Skip to main content

Pulmonary embolism(PE) is not that common. It appears with a frequency of less than one third that of acute myocardial infarction(AMI). More importantly about 70% or PE’s are diagnosed at autopsy. This is worrying. I wonder how many sub-clinical PE’s people are walking around with and if a significant number, do they matter?

Thos patients presenting with a PE causing haemodynamic compromise are on the wrong side of the spectrum. In this group the mortality is around 30%.

In PE as in chest pain patients, it is the lower risk group that we are most interested in, from an emergency department disposition point of view, as all others are admitted and worked up. It is the low risk or no risk we need to determine and be able to use a strategy that allows us to safely send those patients home.

As with acute coronary syndrome (ACS), risk stratification is the approach to take. Nothing reduces the risk to zero. No single investigation is that good that we can rest back and say, it’s all normal so that’s OK. A negative CT angiogram on a patient with suspected PE, still carries a risk of about 1.5% of that patient having a PE. Regardless of what we do, PERC negative, D-Dimer negative, CTPA negative, the risk can only be reduced down to about 0.5% at the most. The trick here is to reduce the risk of missing PE to a minimum, whilst also reducing potentially harmful investigations to the patient.

In medicine, there is a place for ‘gestalt’, our ability to judge that something is wrong. It means that the sum of the parts is greater to the sum of the whole. It is our ability to judge, from the end of the bed, that something is not right. It is important. We will talk about gestalt again soon.


There are many risk factors for PE, including hereditary factors such as deficiencies in antithrombin and protein C or S and there are acquired factors such as age, surgery, decreased mobility, trauma, pregnancy, the oral contraceptive pill and more. About 20%-25% of cases are said to be idiopathic. This only means that we haven’t found the cause yet.


We need to rely on our history and examination, so symptoms and signs are important to us. How helpful are they? Stein et al. Chest 1991;100:598-603 demonstrated that they may not be as helpful as we think. What is important is that 97% of patients will present with:






Can scoring systems give us the edge when we are trying to diagnose?

Probably the most well known and validated and most used is the Well’s Score.

If you have an iPhone, you can visit and get the medical calculator on which you can find the Well’s Score:

The Well’s Score then allows a calculation of pretest probability.

<2     = low probability

2-6    = moderate

>6     = high

So the Well’s Score can be thought of as covering history and examination, but it also covers ‘gestalt’. In fact, the gestalt part is worth 3 points. If you think this is a PE, then the patient falls into the moderate pretest probability group immediately.

Due to the fact that people are suspicious of our ability to use gestalt (which is actually very good), the Geneva Score was developed. Everything is now quantitative.

A score of >3 and that patient is no more in the low probability group.

The concern that I have about applying rules like this without thinking, without gestalt being involved, means that everyone who is over 65 years of age, with a heart rate of 75bpm and has a Geneva score of 4 puts them in an intermediate risk group. OK, CT’s for everybody? So the rules need to be applied to each case, rationally, when you have thought about the problem. They are not a screening tool.


There are some investigations that are continually criticised in the literature as not being of any use in pulmonary embolism. One of these is the arterial blood gas. The PaO2 is >85mmHg is 18% of patients with PE and the A-a gradient is normal in 6% of patients with PE. So based on this, some teachers say don’t do it. Here’s the problem I have with that. If you are unsure and you just can’t make your mind up i.e. your gestalt is on the borderline of 10-15% then these test may help you make a decision. Remember the PaO2 is abnormal in 82% and the A-a gradient is abnormal in 94%. So when they are abnormal, they can assist you in heading for one particular direction. The other thing to look for in the gases is the CO2. Respiratory rate is difficult to measure and a low CO2 can show you the patient is tachypnoeic.

A reminder about the A-a gradient. It measures the difference between the Alveolar and arterial concentration of oxygen. A raised A-a gradient suggests a diffusion defect and thus a ventilation perfusion mismatch.


The gas equation

At 21% oxygen and at sea level, A-a gradient = [(150-pCO2/0.8)-paO2].

For example in a patient with pCO2 of 28 and paO2 of 78, the A-a gradient = [(150-28/0.8) – 78] = 37.

A normal A-a gradient is <15 and it rises with every decade of life.

The full equation is: FiO2[(Patm-PH2O)-PaCO2/0.8] – PaO2


I still believe there is a place for these investigations and use them, to assist me in making decisions.



This is the most loved/hated test, depending on how it is used. If used incorrectly, it can mean more radiation exposure to the patient.

The quantitative test is better. The way to use the test is in a patient where you have a low pre-test probability. If the D-Dimer is normal and the Well’s Score is <4, that patient can be ruled out.

Here is the catch – if the D-Dimer is high, and you did it thinking that the patient may have a PE, then you need to investigate further.

Who not to do a D-Dimer on, as it will be elevated:

1.  Age >70

2.  Patients with malignancy

3.  Surgery within 7 days

4.  Pregnancy


If your gestalt is high, there is about an 80% chance the D-Dimer will be elevated. Again, remember that if you did it to rule out a PE and it comes back elevated, you cannot ignore it, you are committed to investigating that patient further.



There has been much discussion about the PERC (pulmonary Embolism Rule-out Criteria) rule. Some critics say it’s not of any use, as it’s mostly common sense. I agree that it’s all common sense, but I see it as something else that I can use, that is validated and accepted to help me in my decision making about who to send home. It has a low specificity but high sensitivity (97.5%). It allows us to rule out patients where your gestalt is low i.e. <15%. In this group, where your gestalt is <15% and they are PERC negative, then no further testing is necessary i.e. no D-Dimer needed, it’s not a PE. Remember, it may be something else.


This is the PERC rule:
If all the above are negative and your gestalt is <15%, you’re done.The chance of a false negative rate is <2%.

But you may ask, “If the rate of missing a PE with a CTPA is about 1.5% or less, why don’t we just do a CTPA on everyone?”

There are two main reasons:

1.  There is an 8% rate of contrast nephropathy. It is difficult to pick who will get this. In one study, those with a normal creatinine were the ones that developed the nephropathy.

2.  Increased risk of malignancy. This may be as high as 1 in 500 lifetime risk at 40 years of age.

What about V/Q Scans?

There is controversy in terms of radiation dose. Certainly the literature is supporting as decreased dose of radiation with new age CT scanners as opposed to V/Q scans. The V/Q is really of greatest use in non-smoking patients with no COPD. The scan also has to be read as normal. If abnormal, the patient will then need a CT scan.

What about the pregnant patient?

There is a higher risk of PE in pregnancy. The issues that arise are related to radiation doses. The D-Dimer increases in pregnancy, so it helps us less. It increases each trimester and there are suggestions that we increase the D-Dimer threshold that is considered acceptable to rule out PE. I’m a little concerned about doing this, as it’s not really mainstream as yet.

What about CTPA vs V/Q? There appears to be greater fetal exposure with a V/Q as the contrast sits in the bladder. There is higher maternal breast exposure with a CTPA and shielding the CT patient may increase the scatter and increase the fetal exposure.

The approach I ususally recommend is that if you are suspicious of a PE in a pregnant patient, first ultrasound the legs. If the ultrasound is normal do a half dose perfusion V/Q scan. If the perfusion is normal, stop there. This will reduce exposure and is an approach well established in most centres. There will be an approach at your hospital, please be sure to know what this is.

I hope the above helps a little, in working out how you will approach the patient with the potential diagnosis of PE.


  • Siew-Lee Thoo says:

    Thanks for this. Why don’t you list infection & inflammatory disorders as causes of raised D-dimer?

  • Matt Davis says:

    Hi Peter,

    With regards to the pregnant patient…

    CTPA with shielding is purported to reduce foetal radiation risk (but still carries risk to mum and iodine load to both).

    There has also been a suggestion that V/Q with IDC may be the best way to minimise the foetal radiation exposure.

    Wondering what others are doing in this group when less invasive tests are inconclusive?



Leave a Reply