z-score trajectories in relation to chronic pain, with an odds ratio of 1.34 and a confidence interval of 0.96 to 1.86 that crossed the threshold of statistical significance. The authors caution against overinterpreting these patterns, but the sex difference is consistent with a growing body of evidence that puberty reshapes the relationship between adiposity, hormones, and pain. Estrogenic influences on pain processing are well documented, and puberty marks a well-known divergence in which chronic pain becomes substantially more common among girls than boys.

Several biological mechanisms could explain the link. Adipose tissue is not metabolically inert; it is an endocrine organ that secretes inflammatory cytokines and leptin, both of which are implicated in pain sensitization. Excess weight also imposes mechanical loads on joints and muscles, a pathway well established in osteoarthritis research, where weight reduction has been shown to alleviate symptoms in affected adults. In children, high adiposity is associated with reduced physical function and mobility, and prior work from the same cohort has shown that pain sites can persist and migrate between childhood and adolescence. Sleep disturbance, emotional distress, and sedentary behavior may further connect fat gain and pain in a self-reinforcing cycle.

The practical implications are considerable. The results suggest that adiposity trajectories could serve as an early identifier of children at elevated risk of chronic pain, opening a window for intervention years before pain becomes established. Because the association was strongest with fat mass rather than body mass index alone, the study also hints that body composition, not merely weight-for-height, may be the more informative clinical target. Limitations remain: the cohort is from a single European city, pain was self-reported, and observational designs cannot exclude residual confounding. Still, the prospective design, repeated adiposity measurements, and large sample size make this one of the most rigorous demonstrations to date that the obesity-pain connection takes root in childhood, with girls appearing most vulnerable as their bodies and hormonal environments change.

Subject of Research: Association between childhood adiposity trajectories and chronic and multisite pain in adolescence

Article Title: Adiposity trajectory as a predictor of chronic and multisite pain in the transition from childhood to adolescence: A prospective study of a large birth cohort

Article References: Fernandes, F., Pereira, R., Lucas, R., Severo, M., Lopes, C., & Talih, M. (2026). Adiposity trajectory as a predictor of chronic and multisite pain in the transition from childhood to adolescence: A prospective study of a large birth cohort. International Journal of Obesity. https://doi.org/10.1038/s41366-026-02227-8

Image Credits: AI Generated

DOI: 10.1038/s41366-026-02227-8

Keywords: adiposity trajectory, chronic pain, multisite pain, childhood obesity, birth cohort, fat mass percentage, BMI z-score, adolescence, puberty, Generation XXI, longitudinal study, girls

The study, conducted within the Brazilian Unified Health System and reported according to STROBE guidelines, followed 111 caregiver–child dyads recruited through consecutive sampling between 2021 and 2023. Children and adolescents aged 8 to 16 years were eligible if they had a prior diagnosis of obesity, defined as a BMI z-score of 2 or higher under World Health Organization criteria, and medical clearance for regular physical activity. Caregivers, the vast majority of whom were mothers—83.8 percent of the sample—had to live in the same household and attend the child’s follow-up visits. Crucially, the researchers implemented no experimental intervention. Every appointment, assessment, and educational encounter followed the clinic’s existing protocol, making this a portrait of obesity care as it is actually delivered in a resource-constrained public system rather than as it might be delivered in an idealized research setting.

The clinical routine began with a mandatory 90-minute educational session delivered by the multidisciplinary team, introducing families to healthy eating, physical activity, hydration, adequate sleep, and screen-time limits based on Brazilian national guidelines. Families then returned within a month for multidisciplinary evaluations involving pediatricians, physical educators, and nutritionists, with follow-up frequency varying from one to three months according to clinical need. At baseline, the children presented with severe obesity: a median BMI z-score of 3.13, a median BMI of about 30.1 kg/m², and a mean age of 12.1 years. Nearly 61 percent were boys. The comorbidity burden was already substantial—21.6 percent of the children had hypertension, alongside notable rates of asthma, hepatic steatosis, anxiety, and insulin resistance—and their caregivers were hardly healthier, with 40.5 percent reporting hypertension and 23.4 percent diabetes, and a median caregiver BMI of 32.0 kg/m².

Using quantile mixed-effects regression adjusted for sex and age, the team tracked anthropometric and behavioral changes at three and six months. The results were modest but statistically meaningful in the short term. Children’s BMI z-score fell by 0.1 at three months, a significant reduction, and body fat percentage declined by 0.4 percentage points over the same period. By six months, however, the momentum faded: the BMI z-score change shrank to a non-significant 0.04, and the improvement in body fat was no longer sustained. In a cohort this severely affected, the researchers caution against dismissing even small shifts, since stabilization or slight decreases in BMI trajectories have been linked to improvements in cardiometabolic risk profiles in other studies. Still, the trajectory of the data itself—an early dip that flattens—tells a story about the limits of low-contact care.

The behavioral picture was even more static. Across the entire follow-up, light physical activity, moderate-to-vigorous physical activity, screen time, and sleep duration showed no significant variation. At baseline, the median child reported four hours of daily screen time and a median of zero minutes of moderate-to-vigorous physical activity per week—figures that barely budged. In the dietary domain, one variable did move: water intake increased modestly and significantly at both assessments, rising by roughly 0.2 liters per day at three months and 0.1 liters at six months. Total caloric intake and sugary beverage consumption remained flat. The researchers interpret this pattern as evidence that standardized guidance at program entry, plus individualized counseling during follow-up, may raise awareness but is rarely sufficient on its own to disrupt well-established behavioral patterns, particularly among socially vulnerable families facing structural and environmental barriers to change.

Perhaps the most striking finding concerned the caregivers. Despite the consistent evidence from intensive family-based intervention programs that child weight reduction often travels with parental weight change, caregiver body composition in this real-world cohort remained remarkably stable across all six months. Weight, waist circumference, BMI, and body fat percentage showed no significant longitudinal variation. Meanwhile, the cross-sectional correlation between caregiver and child BMI stayed moderate and consistent at every time point—0.293 at baseline, 0.274 at three months, and 0.304 at six months—confirming that families share weight profiles. Yet when the researchers examined whether changes moved together, the synchronization dissolved: changes in pediatric BMI z-score were not significantly correlated with changes in caregiver BMI over time, with a Spearman coefficient of just 0.134. In other words, caregivers and children started from similar places, but they did not travel together.

A responder analysis sharpened that insight. Among the 99 dyads with complete follow-up data, 49.5 percent were classified as responders, having achieved a reduction in BMI z-score of at least 0.1 at six months. The single factor distinguishing responders was age: responders were significantly younger, averaging 11.5 years versus 12.5 years for non-responders, a small-to-moderate effect. Baseline caregiver BMI did not differ between the groups. This points to an early window—before obesity becomes deeply entrenched—as the most favorable period for behavioral adaptation, and it suggests that by the time children reach tertiary care with severe, long-standing obesity, the condition is sustained by an interlocking web of behavioral, environmental, and biological mechanisms. Neuroendocrine adaptations and genetic predisposition can elevate the biological set point and blunt responsiveness to lifestyle-based approaches, while emotional eating—using food to cope with stress or distress—may further undermine behavioral guidance.

The authors are careful to frame these findings not as a failure of clinicians but as an expected output of a pragmatic care model. Compared with structured, high-contact family-based behavioral treatments, which have demonstrated stronger results in trials and primary-care implementations, the routine care evaluated here involved variable consultation frequency and few structured behavioral components. Knowledge-transfer strategies, the evidence suggests, rarely produce sustained lifestyle change without ongoing behavioral support, especially in populations contending with food insecurity, financial constraints, and limited access to safe spaces for physical activity. The stability of caregiver BMI throughout follow-up implies that the shared household routines and environmental constraints shaping the child’s behavior persisted largely untouched, potentially capping the effectiveness of recommendations aimed at the child alone.

The study’s limitations are those inherent to real-world observational designs: no comparison group, a short six-month window, a sample sized by clinic flow, losses to follow-up among 12 dyads, and no socioeconomic data. The findings also cannot isolate the effect of the initial educational session from the surrounding multidisciplinary care. Even so, the takeaway is clear and actionable. Routine multidisciplinary care in this public setting may help prevent further deterioration in children with severe obesity, but achieving substantial improvement will demand more: structured behavioral components, increased contact frequency, active engagement of caregivers as agents of change, stronger early detection and prevention in primary care, and consistent implementation of the public policies Brazil already has on paper. As childhood obesity continues to climb, the gap between what clinics can do alone and what families and health systems must do together has rarely been drawn in sharper relief.

Subject of Research: Longitudinal changes in pediatric obesity and caregiver-child BMI associations during routine multidisciplinary outpatient care

Article Title: Family Dynamics and Longitudinal Trends in Pediatric Obesity: A Real‐World Observational Study in a Public Outpatient Clinic

Article References: de Freitas, F., da Paz, M. M., Zago, M. R., Vitolo, M. R., Antônio, M. Â., & Brandão, M. Â. B. (2026). Family Dynamics and Longitudinal Trends in Pediatric Obesity: A Real‐World Observational Study in a Public Outpatient Clinic. Obesity Science & Practice, 12(5), Article e70192. https://doi.org/10.1002/osp4.70192

Image Credits: AI Generated

DOI: 10.1002/osp4.70192

Keywords: pediatric obesity, BMI z-score, family dynamics, caregiver-child dyads, multidisciplinary care, observational study, public health, lifestyle behaviors, adolescent health, Brazil, real-world evidence, weight management