1. Introduction

Shock may result in an acute blood flow reduction, metabolic abnormalities, anaerobic metabolism, cellular and organ dysfunction, and irreversible damage and death if prolonged [1]. The shock resulting from inevitable hemorrhage during surgery and traumatic injury is a fatal complication (up to 50% in mortality), frequently leading to early deaths [2,3]. Estimation of shock severity is crucial to guide clinicians for making treatment plans and for evaluating therapeutic effects [4].

The widely used clinical method to determine shock severity is to measure blood sample oxygen indices. Various oxygen-related indices were proposed as shock indicators, such as oxygen delivery (DO₂), oxygen consumption (VO₂), blood lactate, pulse oxygen saturation (SpO₂), central venous oxygen saturation (ScvO₂), artery oxygen saturation (SaO₂), and partial pressure of oxygen (PO₂). Among these indicators, only SpO₂ can be detected noninvasively by a finger-sensor. However, finger pulse oximetry works ineffectively during shock because of the poor peripheral blood perfusion in the finger [5]. The global indicators of DO₂ and VO₂ may not catch up regional hypoperfusion [6]. Blood lactate is not sensitive to pre-existing medical conditions [7]. PO₂ has been used for monitoring shock in an uncontrolled study with a small number of patients [4]. ScvO₂ [8] and SaO₂ are accepted to be accurate and objective for shock monitoring [9], and ScvO₂ is commonly regarded as a gold standard [8]. However, the invasive and intermittent procedures to obtain these acceptable indicators (i.e., PO₂, ScvO₂, SaO₂) limit the use for continuous shock monitoring. As a result, the clinician may miss the best time window to rescue the patient. Therefore, it’s crucial to explore a noninvasive technique to continuously monitor shock.

Near-infrared spectroscopy (NIRS) permits noninvasive and continuous measurements of tissue blood oxygenation. The NIRS devices are usually portable, permitting bedside monitoring. There have been recently exploratory studies using NIRS for shock monitor [4, 9–15]. Some of them were carried out on animal models [12,13] or healthy human subjects [14]. A few were performed in patients either invasively with optical sensors inserted into subcutaneous and intramuscular sites [4,15] or noninvasively with optical probes placed on the peripheral muscles [9–11]. However, shock usually results in a poor peripheral blood circulation [5], leading to an insensitive measurement.

Shock occurs mostly because of insufficient blood oxygen to the body. Tissue blood oxygen saturation (StO₂) measured at an appropriate region may be a good indicator of shock state [9–11]. In this study, we investigated the correlations between StO₂ and ScvO₂, SaO₂, and PO₂, and explored the use of StO₂ as an indicator for shock.

2. Methods and materials

2.1. Spatially-resolved NIRS

We have developed a SR-NIRS to monitor StO₂ in the region surrounding the internal jugular vein [16]. Two detectors and a 3-wavelengths (735, 805, and 850 nm) integrated LED were integrated into a probe (Fig. 1(a)) [17,18]. The LED was controlled by a micro-controller to be “on” or “off” alternatively. Relatively larger source-detector distances (2.4 cm and 2.7 cm) were used to probe deeper tissue property around the internal jugular vein, which located ~1 cm beneath the skin/probe (Fig. 1(b), 1(c), Fig. 2), and thus yielding more sensitive signals for the comparison of StO₂ and ScVO₂ measurements. The diameter of each detector was ~0.6 cm, making it impossible to arrange S-Ds in a linear pattern. Thus they were configured in a shape of triangle. Figure 2(b) shows Monte Carlo Simulation [19–21] of NIRS signal sensitivity profile with our probe design, suggesting that the sampling volume of our probe covered the tissue surrounding the central internal jugular vein. The size and depth of the veins in shock patients were determined by ultrasonographs. The thicknesses of skin and fat over the measured population were 0.2 ± 0.04 cm (mean ± standard deviation) measured by a Vernier caliper.

 

Fig. 1 Portable spatially-resolved near-infrared spectroscopy (SR-NIRS) device for bedside shock monitoring. (a) SR-NIRS probe placement; (b) SR-NIRS shock monitor device; (c) Bedside shock monitoring.

Download Full Size | PDF

 

Fig. 2 (a) Ultrasonograph of internal jugular central vein and (b) Monte Carlo simulation of photon paths within the tissues surrounding the veins.

Download Full Size | PDF

The calculation of local StO₂ was accomplished by converting from the detected light intensity changes based on the modified Beer-Lambert law [22,23]. The following formula describes the relationship between the measured optical density (OD), tissue optical properties, and the source-detector spacing (ρ):

Here, ρ₀ is the average of the chosen maximum and minimum source-detector separations in the probe, R_app is the spatial dependence of the approximated diffuse reflectance, and${\mu }_{eff}=\sqrt{3{\mu }_a({\mu }_a+{\mu }_s(1-g))}$. ${\mu }_a$ is the absorption coefficient, ${\mu }_s$is the scattering coefficient, and g is the anisotropic factor. cal means the calibration control. We then plotted OD changes as a function of ρ and the slope_λi of the plotted curve was calculated. The correlation between slope_λi and μ_aλi was calculated as follows:

The relationship between ${\mu }_a$ and the concentrations of deoxy-hemoglobin (Hb) and oxy-hemoglobin (HbO₂) can be described as follows:

Here, ε_Hbλi and ${\epsilon }_{Hb{O}_2{\lambda }_{\text{i}}}$represents the extinction coefficient of Hb and HbO₂ respectively. The mathematical definition of StO₂ is given by:

By substituting Eqs. (3) and (4) into Eq. (2), StO₂ was obtained as follows:

2.2 Subjects and experimental protocol

Twenty-five patients with shock were recruited from Xinhua Hospital in Shanghai of China with their written informed consents approved by the ethics committee (Approval No. XHEC-D-2014-005). Patients were 67.3 ± 16.6 years old. StO₂ measurements were performed using our SR-NIRS device in conjunction with other conventional shock monitoring procedures. The StO₂ measurements were carried out either at the end of shock treatment (8 males and 5 females) or one day after (7 males and 5 females). The internal jugular central venous catheterization could not be performed on the former group of patients since the measurement site was still occupied to perform tracheal intubation to obtain respiration-oxygen delivery at that point.

2.3 Data collection and analysis

We used the ultrasound to locate the internal jugular central vein and to aid the placement of SR-NIRS probe (see Fig. 2(a), Fig. 1(b), 1(c)). The probe position was fine-tuned and fixed until the signal became stable. After collecting SR-NIRS data for 5 minutes, blood sampling through the central venous catheter was performed to obtain the indices of ScvO₂, SaO₂, and PO₂. We also collected SpO₂ data from finger sensor.

All data analysis was performed with MATLAB R2010b, MathWorks, USA. Statistical significance was declared for p values < 0.05. We performed correlation analysis of StO₂ with ScvO₂, PO₂, and SaO₂ respectively and reported the Pearson’ correlation coefficient r, which ranged from 0 to 1 to estimate linear correlation extent. To quantify the relationship of StO₂ with combinations of the acceptable blood oxygen indices, we fit multiple linear regression models. We also test the consistency between StO₂ and ScvO₂ and reported Lin’s concordance coefficient [24], which combines a measure of precision and accuracy. A Bland-Altman plot was carried out to visualize the extent of agreement between the indicators.

3. Results

SpO₂ measurements were not reliable on our patients with shock; almost all showed low invalid values or no readout except one patient had a value 95% which was in the valid range.

The comparisons between the StO₂ measured by SR-NIRS and other shock-monitoring clinical measures are shown in Fig. 3, 4 and Table 1. A significant linear relationship was observed (r > 0.843, p < 0.001) between the StO₂ and ScvO₂ (see Fig. 3(a)), with a slope (95% confidence interval) of 0.96 (0.53 to 1.39) (see Fig. 4(a)). Only three data points fell outside 95% confidence level. The Bland-Altman plot (Fig. 4(b)) for these two measurements also demonstrated an agreement, with a mean difference of 4.16% (−6.87% to 15.20%). The concordance coefficient between StO₂ and ScvO₂ of 0.804 is high. Additionally, StO₂ was found to be correlated with SaO₂ (r = 0.501, p = 0.011; see Fig. 3(b)) and PO₂ (r = 0.465, p = 0.019; see Fig. 3(c)) and there were significant regressions between StO₂ and other acceptable blood oxygen indices (Table 1). The corresponding residual analysis showed only two patients had outlier data, also indicating the robust regression relationships.

 

Fig. 3 The correlations between StO₂ and conventional blood oxygen indices, represented by scatter plot with least-squares fitted lines added.

Download Full Size | PDF

 

Fig. 4 NIRS-measured StO₂ compared to the gold standard ScvO₂ from the blood sample test. (a) The solid line represents the best fit to the data; the dashed line indicates the perfect equivalent; and the dot line denotes the 95% confidence interval for StO₂. (b) Bland-Altman plot of the deference between the StO₂ and ScvO₂ as well as the mean difference (solid horizontal line) between these two parameters. The dash lines indicate the 95% limits of an agreement.

Download Full Size | PDF

Table 1. Multiple linear regression analysis (MLR) of StO₂ with conventional blood oxygen indices

View Table

There were no significant differences in StO₂, SaO₂ and PO₂ between the measures at different days (i.e., the day at treatment versus one day after treatment) (p > 0.16), hence it is not unreasonable to speculate that ScvO₂ might show obscure difference as well. It may contribute to the effective resuscitation on the shocks.

4. Discussion and conclusion

Continuous monitoring of shock is crucial for clinicians to perform treatment timely and evaluate therapeutic effect so that irreversible injuries can be avoided. However, currently available methodologies for shock monitoring rely on frequent blood sampling with internal jugular central venous catheterization, which is invasive and discontinuous. In this study, we demonstrated that the StO₂ level measured at the tissues surrounding the internal jugular vein by noninvasive SR-NIRS may be a shock indicator. The SR-NIRS used in this study enabled absolute measurement of StO₂. Data from 25 patients with shock demonstrate significant correlations between StO₂ and the gold standard ScvO₂ as well as other acceptable blood oxygen indices. These results support the use of SR-NIRS as a noninvasive, continuous tool for bedside monitoring of shock.

Our study is the first exploration of the StO₂ measured by NIRS at the region surrounding the internal jugular central vein as a new shock indicator. Previous studies using NIRS to monitor StO₂ in shocks were performed in the peripheral area including finger, deltoid muscle, subcutaneous tissue, and thenar eminence [11,13]. However, shock usually results in poor peripheral blood circulation, making peripheral measurements unreliable, as seen in this study with SpO₂ measurements. By contrast, we chose the tissue surrounding internal jugular central veins because the gold standard ScvO₂ sampling is usually taken from this vein [11] and this site roughly belongs to splanchnic circulation which is crucial for shock monitoring. We may also explore other potential sites such as patient’s head in the future as cerebral circulation is apparently critical for shock monitoring.

The StO₂ measured by NIRS showed strong correlation and high concordance with the gold standard ScvO₂, which might be because a portion of the signal ScvO₂ goes into the StO₂ (see Fig. 2(b)) measured in the tissue surrounding the internal jugular central vein. Meanwhile, the StO₂ showed weak correlations with other indicators (i.e., SaO₂ and PO₂), indicating that the StO₂ measurement performed closer to the vein for the ScvO₂ test is supper than other places for monitoring shock states. However, the thresholds to separate healthy subjects and shock patients using ScvO₂ and StO₂ have not been established yet, and further investigations including healthy subjects are needed.

In summary, we have demonstrated that NIRS measurements of StO₂ in the internal jugular central venous region provide valuable information for shock monitoring. The NIRS-measured StO₂ has been verified by comparing to the conventional blood oxygen indices, especially ScvO₂. SR-NIRS device can be used at the bedside for continuous monitoring of shock, and thus has the potential to become a clinical tool for shock management.