Rising Lactate Levels May Signal Deadly Neonatal Bowel Disease Early

Necrotizing enterocolitis, or NEC, remains one of the most feared diagnoses in any neonatal intensive care unit. The condition, in which portions of a newborn’s intestine become inflamed and begin to die, strikes vulnerable infants, often those born prematurely, and can progress from subtle feeding intolerance to fulminant sepsis and death within hours. Despite decades of research, clinicians still lack reliable tools to identify, early in the disease course, which babies will deteriorate and which will recover with medical management alone. A new exploratory study published in BMC Pediatrics by Aina Qu, Si Kong, and Zhaoping Wang of Taixing People’s Hospital Affiliated to Yangzhou University suggests that the answer may lie not in any single laboratory value, but in the shape of a laboratory value’s movement over time.

The researchers focused on lactate, a metabolite that has long occupied a central place in critical care monitoring. Lactate accumulates when tissues are deprived of adequate oxygen or when cellular metabolism is disrupted, and elevated blood lactate is a well-established marker of shock and tissue hypoperfusion in adults and children alike. In neonates with NEC, however, lactate has traditionally been interpreted as a static snapshot: a single number drawn at a single moment, compared against a threshold. The Chinese team reasoned that this approach discards most of the information contained in the data. A lactate level of 6 millimoles per liter means something quite different depending on whether it is falling from 9 or climbing from 3, and capturing that direction and velocity could, in principle, reveal a patient’s trajectory toward recovery or collapse.

To test that idea, the investigators turned to a statistical technique known as group-based trajectory modeling, or GBTM. Rather than treating each patient’s serial lactate measurements as an individual curve to be eyeballed, GBTM uses finite mixture modeling to identify latent subgroups within a population—clusters of patients whose measurements follow statistically similar patterns over time. Each patient is assigned to a trajectory group based on a posterior probability, and in this study the researchers required that probability to exceed 0.7 before an assignment was accepted, a threshold that guards against forcing patients into groups they only weakly resemble. The approach, borrowed from criminology and developmental psychology where it has been used to chart patterns of behavior over life courses, is gaining traction in clinical medicine precisely because real-world intensive care data are messy, irregularly sampled, and incomplete.

The messiness was substantial. The team analyzed 106 neonates with NEC drawn from the Pediatric Intensive Care Database, a publicly available, de-identified repository established with approval from the Institutional Review Board of the Children’s Hospital of Zhejiang University School of Medicine. Because the study was a secondary analysis of de-identified data involving no more than minimal risk, patient informed consent was waived, and the work was conducted in accordance with the Declaration of Helsinki. The cohort was divided into three clinically meaningful strata: 63 neonates managed without surgery, 18 who underwent surgery within three days of diagnosis, and 25 who went to surgery later than three days. Lactate measurements taken within the first 24 hours after surgery were excluded from the modeling, a methodological choice designed to prevent the metabolic perturbation of the operation itself from contaminating the preoperative and postoperative trajectory patterns the researchers wanted to isolate.

What emerged from the modeling was striking. In the non-surgical group, GBTM identified a small but ominous trajectory pattern: a high-and-rising lactate course in which patients began with an initial lactate of 11.39 millimoles per liter and climbed by an additional 4.93 millimoles per liter per day. Only two of the 63 non-surgical neonates, or 3.2 percent, followed this pattern—but both died within 48 hours, a mortality rate of 100 percent for that subgroup. The finding, though based on a tiny number of patients, illustrates the central promise of trajectory analysis: a pattern that unfolds over serial measurements may flag impending catastrophe earlier than any single threshold crossed at one point in time.

The surgical strata told a different and, in some respects, equally informative story. All 43 infants who underwent surgery, whether early or late, survived their hospitalization, and their lactate trajectories were characterized as stable or clearing—that is, the models captured lactate values that either held steady at moderate levels or declined over the observation window. The contrast between the two non-surgical infants on a steeply rising course and the surgical patients whose lactate curves bent toward resolution offers a tantalizing hypothesis: that the direction of lactate change, rather than its absolute value, may help distinguish babies who are failing medical management from those whose disease is controlled, and may even inform the timing of the difficult decision to operate.

Because exploratory models built on small samples can be fragile, the authors took steps to assess robustness. They applied bootstrap resampling with 300 replicates to the non-surgical group, repeatedly refitting the trajectory model to resampled versions of the data to see whether the estimated parameters held steady. They did: every parameter in the non-surgical group’s model showed a coefficient of variation below 30 percent, indicating that the trajectory estimates were not artifacts of a few unusual patients or a chance configuration of the dataset. This kind of stability check is an important, if often overlooked, discipline in trajectory modeling, where the number and shape of latent groups can otherwise shift unsettlingly with small perturbations of the input data.

The study’s limitations are as instructive as its findings. With 106 patients in total and only two infants in the fatal high-and-rising group, the results are explicitly exploratory and cannot support clinical decision-making on their own. The retrospective design means the lactate measurements were drawn from routine clinical practice rather than a standardized protocol, introducing the very missingness and irregular sampling that the authors acknowledge as a defining challenge of real-world data. The Bayesian Information Criterion was used to guide model selection, but model choice in GBTM always involves judgment as well as statistics, and different criteria can favor different numbers of groups. Most importantly, no prospective validation has yet been performed; the association between a rising lactate trajectory and early mortality in NEC remains a hypothesis generated by the data, not a confirmed predictive rule.

Even so, the conceptual contribution is significant. Neonatology has long sought dynamic risk markers for NEC, and serial biomarker modeling of the kind demonstrated here offers a framework that could eventually incorporate not just lactate but other laboratory and physiologic signals—C-reactive protein, platelet counts, blood gas parameters—into composite trajectory profiles. If future prospective studies confirm that a steeply rising lactate course precedes death within 48 hours in medically managed NEC, bedside dashboards could one day alert clinicians to a deteriorating trajectory hours before conventional thresholds are breached, buying precious time to escalate care or intervene surgically. For now, the study stands as a careful, methodologically transparent proof of concept: in the fragile population of newborns with necrotizing enterocolitis, the trend of a number may matter as much as the number itself, and the mathematics of trajectories may help clinicians read that trend before it is too late.

Subject of Research: Lactate trajectory patterns as prognostic markers in neonatal necrotizing enterocolitis

Article Title: Lactate trajectory patterns in neonatal necrotizing enterocolitis: an exploratory retrospective cohort study using group-based trajectory modeling

Article References: Qu, A., Kong, S., & Wang, Z. (2026). Lactate trajectory patterns in neonatal necrotizing enterocolitis: an exploratory retrospective cohort study using group-based trajectory modeling. BMC Pediatrics. https://doi.org/10.1186/s12887-026-07788-1

Image Credits: AI Generated

DOI: 10.1186/s12887-026-07788-1

Keywords: necrotizing enterocolitis, lactate, neonatology, group-based trajectory modeling, trajectory analysis, predictive markers, neonatal sepsis, retrospective cohort study, pediatric intensive care, mortality risk, biomarkers, metabolic pathways