Arterial lines

Getting it in is one thing. Reading it correctly is another thing entirely

Most of what goes wrong with arterial lines does not happen at insertion. It happens afterwards, quietly, in the form of a number that is wrong and believed.

A trainee who can cannulate a radial artery but cannot tell an overdamped trace from a genuine fall in pressure is more dangerous than one who cannot get the line in at all, because the second calls for help and the first starts a vasopressor. So this piece gives the insertion its due and then spends most of its length on the waveform and on troubleshooting, which is the part nobody teaches formally.

When it is worth one

Beat-to-beat pressure where the pressure is changingVasopressor or vasodilator titration, shock of any cause, a patient who will swing. This is the main one, and the value is in seeing the trend within seconds rather than every fifteen minutes.
Where the cuff cannot be trustedProfound vasoconstriction, atrial fibrillation and other irregular rhythms, morbid obesity, extensive burns, and the very oedematous patient. A cuff on a shut-down arm reads badly and reads badly in a consistent direction.
Frequent samplingRepeated blood gases in respiratory failure or ketoacidosis. If you are stabbing a patient four times a day, the line is kinder than the alternative.
Waveform-derived informationCardiac output and stroke volume variation from pulse contour analysis, and the crude but useful information in the shape of the trace itself.

Not an indication

“They are in intensive care.” A stable patient on no vasoactive support, whose gases you are not checking, does not need an arterial line, and a line that exists because everybody here has one is a line that will still be there on day nine. Ask at every ward round what decision the trace is changing.

Where to put it

Site	What you are trading
Radial	First choice for almost everyone. Superficial, easy to compress, and the hand usually has a dual supply. It is also the most peripheral, so in shock and after cardiopulmonary bypass it can read lower than central pressure — more on that below.
Femoral	Bigger, easier when everything is shut down, and closer to the aorta so more representative in severe shock. Harder to keep clean, harder to compress, and it tethers the patient to the bed.
Dorsalis pedis / posterior tibial	Useful alternatives with good collateral supply. Being furthest from the heart they exaggerate the systolic peak most — the mean is still reliable.
Brachial	Large and tempting, and an end artery at the elbow with the median nerve beside it. Many units avoid it in children and use it only when there is nothing else.
Axillary	Central pressures and a free hand, at the cost of a site that is difficult to compress and close to the brachial plexus. A considered choice, not a default.

Where not to

Through infected or burned skin. Distal to a dialysis fistula or graft. In a limb with known vascular compromise, a previous cut-down, or Raynaud's or a similar vasospastic disorder. And not in the same limb as a non-invasive cuff, or every cycle will corrupt your trace.

Coagulopathy is a relative contraindication, not an absolute one — but it changes the site you choose towards somewhere compressible, and it changes who does it.

A word about the Allen test

It is still taught and still done, and the evidence that it predicts anything is weak. It has poor reproducibility, and studies have not shown that an abnormal result predicts ischaemic complications or that a normal one prevents them. Ischaemic complications after radial lines are rare regardless.

Do it if your unit expects it and document it. Do not let a normal result reassure you into ignoring a hand that looks wrong afterwards, and do not treat an abnormal one as the only reason to choose another site. The examination that matters is the one you do the next morning.

Getting it in

Position deliberately. Wrist extended over a roll of gauze, thumb taped down in slight abduction. Do not hyperextend — past about forty-five degrees the artery flattens and flow falls, and you make a hard job harder. If the pulse disappears when you position, you have gone too far.
Use local anaesthetic in the conscious patient. It is a painful procedure, and pain causes vasospasm, which is the commonest reason a second attempt is harder than the first.
Feel the course, not just the point. Palpate along two or three centimetres so you know the direction. You are aiming along an artery, not at a spot.
Consider ultrasound from the start. It improves first-pass success, and in a shut down or oedematous patient it turns a blind procedure into a visible one. Short axis is easier to learn; long axis shows you the needle.
Enter shallow, 30 to 45 degrees. Advance slowly and watch for flashback. Blood in an artery comes back briskly and pulsatile; a lazy dribble usually means a vein.
Then choose your technique. Direct cannulation: drop the angle almost flat and advance the cannula off the needle. Transfixion: go deliberately through both walls, withdraw the needle, then pull back slowly until pulsatile flow returns and thread. Seldinger: pass the wire, and if it will not go, do not push — withdraw and start again.
Confirm before you connect. Pulsatile flow at the hub, and a proper arterial waveform once transduced. A damped, venous-looking trace at the moment of insertion is a reason to look again, not to accept it.
Secure it properly and label it. This is not cosmetic. The line must be unmistakably an arterial line to somebody who has never met the patient.

When you miss, and when it is genuinely hard

Everybody misses. The difference between an operator who gets there and one who ends up handing over a spasmed wrist and a haematoma is almost entirely what happens in the sixty seconds after the first failure.

The first rule after a miss

Take your hand off and press. Thirty seconds of firm, deliberate pressure over the puncture, every time, before you do anything else. A haematoma you allow to form is a haematoma you will then have to work through — it obscures the pulse, distorts the anatomy and makes the artery harder to feel for an hour.

The instinct is to keep the needle in and fish for it. Resist it. Withdrawing and repositioning blindly under the skin is how false passages, spasm and multiple punctures of the back wall happen.

Work out which way you missedFlashback then lost it: you were in and went through the back wall, or the cannula did not follow the needle — try transfixion next time, or come in flatter. No flashback at all: you were never in, so you are off to one side or too shallow. Slow venous flow: too deep and medial. Each of those has a different correction, and guessing wastes your next attempt.
Reset the field before you go againRe-palpate. Re-position the wrist — the roll has usually slipped and the extension has crept up. More local anaesthetic if they are conscious. New cannula, always; a used tip is burred and will not thread.
Give the spasm a minuteArteries respond to being poked by shutting down, and the pulse you cannot feel at minute one is often back at minute three. Warming the hand helps and costs nothing.
Change something, not nothingA third attempt identical to the first two is not a plan. Change the site, change the technique, change the equipment, or change the operator.
Two attempts, then hand overThe same discipline as every other cannulation. After two failures your hands are less steady and the target is worse than when you started, and the next person is more likely to succeed on their first attempt than you are on your third.

The genuinely difficult one

Profound vasoconstriction, oedema, obesity, a child, arteriopathy, or the wrist that three people have already had a go at. In roughly this order:

Warm the limb. A warm pack or warm water for a few minutes. The most under-used intervention in the whole procedure, and it turns an impalpable radial into a palpable one more often than anything else on this list.

Get the ultrasound out, and use it properly. Not as a confirmation that an artery exists — as the way you do the procedure. Short axis to find it and centre it, then long axis, or in-plane, so you watch the needle tip the whole way. Compress lightly to distinguish artery from vein, then let go, because pressing hard flattens the very vessel you are aiming at.

Use a Seldinger kit. An integrated guidewire kit forgives a tip that is only just in the lumen, where a straight cannula will not thread at all. If you have them, reach for one on the second attempt rather than the fourth.

Move site rather than persist. The other radial is untouched and the dorsalis pedis and posterior tibial are genuinely good alternatives that people forget. In shock the femoral is easier, bigger and more representative of central pressure, and choosing it early is a decision rather than a defeat.

And ask whether you need one at all right now. A cuff, a blood gas from a stab, and a senior in twenty minutes is sometimes the right answer at three in the morning.

Two techniques worth having in reserve

Transfixion. Deliberately pass through both walls, remove the needle, then withdraw the cannula slowly until pulsatile flow appears, and thread. It is the classic answer to the artery that keeps rolling or the one you can enter but never thread, and it is a technique to have practised before you need it.

Palpation with the non-dominant index finger resting just proximal to your entry point, staying there throughout. You keep the target located while you work, the same principle as the traction thumb in peripheral cannulation.

The error that causes the worst harm

Drugs given into an arterial line. It is rare, catastrophic and entirely a systems problem: distinctive labelling, no injection ports, red-line tubing, and a habit of tracing every line back to the patient before you inject anything.

If it happens: do not pull the cannula out — it is your access for treatment. Stop the injection, leave the cannula in place, give analgesia, and escalate immediately to intensive care seniors and vascular surgery. Management is time-critical and beyond the scope of a trainee working alone.

Reading the waveform

The trace is not just a number generator. Its shape carries information, and the first thing to decide about any arterial trace is whether you believe it.

The upstrokeSteepness reflects contractility and outflow. A slow, slurred upstroke with a late peak suggests a fixed obstruction such as aortic stenosis.
The dicrotic notchAortic valve closure. It sits high in a vasoconstricted patient and slides low, or disappears, in vasodilation and severe shock. Its position is a crude, free estimate of systemic resistance.
Respiratory swingMarked variation in systolic pressure with the ventilator suggests fluid responsiveness — with all the caveats about tidal volume, rhythm and spontaneous effort that make it less useful than it looks.
Pulse pressureNarrow and rising diastolic means vasoconstriction, often hypovolaemia. Wide with a low diastolic means vasodilation. This is the same reasoning the dengue piece is built on, and it applies at every bedside.

Troubleshooting, which is most of the job

Start here: the square wave test

Pull the flush for a second and let go. The trace should shoot to the top, drop sharply below baseline, and then settle.

One to two oscillations before it settles — the system is properly damped. Believe the numbers.

More than two oscillations — underdamped. The system is ringing, and the systolic is being overestimated and the diastolic underestimated. Common with long or stiff tubing, tachycardia, and a hyperdynamic circulation.

Slow, sluggish return with no oscillation — overdamped. Systolic underestimated, diastolic overestimated. This is the one that gets people started on noradrenaline they do not need.

In both cases the mean arterial pressure is the least affected number on the screen. When you are unsure whether to believe a trace, trust the MAP and treat the patient.

What you see	What to do
Damped trace, pressure falling	Before you treat the number: check the tubing for kinks, the three-way taps, and the wrist position. Look for air bubbles anywhere in the line, including at the transducer. Flush and repeat the square wave. Then look at the patient.
Positional trace — good in one wrist position, damped in another	The catheter tip is against the vessel wall or the wrist is over-extended. Reposition, resplint more neutrally, and re-tape. If it only ever works at one impossible angle, it will fail tonight.
Flushes freely but will not aspirate	Tip against the wall, a small thrombus, or spasm. Reposition, withdraw a millimetre or two, and try again gently. Never force a flush against resistance — that is how emboli go retrograde.
Blood tracking back up the tubing	The pressure bag has fallen below arterial pressure or is empty. Re-inflate to 300 mmHg. If it keeps happening, the bag has a leak or nobody is checking it.
Flat line, no trace at all	Treat as a disconnection until proven otherwise and look at the patient and the cannula: an open arterial line will exsanguinate. Then check the tap, the transducer cable and the module.
Trace looks right but disagrees with the cuff	Check levelling and zeroing first, then the square wave, then take a manual cuff yourself. If all three are right and they still disagree, see the box below.
Hand cool, mottled, painful, or with poor capillary return	Remove the line. Now. Do not wait for the morning round to decide, and document the hand at every shift change from the moment the line goes in.

Zeroing and levelling are two different things

Zeroing tells the transducer what atmospheric pressure is. Done once at setup and whenever the numbers look implausible. Levelling sets the transducer at the phlebostatic axis — fourth intercostal space, mid-axillary line — and must be repeated every single time the bed height or the patient's position changes.

The arithmetic is unforgiving: roughly 7.5 mmHg for every 10 cm of error. A transducer taped to the bed rail and left there while the bed is raised will read high and nobody will question it, because the trace looks perfect.

A beautiful waveform tells you nothing about whether the transducer is in the right place.

When the radial line is telling the truth and still misleading you

In profound vasodilation — severe septic shock, and classically in the period after cardiopulmonary bypass — a radial line can read substantially lower than central pressure. The gradient is real, not an artefact, and it can be tens of millimetres of mercury.

So the patient who looks well perfused, with warm peripheries and good urine output, whose radial line says the mean is 55, may not need more noradrenaline. If the two do not fit, consider a femoral line and compare, and treat the patient in front of you rather than the number.

Sampling, and the anaemia we cause

Waste the dead space, but only thatToo little and your sample is diluted with flush solution — a falsely low haemoglobin, a falsely low glucose, a falsely low potassium and a confusing gas. Too much, repeated six times a day, and you are bleeding the patient.
Iatrogenic anaemia is not trivialDaily phlebotomy contributes measurably to transfusion in intensive care. Use paediatric or low-volume tubes where the assay allows, and use a closed in-line sampling system if you have one.
Ask before you take itThe commonest reason an arterial gas is done is that the patient has an arterial line. That is not a reason.

How long it stays

Not on a scheduleRoutine replacement at fixed intervals is not recommended and simply creates a new procedure with a new set of risks. Change it for a reason: it is not working, the site is inflamed, or it is the suspected source of infection.
Look at the site and the hand every shiftRedness, tenderness, discharge, and the colour and warmth of the fingers. Write it down. This is the observation that finds the problem early.
Take it out when the reason has goneOff vasopressors, gases no longer being checked, stable overnight — and the line comes out. Arterial lines are unusually easy to forget because they cause so little trouble until they cause a great deal.
Press properly on removalFive minutes of firm direct pressure as a minimum, longer if coagulopathic or anticoagulated, and then look again ten minutes later. Most arterial line haematomas come from somebody pressing for two minutes and walking away.

Challenge for you

What is your tip for the arterial line that will not go in?

Everybody who does enough of these has one thing they reach for when it is not working — a way of positioning the wrist, a trick with the ultrasound, a moment when they stop and warm the hand, a rule about when to call. Most of it has never been written down anywhere, which is exactly why it is worth collecting.

Send me yours and I will publish the set, with names on them or without, as you prefer. The near misses are welcome too — the arterial line that fooled you is as instructive as the one you got.

Send me your tip

The button opens a reply straight to me. If your device blocks it, write to icuandyou@icloud.com instead.