Comparison of Transcutaneous and Capillary Measurement of PCO in 2

Hypercapnic Subjects
Sven Stieglitz MD, Sandhya Matthes, Christina Priegnitz MD, Lars Hagmeyer MD, and
Winfried Randerath MD

BACKGROUND: Measurement of PCO2 is vital in determining effective alveolar ventilation. How-

ever, obtaining capillary PCO2 by a skin prick of the earlobe is painful, and nocturnal measurements
disturb sleep. End-expiratory measurement of PCO2 is also well established, but there is a low
precision in predicting arterial or capillary CO2. The purpose of the study was to evaluate nocturnal
measurement of noninvasive, transcutaneous PCO2 (PtcCO2) measurement in hypercapnic subjects.
METHODS: In this prospective study, 31 subjects with chronic hypercapnic failure—in a stable
phase of the underlying disease—and a control group of 12 healthy volunteers were included.
Transcutaneous measurements were taken by the Tosca sensor (Radiometer, Copenhagen, Den-
mark) over a period of at least 6 h during the night. A capillary blood gas was measured at midnight
and 4:00 am. RESULTS: The mean nocturnal capillary P CO 2 (P capCO 2 ) of subjects was
50.6 ⴞ 10.2 mm Hg. In the 31 subjects with known hypercapnic respiratory failure, the correlation
between PtcCO2 and PcapCO2 at midnight was 0.86 and at 4:00 am r ⴝ 0.80. The bias of the
hypercapnic subjects was d ⴝ ⴙ 4.5 with a limit(s) of agreement of 2 SD ⴝ 13.0. The process of
blood sampling caused no significant change in PtcCO2. CONCLUSIONS: Our study evaluated
transcutaneous capnography as a continuous nocturnal measurement in hypercapnic subjects. We
found a good agreement between the methods. Because CO2 is not constant in patients with
respiratory failure, but instead fluctuates, we would recommend the continuous transcutaneous
measurement of PCO2 as our method of choice in the diagnosis of nocturnal hypercapnia. Key words:
chronic obstructive pulmonary disease; hypercapnic failure; hypoventilation; noninvasive measurement
of carbon dioxide pressure; obesity-hypoventilation syndrome; sleep-disordered breathing. [Respir Care
2016;61(1):98 –105. © 2016 Daedalus Enterprises]

Introduction (PCO2 ⬎ 45 mm Hg) in which alveolar ventilation is ab-

normally low in relation to oxygen uptake or carbon di-
The assessment of PCO2 is the most important parameter oxide output.1 The accepted standard of measuring PCO2 is
for evaluation of alveolar ventilation and is therefore es- analyzing CO2 through an arterial blood gas analysis. Be-
sential in the diagnosis and treatment of hypoventilation cause obtaining a blood gas sample from an artery is an
invasive and painful procedure2 with the risk of arterial
lesion, this method is mostly used in the ICU or during
surgery. An alternative method is measuring capillary PCO2
Dr Stieglitz is affiliated with the Department of Pneumology and Cardi-
(PcapCO2) using blood obtained from a skin prick of the
ology, Petrus Hospital Wuppertal, Academic Teaching Hospital of the
University of Duesseldorf, Wuppertal, Germany; Ms Matthes and Drs earlobe, which is a simple and less painfulness method of
Priegnitz, Hagmeyer, and Randerath are affiliated with the Clinic for obtaining PCO2.3 With the exception of severe shock, cap-
Pneumology and Allergology, Centre for Sleep and Ventilation Medi-
cine, Bethanien Hospital, Institute of Pneumology, University Witten/
Herdecke, Solingen, Germany.

A version of this paper was presented during a poster session of the 47th Correspondence: Sven Stieglitz MD, Department of Pneumology and
Congress of the German Society of Pneumology, held——March 1, 2006, Cardiology, Petrus Hospital Wuppertal, Carnaper Strasse 48, 42283 Wup-
in Nürnberg, Germany. pertal, Germany. E-mail: [email protected].

The authors have disclosed no conflicts of interest. DOI: 10.4187/respcare.03917

98 RESPIRATORY CARE • JANUARY 2016 VOL 61 NO 1

 COMPARISON OF TRANSCUTANEOUS AND CAPILLARY PCO2

illary blood gas analysis accurately reflects arterial PCO2

(PaCO2), with earlobe sampling being slightly more accu- QUICK LOOK
rate than fingertip sampling, as concluded by a 2007 meta- Current knowledge
analysis.4
Monitoring arterial carbon dioxide (PaCO2) from arterial
PCO2 can also be measured noninvasively and continu-
blood samples is the standard of care for evaluating the
ously through the end-tidal measurement of expired air
effectiveness of alveolar ventilation. Alternatives to
(PETCO2). Its routine use is recommended in every patient
blood gas analysis include airway carbon dioxide and
with invasive ventilation during anesthesia,5 although there
transcutaneous monitoring. The former is affected by
is low precision in predicting arterial or capillary PCO2.6-8
the presence of ventilation perfusion matching in the
Nevertheless, detecting PETCO2 via a face mask is not ac-
lung and cardiac output, whereas the latter is affected
curate9 because of the leakage of the masks and is there-
by tissue perfusion (cardiac output, skin temperature,
fore not recommended during sleep or noninvasive venti-
and vasoconstriction).
lation.10 This also applies to the measurement of PETCO2
via a nasal cannula, which allows a qualitative analysis of What this paper contributes to our knowledge
ventilation and may be suitable for monitoring during an-
esthesia11 but does not allow a reliable qualitative calcu- In a small group of spontaneously breathing, normal
lation of PaCO2. and hypercapnic subjects, transcutaneous monitoring of
Measurement of PCO2 during sleep is important because CO2 closely tracked arterialized capillary blood CO2 at
respiratory mechanics change during sleep and abnormal rest and during exercise. The limits of agreement be-
gas exchange worsens, or may only be present during tween measurements were greater in hypercapnic
sleep, in disorders such as COPD, scoliosis, or obesity- subjects.
hypoventilation syndrome.12 If PCO2 fluctuates during the
night in subjects with respiratory failure (Fig. 1),13 taking
only one blood sample at a random single point of the
night may lead to an over- or underestimation of respira- lung function. Only subjects with stable disease of chronic
tory failure. hypercapnic failure due to obesity-hypoventilation syn-
The measurement of transcutaneous PCO2 (PtcCO2) has drome, COPD, or thoracorestrictive diseases like scoliosis
been examined in critically ill14-16 or ventilated adults17,18 were included. Patients who needed emergency and inten-
and newborns,19 as well as during bronchoscopy20 or inter- sive care treatment such as intubation or the application
hospital ground transport. The aim of our study was to of vasoactive drugs were excluded. Exclusion criteria
compare PtcCO2 with PcapCO2 in hypercapnic subjects dur- also included home mechanical noninvasive ventilation
ing the night. Capillary instead of arterial blood gas mea- before study inclusion as well as lack of informed con-
surement was chosen because of its impact on routine sent. To prove the accuracy of the system in chronic
clinical practice. It was also planned to evaluate whether hypercapnic failure, there was a control group of 12
the process of blood sampling from the earlobe leads to a subjects with normocapnia.
change in the blood gases because obtaining a blood sam-
ple from a sleeping subject may awaken the subject or at Design
least disturb the subjects’ breathing pattern, which could
theoretically underestimate PCO2. Measurements were made as long-term measurement
during a sleep period of at least 6 h in a hospital room.
Methods Capillary blood gas analysis was performed at midnight
and 4:00 am after rendering the earlobe hyperemic by a
Subjects nicoboxil/nonivamide-containing cream. Blood gas analy-
ses were performed twice because of the variation of PCO2
We conducted a prospective study at a university-affil- during sleep and served also to detect a drift of PtcCO2.
iated pulmonary care center. The study protocol was ap- Drift was defined by an increase of PtcCO2-PcapCO2 differ-
proved by the local ethics committee of the University of ence overnight. The samples were analyzed immediately
Witten/Herdecke. After giving their written informed con- (Ecosys II, Eschweiler, Kiel, Germany). Transcutaneous
sent, 31 subjects with either proven or suspected hyper- long-term monitoring of PtcCO2 by an earlobe sensor was
capnia who were admitted to hospital were included in the performed during the whole night. The PtcCO2 value
trial (from May 2006 to August 2007). Hypercapnia was was taken from the CO2 monitor at the time of the blood
suspected on the grounds of clinical assessment in com- gas analysis. To detect sample error caused by taking a
bination with either the results of a previous blood gas blood gas sample during sleep, PtcCO2 values were addi-
analysis with elevated PCO2 or by severe limitation of the tionally measured 30 min before and 30 min after the

RESPIRATORY CARE • JANUARY 2016 VOL 61 NO 1 99

 COMPARISON OF TRANSCUTANEOUS AND CAPILLARY PCO2

Fig. 1. Example of long-term capnometry (9 h) with simultaneous measurement of oxygen saturation (top), heart rate (middle), and carbon
dioxide (bottom) by a single Tosca earlobe electrode in a subject with intermittent nocturnal hypercapnia. Each period of hypoventilation
(hypercapnia) is accompanied by desaturation and increase of the heart rate except midway through the plot, when both transcutaneous
PCO2 and SpO2 drop, which is probably due to an artifact of the PCO2 measurement.

blood sample (11:30 pm, midnight, 12:30 am; 3:30 am, after blood sampling were calculated through the analysis
4:00 am, 4:30 am).21 In addition, a 6-min walk test and a of variance. The drift of transcutaneous capnography was
cycle ergometer exercise test were performed whenever calculated by comparing the difference between PtcCO2
possible. and PcapCO2 at midnight with the values at 4:00 am using
the Wilcoxon test. PtcCO2 and PcapCO2 were correlated by
Transcutaneous Capnometry linear regression, with calculation of Pearson coeffi-
cient of correlation. The bias and the agreement be-
The tc Sensor 92 of the Tosca 500 monitor (Radiometer, tween the 2 different methods measuring CO2 was cal-
Copenhagen, Denmark) offers simultaneous monitoring of culated as described by Bland and Altman.24 The bias
PtcCO2, oxygen saturation (SO2pO2), and pulse rate by a (d) was defined as the mean difference between PtcCO2
single sensor. The Tosca uses a temperature of 42°C, which and PcapCO2. The limits of agreement between the 2
lowers the reactivity to rapid changes in PaCO2, but avoids methods are defined as d ⫾ 2s with s being the standard
heat damage and still shows a good correlation to PaCO2. deviation of d.
Because heat also increases the metabolism of poorly
oxygenated skin layers (anaerobic factor) by approximately
Results
4.5%/°C, PtcCO2 is measured higher than capillary CO2.22
PtcCO2 is not measured directly but potentiometrically de-
rived by determining the pH of the electrolyte layer sep- The 12 healthy volunteers hat a normal daytime PcapCO2 at
arated from the skin by a highly permeable membrane rest (36.9 ⫾ 3.2 mm Hg) and at exercise (38.5 ⫾ 0.8 mm Hg),
and calculated by the equation of Severinghaus: as well at night (37.6 ⫾ 6.1 mm Hg). In contrast, the 31
PaCO2 ⫽ (tcCO2 – 5 mm Hg)/10(l0.019 ⫻ [T - 37]),23 where T subjects with suspected hypercapnia showed a higher day-
is the temperature in degrees celsius and PaCO2 time PcapCO2 at rest (47.2 ⫾ 10.0 mm Hg, P ⬍ .001) and at
the estimated arterial value displayed on the screen of the exercise (49.5 ⫾ 5.4 mm Hg, P ⫽ .02), as well as at night
monitor after temperature correction. Before starting the (50.6 ⫾ 10.2 mm Hg, P ⬍ .001) (for details, see Table 1).
measurement each night, a calibration of the system was The main reasons for hypercapnia were COPD,
carried out. The sensor was placed at the subject’s earlobe. obesity-hypoventilation syndrome, and thoracorestrictive
The Tosca sensor may remain during a period of 12 h. In diseases like scoliosis.
vivo calibration using a reference capillary blood sample Because some blood gas analyses were not performed,
was not performed. only 51 of 62 possible paired measurements of PtcCO2
before/after blood gas sampling were obtained. At 30 min
Analysis before/after blood gas sampling, there was no signifi-
cant change of PtcCO2, indicating that there was no
To evaluate changes of blood gas analysis by the sam- interference caused by blood sampling (Fig. 2, A and
pling capillary blood, the PtcCO2 values 30 min before and B).

100 RESPIRATORY CARE • JANUARY 2016 VOL 61 NO 1

 COMPARISON OF TRANSCUTANEOUS AND CAPILLARY PCO2

Table 1. Anthropometric Data and Results of Different Capillary Blood Gas Analysis in Subject Groups

Healthy Controls Hypercapnic Subjects Obesity-Hypoventilation COPD Thoracorestrictive

(n ⫽ 12) (n ⫽ 31) Syndrome (n ⫽ 10) (n ⫽ 15) (n ⫽ 6)

Age (y) 42.2 ⫾ 18.6 67.2 ⫾ 11.5 65.5 ⫾ 14.7 67.7 ⫾ 9.6 68.5 ⫾ 11.5
Male 4 16 5 9 2
Female 8 15 5 6 4
Height (cm) 165.5 ⫾ 13.2 163.7 ⫾ 10.4 162.1 ⫾ 11.4 166.0 ⫾ 9.7 160.8 ⫾ 11.0
Weight (kg) 79.8 ⫾ 19.0 82.3 ⫾ 20.4 103.4 ⫾ 17.7 72.4 ⫾ 11.8 71.7 ⫾ 14.4
BMI (kg/m2) 29.1 ⫾ 6.1 30.9 ⫾ 7.9 39.5 ⫾ 5.8 26.4 ⫾ 5.0 27.9 ⫾ 5.7
6MWT (m) NA 253.0 ⫾ 20.2 411.0 ⫾ 270.1 74.5 ⫾ 75.7 294.0
Daytime PO2 (mm Hg) 87.5 ⫾ 19.0 59.1 ⫾ 17.6 59.6 ⫾ 12.4 55.6 ⫾ 6.4 67.3 ⫾ 38.2
Daytime PCO2 (mm Hg) 36.9 ⫾ 3.2 47.2 ⫾ 10.0 43.9 ⫾ 5.9 44.1 ⫾ 4.2 61.3 ⫾ 15.0
Exercise PCO2 (mm Hg) 38.5 ⫾ 0.8 49.5 ⫾ 5.4 46.4 48.2 ⫾ 4.8 57.1 ⫾ 0.9
Nocturnal PCO2 (mm Hg) 37.6 ⫾ 6.1 50.6 ⫾ 10.2 48.5 ⫾ 7.0 47.7 ⫾ 07.7 62.1 ⫾ 13.3

Exercise PCO2 was obtained by cycle ergometer. Daytime and nocturnal PO2 and PCO2 were obtained by capillary blood gas analysis.
BMI ⫽ body mass index
6MWT ⫽ 6-min walk test
NA ⫽ not applicable

measurements with only 68 of 86 pairs due to missing

data), indicating that there was no variation over time
(drift during the night).
The correlation between PtcCO2 and PcapCO2 in the nor-
mocapnic control group at midnight was r ⫽ 0.78 and at
4:00 am r was 0.71. In the hypercapnic subjects, a total of
55 paired measurements by earlobe sensor and by capil-
lary blood gas analysis were obtained. Considering only
the 31 subjects with hypercapnia, the correlation between
PtcCO2 and PcapCO2 at midnight was 0.86 and at 4:00 am
r ⫽ 0.80 (Fig. 3, A and B). Summarizing subjects and
control group, the correlation between PtcCO2 and PcapCO2
at midnight was r ⫽ 0.84 and at 4:00 am r was 0.84.
Furthermore, the difference between PtcCO2 and PcapCO2
was constant between the different levels of CO2.
Because correlation measures only relationship and not
agreement between 2 methods, we calculated the bias and
limits of agreement as described by Bland and Altman.24
The bias of the transcutaneous capnography of healthy
controls was 4.9 mm Hg, and the limits of agreement were
8.8 mm Hg. The bias of the hypercapnic subjects was
d ⫽ ⫹ 4.5 mm Hg with a limits of agreement of 13.0 mm Hg
(Fig. 4).

Discussion
Fig. 2. Subjects and control group are shown together. Invasively
sampling blood gas analysis did not influence the ventilation mea-
sured by the short-term trend of transcutaneous PCO2 30 min be-
We evaluated the accuracy of measuring noninva-
fore and after initiation (A) and 30 min before and after 4:00 am (B).
PtcCO2 ⫽ transcutaneous PCO2. sively by the Tosca sensor in hypercapnic subjects dur-
ing the night. Our studies21 are the only studies to ex-
amine transcutaneous capnography via the comparison
Furthermore, the difference between PtcCO2 and PcapCO2 of capillary instead of arterial blood gas measurement
did not change between midnight and 4:00 am (subjects with PtcCO2. This is important because of its impact on
n ⫽ 24, P ⫽ .19; volunteers n ⫽ 10, P ⫽ .20; 34 paired routine clinical practice.

RESPIRATORY CARE • JANUARY 2016 VOL 61 NO 1 101

 COMPARISON OF TRANSCUTANEOUS AND CAPILLARY PCO2

Fig. 4. Bland-Altman plot graphing the bias and the limits of agree-
ment on the vertical against the average of the 2 different methods
on the horizontal. The plot shows the results of the hypercapnic
subjects only (values of baseline and 4:00 am together). Bias and
limits of agreement first of all show good accordance of transcu-
taneous PCO2 and capillary PCO2. The bias was d ⫽ ⫹4.5 with a
limits of agreement of 2 SD ⫽ 13. Second, the bias does not
increase with increasing hypercapnia. PtcCO2 ⫽ transcutaneous
PCO2, PcapCO2 ⫽ capillary PCO2.

In our example (Fig. 1), PCO2 is not constant during the

night in subjects with ventilatory failure. The greatest in-
crease in PCO2 is during the REM phase of sleep because
Fig. 3. Demonstrated are the subjects with hypercapnia only. Dif-
ferences between transcutaneous PCO2 and capillary PCO2 at base- skeletal muscle hypotonia during REM includes the ac-
line (A) show a correlation of r ⫽ 0.8615 (interception y-axis ⫺2.03) cessory muscles of respiration (sparing only the diaphragm).
and at 4:00 am with (B) a correlation of r ⫽ 0.8061 (interception The capillary blood of the earlobe takes a while to equil-
y-axis is 11.38). PtcCO2 ⫽ transcutaneous PCO2, PcapCO2 ⫽ capil- ibrate with the arterial CO2, so that tcCO2 has a certain
lary PCO2.
delay when PCO2 is changing: one study analyzing the
TCM (Radiometer, Copenhagen, Denmark) found a lag of
We found a good but not an ideal agreement compared 17 s with increasing and a lag of 78 s (90% response time)
with capillary blood gas analysis. Theoretically, the ideal with decreasing PCO2.29 Thus, the lag of the systems ex-
agreement should have a bias of 0. However, there will plains why a bias of 0 cannot be found in subjects with
never be a bias of 0, regardless of which kind of transcu- rapidly changing PCO2 levels, as can be observed in sub-
taneous monitoring system is used. The first reason is that jects with hypercapnic failure during sleep.
transcutaneous PCO2 is a different parameter compared with The third important reason explaining the difference
capillary or arterial PCO2. PtcCO2 measures epidermal CO2 between transcutaneous and capillary PCO2 is that both
content and is not a direct measurement of arterial or cap- analytical apparatus also have a certain error in measuring.
illary CO2. Moreover, there is additional difference be- Therefore, the error in measurement is not only caused by
tween PtcCO2 and PcapCO2 caused by the CO2 production by the transcutaneous PCO2 measurement but also by analyz-
the living epidermal cells themselves, which contributes to ing the capillary PCO2, although there is a systemic rather
the capillary CO2 level at a constant amount (metabolic than a random difference between the 2 values.
constant). The PtcCO2 is increased by about 4 –5 mm Hg Surprisingly, the agreement between PtcCO2 and PcapCO2
due to the metabolic rate. There is a correction algorithm did not worsen with the magnitude of hypercapnia, al-
used by manufactures of PtcCO2 systems as described by though one could expect this due to statistical reasons: if
Severinghaus et al.25 Metabolic processes26 as well as there would be a constant deviation of, eg, 10%, then the
the capillary blood flow27 of the earlobe contribute to absolute magnitude of difference would increase with in-
the value that is measured transcutaneously. Both fac- creasing PCO2.
tors are highly influenced by cardiac output and general One problem with the long-term measurement of PCO2
circulation.28 throughout the night, as in our study, might be that the
There is a second reason for the difference between PtcCO2 could increase because of heating the skin, thereby
transcutaneous and capillary PCO2 in our study population: increasing metabolism and production of CO2 (sensor tem-
we examined only subjects with hypercapnia during sleep. perature of the Tosca is 42°C). Occasionally a drift of

102 RESPIRATORY CARE • JANUARY 2016 VOL 61 NO 1

 COMPARISON OF TRANSCUTANEOUS AND CAPILLARY PCO2

Table 2. Correlation Between PtcCO2 and Arterial (Capillary) PCO2, Bias, and Limits of Agreement for Different Capnographs According to
Available Published Data

Subjects Capillary/ d SD Limits of Agreement

Capnograph (n) Study Population Arterial PCO2 r (mm Hg) (mm Hg) d – 2 SD d ⫹ 2 SD First Author
(mm Hg)
Kontron Sensor 51 28 Healthy volunteers NA 0.97 NA NA NA NA Pilsbury30
Kontron 7640 5 Ventilated brain-dead NA 0.97 NA NA NA NA Hanly31
subjects
Sensormedics 17 Subjects undergoing 50.2–50.8 0.67 10.5 15.1 –19.7 40.7 Sanders9
polysomnography
Sensormedics transcutaneous 22 Subjects under general 28–62 0.92 NA NA NA NA Reid32
CO2 monitor anesthesia
Sensormedics transcutaneous 47 Adult ICU subjects NA 0.93 NA NA NA NA Mahutte33
gas system
Sensormedics Transcend 30 Ventilated adult ICU NA NA –0.5 5.2 –11.0 9.9 Blanchette34
PO2-PCO2 monitor subjects
Sensormedics Fastrac 26 Hemodynamic stable 26–71 0.96 0.75 2.6 –4.5 6 Janssens, 199817
subjects requiring
intra-arterial
radial catheter
Tina TCM–3 9 Ergometer exercise ⫾ NA 0.84 NA NA NA NA Hoffmann35
external CO2 loading
Tina TCM–3 20 Stable respiratory out- 19–53 0.90 NA NA NA NA Kesten29
patient subjects
Tina TCM–3 20 10 postoperative subjects, 37.7 ⫾ 51.9 0.87 7.5 5.6 –3.8 18.7 Healey36
10 weaning subjects
Tina TCM–3 24 Subjects with breathlessness 27–57 NA 0.2 1.0–1.7 –1.7 2.2 Sridhar37
performing bicycle
ergometry
Tina TCM–3 251 Neonates, children with NA 0.93 1.3 3.9 –6.5 9.1 Palmisano38
cystic fibrosis, subjects
under general anesthesia
Tosca 29 Healthy subjects during 37.0 ⫾ 5.1 0.5 NA NA NA NA Randerath21
sleep
Tosca 29 Healthy subjects during 37.0 ⫾ 5.6 0.72 NA NA NA NA Randerath21
sleep
Tosca 18 Critically ill subjects NA 0.88 NA NA 3⫾7 Senn14

Manufacturers are as follows: Kontron (St Quentin En Yvelines, France), Sensormedics (Yorba Linda, California), Tosca 500 and Tina TCM–3, Radiometer (Copenhagen, Denmark).
r ⫽ correlation
d ⫽ bias
NA ⫽ not available

PtcCO2 values was recorded, so that re-calibration of the evaluation for suspected obstructive sleep apnea or noc-
systems during the night was required.30 In this respect, we turnal hypoventilation. Each subject received an indwell-
found no significant drift throughout the night by Tosca in ing arterial catheter for blood sampling during sleep. Nine-
our study. teen subjects were spontaneously breathing room air, 13
One of the questions answered in our study was whether received supplemental oxygen via nasal cannula, and 22
disturbances by the process of sampling capillary blood in were ventilated (mostly bi-level). Only 17 subjects of the
a patient during sleep could be detected. In our study, we last group were analyzed regarding the transcutaneous mea-
found no statistically significant disturbances incurred by surement. Three different capnographs (from Novametrics
the sampling process. Medical Systems [Wallingford, Connecticut], Radiometer
Thus, we found a good but not perfect agreement be- [Copenhagen, Denmark], and Sensormedics Corporation
tween PtcCO2 and PcapCO2. What is the clinical implication? [Yorba Linda, California]) were evaluated. In that study,
Should PtcCO2 or PcapCO2 be preferred in clinical routine to no agreement could be found between either PETCO2 or
diagnose hypercapnia? In fact, accuracy of PCO2 measure- PtcCO2 in relation to PaCO2: the average PaCO2 was elevated
ment is not the same as accuracy of detecting hypercapnia, to approximately 50 ⫾ 10 mm Hg, and only the capno-
although there are only a few studies examining this is- graph of Radiometer showed a minor difference
sue.39 There are few studies examining the agreement of of ⫹0.14 mm Hg. However, the error was ⫾ 10.9 mm Hg,
PtcCO2 and PaCO2 in patients during sleep. Sanders9 exam- therefore being wide and having an off-setting scatter.
ined the accuracy of end-tidal carbon dioxide tension as Therefore, the authors concluded that there is no agree-
well as PtcCO2 in subjects undergoing polysomnographic ment between transcutaneous and arterial PCO2. In another

RESPIRATORY CARE • JANUARY 2016 VOL 61 NO 1 103

 COMPARISON OF TRANSCUTANEOUS AND CAPILLARY PCO2

study, 29 healthy volunteers21 were evaluated via Tosca. ACKNOWLEDGMENTS

The conclusion by the authors was that the investigated We thank Karin Steinraths for data collection and help in the transcuta-
systems enabled stable measurement of PtcCO2 without any neous measurement of PCO2. We also thank Norbert Anduleit for help in
drift. Nevertheless the correlation at 2 different points of statistical analysis.
time were only 0.5 and 0.72. Hanly et al31 examined trans-
cutaneous measurement of PCO2 in 16 subjects with con- REFERENCES
gestive heart failure and found that congestive heart failure 1. Gold W. Pulmonary function testing. In: Murray J, Nadel J, editors.
subjects with daytime hypocapnia are more likely to de- Textbook of respiratory medicine; Vol. 1. Philadelphia: Elsevier,
velop Cheyne Stokes respiration during sleep. However, 2000;671-740.
2. Giner J, Casan P, Belda J, González M, Miralda RM, Sanchis J. Pain
the subjects did not have arterial blood gas sampling at the during arterial puncture. Chest 1996;110(6):1443-1445.
time of sleep study. Estimation of PaCO2 was validated in 3. Pitkin AD, Roberts CM, Wedzicha JA. Arterialised earlobe blood
5 brain-dead subjects. In this case, they found an excellent gas analysis: an underused technique. Thorax 1994;49(4):364-366.
correlation of 0.97. 4. Zavorsky GS, Cao J, Mayo NE, Gabbay R, Murias JM. Arterial
versus capillary blood gases: a meta-analysis. Respir Physiol Neu-
Johnson et al18 evaluated PtcCO2 in a specialized wean- robiol 2007;155(3):268-279.
ing unit in tracheostomized subjects with prolonged wean- 5. Whitaker DK, Brattebo G, Smith AF, Staender SE. The Helsinki
ing failure during daytime spontaneous breathing trials, Declaration on Patient Safety in Anaesthesiology: putting words into
the first nights off the ventilator and during bronchoscopy, practice. Best Practice Res Clin Anaesthesiol 2011;25(2):277-290.
6. Hinkelbein J, Floss F, Denz C, Krieter H. Accuracy and precision of
and compared the results with simultaneous measurement three different methods to determine Pco2 (Paco2 vs. Petco2 vs.
of arterial PCO2 and PETCO2. In this study, PaCO2 was closely Ptcco2) during interhospital ground transport of critically ill and
matched to PtcCO2 (mean ⫾ SD difference of 0.5 ⫾ ventilated adults. J Trauma 2008;65(1):10-18.
4.1 mm Hg). There was a greater difference between PaCO2 7. Wahba RW, Tessler MJ. Misleading end-tidal CO2 tensions. Can J
Anaesth 1996;43(8):862-866.
and PETCO2 (3.7 ⫾ 7.7 mm Hg during prolonged exhala- 8. Belpomme V, Ricard-Hibon A, Devoir C, Dileseigres S, Devaud
tion, and 6.8 ⫾ 7.2 mm Hg during tidal breathing).18 ML, Chollet C, Marty J. Correlation of arterial PCO2 and PETCO2
Is there any recommendation for one of the capnographs in prehospital controlled ventilation. Am J Emerg Med 2005;23(7):
852-859.
in published data relating to accuracy? Janssens et al17
9. Sanders MH, Kern NB, Costantino JP, Stiller RA, Strollo PJ Jr,
overviewed different capnographs. We updated the over- Studnicki KA, et al. Accuracy of end-tidal and transcutaneous PCO2
view and added the average PCO2 and the study population monitoring during sleep. Chest 1994;106(2):472-483.
of the published studies (Table 2). The study populations 10. Sleep-related breathing disorders in adults: recommendations for syn-
drome definition and measurement techniques in clinical research.
are extremely heterogeneous, ranging from ventilated brain
The Report of an American Academy of Sleep Medicine Task Force.
dead subjects to healthy volunteers with ergometer exer- Sleep 1999;22(5):667-689.
cise. One must also consider that, in the different studies, 11. Weiss M, Frei M, Buehrer S, Feurer R, Goitein G, Timmermann B.
not only were different capnographs analyzed, but also Deep propofol sedation for vacuum-assisted bite-block immobiliza-
tion in children undergoing proton radiation therapy of cranial tu-
different blood gas analyzers, which themselves have cer-
mors. Paediatr Anaesth 2007;17(9):867-873.
tain variations and errors in analyzing PcapCO2 or PaCO2. 12. Duron B, Marlot D. Intercostal and diaphragmatic electrical activity
Our study has some limitations that must be acknowl- during wakefulness and sleep in normal unrestrained adult cats. Sleep
edged. First, the number of subjects is limited. We tried to 1980;3(3-4):269-280.
13. Casey KR, Cantillo KO, Brown LK. Sleep-related hypoventilation/
improve the sample size by measuring each subject twice.
hypoxemic syndromes. Chest 2007;131(6):1936-1948.
Second, we did not measure arterial PCO2 additionally. The 14. Senn O, Clarenbach CF, Kaplan V, Maggiorini M, Bloch KE. Mon-
conclusion that arterial and capillary PCO2 are equal is itoring carbon dioxide tension and arterial oxygen saturation by a
drawn by other studies. Finally, in method-comparison single earlobe sensor in patients with critical illness or sleep apnea.
Chest 2005;128(3):1291-1296.
methodology, the methods to be compared need to mea-
15. Bendjelid K, Schütz N, Stotz M, Gerard I, Suter PM, Romand JA.
sure the same variable, which is not the case with trans- Transcutaneous PCO2 monitoring in critically ill adults: clinical eval-
cutaneous measuring of PCO2 where metabolic processes uation of a new sensor. Crit Care Med 2005;33(10):2203-2206.
of the earlobe contribute to the value that is measured 16. Tatevossian RG, Wo CC, Velmahos GC, Demetriades D, Shoemaker
WC. Transcutaneous oxygen and CO2 as early warning of tissue
transcutaneously.40
hypoxia and hemodynamic shock in critically ill emergency patients.
Crit Care Med 2000;28(7):2248-2253.
17. Janssens JP, Howarth-Frey C, Chevrolet JC, Abajo B, Rochat T.
Conclusions
Transcutaneous PCO2 to monitor noninvasive mechanical ventila-
tion in adults: assessment of a new transcutaneous PCO2 device.
Chest 1998;113(3):768-773.
Transcutaneous CO2 monitoring reliably reproduced 18. Johnson DC, Batool S, Dalbec R. Transcutaneous carbon dioxide
capillary PCO2 measurements in normocapnic and hyper- pressure monitoring in a specialized weaning unit. Respir Care 2008;
capnic subjects. 53(8):1042-1047.

104 RESPIRATORY CARE • JANUARY 2016 VOL 61 NO 1

 COMPARISON OF TRANSCUTANEOUS AND CAPILLARY PCO2

19. Bernet-Buettiker V, Ugarte MJ, Frey B, Hug MI, Baenziger O, Weiss 30. Pilsbury D, Hibbert G. An ambulatory system for long-term contin-
M. Evaluation of a new combined transcutaneous measurement of uous monitoring of transcutaneous PCO2. Bull Eur Physiopathol
PCO2/pulse oximetry oxygen saturation ear sensor in newborn pa- Respir 1987;23(1):9-13.
tients. Pediatrics 2005;115(1):e64-e68. 31. Hanly P, Zuberi N, Gray R. Pathogenesis of Cheyne-Stokes respi-
20. Chhajed PN, Rajasekaran R, Kaegi B, Chhajed TP, Pflimlin E, Le- ration in patients with congestive heart failure: relationship to arterial
uppi J, Tamm M. Measurement of combined oximetry and cutaneous PCO2. Chest 1993;104(4):1079-1084.
capnography during flexible bronchoscopy. Eur Respir J 2006;28(2): 32. Reid CW, Martineau RJ, Miller DR, Sullivan PJ, Hull KA, Samson
386-390. B. A comparison of end-tidal and transcutaneous PCO2 measure-
21. Randerath WJ, Stieglitz S, Galetke W, Anduleit N, Treml M, Schäfer ments during anaesthesia. Can J Anaesth 1990;37(4 Pt 2):S89.
T. Evaluation of a system for transcutaneous long-term capnometry. 33. Mahutte CK, Michiels TM, Hassell KT, Trueblood DM. Evaluation
Respiration 2010;80(2):139-145. of a single transcutaneous PO2-PCO2 sensor in adult patients. Crit
22. Eberhard P. The design, use, and results of transcutaneous carbon Care Med 1984;12(12):1063-1066.
dioxide analysis: current and future directions. Anesth Analg 2007; 34. Blanchette T, Dzidzio J. Transcoutaneous PCO2 and end-tidal PCO2
in ventilated adults. Respir Care 1992;37:240-248.
105(6 Suppl):S48-S52.
35. Hoffmann U, Essfeld D, Stegemann J. Comparison of arterial, end-
23. Severinghaus J. Transcutaneous PO2-PCO2 electrode with electro-
tidal and transcutaneous PCO2 during moderate exercise and exter-
chemical HCO3 -stabilization. J Appl Physiol 1981;51(4):1027-1032.
nal CO2 loading in humans. Eur J Appl Physiol Occup Physiol
24. Bland JM, Altman DG. Agreed statistics: measurement method com-
1990;61(1-2):1-4.
parison. Anesthesiology 2012;116(1):182-185.
36. Healey CJ, Fedullo AJ, Swinburne AJ, Wahl GW. Comparison of
25. Severinghaus JW, Stafford M, Bradley AF. tcPCO2 electrode de-
noninvasive measurements of carbon dioxide tension during with-
sign, calibration and temperature gradient problems. Acta Anaesthe- drawal from mechanical ventilation. Crit Care Med 1987;15(8):764-
siol Scand Suppl 1978;68:118-122. 768.
26. Grønlund J. Evaluation of factors affecting relationship between trans- 37. Sridhar MK, Carter R, Moran F, Banham SW. Use of a combined
cutaneous PO2 and probe temperature. J Appl Physiol 1985;59(4): oxygen and carbon dioxide transcutaneous electrode in the estima-
1117-1127. tion of gas exchange during exercise. Thorax 1993;48(6):643-647.
27. Waxman K, Annas C, Daughters K, Tominaga GT, Scannell G. A 38. Palmisano BW, Severinghaus JW. Transcutaneous PCO2 and PO2:
method to determine the adequacy of resuscitation using tissue ox- a multicenter study of accuracy. J Clin Monit 1990;6(3):189-195.
ygen monitoring. J Trauma 1994;36(6):852-856; discussion 856-858. 39. Rodríguez DA, Jover L, Drakulovic MB, Gómez FP, Roca J, Albert
28. Kagawa S, Severinghaus JW. Errors in monitoring transcutaneous Barberà J, et al. Below what FEV1 should arterial blood be routinely
PCO2 on the ear. Crit Care Med 2005;33(10):2414-2415. taken to detect chronic respiratory failure in COPD? Arch Bronco-
29. Kesten S, Chapman KR, Rebuck AS. Response characteristics of a neumol 2011;47(7):325-329.
dual transcutaneous oxygen/carbon dioxide monitoring system. Chest 40. Hanneman SK. Design, analysis, and interpretation of method-com-
1991;99(5):1211-1215. parison studies. AACN Adv Crit Care 2008;19(2):223-234.

RESPIRATORY CARE • JANUARY 2016 VOL 61 NO 1 105