Apnea testing for the diagnosis of brain
Apnea testing (AT) is a conditio sine qua non in
determining brain death or brain stem death (1)
worldwide (2, 3), although formal AT at a specific
target PaCO2 is required in only 59% of the
countries (3). It is an important sign of loss of
brain stem function and signifies that breath as an
essential element of life has vanished from man. In
the Bible, Genesis 2:7 reads as follows: ?The Lord
God formed the man from the dust of the ground
and breathed into his nostrils the breath of life, and
the man became a living being.? AT is, however, the
most time-consuming, difficult and potentially
harmful of all clinical assessments. Hypotension
of various degrees is a frequent and rather harmless
concomitant (4, 5), but cardiac arrhythmia (6) and
even asystole may become threatening (4, 7, 8).
Disagreement prevails as to which parameters have
to be applied and how to proceed for best
performance. This paper reviews the preconditions
and procedures of AT and addresses some special
problems and pitfalls that have to be overcome in
clinical practice to obtain valid results.
Materials and methods
A review was made concerning pertinent articles
worldwide on AT for diagnosing brain death using
bases (MEDLINE, PubMed) and monographs.
The key words brain, cerebral, death, apnea,
apnoea, and test* were used. After collecting data
the papers were scrutinized for special problems
such as hypotension, excessive hypercarbia, hyp-
oxia, acidosis, cardiac arrhythmias or cardiac
arrest, and pulmonary disorder and the respective
bibliographies searched for any relevant additional
publications. While not intending to be encyclo-
pedic this paper tries to address the most important
points with AT for brain death. The problems,
proposals and results were evaluated against the
background of our own experience of more than
2000 AT to date.
Physiology of respiration
Neurons for modulation and rhythmogenesis of
automatic respiration are located in the pons and
medulla oblongata. Chemoreceptors are seated in
the ventral respiratory group (VRG), a longitud-
inal tissue column along the nucleus ambiguus.
Neurons of the dorsal respiratory group (DRG)
sitting in the ventral portions of the nucleus tractus
Acta Neurol Scand 2005: 112: 358–369 DOI: 10.1111/j.1600-0404.2005.00527.xCopyright ? Blackwell Munksgaard 2005
Acta Neurol Scand 2005: 112: 358–369. ? Blackwell Munksgaard 2005.
Objectives – A review is given on various methods, preconditions and
pitfalls of apnea testing for the diagnosis of brain death. Materials and
methods – An extensive medical data base search was implemented by
information gathered from books and our own experience with more
than 2000 apnea tests. Results – While testing for apnea (AT) is
considered indispensable worldwide, recommendations and handling
differ. Rather than relying on elapsed time, a specific target value for the
partial arterial pressure of carbon dioxide (PaCO2) should be aimed at
being the maximum physiological stimulus for respiration.
Methodological points are elaborated upon in detail for apneic
oxygenation and hypoventilation. Conclusion – AT is an indispensable
element of diagnosing brain death. Although with proper handling and
adequate precautions AT is safe, it should be performed as a last resort.
An international agreement on target values for the PaCO2is desirable.
C. J. G. Lang, J. G. Heckmann
Neurologische Universit?tsklinik, Erlangen, Germany
Key words: brain death; apnea testing; respiration;
artificial ventilation; blood gas check
Christoph J. G. Lang, Neurologische Universit?tsklinik,
Schwabachanlage 6, 91054 Erlangen, Germany
Tel.: +49 9131 8534339
Fax: +49 9131 8536596
Accepted for publication August 15, 2005
solitarii do not participate in rhythmogenesis (9)
(Fig. 1) and are mainly inspiratory.
The pontine respiratory group modulates brea-
thing frequency. Automatic breathing can be
overridden by voluntary control. The most efficient
respiratory stimulus is the partial arterial pressure
of carbon dioxide (PaCO2) (Fig. 2), hence it is used
for maximal stimulation in AT. The slope of the
response curve reflects the sensitivity of respiratory
regulation and amounts to approximately 2–3 l/
Chemosensitive cells just below the ventral
surface of the medulla oblongata are stimulated
by a rise in PaCO2more than by a drop in pH, as
the blood–brain barrier is poorly permeable to ions
such as H+but not to CO2(10). With preserved
circulation, any change in respiration will promptly
result in feedback to the lower brain stem. Respir-
atory changes of PaCO2and pH are highly and
inversely correlated. Under otherwise normal con-
ditions the CO2response curve rises up to a level of
about 70 mmHg.When
exceeds this level a narcotic effect ensues and
ventilation decreases. In wakeful persons a rise in
PaCO2causes dyspnea (air hunger) and an increase
in respiratory frequency rather than in single
respiratory volume. Different brain disorders
cause different alterations of breathing: brain
stem disorders may lead to rapid superficial brea-
thing; hypoxia, hypercapnia or intoxications may
yield deep breathing (Kussmaul type); with dis-
turbed pontomedullary perfusion apneusis (abnor-
mally prolonged inspiration) may be seen; with
increased intracerebral pressure respiration may
become irregular and aperiodic (Biot type); hyp-
oxia or opiate intoxication may induce periodically
increasing and decreasing respiratory movements
(Cheyne-Stokes type); and finally gasping followed
by periods of apnea is seen with severe damage to
the brain stem respiratory centers during agony
before terminal apnea sets in (11).
Techniques of apnea testing
There are mainly two techniques for ascertaining
sufficient oxygenation during AT: Apneic oxygen-
ation and hypoventilation. In the first case the
patient is disconnected from the respirator and
receives pure oxygen at a rate between 4 and 10 l/
min via a catheter (e.g. a 16 French suction
catheter or cannula) that is inserted into the
endotracheal tube down to the level of the carina
Gas convection ensures that the alveoles are
ventilated enough to transport oxygen into the
bloodstream even if there are no respiratory
movements (13). In the second case the patient is
not disconnected from the respirator but minute
volume is reduced to a very low level (0.5–2 l/min)
using for example (synchronized) intermittent
Figure 2. Correlation between arterial partial pressure of
carbon dioxide (PaCO2) and respiratory minute volume. It
increases up to levels of about 70 mmHg. It can be seen from
the graph that during the linear phase an increase in PaCO2of
10 mmHg corresponds to an increase in respiratory minute
volume of about 30 l/min (11).
Figure 1. The brain stem respiratory centers and chemosensi-
tive areas. On the left, a sagittal ventral view is given; on the
right, a cross-section at the level of the arrow is shown. M is the
rostral area (Mitchell), S the intermediate area (Schla ¨ fke), and
L the caudal area (Loeschcke). Cranial nerves are numbered in
roman numerals. CSF, cerebrospinal fluid; ECF, extracellular
fluid. The black dots on the right-hand side of the figure denote
arteries and veins, the gray dots are chemosensitive areas,
situated within the medulla oblongata about 100–400 lm
below its surface (10).
mandatory volume ventilation [(S)IMV] and pure
oxygen for inspiration. The patient is not discon-
nected until the final phase when the required
PaCO2 is attained. This has the advantage of
preventing tracheo-pulmonary complications and
maintaining sufficient oxygenation at the same
time with a high degree of probability while still
allowing the examiner to detect any spontaneous
breathing. This is our preferred means. A further
modification has been described by Al Jumah et al.
(14) using biphasic intermittent positive airway
pressure (BIPAP). This method is also called bulk
diffusion. Not disconnecting the patient from the
respirator allows the maintenance of a continuous
flow of oxygen and positive endexpiratory pressure
(PEEP) (15, 16) without mechanical ventilation
while the patient remains attached to the IMV
Requirements and preconditions
Partial tension of oxygen – There are no explicit
recommendations for the partial arterial tension of
oxygen (PaO2) except preoxygenation with 100%
O2for some time, mostly for 10 min, and avoid-
ance of hypoxia. Patients given room air during a
10-min apnea test (i.e. disconnected from the
respirator) became hypoxic (8). The Quality Stand-
ards Subcommittee of the American Academy of
Neurology (17) recommends a normal PO2 or
preoxygenation to obtain
‡200 mmHg. If the patient is disconnected without
supplying oxygen to the trachea, the level may
drop to dangerously low levels within minutes even
after preoxygenation (18, 19). We have, however,
seen one temporary PaO2level as low as 14 mmHg
without serious harm to the cardiovascular system.
If continuous or intermittent oxygen supply is
preceded by denitrogenation of blood gases, high
PaO2levels can be sustained for very long periods
of time (13).
Partial tension of carbon dioxide – A normal arterial
PCO2or PaCO2of ‡40 mmHg is recommended by
the Quality Standards Subcommittee of the Ameri-
can Academy of Neurology (17) before AT. Belsh
and Schiffman (20) recommend a starting PaCO2
of 36 mmHg or higher. Telleria-Diaz (21) thought
it possible to consider 20 mmHg above the starting
PaCO2level. The requirements for the final PaCO2
differ according to national guidelines. In many
national guidelines a distinct PaCO2is prescribed
for maximal stimulation of the respiratory centers.
This is 50 mmHg ¼ 6.7 kPa [e.g. in the UK (1, 22),
Switzerland (23) and Portugal (24)] or 60 mmHg ¼
7.98 kPa [e.g. in the USA (25), Canada (26) and
Germany (27)]. Both of these target levels are
illustrated in Fig. 4. The claim of Wawersik (28)
with reference to Belsh et al. (29) and Kaufmann
Figure 3. Apneic oxygenation. The patient is disconnected
from the respirator while receiving pure oxygen via a catheter
inserted into the endotracheal tube. HR, heart rate; ABP,
arterial blood pressure; SpO2, oxygen concentration as meas-
ured by pulse oxymetry; RESP, respiration rate (12).
Figure 4. Rise of PaCO2during apnea testing. The dotted lines
represent the levels of PaCO2that are required in the UK
(lower line, 6.65 kPa ¼ 50 mmHg) and the USA (upper line,
7.98 kPa ¼ 60 mmHg). This figure illustrates that 14 min of
apneic oxygenation may not suffice to reach the target value
Lang & Heckmann
and Lynn (30) that recommended target values
range from 44 to 90 mmHg is mistaken, because
both sources adhere to the President’s Commission
suggesting a PaCO2of 60 mmHg.
venous PCO2being greater than arterial PCO2is
reversed (32). Based on empirical findings derived
from physiological observations, a PaCO2 of
about 60–70 mmHg would seem ideal as there is
a rather linear increase in respiratory minute
volume up to this level and a decrease thereafter
(33). While Ropper et al. (34) found that sponta-
neous breathing always began at CO2 pressures
lower than 40 mmHg, onset of respiration has
been described with values of 47 and 54 mmHg
(35) in adults and even higher values in children
(see below). We ourselves have seen a 64-year-old
man who started breathing at a PaCO2 of
53.6 mmHg and a 6-year-old girl who started
breathing at a PaCO2of 54.8 mmHg. Therefore, a
target of 60 mmHg appears to be more appropri-
ate than 50 mmHg.
Hydrogen ion concentration – There are no specific
recommendations for the pH. Its drop is highly
correlated with the rise of PaCO2 and quickly
restored with normoventilation or mild hyperven-
tilation. We have seen uneventful acidosis as low as
6.808. If respiratory acidosis is corrected during
AT – e.g. by bicarbonate solutions – alkalosis will
should not be corrected by buffer or alcaline
solutions during AT; instead, a normal pH or a
value in the low basic range should be ascertained
at the onset of AT.
Temperature – Testing at body temperatures below
32?C (32.2?C, 36) is discouraged by most authors
(19). In these cases the body must be warmed.
Some authors recommend warming up to at least
36 or 36.5?C in every case (12, 17, 37, 38) which
may be a very time-consuming measure but can
shorten AT thereafter. Maekawa et al. (38) have
given a formula describing the dependency of AT
duration on body temperature (y ¼ 0.54x)15.2,
where x is body temperature in ?C and y is
DPaCO2). At any rate, correction of blood gas
values for actual body temperature is deemed
Duration of apnea testing – Some authors have
recommended 10 min, even if blood gas levels
cannot be determined (34), others have consid-
ered 15 min as sufficient (40, 41). A duration of
£3 min used by clinicians in the USA, as reported
by Earnest et al. (42), was clearly insufficient.
However, fixed durations are not yielding reliable
results (31), as the target value of 6.65 kPa
(50 mmHg) or 7.98 kPa (60 mmHg) may not be
reached and the necessary duration cannot be
firmly predicted assuming a predetermined linear
rise of the pCO2, e.g. 0.33 kPa (2.5 mmHg) or
0.42 kPa (3.2 mmHg) per minute (19, 31, 35). The
rise in PaCO2predicted from normal physiology is
3–4 mmHg (43), but may vary considerably under
conditions of brain death (cf. Fig. 4).
The rise during the first 4 min seems to be about
twice or thrice as steep as thereafter (18, 35).
Increases as steep as 12 mmHg have been observed
during the first minute (43, 44). Paret and Barzilay
(45) have proposed an algorithm for estimating
PaCO2 using the formula lnPaCO2¼ 0.69 +
0.072 + 0.86lnPaCO2O where ln is the natural
logarithm and PaCO2O the PaCO2at the begin-
ning of AT. We discourage a time-locked proce-
dure and strongly recommend arterial blood gas
determinations which are prescribed by many
guidelines. The actual German protocol for the
determination of brain death (27) requires the final
PaCO2 to be recorded. To our knowledge the
duration of the apnea test may be extremely
variable and last between 1 min and more than
1 h. The longest time we have met was 71 min. The
use of (S)IMV ventilation instead of apneic
oxygenation will not substantially prolong the
Duration of observation of apnea – The patient
should be observed during the whole procedure
for any respiratory movements and a sufficient
time, about half a minute (36), after the recom-
mended partial gas tension levels have been
reached. To our knowledge 30–60 s appear to be
sufficient indeed. Longer periods as mentioned by
Kunesch et al. (46) are unnecessary and may result
in profound hypoxia or acidosis. As soon as there
is spontaneous breathing during AT, it must be
Proposed increase over baseline – As a rapid increase
in PaCO2to 20 mmHg above normal baseline is
considered a strong stimulus for the respiratory
centers, such an increase is recommended when
the baseline PaCO2is at or above 36–40 mmHg
(17, 29, 47, 48). This means that a PaCO2 of
about 60 mmHg will be reached. We strongly
recommend determining final blood gases in each
case and not to rely on an anticipated increase in
PaCO2. As stated above, the rate of increase may
vary considerably but is usually biphasic, being
steeper during the first few minutes than thereaf-
Blood pressure – A pretest systolic blood pressure
(BP) of ‡90 mmHg is recommended by the Quality
Standards Subcommittee of the American Acad-
emy of Neurology (17). The German protocol even
recommends 100 mmHg preonset. If during AT
there is a fall in BP it is recommended to not let it
drop below 80 mmHg, but we have seen values as
low as 50 mmHg that promptly recovered after-
wards. Usually the BP response is biphasic, being
mildly hypertensive with hyperoxygenation and
mildly hypotensive with hypercarbia (50) (Fig. 5).
We have never met serious problems with hypo-
tension which could always be controlled by fluid
balance/plasma expansion or catecholamines or
Fluid balance – Euvolemia or a positive fluid bal-
ance during the previous 6 h is recommended by
the Quality Standards Subcommittee of the Ameri-
can Academy of Neurology (17).
Medication – Apnea testing must not be performed
under the influence of drugs that may paralyze
respiratory muscles, i.e. relaxants such as pancuro-
nium, and apnea has been deemed possible by high
doses of opioids and narcotics such as barbiturates.
Problems and pitfalls
Barotrauma – Some researchers have mentioned
the possibility of barotrauma (51) which, however,
is considered extremely rare (52). Eight cases have
been described in the literature (51–57) and some
have never encountered such a complication in
over 20 years (52), while others found one instance
among 63 within one and a half year (53). Tension
pneumothorax and pneumoperitoneum may ensue
massive oxygen insufflation into the endotracheal
tube with an oxygen supply catheter that obliter-
ates the tube or a valve mechanism which makes
the escape of the insufflated gas impossible (53–55).
Figure 5. Trend monitoring during apnea testing which was performed using synchronized intermittent mandatory volume (SIMV)
ventilation. The first downward arrow (from left to right) marks the beginning of hyperoxygenation with pure oxygen (sO2100%),
the second the beginning of hypoventilation, the third an intermediate blood gas check and the fourth the last blood gas check
necessary for conformation of the required PaCO2level (here PaCO2¼ 62.3 mmHg) after which the patient was disconnected and
observed for one more minute. Note the initial increase and subsequent decrease in arterial blood pressure. The top row illustrates
heart frequency (HF), the middle row intraarterial blood pressure (APS, arterial pressure systolic; MAP, mean arterial pressure;
APD, arterial pressure diastolic), and the bottom row pulmonary arterial (PAS, pulmonary arterial systolic; PAM, pulmonary
arterial mean) and central venous pressure (ZVD).
Lang & Heckmann
Limitation of the insufflation rate, appropriate
diameter relation between catheter and endotra-
cheal tube, a not too deep insertion, and close
observation of the thorax prevent this unwanted
effect. It is best avoided by keeping the patient on
the respirator and disconnecting it only during the
final observational period after the target PaCO2
has been reached. The bulk diffusion technique
that has been described (14) also helps in avoiding
this complication. The hypoventilation method we
and others (58) use adhibiting (S)IMV ventilation
of about 1 l/min is another safe method. We have
never experienced lung trauma.
should be clearly avoided because they may
result in CO2 narcosis, although general recom-
mendations do not exist. A reduction in respirat-
ory drive under otherwise normal conditions sets
in at about 90 mmHg. In a paper by Rowland
et al. (59) the PaCO2level after 15 min of AT in
nine children was between 50 and 116 mmHg.
The highest level we have ever observed was
132.6 mmHg. There are no generally accepted
PaCO2 values for patients whose natural respir-
ation is adapted to a PaCO2 of more than
45 mmHg. In these cases confirmation of loss of
brain stem functions by instrumental investiga-
tions is recommended (27).
hypercarbia – Values >120 mmHg
Hypoxia – Values <60 mmHg should be avoided.
As mentioned above we have met much lower
uneventful temporary values. As we use controlled
hypoventilation with pure oxygen we have not seen
severe hypoxia any more. With adequate precau-
tions and use of 100% oxygen such values should
not occur and are to be prevented even with
pulmonary problems such as lung contusion. If the
respiratory center is being driven by hypoxia in
patients known to be adapted to high PaCO2
values, AT may be used that includes cautious
lowering of PaO2. There are, however, no accepted
criteria for this type of test (58). In a study by
Ferbert et al. (18) asystole ensued at a PaO2of
about 35 mmHg (Fig. 6). Artificial CO2augmen-
tation without reduction or even with an increase
in respiratory minute volume may be advantageous
in these cases (7).
If pulse oxymetry is used, values should not drop
below 80%. Inadequate preoxygenation is one of
the main reasons for complications during AT (5).
Neither hyperoxic hypoventilation nor apneic
oxygenation should result in hypoxia if handled
properly. In a prospective study of 36 patients (60)
no relevant hypoxia (PaO2< 80 mmHg) was
observed with apneic oxygenation.
Respiratory acidosis – Values <pH 7.2 or 7.0
should be avoided according to current recom-
mendations. In the study by Rudolf et al. (60) the
mean pH at the end of AT was 7.18 in group 1
(observation time at least 5 min after obtaining a
PaCO2of 40 mmHg) and 7.13 in group 2 (PaCO2
at least 60 mmHg). In the study by Rowland et al.
(59) the pH of one child came close to 6.9 and in
the case report by Brilli and Bigos (61) it was 6.94.
We have seen values as low as 6.808 without
complications and consider values down to 7.0 as
acceptable. According to a report by Oikkonen
et al. (62) apneic oxygenation could lead to
profound respiratory acidosis (pH 6.72) and
hypercarbia (PaCO2250 mmHg), but if there was
‡67 mmHg), cardiac sinus rhythm and a low but
reasonable arterial pressure (75/55 mmHg) was to
be kept up for hours. Work performed by Orlia-
guet et al. (63) has shown that AT is well tolerated
despite severe respiratory acidosis.
Hypotension – Systolic values <70 or 60 mmHg,
diastolic values <40 mmHg and a mean arterial
pressure below 50 mmHg should be avoided. The
drop in BP is most likely due to a reduction in the
effect of catecholamines by acidosis. Usually there
is a mild increase in BP with hyperoxygenation and
a somewhat more marked decrease with hypercap-
nia (4, 50). To our knowledge these variations in
BP never become threatening or pose serious
problems if they are corrected, if necessary, using
intravenous fluids, albumin, dopamine, dobuta-
mine or (nor)epinephrine. Close monitoring, how-
ever, is strongly recommended (4, 64). The degree
of change in mean arterial BP (MAP) can even be
estimated before AT using an algorithm proposed
Figure 6. Sharp drop of PaO2(dotted line) and slow increase
in PaCO2(straight line) with disconnection from the respirator
(first arrow, ?Diskonnektion?). Asystole (second arrow, ?Asys-
tolie?) occurred as a consequence of hypoxia, not hypercapnia
or acidosis (18).
by Paret and Barzilay (45). The formula is
MAP ¼ 78.22 ) 1.7t ) 0.33PaCO2O + 0.038-
PaO2O, where t is the time in minutes and PaO2O
is the PaO2at the beginning of AT. We suggest
that in a brain dead patient whose parasympathetic
cardiac regulation is lost, small changes in cate-
cholamine perfusion rate may result in unusually
Hypertension – An increase in mean pulmonary
arterial pressure has been noted by Nicolas et al.
(65). There was a raise by 10 mmHg which
correlated with an increase in PaCO2(and a drop
in pH). This effect is reversible and not deleterious
for right ventricular function (66). We have most
often noted a mild increase in systemic BP during
hyperxoygenation and a decrease with hypercarbia
Cardiac arrhythmia and arrest – Cardiac arrhythmi-
as induced by AT are rare – <1% according to
Goudreau et al. (5) – and cardiac arrest, as
occurred in 2 of 63 cases in an Argentine series
(53), should be avoided at any rate. The reason is
most often excessive acidosis or hypoxia which is
frequently heralded by the onset of new cardiac
arrhythmias or a marked drop in heart rate (6). We
have seen one cardiac arrest with inadvertent
hypoxia that could be restored and one permanent
cardiac arrest in a multitrauma victim who had
suffered severe multiple organ damage including
cardiac and lung contusion.
Increase of intracranial pressure – Intracranial (IC)
pressure is usually not monitored during AT except
in some neurosurgical intensive care units. Because
theoretically AT may increase IC pressure via local
remnant hypercapnic vasodilation and ensuing
increase of cerebral blood volume, it should be
the final resort of all clinical tests (Fig. 7).
The Japanese guidelines place AT at the end of
all tests, after a flat EEG has been demonstrated.
Because in primary brain stem lesion a slowing of
the electroencephalogram (EEG) has been dem-
onstrated, Schwarz et al. (67) also advocated that
AT should be carried out after an isoelectric
EEG. We ourselves have documented a case of
loss of all brain stem functions without any EEG
change during AT. A paper by Coimbra (68) has
caused some concern as it suggests the possibility
that a global reduction in blood flow may be
further reduced and rendered irreversible by AT.
This fear is neither warranted nor supported by
empirical evidence. The concept of ischemic pen-
umbra is applicable to a two- or three-compart-
ment model where there is viable tissue, ischemic
penumbra, and infarcted tissue. It is conceived as
a focal and temporal process (69, 70). It has never
been demonstrated that global cerebral ischemic
penumbra is transformed into global brain infarc-
tion by AT. However, conclusive evidence would,
for example,require prospective
always a transient phenomenon (72) the long
duration of examining for brain death in itself is
an argument against its impact in this particular
situation. In cranio-caudal cerebral herniation the
respiratory centers are the last to loose their
function and as long as they are functional they
will promptly respond to an increase in PaCO2
before the effects of raising IC pressure become
effective. Any spontaneous respiration will imme-
diately terminate AT. The potential action of CO2
as a vasodilator (73) and an agent apt to raise IC
pressure is counteracted and limited by the rise in
PaO2. In those cases where we and our colleagues
from Munich (H. Angstwurm, personal commu-
nication) had the opportunity to register IC
pressure during AT no further increase was
noted. We have seen single patients in whom
respiration was not lost despite one or multiple
AT, some of whom underwent a certain degree of
recovery thereafter. A number of patients with
preserved EEG, despite loss of all brain stem
functions, went on producing electric brain activ-
ity for hours or days although AT was carried out
several times. Moreover, patients with preserved
cerebral blood flow may be brain dead due to
parenchymal damage without herniation (74). If
serious concerns exist, they may be settled by
doing an atropine test (see below) or performing
Figure 7. Simultaneous registration of heart rate (HF, beats
per minute), blood pressure (ART, systolic and diastolic in
mmHg), and intracerebral pressure (ICP, mmHg). Hyperoxy-
genationstartedat 11:10 hours,
11:15 hours. A PaCO2 of 60.5 mmHg was reached at
11:36 hours and half a minute later normoventilation resumed.
There is a mild increase in ICP from 7 to a maximum of
Lang & Heckmann
an adequate instrumental investigation before-
False readings – We have repeatedly noted that very
sensitive respirators may be triggered by heartbeat-
driven thorax excursions causing minimal air flow.
Wijdicks (12) also mentioned this possibility when
he remarked that with continuous positive airway
pressure (CPAP) settings as low as two, false
readings (?spontaneous? respiratory rates of 20–30/
min) may occur. Similar may be the case with
BIPAP respiration when airway pressure decreases
in time with cardiac contraction (76). In these
instances disconnecting the patient for observation,
raising the trigger flow rate settings or reducing the
assisted spontaneous breathing (ASB) level is
Spinal reflexes – Complex spinal cord symmetric
upper limb movements resembling decerebrate
posture have been described that were triggered
not only by each mechanical pulmonary insuffla-
tion, but also by superficial pressure and noxious
stimuli applied to the arms, thorax, or abdomen.
They were abolished by disconnection from the
respirator (77). Lazarus? sign may be observed
during AT or after removal from the respirator
Children – As it is assumed that the threshold for
maximum stimulation is similar to that observed in
adults, children should be handled accordingly (79,
80). However, as conditions in newborns, infants
and even older children may differ from those
found in adults due to a still immature brain stem
or an open fontanel, new procedures are needed.
There is a case report of a 3-year-old child with
chronic severe neurological dysfunction who took
a single breath after 8 min and 45 s of apneic
oxygenation at a PaCO2of 112 mmHg determined
thereafter but ultimately died 5 days later (61).
Another report is that of a 3-month-old infant who
after fulfilling the 1987 Task Force Criteria of
pediatric brain death developed two or three
irregular breaths days later but also finally died
(80). In a 4-year-old child a higher PaCO2thresh-
old was assumed as the child had minimal respir-
atory effort at 91 mmHg (81). Levin and Whyte
reported an infant who started breathing at
59 mmHg (82). In a study by Ashwal (83) the
mean PaCO2at the end of AT in children was
73.3 mmHg PaCO2. There is a most unusual report
from Japan where a 3-month-old girl suffering
from hypoglycemiadeveloped low-frequency
irregular spontaneous respirations after fulfillment
of current criteria for brain death (84) at PaCO2
levels between 30.1 and 73.5 mmHg. None of these
children recovered, but they all finally succumbed
to circulatory failure. This raises the question
whether in children of a certain age higher
PaCO2levels should be requested or whether the
demonstration of loss of brain perfusion should be
mandatory. Given these observations 5 min of
apneic oxygenation – as were actually performed
by some clinicians – is definitely inappropriate (42,
79). The Portuguese guidelines (85) suggest differ-
ent PaCO2levels for adults (50 mmHg, 6.65 kPa)
and children (60 mmHg, 7.98 kPa). According to
the German guidelines (27), in children under the
age of 2 years AT has to be performed two times
by two examiners 24 or 72 h apart. Although there
are reasons to believe that preterm infants do not
behave differently from full-term newborns, there
are no sufficiently valid recommendations within
this age group (86).
Monitoring – End-tidal capnometry, pulse oxyme-
try, transcutaneous blood gas determination (87),
and intra-arterial blood gas analysis may all be
used (50, 88). However, in vitro measurements, i.e.
regular blood gas determinations, remain the gold
standard and should not be replaced by other
means. Some authors recommend determining a
basal value and then to do subsequent checks
approximately every 5 min. Having obtained two
or three values the increase in PaCO2 may be
extrapolated rather precisely. We feel that the
number and timing of blood gas checks should be
handled individually according to initial values,
monitoring, and the expertise of the examiner.
Close monitoring is especially useful and necessary
with artificial CO2augmentation, depending on the
anticipated increase in CO2(7). Final values should
always be corrected for body temperature. The use
of a (portable) blood gas check equipment at the
bedside is an especially efficient method as it allows
performing the blood gas analyses and observing
the patient at the same time by one examiner.
Artificial CO2 insufflation – The President’s Com-
mission already suggested the use of an oxygen and
carbon dioxide mixture (33) in order to raise
PaCO2. If an especially long duration of observa-
tion is expected, if an increase in PaCO2is not
easily achieved or very lengthy, or if PaO2drops
excessively with hypoventilation, this method may
be used (7). Careful and close monitoring must be
ascertained, but if handled properly the test dur-
ation may be shortened considerably, rendering the
procedure safer in some cases (89, 90) (Fig. 8).
Replacement of apnea testing by other means – In
cases where correct AT is not possible, e.g. because
of the impossibility to reach the required PaCO2
values or when a dangerous drop in PaO2 is
unavoidable such as may be the case in severe
thorax trauma or other pulmonary problems,
instrumental brain stem testing like arteriography,
perfusion SPECT, ultrasound Doppler, or evoked
potentials (AEP [auditory evoked potentials], SEP
[somatasensory evoked potentials]) may replace
this part of the examination according to the
German guidelines (27). The same procedure was
recommended in patients who were adapted to a
PaCO2of more than 45 mmHg (39).
Recommendations and proposals
Recommendations for AT for the determination of
brain death have been given by various committees
or persons of different nations (27, 91), including
explicit instructions and block diagrams. AT
should be the last resort or postponed until an
atropine test (26, 92) – usually by injection of 2 mg
– is negative (93). If it is undoubtedly positive, AT
is not warranted. Care has to be taken not to inject
atropine into a line that was used for catecholam-
ines as the washout of even small quantities may
yield erroneous results. Brain stem respiratory
centers and vagal neurons are situated in close
vicinity such that both functions are highly corre-
lated. In our clinical practice we have seen only one
discrepant result (negative atropine test, but still
respiration on AT) among more than 100 cases. If
atropine testing is negative, AT must be carried
out. Hypoventilation after preoxygenation is our
preferred means. A body temperature of >32?C
should be ascertained. BP, preferentially monit-
ored intraarterially, should be no lower than
80 mmHg initially and the pH preferably in the
normal to mildly alkalotic range. If an unwanted
drop of BP is seen during testing it can be raised
using vasopressor agents or volume substitution or
both. After preoxygenation with 100% O2 the
highest possible PaO2 should be achieved, then
ventilatory volume reduced to approximately 5–
2 l/min [e.g. by (S)IMV ventilation, alternatively
by disconnection from the respirator and insuffla-
tion of O2into the trachea at the level of the carina
at a rate of about six or more liters per minute
under close observation until requirements are met
(12, 94)], and the PaCO2 be checked after an
appropriate period of time depending on the initial
blood gas values corrected for body temperature.
Pulse oxymetry, end-tidal capnometry, a BP and
heart rate monitor, an IA line and the feasibility of
rapid determination of arterial blood gases are very
helpful; an optimal setting would be a bedside
blood gas check device that enables the examiner
to perform blood gas analyses without loosing the
patient out of sight. As soon as the required PaCO2
level is reached, the patient should be disconnected
from the respirator (if he is not already) and
observed for an appropriate period of time (30–
60 s), then reconnected and mildly hyperventilated
for some minutes until initial values are restored
approximately except in those cases where AT is
the final test and the respirator may remain
switched off thereafter. We use a fine thread for
the detection of respiratory activity which is
sensitive enough to show displacement by heart-
beats, thus proving that the system is capable of
monitoring even the slightest breathing. Other
methods such as a moist thin piece of soft paper,
a cold mirror or blank piece of metal, however,
may serve as well.
Testing for apnea is considered indispensable for
the determination of brain death worldwide and
may be safely performed under almost any
Figure 8. Comparison of the rise of PaCO2with hypoventila-
tion (top) and artificial CO2insufflation (bottom). The increase
in PaCO2is much steeper with the second condition (7).
Lang & Heckmann
circumstances, if adequate precautions are met (12,
25, 64), although it remains a formidable task to
the examiners. The demonstration of loss of
breathing bears high intuitive evidence that life
has vanished from man also for laypersons. It is
not as difficult or harmful as some authors want to
make us believe. To our knowledge AT can be
performed lege artis in 99.8% of all cases if
currently available possibilities are exhausted; the
remaining ones may be handled using subsidiary
instrumental tests. The complication rate (e.g.
cardiac arrest) is very low. Differences in national
guidelines exist and there is at least one country
(Portugal) recommending different procedures in
children and adults. As with criteria for brain
death itself, it would be utterly desirable to have
identical guidelines worldwide including the target
PaCO2which, to our opinion, should be at least
60 mmHg in adults. This would abolish the vexing
statement raised by Levin and Whyte (82) that
clinical criteria – i.e. for AT – make it possible to
be brain dead in one country and not in another.
We gratefully appreciate the helpful comments of an anony-
1. Pallis C, Harley DH. ABC of brainstem death, 2nd edn.
London: BMJ Publishing Group, 1996.
2. Haupt WF, Rudolf J. European brain death codes: a com-
parison of national guidelines. J Neurol 1999;246:432–7.
3. Wijdicks EFM. Brain death worldwide. Accepted fact but
not global consensus in diagnostic criteria. Neurology
4. Jeret JS, Benjamin J. Risk of hypotension during apnea
testing. Arch Neurol 1994;51:595–9.
5. Goudreau JL, Wijdicks EF, Emery SF. Complications during
apnea testing in the determination of brain death: predis-
posing factors. Neurology 2000;55:1045–8.
6. Frumin MJ, Epstein RM, Cohen G. Apneic oxygenation in
man. Anesthesiology 1959;12:789–98.
7. Lang CJ. Apnea testing by artificial CO2augmentation.
8. Marks SJ, Zisfein J. Apneic oxygenation in apnea tests for
9. Bolton CF, Chen R, Wijdicks EFM, Zifko UA. Neurology of
breathing. Philadelphia: Butterworth Heinemann, 2004;27.
10. Tammeling GJ, Quanjer PH. Physiologie der Atmung II.
Frankfurt: pmi, 1984;203.
11. Richter DW. Rhythmogenese der Atmung und Atmungs-
regulation. In: Schmidt RF, Thews G, eds. Physiologie des
Menschen, 27th edn. Berlin: Springer, 1997;592–603.
12. Wijdicks EFM (ed.). Clinical diagnosis and confirmatory
testing of brain death in adults. In: Brain death. Philadel-
phia: Lippincott Williams & Wilkins, 2001;61–90.
13. Pitts LH, Kaktis J, Caronna J, Jennett S, Ho JT. Brain death,
apneic diffusion oxygenation, and organ transplantation.
J Trauma 1978;18:180–3.
14. Al Jumah M, McLean DR, Al Rajeh S, Crow N. Bulk dif-
fusion apnea test in the diagnosis of brain death. Crit Care
15. Perel A. Apnea test for brain death. Arch Neurol
16. Perel A, Berger M, Cotev S. The use of continuous flow of
oxygen and PEEP during apnea in the diagnosis of brain
death. Intensive Care Med 1983;9:25–7.
17. Quality Standards Subcommittee of the American Academy
of Neurology. Practice parameters for determining brain
death in adults. Neurology 1995;45:1012–4.
18. Ferbert A, Buchner H, Hacke W. Der Apnoe-Test in der
Hirntoddiagnostik. Intensivmed 1987;24:405–8.
19. Milhaud A, Riboulot M, Gayet H. Disconnecting tests and
oxygen uptake in the diagnosis of total brain death. Ann N
Y Acad Sci 1978;315:241–51.
20. Belsh JM, Schiman PL. Apnoea testing to confirm brain
death in clinical practice. J Neurol Neurosurg Psychiatry
21. Telleria-Diaz A. Prueba de apnea en la determinacio ´ n de la
muerte por criterios cerebrales. Rev Neurol 1998;27:108–
22. Diagnosis of brain death. Lancet 1976;2:1069–70.
23. Moeschler O, de Tribolet N. De ´ finition de l’apne ´ e dans le
diagnostic de mort ce ´ re ´ brale. Schweiz
24. Garcia C, Ferro JM. European brain death codes: Portu-
guese guidelines. J Neurol 2000;247:140.
25. Wijdicks EFM. Determining brain death in adults. Neur-
26. Canadian Neurocritical Care Group. Guidelines for the
diagnosis of brain death. Can J Neurol Sci 1999;26:64–6.
27. Wissenschaftlicher Beirat der Bundesa ¨ rztekammer. Rich-
tlinien zur Feststellung des Hirntodes. Dritte Fortschrei-
bung 1997 mit Erga ¨ nzungen gema ¨ ß Transplantationsgesetz
(TPG). Dt A¨rztebl 1998;95:B-1509–16.
28. Wawersik J. Dissoziierter Hirntod. In: Lawin P, ed. Praxis
der Intensivbehandlung, 6th edn. Stuttgart: Thieme,
29. Belsh JM, Blatt R, Schiman PL. Apnea testing in brain
death. Arch Intern Med 1986;146:2385–8.
30. Kaufmann HH, Lynn J. Brain death. Perspect Neurosurg
31. van Donselaar CA, Meerwaldt JD, van Gijn J. Apnoea
testing to confirm brain death in clinical practice. J Neurol
Neurosurg Psychiatry 1986;49:1071–3.
32. Rigg CD, Cruickshank S. Carbon dioxide during and after
the apnoea test – an illustration of the Haldane effect.
33. Report of the Medical Consultants on the diagnosis of
death to the president’s commission for the study of ethical
problems in medicine and biomedical and behavioral
research. JAMA 1981;246:2184–6.
34. Ropper AH, Kennedy SK, Russell L. Apnea testing in the
diagnosis of brain death. J Neurosurg 1981;55:942–6.
35. Schafer JA, Caronna JJ. Duration of apnea needed to
confirm brain death. Neurology 1978;28:661–6.
36. Guidelines for the determination of death: Report of the
medical consultants to the president’s commission for the
study of ethical problems in medicine and biomedical and
behavioral research. Neurology 1982;32:395–9.
37. Dominguez-Roldan JM, Barrera-Chacon JM, Murillo-Cabe-
zas F, Santamaria-Mifsut JL, Rivera-Fernandez V. Clinical
factors influencing the increment of blood carbon dioxide
during the apnea test for the diagnosis of brain death.
Transplant Proc 1999;31:2599–600.
38. Maekawa T, Sadamitsu D, Tateishi A. Safety applications in
apnea test. Clin Neurol 1993;33:1331–3 (in Japanese).
39. Hacke W, Angstwurm H, Fo ¨rderreuther S et al.
chfu ¨ hrung des Apnoetests bei der Hirntoddiagnostik. Akt
Neurol 2002;29:83–4, 310.
40. Link J, Schaefer M, Lang M. Concepts and diagnosis of
brain death. Forensic Sci Int 1994;69:195–203.
41. Rohling R, Wagner W, Mu ¨hlberg J, Link J, Scholle J,
Rosenow D. Apnea test: pitfalls and correct handling.
Transplant Proc 1986;18:388–90.
42. Earnest MP, Beresford HR, McIntyre HB. Testing for ap-
nea in suspected brain death: methods used by 129 clini-
cians. Neurology 1986;36:542–4.
43. Shapiro BA. The apnea–PaCO2relationship: some clinical
and medico-legal considerations.
44. Stock MC, Schisler JQ, McSweeney TD. The PaCO2rate of
rise in anesthetized patients with airway obstruction. J Clin
45. Paret G, Barzilay Z. Apnea testing in suspected brain dead
children – physiological and mathematical modelling.
Intensive Care Med 1995;21:247–52.
46. Kunesch E, Birken T, Classen J. Durchfu ¨ hrung des
Apnoetests beider Hirntoddiagnostik.
47. Benzel EC, Mashburn JP, Conrad S, Modling D. Apnea
testing for the determination of brain death: a modified
protocol. J Neurosurg 1992;76:1029–31.
48. Wijdicks EFM. The diagnosis of brain death. N Engl J Med
49. Benzel EC, Gross CD, Hadden TA, Kesterson L, Landreneau
MD. The apnea test for the determination of brain death.
J Neurosurg 1989;71:191–4.
50. Lang CJG. Blood pressure and heart rate changes during
apnoea testing with or without CO2insufflation. Intensive
Care Med 1997;23:903–7.
51. Vivien B, Haralambo MS, Riou B. Barotraumatisme lors du
test d’apne ´ e chez des patients en e ´ tat de mort ence ´ phalique.
Ann Fr Anesth Re ´ anim 2001;20:370–3.
52. Seshia S, Teitelbaum J, Young B. Response to Lessard et al.
Can J Neurol Sci 2000;27:353–4.
53. Saposnik G, Rizzo G, Adriana V, Sabbatiello R, Deluca JL.
Problems associated with the apnea test in the diagnosis of
brain death. Neurol India 2004;52:342–5.
54. Bar-Joseph G, Bar-Lavie Y, Zonis Z. Tension pneumothorax
during apnea testing for the determination of brain death.
55. Brandstetter RD, Choi MY, Mallepalli VN, Klass S. An
unusual case of pulmonary barotrauma during apneic
oxygenation testing. Heart Lung 1994;23:88–9.
56. Lessard M, Mallais R, Turmel A. Apnea test in the diag-
nosis of brain death. Can J Neurol Sci 2000;27:353.
57. Saposnik G, Rizzo G, Deluca JL. Pneumothorax and pneu-
moperitoneum during the apnea test. Arq Neuropsiquiatr
58. Gutmann DH, Marino PL. An alternative apnea test for the
evaluation of brain death. Ann Neurol 1991;30:852–3.
59. Rowland TW, Donnelly JH, Jackson AH. Apnea docu-
mentation for determination of brain death in children.
60. Rudolf J, Haupt WF, Neveling M, Grond M. Potential pit-
fallsin apnea testing.
61. Brilli RJ, Bigos D. Altered apnea threshold in a child with
suspected brain death. J Child Neurol 1995;10:245–6.
62. Oikkonen M, Aittoma ¨ki J, Saari M. Apnoeic oxygenation of
a brain-dead patient. Lancet 1987;2:908–9.
J Clin Anesth
63. Orliaguet GA, Catoire P, Liu N, Beydon L, Bonnet F.
Transesophageal echocardiographic assessment of left
ventricular function during apnea testing for brain death.
64. Jeret JS. Complications during apnea testing in the deter-
mination of brain death: predisposing factors. Neurology
65. Nicolas F, Combes JC, Louvier N, d’Athis P, Freysz M.
Severe pulmonary arterial hypertension during the apnea
test for brain death. Transplant Proc 1996;28:375.
66. Combes JC, Nicolas F, Lenfant F, Cros N, d’Athis P, Freysz
M. Modifications he ´ modynamiques induites par le test
d’apne ´ e chez le patient en e ´ tat de mort ce ´ re ´ brale. Ann Fr
Anesth Re ´ anim 1996;15:1173–7.
67. Schwarz G, Litscher G, Pfurtscheller G, Schalk HV,
Rumpl E, Fuchs G. Brain death: timing of apnea testing
in primary brain stem lesion. Intensive Care Med
68. Coimbra CG. Implications of ischemic penumbra for the
69. Fisher M. The ischemic penumbra: identification, evolution
and treatment concepts. Cerebrovasc Dis 2004;17(Suppl.
70. Touzani O, Roussel S, MacKenzie ET. The ischaemic pen-
umbra. Curr Opin Neurol 2001;14:83–8.
71. Heiss WD. Imaging the ischemic penumbra and treatment
effects by PET. Keio J Med 2001;50:249–56.
72. Hakim AM. Ischemic penumbra: the therapeutic window.
Neurology 1998;51(Suppl. 3):S44–6.
73. Heckmann JG, Erbguth FJ, Hilz MJ, Lang CJG, Neundo ¨rfer
B. Die Hirndurchblutung aus klinischer Sicht. Med Klin
74. Tool GJ, Lansky LL, Hughes JR et al. Pitfalls in diagnosing
brain death in infancy. J Child Neurol 1987;2:134–8.
75. Moskopp D, Kurthen M, Wassmann H. Klinische und
76. Willats SM, Drummond G. Brainstem death and ventilator
trigger settings. Anaesthesia 2000;55:676–84.
77. Martı ´-Fa `bregas J, Lo ´pez-Navidad A, Caballero F, Otermin
P. Decerebrate-like posturing with mechanical ventilation
in brain death. Neurology 2000;54:224–7.
78. Do ¨s ¸ emeci L, Cengiz M, Yılmaz M, Ramazanog ˆ lu A. Fre-
quency of spinal reflex movements in brain-dead patients.
Transplant Proc 2004;36:17–9.
79. Outwater KM, Rocko MA. Apnea testing to confirm brain
death in children. Crit Care Med 1984;12:357–8.
80. Ashwal S. Clinical diagnosis and confirmatory testing of
brain death in children. In: Wijdicks EFM, ed. Brain
death. Philadelphia: Lippincott Williams & Wilkins,
81. Vardis R, Pollack MM. Increased apnea threshold in a
pediatric patient with suspected brain death. Crit Care
82. Levin SD, Whyte RK. Brain death sans frontie ` res. N Engl J
83. Ashwal S. Brain death in early infancy. J Heart Lung
84. Okamoto K, Sugimoto T. Return of spontaneous respiration
in an infant who fulfilled current criteria to determine
brain death. Pediatrics 1995;96:518–20.
85. Guia de diagno ´ stico de morte cerebral. Acta Me ´ d Port
86. Ashwal S. Brain death in the newborn. Current perspec-
tives. Clin Perinatol 1997;24:859–82.
BrazJ MedBiol Res
Lang & Heckmann
87. Lang CJG. Apnea testing guided by continuous transcu- Download full-text
taneous monitoring of partial pressure of carbon dioxide.
Crit Care Med 1998;26:868–72.
88. Lang CJG, Heckmann JG, Erbguth F et al. Transcutaneous
and intra-arterial blood gas monitoring – a comparison
during apnoea testing for the determination of brain death.
Eur J Emerg Med 2002;9:51–6.
89. Jeret JS. Apnea testing. Neurology 1996;46:1192.
90. Melano R, Adum ME, Scarlatti A, Bazzano R, Araujo JL.
Apnea test in diagnosis of brain death: comparison of two
methods and analysis of complications. Transplant Proc
91. Wijdicks EFM. In search of a safe apnea test in brain death:
is the procedure really more dangerous than we think?
Arch Neurol 1995;52:338.
92. Ouaknine GE, Mercier C. La valeur du test a ` l’atropine dans
la confirmation de la mort ce ´ re ´ brale. Union Me ´ d Can
93. Siemens P, Hilger HH, Frowein RA. Heart rate variability
and the reaction of heart rate to atropine in brain dead
patients. Neurosurg Rev 1989;12(Suppl. 1):282–4.
94. Lang CJG, Heckmann JG, zu Kunesch E et al. Durchfu ¨ hrung
des Apnoetests bei der Hirntoddiagnostik. Akt Neurol
2002: 29:83–84; 310–1.