Tracheal intubation, the placement of a tube into the windpipe to secure an infant’s airway, is among the highest-stakes procedures performed on newborns. It requires a clinician to thread a tube through the vocal cords of a patient who may weigh less than a kilogram, often while that patient is critically ill. The procedure is known to carry a substantial burden of adverse events, including oxygen desaturation, bradycardia, and in rare but devastating cases, cardiac arrest. Because newborns have a predominantly parasympathetic cardiovascular system, their hearts are especially vulnerable to vagal stimulation, which can be triggered by laryngoscopy and airway manipulation, slowing the heart rate dangerously.

Atropine works by blocking acetylcholine receptors at the muscarinic sites of the heart, effectively dampening the vagal reflexes that can cause the heart to slow. The logic behind its use during intubation has been straightforward: if airway manipulation can provoke reflex bradycardia, and if severe bradycardia can cascade into cardiac arrest, then preemptively blunting that reflex should protect the infant. That reasoning, drawn largely from pediatric anesthesia and emergency medicine, has made atropine a standard component of premedication regimens in neonatal units around the world, alongside sedatives and, in some centers, neuromuscular blocking agents.

To test that assumption at scale, a team led by Taylor Sawyer of the University of Washington School of Medicine and including co-senior authors Elizabeth Foglia and Akira Nishisaki turned to the NEAR4NEOS registry, an international collaborative that prospectively collects standardized data on every intubation attempt at participating neonatal units. The researchers analyzed 13,085 neonatal tracheal intubations performed between 2014 and 2023, making this one of the largest datasets ever assembled to examine the safety of neonatal airway management. Atropine was administered in 8,189 of those procedures, or 62.6 percent, underscoring just how deeply entrenched the practice is.

The team used multivariable logistic regression to compare outcomes between intubations in which atropine was given and those in which it was not, adjusting for confounding factors such as patient characteristics and clinical context. The primary outcome was cardiac arrest during the procedure. The result was striking: atropine use was not associated with reduced odds of cardiac arrest, with an adjusted odds ratio of 0.52 and a confidence interval spanning from 0.25 to 1.09, a range that crosses the threshold of statistical significance. In other words, the data could not demonstrate that the drug prevents the very complication it is most often given to avert.

The secondary findings were equally revealing. Atropine did deliver on one of its promised effects: it was associated with substantially lower odds of severe bradycardia, with an adjusted odds ratio of 0.33, meaning the drug cut the risk of dangerous heart rate slowing by roughly two-thirds. But that cardiac benefit came with trade-offs. Infants who received atropine had higher odds of oxygen desaturation during the procedure, with an adjusted odds ratio of 1.35, and lower odds of success on the first intubation attempt, with an adjusted odds ratio of 0.76. First-attempt success matters enormously in neonatal care, because repeated attempts prolong airway manipulation, increase physiological stress, and compound the risk of harm.

The desaturation finding has a plausible physiological explanation. Atropine increases heart rate and cardiac output, which can accelerate the consumption of oxygen reserves in an infant whose lungs are being ventilated intermittently or not at all during the attempt. A faster heart rate in a tiny preterm baby may burn through the limited oxygen buffer more quickly, causing saturation levels to fall sooner. Meanwhile, the mechanism behind the reduced first-attempt success is less clear. It may reflect confounding by indication, since clinicians may reach for atropine in sicker, more fragile infants in whom intubation is intrinsically harder, or it may reflect a genuine drug-related effect that further research will need to disentangle.

The study arrives amid a growing body of evidence questioning routine atropine in neonates. A randomized clinical trial published in The Journal of Pediatrics in 2025 by Afifi, El-Naggar, Narvey and colleagues compared atropine with placebo for nonemergent neonatal intubation and found the drug offered no clear advantage, an accompanying commentary asking pointedly whether routine pre-intubation atropine does more harm than good. Earlier work in pediatric critical care, including studies by Jones and colleagues on more than three hundred critical care intubations, had likewise suggested that atropine’s effects on rhythm disturbances during intubation are more complicated than the simple reflex-blocking model implies, and that mortality benefits observed in some pediatric cohorts appeared unrelated to the drug’s influence on bradycardia.

For clinicians, the new findings do not amount to an immediate call to abandon atropine, but they do challenge the reflexive use of the drug in every neonatal intubation. The authors themselves conclude that the association between atropine and increased desaturation and lower first-attempt success warrants further study, and the observational design of the registry analysis means that unmeasured differences between patients who received the drug and those who did not could still influence the results. Randomized trials, such as the recent placebo-controlled study, will remain essential for establishing causation. In the meantime, the results give neonatologists something they have lacked: large-scale data suggesting that the most common cardiac premedication in the NICU may not prevent the worst outcome, and may carry costs of its own.

The broader significance of the study lies in what it says about how neonatal practice evolves. Many interventions in the NICU were adopted on physiological reasoning and small studies rather than rigorous outcome data, and the NEAR4NEOS registry has become a powerful instrument for auditing those practices across hundreds of thousands of recorded procedures. Previous reports from the same collaboration have shown that premedication with sedation and neuromuscular blockade is associated with fewer adverse events, that team stress correlates with complications, and that training level affects success rates. This latest analysis adds a crucial nuance to that picture, suggesting that the individual components of premedication regimens deserve scrutiny on their own terms. As the authors and their international collaborators continue to accumulate data, the question is shifting from whether atropine blunts a reflex, which it clearly does, to whether blunting that reflex, at the cost of faster desaturation and more attempts, actually serves the smallest patients best.

Subject of Research: Atropine use and adverse events during neonatal tracheal intubation

Article Title: Association of atropine use during neonatal intubation with cardiac arrest, severe bradycardia, and other adverse events: a report from NEAR4NEOS

Article References: Sawyer, T., Narvey, M., Tyler, M. D., Abou Mehrem, A., Iben, S., DeMartino, C., Tingay, D., Hodgson, K., Kim, J., Herrick, H., Unrau, J., Puia-Dumitrescu, M., Shay, R., DeMeo, S., Assaad, M.-A., O’Shea, J., Brajkovic, I., Quek, B. H., Glass, K., … DeMartino, C. (2026). Association of atropine use during neonatal intubation with cardiac arrest, severe bradycardia, and other adverse events: a report from NEAR4NEOS. Journal of Perinatology. https://doi.org/10.1038/s41372-026-02916-2

Image Credits: AI Generated

DOI: 10.1038/s41372-026-02916-2

Keywords: atropine, neonatal intubation, cardiac arrest, bradycardia, oxygen desaturation, NEAR4NEOS, premedication, neonatology, NICU, patient safety, Journal of Perinatology, registry study

Type 1 diabetes is an autoimmune disease in which the insulin-producing beta cells of the pancreas are destroyed, meaning patients must rely on externally administered insulin for life. Because carbohydrate is the macronutrient with the most direct influence on blood glucose, the amount of carbohydrate a patient consumes has always been a central variable in insulin dosing. Conventional dietary guidance has generally encouraged moderate or higher carbohydrate intake combined with carefully matched insulin, partly out of concern that severe carbohydrate restriction could provoke dangerous metabolic disturbances in growing children. Yet in recent years a growing number of parents have begun experimenting with low and very low carbohydrate feeding strategies, often reporting striking improvements in their children’s glucose readings and a marked reduction in the amount of insulin needed each day.

Professor Christian Denzer, of the Department of Paediatrics and Adolescent Medicine, Division of Paediatric Endocrinology and Diabetes, at University Medical Center Ulm in Germany, who led the research together with Dr Belinda Lennerz of Boston Children’s Hospital and Harvard Medical School, explained the motivation behind the study. These diets are becoming more popular among people with diabetes, he noted, because patients themselves notice encouraging results, including much better blood sugar control and lower insulin needs. At the same time, he acknowledged that theoretical safety concerns have persisted, including the possibility of nutrient deficiencies, elevated cholesterol, low blood sugar episodes and ketoacidosis, a life-threatening complication of diabetes. In children, there is the additional worry that any restrictive diet could interfere with growth during critical developmental windows.

Until now, the evidence base for such diets in paediatric type 1 diabetes has been thin. Several observational studies and small interventional trials have corroborated the remarkable benefits on blood sugar and insulin requirements that patients report anecdotally, and no major safety events have been published to date. An earlier observational study led by Dr Lennerz, involving 316 children and adults, found major glycaemic benefits with mixed effects on cholesterol and no apparent adverse effect on growth or acute diabetes complications. But those studies were too small and too short to settle the question. What was needed, the researchers argued, was a large, long-term analysis capable of detecting rare complications and subtle effects on growth across a broad paediatric population.

To address this gap, the team turned to a diabetes prospective follow-up registry containing 25 years of longitudinal data collected from paediatric diabetes centres across Germany. From this resource they identified 45,569 children with type 1 diabetes, 55 percent of them male, with an average age of 10.3 years, whose carbohydrate intake had been recorded at least twice over a period of a year or more. This scale makes the analysis the largest of its kind ever performed, dwarfing previous investigations by orders of magnitude and providing the statistical power needed to examine both common outcomes and rare adverse events.

The children were grouped according to the carbohydrate percentage of their estimated energy intake. A very low carbohydrate group, comprising 219 children, consumed less than 10 percent of their energy from carbohydrate. A low carbohydrate group of 4,837 children obtained between 10 and less than 26 percent. The remaining 40,513 children followed a standard or higher carbohydrate diet of at least 26 percent. Average follow-up periods were 2.18 years for the very low carbohydrate group, 2.40 years for the low carbohydrate group and 3.67 years for the standard carbohydrate group. The researchers tracked height and weight, adjusting for the fact that the children were still growing, alongside changes in glycaemic trajectories, insulin use and the rates of the two most feared severe metabolic complications of diabetes: hypoglycaemia and diabetic ketoacidosis.

The analysis did not flag up any safety concerns in the areas examined. There was no evidence that a very low carbohydrate diet stunted growth, and no association between very low carbohydrate intake and either severe hypoglycaemia or diabetic ketoacidosis. This absence of harm signals is significant because both complications represent the principal acute dangers of type 1 diabetes. Severe hypoglycaemia, in which blood glucose falls to dangerously low levels, can cause seizures, loss of consciousness and, in extreme cases, death. Diabetic ketoacidosis arises when the body, starved of insulin, begins breaking down fat at an uncontrolled rate, producing acidic ketone bodies that overwhelm the blood’s buffering capacity and can rapidly become fatal without emergency treatment. Critics of carbohydrate restriction have long argued that pushing children into a state of relative ketosis might tip vulnerable patients toward ketoacidosis, but the registry data found no support for that fear.

On the benefit side, lower carbohydrate intake was associated with several favourable changes. Body mass index increased less over time in children following very low and low carbohydrate diets than in those on standard diets, meaning their weight remained closer to a healthier range as they grew. Levels of HbA1c, the laboratory measure that reflects average blood sugar over the preceding two to three months, were lower at the end of follow-up and rose less steeply over time in the lower carbohydrate groups. Doses of basal insulin, one of the two types of insulin typically prescribed for type 1 diabetes, followed the same pattern, being lower at the end of the observation period and increasing less over time among children eating less carbohydrate.

Most notably, the study examined a composite metric known as IDAA1c, which combines blood sugar levels with the amount of insulin a person requires. Dr Lennerz explained the significance of this measure: a lower IDAA1c means that better blood sugar control was achieved without simply using more insulin, strengthening the finding that very low and low carbohydrate diets were associated with both better glycaemic control and lower insulin requirements. This distinction matters because a patient could in principle achieve normal glucose readings by escalating insulin doses, but doing so increases the risk of hypoglycaemia and weight gain. A dietary approach that improves glucose control while simultaneously reducing insulin need therefore represents a genuinely favourable metabolic profile rather than a trade-off between competing risks.

The analyses were adjusted for factors including sex, age, migration background and diabetes duration, but the authors caution that the observational design leaves room for residual confounding. Families who choose very low carbohydrate diets may differ from other families in ways the registry cannot fully capture, including health consciousness, socioeconomic resources and the accuracy of self-reported dietary intake. Dr Lennerz concluded that reducing carbohydrate intake improved glycaemic control while reducing insulin needs without negatively impacting growth or increasing the rate of acute diabetes complications, describing the results as very encouraging but emphasising that they remain observational in nature. Additional research, she said, is needed to confirm the findings and establish the ideal amount of carbohydrate intake to maximise benefits while avoiding adverse effects and promoting food variety. For now, the study offers the strongest evidence yet that carbohydrate restriction in children with type 1 diabetes is not the hazard many clinicians feared, while leaving open the question of where the optimal balance lies.

Subject of Research: Effects of very low carbohydrate diets on growth and metabolic safety in children with type 1 diabetes

Article Title: Very low carbohydrate diets linked to health benefits in children with type 1 diabetes, largest study of its kind finds

Article References: Very low carbohydrate diets linked to health benefits in children with type 1 diabetes, largest study of its kind finds. (n.d.). Original publication

Image Credits: AI Generated

DOI: Not provided

Keywords: type 1 diabetes, very low carbohydrate diet, paediatrics, HbA1c, IDAA1c, insulin requirement, hypoglycaemia, diabetic ketoacidosis, growth, registry study, EASD, nutrition

Researchers led by Jorge Maese-Calvo of Fundación COMPUTAEX, together with clinicians and investigators across Spain’s Extremadura region, have developed an interpretable machine learning approach that predicts whether an individual heart failure patient will be readmitted within 30 days of discharge. Their work, published in Medical & Biological Engineering & Computing, draws on a remarkable asset: a population-based registry covering two decades of heart failure hospitalizations in a rural Spanish healthcare setting. The registry, spanning the years 2000 to 2019, includes nearly 5,000 admissions and had never before been used for artificial intelligence modeling, making it a uniquely grounded testing ground for predictive algorithms in an underserved population.

The scale of the problem the team tackled is hard to overstate. Heart failure affects an estimated 64 million people worldwide, and its prevalence continues to climb as populations age. In the Spanish cohort studied, 8.2 percent of patients were readmitted within 30 days of discharge. While that figure may sound modest, each readmission represents a clinical setback for the patient, a financial burden for the health system, and a missed opportunity for the kind of targeted post-discharge support that could have kept the person at home. Hospital systems in the United States and Europe have introduced penalty and incentive programs around readmission rates, but risk stratification at the moment of discharge remains crude, often relying on clinicians’ intuition or simple scoring rules that capture only part of the picture.

The researchers’ strategy was deliberately pragmatic. Rather than demanding exotic biomarkers or data types that rural hospitals cannot easily collect, they built their models from routinely gathered clinical variables: demographic details, laboratory values, comorbidities, admission characteristics, and length of stay. Three machine learning algorithms were trained and compared against conventional binary logistic regression, the statistical workhorse that has historically served as the default for clinical risk prediction. The contenders were a random forest, an ensemble method that aggregates hundreds of decision trees; extreme gradient boosting, known as XGBoost, which builds trees sequentially to correct earlier errors; and a support vector machine, which finds optimal boundaries separating patient groups in high-dimensional space.

The results were clear. The random forest model emerged as the strongest performer, achieving an area under the receiver operating characteristic curve, or AUC, of 0.812, with a 95 percent confidence interval of 0.744 to 0.867. An AUC of 0.5 indicates performance no better than a coin flip, while 1.0 represents perfect discrimination; a value above 0.80 reflects clinically useful discrimination. By contrast, binary logistic regression managed an AUC of 0.686, with a confidence interval of 0.617 to 0.755. The gap matters because clinical prediction is unforgiving: a model that merely matches traditional statistics offers no reason to change practice, but a model that meaningfully outperforms it can reshape how discharge decisions are made. The advantage of tree-based ensembles lies in their ability to capture nonlinear relationships and interactions between variables, such as the way renal dysfunction may amplify risk differently depending on a patient’s age or prior cardiac history, relationships that rigid regression forms can miss.

What distinguishes this study from many AI-in-medicine efforts is its insistence on interpretability. Black-box models have faced justified skepticism from clinicians who need to understand why an algorithm flags a particular patient as high risk before acting on that information. To open the box, the team applied SHAP, or SHapley Additive exPlanations, a technique borrowed from cooperative game theory that assigns each predictor a quantified contribution to every individual prediction. SHAP values reveal not only which variables matter most overall but also the direction and consistency of their effects, exposing whether, for instance, longer hospital stays push risk up or down across the cohort or whether the relationship shifts in different patient subgroups. This transparency, the authors argue, is what transforms a prediction from an oracle’s pronouncement into a clinical instrument.

The SHAP analysis identified a compact set of dominant predictors. The period in which the patient was admitted emerged as influential, reflecting how the clinical profile and outcomes of heart failure hospitalizations evolved across the two decades covered by the registry, as treatment practices and diagnostic standards changed. Markers of renal dysfunction ranked among the strongest signals, consistent with a growing body of evidence linking impaired kidney function to heart failure readmissions; the cardiorenal interplay means that failing kidneys often herald decompensating hearts. A history of prior heart failure, older age, and longer length of hospital stay completed the top tier of predictors. Each of these is available to clinicians before discharge, which means the model’s inputs place no new burden on resource-constrained rural facilities.

The implications reach beyond the algorithm’s raw performance. In rural health systems, where access to cardiologists, structured follow-up clinics, and home-monitoring programs is often limited, every intervention dollar must be spent where it counts. A transparent risk score computed at discharge could direct scarce resources, such as early nurse visits, telemonitoring enrollment, or expedited outpatient appointments, toward the patients most likely to return. Because the model explains itself, a hospitalist in a small regional hospital could see that a particular patient’s risk is driven chiefly by declining renal markers and a prior heart failure admission, and tailor the discharge plan accordingly. The study was supported by Spain’s Ministry of Science and Innovation through the Complementary Plan for Applied Biotechnology in Health, backed by Next Generation EU funds, reflecting a broader European push to bring precision medicine tools into everyday care.

The authors are careful about what their results do and do not establish. The model was trained and validated within a single regional registry, and its performance in other populations, health systems, and countries remains to be demonstrated. External validation in independent cohorts and prospective evaluation of real-world impact, measuring whether the model actually reduces readmissions when deployed at the bedside, are required before clinical implementation. The underlying data are not publicly available due to privacy and ethical restrictions, though access may be considered with appropriate approvals. These caveats are standard in the field, and the study’s methodology aligns with emerging international guidance such as the TRIPOD+AI reporting framework and the FUTURE-AI consensus on trustworthy clinical artificial intelligence, both of which emphasize rigorous validation, transparency, and fairness. Even with those limitations, the work represents a meaningful step forward. It demonstrates that interpretable machine learning, fed by the ordinary data that any hospital already collects, can outperform traditional statistical models and deliver predictions clinicians can actually understand. For rural communities bearing a disproportionate share of the heart failure burden, that combination of accuracy, accessibility, and transparency could prove decisive in keeping vulnerable patients from cycling back through hospital doors.

Subject of Research: Interpretable machine learning prediction of 30-day readmission after heart failure hospitalization using a 20-year rural population registry

Article Title: Predicting 30-day readmission after heart failure hospitalization using interpretable machine learning: evidence from a 20-year population-based rural registry

Article References: Maese-Calvo, J., Paredes-Calderón, A., Nunez-Bayon, M., Arévalo-Lorido, J. C., Mayoral-Testón, N., Nevado-Nogales, C., José Zaro-Bastanzuri, M., González-Fernández, R., Hernández-Rollán, N., Corral-García, J., Rico-Gallego, J. A., & Fernández-Bergés, D. (2026). Predicting 30-day readmission after heart failure hospitalization using interpretable machine learning: evidence from a 20-year population-based rural registry. Medical & Biological Engineering & Computing. https://doi.org/10.1007/s11517-026-03657-2

Image Credits: AI Generated

DOI: 10.1007/s11517-026-03657-2

Keywords: heart failure, hospital readmission, machine learning, random forest, XGBoost, SHAP, interpretable AI, rural health, renal dysfunction, clinical decision support, risk prediction, registry study