Introduction

Assessment of regional tissue oxygenation (rSO2) utilizing near-infrared spectroscopy (NIRS) is becoming more common in the neonatal intensive care unit (NICU).1,2,3,4,5,6,7,8,9,10 While normative studies are required to delineate expected NIRS values and signal behavior in neonates, evidence supporting the potential utility of NIRS monitoring is growing. Currently, cerebral rSO2 monitoring is most prevalent in the neonatal literature; however, recent studies continue to advance the use of peripheral NIRS monitoring.4,9,11,12,13,14,15,16 This review aims to highlight current evidence and assist in structuring the bedside use of noninvasive NIRS-based rSO2 monitoring in the NICU setting.

Background

NIRS provides regional hemoglobin oxygenation status using a technique similar to pulse oximetry (SpO2). Both forms of monitoring take advantage of differential absorption spectra between oxygenated and deoxygenated hemoglobin to visible light in the near-infrared range.1,4,9,17,18,19,20,21 This difference is then expressed as a ratio [(oxyhemoglobin/oxyhemoglobin+deoxyhemoglobin) ×100] for real-time trending. The primary contrast between SpO2 and NIRS involves hemoglobin oxygenation data processing prior to display on bedside monitors.1,4,9,17

In pulse oximetry, hemoglobin oxygenation data are reported solely from pulsatile, or arterialized, sources with non-pulsatile sources (e.g., venous, capillary) mathematically subtracted from the data output stream.18,22 Peripheral pulsatile vessels conveying SpO2 data are anatomically defined as arteries originating from the aorta. Therefore, as arteries typically contain highly oxygenated blood, SpO2 represents a global estimate of precapillary oxygen delivery to peripheral tissues.17,22

In contrast to pulse oximetry, NIRS expresses full tissue hemoglobin oxygenation without subtraction of non-pulsatile data. Thus, NIRS represents the regional oxygenated to total hemoglobin ratio (rSO2) for the combined arterial, capillary, and venous hemoglobin sources underlying a given sensor.1,4,5,9,10,14,17,23,24,25,26 Anatomically, at any given time, the blood contained within an individual tissue segment exists in a generally accepted vascular distribution of approximately 20% arterial, 75% venous, and 5% capillary.9,25,27,28 Therefore, as NIRS provides predominantly post-capillary tissue oxygenation information, it may be considered a surrogate estimate of local tissue oxygen utilization.

The utility of NIRS in clinical practice may be considered in relation to the theoretical critical oxygen delivery point concept.1,29,30 According to this postulate, tissue oxygen utilization remains relatively constant across a range of oxygen delivery with oxygen extraction stable to meet tissue metabolic requirements. Initially, decreases in oxygen delivery produce subtle increases in oxygen extraction to maintain tissue homeostasis. However, with further oxygen deprivation, a threshold is crossed (“the critical O2 point”) beyond which ongoing tissue oxygen utilization becomes oxygen delivery-dependent. Below the critical O2 point, dramatic increases in oxygen extraction are theorized as necessary to maintain tissue metabolic needs. This circumstance is often accompanied by clinical signs consistent with oxygen delivery-consumption mismatch (see Fig. 1).1,30

Fig. 1
figure 1

Theoretical relationship described between oxygen delivery and both oxygen utilization/consumption and oxygen extraction

Full size image

Integration of SpO2 and NIRS data allows for crude estimation of tissue oxygen extraction. Fractional tissue oxygen extraction (FTOE) is calculated as follows:8,9,31,32,33,34,35,36,37,38,39,40,41

$${\mathrm{FTOE}} = \left[ {\left( {{\mathrm{SpO}}_2-{\mathrm{regional}}{\mathrm{rSO}}_2} \right){\mathrm{/SpO}}_2} \right]$$

In this approach, the numerator represents local tissue oxygen extraction, while the SpO2 in the denominator converts this extraction into a function based on oxygen delivery. While common in clinical studies, FTOE cannot distinguish rSO2 changes resulting from diminished tissue perfusion from those occurring due to increased tissue oxygen extraction/utilization. However, it is feasible that FTOE trending may provide noninvasive clinical decision support at the bedside.

Current state of evidence

Dozens of examples of NIRS monitoring in the NICU setting have been described in the literature.3,8,9,10,15,17,42,43 While cerebral monitoring is most frequently studied, peripheral monitoring of renal, splanchnic, and/or peripheral muscle rSO2 has been reported as well. In these reports, a wide range of NIRS monitoring practices have been described as potential adjuncts in the management of an equally broad number of neonatal conditions.5,8,9,10,11,16,44,45,46,47 These data have highlighted the potential utility of “oxygenation adequacy” monitoring with a goal of improving neonatal outcomes.1,3,8 However, in a recent survey of NIRS uptake and usage among NICU clinicians, concerns remain regarding actual evidence-based clinical utility among the general NICU population.48

Currently, cerebral rSO2 monitoring has evolved into routine practice in certain limited scenarios. Over the past decade, the neonatal neurocritical care paradigm, incorporating cerebral imaging, amplitude-integrated electroencephalography, and cerebral NIRS, has evolved for management of neonates at high risk for neurologic disorders.4,34,49,50,51,52,53,54,55 Additionally, the SafeBoosC studies have examined the practicality and utility of cerebral rSO2 monitoring and guideline-driven targeting on clinical outcomes among extremely premature infants.56,57,58,59,60,61,62 Ongoing research into these paradigms remains an area of highly active inquiry.

In contrast, no specific framework currently exists to routinely incorporate both cerebral and peripheral NIRS monitoring in neonatal critical care.3,5,9,54 Interestingly, given the physiologic redundancy of cerebral perfusion, along with well-described neuroprotective blood flow redistribution during neonatal distress, one can argue that peripheral rSO2 monitoring may be more sensitive than cerebral for demonstrating acute perturbations in physiologic oxygenation homeostasis.27,37 Synthesizing these NIRS monitoring practices into a coherent bedside strategy remains a work in progress.

In general, there are two approaches to NIRS monitoring in the NICU setting. In this review, disease- or event-specific monitoring, referred to as “responsive” monitoring, is defined as goal-oriented use of NIRS for specific clinical or bedside purposes. The second approach, referred to as “routine” monitoring, concerns the continuous use of NIRS in a framework similar to current bedside cardiopulmonary vital sign monitoring practices. The remainder of this review is dedicated to delineating the potential benefits and pitfalls of these neonatal NIRS monitoring strategies.

“Responsive” monitoring

The “responsive” rSO2 monitoring paradigm concerns the use of NIRS to generate specific information for particular clinical management decisions. In this decision-support framework, responsive monitoring may also be considered as a method to gauge clinical responses to therapeutic maneuvers.1,5,8,9,10 This form of NIRS monitoring has also been periodically used to examine the effects of routine NICU care practices on tissue oxygen utilization. In this section, we will review evidence regarding responsive NIRS monitoring on the clinical evaluation of anemia, patent ductus arteriosus, and feeding intolerance. We will additionally highlight examples of NIRS-based procedural monitoring to demonstrate the responsive paradigm.

Anemia

NIRS monitoring has been utilized in numerous studies to evaluate the central and peripheral oxygen extraction and utilization effects of anemia. These studies generally report baseline central and peripheral rSO2 and/or FTOE followed by responses to transfusion. As expected, packed red blood cell transfusion, by increasing global oxygen carrying capacity, results in decreased oxygen extraction evidenced by a rise in regional rSO2.32,36,63,64,65,66,67,68,69,70,71,72,73 Interestingly, these responses to transfusions, which knowingly increase both overall blood volume and the concentration of chromophores per unit blood volume, are tissue specific, with peripheral tissues often demonstrating a more robust response compared to the brain.36,37,63,66

Using transfusion responses as a guide, current research seeks to derive anemia-specific cerebral and/or somatic rSO2 or FTOE patterns to assist in transfusion practices.1,10,37,73,74 In 2002, Wardle et al.70 conducted a study using elevated forearm FTOE as a transfusion threshold, though data were skewed by transfusions provided for clinical concerns within the NIRS-based transfusion group. Defining FTOE-based transfusion thresholds remains a work in progress, especially as oxygen extraction behavior in anemia differs in a tissue-specific manner.36,37,63,66,73 It appears that NIRS monitoring can provide important information for transfusion management, although specific practices remain elusive at present.

In addition, several reports have identified NIRS-based patterns as potentially predisposing factors for transfusion-related acute gut injury, or necrotizing enterocolitis occurring in close contiguity following blood transfusion. In some reports, reductions in splanchnic NIRS and/or increases in signal variability have been associated with necrotizing enterocolitis following packed red blood cell transfusion.75,76,77,78 However, another study demonstrated no differences in NIRS parameters following transfusion.72 Larger studies into the relationship between blood transfusions and splanchnic oxygenation are required to resolve these disparate findings.

Hemodynamically significant patent ductus arteriosus

The evaluation of suspected hemodynamically significant patent ductus arteriosus (hsPDA) poses an interesting avenue of inquiry for neonatal NIRS monitoring. Physiologically, ductal flow across an hsPDA is inferred to cause pulmonary overcirculation at the expense of systemic flow.79,80,81 Currently, management of hsPDA remains a highly controversial issue in neonatology.79,80,82,83,84,85,86,87,88 Indeed, the definition of hemodynamic significance remains contested at present.79,81 In addition, agreed-upon specific echocardiographic parameters to define hsPDA remain undefined.89 Finally, adjunctive PDA-related management, including fluid intake practices, also controversial, limit the generalizability of PDA-related studies.81,90 Given the above, the utility of NIRS within hsPDA management has yet to be clearly defined.

Several studies have assessed whether NIRS monitoring may aid in determining the hemodynamic significance of PDA-related shunting. While some reports have demonstrated physiologically plausible rSO2 decreases and/or FTOE increases in the presence of an hsPDA, these findings have not been consistent across studies.91,92,93,94,95 In addition, responses to both medical and surgical hsPDA therapies have been evaluated for their effects on tissue oxygenation and oxygen extraction.96,97,98,99

Whether cerebral and/or peripheral NIRS monitoring may aid in PDA management remains speculative. Moreover, significant differences exist between cerebral and somatic oxygenation in the context of hsPDA, predominantly accounted for by cerebral autoregulation.91 Further evaluation of peripheral NIRS monitoring is required to determine potential sensitivity compared to cerebral rSO2 as a marker for tissue oxygenation failure in the context of hsPDA. Given the physiology of hsPDA-related shunting and its plausible effects on tissue oxygen delivery/consumption balance, future larger-scale studies with standardized PDA-related definitions and NIRS data collection practices are needed.

Feeding intolerance/necrotizing enterocolitis

Numerous reports have been published on the potential utility of splanchnic NIRS monitoring for the evaluation of feeding intolerance and necrotizing enterocolitis.11,16,100 Common to these studies is a splanchnic observational data set, often paired with cerebral monitoring, that demonstrates changes occurring during various feeding regimens. Moreover, several reports have evaluated changes in splanchnic NIRS data as a potential warning sign for future development of necrotizing enterocolitis (NEC), although with widely variable results.78,100,101,102,103,104 Importantly, these numerous papers highlight the difficulty of analyzing real-time splanchnic monitoring data, predominantly due to inherent signal variability.

For example, initial reports demonstrated daily mean splanchnic rSO2 and FTOE values in premature neonates with expected changes occurring during the first two postnatal weeks.105 This same study reported decreased splanchnic rSO2 and increased FTOE in infants with feeding intolerance and further noted persistently low rSO2 with low variability as a potential risk factor for NEC, a finding replicated in a larger prospective observational study.103 In another report, splanchnic NIRS monitoring was able to aid in distinguishing surgically complicated NEC from uncomplicated cases, but not differentiating the absence of NEC from active disease.104

Other studies have reported correlations between splanchnic NIRS and superior mesenteric artery Doppler values with promising results.101,106,107 While splanchnic NIRS may provide valuable information on intestinal perfusion, these correlative data cannot take into account changes in tissue oxygen demand, for example as a result of feeding. Furthermore, recent work has demonstrated a positive correlation between intestinal peristaltic activity and splanchnic rSO2 in premature neonates, with implications for prospective assessment of feeding tolerance.108

Additional studies have investigated responses to various feeding regimens, including continuous vs. bolus feeding,109 and breast milk vs. formula feeding.110 Splanchnic rSO2 data has also been reported to correlate with early feeding intolerance in premature neonates, both using continuous and bolus feeding regimens.111,112,113 Similar reports have focused on the intrauterine growth restriction population given the high incidence of feeding intolerance in this group.114

When viewed as a series, these studies suggest a pressing need for further inquiry. Importantly, splanchnic NIRS monitoring data is often highly variable at baseline, with frequent signal drop-out. Inherent splanchnic rSO2 quiescent variability has been reported as approaching 25%, with differences in variability depending on how raw splanchnic data are averaged.27 Abdominal sensor placement, specifically overlying the liver vs. the infraumbilical position, can affect splanchnic NIRS monitoring values and variability.115 It thus follows that feeding- or NEC-related changes in splanchnic rSO2 need to be carefully assessed in the context of this variability. Future work is crucial to define appropriate data management techniques to account for this variability in addition to assessing comparisons between splanchnic and other, less variable monitoring sites.

Procedure-related monitoring

Tissue rSO2 monitoring can demonstrate subtle changes in tissue oxygen utilization behavior, often before the onset of clinical symptoms.1,8,9,10 Therefore, one can argue that NIRS monitoring could provide bedside information on factors related to routine NICU care. Currently, studies on specific procedure-related monitoring are most common in the perioperative environment.116,117,118 However, NICU procedure-related monitoring has demonstrated some interesting results that warrant further investigation.

For example, several reports have demonstrated decreases in cerebral oxygenation following umbilical arterial blood drawing.119,120,121,122,123 In one study across a broad range of gestational ages, cerebral rSO2 decrements were attributed to faster blood draws.123 In another report specific to very low birth weight neonates, larger volumes of blood draws were more likely to be associated with cerebral rSO2 decreases.122 In a further study on very low birth weight neonates, the common occurrence and natural course of these decrements was demonstrated, with an additional observation of a prolonged cerebral rSO2 recovery time in some neonates.120 Finally, a follow-up study on cerebral rSO2 monitoring demonstrated cerebral decrements occurring most commonly among neonates with lower baseline SpO2, PaO2, and cerebral rSO2, thus insinuating subject-related factors as contributing to sensitivity to this common procedure.119

Additional studies have assessed the relationship between neonatal head and/or body positioning and tissue rSO2 parameters. In one study evaluating a series of different body positions in premature neonates, no position-related cerebral rSO2 differences were observed.124 Similar findings were also observed for both cerebral and mesenteric rSO2 in clinically stable very low birth weight neonates following both pre- and postprandial body position changes.125 Two additional studies have employed cerebral NIRS monitoring to demonstrate that responses to head-tilting maneuvers, as indicators of neonatal cerebrovascular control, differ between premature and term neonates,126 and in response to prone positioning.127

These reports indicate the versatile nature of NIRS monitoring among the neonatal population. As familiarity with this monitoring technique increases, further publications will likely demonstrate unique and interesting uses of rSO2 monitoring. While NIRS monitoring could conceivably allow for more personalized, patient-centered approaches to NICU care, whether this modality will ultimately improve short- and long-term outcomes remains unknown at this time.

“Routine” monitoring

The “routine” NIRS monitoring approach involves rSO2 monitoring as a continuous, trendable metric used in routine neonatal practice. In this paradigm, cerebral or peripheral rSO2 and/or FTOE is considered as an additional bedside vital sign. As above, several studies have demonstrated that NIRS-specific changes often occur prior to the onset of clinically noticeable sequelae. At present, however, whether this monitoring strategy can function as a real-time early-warning tool has yet to be determined. In this section, we will review important information and pitfalls concerning the signal variability observed in central and peripheral rSO2 monitoring. We will also explore some current approaches utilizing routine neonatal NIRS monitoring, including neurocritical monitoring and perioperative management. Finally, we will review correlations between rSO2 data and invasive monitoring data.

Normative data/variability

Several studies have reported reference ranges for both cerebral and peripheral rSO2 among neonates. These studies include both full-term and premature neonates and, taken together, provide data on cerebral, renal, and splanchnic rSO2. Furthermore, while most studies focus on short-term NIRS monitoring (e.g., during birth transition),38,53,128,129,130 others have sought to demonstrate normative NIRS ranges over longer postnatal intervals.14,31,103,105,131,132

In addition to establishing normative ranges, the inherent variability within individual rSO2 data streams has been described, although in small studies.14,27,52,53 In most studies in which cerebral and peripheral NIRS monitoring is performed, variability differences between NIRS signals have been observed. In Table 1, selected studies are displayed to demonstrate the varied approaches that have been used in managing rSO2 information as a continuous data stream. Determining how to account for normal baseline variability (“signal noise”) within NIRS data sets remains as yet undefined, especially with regard to splanchnic monitoring.11,16,27 Clearly, future studies are required to standardize data collection and management techniques to properly evaluate the use of NIRS as a routine monitoring technique.

Table 1 Comparisons of NIRS-based oxygenation studies
Full size table

Additional attention to splanchnic NIRS monitoring is specifically warranted in this regard. As above, multiple studies have examined the potential utility of splanchnic NIRS monitoring, especially regarding feeding intolerance and necrotizing enterocolitis.11,16 However, whether splanchnic variability results from actual changes in mesenteric oxygen delivery/consumption balance vs. abdominal sensor placement over a hollow anatomic cavity has yet to be conclusively determined. Until larger-scale, standardized normative NIRS data are obtained to address this quandary, careful critical attention is required in assessing the actual bedside utility of splanchnic rSO2 monitoring.

Neurocritical monitoring/cerebral autoregulation

Over recent years, cerebral rSO2 monitoring has been incorporated into the neurocritical monitoring paradigm. In this approach, ongoing neurologic assessment is performed for neonates at highest risk for neurodevelopmental impairment. In addition to neurologic imaging and amplitude-integrated electroencephalography, changes in cerebral oxygenation are considered in managing this vulnerable population.52,53,133

Continuous, routine cerebral rSO2 monitoring has also been used to describe cerebral autoregulation in neonates, or an ability to maintain stable cerebral perfusion across a range of blood pressure.2,43,134,135 Furthermore, several reports have described decreases in cerebral autoregulatory capacity, or pressure-passive circulation, as an indicator of worsening clinical status among neonates.50,118,134,136,137 Moreover, these NIRS-based changes may be observable earlier than evidence of clinical deterioration, thus promoting careful cerebral rSO2 monitoring as a potential early-warning tool.51,138,139,140 Although certainly promising, further inquiry is needed to assess how to best incorporate this strategy into routine practice and whether actual neurodevelopmental outcomes can be improved with this approach.4,23,42,54,135,141,142

Attempting to maintain cerebral oxygenation within a set range has also been an active area of inquiry.143 In the original SafeBoosC observational trial, investigators demonstrated the feasibility of maintaining preterm neonates within a predefined cerebral range (55–85%).144 Subsequently, the SafeBoosC-II trial randomized extremely premature neonates to either guideline-based cerebral NIRS targeting using the above range vs. NIRS monitoring with no targeting.56,57 In the SafeBoosC-II trial, subjects in the targeting group were successfully maintained within range compared to the control group,44 with NIRS-based alarms guiding interventions in approximately 25% of cases.62 Thus, an increased number of bedside interventions occurred in the NIRS targeting group, with uncertain impact. However, this feasibility study was not powered for clinical outcomes. Further analyses demonstrated no statistically significant differences between groups with regard to cranial ultrasound and magnetic resonance imaging (MRI) evidence of brain injury;60 amplitude-integrated electroencephalography and brain injury-related biomarkers;58,61 or 2-year neurodevelopmental outcomes.59 A further study, adequately powered for clinical outcomes, is currently being planned to comprehensively assess the cerebral rSO2 targeting approach among premature neonates (SafeBoosC-III; see http://www.safeboosc.eu).

Cerebral rSO2 targeting can also be considered from a “pattern recognition” framework.10 In this approach, rSO2 targeting is based on the appearance of departures from baseline on longitudinal monitoring. At present, cerebral rSO2 and FTOE baseline values have been best described among the preterm population during the first 72 postnatal hours.31 In this paradigm, careful multimodal analysis is performed to determine and act upon the most likely etiology of cerebral oxygen delivery/consumption imbalance (e.g., hypo/hyperoxia, hypo/hypercapnia, anemia, hypotension, hypoglycemia, intracranial hemorrhage, and impaired cerebral autoregulation). Although further study is needed, this pattern recognition-based paradigm appears promising as a likely future practice to incorporate routine NIRS monitoring into NICU care.

Perioperative and ECMO monitoring

One of the most commonly reported uses of NIRS monitoring in the neonatal population occurs in the perioperative and extracorporeal membrane oxygenation (ECMO) setting. Particularly, cerebral and occasionally peripheral rSO2 monitoring are frequently utilized in the context of congenital heart disease repair both intraoperatively and in postoperative management. In the operating room, NIRS-based evaluation of tissue oxygen delivery/consumption balance has become a fairly standard adjunctive vital sign used to aid in optimizing tissue perfusion during surgery and/or ECMO.117,145,146,147

In addition, perioperative rSO2 monitoring has demonstrated efficacy in early recognition of low perfusion states with important implications for operative morbidity and mortality.136,139,148,149,150,151 Furthermore, postoperative cerebral rSO2 monitoring has been observed to correlate with short-term lactates and both survival and neurodevelopmental outcomes following cardiac surgery.116,152,153,154,155 In addition, changes in both intraoperative and postoperative renal rSO2 parameters have been shown to correlate with later development of acute kidney injury.156,157 Splanchnic monitoring has additionally been investigated as potentially associated with the development of necrotizing enterocolitis among postoperative cardiac surgery patients.102 Finally, cerebral rSO2 monitoring has been utilized in neonates undergoing ECMO support,158 with one study demonstrating a correlation between cerebral rSO2 values and overall survival in this population.145

Perioperative NIRS monitoring has also been described following non-cardiac procedures.159 For example, cerebral and renal rSO2 have been reported as providing useful information during and following alimentary tract procedures,160 although specific management recommendations remain lacking. In addition, intraoperative cerebral monitoring was described as providing clinically actionable information allowing for respiratory adjustments during esophageal atresia repair.161,162

While standardization and best practices have yet to be formally derived, these examples offer insight into the possible benefits of both central and peripheral NIRS monitoring among critically ill neonates. Further research is required to aid in extrapolating the experience of perioperative NIRS monitoring to NICU-specific practices.

Correlational analyses

Several reports have demonstrated correlations between conventional monitoring data and tissue-specific NIRS-based parameters. These small-scale observational studies have investigated these relationships both in the NICU and among the operative and ECMO populations. Although larger, standardized studies are required to further investigate these relationships, the available data do support the physiologic plausibility underlying NIRS monitoring.

For example, in an observational analysis of cerebral NIRS, both rSO2 and FTOE correlated with heart rate, respiratory rate, and pulse oximetry among 30 to 42 weeks gestational age neonates during the first six postnatal hours.163 However, no correlation was observed between cerebral NIRS and arterial oxygenation in a separate study.164 Interestingly, in another study, cerebral FTOE, although not pulse oximetry, was found to be associated with severe retinopathy of prematurity among preterm neonates <30 weeks gestation.41 In addition, significant correlations were observed between both cerebral and splanchnic rSO2 and FTOE with mean blood pressure in a small cohort of neonates ranging from 22 to 42 weeks gestational age.35 In another study, renal rSO2 was shown to directly correlate with urine output among a cohort of neonates on ECMO for congenital diaphragmatic hernia.146 Finally, cerebral rSO2 has been observed to correlate with brain MRI-based regional cerebral blood flow values among neonates with hypoxic–ischemic encephalopathy.165

Other studies have sought to determine correlations between invasive monitoring parameters and NIRS-based values. In one study, cerebral and renal FTOE were observed to inversely correlate with hematocrit level, although a similar relationship could not be demonstrated for splanchnic oxygenation likely due to excessive signal variability.37 In addition, as previously discussed, physiologically plausible responses to packed red blood cell transfusion further imply a correlation between rSO2 and FTOE values and degree of anemia. However, at present, organ-specific FTOE-based transfusion thresholds have not yet been defined.73,74 Although beyond the scope of this NICU-specific review, it is worth mentioning that several studies have reported plausible correlations between tissue oxygenation values and invasive monitoring parameters among the pediatric and adult operative and intensive care populations.

Future direction/vision

Tissue oxygenation monitoring offers an interesting and objective perspective on a neonate’s overall clinical status. Using this noninvasive approach, physiologic changes in tissue oxygen delivery/utilization behavior can be observed in real time. However, as we have described in detail, carefully designed broad-based future studies are required to maximize the potential of this monitoring approach.

The routine monitoring approach as described above requires familiarity with normal tissue-specific oxygenation behavior. Comparing NIRS monitoring to pulse oximetry, there is a broad difference in experience regarding the perception of “normal” between these two monitoring approaches. Moreover, when to intervene and what to expect in response to a pulse oximetry-based intervention is already well understood. Despite numerous studies, this is not necessarily the case for NIRS monitoring at this time.

One concern is that currently available NIRS monitoring devices report raw rSO2 data in short intervals with trends typically available as a horizontal graph. As numerous factors affect real-time rSO2, with individual tissues demonstrating unique oxygenation behavior, the practicality of using real-time raw rSO2 signals in bedside NICU care is challenging. An interesting improvement would be the addition of pulse oximetry data to NIRS devices. This would allow more rapid analysis of a given perturbation in rSO2 signals along with allowing real-time tissue-specific FTOE trending.

In addition, NIRS signals are much more variable than pulse oximetry. It is therefore difficult to determine in real-time whether an individual rSO2 departure from baseline represents an actual clinical event vs. a momentary change. Thus, it is currently uncertain whether real-time raw rSO2 displays on NIRS devices are the best method to report this tissue-specific behavior. Another advance could have future devices displaying a more integrated NIRS profile on-screen. Such a display would include raw NIRS (and/or FTOE) data along with short- and longer-term trends to allow for enhanced contextualization and improved clinical-decision support.

Most importantly, large-scale studies are required to establish a normative data set, including region-specific baseline variability, among the overall NICU population. To accomplish this goal, a dedicated research consortium may be considered to assemble these data in a reference registry from which normative analyses may be performed. Discovering the true potential of routine rSO2 monitoring depends on the establishment of agreed-upon norms, standards, and expectations.

Conclusion

Current NICU monitoring practices are primarily centered around indirect measures, including cardiorespiratory vital signs and serum markers of impaired oxygen delivery/consumption balance. NIRS monitoring, on the other hand, offers measured direct real-time tissue oxygen extraction/utilization behavior in a noninvasive monitor. While studies have shown potential uses for NIRS monitoring in the NICU, elucidating specific strategies, best practices, and effects on long-term outcomes remains a work in progress.