Comparison of non-invasive peripheral venous saturations with venous blood co-oximetry
The estimation of venous oxygen saturations using photoplethysmography (PPG) may be useful as a noninvasive continuous method of detecting changes in regional oxygen supply and demand (e.g. in the splanchnic circulation). The aim of this research was to compare PPG-derived peripheral venous oxygen saturations directly with venous saturation measured from co-oximetry blood samples, to assess the feasibility of non-invasive local venous oxygen saturation. This paper comprises two similar studies: one in healthy spontaneously-breathing volunteers and one in mechanically ventilated anaesthetised patients. In both studies, PPG-derived estimates of peripheral venous oxygen saturations (SxvO2) were compared with co-oximetry samples (ScovO2) of venous blood from the dorsum of the hand. The results were analysed and correlation between the PPG-derived results and co-oximetry was tested for. In the volunteer subjects,moderate correlation (r = 0.81) was seen between SxvO2 values and co-oximetry derived venous saturations (ScovO2), with a mean (±SD) difference of +5.65 ± 14.3% observed between the two methods. In the anaesthetised patients SxvO2 values were only 3.81% lower than SpO2 and tended to underestimate venous saturation (mean difference = –2.67 ± 5.89%) while correlating weakly with ScovO2 (r = 0.10). The results suggest that significant refinement of the technique is needed to sufficiently improve accuracy to produce clinically meaningful measurement of peripheral venous oxygen saturation. In anaesthetised patients the use of the technique may be severely limited by cutaneous arteriovenous shunting.
KeywordsVenous oxygen saturation Pulse oximetry Photoplethysmography Respiratory-induced intensity variations Non-invasive monitoring
Photoplethysmography (PPG), a fundamental technology behind the widely used pulse oximeter, is an optical technique that detects periodic blood volume changes in the tissue bed . Pulse oximetry algorithms calculate arterial oxygen saturations (SpO2) using the ratio of absorbance of red and infrared light by haemoglobin, exploiting the pulsatile nature of arterial blood to discriminate between arteries and non-pulsatile absorbers in the tissue, such as skin, muscle, bone and pigments .
SpO2 is by far the most common application of PPG, and while undoubtedly useful, it only gives information about the incoming supply of oxygen. Knowledge of the oxygen saturations in the venous, as well as the arterial blood, reveals more about the balance of oxygen supply and demand in the monitored tissues than the arterial saturations alone. Of the factors affecting the arteriovenous difference in oxygen content (ΔO2) across a vascular bed, blood flow through the organ has the greatest influence. Oxygen consumption and haemoglobin concentration also influence ΔO2, as demonstrated by the modified Fick equation , but in the short term, these tend not to change significantly . This concept is also applicable to the body as a whole, and is the physiological basis of using mixed venous saturations (SvO2) as an indication of cardiac output in the assessment of global oxygen delivery. Measurement of SvO2, however, requires the placement of a pulmonary artery catheter, which is not without significant risks, and the use of which has seen a steady decline in recent years [5, 6].
Previous research suggests that suitable signal analysis of PPG signals in the respiratory frequency range can produce estimations of SxvO2 . Although saturations lower than those of arterial blood have been obtained from analysis of oesophageal  and peripheral  PPG signals, these estimations have not to date been validated by comparison with co-oximetry of venous blood, considered widely to be the gold standard [18, 19].
The aim of this research was to compare PPG-derived local venous oxygen saturations directly with venous saturation measured from co-oximetry blood samples. The paper describes two separate clinical studies. In the first study, PPG-derived SxvO2 was recorded from awake volunteers breathing through a flow resistor to simulate mechanical ventilation and compared to peripheral venous co-oximetry sample values (ScovO2) obtained form the dorsal vein of the hand. In the second study, similar measurements were performed in mechanically ventilated anaesthetised patients.
The measurement system used in both studies consists of a custom made finger pulse oximeter, airway pressure sensor and peripheral venous pressure sensor, attached to the study main monitoring unit.
2.1.1 PPG probe
The study pulse oximeter is a standard commercial finger probe shell (GE Datex-Ohmeda®, Helsinki, Finland), with the spring clip removed. The shell was fitted with red (660 nm) and infrared (940 nm) light emitting diodes (LEDs) in the upper part of the probe. A photodetector collects the proportion of light not absorbed by tissue on the lower side of the probe. The output feeds into the study monitoring unit where it is amplified using a transimpedance amplifier. A demultiplexer then separates the red (R) and infrared (IR) signals, which are then further separated into alternating current (AC) components, which represent the pulsatile components of the PPG, and direct current (DC) components, which accounts for non-pulsatile tissues in the finger. This separation is achieved using an active filter with pass band ranging from 0.38 to 28.2 Hz. All four signals (ACR, DCR, ACIR and DCIR) are filtered to remove artifact using an active low pass filter of cut-off frequency 16.7 Hz to remove coupled mains and other interference. All filters used in the circuit are based on a second-order Butterworth-type Sallen and Key topology. The signals are digitised using a16-bit data acquisition card (National Instruments Inc., Austin, TX, USA) located within the main monitoring unit and interfaced to a battery-powered notebook computer via a USB cable. All data acquisition functions were performed by a virtual instrument (LabVIEW®) on the notebook computer.
2.1.2 Airway pressure monitoring
For the volunteer studies, a mouthpiece consisting of a narrow tube (8 mm internal diameter) was used as a flow resistor to allow significant airway pressures to be generated during forced breathing. A port on the mouthpiece was used to measure pressure at the mouth to give an estimation of airway pressure, connected to a signal conditioned pressure sensor (Honeywell Inc., Freeport IL, USA) via an airway gas sampling line. For the studies on mechanically ventilated patients, a short tube with a luer lock gas sampling port was placed between the ventilator tubing and the respiratory filter, and connected to the pressure sensor, as in the volunteer studies.
2.2 Experimental methods
2.2.1 Volunteer study
The study on awake volunteers was granted ethical approval by the Senate Research Ethics committee at City University, Northampton Square, London EC1V 0HB, United Kingdom and informed consent was obtained from all participants. ASA Grade 1 and 2 Consenting adults were recruited (9 M, 10 F, mean ± SD age 28.5 ± 7.83 years). Participants with respiratory disease including asthma, vascular disease and clotting disorders were excluded. Subjects were seated in a chair with the right hand resting on a table in front of them. The pulse oximeter probe was placed on the index finger of the right hand. The subjects performed 2 min of voluntary forced inspiration and expiration, in time with a periodic signal, through a narrow tube. These manoeuvres were designed to generate consistent alternating positive and negative airway pressures. The airway pressure, monitored from the mouthpiece, was displayed in real time on a computer screen so that the volunteers could aim for a prescribed rate (12 breaths per minute), rhythm (inspiratory to expiratory time ratio 1:1) and a target airway pressure (+15 to −15 cmH2O). Room temperature was controlled at 22 °C and the monitored hand was maintained at the level of the heart. Restrictive clothing which could impede venous return from the arm was removed. At the end of the monitoring period a venous blood sample was taken from the dorsum of the monitored hand and analysed immediately in a Radiometer ABL80 co-oximeter (Radiometer Inc., Copenhagen, Denmark).
2.2.2 Ventilated patient study
The study conducted on anaesthetised patients was granted full ethical approval by the East London and City Research Ethics Committee (study reference: 11/LO/0048). Patients requiring general anaesthesia and tracheal intubation were recruited to the study at the Royal London Hospital and written informed consent was obtained from all participants. Measurements were made on 40 ASA Grade 1–3 patients (18 M, 22 F, mean ± SD age 45.8 ± 16.6 years) in the anaesthetic room. Patients with significant respiratory disease, peripheral vascular disease and clotting disorders were excluded. Ambient temperature was in the range 19–23 °C in all cases. After induction of anaesthesia and intubation, patients were ventilated with an airway pressure of 15 cmH2O initially, although the pressure was adjusted in a few cases if dictated by clinical requirements. The study monitoring apparatus (finger probe, peripheral venous pressure transducer and gas sampling line) were then attached to the patient. PPG signals and airway pressures were then recorded for 10 min, at the end of which, a venous sample was taken from the dorsum of the hand and analysed using a co-oximeter (Radiometer ABL800) located near the anaesthetic room. The study apparatus was removed before the patient was transferred to the operating theatre.
2.3 Signal processing
2.4 Statistical analysis
3.1 Volunteer study
3.2 Ventilated patient study
In the volunteer group, venous oxygen saturations that were significantly lower than those of arterial blood were calculated from PPG analysis in the respiratory frequency range (SxvO2), and moderate correlation was found between these values and those obtained by co-oximetry (ScovO2). Note that a normally-breathing control group was not included as the respiratory component of the PPG is difficult to resolve under such circumstances. The effect of the airway pressure itself was the subject of an earlier volunteer study using a similar experimental setup  which concluded that forced breathing caused a fall in reported arterial oxygen saturation of 1.9% compared to control measurements. SxvO2 values in the anaesthetised subjects were also found to be significantly lower than arterial saturations (SpO2), but in contrast to the volunteer subjects, SxvO2 values were high and correlated poorly with co-oximetry. This may be largely attributed to cutaneous arteriovenous (AV) shunt induced by anaesthesia .
The circulation of the skin has a highly specialised vascular arrangement. At rest and in cool ambient temperature, a state of relative vasoconstriction prevails, controlled by the sympathetic innervation. Blood flow can be greatly increased through vasodilation, under the central control of the hypothalamus, in conditions where extensive heat transfer is required . The arteries supplying the skin are located deep in the hypodermis and give rise to plexuses of anastomosing vessels. Numerous shunts provide direct AV communications which play an important role in thermoregulation, and are densely innervated with sympathetic fibres . In addition, the dermis in the finger tips, from which PPG is recorded, contains highly convoluted AV anastomoses known as glomus bodies, providing more direct AV mixing .
When thermal or pharmacological vasodilation occurs, the oxygen content of peripheral venous blood rises significantly. In previous work studying the arterialisation of venous blood in the heated hand, venous blood gas values were found to be consistently in the arterial range . Intravenous anaesthetics, volatile agents and opioids, all of which were administered to the study subjects, cause impairment of the hypothalamic response to hypothermia and direct vasodilator action causing the skin to appear flushed and veins to become prominent. The extensive AV mixing is likely to have resulted in the high oxygen saturations observed in the dorsal vein blood samples (ScovO2) in the anaesthetised subjects and may account for the poor correlation between SxvO2 and ScovO2. The PPG-derived SxvO2 estimates were significantly lower than ScovO2 in this group, which suggests that arterialisation of dorsal vein blood is occurring.
A second limitation of the anaesthetised patients study was highlighted by observation of very high peripheral venous oxygen saturation values. The administration of high flow oxygen to patients prior to anaesthesia, which continued after anaesthetic induction to allow for the full action of muscle relaxants before tracheal intubation, probably results in the tissue beds being flooded with oxygen resulting in tissue hyperoxia and resulting venous hyperoxaemia. Although the fraction of inspired oxygen (FiO2) was reduced to 0.3 as soon as the airway was secured to minimise the effect of preoxygenation, the short period of monitoring in the anaesthetic room before transfer into the operating theatre did not allow enough time for return of arterial oxygen tension to normal values. This limitation could perhaps be overcome in future studies by recording PPG during or after the surgical procedure, which would provide time for washout of excess oxygen, thus allowing for a meaningful difference in arterial and venous oxygen saturations.
In the volunteer group, room temperature was controlled, and although this does not eliminate individual differences in cutaneous vascular tone and AV shunting, it has been possible to obtain consistently low SxvO2 that correlated well with the ScovO2 values. There was, however, wide spread of data, which may be attributed to intersubject variability in technique of forced breathing. The calculation of SxvO2 depends on the ratio of the signal amplitude between the red and infrared peaks in the respiratory frequency range. Inconsistent breathing technique causes the respiratory frequency to become indeterminate, causing errors in the calculated amplitudes of the respiratory red and infrared signals at the respiratory frequency. This affects the ratio-of-ratios and thus the calculated saturations.
The finger is often chosen as a convenient monitoring site for studying PPG, and indeed for routine clinical monitoring, but it is clear from these studies and previous work that further research is needed to enhance our understanding of the behaviour of the venous and arterial compartments within the tissues of the finger and the effect of this behaviour on pulse oximetry signals. Furthermore the effect of other physiological factors such as cutaneous arteriovenous shunt and how these are affected by physical variables such as ambient temperature warrant further study.
In conclusion, these results suggest that measurement of peripheral venous oxygen saturation remains a challenge. Although there is growing interest in utilising this concept in the monitoring of regional circulations, refinement of the technique, including application of more suitable algorithms and calibration functions, are perhaps needed to sufficiently improve accuracy to produce clinically meaningful results from peripheral sites. Cutaneous arteriovenous shunt induced by anaesthesia may especially limit the use of this technique in anaesthetised patients. Application of suitable PPG sensors  to the splanchnic circulation however, where AV anastomoses do not interfere with blood flow, could provide valuable clinical tools for assessment of oxygen uptake and metabolic status of vital organs.
Compliance with ethical standards
Conflict of interest
The authors declare that they have no conflict of interest.
- 2.Moyle JTB. Pulse oximetry. 2nd ed. London: BMJ; 2002.Google Scholar
- 6.Harvey S, Harrison DA, Singer M, Ashcroft J, Jones CM, Elbourne D, Brampton W, Williams D, Young D, Rowan K. Assessment of the clinical effectiveness of pulmonary artery catheters in management of patients in intensive care (PAC-Man): a randomised controlled trial. Lancet. 2005;366:472–7.CrossRefPubMedGoogle Scholar
- 7.Phillips JP, Belhaj A, Shafqat K, Langford RM, Shelley KH, Kyriacou PA. Modulation of finger photoplethysmographic traces during forced respiration: venous blood in motion? In: Conference proceedings of the IEEE engineering in medicine and biology society; 2012. p. 3644–7.Google Scholar
- 9.Phillips JP, Belhaj A, Langford RM, Kyriacou PA. Effect of respiratory-induced intensity variations on finger SpO2 measurements in volunteers. In: Conference proceedings of the IEEE engineering in medicine and biology society; 2013. p. 3937–40.Google Scholar
- 18.Shapiro BA, Peruzzi WT, Kozelowski-Templin R. Clinical application of blood gases. 5th ed. St. Louis: Mosby-Year Book; 1994.Google Scholar
- 22.Krouwer JS. Why Bland–Altman plots should use X, not (Y + X)/2 when X is a reference method. Stat Methods. 2008;27:778–80.Google Scholar
- 23.Phillips JP, Belhaj AM, Langford RM, Kyriacou PA, Effect of respiratory-induced intensity variations on finger SpO2 measurements in volunteers. In: Conference proceedings of the IEEE engineering in medicine and biology society; 2013. p. 3937–40.Google Scholar
- 25.Rhoades R, Bell DR. Medical physiology: principles for clinical medicine. 4th ed. London: Wolters Kluwer/Lippincott Williams and Wilkins; 2013.Google Scholar
- 26.Boron WF, Boulpaep EL. Medical physiology: a cellular and molecular approach. 2nd ed. Philadelphia: Saunders/Elsevier; 2012.Google Scholar
- 27.Young B, Heath JW, Burkitt HG, Wheater PR. Wheater’s functional histology: a text and colour atlas. 4th ed. New York: Churchill Livingstone; 2000.Google Scholar
Open AccessThis article is distributed under the terms of the Creative Commons Attribution 4.0 International License (http://creativecommons.org/licenses/by/4.0/), which permits unrestricted use, distribution, and reproduction in any medium, provided you give appropriate credit to the original author(s) and the source, provide a link to the Creative Commons license, and indicate if changes were made.