Main

Failure of oxygen delivery leads to an increase in extraction of oxygen from the blood, and this results in a fall in Svo2. Measurement of mixed Svo2 using a pulmonary artery catheter has become an established form of monitoring in adult intensive care(1). However, this technique is not without its hazards(2) and is technically not feasible in neonatal intensive care. An important response to a decrease in oxygen delivery is redistribution of blood flow from the peripheries to the brain and myocardium. In the absence of any factor affecting local blood flow, measurement of peripheral Svo2 may therefore provide an index of the adequacy of global oxygenation.

Changes in peripheral oxygen tension have been shown to precede other indicators of shock in animal models(3) and in critically ill adults(4). Peripheral Svo2 may be measured by co-oximetry of venous blood. However, frequent measurements are impractical. We have developed a noninvasive, easily repeatable method of peripheral Svo2 measurement using NIRS which may prove to be clinically useful.

NIRS can be used to detect changes in the concentrations of oxygenated and deoxygenated hemoglobin (Hbo2 and Hb) and myoglobin (Mbo2 and Mb) in tissue(5, 6). The absorption properties of hemoglobin and myoglobin in the oxygenated and deoxygenated forms are identical for the wavelength used(7). The chromophores measured can therefore be considered as Hbo2/Mbo2 and Hb/Mb.

NIRS does not measure absolute concentrations of the chromophores, but it does measure changes in their concentrations, and the method described in this report makes use of that fact. Occlusion of venous drainage for a few seconds causes a rise in the measured concentration of both Hbo2/Mbo2 and Hb/Mb. It is reasonable to assume that the tissue myoglobin content will remain constant, and the oxygenation state of myoglobin will not be affected by a short venous occlusion lasting just 10 s. Changes from the baseline during venous occlusion should therefore be due solely to changes in Hbo2 and Hb. If all of the increase in the hemoglobin content of the tissues is due to an increase in the venous blood volume, then the Svo2 may be calculated from: Equation where [Hbo2] = change in concentration of Hbo2 and [HbT] = change in the total concentration of hemoglobin (Hbo2 + Hb).

The aim of this study was to test the hypothesis that NIRS with intermittent venous occlusion can be used to measure peripheral Svo2 in neonates.

METHODS

Sixteen infants undergoing neonatal intensive care were studied. NIRS(Hamamatsu NIRO 500, Hamamatsu Photonics K.K., Hamamatsu, Japan) was performed over the forearm muscles with the optodes positioned on either side of the forearm. A 0.5-s sample time was used. Venous occlusion was caused by inflating a sphygmomanometer cuff around the upper arm to a pressure which resulted in a clear rise in the Hbo2 and Hb signals. The cuff was released after 15 to 30 s. The OD changes for each wavelength were stored in a computer file for later analysis. Concentration changes of Hbo2 and Hb were calculated using the appropriate absorption coefficients. As no path length factor was used in the calculation, the changes were expressed in units of mM·cm. The concentration changes were plotted against time for each occlusion. The data were visually inspected, and an occlusion was determined as satisfactory for further analysis if there was a steady baseline before the occlusion, a rise in both Hbo2 and Hb during the occlusion, and a return to the same baseline after release of the occlusion. A typical occlusion is illustrated in Figure 1.

Figure 1
figure 1

NIRS data from a single measurement of peripheral Svo2. ○, Hbo2; •, Hb.

Full size image

Peripheral Svo2 was calculated in three different ways as previously described(8) and is illustrated in Figure 1. The first method calculated peripheral Svo2 from the area under the curves. The second calculated peripheral Svo2 from the initial concentration changes during the first 5 s of venous occlusion. The third method calculated peripheral Svo2 at each point during the peak concentration change. If the peak change lasted for more than 5 s, data from only the first 5 s were analyzed.

Immediately after several satisfactory occlusions, blood was taken from a forearm vein, and peripheral Svo2 was measured by co-oximetry (IL 482, Instrumentation Laboratory, Lexington, KY). HbF was also measured in each infant, and the co-oximetry data were adjusted accordingly(9). Two infants had blood drawn from a 24-gauge cannula sited in the antecubital vein, the others had blood taken by venepuncture.

Correlation between the two measurements was assessed, and for each subject the peripheral Svo2 measured by co-oximetry was compared with the mean of the NIRS values, using the method of Bland and Altman(10). All studies were approved by the Paediatric Research Ethics Committee of the Royal Liverpool Childrens Hospital, and informed parental consent was obtained.

RESULTS

There were seven male and nine female neonates with a median age of 8.5 d(range 1 to 29 d). Their median gestational age was 30 wk (range 25 to 39 wk). The median peripheral Svo2 by co-oximetry was 75.9% (range 60.8 to 90.7%). Results for individual subjects are presented inTable 1.

Table 1 Svo2 by each method for each subject
Full size table

Correlation and agreement between the co-oximetry and NIRS measurements is presented in Table 2 and Figure 2. The mean difference was slightly less when the initial NIRS change was used to calculate peripheral Svo2, but the limits of agreement were very similar. Peripheral Svo2 by co-oximetry was generally slightly higher than that measured by NIRS, and this difference became more pronounced at higher levels of venous saturation. This relationship was significant(r = 0.528, p < 0.05, n = 16).

Table 2 Correlation and agreement between each method of NIRS analysis and co-oximetry
Full size table
Figure 2
figure 2

Agreement between co-oximetry and NIRS (initial change method) measurement of peripheral Svo2. - - - - - = mean difference 6%;····· = limits of agreement -5.1% to 17.1%.

Full size image

Reproducibility of the NIRS measurement was assessed by calculating the 95% confidence limits of the mean for those 12 subjects with five or more measurements. The confidence interval was smaller than ±6% in 11 of these 12 subjects, and in the other subject (no. 7) the confidence interval was ±7.3%. The mean (±SD) coefficient of variation for the NIRS measurement in these 12 subjects was 5.5% (±2.6%). The co-oximeter value was included within the confidence interval of the mean NIRS measurement in only four subjects.

DISCUSSION

We have described a repeatable noninvasive method for the measurement of peripheral Svo2 in neonates. This method measures Svo2 of the whole forearm, deep and superficial tissues. Co-oximetry of blood from a superficial vein measures superficial Svo2 only. As we do not know how well superficial Svo2 reflects Svo2 of the whole forearm, the measurements made in this study are best considered as two related variables, rather than two measurements of the same variable. A comparison between the two methods is therefore best made from the correlation between the two. Because the difference between superficial and deep Svo2 in neonates is not known, and the values may in fact be very close, we have also included an assessment of the agreement between the two methods by the method of Bland and Altman(10). There is a strong correlation between Svo2 by NIRS and Svo2 by co-oximetry of superficial venous blood. We have also demonstrated reasonable agreement between the two methods. This suggests that we have successfully measured of Svo2 by NIRS.

The degree of agreement was similar for all three methods of analysis of the NIRS data. In a previous study(8), in which we measured cerebral venous saturation with a similar NIRS technique and compared the results with those obtained by co-oximetry of blood from the jugular bulb, we noted that venous saturation measurements made using the NIRS data from immediately after the start of the venous occlusion provided better agreement with the co-oximetry measurement than measurements made from the area under the curves or from the peak concentration changes. Our interpretation of this was that the initial changes in the NIRS Svo2 measurements reflected most closely the preocclusion situation. We have also seen that the Hbo2:HbT ratio fell during prolonged venous occlusion in the adult forearm (our unpublished observations). We consider that these observations demonstrate that prolonged venous occlusion may impair tissue oxygenation. However, this effect did not appear to be important during a short venous occlusion in the forearm of neonates.

We have considered the possible explanations for the differences between the two measurements. The co-oximetry value was generally higher than the NIRS value, and as the co-oximetry value was usually outside the confidence interval for the NIRS value, this difference did not appear to be accounted for by poor reproducibility of the NIRS measurement. There are several methodologic reasons why we would not expect complete agreement between the two techniques. First, as discussed above, the NIRS method measured Svo2 from all forearm tissues whereas the co-oximetry method measured only superficial Svo2. The difference between superficial and deep Svo2 is not known, but there are two venous systems in the forearm. The superficial veins of the forearm carry blood from the skin and s.c. tissues. The deeper tissues are drained by the venae comites. It seems likely that the superficial tissues would have a lower rate of oxygen consumption than the deeper tissues, and the superficial veins would therefore have a higher oxyhemoglobin saturation than the deeper veins. This hypothesis was supported by our own observations when we used this technique with different optode spacings to measure superficial and deep venous saturation in the forearms of adults and found superficial venous saturation to be higher than deep venous saturation (our unpublished observations). This could explain why co-oximetry of blood from the superficial veins generally gave a higher measurement than did the NIRS measurement. The difference between NIRS and co-oximetry measurements appeared to be greater in those subjects with high co-oximetry values, and this might have been due to these individuals having less mixing between the deep and superficial venous systems.

Second, measurements made by the two techniques were not made at exactly the same time. Third, the process of venepuncture sometimes disturbed the babies and may itself have affected peripheral Svo2.

One of the prime objectives of successful intensive care is the maintenance of an adequate supply of oxygen to meet tissue needs. Peripheral Svo2 monitoring may provide useful information about changes in the relationship between oxygen delivery and oxygen consumption. There may, then, be a role for a method which assesses peripheral Svo2 noninvasively and which can be repeated every few minutes as a monitor of tissue oxygenation.

This study has shown the feasibility of frequent noninvasive measurements of peripheral Svo2. The technique may turn out to be to be a useful guide to tissue oxygenation in sick neonates, but further clinical studies are required before it can be considered as a useful addition to the clinical monitoring armamentarium.