Best Evidence Topics
  • Send this BET as an Email
  • Make a Comment on this BET

In Pulmonary Embolus is the Aa gradient a useful diagnostic tool?

Three Part Question

In [a patient with a suspected PE] is [an abnormal aa gradient] a [reliable indicator of a PE]

Clinical Scenario

A patient attends the Accident and Emergency department with a good clinical history of pulmonary embolus. One wonders two things: - 1) Does a normal arterial alveolar gradient excludes the suspected diagnosis? 2) Does an abnormal arterial alveolar gradient, which confirms the diagnosis necessitate further investigations?

Search Strategy

PubMed on the worldwide web
[{arterial alveolar gradient}] and [{pulmonary embolus}]

Search Outcome

37 papers were found. 33 were excluded on the basis of irrelevance, use of other parameters i.e. not solely the Aa gradient, and use of animal subjects. Papers not in English and papers whose population was strictly geriatric or pregnant, were also excluded.

Relevant Paper(s)

Author, date and country Patient group Study type (level of evidence) Outcomes Key results Study Weaknesses
McFarlane MJ, Imperiale TF
The derivation set came from records of all patients records at the Cleveland MetroHealth Medical centre who received a V/Q scan for suspected PE in 1988-89. Validation set came from patients who had V/Q scans between 1987-90. Demographic and clinical data was obtained. A normal A-a gradient was defined as less than or equal to Age/4 + 4. Of 873 in the derivation set, 540 had simultaneous room air ABG's. Among 805 in the validation set, 489 had simultaneous room air ABG'sRetrospective data analysis Statistical analysis - Chi squared and Fisher's exact test for categorical data. Students t-test for continuous data.Derivation set. PE in72 of 540 (13%) with V/Q scan, acceptable ABG's had a normal Aa gradient.5 of 72 (7%) had a PE. 4 of the 15 patients with normal Aa gradientwho had a previous PE/DVT had a PENot every patient had gold standard of angiography. Study may perform better due to low prevalence of condition. Retrospective analysis dependent on clinicians recording of data. Incorporation bias with respect to ABG results
Validation set. 75 of 489 (15%) had PE. Among 384 with no previous PE/DVT 54 (14%) had a normal Aa gradientOnly 1 out of 54 with normal Aa gradient had a PE
Stein PD, Goldhaber SZ, Henry JW
January 1995
Patients were taken from the multicentre collaborative study of PIOPED (Prospective Investigation of Pulmonary Embolism Diagnosis). All patient data was utilised, including those who were not randomised for angiographic pursuit, but who had pulmonary angiograms if requested by the attending physician. The diagnosis of PE was established via pulmonary angiography. There were 280 patients with PE who had an A-a gradient calculated on room air. For comparison patients with suspected PE were evaluated where the diagnosis was excluded by angiography. There were 499 who had an A-a gradient measured on room air. Patients were then categorised as (1) entire cohort, (2) no prior cardiopulmonary disease and no prior PE, and (3) no prior DVT or PERetrospective data analysisNormal A-a gradient defined as less than or equal to 20mmHg11-14% of patients with PE had a normal A-a gradient.No relative weaknesses found - a comprehensive study exploring many avenues, expanding on a previous multicentre study
Normal A-a gradient using the equation age/4+48-10% of patients with PE had a normal A-a gradient
Normal A-a gradient using age related values from literature20-23% of patients with PE had a normal A-a gradient
A-a gradient showed a linear correlation with the severity of PEAssessed by pulmonary artery mean pressure and number of V/Q defects, and statistical analysis
Overton DT, Bocka J
July 1988
William Beaumont Hospital, Royal Oak, Michigan between February 1967 to January 1987. Review of all charts of patients undergoing pulmonary angiography. Excluded were patients undergoing angiography for reasons except PE, no ABG results at room air, negative pulmonary angiography, lack of medical records.Retrospectve data analysis64 of 194 patients met criteria. 3 patients had a normal Aa gradient3 pateints who had a normal Aa gradient also had a PELimited criteria preventing analysis of many numbers of patients. Study spanned 2 decades where diagnostic and clinical approaches can differ. Population chosen not representative of the population of patients as a whole.
Jones JS, Neff TL, Carlson SA
All emergency department patients who underwent pulmonary angiography for suspected PE at Butterwirth hospital January 1992-December 1995. Exclusion criteria - patients on supplementary oxygen (unless removed for 20 mins), no information about prior cardiopulmonary diseaseRetrospective data analysis Blinded data analysis and review of accuracy of data analysis152 of 197 selected. 59 had demonstrated emboliAa gradient failed to exclude PE in 35% of patients without prior cardiovascular disease and 25% with prior cardiovascular disease. Aa gradient shosed a linear correlation with severity of PEselection bias using only patients who had pulmonary angiography. Omission of more seriously ill patients


A-a gradient is a simple test often misunderstood and underused. Its limitations as a specific indicator for pulmonary embolism are highlighted in the literature, however this is also one of its strengths. Essentially it is a test of V/Q mismatch and in the right clinical context can provide valuable information on the severity of the condition. The highlighted studies are some years old, and are all retrospective data analyses. In order to extrapolate the full diagnostic usefullness, sensitvity and specificity a prospective study must be conducted. In the interim it remains a useful tool in the armoury of a judicious clinician.

Clinical Bottom Line

A-a gradient is a useful bedside parameter to aid in the diagnosis of pulmonary embolus. It has more relevance in patients without pre-existing cardiopulmonary disease, and previous PE/DVT. Clinical suspicion is paramount, and a normal A-a gradient by no means excludes the diagnosis or should prevent further diagnostic tests. It may however decrease the need of further investigations in some patients.


  1. McFarlane MJ, Imperiale TF Use of the alveolar-arterial oxygen gradient in the diagnosis of pulmonary embolism Am J Med 1994 Jan;96(1):57-62
  2. Stein PD, Goldhaber SZ, Henry JW Alveolar-arterial oxygen gradient in the assessment of acute pulomonary embolism Chest 1995 Jan;107(1):139-43
  3. David T. Overton, Joseph J. Bocka The Alveolar-Arterial Oxygen Gradient in Patients With Documented Pulmonary Embolism Arch Intern Med 1988 Jul;148(7):1617-9
  4. Jeffery S. Jones, Timothy L. Neff, Scott A Carlson Use of the AlveolarAarterial Oxygen Gradient in the Assessment of Acute Pulmonary Embolism Am J Emerg Med 1998 Jul;16(4):333-7