OBJECTIVE. The absence of cerebral autoregulation in preterm infants has been associated with adverse outcome, but its bedside assessment in the immature brain is problematic. We used spatially resolved spectroscopy to continuously measure cerebral oxygen saturation (expressed as a tissue-oxygenation index) and used the correlation of tissue-oxygenation index with spontaneous fluctuations in mean arterial blood pressure to assess cerebral autoregulation.
PATIENTS AND METHODS. The tissue-oxygenation index and mean arterial blood pressure were continuously measured in very premature infants (n = 24) of mean (±SD) gestational age of 26 (±2.3) weeks at a mean postnatal age of 28 (±22) hours. The correlation between mean arterial blood pressure and tissue-oxygenation index in the frequency domain was assessed by using cross-spectral analysis techniques (coherence and transfer-function gain). Values of coherence reflect the strength of linear correlation, whereas transfer-function gain reflects the amplitude of tissue-oxygenation index changes relative to mean arterial blood pressure changes.
RESULTS. High coherence (coherence ≥ 0.5) values were found in 9 infants who were of lower gestational age, lower birth weight, and lower mean arterial blood pressure than infants with coherence of <0.5; high-coherence infants also had higher median Clinical Risk Index for Babies scores and a higher rate of neonatal deaths. Coherence of ≥0.5 predicted mortality with a positive predictive value of 67% and negative predictive value of 100%. In multifactorial analysis, coherence alone was the best predictor of mortality and Clinical Risk Index for Babies score alone was the best predictor of coherence.
CONCLUSIONS. High coherence between mean arterial blood pressure and tissue-oxygenation index indicates impaired cerebral autoregulation in clinically sick preterm infants and is strongly associated with subsequent mortality. Cross-spectral analysis of mean arterial blood pressure and tissue-oxygenation index has the potential to provide continuous bedside assessment of cerebral autoregulation and to guide therapeutic interventions.
- cerebrovascular autoregulation
- cerebral oxygenation
- near-infrared spectroscopy
- premature infants
- spectral coherence
Cerebral autoregulation limits cerebral blood flow (CBF) variation over a range of cerebral perfusion pressures1 and is a key physiologic mechanism for ensuring adequate perfusion and oxygenation of the brain. In the preterm brain, however, the presence and the characteristics of cerebral autoregulation are still poorly characterized despite intensive study. A number of studies have suggested that cerebral autoregulation is lost in sick preterm infants, predisposing them to hemorrhagic and ischemic cerebral injury.2–4 However, autoregulation may be absent not only in sick preterm infants but also in those who are clinically well.5,6 On the other hand, there is evidence that CBF is independent of mean arterial blood pressure (MAP) over a wide pressure range in preterm infants, suggesting that autoregulation may actually be effective in the immature brain.7,8
Because autoregulation is a dynamic process,2,9 new approaches to assessing cerebral perfusion and oxygenation continuously in the preterm brain are needed, if the uncertainty surrounding autoregulation is to be resolved. Conventional near-infrared spectroscopy (NIRS) is able to quantify relative changes in hemoglobin concentrations but is limited to intermittent absolute measurements of cerebral hemodynamics10,11 and oxygenation.12 Spatially resolved spectroscopy (SRS), a recent development of NIRS, provides a continuous, quantitative measurement of cerebral oxygenation expressed as a tissue-oxygenation index (TOI).13,14 TOI, the ratio of oxyhemoglobin to total hemoglobin, reflects oxygen saturation in all of the vascular compartments but is most influenced by the venous compartment, which represents ∼75% of total cerebral blood volume.15 Assuming constant cerebral metabolic rate for oxygen, changes in cerebral venous oxygen saturation and, therefore, changes in TOI will parallel variations of CBF according to the Fick principle. Should autoregulation fail, changes in CBF will follow changes in MAP, and, hence, TOI as a surrogate for CBF should also parallel MAP.
In this study, we exploited spontaneous variations in MAP and TOI to assess cerebral autoregulation in very preterm infants. We aimed to use cross-spectral analyses (coherence and transfer function analyses) to quantify the frequency-dependent covariation of MAP and TOI as a means of assessing autoregulation. We hypothesized that high values of the coherence and transfer functions, reflecting impaired autoregulation, would be related to adverse clinical indices in sick preterm infants.
A cohort of extremely preterm infants with gestational ages of <32 weeks and with indwelling arterial catheters was studied in the NICU at University College London Hospital. Infants with major congenital anomalies were excluded. The study had ethical approval from Joint University College London/University College London Hospital Committees on the Ethics of Human Research. Written, informed parental consent was obtained before each study. Twenty-four infants (15 male and 9 female) born at mean (±SD) gestational age of 26 (±2.3) weeks, with mean birth weight of 815 (±253) g, were studied at mean postnatal age of 28 (±22) hours. All of the infants were intubated and receiving mechanical ventilation during the study period. The Clinical Risk Index for Babies (CRIB) was calculated on the basis of birth weight, gestational age, congenital malformation, base excess, and fractional inspired oxygen in the first 12 hours of life.16 Cranial ultrasounds were performed daily from the time of admission. Clinical management of the infants was at the sole discretion of the attending neonatologist.
SRS (NIRO-300 spectrophotometer, Hamamatsu Photonics KK, Hamamatsu City, Japan) uses continuous wave light and light detection at multiple distances. The NIRO-300 system transmits light at 4 wavelengths in the NIRS at 775, 810, 850, and 905 nm, via a fiber optic bundle, delivered to the infant by an emission probe. Three aligned photodetectors are housed in a single probe, which is spaced 4 cm from the emission probe. Both the emission and detection probes are placed over the temporoparietal region of the infant. SRS uses the slopes of near infrared light attenuations versus distances measured from all 3 of the photodetectors and computes continuously the absolute ratio of oxyhemoglobin to total hemoglobin and, hence, an absolute average tissue-oxygen saturation (TOI).13,14
Measurements of Physiologic Variables
MAP waveform was continually measured from an indwelling peripheral or umbilical arterial catheter connected to a transducer monitoring system zeroed at the midmaxillary line, recorded, and displayed in real time at the bedside (Agilent Viridia CMS 2001 modular system, Phillips Medical, Da Best, Netherlands). Arterial oxygen saturation, transcutaneous partial pressure of carbon dioxide, and heart rate were simultaneously recorded using the same bedside system. Using in-house software written in Labview (National Instruments, Austin, TX), both the physiologic data from the Agilent system and the optical data from the NIRO-300 were collected simultaneously and downloaded onto a personal computer for subsequent analysis.
Studies were performed over 2 to 4 hours in each infant. Physiologic data including MAP were collected at 1 Hz. TOI data were collected at 6 Hz and then downsampled to 1 Hz to allow offline coherence and transfer-function analysis along with MAP.
Uninterrupted recordings of 20 minutes with stable arterial oxygen saturation were used for coherence analysis. SRS recordings were interrupted if movement artifacts caused a baseline shift. In practice this resulted in interruption of recordings during spontaneous head movements, physical examinations, radiographs, clinical procedures, and other events requiring handling of the infant. Arterial blood gas sampling also interfered with recordings because of loss of the arterial pressure signal.
Correlation between continuous measurements of MAP and cerebral TOI were quantified in a frequency-specific manner by established spectral analysis techniques, the coherence function and the transfer function analysis.17,18 Computations were performed by using a scientific software package (MatLab; MathWorks, Inc, Natick, MA). Each 20-minute epoch was divided into 5 segments of 10 minutes, with successive segments overlapped by 75%. Averages for MAP and TOI obtained in each segment were subtracted from the data to eliminate any contribution of the DC component to low-frequency power. These time domain data were then multiplied by a full cosine-tapered (Hanning) window to minimize spectral leakage and transformed into power spectral densities (PSDs) by Fourier transform. The PSDs obtained for each of the 5 overlapping segments were averaged to minimize contributions from variable noise and to sharpen reproducible spectral components. PSD of TOI and MAP (PTOI[f] and PMAP[f]), and their cross PSD (PTOI,MAP[f]) were used to generate the coherence function for the 2 signals |PTOI,MAP(f)|2/[|PTOI(f)| |PMAP(f)|], and their transfer function, PTOI,MAP(f)/PMAP(f).
The coherence function characterizes the frequency-dependent correlation of MAP with TOI. The numeric value of the coherence function (Coh) at a given frequency ranges between 0.0 and 1.0, depending on the degree of correlation between the 2 waveforms; a Coh value of 1.0 indicates perfect frequency-specific correlation, whereas 0.0 indicates complete lack of correlation. By convention, Coh values of ≥0.5, are considered to represent significant concordance between the 2 waveforms.2,17 High concordance between MAP and TOI and, therefore, high Coh values, can be anticipated when autoregulation is impaired, as in this circumstance where CBF and oxygen delivery become pressure passive.
The transfer function models a linear input-output system with the MAP signal as the input and the TOI signal as the output. The magnitude of the transfer function (transfer-function gain [G]; percentage per millimeter of mercury) represents the magnitude of change in the output (TOI) arising from a unity change in the input (MAP) at a given frequency. Thus, impaired autoregulation, signified by change of large magnitude in TOI for a change in MAP, would give rise to a high magnitude for G. Combining the use of Coh and G for simultaneous measurements of MAP and TOI, frequency-dependent (dynamic) cerebral autoregulation can be quantified.
Coh and G values were computed for frequencies between 0.003 and 0.100 Hz. The lower limit of 0.003 Hz was chosen to be above the lower frequency limit dictated by the length of the data segment (20 minutes). According to Nyquist sampling theorem, our data sampling frequency of 1 Hz would allow detection of frequencies ≤0.5 Hz. We chose the upper limit to be 0.1 Hz because autonomic vascular control and cerebral autoregulatory processes occur well below this frequency.19
For each 20-minute segment, the Coh and G values were averaged over the frequency bands of 0.003–0.020 Hz (ultralow frequency [ULF]), 0.020–0.050 Hz (very low frequency [VLF]), and 0.050–0.100 Hz (low frequency [LF]). Where more than one 20-minute segment was analyzed in an infant, the maximum Coh and G values were selected for further analysis.
Cranial ultrasonography was performed on admission and daily thereafter. Standard coronal and sagittal views were recorded. Intraventricular hemorrhage (IVH) was categorized and graded according to previously published criteria.20 Specifically, in grade I IVH, echodensity is confined to the germinal matrix. In grade II IVH, echodensity is present in a nondistended lateral ventricle, whereas in grade III lesions, echodensity is present in a distended lateral ventricle. In grade IV IVH, echodensity has extended into the periventricular parenchyma.
Infant data were contrasted after grouping according to (1) high (≥0.5) or low (<0.5) Coh values, (2) survival or death, and (3) treatment or no treatment with an inotropic agent. Differences between groups were compared with Student's t test for continuous variables and by the Fisher's exact test for dichotomous variables (SigmaStat; SPSS Inc, Chicago, IL). Relationships of Coh to G and clinical variables were determined using Pearson product moment correlation. To determine predictive models for Coh and mortality, we used best-subset multiple linear and logistic regression (Statistica; StatSoft, Inc, Tulsa, OK). In all of the tests, a P value of <.05 was considered significant.
Successful recordings were obtained from 24 infants. After exclusion of segments containing movement artifacts and other interruptions, an average (±SD) of 2.6 (±1.8) 20-minute segments were analyzed for each infant. Six infants died during the neonatal period, of which 5 died within the first week of life. Causes of death included respiratory failure (n = 1), necrotizing enterocolitis (n = 4), and withdrawal of intensive care after severe (grade IV) IVH (n = 1). Seven infants were found to have moderate-to-severe IVH (grade II–IV IVH) on cranial ultrasound examinations; 6 infants had grade II IVH, and 1 infant had grade IV IVH. At the time of study, 8 infants were receiving treatment with inotropic medications for hypotension (all were treated with dopamine [5–20 μg/kg per minute], and 3 were also treated with dobutamine [5–20 μg/kg per minute]).
Coherence and Cerebral Hemodynamics
Group Coh values (mean ± SD; n = 24) were 0.46 ± 0.12 for ULF, 0.41 ± 0.09 for VLF, and 0.35 ± 0.06 for LF. In 9 infants, individual Coh values of ≥0.5 were observed at the ULF range, corresponding with a cycle length of 5 minutes and 50 seconds. Table 1 details the clinical characteristics of the 9 infants with Coh values of ≥0.5 in the ULF range. Figure 1 illustrate the group-averaged Coh in the frequency domain for the 9 infants with ULF Coh values of ≥0.5 and the remaining 15 infants with ULF Coh values of <0.5.
Figure 2 illustrate the variable Coh indicating the degree of concordance between the MAP and TOI waveforms. Figure 2A shows simultaneous recordings of MAP and TOI in an infant exhibiting strong correlation in the time domain and a high Coh value of 0.6 in the ULF range of the frequency domain. By contrast, the recording in Fig 2B illustrates little evident fluctuation of TOI in association with spontaneous fluctuations in MAP; the Coh value was 0.3 in the ULF range in this infant. Figure 3 illustrates the variable G, indicating the magnitude of TOI changes in correlation with MAP fluctuations. The infant in Fig 3A demonstrated a much higher magnitude in TOI changes in correlation with fluctuations in MAP, whereas the infant in Fig 3B had low magnitude of TOI changes; and the G values for the 2 infants were 0.8% and 0.4%/mmHg, respectively, in the ULF range. Notably, these 2 infants both showed concordant changes in MAP and TOI in the time domain and high Coh value both at 0.6 in the ULF range.
Clinical Variables and Cerebral Hemodynamics
Infants who had ULF Coh value of ≥0.5 (n = 9) are contrasted with infants with ULF Coh values of <0.5 (n = 15) in Table 1. With ULF Coh values of ≥0.5, infants were of lower gestational age, birth weight, and MAP and had higher CRIB scores (P < .05). Not withstanding the dissociation of Coh and G in individuals as illustrated in Fig 3B, G values in the ULF range were significantly greater (P < .05) on average in infants with ULF Coh values of ≥0.5. No differences were found in the postnatal age, level of arterial carbon dioxide tension, averaged TOI, the proportion of infants requiring inotropic medication, or the incidence of moderate-to-severe IVH (grade II–IV IVH; Table 1).
Univariate analysis revealed that ULF Coh was positively correlated with CRIB score (r = 0.57; P < .01), gestational age (r = 0.41; P < .05), birth weight (r = 0.41; P < .05), and ULF G values (r = 0.45; P < .05). Multiple linear regression of ULF Coh with the significant variables in Table 1 (gestational age, birth weight, MAP, and CRIB scores) showed that, of the possible subsets, CRIB score alone was the best predictor of ULF Coh values (P = .004).
All 6 of the infants who died had ULF Coh values of ≥0.5. The ULF Coh value (mean ± SD) of infants who died during the neonatal period was 0.59 ± 0.10, significantly higher than the ULF Coh values of infants who survived (0.41 ± 0.09; P < .001). The ULF G of infants who died also tended to be higher than that of infants who survived, although it did not reach statistical significance (0.78% ± 0.29 vs 0.53% ± 0.27%/mmHg; P = .07). ULF Coh values of ≥0.5 predicted mortality with positive predictive value of 67% (95% confidence interval [CI]: 30%–92%) and negative predictive value of 100% (95% CI: 78%–100%). Best subset multiple logistic regression of mortality with predictive variables (gestational age, birth weight, MAP, CRIB score, ULF Coh, and ULF G) showed that ULF Coh alone was the best predictor of mortality (P = .002).
The ULF Coh value (mean ± SD) of infants receiving inotropic medications was 0.54 ± 0.1, higher than the value in infants who were not (0.42 ± 0.1; P = .013). The ULF G of infants receiving inotropic medications was also higher than the value in infants who were not on inotropic support (0.78% ± 0.32% vs 0.49% ± 0.23%/mmHg; P = .02). At the time of study, there was no difference in the MAP between infants who were receiving inotropic support and those who were not (31.7 ± 5.2 vs 34.5 ± 4.9 mmHg; P = .2).
Seven infants had grade II–IV IVH. Of these, 2 infants had developed IVH before the NIRS study; 1 had a ULF Coh value of ≥0.5, and the other had a ULF Coh value of <0.5. Five infants had normal cranial ultrasonography results on the day of the NIRS recording, then subsequently developed grade II-IV IVH; 2 of these had ULF Coh values of ≥0.5, and the other 3 had ULF Coh values of <0.5. Notably, all except 1 of these infants developed grade II IVH. In the single infant who later developed grade IV IVH, the previous ULF Coh value was ≥0.5
This is the first study to use the relationship between MAP and TOI in extremely low birth weight infants for the assessment of cerebral autoregulation and its relationship to clinical characteristics. We found high coherence between MAP and TOI in the ULF range, indicating impaired cerebral autoregulation, in sick preterm infants of whom two thirds died in the neonatal period. This subgroup with high ULF Coh values was characterized by lower gestational age, lower birth weight, lower MAP, use of inotropic agents, higher CRIB scores, and higher mortality. In addition, these infants also exhibited high gain values for the MAP-TOI transfer function. These findings demonstrate that, in this subgroup of infants, cerebral perfusion and oxygenation were directly dependent on arterial blood pressure, and, hence, they are consistent with the presence of a “pressure-passive” cerebral circulation.
Coherence of MAP with intravascular oxygenation (HbD) measured by NIRS has been used previously as an index for autoregulation.2,6,21 In animal studies, HbD has been shown to correlate with CBF in induced hypotension22 or hydrocephalus.23 However, HbD may have theoretical limitations when used as a marker for CBF, because it requires the assumption of a cerebral blood volume that is either constant or changed in the same direction as CBF. The assumption may not hold in conditions such as mild and moderate hypotension when autoregulation results in cerebral vasodilatation.24 The resultant increase in cerebral blood volume in response to hypotension would lead to an increase in HbD, although CBF may have remained unchanged or even reduced. By contrast, TOI is not affected by changes in cerebral blood volume, because it represents the ratio of oxygenated to total oxyhemoglobin in the cerebral vasculature. However, it would be useful to confirm the relationship between TOI and CBF in animal models.
Our study confirms and extends the results obtained in previous studies of infants using NIRS and coherence analysis.2,6,21 The novel measures of dynamic autoregulation (ULF Coh and ULF G) that we obtained were strongly related to clinical features and to subsequent mortality. Our results are similar to the findings of a recent study that demonstrated association of cerebral pressure passivity with low gestational age, low birth weight, and systemic hypotension.6 Of the clinical variables examined, CRIB score was most predictive of high ULF Coh values. In turn, ULF Coh value was the most highly predictive of mortality, exceeding all of the clinical variables examined. Importantly, as a predictor of mortality, ULF Coh value was superior to CRIB score, an index that has been shown previously to be highly predictive of mortality.16,25 However, the 95% CI for the positive predictive values of ULF Coh for mortality was wide, probably because of the small population in this study. On the other hand, the narrower 95% CI for the negative predictive values suggested that infants with low ULF Coh values have a high chance of survival. Additional studies using larger populations with follow-up neuroimaging and developmental assessment will be required to assess the clinical value of this approach.
Surprisingly, high ULF Coh was not clearly associated with the presence of cerebral lesions assessed by daily cranial ultrasonography, similar to the finding by Soul et al.6 Of the 5 infants who had NIRS recording before development of grade II–IV IVH, only 2 had ULF Coh values of ≥0.5. Our results do not support the concept that impaired autoregulation is predictive of the subsequent development of IVH. However, the number of infants with such lesions in our study was small, and only 1 infant developed severe IVH. Hence, our study was not powered to test this hypothesis. In a previous study, impaired autoregulation, identified using coherence between MAP and HbD measured by NIRS, was associated with severe cerebrovascular lesions (grade III–IV IVH and periventricular leukomalacia) in preterm infants.2 However, in that study, cranial ultrasonography was not performed before the third day of life, and, hence, it was unclear whether the impaired autoregulation preceded the occurrence of cerebral lesions. Resolution of these issues would require a study of a larger group of preterm infants with serial NIRS studies coupled with daily cranial ultrasonography to identify cerebrovascular lesions.
The physiologic mechanism of autoregulation is not a simple “all-or-none” phenomenon but rather a dynamic, frequency-sensitive system.5,26 At higher frequency of blood pressure oscillation, autoregulation is less effective, allowing concordant variation of CBF with blood pressure. Thus, the autoregulation mechanism behaves as a high-pass filter.17,27,28 Because of the frequency-dependent characteristics of autoregulation, dynamic measurements of CBF would more readily detect CBF changes occurring in response to rapid and transient MAP variations.5,26 On the other hand, “static” single estimates of CBF made using 133Xenon clearance29 or NIRS7,8 may fail to detect impaired autoregulation. This may explain the previous finding that autoregulation was present in preterm infants.7,8 Autoregulation may become critical in the ultralow-frequency range of MAP variations, because the amplitude of spontaneous MAP variations increases as their frequency decreases.30 We, therefore, used uninterrupted epochs of relatively long duration to assess ULF components with increased reliability. As demonstrated in this study and others,2,6 it is likely that loss of ULF autoregulation is of the greatest clinical and pathophysiological importance.
Previous studies of dynamic autoregulation based on Doppler measurements of CBF velocity (CBFV) and MAP-CBFV coherence analysis in healthy adults demonstrated increasing coherence with higher frequencies, in keeping with the inability of vascular smooth muscle to respond at higher frequencies.17,26–28 This is consistent with the notion that autoregulation functions as a high-pass filter. In contrast, coherence analyses in this study and other's2 that are based on NIRS techniques have not demonstrated the “high-pass filter” characteristics. The contrasting findings may be explained by the difference in the response time of the 2 physiologic systems examined. Doppler measurement detects the rapid changes of CBFV, which occur within 1 heartbeat of a change in MAP.31 On the other hand, whereas changes in cerebral oxygenation measured by NIRS will reflect fluctuations in CBF, they will occur at a slower rate than CBF because of the time delay in achieving equilibrium of oxygenation in the venous blood pool. Regardless of which of the 2 systems are examined, increased coherence in the very low frequency range has been identified as clinically significant.2,6,32 In this respect, NIRS is better suited to the prolonged measurement needed to assess autoregulation in the lower frequency range, whereas the cerebral Doppler ultrasound technique is not so easily applied because of the need to maintain a constant insonation angle to the vessel of interest.
According to the Fick principle, TOI will correlate with CBF provided that cerebral metabolic rate for oxygen and cerebral arterial oxygen saturation remain constant. In our study, only changes in cerebral metabolic rate for oxygen and arterial oxygen saturation that occur with frequencies identical to MAP variations would cause a major error in measurements of Coh and G. We minimized any potential interference from changes in metabolic rate and arterial oxygenation by selecting periods when the infants were relatively stable with minimal fluctuations in arterial saturation.
It is important to recognize the potential limitations of the use of coherence and transfer function analyses to investigate moment-to-moment autoregulatory mechanisms. Our analyses were based on the assumption that the MAP-TOI relationship could be modeled as a linear system. This approach has the potential to produce misleading results in a system with nonlinear properties. For example, low coherence may result when the output (TOI) is related to the input (MAP) in a nonlinear manner.27 It is arguable that a nonlinear model may be more appropriate for assessment of autoregulation because of the nonlinear responses of key physiologic parameters, such as cerebrovascular resistance and arteriolar diameter, to changes in MAP.33 However, the assumption of linearity can be justified in the case of spontaneous changes in MAP, because the changes are relatively small.34 Linear spectral analysis techniques also require the signals to be stationary, that is, it is assumed that their dynamic properties do not change with time. Because biological signals such as MAP and TOI have spontaneous fluctuations of different amplitude and periodicity, they are not perfectly stationary signals. It is possible that nonstationary elements may have contributed to the variation of Coh and G that we observed in different segments of data collected in the same infant. Variation in Coh and G in the same infant may also have arisen from periodic variation of the autoregulatory capacity.6 It is possible that certain periods demonstrating impaired autoregulation were not captured or excluded because of movement artifacts or unstable oxygen saturations. We, therefore, selected the period with maximum Coh value in the ULF range in each infant for analysis, because periods of ULF variations likely represent the greatest clinical risk.2 Despite these limitations, the variation in Coh that we observed was more likely to have a biological basis rather than a methodologic one, because the maximum ULF Coh value was strongly concordant with clinical indices and with mortality.
This study provides new information about cerebral autoregulation in preterm infants undergoing intensive care. We have demonstrated that impaired autoregulation, signified by high coherence between MAP and TOI, was present in a subgroup of clinically sick infants and was strongly associated with subsequent mortality. Additional studies in a larger population of infants are needed to confirm the findings. Spatially resolved NIRS has the potential to provide continuous assessment of cerebral autoregulation at the bedside, and, hence, this approach may guide therapeutic interventions in critically ill infants.
Dr Wong was supported by the Victoria Fellowship (state government of Victoria), the Eric Burnard Fellowship (Royal Australasian College of Physicians), and an Australian postgraduate award. Project support was provided by the National Health and Medical Research Council (Australia).
We thank the parents of all of the infants in our study for their participation and trust, and the neonatal nursing and medical staff for their assistance.
- Accepted August 2, 2007.
- Address correspondence to Adrian M. Walker, PhD, Ritchie Centre for Baby Health Research, Monash Medical Centre, 246 Clayton Rd, Clayton, Victoria 3168, Australia. E-mail:
The authors have indicated they have no financial relationships relevant to this article to disclose.
- ↵Lassen NA. Cerebral blood flow and oxygen consumption in man. Physiol Rev.1959;39 (2):183– 238
- ↵Tsuji M, Saul JP, du Plessis A, et al. Cerebral intravascular oxygenation correlates with mean arterial pressure in critically ill premature infants. Pediatrics.2000;106 (4):625– 632
- ↵Tyszczuk L, Meek J, Elwell C, Wyatt JS. Cerebral blood flow is independent of mean arterial blood pressure in preterm infants undergoing intensive care. Pediatrics.1998;102 (2 pt 1):337– 341
- ↵Munro MJ, Walker AM, Barfield CP. Hypotensive extremely low birth weight infants have reduced cerebral blood flow [published correction appears in Pediatrics. 2005;115(6):1794–1795]. Pediatrics.2004;114 (6):1591– 1596
- ↵Wyatt JS, Cope M, Delpy DT, et al. Quantitation of cerebral blood volume in human infants by near-infrared spectroscopy. J Appl Physiol.1990;68 (3):1086– 1091
- ↵Matcher SJ, Kirkpatrick P, Nahid K, Cope M, Delpy D. Absolute quantification methods in tissue near infrared spectroscopy. Proc Soc Photo Opt Instrum Eng.1995;2389 :486– 489
- ↵Suzuki S TS, Ozaki T, Kobayashi Y. A tissue oxygenation monitor using NIR spatially resolved spectroscopy. Proc Soc Photo Opt Instrum Eng.1999;3597 :582– 592
- ↵Stearns SD, David RA. Signal Processing Algorithms. Englewood Cliffs, NJ: Prentice-Hall;1988
- ↵Volpe JJ. Intracranial hemorrhage: germinal matrix-intraventricular hemorrhage of the premature infant. In: Volpe JJ, ed. Neurology of the Newborn. 4th ed. Philadephia, PA: WB Saunders;2001:428– 448
- ↵Sarquis AL, Miyaki M, Cat MN. The use of CRIB score for predicting neonatal mortality risk [in Portuguese]. J Pediatr (Rio J).2002;78 :225–229
- ↵Latka M, Turalska M, Glaubic-Latka M, Kolodziej W, Latka D, West BJ. Phase dynamics in cerebral autoregulation. Am J Physiol Heart Circ Physiol.2005;289 (5):H2272– H2279
- ↵Parati G, Saul JP, Di Rienzo M, Mancia G. Spectral analysis of blood pressure and heart rate variability in evaluating cardiovascular regulation: a critical appraisal. Hypertension.1995;25 (6):1276– 1286
- ↵Zhang R, Zuckerman JH, Iwasaki K, Wilson TE, Crandall CG, Levine BD. Autonomic neural control of dynamic cerebral autoregulation in humans. Circulation.2002;106 (14):1814– 1820
- Copyright © 2008 by the American Academy of Pediatrics