ABG Analysis Case 1

Answer the questions at the bottom of the page when all data has been reviewed

History & Examination

A wealthy 60 year old divorcee is taken ill at the National Gallery and is brought by ambulance to Hospital. A US national, she is in London on vacation. She had flown first class from Galveston, Texas, into Heathrow some days earlier. It had been a very stressful time in her life, her estranged mother having died a few weeks before the trip. In addition, her son was on active duty in the Middle East with the US military at the time of her visit.

She is fully communicative, clearly intelligent and likeable. A lifelong non-smoker and former sportswoman, she is statuesque and is normally fit and well. She had felt ‘fine’ on the morning of presentation but experienced a number of episodes of dizziness and palpatations associated with severe anxiety while in the National Gallery. She had never experienced anything like this before. With initially low oxygen saturations, the ambulance crew had started her on oxygen which was continued in ER/ED. She reported no other symptoms. She is convinced to stay pending the results of some tests.

Respiratory examination was normal with vesicular breath sounds throughout both lung fields. Percussion note was resonant throughout. No murmurs or added sounds present on CVS examination. Abdominal and neurological examinations were normal.

The laboratory investigation results are complete by the next shift. Following review of the investigations below, the doctor on that shift diagnoses ‘stress-related’ symptoms. The patient is reassured and discharged. Please review the case and answer the questions below.


Sylvia Avery bloods now




Sylvia Obs anon


Sylvia fluids anon


Q1. What abnormalities, if any, are present on the tests carried out at presentation?

Q2. What diagnosis should have been considered at initial presentation?

Q3. What imaging test would have been useful in making/excluding the diagnosis in Q2?

Q4. Is there any other test that should have been carried out at presentation?

Q5. Assuming we have missed the diagnosis in Q2, what is the risk to the patient?


Q1. What abnormalities, if any, are present on the tests carried out at presentation?

Sylvia ABG CSi

Her ABG result is grossly abnormal. On casual glance, all values reported on her ABG appear to be normal. The second doctor may have missed the fact that she was on oxygen (4L) when the ABG sample was taken (14.30 PM, see observation chart). The FiO2 has not been recorded on the readout. The paO2 on an ABG must be interpreted in terms of the level of inspired oxygen and the degree of alveolar ventilation (the paCO2). In this case, her A-a gradient is markedly elevated.

(remember mmHg = kPa x 7.5)

The predicted A-a gradient for her age is (Age/4 + 4) mmHg = 19 mmHg

This needs to be corrected for the FiO2 (see A-a gradient article below) giving a predicted value of 30 mmHg

Her alveolar pO2 (pAO2) on 36% oxygen  is 0.36(760 – 47) – 35/0.8 = 213 mmHg (See A-a gradient article below)

Her A-a gradient is, therefore, 213 – 85 (pAO2 – paO2) = 128 mmHg

This grossly elevated A-a gradient is difficult to explain in a woman who has never smoked and was an athlete. In addition, her CXR is normal.

Her ECG shows a non-specific abnormality-a sinus tachycardia. It may be non-specific but it is an abnormality nonetheless and requires some explanation.

Q2. What diagnosis should have been considered at initial presentation?

Pulmonary embolism (PE)

This lady presented following a syncopal episode. Syncope is a very common presentating complaint in ED and frequently no underlying cause is diagnosed. Identifying PE as a cause of syncope is a challenge. Without subjecting large numbers of patients to unnecessary investigations, occasional cases were PE is the cause of syncope will always be missed. However, there were clues in this case. Even though this stoical woman complained of no respiratory symptoms at presentation, she had been started on oxygen because of a low SpO2 and did have an ABG taken. Evidence of a problem was present. Analysis of her ABG shows a markedly raised A-a gradient. If the ABG had been analysed properly, this finding would have had to have been explained. She has recently been on a long haul flight (Galveston to London is 10 hours). The relationship between flying and venous thromboembolism (VTE) is controversial but there does appear to be an increased risk of VTE even in individuals flying first class (ie even people with leg room!).

Q3. What imaging test would have been useful in making/excluding the diagnosis in Q2?

CT Pulmonary Angiogram (CTPA)

CTPA is the imaging method of choice to exclude the diagnosis of PE and the protocol can be modified to also exclude underlying DVT. A negative CTPA has a high negative predictive value for PE. Most patients with a negative CTPA can be discharged safely without anticoagulation.

Q4. Is there any other test that should have been carried out at presentation? 

A serum D-dimer level. It might have helped but I suspect that if the doctor had thought of the diagnosis of PE the patient would have been imaged anyway.  A pathological clot anywhere in the body may result in a rise in the level of breakdown products of fibrin in the systemic circulation (elevated D-dimers). The interpretation of a D-dimer test result depends on the pre-test probability of pulmonary embolism (PE) in a patient. The ‘Wells score’  (see below) may be used to assess the pre-test probability of PE.

In patients with low or intermediate probability of PE, a negative D-dimer (at or below normal concentration) effectively rules out the diagnosis. At high probability, it does not. Cases of PE with negative D-dimers have been described. Also, remember that diverse conditions unrelated to PE can elevate the D-dimer level. Therefore, elevated D-dimers have a poor positive predictive value for diagnosing PE. A D-dimer result must be interpreted in clinical context.

In reality, if, with experience, your ‘gut-feeling’ on detailed and careful consideration of a case and all associated findings is that a patient may have had a PE, you should image them. If, on the other hand, your instincts tell you that they have not, a negative D-dimer result is reassuring for the doctor.

The only doctors who don’t make mistakes are the ones who do nothing or who wont take responsibility. The purpose of these cases is not to frighten you into investigating everything. As we will see on this site, over-investigation can lead to catastrophic problems! However, this patient’s situation might have been avoided by asking at initial presentation, does everything add up to a benign cause? How can we explain the tachycardia and most significantly the grossly elevated A-a gradient in the presence of a normal chest X-ray. Particularly as she is a non-smoker.

Finally, it is worth noting that a person with a PE may have a completley normal ABG (even the A-a gradient may be normal) and a negative D-dimer. In reality, missing PEs can be a very difficult trap to avoid.

With the benefit of hindsight, if we calculate our patient’s Wells score at presentation: that is if we had detected and were looking for a cause of the ABG abnormalities, she is tachycardic (+1.5), in the presence of a normal CXR, then PE must be as likely as any other cause (+3). Her lower limbs were not examined. So, minimum total score = 4.5 (intermediate probability)

Well's criteria now


Q5. Assuming we have missed the diagnosis in Q2, what is the risk to the patient?

‘We are constantly misled by the ease with which our minds fall into the ruts of one or two experiences.’  Sir William Osler.

Controversial. The generally quoted figure for mortality in ‘missed’ PE is between 20% and 30%. This is based on studies from several decades ago. These studies have methodological problems and recent re-analysis of the limited data available suggest that the actual figure may be less than 5%. This is thought provoking as it makes you wonder whether treatment has any great impact at all! However, if you ever see the scenario described here unfold, it will stay with you.

Subsequent Course


Twenty four hours after leaving the hospital she collapses in the street. CPR is immediately commenced by a trained bystander. When an ambulance crew arrive, no pulse is present. The crew continue with CPR on the journey back to the Hospital.

The crew call ahead and report bradycardia and right bindle branch block on the ECG readout in the ambulance. No cardiac output is associated with this rhythm. The patients notes from the previous day are available to a senior ED doctor for review. On arrival in ED, intravenous thrombolysis is immediately administered. After a further hour and repeat administration of IV thrombolytics there is no recovery in output and CPR is stopped with the agreement of all present. Autopsy confirmed the presence of multiple bilateral peripheral pulmonary emboli and a recent large saddle embolism at the bifurcation of the pulmonary artery.


The A-a gradient: an essential part of ABG analysis

The normal A-a gradient

When analysing a paO2 value on an arterial blood gas we must factor in both the degree of alveolar ventilation and the level of inspired oxygen (the FiO2). This is achieved by calculating the ‘A-a gradient’. It is well worth getting to grips with this topic.

In order to ensure elimination of all CO2 produced during metabolism, centers in the brainstem control the volume of air per unit time entering the alveoli (‘the alveolar ventilation’). This is controlled by altering the depth and rate of respiration. In addition to the elimination of CO2, a normal level of alveolar ventilation maintains the partial pressure of oxygen in the alveoli (the pAO2) at a level well above that required to saturate hemoglobin in the arterial blood.

At a given level of alveolar ventilation the pAO2 is also critically dependent on the level of inspired oxygen in inhaled air. Normally 21%, we can increase this and consequently the pAO2 by adding oxygen to the inhaled air.

Diffusion of oxygen across the alveolar-capillary membrane is not a perfectly efficient process. Due to normal physiological effects, the level (i.e concentration = partial pressure) of oxygen in the arterial blood returning to the left ventricle from the lungs (the paO2) never quite reaches the level of the pAO2. There is a difference, (a ‘gradient’)  between the alveolar (pAO2) and the arterial (paO2) concentration of oxygen. pAO2 – paO2 is termed the A-a gradient. There is a normal range of values for the A-a gradient depending on the subjects age and the level of inspired oxygen. In the presence of disease, the A-a gradient may rise above this normal range. Therefore, if we detect an A-a gradient above the normal expected range on the basis of the patients age we must explain this finding.

How do we calculate the alveolar pAO2?

We can calculate the pAO2 in mmHg from the values recorded on the ABG using the following equation (‘the alveolar gas equation’).

pAO2 = FiO2(atmospheric pressure – vapour pressure in the airways) – paCO2/respiratory quotient

At sea level breathing room air:

pAO2 (mmHg) = 0.21(713) – paCO2(mmHg)/0.8

Note: in this equation FiO2 is expressed as a decimal fraction (21% = 0.21).

Note: If you normally work in kPa (eg UK and Ireland), it is very useful to be able to work in the US units in these calculations. It is simple, mmHg = 7.5 x kPa.

The effects of oxygen therapy

In a patient, we can adjust the FiO2 and, therefore, the pAO2, by mixing oxygen with the inhaled room air. Our patient was receiving 4L of oxygen per minute  (we presume via nasal prongs as this is usually the case for patient comfort). What FiO2 does this generate? You will need to know the relationship between litres of oxygen administered per minute (via nasal prongs) and FiO2. The relationship is shown below (‘+3 then all +4’)
o2 litres2

(40% or thereabouts: maximum achievable with nasal canula)

 Calculating the A-a gradient in our patient 

Sylvia ABG CSi

So, in our patient

pAO2 = 0.36(713) – 35/0.8 = 213 mmHg

Therefore, observed A-a gradient = (pAO2 – paO2) = 128 mmHg

Calculation of the predicted A-a gradient in our patient

Predicted A-a gradient is given by:

Age/4 + 4 = 19 mmHg (if breathing room air)

However, the A-a gradient rises stepwise with increases in FiO2 in normal individuals. We need to add 7.5 mmHg to the predicted A-a gradient for every 10% increase in the FiO2.

So, predicted A-a gradient for a 60 year old woman on 4L (36%) O2 should be close to:

(60/4 + 4) + 11 = 30 mmHg

Despite a superficially normal ABG, our patient had a markedly elevated A-a gradient. In effect, this analysis is telling us that in a lady of this age, breathing this concentration of oxygen (FiO2) at this level of alveolar ventilation (paCO2), we should expect a much higher paO2. Something is reducing the level of oxygenation of the blood flowing through the lungs to the left ventricle. In this case that something was probably a pulmonary embolism (or emboli).

The calculations above look onerous but can by carried out by apps available on your mobile telephone device. Usually it makes little difference but very occasionally, it makes all the difference.