The investigation took a two-pronged approach. On the population side, the team mined the Global Burden of Disease 2021 results tool, selecting heart failure as a Level 1 impairment and then restricting the analysis to the neonatal period, the first twenty-eight days of life. This yielded prevalence and years lived with disability estimates for Chinese newborns stretching from 1990 to 2021. On the clinical side, the researchers performed a secondary analysis of a pre-existing pediatric heart failure cohort that had enrolled patients at thirty centers spanning twenty Chinese provinces between 2013 and 2022. From that larger cohort, they isolated 212 neonates, of whom 23 died in hospital, and asked which baseline characteristics were associated with those deaths.

The population-level findings paint a picture of modest but persistent growth. Between 1990 and 2021, the prevalence of neonatal heart failure in China rose by 8.64 percent, while the rate of years lived with disability, a measure that captures the functional burden of disease, increased by 8.57 percent. Both metrics showed an estimated annual percentage change of 0.38 percent, indicating a slow, steady climb rather than any dramatic surge. That trajectory likely reflects a combination of forces: improved survival of critically ill newborns who subsequently live with cardiac impairment, better recognition and diagnosis of heart failure in neonatal units, and shifting patterns of underlying disease across three decades of rapid health-system change in China.

Perhaps the most consequential epidemiological finding concerns causation. In 2021, complex congenital heart anomalies emerged as the leading contributing cause of neonatal heart failure in China. This matters because it defines where resources must flow. Structural heart disease in newborns demands surgical and catheter-based interventions, specialized cardiac intensive care, and prenatal detection programs that can route expectant mothers to equipped centers before delivery. A burden dominated by congenital anomalies is fundamentally different from one dominated by cardiomyopathy or infection, and the Chinese data suggest that the structural component now sits at the top of the list.

The clinical half of the study, however, delivers a more sobering message about prediction. The researchers focused on four clinically selected baseline variables: postnatal age, gestational age, the presence of a complex congenital heart anomaly, and severe infection at admission. Because 23 deaths among 212 patients is a small number of events for statistical modeling, they used a Firth penalized logistic regression, a technique designed to produce more stable estimates when outcomes are rare and conventional maximum-likelihood methods can fail. The primary complete-case analysis included 184 neonates and all 23 deaths, with additional analyses addressing missing data, the clustering of patients within centers, and internal validation of the model’s performance.

The point estimates told an intuitively plausible story. Each additional day of postnatal age was associated with lower odds of in-hospital death, with an odds ratio of 0.95, and each additional week of gestational age carried an odds ratio of 0.92, suggesting that more mature and older newborns fared somewhat better. Conversely, complex congenital heart anomalies roughly doubled the odds of death, with an odds ratio of 2.28, and severe infection at admission raised the odds by a factor of 2.68. But here lies the study’s central caveat: every one of those confidence intervals crossed the null value of one. The interval for postnatal age ran from 0.89 to 1.01, for gestational age from 0.80 to 1.05, for complex anomalies from 0.68 to 7.59, and for severe infection from 0.81 to 8.94. In plain terms, the data cannot rule out the possibility that none of these factors is truly associated with mortality.

The authors are refreshingly candid about this limitation, describing the associations as imprecisely estimated and explicitly framing them as exploratory. They caution that larger, prospectively characterized cohorts are needed before any of these variables can be used for risk stratification at the bedside. That caution is well placed. In rare-outcome research, wide confidence intervals are not a technical footnote; they are the difference between a finding that can guide clinical decisions and one that merely generates hypotheses for the next study. A clinician reading these results should come away informed about what is uncertain, not reassured by the direction of the point estimates.

What the model did achieve was respectable discrimination. The apparent area under the receiver operating characteristic curve, a measure of how well the model separates those who die from those who survive, was 0.736, with a bootstrap confidence interval of 0.628 to 0.832. After correcting for optimism, the statistical inflation that occurs when a model is evaluated on the same data used to fit it, the AUC settled at 0.695. That optimism-corrected figure sits in the range often considered acceptable for a parsimonious four-variable model, though it falls short of the performance that would justify deployment as a standalone triage tool. It suggests the chosen variables carry real signal, but that signal is diluted by small numbers and by the heterogeneity inherent in a condition with many causes.

The methodological architecture of the study deserves attention in its own right. Secondary analyses of existing cohorts are an efficient way to extract clinical insight from data already collected, and the researchers layered on the safeguards that such analyses require: multiple imputation or sensitivity handling of missing values, accounting for the fact that patients nested within the same hospital share institutional practices, and bootstrap-based internal validation rather than the naive presentation of uncorrected performance statistics. Pairing this clinical modeling with the GBD 2021 framework also demonstrates a template for rare pediatric conditions worldwide, where dedicated national registries are scarce but global burden estimates and multicenter hospital cohorts can be combined to triangulate both the size of a problem and its determinants.

The broader implications reach beyond China. Neonatal heart failure sits at the intersection of congenital cardiology, neonatal intensive care, and infectious disease, and the finding that complex congenital anomalies now lead the causal profile underscores the importance of prenatal screening, timely referral, and equitable access to pediatric cardiac surgery. Meanwhile, the mortality analysis, though inconclusive, flags severe infection at admission as a factor worth watching, consistent with the well-established interplay between sepsis and cardiac dysfunction in newborns. For researchers, the message is clear: the questions are well posed, the analytical tools are available, and what is needed now is scale. A prospective cohort large enough to yield dozens or hundreds of neonatal deaths would transform these wide, uncertain intervals into estimates sharp enough to build genuine risk scores. Until then, the study stands as both a valuable map of the epidemiological terrain and an honest acknowledgment of how much remains unmapped in the care of the smallest hearts.

Subject of Research: Epidemiology and in-hospital mortality risk factors of neonatal heart failure in China

Article Title: Epidemiological profile and factors associated with in-hospital mortality in neonatal heart failure: a secondary analysis of GBD 2021 and a Chinese multicenter cohort

Article References: Aini, M., Yuan, Y., Liu, L., Yan, X., Li, J., Huang, S., Pan, B., & Tian, J. (2026). Epidemiological profile and factors associated with in-hospital mortality in neonatal heart failure: a secondary analysis of GBD 2021 and a Chinese multicenter cohort. BMC Pediatrics. https://doi.org/10.1186/s12887-026-07826-y

Image Credits: AI Generated

DOI: 10.1186/s12887-026-07826-y

Keywords: neonatal heart failure, Global Burden of Disease 2021, in-hospital mortality, congenital heart anomaly, China, multicenter cohort, Firth penalized logistic regression, years lived with disability, neonatology, severe infection, risk stratification, pediatric cardiology

The research team, led by Xu Yang and Li-Min Xin of Beijing Huilongguan Hospital at Capital Medical University, together with senior author Tian-Mei Si of Peking University Sixth Hospital, recruited participants from a Chinese multicenter cohort known as ODMPIM — Objective Diagnostic Markers and Personalized Intervention in MDD Patients. The trial was registered on ClinicalTrials.gov as NCT02023567 in December 2013, and the study drew on nine psychiatric and general hospitals across China, from Peking University Sixth Hospital and Beijing Anding Hospital to sites in Tianjin, Shanxi, Hebei, Dalian and Shanxi Medical University. Ethics committees at every participating institution approved the protocol, and all participants gave written informed consent under the Declaration of Helsinki.

What makes the design unusually clean is the drug-naïve, first-episode restriction. Most studies of cognition in depression examine patients who have already been exposed to antidepressants, sometimes for years, which makes it difficult to separate the effects of the illness itself from the effects of treatment, chronicity or prior episodes. By enrolling only people who had never received psychotropic medication and were experiencing their first depressive episode, the researchers could ask a purer question: what does depression, in its untreated first appearance, actually do to the thinking brain — and does the answer depend on when in life the disorder first emerged?

The cutoff chosen was 30 years of age at onset. Patients whose depression first appeared before 30 were classified as younger-onset; those whose illness began at 30 or later were classified as older-onset. All patients completed the 17-item Hamilton Depression Rating Scale to quantify depressive severity, the Mini-International Neuropsychiatric Interview to confirm diagnosis, and the Childhood Trauma Questionnaire-Short Form to capture early adversity across emotional, physical and sexual abuse and neglect domains. Healthy controls completed the trauma questionnaire and the same cognitive battery, allowing the team to benchmark both patient subgroups against people without the disorder.

The cognitive assessment spanned five domains, each tapped by established neuropsychological instruments. Processing speed was measured with the Digit Symbol Coding Test and the Color Trial Test. Executive function was probed through the Stroop Color-Word Test and Auditory Verbal Fluency. Attention was assessed with the Continuous Performance Test-Identical Pairs, a demanding task requiring sustained vigilance to fleeting stimulus matches. Verbal learning and memory relied on the Hopkins Verbal Learning Test-Revised, while visual learning and memory used the Brief Visual Memory Test-Revised. Statistical analysis proceeded through multivariate analysis of covariance, followed by correlation and multiple linear regression models that asked which clinical variables best predicted cognitive performance.

The headline result is a stark asymmetry. Younger-onset patients showed impairments across all five cognitive domains relative to healthy controls — processing speed, attention, executive function, verbal learning and visual learning were all affected. Older-onset patients, by contrast, displayed a narrower profile, with deficits confined to processing speed, verbal learning and executive function, while their attention and visual memory performance did not differ significantly from the healthy comparison group. When the two patient subgroups were compared directly, the younger-onset group performed worse than the older-onset group specifically in processing speed and in learning on the Hopkins Verbal Learning Test-Revised, pointing to speed of mental processing and the capacity to acquire new verbal material as the domains most sensitive to early illness onset.

Equally revealing were the results that failed to materialize. Both patient subgroups reported significantly higher levels of childhood trauma than healthy controls, yet the rates of current suicidal ideation — assessed with the Mini-International Neuropsychiatric Interview — were similar between the younger- and older-onset groups. And in the regression models, neither total childhood trauma scores nor Hamilton depression severity predicted cognitive performance. This dissociation matters: it suggests that the cognitive differences between onset groups are not simply a byproduct of more severe depression or a heavier load of early adversity, but may reflect something intrinsic to the developmental timing of the disorder itself.

The regression analyses did surface other predictors. Younger age-at-onset and male sex both predicted poorer processing speed and weaker executive function, reinforcing the picture that an early-arriving illness, particularly in men, is associated with the most pronounced slowing and the greatest difficulty in flexible, goal-directed thinking. In a counterintuitive twist, the presence of suicidal ideation was associated with better processing speed. The authors frame the study as exploratory, and this association in particular invites cautious interpretation — but it hints that the cognitive signatures of depression may not map neatly onto clinical risk in straightforward ways, and that faster processing in some patients may coexist with acute psychological distress.

Why might an earlier onset leave a wider cognitive footprint? The authors’ framing points toward neurodevelopment. Cognitive abilities such as processing speed and verbal learning continue to mature through the third decade of life, and an illness that disrupts brain function during or before this window may interfere with abilities that are still consolidating, or may mark a more neurodevelopmentally rooted form of depression. A later onset, arriving after cognitive systems have fully matured, may spare some domains — attention and visual memory in this dataset — while still eroding speed, verbal learning and executive control. Age-at-onset, in other words, may act as a clinically relevant marker of cognitive heterogeneity, a way of subtyping major depressive disorder that carries direct implications for rehabilitation, occupational counseling and the monitoring of functional recovery in newly diagnosed adults.

The study’s limitations temper but do not erase its significance. The cross-sectional design captures a single moment in time and cannot establish whether early-onset patients will remain more impaired over the course of illness or treatment. The 30-year cutoff, while clinically meaningful, is ultimately a convention, and the authors describe their cognitive patterns as exploratory rather than definitive. Still, the scale of the cohort, the rigor of the five-domain battery and, above all, the drug-naïve first-episode design give the findings unusual weight. For a disorder diagnosed in hundreds of millions of people worldwide, the message is concrete: when a young adult walks into a clinic with a first episode of depression, the age at which the illness began may be one of the most informative numbers on the chart — a signal of how widely cognition has been affected, and of which patients deserve the closest attention to the thinking skills that underpin work, study and daily life.

Subject of Research: Age-at-onset differences in neurocognition, childhood trauma and suicidal ideation in drug-naïve first-episode major depressive disorder

Article Title: Neurocognition, childhood trauma and suicidal ideation in drug-naïve first-episode major depressive disorder: a comparison of younger- versus older-onset patients

Article References: Yang, X., Xin, L.-M., Fan, H.-Z., Lin, J.-Y., Yan, F., Yu, X., Su, Y.-A., & Si, T.-M. (2026). Neurocognition, childhood trauma and suicidal ideation in drug-naïve first-episode major depressive disorder: a comparison of younger- versus older-onset patients. BMC Psychiatry. https://doi.org/10.1186/s12888-026-08695-6

Image Credits: AI Generated

DOI: 10.1186/s12888-026-08695-6

Keywords: major depressive disorder, age-at-onset, neurocognition, childhood trauma, suicidal ideation, drug-naïve, first-episode, processing speed, executive function, verbal learning, BMC Psychiatry, multicenter cohort

The work was led by Lysander Grospeaud, Romain Sonneville, and Etienne de Montmollin of AP-HP’s Hôpital Bichat-Claude Bernard in Paris, together with the EURECA Investigator Study Group, a broad network of intensive care clinicians spanning Europe and beyond. The parent EURECA cohort, funded by the European Society of Intensive Care Medicine, prospectively enrolled adult patients with meningoencephalitis requiring intensive care across multiple international centres, and an earlier publication in 2023 described the clinical features, etiologies, and outcomes of that population. The new scientific letter builds on that foundation by focusing specifically on the causes of death among patients who did not survive, and on the factors that pushed individual patients toward one mode of dying rather than the other.

This distinction matters because the two death pathways demand different clinical strategies. A death driven by neurologic causes, such as intractable intracranial hypertension, malignant brain edema, or refractory status epilepticus, invites aggressive neurocritical care: intracranial pressure monitoring in selected patients, osmotic therapy, seizure control, and early neuroimaging to detect surgical complications. A death driven by systemic causes, by contrast, reflects the failure of organs far from the central nervous system and calls for the full arsenal of general intensive care, from ventilatory and circulatory support to the prevention of ventilator-associated pneumonia, thromboembolism, and other intensive care complications. Knowing which patients are most likely to die through which route could, in principle, allow clinicians to target monitoring and preventive interventions more precisely.

The methodological strength of the EURECA approach lies in its prospective, multicenter design. Prospective enrolment means that data on presentation, etiology, treatment, and complications were collected according to a predefined protocol rather than reconstructed retrospectively from charts, reducing the information bias that plagues much of the literature on central nervous system infections. Multicenter recruitment across many countries guards against the idiosyncrasies of any single health system or referral pattern, producing a picture of severe meningoencephalitis that is more likely to generalise to intensive care units worldwide. The ancillary analysis reported in the new letter was conducted on anonymized data from the cohort and did not require additional ethical approval beyond the original authorisations, which were granted by the Comité de protection des personnes Sud Méditerranée 2, with national and local approvals at each participating site.

Severe meningoencephalitis remains one of the most feared presentations in critical care. Bacterial meningitis, most notably pneumococcal disease, has long been recognised as a devastating illness in adults, with landmark cohort studies from the Netherlands in 2004 and 2006 documenting high mortality and a heavy burden of intracranial complications such as cerebral herniation, cerebrovascular injury, and hearing loss. More recent work, including a 2024 retrospective multicenter cohort on severe pneumococcal meningitis, has reinforced how frequently these intracranial catastrophes occur even under modern care. Viral encephalitis, dominated by herpes simplex virus in adults, adds a different profile of injury, with seizures, limbic involvement, and prolonged encephalopathy, and a 2025 analysis in JAMA Network Open by members of the same French group examined long-term outcomes in critically ill adults with acute encephalitis. The EURECA programme sits squarely within this tradition, but its prospective international design gives it unusual weight.

By classifying deaths as neurologic or systemic, the new analysis reframes mortality in these infections not as a single undifferentiated outcome but as the endpoint of at least two distinguishable clinical trajectories. Patients on the neurologic trajectory are those in whom the brain injury itself proves fatal, whether through direct tissue destruction, mass effect and herniation, vascular complications, or uncontrollable electrical storm. Patients on the systemic trajectory die of the consequences of critical illness: refractory shock, respiratory failure, renal and hepatic dysfunction, and the cascade of organ support failures that define multi-organ dysfunction syndrome. The determinants of each trajectory are unlikely to be identical, and the study’s central contribution is to identify which baseline and early-course variables associate with each.

The clinical implications of such a stratification are considerable. If, for example, markers of severe brain injury at presentation, such as profound coma, seizures, or characteristic imaging findings, are found to predict neurologic death, then those patients might be candidates for earlier intracranial pressure monitoring, more aggressive neuroimaging, or transfer to centres with neurosurgical capability. Conversely, if systemic death is predicted by factors such as older age, pre-existing comorbidity, immunosuppression, or the causative organism, then those patients might benefit from intensified general critical care, earlier escalation of organ support, and vigilant prevention of intensive care unit complications. In both cases, the value of the analysis lies in converting a binary outcome, survival or death, into a more nuanced map of risk that can guide bedside decisions.

The study also speaks to a broader shift in how critical care research approaches central nervous system infections. For decades, outcome studies in this field reported mortality as a single figure, obscuring the fact that some patients die of their brain disease while others die of everything else. Distinguishing cause-specific death allows the field to ask sharper questions: Are intracranial complications becoming less lethal as neurocritical care improves? Is systemic mortality rising as patients survive longer on organ support? Are the two trajectories converging or diverging over time? Answering such questions requires the kind of granular, prospectively collected cause-of-death data that the EURECA investigators have now assembled, and it sets a template for future cohort studies in related conditions, from cerebral malaria to autoimmune encephalitis requiring intensive care.

There are, inevitably, limits to what any observational analysis can establish. Assigning a single cause of death in a critically ill patient is inherently judgment-dependent, and even with predefined criteria, the boundary between a neurologic and a systemic contribution can blur when brain-injured patients develop infections, thrombosis, and organ failure as downstream consequences of immobility and prolonged ventilation. The investigators addressed this by relying on the structured data collection of the parent cohort and by conducting the analysis on anonymized data under a predefined statistical plan, but cause-of-death classification remains an area where independent replication is valuable. The authors note that the data supporting the findings are available upon reasonable request, together with the study protocol and statistical analysis plan, which should facilitate scrutiny and reuse.

For clinicians, the practical message of the EURECA analysis is that death in severe meningitis and encephalitis is not monolithic, and that the fight against it must be waged on two fronts simultaneously. For researchers, the study demonstrates the power of large prospective international cohorts to resolve questions that single-centre series cannot touch. And for patients and families, it offers a sober but important form of transparency: an honest accounting of how these devastating infections kill, gathered from dozens of intensive care units and hundreds of clinical decisions, and published so that the next generation of care can be built on evidence rather than intuition. As the EURECA investigators continue to mine their cohort, the hope is that identifying the determinants of each death pathway will translate into fewer deaths on both of them.

Subject of Research: Causes of death and their determinants in critically ill adults with severe meningitis and encephalitis

Article Title: Determinants of neurologic causes and systemic causes of death in severe meningitis and encephalitis: insights from the international prospective multicenter EURECA study

Article References: Grospeaud, L., Sonneville, R., de Montmollin, E., on behalf of the EURECA Investigator Study Group, Aissaoui, N., Argaud, L., Audibert, J., Bailly, P., Balandín Moreno, B., Barbier, F., Ben Hadj Salem, O., Beuret, P., Boronat, P., Bourenne, J., Bruel, C., Bruneel, F., Calamai, I., Cariou, A., Chabanne, R., … Voiriot, G. (2026). Determinants of neurologic causes and systemic causes of death in severe meningitis and encephalitis: insights from the international prospective multicenter EURECA study. Intensive Care Medicine. https://doi.org/10.1007/s00134-026-08603-2

Image Credits: AI Generated

DOI: 10.1007/s00134-026-08603-2

Keywords: meningitis, encephalitis, intensive care, critical illness, neurologic death, systemic organ failure, EURECA study, central nervous system infection, sepsis, intracranial pressure, prognosis, multicenter cohort