Monaldi Arch Chest Dis
2008; 69: 3, 114-118
Carbon dioxide rebreathing
in non-invasive ventilation.
Analysis of masks, expiratory ports
and ventilatory modes
D. Samolski, N. Calaf, R. Güell, P. Casan, A. Antón
Portable ventilators used in non-invasive ven-
tilation (NIV) may provoke carbon dioxide (CO2)
rebreathing due to the use of a single-limb circuit
. The mode of ventilation, the type of mask, the
expiratory device and the expiratory pressure level
setting may also play a key role in this event [2-6].
Several masks currently used in positive pres-
sure ventilation have exhaust vents (EV) that func-
tion as an expiratory port. In addition, some au-
thors [2, 7] have shown that specific anti-rebreath-
ing expiratory devices (ARD) minimise or prevent
rebreathing when used in conjunction with her-
metic masks. To our knowledge, no studies have
been conducted to determine whether the newly
designed EV equipped masks act in a similar way
when used in combination with portable pressure-
While using volume-cycled ventilation, exha-
lation is allowed by a pneumatic valve that, if po-
sitioned next to the mask, is uncomfortable be-
cause of its size and form. It is therefore necessary
to add a flexible cylindrical spacer of variable size
to separate the mask from the valve. Such a struc-
ture may interfere with the expiratory phase and
generate a dead space that could potentially cause
rebreathing. No studies have evaluated this issue.
The aim of the present study was to assess the
ability of EV equipped nasal and facial masks to
prevent rebreathing while using a portable pres-
sure-cycled ventilator, and to quantify the mini-
mum EPAP level needed to do this. Using volume-
cycled ventilation, we also aimed to evaluate the
maximal volume of a cylindrical spacer still effec-
tive in preventing rebreathing.
Materials and methods
The study was conducted in the pulmonary
function laboratory of our hospital. All volunteers
provided written consent prior to participate in this
study. Moreover, the project was approved by the
Hospital Ethics Committee.
Five healthy volunteers were evaluated. None of
them had a prior history of smoking or respiratory
diseases or were using drugs at the time of the study.
– Pressure-cycled ventilation: a VPAP III
ventilator (ResMed, North Ryde, Australia) with
Keywords: CO2rebreathing, Nasal and facial masks, Expiratory devices, Pressure and volume-cycled ventilators.
Respiratory Dept. Hospital de la Santa Creu i Sant Pau, Barcelona, Spain.
Supported in part by RTIC-ISCiii - Red Respira and ALAT-SEPAR grants.
Correspondence: Daniel Samolski M.D., Respiratory Dept., Hospital de la Santa Creu i Sant Pau, Sant Antoni Mª Claret 167,
08025 Barcelona, Spain; e-mail: email@example.com
ABSTRACT: Carbon dioxide rebreathing in non-invasive
ventilation. Analysis of masks, expiratory ports and ventilatory
modes. D. Samolski, N. Calaf, R. Güell, P. Casan, A. Antón.
Background and Aim. Carbon dioxide (CO2) re-
breathing is a complication of non-invasive ventilation
(NIV). Our objectives were to evaluate the ability of masks
with exhaust vents (EV) to avoid rebreathing while using
positive pressure (PP) NIV with different levels of expira-
tory pressure (EPAP). Concerning volume-cycled NIV, we
aimed to determine whether cylindrical spacers located in
the circuit generate rebreathing.
Materials and methods. 5 healthy volunteers were eval-
uated. Bi-level PP was used with 3 nasal and 2 facial masks
with and without EV. Spacers of increasing volume at-
tached to nasal hermetic masks were evaluated with vol-
ume NIV. Inspired CO2fraction was analyzed.
Results. Rebreathing was zero with all nasal masks
and EPAP levels. Using facial masks 1 volunteer showed
rebreathing. There was no rebreathing while using all the
Conclusions. In healthy volunteers, nasal and facial
masks with EV prevent rebreathing. In addition, the use
of spacers did not generate this undesirable phenomenon.
Monaldi Arch Chest Dis 2008; 69: 3, 114-118.
CO2REBREATHING IN NON-INVASIVE VENTILATION
its own single-limb circuit was used for the study.
Three different nasal masks and two facial masks
were evaluated. Two nasal masks had an EV (Ul-
traMirage and Mirage, ResMed, North Ryde, Aus-
tralia). The one out of EV (UltraMirage, ResMed,
North Ryde, Australia) was connected to an ARD
(Plateau valve, Respironics, Murrysille, PA, USA).
Fig. 1. One facial mask (UltraMirage, ResMed,
North Ryde, Australia) was EV equipped, and the
other (Mirage, ResMed, North Ryde, Australia),
without EV, was connected to the ARD. Using
sealed envelopes each containing the specification
for a single mask and pressure level combination
among those to be tested, volunteers were random-
ly crossed over to receive NIV sessions with the
type of mask and pressure selected. We set a fixed
inspiratory pressure (IPAP) of 14 cmH2O and an
increasing expiratory pressure (EPAP) of 4, 6, 8
– Volume-cycled ventilation: we used a Breas
PV 501 ventilator (Breas Medical, Swedish) with
its own single-limb circuit and expiratory valve.
An hermetic nasal mask was employed (UltraMi-
rage without EV, ResMed, North Ryde, Australia).
The cylindrical spacers located between the mask
and the expiratory valve had three different vol-
umes (43, 85 and 176 ml.) that were selected in
random order using sealed envelopes each con-
taining the specification for a volume to be tested.
All volunteers underwent NIV sessions receiving a
tidal volume of 8-10 ml/kg, a respiratory rate of
16-18 cycles/minute and an I/E relation of 1:2. Fig. 2.
The inspired CO2fraction (FiCO2) was contin-
uously analysed using a thin sampling tube located
in the mask, close to the nostrils. This tube was
connected to a paramagnetic CO2 analyser (NDIR)
with a 0-10% response lower than 130 millisec-
onds and a precision of +/- 0.05% (Medical graph-
ics system, St. Paul, Minnesota, USA). Fig. 3. The
paramagnetic CO2 analyser was calibrated using
atmospheric air and a calibration gas mixture of
CO2(6.87%), O2(50.3%) and N2(42.83%) (Abel-
lo Linde SA, Barcelona, Spain). The gain and the
zero levels were checked prior to each change in
the pressure level, the type of mask or the spacer’s
volume. Moreover, we performed the calibration
before carrying out the study in each volunteer.
Volunteers were seated in a semi-recumbent po-
sition and each mode of ventila-
tion was evaluated for five min-
utes. They were advised to keep
their mouths closed while using
nasal masks to avoid leaks
through it. A wash-out period
consisting of 15 minutes of room
air breathing was performed be-
tween each NIV session.
The FiCO2level was mea-
sured in 15 respiratory cycles
after adaptation to each level of
ventilation. The average FiCO2
level from these breaths was
then considered for further
analysis. According to the study
design, differences of the mean
FiCO2 level in all the ventilato-
ry conditions tested were
planned to be evaluated with a
Friedman 2 way ANOVA non
parametric test since low FiCO2
values, but different from zero,
were expected before the study
completion. Differences will be
considered statistically signifi-
cant when p < 0.05.
Ventilation was well tolerat-
ed in all volunteers regardless
of the mode of ventilation, the
interface or the volume of the
With reference to pressure-
cycled ventilators, as shown in
Fig. 4, FiCO2was zero while
Fig. 2. - Volume-cycled ventilator and cylindrical spacer between the mask and the expiratory valve.
Fig. 1. - Pressure-cycled ventilator and nasal mask with different expiratory ports.
D. SAMOLSKI ET AL.
using all the nasal masks tested and their corre-
sponding expiration ports. An EPAP as low as 4
cmH2O was useful to prevent rebreathing in all the
As Fig. 5 shows, FiCO2was zero while using
the EV equipped facial mask, whatever the EPAP
used. The hermetic mask connected to an ARD
showed no rebreathing in all except one volunteer.
In this individual, the FiCO2 was 0.2% and it was
not modified with the increase in EPAP.
While using volume-cycled ventilators, re-
breathing was zero for all the volumes of spacers
studied, as seen in Fig. 6.
This study shows that newly designed nasal
and facial masks with EV prevent CO2rebreath-
ing, even with an EPAP as low as 4 cmH2O in nor-
mal subjects. Likewise, a cylindrical spacer locat-
ed between the mask and the expiratory valve did
not generate rebreathing during volume ventila-
The phenomenon of rebreathing is generally
considered a potential cause of failure of NIV
treatment . Since the first description by Fergu-
son et al  in 1995, little has been published on
this subject. It is well known that the use of a sin-
gle-limb circuit is its basic causal mechanism and
that it can be modified by the mode of ventilation
[2, 3] and the type of mask or expiratory port [4, 5,
9-11]. The use of positive pressure during the ex-
piratory phase could help with the “wash out of
CO2” through the holes of the mask. As a matter of
fact, an appropriate EPAP level has been claimed
to be critical in preventing rebreathing [2, 11].
Ferguson et al , using a specific ARD,
found that an EPAP level of 6-8 cmH2O was nec-
essary to prevent rebreathing. In our work, using a
mask with EV, we were able to reduce the EPAP to
4 cmH2O without rebreathing. Our results are in
agreement with those described by Schettino et al
 who also used masks with EV. This improve-
ment in preventing CO2rebreathing could be re-
lated to improve CO2 kinetics with this type of
masks with EV . Another important aspect to
Fig. 3. - Continuous analysis of CO2inside the mask.
Horizontal arrow: end tidal CO2(ETCO2mmHg). Vertical arrow: inspired CO2fraction (FiCO2%).
Fig. 4. - Inspired CO2fraction (FiCO2) during nasal mask ventilatory assistance with increasing levels of EPAP.
Top: nasal mask (UltraMirage, ResMed, Australia) with exhaust vent (EV). Middle: nasal mask (Mirage, ResMed, Australia) with EV.
Bottom: Hermetic nasal mask (UltraMirage without EV, ResMed, Australia) joined to an anti-rebreathing expiratory port (Respironics, USA).
Data obtained from volunteer #5.
CO2REBREATHING IN NON-INVASIVE VENTILATION
take into consideration is that the expiratory de-
vices could worsen expiratory resistance. This sit-
uation was well described by Lofaso et al  who
suggested that this type of expiratory plateau port
could increase the positive end-expiratory pressure
(PEEP) due to its higher expiratory resistance.
Hence, the increase in expiratory resistance should
be taken into account when using ARDs.
Schettino et al  evaluated the effects of
mask volume in rebreathing. These authors report-
ed, as was expected, that the volume of the mask
was associated with CO2rebreathing. Saatci et al
 found that dead space is higher with the facial
mask but that it could be partially offset using ap-
propriate levels of positive pressure or adding an
EV to enhance mask ventilation. In our work, on-
ly one volunteer showed an increase in FiCO2
while using a hermetic facial mask connected to an
ARD. This phenomenon was not modified even
when the EPAP level was increased. On the other
hand, we did not observe rebreathing with the fa-
cial mask with EV, even with EPAP level as low as
The addition of a spacer with an internal vol-
ume up to 176 ml, acting as dead volume, did not
generate rebreathing. This finding suggests that
pneumatic valves are effective in preventing re-
breathing. On the other hand, we cannot rule out
the possibility that a greater volume of the spacers
could generate rebreathing. However, with a spac-
er as large as 176 ml. (the highest volume in our
study), the mask can be comfortably separate
from the expiratory valve. Thus, longer spacers
with higher internal volume seem unnecessary. To
our knowledge, this is the first study addressing
Reports in literature about the clinical signifi-
cance of rebreathing are few and contradictory.
Comparing an ARD with a standard valve, Hill et
al  reported no clinical or gasometric differ-
ences in patients undergoing chronic NIV. By con-
trast, Farré et al  suggested that rebreathing
could be so deleterious to lead to CPAP failure, an
event that could be prevented by using an anti-re-
Our study in healthy volunteers show that the
masks tested had little or no CO2rebreathing.
However, the “dead space effect” of these masks
may negatively affect already hypercapnic patients
who show a rapid and shallow breathing. For this
reason, further studies on this kind of patients are
warranted to assess the CO2wash-out effective-
ness of newly designed masks during assisted ven-
Fig. 5. - Inspired CO2fraction (FiCO2) during facial mask ventilatory assistance with increasing levels of EPAP.
Top: Facial mask with EV (UltraMirage, ResMed, Australia). Bottom: Hermetic facial mask (Mirage, ResMed, Australia) joined to an anti-
rebreathing expiratory port.
Data obtained from volunteer #3.
Fig. 6. - Inspired CO2fraction (FiCO2) inside the mask while using flexible cylindrical spacers of increasing volume.
Data obtained from volunteer #2.
D. SAMOLSKI ET AL.
Technical remarks and limitations of the study
Rebreathing can be detected by measuring
CO2inside the mask during the ventilatory cycle.
It is not clear how or at what moment CO2should
be measured to quantify rebreathing. Some au-
thors suggest that end-tidal of CO2 (ETCO2) is the
best parameter to evaluate this phenomenon [5, 7,
11]. Others consider that FiCO2is better than ET-
CO2 measurement [2, 9]. In our study, we
analysed the CO2 present inside the mask at the
beginning of the inspiration (proto-inspiration)
because, in our opinion, it better reflects the
amount of CO2rebreathing, as this is the real
quantity of this gas that the patient will inspire.
Likewise, it is logical to suppose that a greater
ETCO2 could generate rebreathing. However, this
depends not only on the amount of expired CO2
but also on the ability of the expiratory port to
eliminate it. In view of these considerations, we
took into account the FiCO2and not the ETCO2to
evaluate the rebreathing.
Secondly, we analysed only a few of the masks
that are currently available to make the study fea-
sible. Moreover, it could be supposed that other
similarly designed interfaces would yield similar
Thirdly, an open ventilatory system, like the
one used in this work, did not allow us to quantify
exactly the ventilatory mask flows; consequently,
the CO2rebreathing was showed as a percentage
and not as a real amount.
Fourthly, we do not use any dyspnoea or toler-
ance scales to assess objectively mask comfort. We
deduced that ventilation was well tolerated in all vol-
unteers from the fact that all of them could fulfill and
complete the whole study without reporting signifi-
cant troubles or complaints with each mask or expi-
ratory port, the mode of ventilation and the spacer.
Finally, with reference to statistical analysis,
the absolute values obtained were not adequate for
applying the planned statistical analysis. FiCO2
amounting almost invariably to zero. However, the
repetition of this result strongly supports our con-
clusions. Thus, we decided to show only raw data
with no more analysis.
In conclusion, we show that in healthy volun-
teers CO2rebreathing is not a common issue in
newly designed masks with incorporated EV even
when EPAP is low or when cylindrical spacers are
added to the circuit during volume-cycled ventila-
tion. Our results should be contrasted in future
with a clinical trial in hypercapnic patients.
Acknowledgements: The authors thank Carolyn Newey
and Daniela Bordet for their assistance in editing the manuscript.
Also they thank Marina Khoury for her assistance in data
1.Hill NS. Problems, remedies, and strategies to optimize
the success of noninvasive ventilation. In: Hill NS, ed-
itor. Non invasive positive pressure ventilation. Ar-
monk NY: Ed. Futura publishing company 2001: 187-
Ferguson GT, Gilmartin M. CO2rebreathing during Bi-
PAP ventilatory assistance. Am J Respir Crit Care Med
1995; 151: 1126-1135.
Lofaso F, Brochard L, Touchard D, Hang T, Harf A, Is-
abey D. Evaluation of carbon dioxide rebreathing dur-
ing pressure support ventilation with airway manage-
ment system (BiPAP) devices. Chest 1995; 108:772-
Zhang X ChR, He G. Modification of facial mask on
the dead space effect in non invasive mask ventilation.
Zhonghua Jie He He Hu Xi Za Zhi 2000; 23: 734-736.
Schettino GP, Chatmongkolchart S, Hess DR, Kac-
marek R. Position of exalation port and mask design af-
fect CO2rebreathing during non-invasive positive pres-
sure ventilation. Crit Care Med 2003; 31: 2178-82.
Farre R, Montserrat JM, Ballester E, Navajas D. Poten-
tial rebreathing after CPAP failure during sleep. Chest
2002; 121: 196-200.
Hill NS, Carlisle C, Kramer N. Effect of a nonrebreath-
ing exhalation valve on long term nasal ventilation us-
ing a bilevel device. Chest 2002; 122: 84-91.
Saatci E, Miller DM, Stell IM,Lee KC, Moxham J. Dy-
namic dead space in face masks used with non-invasive
ventilators: a lung model study. Eur Respir J 2004; 23:
Taccone P, Hess D, Caironi P, Bigatello LM. Continu-
ous positive airway pressure delivered with a helmet.
Effects on carbon dioxide rebreathing. Crit Care Med
2004; 32: 2090-2096.
Racca F, Appendini L, Gregoretti C. Effectiveness of
mask and helmet interfaces to deliver noninvasive ven-
tilation in human model of resistive breathing. J Appl
Physiol 2005; 99: 1262-1271.
Mediano O, Garcia Rio F, Villasante C. Comparacion
de la reinhalación de anhídrido carbónico originada por
3 mascarillas durante la aplicacion de CPAP. Arch
Bronconeumol 2006; 42: 189-93.