Main

The indications for blood transfusions in the neonatal period are not particularly well defined. Some infants seem to tolerate a low Hb with no apparent clinical problems, but others manifest a variety of symptoms at a similar Hb. These symptoms may be related to poor tissue oxygenation and have therefore been attributed to anemia. Strategies for blood transfusions for preterm babies have been developed to maintain a higher Hb in sick babies, whereas older or less intensively treated babies are transfused only when the Hb falls to low levels or symptoms develop(1). These recommendations aim to maintain tissue oxygenation at a suitable level in all babies but are based on clinical experience and consensus opinion rather than scientific evidence.

Currently total Hb or hematocrit are used as markers of when to transfuse, but an indicator that reflects inadequate oxygen delivery to meet oxygen demands is likely to be better(2,3). Fractional oxygen extraction is a measure of the balance between oxygen delivery and oxygen consumption: Equation 1 Oxygen delivery can be calculated from the blood flow and the arterial oxygen content: Equation 2 Oxygen consumption can be calculated from the blood flow, arterial oxygen content, and the venous oxygen content using the Fick principle: Equation 3 Arterial and venous oxygen content may in turn be calculated from: Equation 4 where 1.39 is the stoichiometric oxygen capacity of Hb (mL/g) and 0.00003 is the Bunsen solubility coefficient for oxygen (mL O2/mL blood). Because the amount of dissolved oxygen is insignificant in comparison with that carried by Hb, it can be ignored. Equation 4 can then be simplified to Equation 5 Combining Equations 2 and 3 with Equation 5, Equation 6, Equation 7 and substituting into Equation 1, Equation 8 Measurements of fractional oxygen extraction for the whole body produce a range of approximately 0.15 to 0.33; that is, the body consumes 15-33% of oxygen transported(2,4).

Measurements of arterial saturation are possible using pulse oximetry. Measurements of peripheral venous saturation and blood flow in the forearm of neonates can be made using near infrared spectroscopy with venous occlusion(5,6). These measurements are simple, involving minimal disturbance of the subject, and repeated measurements are possible(5,6).

The measurements of forearm blood flow, venous saturation, arterial saturation, and Hb allow the calculation of fractional oxygen extraction, oxygen delivery, and oxygen consumption in peripheral tissues. Provided oxygen consumption remains constant, oxygen extraction will increase when oxygen delivery decreases. Fractional oxygen extraction may then identify changes in oxygen availability to the tissues. We hypothesized that fractional oxygen extraction could be used as an indicator of the need to increase oxygen delivery by blood transfusion.

The objective of this study was to test the hypothesis that fractional oxygen extraction measured by this technique was high in the presence of anemia and decreased when anemia was corrected. We aimed to establish a normal range of forearm fractional oxygen extraction, oxygen delivery, and oxygen consumption measured by pulse oximetry and near infrared spectroscopy, to establish the relationship between these and other measures, which are known to be important in determining oxygen availability in preterm babies, and to study the changes that occur in tissue oxygenation after blood transfusions.

METHOD

Near infrared spectroscopy. Venous saturation measurements. Measurements were made using the Hamamatsu NIRO 500(Hammatsu UK Ltd) and a pulse oximeter (Datex, Finland) in beat to beat mode. Venous saturation was measured using a technique that has been described elsewhere in detail(5). The monitoring optodes were positioned on the upper forearm and the interoptode distance measured using calipers (usually 1.5-2.5 cm). The optodes were held in place using a small Velcro band (Ohmeda). A brief venous occlusion was made with a blood pressure cuff around the upper arm manually inflated to 30 mm Hg(5). Measurements made were analyzed off-line using a spreadsheet package (Lotus 1-2-3) using the concentration changes during the first 5 s of the occlusion. Venous saturation was measured five times during a 2-min period for each fractional oxygen extraction measurement and the mean value was used. Measurements were made when babies were quiet and at rest. The pulse oximeter optodes were positioned on the same limb or another limb, once it had been determined that there was no difference in arterial saturations in different limbs. Arterial saturation was recorded every 0.5 s and the mean value at the time of each measurement of Venous saturation was used to calculate fractional oxygen extraction. Arterial saturations were maintained in the range 88-94% for babies less than 32 wk and greater than 92% in babies more than 32 wk(7). The differential path length factor used was 3.59, which is that determined for the adult forearm, because path length measurements for the neonatal forearm are not available(8). Measurements took approximately 10 min to complete.

Blood flow measurements. Measurements of forearm blood flow were also made from the near infrared spectroscopy data using a method adapted from that described and validated in adult patients by de Blasi et al.(6). In a steady state, arterial and venous flow are equal. During the initial part of a venous occlusion, Hb accumulates in the tissues due to cessation of venous flow, and the Hb flow is equal to the rate of tissue Hb accumulation during the initial part of the occlusion. Near infrared spectroscopy measures changes in total tissue Hb concentration(ΔHbT), and in this study we measured Hb flow by the slope of a line through the ΔHbT values during the first two seconds of occlusion using a least squares method. Blood flow (mL/100 mL/min) was calculated by dividing Hb flow (µmol/100 mL/min) by venous Hb concentration (µmol/mL).

Patient population. Babies were not included in the study if they had received a transfusion in the previous 3 d or were treated with muscle relaxants or inotropes. There groups of patients were studied. Group 1 was a control, nontransfusion group. Babies being nursed in the neonatal unit were of any gestational or postnatal age, who were stable with normal arterial saturation, blood gases, not receiving respiratory or inotropic support, and not considered to need transfusions by the clinicians responsible for their care. Group 2 were the asymptomatic transfusion group. Babies in this group were healthy but their Hb was below the limits set by our unit blood transfusion protocol, which has been used in previous studies(9). These babies did not have symptoms ascribable to anemia. This protocol is summarized below.

Blood transfusion protocol. Babies were transfused 1) at Hb of 14 g/dL if inspired oxygen concentration was >35% or mean airway pressure was >6 cm H2O; at Hb of 12 g/dL if inspired oxygen concentration was <35% or mean airway pressure was <6 cm H2O, or if significant apneas or bradycardias (>9 in 12 h); 3) heart rate >180/min or respiratory rate >80/min for 24 h; and 4) weight gain <10 g/d for 4 d on 100 kcal/kg/d.

Group 3 were the symptomatic transfusion group. Using the same protocol(9), these babies were considered to require a blood transfusion because of a low Hb and symptoms ascribable to their anemia. Although the protocol was used as a guideline for clinicians, some transfusions were given to babies with symptoms not defined in the protocol but which were felt to represent symptomatic anemia. These babies were included in this group.

Measurements of peripheral venous saturation, arterial saturation, blood flow, Hb, and HbF were made in each baby during the 8 h preceding the transfusion. HbF was measured wherever possible, but some babies did not have measurements of HbF because insufficient blood was taken. The clinicians had no knowledge of the near infrared spectroscopy measurements in any of the groups. In the transfused babies a second measurement of venous saturation, arterial saturation, blood flow, Hb (and HbF where possible) was made 12-24 h after the blood transfusion.

The pretransfusion RCV was calculated using the measurements of HbF by dilution with a known amount of Hb A from the transfusion(10). RCV calculations were accepted only if the HbF fraction was greater than 20% before transfusion(11). Transfusion comprised approximately 20 mL/kg of packed cells over a 4-h period. As with a previous study where RCV was calculated using this technique(11), we assumed that each 20 mL/kg transfusion of packed cells contained approximately 13.3 mL/kg of red cells.

This study was approved by the local research ethics committee, and written parental consent was obtained.

Statistics. Data were analyzed using SPSS for Windows(Version 4). The venous saturation and blood flow for each baby was determined as the mean of five venous occlusions. The relationship between variables was assessed using Pearson's correlation coefficients and multiple linear regression. Differences between groups were assessed using ANOVA and then unpaired t tests or Mann Whitney U tests as appropriate. Changes after transfusions were analyzed using paired t tests and Wilcoxon rank sum test as appropriate for parametric and nonparametric data.

RESULTS

Descriptive data for the three groups are summarized in Table 1. The asymptomatic transfusion group (group 2) had lower gestational ages than did the control group (p < 0.001) and were lighter (p < 0.001). The symptomatic transfusion group (group 3) were also of lower gestational age than the control group (p = 0.046) and were lighter (p = 0.003). There were no differences in the postnatal ages in any of the groups at the time of measurement (p = 0.97).

Table 1 Results in the three study groups
Full size table

Group 1 (control subjects). Fifty-two nonanemic babies were studied. The results are summarized in Table 1. Fractional oxygen extraction ranged between 0.231 and 0.463: the 5th, 10th, 50th, 90th, and 95th centiles were 0.247, 0.279, 0.357, 0.436, and 0.460, respectively.

Group 2 (asymptomatic). Twenty-four babies were studied who had no symptoms but who received blood transfusions according to the protocol. None of these babies was receiving muscle relaxants, and 13 of the 24 were ventilated. The fractional oxygen extraction before transfusion was similar to that of the control group (p = 0.22; Fig. 1). The mean Hb and the median forearm blood flow in this group before transfusion were not significantly different from that of the control group (p = 0.34 and 0.83, respectively) (Table 1). There was no difference in oxygen delivery(p = 0.81) or oxygen consumption (p = 0.50) between this asymptomatic group before transfusion and the control subjects (Table 1). Before transfusion only one of these babies had a fractional oxygen extraction greater than 0.46, the 95th centile for the control group (Fig. 2). Babies who were ventilated(n = 13) had a mean (SD) fractional oxygen extraction of 0.34(0.06) before transfusion. This was similar (p = 0.71) to the mean(SD) fractional oxygen extraction [0.35 (0.06)] of the 63 nonventilated babies in groups 1 and 2.

Figure 1
figure 1

Fractional oxygen extraction results in the three groups studied; bars represent mean values for each group.

Full size image
Figure 2
figure 2

Relationship between HbF and fractional oxygen extraction in all babies studied; n = 66, r= 0.49, p < 0.001.

Full size image

After transfusion, fractional oxygen extraction remained similar to the pretransfusion value (n = 24, p = 0.74; Table 1). There was, however a significant increase in Hb (p < 0.001) and in oxygen delivery (p = 0.03; Table 1). The mean increase (95% confidence intervals) in oxygen delivery was 35.0 (4.3-65.6) µmol·100 mL-1 min-1.

The forearm blood flow after transfusion was not significantly different from the pretransfusion value (Table 1) (p = 0.86) with a mean change (95% confidence intervals) of 1.0 (-2.8-4.9) mL·100 mL-1 min-1. There was also no significant change in oxygen consumption (p = 0.10; Table 1) [mean change in oxygen consumption (95% confidence intervals) 9.2 (-1.2-19.6) µmol·100 mL-1 min-1).

Group 3 (symptomatic). Eighteen babies had symptoms which were attributed to anemia and were therefore transfused. The fractional oxygen extraction before transfusion was higher (p < 0.01) and the mean Hb lower than the control group (p < 0.001)(Table 1). The forearm blood flow (p = 0.81), oxygen consumption (p = 0.56), and oxygen delivery (p = 0.68) were not significantly different from those of the control group (Table 1). Six (33%) of these babies had values of fractional oxygen extraction greater than 0.46, the 95th centile for the control group, and 16 (89%) had values greater than the 50th centile (Fig. 1 and Table 2).

Table 2 Symptomatic transfused babies
Full size table

Six babies had either a raised C-reactive protein and/or positive blood cultures (coagulase-negative staphylococci in all cases). If these six infected babies were excluded, the mean (SD) fractional oxygen extraction of this group (n = 12) was 0.44 (0.07), which remained significantly higher than that of control subjects (p < 0.001). Five of the six infected babies had a fractional oxygen extraction less than 0.46 (Table 2).

The mean (SD) fractional oxygen extraction of the 12 babies who were transfused because of symptoms set out in the protocol (0.43 ± 0.07) was not significantly higher than the fractional oxygen extraction of the six babies who were transfused because of other symptoms (0.42 ± 0.05)(p = 0.55).

The fractional oxygen extraction of symptomatic babies fell significantly after transfusion (p = 0.001; Table 1). The forearm blood flow was unaffected by transfusion (p = 0.44)(Table 1), the mean change (95% confidence intervals) being -1.8 (-8.1 to 4.6) mL·100 mL-1 min-1. The mean change in oxygen delivery after transfusion in symptomatic babies was +23.3µmol·100 mL-1 min-1, but this was not statistically significant (95% confidence intervals -33.3 to 80.0 µmol·100 mL-1 min-1, p = 0.43). Oxygen consumption changed less after transfusion with a mean (95% confidence intervals) change of 3.0 (-17.9 to 23.9) µmol·100 mL-1 min-1, p = 0.86.

Peripheral fractional oxygen extraction in all groups. Forearm fractional oxygen extraction measurements correlated positively with HbF(n = 66, r = 0.49, p = <0.001; Fig. 2) and oxygen consumption (n = 87,r = 0.33, p = 0.001) and negatively with Hb (n = 94, r = -0.21, p = 0.04; Fig. 3). There was no relationship between forearm fractional oxygen extraction and forearm blood flow (r = 0.08,p = 0.45), oxygen delivery (r = 0.04, p = 0.97), postnatal age (r = -0.08, p = 0.44), gestational age (r = 0.08, p = 0.45), or heart rate (r = 0.09, p = 0.59).

Figure 3
figure 3

Relationship between Hb concentration and fractional oxygen extraction in all babies studied; n = 94, r = -0.21, p = 0.04.

Full size image

On multiple regression analysis with fractional oxygen extraction as the dependent variable and Hb, HbF, gestational age, forearm blood flow, and current weight as the independent variables, only forearm blood flow(p = 0.046) and HbF (p < 0.0001) were independently associated with fractional oxygen extraction (multiple r = 0.58, r2 = 0.34, n = 57).

RCV. RCV measurements were made in 19 babies who received transfusions. There was a significant correlation between the pretransfusion RCV and fractional oxygen extraction in these babies (r = -0.48,p = 0.04; Fig. 4), and between pretransfusion RCV and Hb (r = 0.6, p = 0.007).

Figure 4
figure 4

Relationship between red cell mass and fractional oxygen extraction in transfused babies; n = 19,r = -0.48, p = 0.04.

Full size image

Comparison of symptomatic group with asymptomatic group. Before transfusion, the mean forearm fractional oxygen extraction was higher(p < 0.001), the median RCV was lower (p = 0.02), and the Hb was lower (p = 0.001) in the symptomatic group when compared with the asymptomatic group (Table 1).

The asymptomatic group had lower values of HbF than did control subjects(p = 0.02) and the symptomatic group (p = 0.012) before transfusions. The HbF of the symptomatic and the control group were similar(p = 0.54) before transfusion (Table 1).

Forearm blood flow. Forearm blood flow did not correlate with the HbF fraction (n = 62, r = -0.02, p = 0.9), or the RCV (n = 19, r = -0.24, p = 0.3). There was a significant positive correlation between forearm blood flow and the postnatal age (r = 0.4, p < 0.001) and a negative correlation between forearm blood flow and gestational age(r = -0.26, p = 0.02).

Variability and repeatability of venous saturation and blood flow measurements. The variability of these measurements was assessed using methods described by Bland and Altman(12). A measure of the repeatability of measurements was made by making two consecutive measurements (mean of five venous occlusions) on the same individual a few minutes apart with repositioning of the spectroscopy optodes before the second measurement.

The mean difference between the first and second measurements (mean of five occlusions) of venous saturation was -0.5%, and the intrasubject coefficient of repeatability (±2 SD differences) was ±9.7%. The SD of venous saturation for the control population studied was 6.1%.

Measurements of blood flow (mean value of five consecutive occlusions) had a mean difference of -1.3 mL·100 mL-1 min-1 and the coefficient of repeatability was ± 14.1 mL·100 mL-1 min-1. The SD of blood flow for the control population studied was 6.0 mL·100 mL-1 min-1.

DISCUSSION

The chief objective of this study was to investigate the usefulness of fractional oxygen extraction as an indicator of tissue oxygenation. We had hypothesized that fractional oxygen extraction was related to the degree of anemia, usually defined as a Hb or RCV below a particular predetermined level. This definition may not, however, be appropriate. We found only a weak correlation between Hb and fractional oxygen extraction. The explanation for this observation is likely to be that the total Hb concentration is a relatively poor indicator of the adequacy of the provision of oxygen to the tissues [discussed in detail by Holland et al.(13)] and may not accurately reflect oxygen availability to the tissues(2,14). A better definition of anemia is the inadequacy of Hb-determined oxygen availability to meet tissue requirements, and fractional oxygen extraction may be a clinically useful additional measurement because it was highest in those babies who had symptoms attributed to anemia.

The most important determinant of fractional oxygen extraction was HbF. This was not unexpected. HbF has a higher oxygen affinity than does Hb A, and its oxyhemoglobin dissociation curve is shifted to the left, resulting in lower oxygen availability to the tissues(15). A similar concept was explored in studies that used the term "available oxygen" derived from the P50 (partial pressure of oxygen at a Hb saturation of 50%) and Hb(14,16). These showed that the Hb oxygen affinity was related to symptoms of anemia. We also found that babies with a low total Hb were less likely to have symptoms attributable to anemia if the HbF was low, because they had received a greater number of transfusions of adult blood. This led to an increased concentration of Hb A and a shift to the right of the Hb-oxygen dissociation curve, resulting in improved oxygen availability and an associated lower fractional oxygen extraction. HbF has also been shown to correlate positively with cerebral blood flow(17), but in that study cerebral oxygen extraction was not measured. In the present study, there was a significant fall in fractional oxygen extraction after babies with symptomatic anemia were transfused, suggesting an improvement in oxygen availability. This may have occurred either as a result of an increase in RCV and/or a decrease in Hb oxygen affinity because of the increased proportion of Hb A.

In asymptomatic babies the changes in oxygen delivery and oxygen consumption after transfusion were as expected; there was an increase in oxygen delivery, but no change in oxygen consumption. This led to the conclusion that oxygen consumption was not oxygen delivery-dependent before transfusion. This contrasted with the situation in symptomatic babies, where there was no apparent change either in oxygen delivery or oxygen consumption after transfusion. The explanation for this unexpected observation is likely to be methodologic. There was an increase in oxygen delivery by a mean of 23.3 µmol·100 mL-1 min-1 in the symptomatic babies in this study, but this did not reach statistical significance, probably because of the wide range of values for oxygen delivery. The methodologic explanation for the oxygen delivery and oxygen consumption results is that both indices rely on measurements of venous saturation and blood flow. Venous saturation has a small coefficient of repeatability (as does arterial saturation(18)), but blood flow measurements have a higher repeatability coefficient, either because of methodologic error or because of greater biologic variability. Fractional oxygen extraction measurements depend only on venous saturation and arterial saturation measurements. It is therefore reasonable to place more weight on the fractional oxygen extraction measurements than on those of oxygen delivery and oxygen consumption, measured by this method. We have therefore concluded that measurements of fractional oxygen extraction are likely to be clinically useful in this situation, whereas measurements of blood flow using near infrared spectroscopy may be less useful because of their variability, although they may cast some light on pathophysiologic mechanisms.

This study used measurements of fractional oxygen extraction to identify babies in whom oxygen availability was significantly diminished. Measurements using this technique have been shown to correlate well with invasive measurements of mixed venous saturation made by a fiberoptic pulmonary artery catheter(19). The peripheral tissues of the upper forearm was chosen because it is likely that changes in oxygenation in the extremities precede changes in more central organs(20). Changes in peripheral oxygen tension have been shown to precede other indicators of shock in animal models(21) and in critically ill adults(22). There are other advantages. The site of monitoring can be easily standardized, and it is a part of the body that is readily accessible in preterm babies. It is not known whether monitoring other sites in the body would be more useful, and this requires further investigation.

The babies measured in this study were studied during rest between feeds. The advantage of measuring during rest is that there is less biologic variability with a narrower normal range. It is, however, likely that the early clinical effects of anemia might be first disclosed by exercise or stress, such as during feeding, and this too requires further investigation.

There is conflicting evidence from previous studies as to whether symptoms of anemia improve after blood transfusion(23–28). In this study, babies with symptomatic anemia had a mean fractional oxygen extraction which was 28% higher than the fractional oxygen extraction of control subjects. However, two-thirds of babies with "symptoms of anemia" did not have high oxygen extractions. This raises the possibility that those babies were transfused unnecessarily and that their symptoms may have been related to other clinical problems, such as infection. One-third of the babies in the symptomatic group had suspected sepsis on the basis of a raised C-reactive protein or bacteremia. Only one of these babies had an abnormally high fractional oxygen extraction. It is possible either that these babies were not truly clinically anemic and that their symptoms were due to infection, or alternatively, their threshold for developing symptoms of anemia may have been lowered.

Other authors have suggested that the RCV may be a better indicator of clinically significant anemia than Hb(11), and Hb correlated only poorly with the RCV in preterm infants(29). Our data showed a significant correlation between RCV and fractional oxygen extraction, but only a poor correlation between Hb and fractional oxygen extraction. However, there were some extreme values of RCV in our study (<10 mL/kg and >50 mL/kg). The reasons for this are likely to be related to error in the measurement of RCV, the most likely source of which is the use of packed cells, which are not always packed homogeneously and may therefore sometimes contain a variable volume of red cells per mL. The clinical usefulness of RCV is also limited generally by the difficulty of measuring it, requiring either a transfusion of donor blood with measurement of HbF before and after transfusion, or the use of biotin-labeled red cells(10,30).

Transfusing blood carries a significant risk. It may be possible to reduce the use of blood transfusions by introducing an objective assessment of tissue oxygenation such as fractional oxygen extraction rather than relying on clinical symptoms and/or Hb. As well as the risks of viral transmission(31), there is also the theoretical risk of the increased generation of oxygen free radicals from free iron, which may be involved in the mechanism for the development of retinopathy of prematurity, bronchopulmonary dysplasia, and intraventricular hemorrhage(32,33). In addition, transfusion impairs the normal erythropoietic response and results in decreased endogenous production of red cells(34). Reducing the number of blood transfusions that babies receive could therefore have significant beneficial clinical effects as well as reducing cost(35). Measurement of peripheral fractional oxygen extraction using near infrared spectroscopy and partial venous occlusion is noninvasive, does not require blood sampling, and is relatively simple to perform. But the clinical usefulness of this new technique needs to be established by formal comparison with conventional methods for determining the need for blood transfusion by a clinical trial. By developing guidelines for transfusion based on measurements of oxygenation, rather than just the Hb concentration, transfusions may be given more appropriately to preterm babies, with subsequent decreases in transfusions, donor exposure, and potential morbidity.