Simple Blood Test Combo Predicts Death Risk in Critically Ill Children

A routine blood calculation that combines inflammation and insulin resistance signals may help doctors identify which critically ill children are most likely to die during their hospital stay, according to a new study published in Pediatric Research. The measure, known as the C-reactive protein-triglyceride glucose index, or CTI, merges two laboratory values that are already drawn from almost every child admitted to an intensive care unit: C-reactive protein, a well-known marker of systemic inflammation, and the triglyceride-glucose index, a surrogate measure of insulin resistance. By multiplying these two dimensions of critical illness into a single number, researchers at Wenzhou Medical University in China found they could stratify mortality risk more effectively than with either component alone, and even outperform elements of established pediatric severity scores.

The retrospective cohort study drew on the Pediatric Intensive Care database, a de-identified repository of records from the Children’s Hospital of Zhejiang University School of Medicine covering the years 2010 to 2019. The analysis included 3,533 critically ill children aged between 29 days and 18 years. The researchers asked two related questions: whether a child’s baseline CTI value on admission predicted death within 30 days, and whether the way that value changed over the course of the hospital stay, its trajectory, carried additional prognostic information. Both answers turned out to be yes, and the magnitude of the associations was striking.

Statistically, each interquartile range increase in baseline CTI, meaning the jump from roughly the 25th to the 75th percentile of the distribution, was independently associated with a 50 percent higher risk of 30-day all-cause mortality, with an adjusted hazard ratio of 1.50 and a 95 percent confidence interval of 1.23 to 1.84. Importantly, this association held after adjustment for confounders, suggesting that CTI captures physiological information beyond what conventional measures provide. The team also demonstrated that CTI discriminated between survivors and non-survivors better than the triglyceride-glucose index alone or C-reactive protein alone, and that adding CTI to the PRISM model, a widely used pediatric risk-of-mortality score, significantly improved its performance.

Beyond the continuous association, the researchers identified a clinically practical threshold. A CTI cutoff of 10.42 divided the cohort into groups with sharply different fates: children above the cutoff faced roughly a twofold higher mortality risk, with an adjusted hazard ratio of 2.05 and a 95 percent confidence interval of 1.46 to 2.88. A single number, computable from blood tests ordered on virtually every ICU admission, separating patients into a high-risk and lower-risk category is exactly the kind of tool that appeals to frontline clinicians, because it requires no new assays, no specialized equipment, and no additional cost.

The technical logic behind the index reflects a growing recognition that inflammation and metabolic dysregulation are not separate phenomena in critical illness but intertwined processes. C-reactive protein rises rapidly in response to interleukin-6-driven inflammatory signaling, particularly in infection and tissue injury. The triglyceride-glucose index, calculated from fasting triglycerides and glucose, serves as a proxy for insulin resistance, which emerges during the metabolic stress response that accompanies severe illness. When these two axes are combined, the resulting score captures a vicious cycle in which inflammatory cytokines disrupt insulin signaling, and insulin resistance in turn amplifies inflammatory pathways, a feedback loop documented in both adult and pediatric critical care research.

Perhaps the most novel contribution of the study is its longitudinal dimension. Using a latent class growth mixed model, a statistical technique that groups patients by the shape of their repeated measurements over time rather than by any single value, the researchers identified distinct CTI trajectory patterns across the hospital stay. Children whose CTI remained persistently high fared the worst, while those with low-stable trajectories and those with U-shaped trajectories, in which the index initially rose and then fell back, showed significantly lower mortality risk than the high-stable group. This finding suggests that the direction of change matters: a falling CTI may signal resolving inflammation and recovering metabolic balance, whereas a stubbornly elevated index flags ongoing physiological derangement.

This trajectory approach mirrors a broader shift in critical care research away from static snapshots and toward dynamic biomarker monitoring. Previous studies have shown that serial lactate measurements, evolving vital sign patterns, and changing triglyceride-glucose values in adults all carry prognostic weight that single time-point readings miss. Earlier work had also hinted at complexity in pediatric populations specifically, with one prior retrospective cohort reporting a U-shaped relationship between the triglyceride-glucose index and mortality in critically ill children, meaning that both very low and very high values were associated with worse outcomes. The new study extends this line of inquiry by adding the inflammatory component and by formally modeling trajectories rather than isolated readings.

The authors, led by Guomiao Zhang, Huiqin Mei, and Qichao Sheng of the Department of Epidemiology and Health Statistics at Wenzhou Medical University, with Guangyun Mao and Dapeng Li as corresponding authors, emphasize that this is the first study to demonstrate the dual prognostic value of both baseline CTI and its longitudinal trajectories in critically ill children. They position CTI as a simple, dynamic tool for mortality risk stratification in pediatric critical care, one that could complement rather than replace established scoring systems such as PRISM and PIM2, which rely on physiological parameters collected at admission and are known to have imperfect calibration across diverse patient populations.

As with any retrospective, single-database study, important caveats apply. The data come from one Chinese children’s hospital, and the analysis relied on the Pediatric Intensive Care database, which was approved by that institution’s review board as a de-identified dataset exempt from additional consent. Retrospective designs cannot establish causation, and the association between high CTI and mortality does not prove that the index itself is harmful; it may simply be a faithful readout of underlying severity. Missing data were handled with multiple imputation, a standard but imperfect technique, and the researchers themselves note in the broader literature that classifying trajectories over time should be done with caution because different statistical methods can yield different groupings. External validation in independent cohorts, ideally prospective and multi-center, will be needed before CTI thresholds are adopted into routine practice.

Nevertheless, the appeal of the finding lies in its accessibility. In intensive care units around the world, particularly in resource-limited settings where advanced biomarkers and sophisticated scoring infrastructure are scarce, the ability to compute a mortality risk indicator from three routine laboratory values, triglycerides, glucose, and C-reactive protein, could meaningfully improve triage, family counseling, and the intensity of monitoring. If future studies confirm the cutoff of 10.42 and the prognostic significance of trajectory patterns, a calculation scribbled on a chart could become an early warning system for the sickest children, flagging those who need escalation of care before irreversible deterioration sets in. For a field where every hour of delay can matter, a free and fast risk signal drawn from blood already being tested is a compelling proposition.

Subject of Research: Prognostic value of the C-reactive protein-triglyceride glucose index and its trajectories for mortality in critically ill children

Article Title: C-reactive protein-triglyceride glucose index and its dynamic trajectories on all-cause mortality in critically ill children: a longitudinal, retrospective cohort study

Article References: Zhang, G., Mei, H., Sheng, Q., Yin, M., Fang, Y., Ding, Q., Liu, K., Shou, Y., Zhang, X., Shi, F., Mao, Q., Zheng, C., Mao, G., & Li, D. (2026). C-reactive protein-triglyceride glucose index and its dynamic trajectories on all-cause mortality in critically ill children: a longitudinal, retrospective cohort study. Pediatric Research. https://doi.org/10.1038/s41390-026-05481-8

Image Credits: AI Generated

DOI: 10.1038/s41390-026-05481-8

Keywords: C-reactive protein, triglyceride-glucose index, CTI, pediatric intensive care, critical illness, mortality prediction, insulin resistance, inflammation, biomarker trajectories, retrospective cohort study, PRISM score, risk stratification