Algorithm for Predicting the Unfavorable Course of Sepsis in Children: Complete Statistical Analysis

1. Trial at a Glance

This page covers the registry results for Algorithm for Predicting the Unfavorable Course of Sepsis in Children (NCT05908162), a registered phase not stated study. The primary outcome is Blood Leukocyte Subpopulations: the Absolute Numbers of Leukocytes of Specific Phenotypes.

2. Clinical Question

3. Trial Design

This is a completed study listed under the study type “other,” with an enrollment of 185. The sponsor is The Republican Research and Practical Center for Epidemiology and Microbiology. The recorded start date is 2021-07-01, and the primary completion date is 2024-12-25. The registry status is COMPLETED.

The record lists four patient groups: patients with sepsis on day 1, patients with sepsis on day 7, patients with severe bacterial infection, and patients with severe viral infection. The severe bacterial infection group is described as pneumonia, and the severe viral infection group as COVID19. A control group also appears in the adverse-event information. These group descriptions do not establish that the groups were assigned as treatment arms or that the same people were followed across time.

The listed intervention is a diagnostic test labeled “Bood leukocyte subsets.” The primary outcome description concerns blood leukocyte subpopulations and total leukocyte numbers. Separately, the study summary describes investigating four diagnostic markers—C-reactive protein, procalcitonin, presepsin and lipopolysaccharide binding protein—and immune-system gene polymorphisms. It states an intention to develop a prediction algorithm from a comprehensive assessment of biochemical and molecular genetic markers.

The record does not state an allocation method. It identifies the groups as not being a like-for-like control comparison, so differences between them should not be interpreted as randomized treatment effects. Further details about recruitment, eligibility, follow-up procedures, or the relationship between the group definitions are not reported in the registry record.

4. Participant Flow and Accounting

The registry reports how participants moved through each stage of the study. All counts below are exactly as posted; missing counts are not inferred from other stages.

Overall Study

5. Baseline Characteristics

The table reproduces the baseline characteristics posted in the registry, which describe the participants analysed at baseline and help show whether the randomized groups were comparable.

6. Primary Endpoint

The registry lists the primary outcome as Blood Leukocyte Subpopulations: the Absolute Numbers of Leukocytes of Specific Phenotypes, measured over 1 month.

The primary outcome is titled “Blood Leukocyte Subpopulations: the Absolute Numbers of Leukocytes of Specific Phenotypes.” It concerns the number of leukocytes and selected cell types in blood, expressed in 10^9 cells/l. The listed cell populations include total leukocytes; lymphocytes; T lymphocytes and their helper, cytotoxic and NKT subsets; NK cells; B lymphocytes; monocytes; and classical, intermediate and nonclassical monocyte subsets. The timeframe is 1 month.

The registry record identifies the outcome parameter as a mean and the dispersion as an inter-quartile range. It describes determining the leukocyte count under a microscope in a Goryaev chamber. These details identify the measurement and the planned summary style; they do not provide the observed values for the cell populations. Numerical results for the primary outcome are not reported in the registry record.

In general, leukocyte subpopulation measurements can characterize the distribution of immune cells in a blood sample. In this study, these measurements sit within a broader stated aim to investigate markers relevant to early diagnosis and prediction of complications and death in sepsis. However, the outcome description alone does not establish that any cell measurement predicts clinical course, nor does it quantify how well measurements distinguish outcomes. The registry record does not report a definition of “unfavorable course” for evaluating the primary outcome.

7. Why No Re-analysis Is Shown

8. Statistical Interpretation

9. Statistical Methods Explained

What does the primary outcome summary describe?

The registry record identifies the primary outcome as absolute leukocyte counts for specified phenotypes, with units of 10^9 cells/l. It names a mean as the outcome parameter and an inter-quartile range as the dispersion measure. In general, a mean describes the arithmetic average of observed values. The inter-quartile range describes the span from the lower to upper quartile and gives information about the middle portion of a distribution. Reporting both can give readers a view of the average and the spread, although the registry record does not provide the actual primary outcome summaries.

The record says that leukocyte count was determined under a microscope in a Goryaev chamber. It lists total leukocytes and several lymphocyte and monocyte phenotypes. The measurement description is important because the definition of each cell population and the measurement process determine what the resulting numbers represent. Further operational details, including how samples were collected, processed or assessed for quality, are not reported in the registry record.

What can a confidence interval tell readers?

In general, a confidence interval expresses the statistical uncertainty around an estimated quantity under the assumptions of the method used. A narrow interval indicates greater precision than a wide interval, while an interval that spans values compatible with different interpretations may leave the estimate inconclusive. An interval does not repair limitations in study design, measurement, or missing data. The registry record does not report confidence intervals for the primary leukocyte outcome or for prediction performance.

For comparisons, an interval around a difference or other contrast can help show both the estimated direction and the plausible range of magnitudes. Interpreting such a comparison requires knowing how groups were defined, whether observations are independent, and whether the analysis accounted for important sources of variation. Those details are not reported in the registry record. The record also does not state whether adjusted estimates were used.

Why does the group structure matter?

The listed groups are clinical categories and time labels, not documented randomized treatment assignments. In general, random allocation can support causal comparisons by balancing measured and unmeasured characteristics on average, whereas comparisons between groups formed by clinical status may reflect differences in severity, timing, or other characteristics. The registry record explicitly cautions that the arms are not a like-for-like control comparison and says allocation is not stated. Therefore, any difference across these groups should be read as an association unless additional design information supports a stronger interpretation.

The labels for sepsis on day 1 and day 7 might suggest measurements at different times, but the record does not state whether the same children contributed to both groups or how repeated observations were analyzed. In general, repeated measurements on one person are correlated. Analyses that treat correlated measurements as independent can misstate uncertainty. The registry record does not describe a repeated-measures method or clarify the participant overlap across groups.

What does over-dispersion mean in general?

For count outcomes, over-dispersion means the observed variation is greater than expected under a model that assumes a particular count distribution. If ignored, this extra variability can make standard errors too small and confidence intervals too narrow. A model that allows greater variation may be more appropriate in such settings. The primary outcome is described as absolute cell counts, but the registry record does not report a statistical model, distributional assumptions, or whether over-dispersion was assessed. It would therefore be inappropriate to infer that a particular count model was used.

What do the planning calculations represent?

The planning illustrations are described as calculations based on the observed effect, using a two-sided significance level of 0.05, equal allocation between two arms and a normal approximation. The effect to detect is expressed as a stated fraction of the observed effect. These calculations are planning illustrations and are not observed power. They do not describe the trial’s original design justification; that justification is not reported in the registry record. The record does not provide the numerical planning estimates in the information available for this analysis.

For time-to-event calculations, the stated method uses the total number of events, following Schoenfeld. Other calculations are described as participants per arm. In general, a planning calculation depends on its assumptions about the target effect, variability, significance level and allocation. If those assumptions differ from the actual study, the illustration may not describe the study’s achieved precision. The record does not report a time-to-event primary outcome or an interim-analysis plan.

More broadly, the registry record lists 4 analyses but does not describe the statistical tests, models, covariates, handling of outliers, or procedures for checking model assumptions. It also does not report whether an analysis plan was prespecified. These omissions limit an independent assessment of how the reported results, if available elsewhere, were generated and how robust they are to alternative analytical choices.

10. Multiplicity, Missing Data and Analysis Populations

The registry record lists 2 outcome measures and 4 analyses, but does not describe how the analyses relate to the outcomes or identify a primary analysis among them. It does not report a multiplicity adjustment or explain whether the analyses were treated as exploratory. In general, testing several outcomes, subgroups or contrasts can increase the chance of finding an apparently notable result by chance. Adjustment methods can control some forms of this problem, while prespecified prioritization can clarify which questions are confirmatory. Without a stated approach, readers cannot determine how multiplicity was addressed.

Missing-data handling is not reported in the registry record. The record gives an enrollment of 185, but does not state how many participants contributed primary outcome measurements, whether measurements were unavailable at the 1 month timeframe, or why data might be missing. In general, the consequences of missing data depend on how much is missing and on the relationship between missingness and the unobserved values or clinical characteristics. Different assumptions and methods can lead to different estimates, so reporting the extent and handling of missingness is important.

The analysis population is also not described. The record does not say whether analyses used all enrolled participants, only participants with available measurements, or another defined set. It does not clarify whether the day 1 and day 7 groups contain repeated observations from the same participants. Without these details, readers cannot assess the denominator for outcome analyses, the handling of participants who did not complete follow-up, or whether the analysis corresponds to the original group definitions.

The registry record does not report an interim-analysis plan, a missing-data sensitivity analysis, or a prespecified strategy for subgroup analyses. The absence of these details should not be interpreted as evidence that such procedures were or were not performed; they are not reported in the registry record.

11. Safety Results

The registry reports serious adverse events as affected participants divided by participants at risk.

Deaths

Other (non-serious) adverse events

Participants with at least one non-serious event at or above the registry's 0% reporting threshold.

Most frequent serious adverse event terms

Serious events with the highest arm-level rate: death (6.0%).

The adverse-event record lists 3 serious events among 50 at risk in the patients-with-sepsis-on-day-1 group, and 3 deaths in that group. The other listed groups—sepsis on day 7, severe bacterial infection, severe viral infection and control—each have zero serious events and zero deaths recorded. The day 1 group also has zero other adverse events recorded; the other groups likewise have zero other adverse events recorded. The listed top serious term is death, with 6.0% shown.

These figures are descriptive entries by group. They do not establish that an event was caused by a diagnostic test or other study procedure. The registry record does not provide event narratives, timing details, severity grading beyond the serious-event category, or a causal assessment. Nor does it describe a randomized comparison that would support attributing differences in event counts to an intervention.

Zeros in a registry entry mean that no events are recorded for that group in the listed fields; they do not, by themselves, establish that no events occurred outside the reporting process or that the groups had equivalent safety. The record also does not state whether safety information was collected over the same duration in all groups. These figures should therefore be read as reported event counts, not as a comparative estimate of risk.

12. What the Result Does, and Does Not, Mean

The record documents a completed study with an enrollment of 185 and a primary outcome focused on absolute blood leukocyte subpopulation counts. It identifies the measurement unit, the cell populations of interest, a 1 month timeframe, and a mean with inter-quartile range as the stated summary format. It also describes a broader objective of investigating biochemical markers and immune-system gene polymorphisms to develop an algorithm for predicting an unfavorable course of sepsis in children.

Those aims do not, by themselves, show that a prediction algorithm was developed, validated or found to be accurate. The registry record does not report numerical primary outcome results, algorithm performance, or an operational definition of an unfavorable course. It therefore does not permit assessment of the size or precision of differences in leukocyte measurements, nor of whether those measurements or other markers predict complications or death.

The groups are not described as randomized, allocation is not stated, and the record cautions that they are not a like-for-like control comparison. Group contrasts, if available elsewhere, should not automatically be interpreted as causal effects. Differences could reflect how participants entered the groups or other clinical characteristics, but the registry record does not provide enough detail to characterize those possibilities.

The adverse-event figures record 3 serious events and 3 deaths among 50 at risk in the sepsis day 1 group, with zeros in those fields for the other listed groups. They are not evidence of a treatment effect or of a causal relationship to a study procedure. The record does not report detailed safety follow-up or causal evaluations.

In short, the registry record communicates the study’s intended measurements and prediction objective, but its available results do not establish the algorithm’s performance or the clinical meaning of any measured differences. Such conclusions would require reported estimates, uncertainty, analysis details and validation information.

13. Why This Trial Matters Statistically

This study is statistically relevant because it brings together measurements from different domains: blood leukocyte subpopulations, biochemical markers and immune-system gene polymorphisms. The stated aim is to combine biochemical and molecular genetic markers in an algorithm intended to predict an unfavorable course of sepsis in children. Evaluating such an aim requires more than identifying associations; it requires clear outcome definitions, transparent model development and evidence about predictive performance.

The registry record also illustrates why the design structure matters. It lists groups defined by clinical condition and time, while stating that allocation is not reported and that the arms are not a like-for-like control comparison. This distinction shapes what statistical conclusions are justified. Descriptive contrasts may characterize the groups, but they do not automatically identify causal effects.

Prediction research raises particular statistical questions. In general, a model should be evaluated not only for how closely it fits the data used to build it, but also for how well its predictions agree with outcomes in data not used for development. Calibration describes agreement between predicted and observed outcome frequencies; discrimination describes how well predictions separate participants with different outcomes. The registry record does not report these performance measures, a validation procedure, or the algorithm itself.

The record’s limited reporting of primary outcome values and analysis methods also highlights the role of precision and transparency. A measurement description alone cannot show the magnitude or uncertainty of an association. Readers need estimates, denominators, methods for handling missing data, and a clear account of how multiple outcomes or analyses were treated. Without them, the stated research objective can be understood, but the strength and reproducibility of the statistical evidence cannot be evaluated from the registry record.

14. Limitations

• The posted summary statistics are the only data these results rest on; patient-level data, full protocols and statistical analysis plans are not available.
• Recalculated results are simple, unadjusted calculations from posted summaries. They do not replace the sponsor's prespecified analysis.

The registry record does not report numerical primary outcome results, confidence intervals, or estimates of association for the listed leukocyte subpopulations. This prevents assessment of the magnitude and precision of the measurements or contrasts. The description identifies the outcome’s unit, timeframe and summary format, but does not give the observed values.

The intended prediction algorithm is not described in enough detail for statistical evaluation. The record does not report the final predictors, the definition of an unfavorable course, the algorithm’s fitted form, or performance measures. It also does not report whether the algorithm was validated in data separate from those used to develop it. The broader summary’s stated goal should therefore not be taken as evidence of achieved predictive performance.

Group structure and participant flow are unclear. Allocation is not stated, and the record cautions that the groups are not a like-for-like control comparison. It does not explain whether the day 1 and day 7 groups include the same children, how participants were selected for the severe infection and control groups, or how enrollment was distributed across groups for the primary outcome. Eligibility criteria and analysis populations are not reported in the registry record.

Important statistical details are also absent: the analysis methods, handling of repeated measurements, missing-data procedures, multiplicity strategy, model assumptions, and any sensitivity analyses are not reported. The record lists 4 analyses but does not explain their methods or relation to the outcomes. These gaps limit independent assessment of bias, precision and reproducibility.

The adverse-event entries provide group counts but not narratives, follow-up duration, or causal assessments. The record does not state whether event collection was comparable across groups. Consequently, the figures are descriptive and cannot establish comparative safety or attribution to a study procedure.

15. Related Tutorials

16. Related Statistical Calculators

17. Sources

• ClinicalTrials.gov: Algorithm for Predicting the Unfavorable Course of Sepsis in Children — NCT05908162.
• Registry record last updated 2025-12-30; status last verified 2025-12.
• Results first posted 2025-12-30.
• Data retrieved from ClinicalTrials.gov on 2026-11-09.
• Group sizes, means, event counts and sponsor-reported p-values are reproduced as posted. Effect estimates, confidence intervals and p-values not attributed to the sponsor are independently calculated from the posted summary statistics.

18. Record Summary

NCT05908162 is a completed study titled “Algorithm for Predicting the Unfavorable Course of Sepsis in Children,” sponsored by The Republican Research and Practical Center for Epidemiology and Microbiology. The registry lists an enrollment of 185, a start date of 2021-07-01 and a primary completion date of 2024-12-25.

The primary outcome concerns absolute numbers of blood leukocyte subpopulations, measured in 10^9 cells/l, with a timeframe of 1 month. The stated summary format is mean with inter-quartile range. The study summary describes a broader objective involving biochemical markers and immune-system gene polymorphisms to develop a prediction algorithm.

The registry record does not report numerical primary outcome results or algorithm performance. It does not state allocation, and it cautions that the listed groups are not a like-for-like control comparison. The record lists 2 outcome measures and 4 analyses but does not report their analysis methods. Adverse-event entries include 3 serious events and 3 deaths among 50 at risk in the sepsis day 1 group; the other listed groups have zero serious events and deaths recorded.