Anda di halaman 1dari 9


33. Simpson PC, et al. High-throughput genetic analysis using 56. Gale BK, et al. A micromachined electrical field-flow fractionation
microfabricated 96-sample capillary array electrophoresis (mu-EFFF) system. IEEE Trans Biomed Eng 1998;45(12):1459–
microplates. Proc Nat Acad Sci USA 1998;95:2256–2261. 1469.
34. Chiem N, arrison HDJ. Microchip-based capillary electrophor- 57. Gale BK, et al. Geometric scaling effects in electrical field flow
esis for immunoassays: analysis of monoclonal antibodies and fractionation. 1. Theoretical analysis. Analyt Chem 2001;73
theophylline. Analyt Chem 1997;69:373–378. (10):2345–2352.
35. Li PCH, Harrison DJ. Transport, manipulation, and reaction 58. Gale BK, et al. Geometric scaling effects in electrical field
of biological cells on-chip using electrokinetic effects. Analyt flow fractionation. 2. Experimental results. Analyt Chem
Chem 1997;69:1564–1568. 2002;74(5):1024–1030.
36. Cabrera CR, Yager P. Continuous concentration of bacteria in 59. Gale BK. Novel techniques and instruments for field flow
a microfluidic flow cell using electrokinetic techniques. Elec- fractionation of biological materials. Abstr Papers Am Chem
trophoresis 2001;22:355–362. Soc 2003;225:U138–U138.
37. Kremser L, et al. Capillary electrophoresis of biological par- 60. Gale BK. Miniaturized field flow fractionation systems. Abstr
ticles: Viruses, bacteria, and eukaryotic cells. Electrophoresis Papers Am Chem Soc 2004;227:U116–U116.
2004;25:2282–2291. 61. Edwards TL, et al. A microfabricated thermal field-flow frac-
38. Priego-Capote F, Castro MDLd. Dual injection capillary elec- tionation system. Analyt Chem 2002;74(6):1211–1216.
trophoresis: Foundations and applications. Electrophoresis 62. Schimpf ME, Polymer analysis by thermal field-flow frac-
2004;25:4074–4085. tionation. J Liquid Chromatogr Related Technol 2002;25
39. Debesset S, et al. An AC electro-osmotic micropump for circular (13-15):2101–2134.
chromatographic applications. Lab Chip 2004;4:396–400. 63. Janca J. Micro-channel thermal field-flow fractionation: High-
40. Glasgow I, et al. Electro-osmotic mixing in microchannels. Lab speed analysis of colloidal particles. J Liquid Chromatogr
Chip 2004;4:558–562. Related Technol 2003;26(6):849–869.
41. Hofmann O, Che D, Cruickshank KA, Muller UR. Adaptation 64. Janca J, Ananieva IA. Micro-thermal field-flow fractionation
of capillary isoelectric focusing to microchannels on a glass in the characterization of macromolecules and particles:
chip. Analyt Chem 1999;71:678–686. Effect of the steric exclusion mechanism. E-Polymers 2003.
42. Tan W, Fan ZH, Qiu CX, Ricco AJ, Gibbons I. Miniaturized 65. Janca J, et al. Effect of channel width on the retention of
capillary isoelectric focusing in plastic microfluidic devices. colloidal particles in polarization, steric, and focusing micro-
Electrophoresis 2002;23:3638–3645. thermal field-flow fractionation. J Chromatogr A 2004;1046
43. Shimura K. Recent advances in capillary isoelectric focusing: (1-2):167–173.
1997–2001. Electrophoresis 2002;23:3847–3857. 66. Bargiel S, et al. A micromachined system for the separation of
44. Kilar F. Recent applications of capillary isoelectric focusing. molecules using thermal field-flow fractionation method. Sens
Electrophoresis 2003;24:3908–3916. Actuators A-Phys 2004;110(1-3):328–335.
45. Simpson DC, Smith RD. Combining capillary electrophoresis 67. Edwards TL. Microfrabricated acoustic and thermal field-flow
with mass spectrometry for applications in proteomics. Elec- fractionation systems. Electrical and computer engineering.
trophoresis 2005;26:1291–1305. Atlanta, GA: Georgia Institute of Technology; Ph.D. thesis,
46. Schmitt-Kopplin P, Frommberger M. Capillary electrophor- 2005. p 300.
esis-mass spectrometry: 15 years of development and applica-
tions. Electrophoresis 2003;24:3837–3867. See also ANALYTICAL METHODS, AUTOMATED; PHARMACOKINETICS AND
47. Sung W-C, Makamba H, Chen S-H. Chip-Based microfluidic PHARMACODYNAMICS; TRACER KINETICS.
devices coupled with electrospray ionization-mass spectrome-
try. Electrophoresis 2005;26:1783–1791.
48. Smyth WF. Recent applications of capillary electrophoreis-
electrospray ionization-mass spectrometry in drug analysis.
Electrophoresis 2005;26:1334–1357.
49. Simo C, Barbas C, Cifuentes A. Capillary electrophoresis-mass CO2 ELECTRODES
spectrometry in food analysis. Electrophoresis 2005;26:1306–
50. Shamsi SA, Miller BE. Capillary electrophoresis-mass spec- University of California in San
trometry: Recent advances to the analysis of small achiral and Francisco
chiral solutes. Electrophoresis 2004;25:3927–3961. San Francisco, California
51. Zamfir A, Peter-Katalinic J. Capillary electrophoresis-mass
spectrometry for glycoscreening in biomedical. Electrophor- METHODS OF MEASURING BLOOD pCO2 BEFORE
esis 2004;25:1949–1963. DISCOVERY OF THE pCO2 ELECTRODE
52. Senk P, Kozak L, Foret F. Capillary electrophoresis and mass
spectrometry for screening of metabolic disorders in newborns.
Bubble Equilibration Methods
Electrophoresis 2004;25:1447–1456.
53. IV WRV, Pasas-Farmer SA, Fischer DJ, Frankenfeld CN, Carbon dioxide in blood is largely in the form of the
Lunte SM. Recent developments in electrochemical detection bicarbonate ion, which could be converted to CO2 gas by
for microchip capillary electrophoresis. Electrophoresis 2004; adding acid and extracting the gas in a vacuum. The
concept of partial pressures gradually stimulated interest
54. Qiu H, Yan J, Sun X, Liu J, Cao W, Yang X, Wang E. Microchip
in measuring pCO2 in the late nineteenth century. Gas
capillary electrophoresis with an integrated indium tin oxide
electrode-based electrochemiluminescence detector. Analyt analysis had been developed earlier, so the first method
Chem 2003;75:5435–5440. was to equilibrate a small gas bubble with a large volume of
55. Wolters AM, Jayawickrama DA, Webb AG, Sweedler JV. NMR blood sample at body temperature, and then remove the
detection with multiple solenoidal microcoils for continuous-flow bubble for gas analysis. Pflüger developed a tonometer for
capillary electrophoresis. Analyt Chem 2002;74:5550–5555. this purpose in the 1870s, and August Krogh used this

method in fish in the early twentieth century. It was

developed into a clinical and laboratory method by Richard
Riley, using a specially adapted syringe with a capillary
attached, which was invented by F. J. W. Roughton and
P. Scholander during World War II. Riley’s bubble method
worked well for pCO2, but poorly for pO2 especially when
blood was saturated with oxygen.

The Henderson–Hasselbalch Method

The most accurate early method was made possible by L. J.
Henderson’s discovery of buffering and his equation in
1908, its logarithmic modification by K. A. Hasselbalch
in 1916, and P. T. Courage’s design of a glass pH electrode
(1925) in which blood could be measured with little loss of Figure 1. Equilibration method for measuring arterial pCO2
CO2 to air. Blood pH was determined, usually at room introduced by Astrup during the Copenhagen polio epidemic,
temperature, there being no thermostated electrodes, and 1952–1954. Log pCO2 plotted versus pH results in straight lines
the Rosenthal temperature correction (0.0147 pH units/ with varying pCO2, and shifts of pH and slope when blood is acidified
8C) was used to compute pH at 37 8C. Plasma CO2 content or alkalinized. The shift gave rise to the concept of base excess.
was determined in the Van Slyke manometric apparatus
that used 1 mL of plasma (after carefully centrifuging blood
under a gas tight seal, a floating cork). This method
reached a precision of 0.3 mmHg pCO2 in studies of the
arterial to alveolar pCO2 difference during surgical
hypothermia at The National Institute of Health (NIH) (5). A carbon dioxide (CO2) electrode was first described by
physiologists Gesell and McGinty at the University of
Astrup and The Equilibration Method Michigan in 1926, for use in expired air, but not in blood
(8). It used the effect of CO2 on the pH of a film of peritoneal
Hundreds of patients with polio needed artificial ventila-
membrane wet with a salt solution. Their paper was redis-
tion in the communicable disease hospital in Copenhagen
covered 40 years later by M. Laver at Massachusetts General
during epidemics in 1950–1952. Poul Astrup, M.D. (Pro-
Hospital who informed Trubohovich of this effect (9).
fessor of Clinical Chemistry, University of Copenhagen,
In August 1954, Richard W. Stow, Ph.D. (Associate
Copenhagen, Denmark, and Director of the Clinical
Professor of Physical Medicine, Ohio State University,
Laboratory, Rigshospitalet, Copenhagen, Denmark) and
Columbus, Ohio) (Fig. 2), a physical chemist, reported
his associates, particularly Ole Siggaard Andersen,
the design of a CO2 electrode at the fall meeting of the
Ph.D., M.D. (Professor of Clinical Chemistry, University
American Physiologic Society in Madison, Wisconsin (10).
of Copenhagen, and Director, Clinical Chemistry Labora-
tory, Herlev Hospital, Copenhagen, Denmark), devised a
way of determining blood pCO2 using only a pH electrode to
measure pH before and after equilibration of a blood sam-
ple with two known concentrations of pCO2(6). Astrup
made use of the little known fact that, as pCO2 is changed,
the relationship of pH to pCO2 in a given blood sample is
semilogarithmic (Fig. 1). By plotting the two measured values
of pH at the known equilibrated pCO2, he could graphically
interpolate the pCO2 from the original sample pH.
From 1954 until the mid-1960s, Astrup’s method was
made widely available by the Radiometer Co. of Copenhagen.
The device had a thermostated capillary pH electrode, refer-
ence electrode, and tiny shaking equilibrator through which
humidified gas flowed. Astrup’s apparatus and method
became obsolete with the introduction of the CO2 electrode.
Ole Siggaard Andersen, Astrup, and others used the
values obtained for pH and pCO2 to calculate bicarbonate,
total CO2, and base excess, a term they introduced as a
quantitative measure of the nonrespiratory or metabolic
abnormality in a whole blood sample. Base excess proved to
be the first accurate index of the nonrespiratory component
of acid–base balance (7). Its first application was only for
blood, but by 1966, it was shown to apply to the extra-
cellular fluid of the entire body if one assumed an average Figure 2. Richard Stow, invented the CO2 electrode in 1954 to
extracellular fluid hemoglobin concentration of 5 g/dL. assist in managing polio patients on ventilators (10).

Figure 4. The concept of a pCO2 electrode. The pH sensitive

surface of a pH glass electrode is covered by a layer of electrolyte
(here in cellophane), and then by a thin layer of a membrane
permeable to CO2, but not to hydrogen ions (here Teflon). The
pH in that film is controlled by the partial pressure of CO2 on
the outside of the outer membrane as CO2 dissolves and reacts
with water to form carbonic acid. The carbonic acid dissociates
into hydrogen and bicarbonate ions. Because the electrolyte has
5–20 mM HCO 3 ions, changes of pCO2 have no measurable effect
on HCO þ
3 . The mass law then requires H to change in direct
proportion to change in pCO2. A doubling of pCO2 doubles Hþ
concentration, which is seen as a 0.3 pH unit fall.

pH electrode soaked in 25 mM NaHCO3 and then covered

the entire tip with a thin rubber dam, later from a surgical
glove. The bicarbonate not only made the device stable, but
doubled the pCO2 sensitivity compared with an electrolyte
Figure 3. Stow’s sketch of his 1954 CO2 electrode. (a) Cable of distilled water (or 1% NaCl). Salt was added to help
connection enclosure. (b) Rubber membrane. (c) Retaining O
stabilize the silver chloride reference electrode.
ring. (h) Chamber for internal pH electrolyte. (j) Reference
electrode of silver chloride, not in contact with internal
In 1957, Stow. Baer and Randall (11) published their
electrolyte, but opening to exterior through port K. discovery of the CO2 electrode without mentioning the need
to add bicarbonate ion, and took no further interest in this
idea. Stow had no interest in a patent, thinking it would
The polio epidemic was raging at the time, and as part of distract him from his job, and also because his university
the physical therapy faculty he had sought some way to only allowed inventors 10% of royalties. As a U.S. govern-
measure pCO2 in the victims. He read in the library about ment employee, Severinghaus was not permitted to patent
specific ion electrodes, and conceived the electrode idea. He it, certainly not with a reluctant coinventor.
had wrapped a thin rubber membrane wet with distilled Severinghaus and co-worker A. Freeman Bradley pro-
water over a homemade combined pH and reference elec- ceeded to investigate and optimize the electrode design and
trode (Fig. 3). When he changed gas pCO2 outside the to test its performance, linearity, drift, and response time.
device, the pH inside changed as a log function of gas They constructed electrodes for laboratory use by several
pCO2. However, he was unable to get stable readings colleagues, but unfortunately made no attempt at commer-
and said he doubted it could be made useful. cial development for 4 years.
After his talk, Severinghaus asked him why he did not Between 1958 and 1960 several other investigators
try adding sodium bicarbonate (NaHCO3) to the water film constructed and published similar CO2 electrodes, in several
in the electrode. He replied that he believed this would instances without being aware of the Stow–Severinghaus
abolish the signal because bicarbonate would buffer the electrode (12–14).
effect of pCO2 on pH. Severinghaus replied that he was
confident that bicarbonate would not block the sensitivity.
Stow agreed that Severinghaus would further investigate CO2 ELECTRODE DESIGN DETAILS
this idea. In September, 1954, after returning from Madi-
son to the National Institutes of Health, Severinghaus A CO2 electrode consists of a slightly spherical surfaced
confirmed the advantage of adding bicarbonate ions. A glass pH electrode and a silver chloride reference electrode.
schematic diagram of his modification of Stow’s electrode Both are mounted in a glass or plastic sleeve holding a
is shown in Fig. 4. He used a Beckman bulb-type pH Teflon or silicone rubber membrane, typically 12 mm thick,
electrode, a chloride-coated silver wire reference, and a over the glass surface, in some cases with a spacer of very
Beckman pH meter. He tied a film of cellophane over the thin lens-cleaning paper between membrane and glass to

in < 1 s by using thin silastic (silicone rubber) membrane,

low bicarbonate concentration (i.e., 1 mequiv/L), and add-
ing carbonic anhydrase to the electrolyte, but the downside
is loss of stability and signal amplitude.
The CO2 electrode is usually calibrated to read in milli-
meters of mercury. It reads the same value for gas and
liquid equilibrated with that gas at the electrode tempera-
ture, usually 37 8C.
A useful test of a leaking membrane is to equilibrate a
dilute solution (e.g., 1 mequiv/L) of HCl, or lactic acid with a
known calibration gas, and test its reading. Any leak will
permit acid entry and an erroneously high pCO2.
Figure 5. Cuvette with blood inlet and outlet connections in a Maintenance requires replacement of the membrane
thermostated water jacket made for the Stow–Severinghaus pCO2
and electrolyte when errors are detected or when drift
electrode (National Welding Co, San Francisco, 1959).
has driven the electrode beyond the ability of the apparatus
to compensate its potential. The pH glass may become so
insure a uniform distribution of the electrolyte that is NaCl impermeable to hydrogen ions that it shows low sensitivity
or KCl with  5–20 mequiv/L of NaHCO3. For use in blood, or slow response after years of use.
the electrode is mounted in a 37 8C cuvette into which a small The amplifier circuit must be electrically isolated from
sample of blood can be injected (typically 50 mL) (Fig. 5). the ground because any ground path leakage will draw
The electrode output voltage is a logarithmic function of current through the silver chloride reference and changes
pCO2,  60 mV for a 10-fold change of pCO2, which induces its potential causing drift. The input impedance of all
a pH change of  1 pH unit. Sensitivity is defined as D pH/ modern pH and pCO2 meter amplifiers is >1011 V.
D log pCO2, where S reaches nearly the ideal maximum
value of 1.0 with HCO 3 concentrations of 5–25 mM (Fig. 6). The Combined Blood Gas Analysis Apparatus
At higher bicarbonate levels, carbonate acts as a buffer,
and reduces both sensitivity and speed of response. In 1956, Leland Clark disclosed his invention of the oxygen
Response is faster at lower bicarbonate concentration, electrode at a meeting in Atlantic City to which Severin-
but carbonic acid pK’ is 6.1, resulting in some change of ghaus had invited physiologists interested in measuring
bicarbonate as pCO2 changes, reducing sensitivity. As pO2. That invention made a huge difference in blood gas
bicarbonate concentration is lowered, sensitivity falls to analysis.
30 mV/decade pCO2 change, or S ¼ 0.5, at zero bicarbonate. While Severinghaus completed his anesthesia residency
The log response is almost linear from 5 to 700 mmHg at the University of Iowa, with help from the physiology
pCO2. The response time to a step change of pCO2 is workshop, he constructed a thermostat into which he
exponential with a 95% response time of  30 s, depending mounted both the Stow–Severinghaus CO2 electrode and
on the membrane thickness and material, bicarbonate ion the Clark O2 electrode in a stirred cuvette with a small
concentration and the thickness of the electrolyte layer blood tonometer. That apparatus was exhibited at the
over the glass electrode surface. It can be made to respond meeting of the American Society of Anesthesiologists in
October 1957 and at the meeting of the Federation of
American Societies of Experimental Biology in Atlantic City
in the spring of 1958 and published in 1958 (15) (Fig. 7).

Figure 6. The first blood gas apparatus, with the Clark pO2
electrode (below) in a stirred cuvette, and the Stow– Figure 7. The first three-function blood gas analyzer, using a
Severinghaus pCO2 electrode above, tilted to keep the internal McInnes Belcher pH electrode (1930) with the pCO2 and pO2
air bubble of the pH electrode away from the tip (1957). electrodes in a 37 8C bath.

In 1958, after moving from the National Institutes of
Health to the University of California, San Francisco,
From 1951 to 1952, the discovery of oxygen related blind-
Severinghaus and Bradley added a pH electrode to the
ness in premature infants created an urgent need for
blood gas electrode waterbath, making the first three-
continuous noninvasive monitoring of blood oxygen. A
function blood gas apparatus (Fig. 8). Forrest Bird,
new solution to the problem came from physiologists study-
Ph.D., M.D. (President, Bird Corporation, Palm Springs,
ing skin respiration. Human skin breathes, taking up
California) had designed popular positive-pressure venti-
oxygen and giving off CO2 to the air. If skin is covered
lators, manufacturing them at the National Welding Co.
(as by a flat unheated pCO2 electrode) the surface tcpO2
in San Francisco. He proposed to manufacture the CO2
falls to zero in a few minutes. However, in 1951 Baumber-
electrode and to make it commercially available. From
ger and Goodfriend showed that if skin blood flow is greatly
1959–1961 the National Welding Co. sold the only avail-
increased by the highest tolerable heat (45 8C), the surface
able pCO2 electrode. The design concept was soon copied
pO2 rises to about paO2 (arterial blood) (17).
and marketed by Beckman, Radiometer, Instrumentation
Within a year after Clark’s invention of the membrane
Labs and later by several other firms.
covered platinum polarographic electrode (18,19), Rooth
used polarography to confirm the Baumberger report (20).
Impact of Blood Gas Analysis Researchers tried unsuccessfully to use chemical vasodi-
lators to make skin pO2 a monitor of paO2. Kwan and Fatt
During the 1960s, blood gas analysis became widely avail-
(21) noted that pO2 of the palpebral conjunctiva measured
able in anesthesia, intensive and critical care facilities,
with an unheated tiny Clark electrode mounted facing
and cardiorespiratory research laboratories. For several
outward on a contact lens over the cornea simulated
years, the Severinghaus paper (15) was among the most
paO2. This device was briefly marketed a decade later,
quoted articles in biologic literature, and blood gases were
but discontinued due to the danger of infection.
called the most important laboratory test for critically ill
In Marburg, Germany, Professor of Physiology Dietrich
patients. Blood gas apparatus now uses automatic self-
Lübbers and students, especially Renate Huch, pursued
calibration and automatic transport of sample and wash-
the concept of heating the skin under an oxygen electrode
ing of cuvettes, printing of results, and often sending the
by heating the electrode itself to as high as 45 8C. They
values to remote terminals. In the United States, regula-
were joined by Patrick Eberhard, and the group soon found
tions have been used by pathologists to require that these
ways of making electrically heated, thermostated oxygen
automated instruments can only be used by licensed
surface electrodes. By 1972, they had shown a good rela-
technicians, usually meaning that the income flows to
tionship between heated skin and arterial blood pO2 in
pathologists. Gone are the days when students, nurses,
infants (22). Several firms began to design electrodes for
residents, and faculty all took part in doing blood gas
this purpose.
A more complete history of the CO2 electrode and
related blood gas technology is available in References DEVELOPMENT OF METHODS AND UNDERSTANDING OF
(9,16). THEORY

By 1977, the Marburg group had published at least 11

papers documenting the validity of transcutaneous oxygen
measurement. At least three commercial tcpO2 electrode
systems were available (Helige, Roche, Radiometer). In
November 1977, some 18 research teams joined for a work-
shop on transcutaneous blood gas methods in San Fran-
cisco, assessing the theory, problems, possibilities, and
progress (23–30). The following summer (1978) many of
these workers joined the Marburg team and others for the
first international congress on transcutaneous blood gas
monitoring, establishing the technology as an essential tool
in neonatology and as useful in many other fields (31,32).
The agreement of tcpO2 with paO2 proved to be a
cancellation of two opposing effects illustrated in Fig. 9

1. Heating of desaturated blood raises its pO2 by 7%/8C,

or 50% at 43 8C, but in saturated blood, as in water,
pO2 rises only 1.3%/8C (35);
Figure 8. Relationship of pCO2 electrode sensitivity to its
internal electrolyte bicarbonate ion concentration. Maximum 2. Skin metabolism at the high temperature consumes
sensitivity occurs at  20 mM HCO 3 , but for faster response,
O2 as it diffuses outward from capillaries through
most electrodes operate at 5–10 mM. living cells, reducing the value to about paO2.

Figure 9. A schema of the effect of both heating of skin surface by

a transcutaneous electrode, and of local metabolism, on the tissue
internal oxygen tension from the arteries out past the capillaries
and the living and dead epidermis to the surface, and through the
Figure 10. The time course of skin pO2 and pCO2 on an arm after
electrode membrane into the cathode that keeps its surface pO2 ¼ 0
sudden circulatory occlusion with a blood pressure cuff. The rate of
by its electrical negative potential (39).
fall of pO2 from a high level is a measure of the skin metabolic rate.
The pCO2 rises at first from metabolic CO2 production, but later at
The outward oxygen diffusion is facilitated by heat that a steeper rate as skin generates lactic acid when skin pO2 reaches
zero. With release of occlusion, the electrode recovery time is
proved to ‘‘melt’’ some skin diffusion barriers (33,36). Skin
delayed by both the skin washin and washout, and by electrode
O2 conductivity C (adult volar forearm) was determined by
equilibration (34).
two groups by comparison of flux with two membranes
(teflon and mylar) of very high and low conductivity. With
Severinghaus (48). Figure 12 schematically shows the
a large gold cathode Clark electrode, C ¼ 15 nL  cm2  s1 
internal design of an early Radiometer combined electrode.
atm1(37) and with a mass spectrometer C ¼ 10 nL  cm2 
Figure 13 shows the electrode with a membrane mounted.
s1  atm1(38).
When a heated combined pO2–pCO2 electrode is first
Skin O2 consumption (VO2) was determined after ther-
attached to skin, the time needed to equilibrate is  5 min
mal vasodilation by the rate of fall of tcpO2 with circulatory
for both electrodes, although the pO2 electrode may show
occlusion (arm cuff) (Fig. 10) (27). Relative skin blood flow
later small changes as thermal vasodilation slowly develops
under the heated electrode was estimated by measuring
(Fig. 14). The response to step changes in alveolar and
the required heating power (39). Analysis of data collected
arterial pCO2 is slower as seen in Fig. 15. Here the response
at two levels of pO2 and two temperatures permitted
is delayed both by the washout or washin of CO2 into the
calculation of blood flow, capillary temperature under a
tissue by blood flow, and the electrode’s own delay. The
heated electrode, and diffusion gradient from capillary to
response is pseudoexponential, a combination of the two
surface (40). Mean adult volar forearm skin VO2 was
delays, resulting in a 95% response times of  10 min.
4.2 0.4 mL  g1min1 at 44 8C and 2.8 0.3 mL  g1  min1
Without correction, tcpCO2 is not similar to paCO2.
at 37 8C. At 44 8C, skin blood flow averaged 0.64 0.17 mL 
Heating of blood (and water) raises pCO2  4.6%/8C
g1  min1, capillary temperature was 43 8C and the diffusion
gradient was 32 7 mmHg.


In 1959, Severinghaus constructed a 37 8C thermostated

open tipped CO2 electrode to determine pCO2 of various
tissue surfaces in animals (Fig. 11) (41). Without heating
the skin well above body temperature, skin pCO2 at 37 8C
climbed steadily over one-half of an hour to > 80 mmHg.
Dog intestinal mucosa and liver surfaces were very high.
Fifteen years later, the success in transcutaneous mea-
surement of oxygen led to design and testing of electrodes
to measure tcpCO2 by Beran et al. (42,43), Huch et al. (44)
and Severinghaus et al. (45,46). Combined tcpO2–tcpCO2 Figure 11. The first tissue surface pCO2 electrode (41) with a
electrodes were initially described by Parker et al. (47) and circulating temperature controlled water jacket.

Figure 14. Initial responses of a combined pO2–pCO2 electrode

when first mounted on skin. Both electrodes need  5 min to
equilibrate, the pO2 showing a small late rise as skin
hyperemia develops from the heating (49).

(41), metabolism adds  3 mmHg pCO2, and the cooling by

skin and blood of the electrode surface further raises the
electrode reading. The effect of heating on blood pCO2 may
be computed as DpCO2 ¼ exp(0.046[T – 37]) (51). The net
effect at 43 8C was found to be tcpCO2 ¼ 1.33paCO2 þ
Figure 12. Schema of the design of a combined tcpO2–tcpCO2 4 mmHg (48,52) or tcpCO2 ¼ 1.4paCO2 (53). This form of
electrode. (a) pO2 cathode, the end of a 12 mm platinum wire
temperature-dependent correction factor was later incor-
fused in glass. (b) A silver wire reference electrode. (c) pH glass
electrode surface. (d) solid silver internal pH electrode (used
porated in most commercial transcutaneous blood gas
to improve heat transfer to skin). (e) Internal pH electrolyte. monitoring apparatus.
(f) Heater Zener diode. (g) Thermistor. (h) Silver body, and With this correction factor, the relationship of tcpCO2 to
reference electrode. J, K, L, M: O rings. N: Lexan jacket. Q: paCO2 is excellent, as shown in Fig. 16. The previous
epoxy. P: Cable (48). correction factors appear to have become incorrect for a
second generation of the Radiometer tcpCO2 electrodes, due
to a design change in the internal temperature coefficient
of the glass pH electrode. The additive factor of 4 mmHg
changed to  8 mmHg in the newer instruments (Kagawa
S, personal communication).
Although tcpCO2 appears to work at 42–43 8C, Tremper
et al. showed that 44 8C was a better temperature when

Figure 15. Transcutaneous pCO2 electrode response to a step

increase of ventilation adjusted by the subject to reduce end tidal
pCO2 suddenly from 40 to 20 mmHg and hold it at a constant level
for 18 min, followed by addition of enough CO2 to inspired gas to
Figure 13. Photograph of combined pO2–pCO2 electrode with raise end tidal pCO2 as quickly as possible to  50 mmHg. The
teflon membrane (48). response time constants (63%) are  5 min (45).

fetal pCO2 or raise pO2(70). During maternal hypo-

capnia, fetal tcpO2 fell due to the Bohr effect, whereas
it rose during hypercapnia (71). Fetal tcpO2 was
considered influenced by local scalp blood flow (72).
Repeated episodes of asphyxia were reported to
express catecholamines, which reduced blood flow
to the fetal skin, artifactually reducing tcpO2(73,74).
Fetal tcpCO2 may have failed to disclose severe
acidosis or circulatory impairment (75).
3. Sleep Studies: Combined pO2–pCO2 electrodes are
used in sleep studies in combination with pulse
oximetry, because nostril sampling of end-tidal
pCO2 is somewhat annoying and more apt to become
plugged or dislodged (76–83). The combined tcpO2–
tcpCO2 electrode made it possible to show that the
ventilatory response to induced mild hypoxia in
sleeping infants changes with age from acute depres-
sion at 1–5 days, to stimulation at 4–8 weeks, and mild
or no stimulation at 10–14 weeks (84). A method was
Figure 16. Transcutaneous pCO2 correlates well with arterial designed for estimating the ventilatory response to
pCO2 in patients during anesthesia or intensive care (48). CO2 during sleep using capnography and tcpCO2(79).
4. Peripheral Circulation: The tcpO2 electrodes are
blood pressure was or had been low (54). The tcpCO2 value extensively used in evaluating arterial disease in
was better than the PETCO2 value (end-tidal or end-expired the peripheral circulation (85–88). A test of ade-
air) in predicting paCO2 (bias and s.d. 1.6  4.3 mmHg) in quacy of peripheral circulation, ‘‘initial slope index’’
anesthetized adults (n ¼ 24) (55). (ISI) was suggested by Lemke and Lübbers (89).
A special advantage of tcpCO2 is that it averages out Blood flow is stopped by an arm cuff above the
breath-by-breath variations, and has almost no inherent electrode and restarted when tcpO2 ¼ 0. The initial
‘‘noise’’ or variability, such that it often is found to be the rate of rise should be a slope per min of at least 75%
best trend monitor for detecting small changes in paCO2 of the preocclusion tcpO2.
such as those induced by experimental variations (anesthe- 5. Skin Circulation: Monitoring the viability of skin
sia, ventilatory settings, posture, FIO2, FICO2, blood pres- after injury or transplant or flap movement (90,91).
sure, pharmacologic agents, etc). 6. Ventilatory Control: In intensive care, transcuta-
neous electrodes greatly increased the safety and
APPLICATIONS simplicity of PEEP optimization and respiratory
management of adults with respiratory distress
Transcutaneous technology is used in many ways, some of syndrome (92). They are widely used simply to
which are discussed in accompanying papers: reduce arterial blood sampling.
7. Hyperbaric Oxygen: Monitoring and guiding hyper-
1. Neonatology: Guidance of O2 therapy remains the baric oxygen therapy, primarily for infections and
most common use of transcutaneous monitoring wound healing (93,94). The tcpO2 tracked paO2 up
(56–58). The suspected etiologic role of hyperoxia to 4-atm hyperbaric pressure in normal subjects
(tcpO2 > 80 mmHg) in retinitis of premature infants (95). Surprisingly, no one has reported using tcpO2
has been confirmed in a cohort study (59). The tcpO2 in hyperbaric treatment of CO poisoning despite the
value can be measured above and below the ductus demonstration by Barker and Tremper in experi-
to demonstrate closure (60). In low birth weight mental CO administration that transcutaneous pO2
infants, tcpCO2 (at 40 8C!) is the best available falls linearly as COHb increases, and reaches about
monitor of ventilation (61). one-fifth of its initial value at the highest COHb
2. Fetal Monitoring: Using specially designed electro- levels despite the maintenance of constant arterial
des attached to the fetal scalp, intrapartum monitor- pO2(96). It is thus unknown whether HBO can
ing revealed some important new pathophysiologic normalize tissue pO2 in the presence of high levels
understanding (62–65). As hoped, changes in tcpO2 of COHb.
rapidly reflected changing maternal and fetal condi- 8. Clinical Physiology: Transcutaneous monitoring has
tions (66). The tcpO2 value fell and tcpCO2 rose with found use in exercise tolerance studies (97,98). End-
contractions during the second stages of labor (67). tidal CO2 is not exactly equal to paCO2 and the
The tcpCO2 value closely followed fetal paCO2(68). difference between them varies with posture and
When there were signs of fetal distress, fetal scalp inspired oxygen concentration. When testing hypoxic
tcpO2 was < 15 mmHg (69). Surprisingly, O2 admin- ventilatory responses by monitoring PETCO2, tcpCO2
istration to mothers with fetal distress did not alter helps to correct these small errors (99).

9. Pharmacologic Research: Transcutaneous monitor- oximetry, neonatologists found that oximetry failed to
ing may be the simplest monitor of the depressant detect hyperoxia adequately (110) and now mostly use both
effects of opiates, sedatives, and anesthetics espe- technologies (111–115). In neonatology, a significant pro-
cially in awake children (100). blem is that the inherent errors of pulse oximetry are  3%,
10. Animal Studies: Intestinal or other tissue animal which could fail to warn of paO2 > 80 unless a set point of
experimental ischemia has been found to be better  90% SpO2 is chosen (116). Some have arbitrarily dis-
detected by the rise of the organ or tissue surface missed transcutaneous monitoring as ‘‘. . .plagued by tech-
pCO2 using tcpCO2 electrodes at body temperature nical problems, . . .Its use in efforts to prevent retinopathy
than by gastric tonometry (101). Both tcpO2 and of prematurity, an eye disease of preterm newborns often
tcpCO2 have been widely used in small and large leading to blindness, proved disappointing’’ (117). To them,
animal studies (102) and to assess the effect of the transcutaneous field served as a model of problems
cardiopulmonary resusitation (CPR) (103). in medical innovation, new technology, and personnel
training. Not everyone agrees with this pessimism. Most
technical problems have been solved, and the occurrence of
ACCURACY blindness in very premature infants is now believed to be
multifactorial, not just due to hyperoxia. Therefore when it
With the widespread use of tcpO2 and tcpCO2 came concern occurs, it is not appropriate to attribute it to failed trans-
about its accuracy and the possible sources and effects of cutaneous methodology.
errors, especially with severe hypotension (28,104). Pea-
body et al.(25) identified two groups of infants in whom
tcpO2 was lower than paO2. These were infants receiving CONCLUSIONS
an intravascular infusion of tolazoline and infants with
mean arterial blood pressures > 2.5 s.d. below the pre- The enthusiasm for transcutaneous blood gas analysis of
dicted average value. Vasoconstrictors also lower the period 1976–1986 was followed by a decrease due to the
tcpO2(105). Both of these situations represent extreme advent of pulse oximetry. The number of papers per year
alterations in peripheral blood flow. Mild hypotension, listing medline keywords ‘‘transcutaneous blood gas’’
hypothermia, anemia, radiant warmers, and bilirubin reached an early peak of 75 in 1979, when the first inter-
lights did not adversely affect transcutaneous accuracy national symposium was devoted to this field, in Marburg
(106). In a large multiinstitutional study of 327 patients and 200 in 1987. However, after 1986 many papers used
older than 1 month, when paO2 was between 80 and the keywords ‘‘transcutaneous blood gas’’ when writers
220 mmHg, Palmisano found the mean bias  s.d. of tcpO2 meant to refer to pulse oximetry.
was 43  40 mmHg, and the slope of the regression was Transcutaneous technology is inherently somewhat
0.65 (107). It was determined that tcpCO2 correlated far complicated. Users must change membranes and calibrate,
better with paCO2: R ¼ 0.929, slope 1.052, bias and change skin sites periodically to avoid burns, beware of
s.d. ¼ 1.3  4.0 mmHg (n ¼ 756). drift or error due to poor circulation or poor skin attach-
Defining a tcpO2 index as tcpO2/paO2, Tremper and ment, and take account of the slower response than given
Shoemaker (108) studied the effect of shock. For 934 data by oximetry. Nonetheless, transcutaneous blood gas mea-
sets taken on 92 patients not in shock, there was a correla- surement continues to be used because of its unique ability
tion coefficient (r) of 0.89 and a tcpO2 index 0.79  0.12 to meet many special situations needing its characteristics
(SD). In five patients with moderate shock, the r was 0.78 of noninvasively and continuously determining partial
and the tcpO2 index was 0.48  0.07. In nine patients with pressures of O2 and CO2. Several professional organiza-
severe shock, there was no correlation between tcpO2 and tions have published guidelines for use of these monitors
paO2 and the tcpO2 index was 0.12  0.12. (118,119).


Skin burns may occur after an electrode has been in one 1. Severinghaus JW. The current status of transcutaneous
place over several hours at 44–45 8C, and sometimes even blood gas analysis and monitoring. Blood Gas news (Radio-
at 43 8C. Long-term monitoring requires site changes, or a meter house organ) 1998;7:4–9.
dual electrode alternating system (109). There may be 2. Severinghaus JW. The Invention and Development of Blood
Gas Apparatus. Anesthesiology 2002;97:253–256.
problems with drift of calibration, membrane failure, par-
3. Severinghaus JW. Severinghaus electrode. In: JR Maltby,
tial loss of skin contact giving errors in both O2 and CO2
editor. Notable Names in Anaesthesia. London: Royal Society
readings. Maintenance of these electrodes requires train- of Medicine Press Ltd.; 2002.
ing and some technical proficiency. 4. Severinghaus JW, Astrup P, Murray J. Blood gas analysis
and critical care medicine. Am J Respir Crit Care Med
IMPACT OF PULSE OXIMETRY 5. Severinghaus JW, Stupfel MA, Bradley AFJ. Accuracy of
blood pH and pCO2 determinations. J Appl Physiol 1956;19:
Pulse oximetry came into widespread use in 1985–1987, 189–196.
and quickly replaced transcutaneous blood gas analysis in 6. Astrup P. A simple electrometric technique for the determi-
many situations. However, after an initial switch to nation of carbon dioxide tension in blood and plasma, total