British Journal of Anaesthesia 96 (3): 335–40 (2006)

doi:10.1093/bja/aei310 Advance Access publication January 16, 2006

The carina as a radiological landmark for central venous

catheter tip position
P. A. Stonelake and A. R. Bodenham*

Department of Anaesthesia, The General Infirmary at Leeds, Great George Street, Leeds LS1 3EX, UK
*Corresponding author. E-mail: Andy.Bodenham@leedsth.nhs.uk
Background. Many publications, including the instructions accompanying central venous
catheters, state that it is negligent to site the catheter tip in the right atrium. If the catheter
tip is above the carina on a post-procedure radiograph then it is generally accepted that the
catheter lies outside the right atrium. It is also recommended that the catheter tip should lie
in the long axis of the superior vena cava without acute abutment to the vein wall. We per-
formed a retrospective audit of the position of central venous catheter tips on routine post-
procedure chest radiographs in intensive care unit patients, to see if these potentially conflicting
requirements had been met.
Methods. We identified 213 central venous catheters suitable for analysis, within a study
population of 200 consecutive cases. We measured the distance of the central venous catheter
tip above or below the carina and the angle of the central venous catheter tip to the vertical
(a surrogate marker for the angle of abutment of the tip to the approximately vertical superior
vena cava wall).
Results. For right-sided catheters there was a high (74/163) number placed with their tips below
the carina, but a very low number (4/163) with their tips at a steep (>40 ) angle to the vertical. For
left-sided catheters very few (7/50) were placed with their tips below the carina, but for those
43 sited above the carina most could be considered to be in suboptimal positions. This was
because they were either too high and had not even crossed the midline (9), or had an acute angle
(>40 ) between the tip and the vertical (27).
Conclusions. We suggest that for left-sided catheters placement of the tip below the carina is
more likely to result in a satisfactory placement.
Br J Anaesth 2006; 96: 335–40
Keywords: anatomy, jugular vein; complications, positional; monitoring, central venous pressure

Accepted for publication: November 23, 2005

There is ongoing controversy as to whether central venous
catheter (CVC) tips should always lie above the pericardial
reflection.1 For CVC tips lying below the pericardial reflec-
tion there is a small but potentially fatal risk of pericardial
tamponade if the CVC tip erodes through the vessel wall.
The frequency of this risk has not been quantified. Other
problems of catheter placement in the right atrium (RA)
include arrhythmias, placement in the coronary sinus, and
tricuspid valve damage.
Vessel wall erosion can also occur when the CVC tip lies
above the pericardial reflection, usually causing hydrothorax
or hydromediastinum from extravasated fluid, but this is less
likely to have a fatal outcome. Perforation of the superior
vena cava (SVC) is probably more likely with left-sided than
with right-sided CVCs.2–4 This is influenced by the steep
angle the left innominate vein makes with the SVC. Here the
catheter will abut the wall of the SVC unless the tip is
advanced around the curve into the lower SVC or RA. It
has been shown in the laboratory that an angle of the CVC tip
to vessel wall of greater than 40 is more likely to lead to
vessel wall perforation.5 Vessel wall erosion is also influ-
enced by the type of fluid infused, the likely mechanism
being chemical trauma, suggesting that lesser angles may
also be damaging.
There is appreciable morbidity from CVC tips lying too
proximal, either in the left or the right innominate vein.
Mechanical or chemical irritation of the vessel wall leads to
pain on injection of drugs, thrombosis and subsequent infec-
tion,6 7 which is more likely in the upper SVC or innominate
veins, especially on the left side.8 Extravasation injury may
result from CVCs lying so proximal that one or more of the
catheter openings lies outside the vessel lumen.

The Board of Management and Trustees of the British Journal of Anaesthesia 2006. All rights reserved. For Permissions, please e-mail: journals.permissions@oxfordjournals.org
 Stonelake and Bodenham

The upper limit of the pericardial reflection cannot be
seen on a plain chest radiograph (CXR), however it is
generally accepted that it is below the carina. This has
been assessed in preserved cadavers,9 fresh cadavers,10 in
anaesthetized children undergoing cardiac surgery11 and in
adults using computerized tomograms.12 The optimum tip
position is more complicated than merely ensuring place-
ment above the carina and is also likely to be different for
left- and right-sided CVCs. It may be particularly difficult
for left-sided CVCs to satisfy all the criteria for optimum
placement. This leaves the clinician with a dilemma: do
they risk the potentially fatal but rare complication from
placement in the RA or the much commoner complications
from higher placement that may still have appreciable mor-
bidity and occasional mortality. For optimum placement of
left-sided CVCs do we need to ignore the recommendation
to site the tip above the carina? With this in mind we decided
to audit our current practice for CVC placement.
Methods
Two hundred consecutive admissions to the adult general
intensive care unit (ICU) at The General Infirmary at Leeds
between December 2003 and April 2004 were identified
retrospectively. Post-insertion chest X-rays (CXRs) corre-
sponding to these admissions were retrieved from the
Picture Archiving & Communication System (PACS—
Agfa-Gevaert Ltd, Brentford, UK). As this audit was per-
formed retrospectively, clinicians inserting CVCs did not
know that an audit would take place. We identified post-
insertion CXRs by examining the radiology database for
each admission chronologically and selecting the first or
second CXR with a CVC in situ. Second CXRs were
chosen where available, as this should have allowed clini-
cians to correct CVCs deemed to be in an unsatisfactory
position. These CXRs were assessed for the following
measurements:
 Vertical distance of the CVC tip above or below the
carina.
 The angle of the distal 1 cm of the tip to the vertical.
Assuming the SVC lies approximately vertical, this angle
approximates the angle of the CVC tip to the wall of
the SVC. The vertical against which the angles were
measured was a line connecting the vertebral spinous
processes.
 CVC type, site of insertion and any evidence of
misplacement were recorded.
All measurements were made using the software available
on the hospital PACS system. Images were numbered then
anonymized before analysis.
We also commented on the overall suitability of CVC
position. We suggested arbitrarily that catheter tips at 5 cm
or greater above or below the carina were likely to be too
high or low respectively.
Although we did not look back at medical records to get
details of placement techniques, we can state that with the
exception of a small number of pre-existing long term i.v.
devices such as Hickman lines all procedures were per-
formed in the ICU or operating theatre without the use of
ECG-guided positioning or X-ray screening.
Results
We identified and examined the radiological records of
200 consecutive ICU admissions. We found 148 patients
who had CVCs visible on CXR—many with multiple
CVCs during their in-patient stay. We identified 243
CVCs. We excluded 30 of these as unsuitable for measure-
ment (Fig. 1). We were not able to record the length of the
CVCs (the two usual multilumen CVC lengths available in
our institution were 16 and 20 cm), nor did we know the
depth to which they were inserted.
Table 1 shows the site of insertion of the 213 CVCs
audited. It is clear that in the population audited the
Table 1 The number (percentage of total) of CVCs according to the site of
insertion
Internal jugular Subclavian Total
Right 148 (69%) 15 (7%) 163 (76%)
Left 40 (19%) 10 (4%) 50 (23%)
Total 188 (88%) 25 (12%) 213

Total CVCs identified

243

CVCs audited CVCs excluded

213 30

Misplaced CVCs
Standard CVC
Long term CVC Dialysis catheters PAFC PAFC sheaths 2
210
3 1 12 15 Left IJ L sub
16 and 20 cm CVCs
Right IJ innom

Fig 1 Details of CVCs audited. CVC, central venous catheter; PAFC, pulmonary artery floating catheter; IJ, internal jugular; innom, innominate vein; sub,
subclavian vein.

336
 CVP catheter tip and carina

internal jugular vein was favoured over the subclavian and
that the majority were inserted from the right-hand side.
The vertical distance of the catheter tip from the carina
was measured. Table 2 shows the proportion of CVCs
located below the carina according to site of insertion.
Figure 2 shows the distribution of the distance of CVC tips
from the carina. Right-sided CVCs, whether placed via the
internal jugular or subclavian routes, were more likely to
lie with their tips below the carina than left-sided CVCs.
CVCs placed via the left internal jugular vein were least
likely to have their tips lying below the carina. Table 3
shows the number of CVCs >50 mm above and below
the carina according to site of insertion.
We also audited the angle the CVC tips made with the
vertical. Right-sided CVCs were found to rarely have a
significant angle between the tip and the SVC—only 4 of
the 163 right-sided CVCs had tips with angles greater
Table 2 The proportion (percentage) of CVCs with tips sited below the carina
according to site of insertion
Internal jugular Subclavian
Right 67/148 (45%) 7/15 (47%)
Left 4/40 (10%) 3/10 (30%)
Total 71/188(38%) 10/25 (40%)
than 5 to the vertical (range 12–90 ). Further data relate
to left-sided CVCs only.
The left-sided CVCs were investigated further to examine
the relationship between tip position and angle to the SVC
(Fig. 3 and Table 4). A high proportion, 27/43 (63%), of
the CVCs above the carina had an angle to the vertical
above 40 . None of the CVCs below the carina had an
angle to the vertical above 40 (in fact they were all less
than 20 to the vertical). The number of CVCs that would
satisfy both criteria for adequate position, that is, above the
carina and an angle to the SVC of less than 40 , was small
[16 of 43 (37%)] and the majority of these (9) were high
up in the innominate vein and had not even crossed the
midline.
These figures show that the majority of right-sided cathe-
ter tips could be considered to be in a satisfactory position
but a smaller number were either too high in the SVC or
too low in the RA. The majority of left-sided catheter tips
could be considered unsatisfactory because of the risk
of abutment at and acute angle to the wall of the SVC or
placement in the innominate vein.
Total A typical example of poor catheter tip position is shown
in Figure 4 where the tip of a left-sided catheter is abutting
74/163 (45%)
7/50 (14%) the right side of the SVC at an acute angle with the risk of
81/213 (38%) impending catheter perforation.

Table 3 Number (percentage) of CVCs >50 mm above and below the carina Discussion
according to site of insertion
This study demonstrates that a high proportion (38% overall)
Right Right Left Left of CVCs were sited with their tips below the carina. The
internal subclavian internal subclavian proportion was higher with right-sided CVCs (45%). We
jugular jugular
suspect these findings are typical of practice elsewhere.
>50 mm above carina 11 (7%) 0 10 (25%) 0 CVCs of standard length of 16 or 20 cm are used within
Within 50 mm of carina 131 (89%) 14 (93%) 29 (73%) 10 (100%) the setting of the ICU and operating theatre. There is
>50 mm below carina 6 (4%) 1 (7%) 1 (2%) 0
Total 148 15 40 10
undoubtedly reluctance from many clinicians to leave por-
tions of the CVC outside the patient for reasons of sterility

Distance of CVC tip from carina

25

20
Right IJ
Number of CVCs

Right SUB
15 Left IJ
Left SUB

10

0
−80 −70 −60 −50 −40 −30 −20 −10 0 10 20 30 40 50 60 70 80
Below carina Above carina
Distance from carina (mm)

Fig 2 Distance of CVC tip from carina. IJ, internal jugular; SUB, subclavian.

337
 Stonelake and Bodenham

Angle of tip and distance from carina for left-sided CVCs

90

80

70

Angle of CVC to vertical 60

50

40

30

20

10

0
−100 −80 −60 −40 −20 0 20 40 60 80 100
Below carina Above carina
Distance from carina (mm)

Fig 3 Scattergraph of catheter tip position and angle to the vertical. Each point represents an individual catheter tip.

Zone C
Zone B

Zone A

Pericardial
reflection Right
atrium

Fig 4 Portable CXR in a critically ill patient shows two CVCs in situ, both
inserted from the left internal jugular vein. One is a standard multilumen
catheter and the other is a dialysis catheter. The tip of the former is poorly
positioned at the junction of the innominate vein and SVC. The stiffer larger
bore dialysis catheter is in an even worse position, abutting and tenting the
wall of the SVC at an acute angle with significant risk of pain, thrombosis,
infection, perforation, and malfunction. See text for further explanation. Fig 5 Stylized anatomical figure dividing the great veins and upper RA into
three zones (A–C), representing different areas of significance for place-
Table 4 Number of subjects and the angle to the vertical and relationship to the ment of CVCs. See text for further description. Zone A, upper RA and lower
carina for CVCs inserted from the left side. *Nine of these were so high in the SVC; Zone B, upper SVC and junction of left and right innominate veins;
innominate vein that they had not even crossed the midline Zone C, left innominate vein.

Angle to CVC at or CVC below

vertical (degrees) above carina carina
more ready to withdraw them to a safer position after post-
0–19 10* 7 insertion chest radiographs.
20–39 6 0 Schematic zones for catheter tip positioning can be
>40 27 0 categorized as shown in Figure 5.1
Zone A represents the lower SVC and upper RA. In this
and security, therefore, most CVCs are inserted to their full zone CVCs placed from the left side are likely to lie parallel
length. Even the shorter 16 cm CVC, if fully inserted from to the vessel walls. However, a part of this zone lies within
the right side, often results in the tip lying below the carina.13 the RA and therefore within the pericardial reflection.
Either shorter CVCs should be used or clinicians should be This may represent a necessary compromise for left-sided

338
 CVP catheter tip and carina
CVCs to ensure they lie parallel to the vessel wall. Right-
sided CVCs in this zone, however, should be pulled back to
zone B. The azygous vein junction with the SVC lies within
this zone and catheters may pass into this system.
Zone B represents the area around the junction of the left
and right innominate veins and the upper SVC. This is a
suitable area for CVCs placed from the right side, however
left-sided CVCs will enter this area at a steep angle (see
Fig. 4) and are at risk of abutting the lateral wall of the SVC
and should ideally be advanced into zone A.
Zone C represents the left innominate vein proximal to the
SVC. CVCs in zone C are probably suitable for short-term
fluid therapy and CVP monitoring, but not for inotrope
infusions or long-term use. The safety of this site has
been questioned.14
Instructions accompanying the packaging of CVCs state
that it is negligent to site the CVC with the tip in the RA. This
is because of the potential risk of pericardial tamponade
if the CVC tip erodes through the vessel wall below the
pericardial reflection. The upper limit of the pericardial
reflection cannot be seen on CXR, but anatomical studies
have shown that it is very unlikely to extend above the level
of the carina.9
Left-sided CVCs were less likely to be placed below the
carina—only 14% in this audit. This is presumably due to the
greater distance of the insertion point from the SVC than for
right-sided CVCs. CVCs 16 cm long may not even reach the
SVC from the left side. A high proportion of left-sided lines
above the carina had a steep (>40 ) angle to the vertical.
These can be further divided into two groups: those lying
within the left innominate vein proximal to the SVC (zone C)
and those where the tip abuts the right internal wall of
the SVC (zone B). The latter group has a higher chance
of vessel wall perforation and associated morbidity. As
the junction of the innominate vein with the SVC cannot
be identified on CXR then it is difficult to differentiate
between these groups.
In both these two groups (zones B and C), however, the
position is unsatisfactory: either because the proximal posi-
tion increases the chance of thrombosis or because there is
an increased risk of vessel wall perforation. Ideally the tip of
left-sided catheters should be sited well below the junction
of the innominate vein with the SVC (so that the CVC tip is
parallel with the SVC2) and above the pericardial reflection.
This audit shows that in practice those left-sided CVC tips
that are vertical are generally sited below the carina. This
renders the advice to site all CVCs above the carina inap-
propriate, particularly if they are inserted from the left side.
The higher incidence of vessel wall perforation seen with
left-sided CVCs can be accounted for by the increased risk
of mechanical and chemical trauma as the CVC tip enters the
SVC from the left innominate vein. The amount of travel of
the CVC within the vessel with changes in position from
supine to upright,15 respiration and arm movement16 has
been shown to be in excess of 2–3 cm. This suggests that
339
those CVCs, which are only just vertical on CXR, may not in
fact always be so and perhaps should be advanced even
further down the SVC.
These factors are less important when CVCs are placed on
a short-term basis such as for monitoring and fluid therapy in
the context of major surgery. However, for those CVCs
likely to remain in situ for a longer period of time and
those used for infusing irritant substances, such as inotro-
pes and TPN, precise placement and the incidence of
complications become more significant.
There are a number of ways to potentially improve prac-
tice. Firstly, different length catheters for right- and left-
sided placement should be stocked and used. A range of
12–30 cm should be suitable for all approaches from the
upper body in different size adults; there should be increased
awareness of the issues discussed above, including special-
ties other than Anaesthesia and Intensive Care Medicine.
Secondly, greater use of ECG guidance and radiological
screening should be made during placement of CVCs.
The limitations of ECG guidance for left-sided placements
have been recently highlighted.14 17 Thirdly, it should be
accepted that left-sided catheters may require placement
within the RA for positioning to be acceptable. We suggest
that for left-sided catheters placement at or below the
carina is more likely to result in a satisfactory tip position.
Finally, it is difficult to aseptically reinsert catheters more
deeply after the insertion procedures are complete, and
impossible if the chosen catheter is too short. Catheters
can however be withdrawn if a CXR shows it has been
inserted too far and clip on retaining wings make this
process easier. Hence it may be preferable to choose and
insert longer catheters initially but be prepared to withdraw
them after imaging.
We conclude that right-sided CVCs should generally be
sited above the carina in line with current guidelines. Left-
sided CVCs should be sited in the SVC with the tip at a
shallow angle to the vessel wall. In reality this will mean that
the tip will often require to be sited below the carina and
pericardial reflection in the upper RA.
References
1 Fletcher SJ, Bodenham AR. Safe placement of central venous
catheters: where should the tip of the catheter lie? Br J Anaesth
2000; 85: 188–91
2 Tocino IM, Watanabe A. Impending catheter perforation of
superior vena cava: radiographic recognition. Am J Roentgenol
1986; 146: 487–90
3 Duntley P, Siever J, Korwes ML et al. Vascular erosion by central
venous catheters. Chest 1993; 101: 1633–8
4 Mukau L, Talamini MA, Sitzmann JV. Risk factors for central
venous catheter-related vascular erosions. J Parenter Enteral
Nutr 1991; 15: 513–16
5 Gravenstein N, Blackshear RH. In vitro evaluation of relative per-
forating potential of central venous catheters: comparison of
materials, selected models, number of lumens and angles of inci-
dence to simulated membrane. J Clin Monit 1991; 7: 1–6
 Stonelake and Bodenham
6 Raad II, Luna M, Khalil SM, et al. The relationship between
the thrombotic and infectious complications of central venous
catheters. JAMA 1994; 271: 1014–16
7 Cadman A, Lawrance JAL, Fitzsimmons L, et al. To clot or not to
clot? That is the question in central venous catheters. Clin Radiol
2004; 59: 349–55
8 Puel V, Caudry M, Le Metayer P, et al. Superior vena cava
thrombosis related to catheter malposition in cancer chemo-
therapy given through implanted ports. Cancer 1993; 72: 2248–52
9 Schuster M, Nave H, Piepenbrock S, et al. The carina as a landmark
in central venous catheter placement. Br J Anaesth 2000; 85: 192–4
10 Albrecht K, Nave H, Breitmeier D, et al. Applied anatomy of the
superior vena cava—the carina as a landmark to guide central
venous catheter placement. Br J Anaesth 2004; 92: 75–7
11 Yoon SZ, Shin JH, Hahn S, et al. Usefulness of the carina as
a radiological landmark for central venous catheter placement
in paediatric patients. Br J Anaesth 2005; 95: 514–17
340
12 Caruso LJ, Gravenstein N, Layon AJ, et al. A better landmark
for positioning a central venous catheter. J Clin Monit Comput
2002; 17: 331–4
13 Russell WC, Parker JL. Thirteen centimetre central venous cathe-
ters, lucky for all? Anaesthesia 2003; 58: 388
14 Schummer W, Herrmann S, Schummer C, et al. Intra-atrial
ECG is not a reliable method for positioning left internal
jugular vein catheters. Br J Anaesth 2003; 91: 481–6
15 Nazarian GK, Bjarnason H, Dietz CA, et al. Changes in tunnelled
catheter tip position when a patient is upright. J Vasc Interv Radiol
1997; 8: 437–41
16 Kalso E, Rosenberg PH, Vuorialho M, et al. How much do
arm movements displace cubital central venous catheters? Acta
Anaesthesiol Scand 1982; 26: 354–6
17 Schummer W, Schummer C, Schelenz C, et al. Central venous
catheters—the inability of ‘intra-atrial ECG’ to prove adequate
positioning. Br J Anaesth 2004; 93: 193–8