Blood Cell Counts Reveal Which Children With Gastroenteritis Get Admitted

Every day, emergency departments and pediatric wards around the world face the same deceptively simple question: when a child arrives with acute gastroenteritis, does that child need to be observed, admitted, or safely sent home? Most cases of vomiting and diarrhea resolve with fluids and time, yet a minority deteriorate quickly, and clinicians have long relied on a blend of clinical judgment, vital signs, and laboratory results to sort the two groups apart. A new exploratory study published in BMC Pediatrics by Mehmet Cengiz of the Department of Pediatrics at Özel Gürlife Hospital in Eskişehir, Turkey, now offers a detailed look at which laboratory values actually track with those real-world escalation decisions, and the answer is striking: among children who were tested, the lymphocyte count stood out as the single most consistent correlate of a clinician’s decision to escalate care.

The study took the form of a single-centre retrospective analysis of 226 children aged zero to eighteen years who presented with acute gastroenteritis between April 2024 and March 2026. Rather than imposing an arbitrary severity score, the researchers classified each child’s disposition exactly as it happened in practice, using a hierarchical scheme with three levels: outpatient discharge, observation only, and hospitalization. Laboratory tests were not drawn uniformly; instead, they were obtained at the discretion of the treating clinician, either at presentation, at the moment of intravenous cannulation, or before fluids were administered. This design choice is crucial, because it means the study captures the messy, decision-driven reality of pediatric emergency care rather than an idealized research protocol.

That reality came with a statistical complication the author was careful to confront head-on. Laboratory testing was obtained in only 43 percent of children who were discharged as outpatients, compared with a near-universal 98 percent of those placed under observation or hospitalized. In other words, the data were not missing at random: clinicians ordered blood work precisely for the children they were already worried about, a phenomenon statisticians call informative missingness. Any comparison between tested and untested children is therefore biased by design, and the analysis was deliberately restricted to the tested subgroup, with the explicit framing that the findings describe associations with real-world management rather than a universal prediction rule for all children with gastroenteritis.

Within that tested subgroup, the differences between children who were escalated and those who were sent home were clear and statistically robust. Compared with outpatients, children escalated to observation or hospitalization had lower lymphocyte counts, higher neutrophil counts, and higher urea values, and all three differences survived correction for multiple comparisons using the Benjamini–Hochberg false discovery rate across eleven laboratory parameters. Group comparisons employed the Mann–Whitney U test for two-group contrasts and the Kruskal–Wallis test with Dunn post-hoc procedures for the three-level disposition comparison, non-parametric methods well suited to the skewed distributions typical of clinical laboratory data. The analyses were conducted in Python 3 using the SciPy and statsmodels libraries, and the full analysis code was made available as supplementary material, an unusual and welcome degree of transparency for a single-centre study.

The three-group comparison added an important nuance. Lower lymphocyte count separated outpatients from children who were escalated, with a Kruskal–Wallis statistic of 20.1 and a p-value below 0.001, but it did not distinguish children who were merely observed from those who were admitted to the ward. This suggests that lymphopenia may function in practice as a signal that a child is sick enough to warrant escalation of some kind, rather than as a fine-grained discriminator between intermediate and intensive levels of care. For a condition in which the pivotal decision is often simply whether to keep a child in the hospital at all, that distinction matters, and it hints that different markers or combinations of markers may be needed to separate observation from admission.

To quantify how strongly each laboratory parameter was associated with escalation, the study turned to multivariable logistic regression, reporting adjusted odds ratios alongside variance inflation factors to check for problematic collinearity between correlated blood counts. In the composite model, lower lymphocyte count emerged as the strongest correlate of escalation, with an adjusted odds ratio of 0.575 per unit increase, a 95 percent confidence interval of 0.419 to 0.789, and a p-value of 0.001. Adding lymphocyte count to the available clinical variables, which included fever, vomiting, and measured temperature, was associated with a statistically significant improvement in model fit within the tested subgroup, with a likelihood-ratio p-value below 0.001, and the apparent area under the receiver operating characteristic curve rose from 0.75 to 0.81. Discrimination for the full escalation model, after bootstrap optimism correction, was a c-statistic of 0.77, down from an apparent 0.82, a correction that guards against the inflated performance that plagues many small clinical prediction studies.

Among the composite inflammatory indices, the neutrophil-to-lymphocyte ratio, a marker widely studied in adult critical care and oncology but far less in pediatric gastroenteritis, showed the numerically highest single-marker discrimination for escalation, with an area under the curve of 0.78 and a 95 percent confidence interval of 0.68 to 0.86. The author was careful to note, however, that this confidence interval overlapped substantially with those of the lymphocyte and neutrophil counts themselves, meaning the ratio cannot claim superiority over its component measurements on this evidence. The platelet-to-lymphocyte ratio was assessed as a second exploratory composite but did not emerge as a standout. In a sobering secondary finding, no laboratory parameter survived false discovery rate correction when the outcome was narrowed specifically to hospitalization, as opposed to the combined escalation outcome, underscoring how much harder it is to predict the most serious dispositions than the broader decision to escalate.

One virological detail caught the eye: rotavirus positivity was associated with admission, with an odds ratio of 2.74 and a p-value of 0.015. This association is biologically plausible, since rotavirus is classically responsible for some of the most dehydrating episodes of childhood gastroenteritis, and clinicians may reasonably have a lower threshold for admitting a child with a confirmed rotavirus infection. It also serves as a reminder that disposition decisions are shaped by the full clinical picture, including etiological diagnosis, and not by blood counts alone. The study’s ethical footing was solid: the Scientific Research Ethics Committee of Eskişehir City Hospital approved the work in April 2026, waiving informed consent for the retrospective analysis of anonymized data, and the study was conducted in accordance with the Declaration of Helsinki, with no funding received and no competing interests declared.

The author is refreshingly candid about the study’s central limitation, which is incorporation bias: the very laboratory values under investigation informed the disposition decisions being studied, so the analysis inevitably finds that the tests correlate with decisions they helped to shape. Combined with the selective, clinician-directed testing, this means the results are hypothesis-generating rather than predictive, and they cannot be generalized to all children with acute gastroenteritis, most of whom never have a blood count drawn at all. Still, the study earns its place in the literature by documenting, with unusual statistical rigor and full methodological transparency, what actually happens in the real world of pediatric emergency care. Lower lymphocyte counts, elevated neutrophil counts, rising urea, and a high neutrophil-to-lymphocyte ratio cluster around the children clinicians choose to keep, and a prospective study that tests these markers before decisions are made, ideally across multiple centres and in the untested majority of children, would be the natural next step toward turning this observational signal into genuinely useful decision support.

Subject of Research: Laboratory correlates of observation and hospitalization decisions in children with acute gastroenteritis

Article Title: Laboratory correlates of observation and hospitalization decisions in children with acute gastroenteritis: an exploratory association study

Article References: Cengiz, M. (2026). Laboratory correlates of observation and hospitalization decisions in children with acute gastroenteritis: an exploratory association study. BMC Pediatrics. https://doi.org/10.1186/s12887-026-07783-6

Image Credits: AI Generated

DOI: 10.1186/s12887-026-07783-6

Keywords: acute gastroenteritis, children, lymphocyte count, neutrophil-to-lymphocyte ratio, hospitalization, rotavirus, pediatrics, laboratory markers, clinical decision-making, retrospective study, diagnostic markers, emergency care