Content uploaded by Nicholas Gant
Author content
All content in this area was uploaded by Nicholas Gant on Mar 26, 2015
Content may be subject to copyright.
Diving and Hyperbaric Medicine Volume 45 No. 1 March 2015
16
The ve-minute prebreathe in evaluating carbon dioxide absorption
in a closed-circuit rebreather: a randomized single-blind study
Carolyn Deng, Neal W Pollock, Nicholas Gant, Jacqueline A Hannam, Adam Dooley,
Peter Mesley and Simon J Mitchell
Abstract
(Deng C, Pollock NW, Gant N, Hannam JA, Dooley A, Mesley P, Mitchell SJ. The five-minute prebreathe in evaluating
carbon dioxide absorption in a closed-circuit rebreather: a randomized single-blind study. Diving and Hyperbaric Medicine.
2015 March;45(1):16-24.)
Introduction: Closed-circuit underwater rebreather apparatus (CCR) recycles expired gas through a carbon dioxide
(CO2) ‘scrubber’. Prior to diving, users perform a five-minute ‘prebreathe’ during which they self-check for symptoms of
hypercapnia that might indicate a failure in the scrubber. There is doubt that this strategy is valid.
Methods: Thirty divers were block-randomized to breathe for five minutes on a circuit in two of the following three
conditions: normal scrubber, partly-failed scrubber, and absent scrubber. Subjects were blind to trial allocation and instructed
to terminate the prebreathe on suspicion of hypercapnia.
Results: Early termination was seen in 0/20, 2/20, and 15/20 of the normal, partly-failed, and absent absorber conditions,
respectively. Subjects in the absent group experienced a steady, uncontrolled rise in inspired (PICO2) and end-tidal CO2
(PETCO2). Seven subjects exhibited little or no increase in minute volume yet reported dyspnoea at termination, suggesting a
biochemically-mediated stimulus to terminate. This was consistent with results in the partly-failed condition (which resulted
in a plateaued mean PICO2 near 20 mmHg), where a small increase in ventilation typically compensated for the inspired CO2
increase. Consequently, mean PETCO2 did not change and in the absence of a hypercapnic biochemical stimulus, subjects
were very insensitive to this condition.
Conclusions: While prebreathes are useful to evaluate other primary functions, the five-minute prebreathe is insensitive for
CO2 scrubber faults in a rebreather. Partly-failed conditions are dangerous because most will not be detected at the surface,
even though they may become very important at depth.
Key words
Scuba diving, rebreathers/closed circuit, carbon dioxide, hypercapnia, rebreathing, capnography, physiology
Introduction
Closed-circuit rebreathers (CCRs) are popular in advanced
recreational diving owing to advantages such as the
minimization of gas consumption, especially during deep
diving, and optimization of decompression. Rebreathers
recycle expired gas around a circle circuit with one-way
valves. Expired carbon dioxide (CO2) is removed as it passes
through a ‘scrubber’ canister containing CO2 absorbent that
is most commonly soda lime (a mixture of sodium hydroxide
and calcium hydroxide). Oxygen metabolised by the diver
is replaced in the circuit to maintain a safe inspired partial
pressure of oxygen (PIO2).
Rebreathers are more complex than open-circuit scuba
equipment and more prone to operator errors.1 Some of
these relate to the CO2 scrubber. The absorbent material
has a finite capacity (approximately 12–15 L CO2·100 g-1)
and must be changed regularly.2 Errors include failing to
replace the absorbent material in a timely manner, incorrect
packing of the absorbent material into the scrubber canister,
incorrect installation of the canister in the rebreather and,
rarely, forgetting to install it entirely. Such errors may
allow expired CO2 to enter the inhaled gas which may in
turn cause symptomatic hypercapnia (often referred to by
divers as CO2 toxicity). There have been deaths during the
use of rebreathers in which hypercapnia is thought to have
contributed, one of which is comprehensively documented
in the medical literature.3 Hypercapnia also enhances the
toxicity of oxygen4,5 and the narcotic effect of nitrogen6
breathed at higher partial pressures.
Most rebreather units do not measure inspired CO2, so most
technical diver training agencies teach divers to conduct a
five-minute ‘prebreathe’ as a means of checking scrubber
function before entering the water. A prebreathe involves
preparing the unit for diving, and then sitting quietly
breathing on the circuit, ideally with the nose blocked. If
the CO2 scrubber is absent or faulty, the diver will re-inhale
expired CO2 and, in theory, should notice the early symptoms
of hypercapnia such as dyspnoea and/or headache. The
five-minute duration is assumed to be sufficiently long for
early symptoms of hypercapnia to reliably manifest, but
the validity of this practice has not been formally tested.
Therefore, we measured the proportion of blinded subjects
who could discern an absent or faulty CO2 scrubber during
a five-minute prebreathe test on a rebreather circuit. A
secondary aim was to derive a physiological interpretation
of the results.
Diving and Hyperbaric Medicine Volume 44 No.1 March 2015 17
Methods
TRIAL DESIGN AND PARTICIPANTS
This was a randomised, single-blind, controlled trial that took
place at the Exercise Metabolism Laboratory, University of
Auckland, in July 2014. The study protocol was approved
by the University of Auckland Human Participants Ethics
Committee (reference 012315).
The subjects were trained, certified and active adult divers.
Preference was given to rebreather divers, but experienced
open-circuit scuba divers were not excluded as they would
be taught the same prebreathe technique and expected to use
it if undertaking a rebreather training course. All subjects
received a participant information sheet, a verbal briefing
and provided written informed consent.
EXPERIMENTAL CONDITIONS & RANDOMIZATION
Twenty prebreathe tests were conducted on a rebreather
in each of the following experimental conditions: normal
scrubber; partly-failed scrubber and absent scrubber as
described in more detail below. To achieve sufficient
numbers of trials in each condition, 30 blinded subjects were
block randomised to prebreathe in two of the three scrubber
conditions with a rest period of at least 20 minutes between
the two experiments. Subjects relaxed between trials in the
presence of study personnel to prevent them discussing
their experience until the study was complete. For each
subject, the order of conditions was constrained so that the
condition likely to result in less CO2 rebreathing was first.
This constraint was concealed from the subjects and was
necessary to prevent an obvious hypercapnia experience
on the first prebreathe from biasing perceptions of scrubber
condition on the second. Similarly, we concealed the block
randomization pattern from subjects who were told, again
to avoid biasing, that any combination of the two conditions
was possible, including breathing on a normal circuit twice.
EQUIPMENT CONFIGURATION
An Inspiration Evolution Plus rebreather (Ambient
Pressure Diving, Helston, Cornwall) was assembled by the
investigators for each prebreathe. The rebreather oxygen
cylinder contained 100% oxygen and the diluent cylinder
contained air. The rebreather oxygen controller was set
to maintain a PIO2 at 0.7 atm (71 kPa) throughout each
experiment. This is a standard setting used by rebreather
divers when at the surface.
Rebreather assembly followed the standard procedure
described by the manufacturer with several exceptions.
First, the CO2 scrubber was configured according to the
allocated condition. In the normal condition, the absorbent
canister was installed as recommended, with the soda lime
material replaced each day (after approximately 80 minutes
of a maximum recommended 180 minutes use). In the
partly-failed condition, the scrubber canister was installed,
but a known assembly error was intentionally committed: a
sealing O-ring that directs all gas flow through the canister
was omitted from the circuit, allowing some expired gas to
bypass the scrubber. In the absent condition, the absorbent
canister was completely omitted.
Second, a disposable anaesthetic circuit antibacterial
filter (Covidien DAR, MA, USA) was incorporated into
the mouthpiece of the rebreather circuit. The filter had
a dual purpose. It served to mask any changes in the
circuit breathing resistance resulting from the scrubber
condition (particularly the absent condition) by imposing
a fixed resistance at the mouth. In addition, replacement
of the mouthpiece and filter for each subject allowed use
of the same rebreather circuit for multiple subjects. In a
supplementary experiment using simple manometry, we
evaluated the efficacy of the filter in masking changes in
circuit resistance related to the scrubber condition and its
contribution to any increase in circuit resistance. With the
filter present or absent, and with the rebreather configured as
for each of the three experimental conditions, we measured
peak inspiratory and expiratory pressures (cm H2O) at the
mouthpiece with a respiratory pressure transducer (MLT844,
AD Instruments, Dunedin) during sinusoidal mechanical
ventilation (17050-2 Lung Simulator, VacuMed, Ventura,
CA) over 1 minute (tidal volume VT 1.5 L; respiratory rate
RR 10 breaths·min-1).
Third, a gas sampling line was attached to the dedicated port
of the mouthpiece filter. This allowed continuous sampling
for rapid response measurement of PIO2 with a paramagnetic
O2 analyser (S-3A, AEI Technologies, Pittsburgh, PA),
inspired CO2 (PICO2), and end-tidal CO2 (PETCO2) with
an infrared CO2 analyser (CD-3A, AEI Technologies,
Pittsburgh, PA). A three-point calibration was performed
at routine intervals for O2 and CO2 using reference gases
spanning the measurement range. A pneumotachometer
(MLT1000L, AD Instruments, Dunedin) was interposed in
the exhale limb of the rebreather circuit for measurement of
VT, RR and minute volume (VE). The device was calibrated
prior to each trial and removed from the circuit at regular
intervals for comparison with an external standard (3L
Calibration Syringe, Hans Rudolph, Shawnee, KS). For
safety, heart rate (HR) and oxygen saturation (SpO2) were
monitored using a pulse oximeter (Rad-5, Masimo, Irvine,
CA) with the audible signal silenced. All physiological
parameters were sampled at 15 second intervals. The
laboratory set-up is illustrated in Figure 1.
EXPERIMENTAL PROCEDURE
Subjects were briefed in a standardised manner prior to
their first prebreathe. They were reminded of the symptoms
of hypercapnia, and it was emphasised that this was an
experiment to determine whether the subjects could detect
Diving and Hyperbaric Medicine Volume 45 No. 1 March 2015
18
a scrubber problem if present; not to determine whether
they could tolerate hypercapnia. Accordingly, the subjects
were asked to terminate the prebreathe test as they would
in a real-world scenario if they detected relevant symptoms.
Subjects donned the rebreather in the sitting position, and
faced away from the monitoring equipment. The breathing
circuit hoses were passed over the shoulders as in normal use.
At commencement of the prebreathe period the mouthpiece
was placed in the subject’s mouth and the nose was occluded
using a nose clip, which is recommended as best practice.
Each prebreathe either continued for five minutes or was
terminated by the subject if he or she discerned symptoms
of hypercapnia. Subjects who terminated the prebreathe early
were asked to describe their symptoms.
OUTCOMES
The primary outcome was a comparison of the proportion
of subjects who detected symptoms of hypercapnia and
terminated the prebreathe in each condition. A secondary
aim was to interpret these results in the context of the
physiological data (PICO2, PETCO2, VT, RR, and VE).
POWER
We considered that 80% sensitivity for detection of a
scrubber problem in the abnormal scrubber conditions would
indicate a potentially useful test. We anticipated that under
the circumstances of the experiment, subjects might exhibit a
high index of suspicion for CO2 scrubber problems, resulting
in some false positives in the group breathing on a normal
rebreather loop. Thus, allowing for a 30% false positive
rate in the normal rebreather condition, we calculated that
to demonstrate a statistically significant difference between
terminations in the normal condition and in each abnormal
condition where the test appeared useful (80% sensitivity)
with 90% power and an alpha value of 0.05, we would need
20 subjects in each group.
STATISTICAL ANALYSIS
Descriptive data are presented as mean ± standard deviation
(SD) or median with ranges, as appropriate. The proportion
of subjects terminating the prebreathe in each condition was
calculated, and these were compared using a two-tail Fisher
exact test (GraphPad Prism ver 6.01, San Diego, CA). The
sensitivity, specificity and positive predictive values of the
prebreathe were calculated.
Results
Baseline characteristics of the groups are described in
Table 1.
PRIMARY OUTCOME
Twenty prebreathe tests were completed in each condition.
The proportion of subjects terminating the prebreathe in each
of the three conditions is shown in Table 2. The sensitivity of
the prebreathe was 10% for the detection of a partly-failed
scrubber, and 75% for detection of an absent scrubber.
The specificity of the prebreathe was 100% as there were
no false positives in the normal condition. The positive
predictive value (PPV) was 100% (albeit in a high prevalence
setting), indicating that all subjects who terminated because
of perceived symptoms of CO2 toxicity were breathing on
a loop with a faulty CO2 scrubber. The negative predictive
value was 80% for an absent scrubber and 53% for a partly-
failed scrubber.
The mean time to termination in the absent scrubber group
was 3 minutes and 41 seconds (range 2 min 1 s to 4 min
52 s). Among the 18 subjects who terminated the prebreathe,
the most frequently reported symptoms of hypercapnia were
Figure 1
Laboratory set-up; the subject breathes from the modified
rebreather whilst seated facing away from the monitors and
recording equipment
Table 1
Descriptive data for participants randomised to the three scrubber conditions; n = 20 for all three conditions
Normal scrubber Partly-failed Absent scrubber
Age (years) mean (SD) 42 (8) 44 (10) 42 (9)
Sex (M/F) 14/6 16/4 14/6
Body mass index (kg·m-2) mean (SD) 28.6 (3.2) 27.7 (3.3) 28.4 (3.7)
Years of diving median (range) 18 (3–45) 14 (1–45) 15 (1–28)
Rebreather divers 15 12 13
Diving and Hyperbaric Medicine Volume 44 No.1 March 2015 19
Table 2
Outcomes (numbers and proportion of subjects who terminated the prebreathe) for each of the three scrubber conditions;
P values are for the comparison with the normal scrubber state
Terminated Not terminated P-value
Normal 0 20
Partly failed 2 18 0.487
Absent 15 5 < 0.0001
Figure 2
Normal scrubber condition (mean ± SD); A – End-tidal (closed circles) and inspired (open circles) PCO2; B – minute ventilation during
the course of a five-minute prebreathe; note in both cases the first reading was made 30 s after commencement of the prebreathe
Figure 3
Partly-failed scrubber condition (mean ± SD); A – End-tidal (closed circles) and inspired (open circles) PCO2; B – minute ventilation
during the course of a five-minute prebreathe; note in both cases the first reading was made 30 s after commencement of the prebreathe,
therefore, these readings are not true baseline values; indicative baselines may be inferred from Figure 2 (normal scrubber condition)
Diving and Hyperbaric Medicine Volume 45 No. 1 March 2015
20
‘shortness of breath’ or ‘increased work of breathing’ (16 of
the 18), followed by ‘dizziness’ or ‘light-headedness’ (3/18).
Cognitive changes (3/18), anxiety (2/18), visual changes
(1/18) and the perception of a ‘racing pulse’ (1/18) were
also reported.
There were no significant differences between the subjects
who were rebreather or open-circuit divers in relation to
the primary outcome. For example, in the absent scrubber
condition 9 of 13 rebreather divers versus 6 of 7 open-circuit
divers terminated the prebreathe (P = 0.61).
Figure 4
Absent scrubber condition (mean ± SD); A – End-tidal (closed circles) and inspired (open circles) PCO2 and B – minute ventilation
during the course of a five-minute prebreathe. Note, in both cases the first reading was made 30 s after commencement of the prebreathe,
therefore, these readings are not true baseline values, indicative baselines may be inferred from Figure 2 (normal scrubber condition)
Figure 5
Effect of the partly-failed and absent scrubber conditions on respiratory rate (closed circles) and tidal volume (open circles) (mean ±
SD) during the course of a five-minute prebreathe; respiratory rate remains relatively unchanged whilst tidal volume increases in both
conditions; note in both cases the first reading was made 30 s after commencement of the prebreathe, therefore, these readings are not
true baseline values
Diving and Hyperbaric Medicine Volume 44 No.1 March 2015 21
EFFECT OF THE THREE CONDITIONS ON
PHYSIOLOGICAL PARAMETERS
The effects of the three experimental conditions on PICO2,
PETCO2 and ventilation during the five-minute prebreathe
period are shown in Figures 2 to 4. In the normal scrubber
condition (Figure 2) these parameters did not change
significantly throughout the prebreathe. A low PICO2
(< 5 mmHg), which did not change, was detected in this
condition.
In the partly-failed condition, the mean PICO2 rose
immediately and by three or four minutes into the prebreathe
had plateaued near 20 mm Hg (Figure 3A). Despite this, the
mean PETCO2 did not change due to a small compensatory
increase in mean ventilation (Figure 3B) achieved
predominantly by an increase in VT (Figure 5).
In the absent scrubber condition, the mean PICO2 and PETCO2
rose inexorably (Figure 4A) despite an increase in mean VE
(Figure 4B); the latter once again explained primarily due
to an increase in VT rather than respiratory rate (Figure 5).
There was, however, marked variability among individuals
in the ventilation response to rising PETCO2 (Figure 6). Some
individuals tolerated increases in PETCO2 to higher than
50 mm Hg with no change or even a decrease in VE, whilst
others quickly increased VE to levels around 40–50 L·min-1
very early as the PETCO2 began to rise.
These observations still applied when subjects were
separated into those who terminated (Figure 6B) and those
who did not (Figure 6A), and into rebreather divers and open-
Figure 6
Subjects in the absent scrubber group, separated into A – those who completed the prebreathe, and B – those who terminated the prebreathe;
each subject is represented by a straight line linking the PETCO2 and VE pairs at the beginning and end of the prebreathe
Table 3
Peak inspiratory and expiratory pressures (cm H2O, mean (SD) shown) required for a breathing simulator to move a 1.5 L tidal volume
around the rebreather circuit in the three scrubber conditions, and in the presence and absence of the mouthpiece filter; data represent
the mean of 10 breaths measured over a 1-min period
Condition Expiratory pressure Inspiratory pressure
Filter only 2.56 (0.03) -2.42 (0.07)
Rebreather + normal scrubber 3.51 (0.02) -4.33 (0.06)
Rebreather + partly failed 3.49 (0.04) -4.14 (0.02)
Rebreather + absent scrubber 2.74 (0.01) -3.24 (0.08)
Rebreather + filter + normal scrubber 4.67 (0.05) -5.35 (0.07)
Rebreather + filter + partly failed 4.50 (0.05) -5.31 (0.07)
Rebreather + filter + absent scrubber 4.21 (0.09) -5.06 (0.05)
∆ normal vs. absent scrubber, no filter 0.77 -1.09
∆ normal vs. absent scrubber, with filter 0.46 -0.29
Diving and Hyperbaric Medicine Volume 45 No. 1 March 2015
22
circuit divers (data not presented). The reported symptoms
precipitating termination were often inconsistent with the
obvious physiological responses. For example, all seven
subjects who terminated despite no significant increase
(≤ 2 L·min-1), no change, or even a decrease in VE still cited
dyspnoea as a precipitating symptom. Heart rate did not rise
as the PETCO2 increased in this group (including the subject
who perceived a “racing heart”); the mean (± SD) heart rate
at minutes 1 to 5, being 73 ± 11, 73 ± 9, 74 ± 10, 76 ± 12
and 72 ± 12 beats·min-1 respectively.
MANOMETRY EXPERIMENT
Peak inspiratory and expiratory pressures recorded at the
mouthpiece with the filter present and absent in each of
the three experimental conditions are shown in Table 3.
As anticipated, the difference in pressures between the full
scrubber and absent scrubber condition was reduced (and
therefore less likely to be apparent to subjects) when the
filter was in place.
Discussion
Rebreathers are complex devices with many failure points
and potential user errors. Errors in preparation, assembly
or installation of the CO2 scrubber may result in CO2
rebreathing and hypercapnia. Hypercapnic events, in turn,
may potentiate oxygen toxicity or precipitate other fatal
accidents. As a screen to detect such errors, most divers are
taught to conduct a five-minute ‘prebreathe’ on the assembled
rebreather circuit prior to diving.
The validity of this prebreathe strategy has been questioned.
A small, non-peer-reviewed study reported that none of 14
subjects terminated a five-minute prebreathe on a rebreather
with no scrubber canister installed.7 Ventilation parameters
were not reported, and it is not clear how the subjects were
briefed. It is therefore difficult to compare the results to
those we report here. Nevertheless, our study also indicates
that the prebreathe strategy is insensitive to failure of the
CO2 scrubber.
Most importantly, we exposed the partly-failed group to a
known assembly error that allowed a fraction of the expired
gas to bypass the CO2 scrubber canister, resulting in a
PICO2 that rose to approximately 20 mmHg over several
minutes. Despite this, 18 of 20 subjects did not terminate
the prebreathe in this condition. Other errors or problems
encountered in the real world may result in more (or less)
inspired CO2 than in this partly-failed scenario, and these
would be correspondingly more (or less) likely to be detected
by a prebreathe. However, since a quarter of our subjects did
not terminate even when allocated to the worst possible CO2
rebreathing scenario (complete omission of the CO2 scrubber
canister) the prebreathe must be considered an insensitive
test over the entire range of errors leading to partial failure.
An interesting physiological consideration in interpreting
these results is “what causes subjects to terminate a
prebreathe?” Although our study was not designed
specifically to answer this question we made some relevant
observations. Our data suggest that an increase in ventilation
is not a prerequisite for subjects to perceive dyspnoea
(Figures 4B and 6). Virtually all terminating subjects,
including those whose ventilation did not increase, cited
shortness of breath as one of the precipitating symptoms.
Thus, it is possible that in at least some subjects termination
is driven biochemically; that is, by symptoms (including the
perception of dyspnoea) mediated by an increasing arterial
PaCO2, rather than by perception of an actual increase in
ventilation. This may help to explain the very poor sensitivity
of the prebreathe in the partly-failed condition. In that
setting (Figure 2), a relatively small increase in ventilation,
certainly below a threshold noticeable to the vast majority of
our subjects, was sufficient to compensate for a PICO2 that
plateaued near 20 mm Hg. This prevented the PETCO2 from
increasing, and therefore the subjects in the partly-failed
group were not exposed to the same biochemical stimulus
(an increasing PaCO2) which seems likely to have driven
termination in the absent scrubber group.
The ability to maintain normocapnia during a surface
prebreathe despite partial scrubber failure should not
be interpreted to indicate that minor degrees of bypass
are benign. Indeed, as has been mentioned previously,
commission of the assembly error we used to produce a
repeatable partly-failed condition is widely reported among
divers, and (anecdotally) has led to hypercapnia-induced
incidents. This apparent inconsistency whereby the same
partly-failed condition causes hypercapnia during diving but
not during a prebreathe can be explained by the derangement
of respiratory control that occurs during a dive.
Static lung loads, external resistance to gas flow, and
increased respired gas density all contribute to an increase
in the work of breathing during a dive.8 It has been known
for decades that in some divers this increased work causes
hypoventilation and CO2 retention, even in the absence of
an increased PICO2.9 This tendency has been characterized
as a propensity for the respiratory controller to sacrifice
tight CO2 homeostasis in order to avoid performing the
respiratory work that homeostasis would require.10 There
is evidence that the presence of inhaled CO2 during exercise
and respiratory loading further blunts respiratory drive,11,12
paradoxically (in the present context), at the very time that
responsiveness is crucial to safety.
Not surprisingly, others have reported that a PICO2 similar
to that in our partly-failed condition is dangerous when
combined with exercise and external breathing resistance
similar to that imposed by a rebreather apparatus. In an
experiment aiming to investigate maximum acceptable
CO2 breakthrough levels in rebreather circuits, a 2% CIO2
(15 mmHg) combined with relevant levels of resistance,
exercise, and oxygen breathing caused dangerous levels of
CO2 retention with poor awareness in many of the subjects.13
Diving and Hyperbaric Medicine Volume 44 No.1 March 2015 23
It was concluded that, for diving safety when using typical
underwater breathing apparatus, PICO2 must be maintained
as close to zero as possible. Thus, we reiterate the point that
divers should not assume partial scrubber failure and CO2
rebreathing at levels similar to those measured in our study
are benign simply because our subjects maintained a normal
PETCO2 in this condition. It is notable that we detected a
very small amount of inhaled CO2 (~ 3 mmHg) even in the
normal scrubber condition (Figure 2). This could have been
due to dead space in the mouthpiece and/or filter, trivial
incompetency in the mouthpiece non-return valves, a very
low level of CO2 bypass at the scrubber or a combination of
these factors. Since we only studied one rebreather, we do
not know whether this is a generalized phenomenon.
A number of subjects exposed to the absent scrubber
condition failed to increase or actually decreased ventilation
as PETCO2 increased (Figure 6). Although this is at odds
with classical descriptions of the PETCO2 /VE response,14,15
substantial variability in the ventilation response to rising
PETCO2 has been reported previously in both non-divers and
divers.14,16–18 There is some evidence that divers are more
prone to abnormal responses and that diving itself conditions
participants to become ‘CO2 retainers’.19 The subjects in
our study were relatively experienced divers. Moreover,
some aspects of our experimental conditions may have been
contributory. For example, the rebreather used in our study
would have imposed greater external breathing resistance
than the low resistance respiratory measurement equipment
typically used in studies of CO2 response, and greater
external resistance may dampen the ventilatory response
to inhaled CO2 as discussed earlier.11,12 In addition, to be
consistent with usual diving practice, the subjects breathed a
high fraction of inspired oxygen (70%), and elevated inspired
oxygen may make a further contribution to dampening the
CO2 response.16
There are several limitations to our study. First, subjects
performed the prebreathe in a laboratory environment that
does not faithfully simulate the distracting conditions on a
dive deck before a dive. We attempted to lessen any impact
of the laboratory setting by maintaining lively conversation
among investigators (without directly involving the subjects)
throughout each prebreathe trial.
Second, unlike a real world scenario in which there would
be a low expectation of problems, and although blinded,
our subjects knew there was a substantial chance of being
randomised to breathe on a loop with a scrubber fault. It
was reassuring that despite this, there were no false positives
among 20 subjects when there was a normal scrubber
in place. Nevertheless, the experiment almost certainly
promoted vigilance and our results arguably represent a
best possible case for prebreathe sensitivity (see also the
fourth point below).
Third, due to difficulties in recruiting 60 subjects for the
study, we block-randomised 30 subjects to two of three
scrubber conditions and imposed a concealed manipulation
on the order of those two conditions such that the condition
least likely to result in hypercapnia was tested first in all
participants. This required the subjects to undertake two
prebreathes at least 20 minutes apart. Since the groups had
some subjects in common, they are not entirely independent.
We also considered the possibility of one exposure to inhaled
CO2 somehow affecting the physiological response to a
second administered in close succession, but others have
shown that this does not happen.14
Fourth, the use of an antibacterial filter did impose a small
increase in the manometric pressures required to move a
fixed gas volume around the circuit (Table 3). This could
have contributed to an increased tendency to retain CO2, but
given subjects in the normal condition (and even the partly-
failed condition) maintained a normal PETCO2, there is little
evidence to suggest a prominent effect in that regard. The
small increase in breathing resistance imposed by the filter
may also have increased sensitivity of the prebreathe to the
fault conditions by increasing the likelihood of dyspnoea
as the PETCO2 increased. Thus, we reiterate that our results
arguably represent a best possible case for prebreathe
sensitivity.
Finally, the investigators were not blinded to scrubber
condition. It is therefore possible that the primary outcome
could have been influenced by subtle differences in the
way we interacted with the subjects. However, there was
little opportunity for this. Once the prebreathe started, no
attempts were made to ask the subjects questions or engage
them in any conversation. Discussion about the state of the
rebreather or the outcomes for other subjects were explicitly
avoided during experimental runs.
We also believe the study has several strengths. First, it is
the only study known to address this issue with blinded
subjects and careful physiological monitoring. Second, the
fact that none of 20 subjects terminated when breathing with
a normal scrubber suggests that expectation of problems
was not excessively high, blinding was effective, and the
slight increase in resistance associated with use of the
antibacterial filter did not substantially increase perceptions
of hypercapnic symptoms. Third, the study incorporated a
repeatable partly-failed condition arising from an assembly
error known to have occurred many times in real-world
diving. The implications for translation of study findings to
the diving community are obvious. Finally, all 30 volunteers
attended the study sessions and completed their allocated
experimental trials. There were no drop outs resulting in
missing data.
Conclusions
The five-minute prebreathe is an insensitive test for CO2
scrubber function in a diving rebreather, even when the
scrubber canister is absent. A prebreathe is nevertheless
recommended for purposes such as checking the function
•
Diving and Hyperbaric Medicine Volume 45 No. 1 March 2015
24
of the oxygen addition system before entering the water,
but a duration less than five minutes should be adequate
for that purpose. Arguably the most important secondary
finding of our study is that partial scrubber failure in a
rebreather is a particularly insidious fault if divers rely on a
prebreathe to detect it. By modestly increasing ventilation,
subjects typically maintain normocapnia during a surface
prebreathe in this condition, resulting in a false negative
that is dangerous because normocapnia is much less likely
to be maintained during the dive itself. These findings raise
concerns around methods for testing and monitoring safe
CO2 elimination in rebreather circuits. Several manufacturers
offer CO2 analyzers in the inhale limb of the rebreather circuit
as an option, but these are not yet mainstream features. We
recommend that rebreather training courses emphasize
the importance of correct packing and installation of CO2
scrubber canisters. There is mounting evidence that divers
are poor at recognizing the early symptoms of hypercapnia
(during both prebreathes and diving) and strategies for
avoidance of hypercapnia should be prioritized.
References
1 Doolette DJ, Mitchell SJ. Recreational technical diving part
1. An introduction to technical diving. Diving Hyperb Med.
2013;43:86-93.
2 Higuchi H, Adachi Y, Arimura S, Kanno M, Satoh T. The
carbon dioxide absorption capacity of Amsorb is half that of
soda lime. Anesth Analg. 2001;93:221-5.
3 Mitchell SJ, Cronje F, Meintjes WAJ, Britz HC. Fatal
respiratory failure during a technical rebreather dive at extreme
pressure. Aviat Space Environ Med. 2007;78:81-6.
4 Arieli R, Ertracht O. Latency to CNS oxygen toxicity in
rats as a function of PCO2 and PO2. Eur J Appl Physiol.
1999;80:598-603.
5 Arieli R, Rashkovan G, Moskovitz Y, Ertracht O. PCO2
threshold for CNS oxygen toxicity in rats in the low range of
hyperbaric PO2. J Appl Physiol. 2001;91:1582-7.
6 Hesser CM, Adolfson J, Fagraeus L. Role of CO2 in
compressed air narcosis. Aerospace Med. 1971;42:163-8.
7 Graham D, Bozanic JE. Prebreathing during closed-circuit
diving apparatus set-up ineffective in assessing scrubber
efficacy. In: Vann RD, Denoble PJ, Pollock NW, editors.
Rebreather Forum 3 Proceedings, Durham, NC: AAUS/DAN/
PADI; 2014. p. 268-71.
8 Doolette DJ, Mitchell SJ. Hyperbaric conditions.
Comprehensive Physiol. 2011;1:163-201.
9 Jarrett AS. Alveolar carbon dioxide tension at increased
ambient pressures. J Appl Physiol. 1966;21:158-62.
10 Poon CS. Ventilatory control in hypercapnia and exercise:
Optimization hypothesis. J Appl Physiol. 1987;62:2447-59.
11 Poon CS. Effects of inspiratory resistive load on respiratory
control in hypercapnia and exercise. J Appl Physiol.
1989;66:2391-9.
12 Poon CS. Effects of inspiratory elastic load on respiratory
control in hypercapnia and exercise. J Appl Physiol.
1989;66:2400-6.
13 Shykoff BE, Warkander DE. Exercise carbon dioxide (CO2)
retention with inhaled CO2 and breathing resistance. Undersea
Hyperb Med. 2012;39:815-28.
14 Sahn SA, Zwillich CW, Dick N, McCullough RE,
Lakshminarayan S, Weil JV. Variability of ventilatory
responses to hypoxia and hypercapnia. J Appl Physiol.
1977;43:1019-25.
15 Sherrill DL, Swanson GD. On-line computer estimation of
carbon dioxide response curves. J Clin Monit. 1986;2:198-202.
16 Lanphier EH, Bookspan J. Carbon dioxide retention. In:
Lundgren CEG, Miller JN, editors. The Lung at Depth. New
York: Marcel Decker; 1999. p. 211-36.
17 Schaefer KE. Respiratory pattern and respiratory response to
CO2. J Appl Physiol, 1958;13:1-14.
18 Davis FM, Graves M, Guy H, et al. Carbon dioxide responses
and breath-hold times in underwater hockey players. Undersea
Biomedical Research. 1987;14:527-34.
19 Kerem D, Melamed Y, Moran A. Alveolar PCO2 during rest
and exercise in divers and non-divers breathing O2 at 1 atm
abs. Undersea Biomedical Research. 1980;7:17-26.
Acknowledgements
The authors thank Mr Martin Parker, Ambient Pressure Diving,
for contributing mouthpiece-filter adaptors, mouthpieces and a set
of rebreather hoses that could be destroyed to allow interposition
of the monitoring equipment. The authors also thank Dr Matthew
Pawley for providing statistical guidance in development of the
randomization protocol. Finally, we acknowledge the assistance
provided by Mr Andrew Simpson of Global Dive, Auckland, in
recruiting the subjects.
Conflict of interest: nil
Funding
This work was supported by a grant from the University of
Auckland Performance Based Research Fund.
Submitted: 27 November 2014
Accepted: 14 January 2015
Carolyn Deng1, Neal W Pollock2,3, Nicholas Gant4, Jacqueline A
Hannam1, Adam Dooley4, Peter Mesley5, Simon J Mitchell1,6
1 Department of Anaesthesiology, University of Auckland
2 Center for Hyperbaric Medicine and Environmental Physiology,
Duke University Medical Center, Durham NC, USA
3 Divers Alert Network, Durham, NC, USA
4 Department of Sport and Exercise Science, University of Auckland
5 Divetec New Zealand, Auckland
6 Department of Anaesthesia, Auckland City Hospital, Auckland,
New Zealand
Address for correspondence:
Associate Professor Simon Mitchell PhD, FANZCA
Head of Department, Department of Anaesthesiology
University of Auckland
Private Bag 92019
Auckland, New Zealand
Phone: +64-(0)9-923-2569
E-mail: <sj.mitchell@auckland.ac.nz>
