Arterial Lines and Central Venous Pressure
How to read the two waveforms you will stare at most in the ICU, what they are actually telling you, and the artifacts that make them lie.
Arterial lines The a-line
An arterial line does two things: it gives you a beat-to-beat blood pressure, and it gives you a place to draw blood gases from without sticking the patient again. The second reason is why most of them are still in on day four.
When one is genuinely indicated
- Any patient on a vasopressor or a vasodilator infusion. Titrating norepinephrine off a cuff that cycles every fifteen minutes is guessing.
- Hemodynamic instability of any cause, where the trend matters more than the number.
- Frequent arterial blood gases, meaning more than a few a day.
- Situations where a cuff is unreliable: severe obesity, profound vasoconstriction, an irregular rhythm at a fast rate, or a patient in whom the cuff and the clinical picture disagree.
Sites, and how to choose
The default. Superficial, compressible, collateral supply from the ulnar artery, and the lowest complication rate. The tradeoff is that it is the most peripheral site, so it exaggerates the systolic pressure and is the first to damp when the patient is vasoconstricted.
Larger, easier in shock, and it reads closer to aortic pressure, which matters when you are making decisions on mean arterial pressure in profound vasoconstriction. Higher infection risk, harder to keep clean, and it tethers the patient to the bed.
Reasonable in experienced hands, but it is an end artery with poor collateral supply at the elbow. Thrombosis there threatens the whole forearm. Many units avoid it for that reason alone.
Fallbacks. Dorsalis pedis overestimates systolic pressure substantially. Axillary reads centrally and is used in long-term support patients, but carries a cerebral embolization concern on the left.
It is traditional, it is quick, and the evidence that it predicts ischemic complications is weak. A normal Allen test does not make a radial line safe and an abnormal one does not reliably predict harm. Do it if your unit requires it, but do not let a reassuring result substitute for checking the hand afterward, which is the thing that actually catches problems.
Reading the arterial waveform
Most people look at the number and ignore the trace. The trace carries information the number does not.
Steepness reflects contractility and the rate of ventricular ejection. A slow, slurred upstroke with a delayed peak is the pattern of severe aortic stenosis. A very steep upstroke with a rapid collapse suggests a wide pulse pressure state: aortic regurgitation, sepsis, anemia, thyrotoxicosis, an arteriovenous fistula.
Aortic valve closure. Its position on the downstroke tracks systemic vascular resistance. A high notch means high resistance, a vasoconstricted patient. A notch that sits low and shallow, almost lost, means low resistance, which in a hypotensive patient points hard at distributive shock. In severe vasodilation the notch can disappear entirely.
Systolic minus diastolic. A narrow pulse pressure in a hypotensive patient suggests low stroke volume: cardiogenic shock, tamponade, hypovolemia. A wide pulse pressure points to vasodilation or aortic regurgitation. It is one of the fastest bedside discriminators you have.
In a passively ventilated patient in sinus rhythm, systolic pressure variation across the respiratory cycle reflects preload dependence. Large swings suggest the patient will respond to fluid. See fluid responsiveness below, including the substantial list of conditions that invalidate it.
As the pulse wave travels distally it narrows and amplifies. A dorsalis pedis line may read a systolic pressure 20 to 30 mm Hg higher than the aorta, and radial reads higher than femoral. The mean arterial pressure changes far less along that path, which is one of several good reasons resuscitation targets are written in MAP rather than systolic pressure.
The exception worth knowing: in profound shock on high-dose vasopressors, radial pressure can under-read central pressure substantially. If the radial line says the MAP is 55 and the patient is warm, making urine and mentating, get a femoral reading before you escalate the pressor.
Damping, and the square wave test
Every arterial trace is a physical system: a column of fluid in stiff tubing connected to a transducer. Get the mechanics wrong and the numbers are wrong, and the trace tells you which way.
The trace looks rounded and sluggish, the dicrotic notch is lost, the upstroke is slow. Systolic reads falsely low and diastolic falsely high. Mean stays roughly accurate. Causes: air bubbles in the line, a kink, a clot at the tip, tubing that is too long or too compliant, a loose connection, a partially closed stopcock, or the catheter tip lying against the vessel wall.
The trace shows a spiked, exaggerated systolic peak with ringing artifact after it. Systolic reads falsely high, sometimes dramatically, and diastolic falsely low. Mean again stays roughly accurate. Causes: stiff non-compliant tubing, tachycardia, high-output states, catheter whip. This is the one that gets a patient treated for hypertension they do not have.
The square wave test, and how to actually read it
Pull the flush valve for one second and let it snap back. The trace shoots to the top of the scale, then drops.
Normal: a sharp drop below baseline followed by one, or at most two, small oscillations before settling.
Overdamped: the trace slides back to baseline with no oscillation at all. Look for a bubble, a kink or a clot.
Underdamped: more than two or three oscillations, ringing on for several cycles. Consider shorter, stiffer tubing, or a damping device.
Do this every shift and before you believe any surprising pressure. It takes four seconds.
Zeroing tells the transducer what atmospheric pressure is. Do it once at setup and after any disconnection.
Leveling is positioning the transducer at the phlebostatic axis, the fourth intercostal space at the mid-axillary line. This is the one that gets missed. Every 10 cm the transducer sits below that axis adds roughly 7.5 mm Hg to the reading, and every 10 cm above subtracts the same. A transducer taped to the bed rail after the head of the bed was raised is the most common cause of a blood pressure that suddenly changed for no reason.
Re-level after every position change. If a pressure moves and the patient looks the same, check the transducer before you change the drips.
Arterial line complications
- Temporary occlusion is common and usually silent. Permanent ischemic damage is rare, on the order of 0.1 percent at the radial site, but it is catastrophic when it happens.
- Check the hand every shift. Color, capillary refill, temperature, sensation. A cold, mottled, painful hand distal to a radial line means the line comes out now, not after rounds.
- Infection is less frequent than with central lines but not zero, and rises with duration and with the femoral site.
- Accidental drug administration into an arterial line can cause distal necrosis. Label it clearly and differently from every venous line at the bedside.
- Bleeding. An arterial line that disconnects can exsanguinate a patient quickly and quietly under a blanket. Keep the connections visible.
- Pseudoaneurysm and AV fistula are late and uncommon.
Central venous catheters
Placed for vasopressors, for drugs that will destroy a peripheral vein, for dialysis or plasmapheresis, for poor peripheral access, and for pressure monitoring. Note that pressure monitoring is last on that list, which is a change from how these were taught twenty years ago.
Easiest under ultrasound, compressible if you hit the carotid, lowest thrombosis rate. Uncomfortable for the patient and awkward in a beard or a cervical collar.
Lowest infection and thrombosis rate of the three, most comfortable long term, and the site of choice for a line staying in a while. The tradeoff is pneumothorax risk and non-compressibility, so it is a poor choice in coagulopathy.
Fastest in an emergency and the safest in a coding patient because you stay out of the way of the airway and the chest. Highest infection and thrombosis rate, so plan to move it.
Tip position matters. The target is the lower superior vena cava at or just above the cavoatrial junction, which on a chest film sits roughly at the level of the carina. Too deep risks arrhythmia and, rarely, perforation into the pericardium. Too shallow risks thrombosis and unreliable pressures. Confirm on the film and write down what you saw.
CVP, and what it actually means
Central venous pressure is the pressure in the thoracic vena cava near the right atrium. It is right atrial pressure, which in the absence of tricuspid disease is right ventricular end-diastolic pressure. Normal is roughly 2 to 8 mm Hg.
It is a poor predictor of whether a patient will respond to fluid. Multiple systematic reviews have found essentially no useful relationship between a static CVP value and the change in cardiac output after a fluid bolus. The correlation is close to what you would get flipping a coin.
This is not a technicality. Whole generations were taught to bolus until the CVP reached 8 to 12, and that target has been abandoned in sepsis guidance for exactly this reason. A CVP of 4 does not mean the patient needs fluid. A CVP of 14 does not mean they will not respond to it.
So what is it good for? Quite a lot, as long as you use it as a pressure rather than as a volume:
- A high CVP is a real finding. It tells you the right heart is failing, obstructed, or overloaded, and it tells you the venous pressure the organs are draining against. A CVP of 20 in a patient with an acute kidney injury is a plausible cause of that injury, because renal perfusion depends on the pressure gradient across the kidney and you have just raised the back pressure.
- The trend within one patient is more useful than any single value, particularly when it moves in response to something you did.
- The waveform carries diagnoses that the number does not. See below.
- Extremes are informative. A CVP of 1 in a shocked patient makes hypovolemia likely. A CVP of 25 makes tamponade, massive PE, RV infarct or severe tricuspid regurgitation the question.
Measure it at end-expiration, where pleural pressure is closest to atmospheric, and read it off the waveform rather than the digital average. On a ventilated patient end-expiration is the trough of the respiratory swing; on a spontaneously breathing patient it is the peak. Positive pressure ventilation and PEEP both raise the measured CVP without changing the volume status underneath it.
The CVP waveform
Three positive waves and two descents, and each one has a mechanical cause.
| Component | Cause | What changes it |
|---|---|---|
| a wave | Atrial contraction | Absent in atrial fibrillation. Large "cannon a waves" when the atrium contracts against a closed tricuspid valve: complete heart block, ventricular tachycardia, junctional rhythm. Tall a waves in tricuspid stenosis, pulmonary hypertension, RV hypertrophy. |
| c wave | Tricuspid valve bulging into the atrium at the start of ventricular systole | Often small and hard to see. |
| x descent | Atrial relaxation and downward pull of the annulus during systole | Prominent in tamponade. Blunted in tricuspid regurgitation. |
| v wave | Atrial filling against a closed tricuspid valve | Giant v waves are the signature of significant tricuspid regurgitation, where the wave merges with the c wave and produces a broad systolic hump. |
| y descent | Tricuspid opening and rapid ventricular filling | Blunted or absent in tamponade. Steep and deep in constrictive pericarditis and in severe tricuspid regurgitation. This one distinction is worth memorizing. |
Tamponade: preserved or prominent x descent, absent y descent. Filling is impaired throughout diastole, so there is no rapid early filling phase.
Constriction: both descents preserved and steep, producing the M or W shape. Early filling is unimpeded until the ventricle hits the rigid pericardium and stops abruptly.
If you find yourself squinting at a CVP trace in a hypotensive patient, this is the thing you are looking for.
Predicting fluid responsiveness
Since static pressures do not answer the question, the useful measures are dynamic: they ask what happens to stroke volume when preload is deliberately changed.
The variation in pulse pressure across a respiratory cycle. Above roughly 13 percent predicts fluid responsiveness reasonably well. Stroke volume variation behaves similarly.
The most broadly valid test, because it works in spontaneous breathing and in atrial fibrillation. Lay the patient flat and raise the legs to 45 degrees, which autotransfuses roughly 300 mL. Measure the change in cardiac output or stroke volume within a minute. A rise of 10 percent or more predicts a response. The catch is that you need a real-time cardiac output measure to see it, because blood pressure alone is an insensitive surrogate.
Give a defined volume over a defined time and measure the effect. Crude, but honest, and it is what most units actually do.
It requires all of the following: a passively ventilated patient making no respiratory effort, a regular rhythm in sinus, a tidal volume of at least 8 mL/kg, an open chest excluded, and no severe right ventricular failure or raised intra-abdominal pressure.
Which describes almost nobody on a modern lung-protective ventilator at 6 mL/kg in atrial fibrillation. Know when the number does not apply, because the monitor will display it regardless.
And the point that gets lost: fluid responsiveness is not the same as needing fluid. Roughly half of healthy people are fluid responsive at any moment. The question is whether raising cardiac output will help this patient, and whether the volume you give to do it will end up in their lungs.
Troubleshooting, in order
The pressure suddenly changed and the patient looks the same
Check the transducer level before anything else. Then check for a bed position change, a new bubble, a loose connection, a stopcock left half-turned, and the arm position. Do a square wave test. Only then start believing the number.
The arterial trace is flattening
Flush and watch. If it improves and then decays, suspect a clot forming at the tip or the tip against the wall. Reposition the wrist. Check the pressure bag is inflated to 300 mm Hg and the flush bag is not empty, which is the single most common cause of a slowly dying a-line.
The CVP is high and you cannot explain it
Work through: is the transducer level. Is there PEEP or auto-PEEP. Is the patient straining, coughing or breath-stacking. Is the tip in the right atrium or too deep. Then the real causes: right ventricular failure, pulmonary embolism, tamponade, tension pneumothorax, severe tricuspid regurgitation, volume overload, raised intra-abdominal pressure.
Next: the Swan-Ganz catheter, which measures what these two cannot, and mechanical circulatory support for when the numbers say the heart is not going to manage on its own. For the respiratory side, see ventilator settings.
This is not medical advice. It is a teaching outline for clinicians and clinicians in training. Insertion technique, anticoagulation and device management are institution-specific. Follow your own protocols, and never act on a number you have not confirmed is real.