Sample entropy misses cardiorespiratory events but flags vulnerable preterm infants after vaccination
Routine immunization in very preterm infants sits at the intersection of prevention and fragility. Protection against vaccine-preventable disease cannot wait, yet the post-vaccination period can expose immature autonomic control in a way that worries clinicians at the bedside. In a prospective single-center study of infants born before 32 weeks’ gestation, investigators examined whether sample entropy from heart rate time series could identify which babies were more likely to develop cardiorespiratory events after immunization.
The answer was mixed. Cardiorespiratory events increased after vaccination, but sample entropy did not separate infants who did from those who did not experience those events.
That negative finding matters because it narrows what this physiologic measure can and cannot do in the neonatal unit. Sample entropy is a non-linear marker of heart rate variability, designed to capture the irregularity of a time series rather than simple average rate or beat-to-beat spread. In theory, lower entropy may reflect a less adaptable autonomic system, one that struggles to buffer stressors such as handling, immune activation, or transient respiratory instability. In practice, however, the study found no association between entropy and the occurrence of post-vaccination cardiorespiratory events themselves.
A physiologic signal emerged only when intervention was needed
The more clinically actionable signal appeared one step later, when the analysis focused on infants who required intervention after vaccination. Lower sample entropy and lower gestational age were independently associated with the need for clinical intervention. Those interventions included bedside responses such as tactile stimulation or respiratory support, making the distinction more than an abstract statistical exercise; it speaks directly to which infant may need closer observation after a routine injection.
The cohort was drawn from routine hospital care between 2018 and 2022. Of 188 enrolled infants, 103 underwent sample entropy analysis at the first vaccination during the primary hospital stay. Mean gestational age was 27.3 weeks, with a range from 23.3 to 31.3 weeks. Heart rate and oxygen saturation were captured with standard monitoring at vaccination timepoints, and sample entropy was calculated from 90-minute electrocardiographic recordings. These details matter because the signal being measured was not an experimental laboratory readout; it came from the same monitoring environment that already supports neonatal care.
The separation between event occurrence and intervention requirement points to a subtle physiologic truth. Some infants may manifest transient instability that resolves without active bedside support, while others cross a threshold where clinicians must intervene. A measure of autonomic irregularity may be more closely tied to that threshold than to the raw presence of a brief cardiorespiratory episode. The data suggest that sample entropy, if used at all, may be better suited to identifying vulnerability severe enough to matter operationally in the nursery.
Why vaccination can transiently disturb preterm physiology
Preterm infants arrive at vaccination with immature respiratory control, limited physiologic reserve, and rapidly changing autonomic regulation. Immune stimulation may transiently challenge that balance. Even when vaccination is essential and should not be delayed without reason, clinicians remain attentive to the possibility of apnea, bradycardia, desaturation, or combinations of these events in the hours after immunization. The study addressed that concern directly, asking whether a quantitative autonomic marker could anticipate which infant would need support.
Sample entropy was an appealing candidate because it captures complexity in heart rate dynamics rather than relying on simple linear variability metrics. In a stable physiologic system, more adaptable cardiac timing can produce richer variability, while lower complexity may suggest a constrained control network. Yet biology rarely aligns cleanly with a single marker. Respiratory drive, maturation, recent handling, illness burden, and gestational age all shape the post-vaccination period. The absence of a relationship between entropy and event occurrence suggests that cardiorespiratory events arise from a broader set of influences than a single heart rate metric can capture.
Lower gestational age remained a companion predictor of intervention need, which is biologically unsurprising. The youngest infants have the least reserve for mounting coordinated responses to stress, and their respiratory and autonomic systems are still developing the feedback control needed to maintain stable oxygenation and heart rate. In that setting, a reduced entropy signal may be less a standalone biomarker than a marker of developmental immaturity layered onto known clinical risk.
What neonatal teams can use, and what they cannot
For now, the study supports a narrow conclusion: sample entropy may help identify infants who are more likely to need clinical intervention after vaccination, but it does not reliably identify which infants will experience cardiorespiratory events in the first place. That limitation constrains immediate bedside use. A tool that cannot predict the event itself cannot replace routine post-immunization monitoring, and it cannot justify relaxing observation practices for infants with reassuring values.
At the same time, the work provides a useful direction for neonatal physiology research. Continuous monitoring data are abundant in the hospital, and measures derived from those data could help transform routine observation into risk stratification. The challenge is selecting metrics with enough specificity to distinguish harmless fluctuation from instability requiring treatment. In this cohort, sample entropy did not reach that level for event prediction, yet it still carried information about clinical intervention.
The study design also frames the result carefully. It was prospective and observational, conducted at a single center, and only 103 of the 188 enrolled infants underwent entropy analysis at the first vaccination during the primary hospital stay. The abstract does not report the full range of vaccination schedules, the precise timing of events after immunization, or the detailed intervention thresholds used at the bedside. Those missing details limit how far the result can be generalized, particularly across centers with different vaccination workflows and monitoring practices.
Even so, the practical message is straightforward. Preterm infants remain at risk for post-vaccination cardiorespiratory instability, and physiologic complexity measures may have a role in identifying which babies need the closest watch. This study leaves sample entropy in an intermediate position: not strong enough to predict cardiorespiratory events, but potentially informative when the question is whether an infant will need active support after routine immunization. The next step is not a broader claim. It is a more precise test of whether this marker can be integrated with gestational age and bedside monitoring into a usable neonatal risk model.