Interpretation of the Anion Gap: Oxygen Pressure Field Theory VIII

OPFT Part VIII:  Interpreting the Anion Gap

The anion gap is the difference in the measured cations and the measured anions in serum, plasma, or urine. The magnitude of this difference (i.e. “gap”) in the serum is often calculated in medicine when attempting to identify the cause of metabolic acidosis. If the gap is greater than normal, then high anion gap metabolic acidosis is diagnosed.

The term “anion gap” usually implies “serum anion gap”, but the urine anion gap is also a clinically useful measure.

(Click Image to View Power Point Presentation)

Authored by:  Gary Grist,  BS, RN, CCP, LCP   

To View all of Gary Grist”s Posts Regarding OPFT-  click here

Mr Gary Grist delivered a seminar on this topic at The Missouri Perfusion Society 16th Annual Meeting titled “Beyond The Fick Equation”  on June 10-12, 2011

Interpretation of the Anion Gap

 The concept that a single number (the anion gap) can indicate the presence of abnormal cell metabolism is quite elegant  (elegance is simplicity carried to the extreme).  Unfortunately, in the real world, nothing is simple.  So too with the anion gap, it is a very controversial number. People usually have one of three opinions about it; it is a very useful marker, it is of no use whatsoever or many people (most perfusionists) have never heard of it. The first controversy centers on the value of a “normal” anion gap.

In lab text books, the normal value and range of the anion gap is 12 +/-4 mEq/L. This means that about 2/3’s of healthy human beings will have an anion gap (AG) of from 8 – 16 mEq/L. However, there are those who contest this value and suggest that the true range for the normal AG is 3 – 11 mEq/L. The argument is based on the fact that when the range of 8-16 mEq/L was set, lab testing of electrolytes was done by flame photometry.

In the last 10 years, however, most labs began testing electrolytes using ion selective electrodes.  The ion selective electrodes give a slightly different value than flame photometry for electrolytes, particularly for Cl ions.  Since the AG is calculated from the electrolyte measurements, this has resulted in a change of the normal AG range from 8 – 16 mEq/L down to 3 – 11 mEq/L.  So, depending on which range you subscribe to (the old range or the new range), an AG of 16mEq/L would be normal or grossly abnormal.

At Children’s Mercy Hospital in Kansas City, Missouri, the perfusionists wanted to know what the true AG range was for normal, healthy children.  We looked at the AG value for about 100 children scheduled for elective heart surgery.  These were children in no apparent distress who either were within the normal growth curve or just beginning to fall off it.  The electrolyte specimen was taken the day prior to surgery from an antecubital venipuncture and measured by ion selective electrodes.  The average and standard deviation for these specimens was 8 +/-3 mEq/L.  This means that about 2/3’s of these healthy patients had AG in the range of 5 – 11 mEq/L.  This supports the ‘new range’ proponents.

(To digress,  these elective heart patients, typically,  have an AG of about 20 mEq/L immediately  after transfer to the ICU post-op.  Over the next 24 to 48 hours, the AG drops to the normal range, closely followed by extubation and discharge to the floor.  The post-operative AG in patients with  more complex lesions and extensive operative course often remains elevated for longer periods, correlating to a longer intubation time and ICU stay.  Occasionally, the traditional interventions fail to reverse a patient’s downward spiral.  The AG in these patients continues to climb post-operatively resulting in an ECMO intervention.  If intervention is soon enough, the AG can be corrected on ECMO, the lethal corner is reduced, intracellular pH is corrected, healing can proceed and the patient usually survives.)

One circumstance which may be responsible for much confusion about the normal AG value concerns the type of blood sample taken for electrolyte measurement.  We noticed that AG  values calculated from blood specimens taken from the venous line of an ECMO circuit were often lower than AG values calculated from blood specimens taken from the post-oxygenator blood line.Post-oxygenator specimens have significantly less CO2 content than pre-oxygenator values and this results in a higher AG value than one taken from a venous site.  We found that arterial samples could have an AG value 1-3 mEq/L greater than the venous sample from the same patient.

None of our lab procedures make any specification about drawing electrolytes from venous or arterial sources.  As a result the “normal” range could be calculated from a mixture of venous and arterial specimens.This would result in a normal range higher than the range calculated from venous samples only. When using AG values, consistency is important.  It is best to measure the electrolytes only from venous samples.  If well ventilated arterial samples are used, the perfusionist should be aware that the AG may be falsely elevated by 1-3 mEq/L.

The second controversy

Has to do with the accuracy of the AG.  The AG is calculated from the measurement of three ions; Na, Cl and HCO3 (or TCO2). Since the measurement of each ion can have an error rate of  as much as 5% (p = 0.05) , using all three ion measurements in a calculation will increase the error rate to as much as 15% (p = 0.15).  This means that roughly 1 of every 6 AG values will be incorrect, so any single AG value by itself could be in question.  In order to reduce the magnitude of this error, AG values should be drawn 3-4 times per day in the critical patient.

All these values should be averaged for the 24 hour period.  This would reduce the effect any outlying electrolyte value would have in calculating the anion gap.  This also means that the AG is most effective as a long term marker, utilizing many measurements rather than taken as a single spot measurement. (Iatrogenic complications can also temporarily effect the AG.  For example, saline boluses or loop diuretics given just prior to taking the blood specimen can result in a false AG value.)

Patient 101

As an example of AG interpretation, take my ECMO patient # 101.  This was a patient with a disease called meconium aspiration syndrome (MAS).  Found only in newborns,MAS can often result in death due to hypoxia and/or lung damage from ventilator use.  In extreme cases, the child requires ECMO.  Pt. #101 ended up on ECMO.  His progress can be neatly summarized by looking just at the daily average AG values which were drawn every 6 hours.

Day 1

The AG values were 15, 16, 10, 10, with the average being 13 mEq/L.  Day 2: 5, 6, 5, 5, with the average being 5 mEq/L.  Day 3: 10, 4, 7, 7, with the average being 7 mEq/L.  The daily averages in order were  13, 5, 7.  The patient had an elevated AG which was corrected within the first 24 hours of ECMO.  The AG values stabilized in the “normal range” for the second and third days, allowing the patient to quickly heal and come off ECMO in only 71 hours.  MAS is considered an “easy”  ECMO diagnosis with few complications, 97% of these patients survive, only 3% expire.

Meet “Mr. 3%”.

Acidotic at birth (ABG 7.22/57/39),  this MAS baby was treated with NaHCO3 and sent to our hospital for possible ECMO.  Upon arrival his BE was +11.6 and his ABG was 7.48/48/13 (yes, thirteen!!).  He looked terrible with cardiac arrhythmias and low blood pressure (so much for base balance being any kind of reliable ongoing metabolic marker).  He was rushed to ECMO.  His AG values , drawn every 6 hours, on Day 1 were  35, 33, 34, 28,  with the average being 33.

Day 2:

24, 21, 20, 24, with the average being 22.

Day 3:

23, 20, 19, 24, with the average being 21.

Day 4: 19, 21, 28, 32, with the average being 24.

Day 5: 25, 20, 20, 25, with the average being 23.

Day 6: 28, 26, 16, 16, with the average being 22.

Day 7: 27, 25, 33, 23, with the average being 27.

Day 8:  25, 27, 25, with the average being 26.

The daily averages in order were: 33, 22, 21, 24,  23, 22, 27, 26.  This child had a BE almost all of the time on ECMO, but he had many complications including renal and liver failure.  Incredibly he was gradually weaned off  ECMO after 168 hours.  You can see when weaning began by the daily AG averages (Day 7 the AG took a big jump).  His last VBG immediately prior to coming off ECMO was 7.44/48/32/+7.2.  He quickly decompensated and his first VBG off ECMO was 7.23/46/51/-8.8.

He had a massive intracranial hemorrhage and died within 12 hours.

Both of these were my cases before I knew anything about Oxygen Pressure Field Theory or anion gaps.  In Pt. 101, I thought I did a good job of making sure the baby was well perfused.  In retrospect, his AG values confirm that perfusion was good and the baby did well.

In Pt. 100 (Mr. 3%) I also thought I was doing a good job of making sure the baby was well perfused.  His blood gases were good, his hemodynamics were good, his hematocrit was good, he wasn’t acidotic on ECMO.  But, he was in multiple organ failure.  He had renal failure necessitating ultrafiltration, liver failure and abnormal coagulation,  a hemopericardium requiring a tap,  and seizures.

On autopsy, he was shown to be a full term infant with MAS and persistent fetal circulation (which is why he went on ECMO), liver insult, kidney insult and severe hypoxic ischemic encephalopathy  and massive brain hemorrhage.  In retrospect, his AG values showed that he never had a chance. While the AG did come down from a high the first day of 33, he never even came close to dropping into the normal range.  Despite what the blood gases said, for 8 days this baby had a huge lethal corner with very bad intracellular acidosis which caused the multiple organ failure.  Once off ECMO he quickly succumbed to the large lethal corner and died.

It is plain that without an understanding of OPFT the perfusionist can be very inconsistent as to the  outcome of long-term extracorporeal perfusion applications, even within the same diagnostic group.  Today, I would adopt a different strategy for Pt. 100.  I would increase ECMO blood flow until the AG stabilized and began dropping.I would continue the high blood flow until the AG dropped into a range where the tissues could begin to heal.  I would keep the blood flow high, regardless of how good the blood gases were, until I was sure that healing of the lungs and other organs was well along, before beginning to wean the blood flow.  In other words, I need to minimize the lethal corner,  maximize perfused capillary density and increase intracapillary blood flow velocity.

So, now we know what the “normal”  AG range is (3 – 11 mEq/L by venipuncture and ion selective electrode).  And we know how to evaluate the data (average multiple measurements).  It is unrealistic to believe that the AG of every patient can be brought into the normal range of 3 – 11 mEq/L.  After all, these patients are critically ill.  In some cases, they are just barely alive with their only link to survival being some kind of electric pump. Some people obviously need support,  such as intra-aortic balloon pump for failure to wean from bypass.  Following the AG in these patients can give important clues as to whether the balloon pump is really effective.  If the AG continues to climb, perhaps LVAD or ECMO support is warranted before organ failure becomes irreversible.


So, monitoring  a potential long-term support patient’s AG prior to implementing extracorporeal support may be useful in timing the need for extracorporeal support.

Patients with a respiratory diagnosis are often placed on ECMO in order to reduce the ventilator damage to the lungs.  Ventilator damage can be crippling or even lethal.  Many of these patients will have a normal AG despite their respiratory disease.  Therefor, it is not necessary to have an elevated AG in order to need extracorporeal support.This is probably why extracorporeal support is more successfull in respiratory patients. On the other hand, patients who are hypoxic even with maximized ventilator use or patients with cardiopulmonary failure often will have elevated AG values, typically resulting in greater mortality.

What AG values indicate the need for intervention?  Well, I’m not sure, but I have some clues. Basically, you play the odds.  In a study at Children’s Mercy Hospital In KCMo we looked at over 100 ECMO children, age newborn through about 3 years old.  The diagnoses included all the usual newborn ECMO diseases of meconium aspiration syndrome, primary pulmonary hypertension (persistent fetal circulation), respiratory distress syndrome, group B streptococcus sepsis, and congenital diaphragmatic hernia.

Other diagnoses included post-cardiotomy, trauma, adult respiratory distress syndrome, myocarditis, bacterial and viral sepsis and pneumonia.  In order to classify the AG values of these patients for comparison we looked at each patient’s cumulative average, that is, the average of all the AG values obtained during each individual ECMO procedure.  For the entire study population of over 100 patients, the average of all the cumulative averages was 11 +/- 2 mEq/L.  This would be the ‘normal’ value for a very sick population as compared to the ‘normal’ value of a healthy population (8 +/-3 mEq/L).  It would be safe to conclude that many ECMO patients have abnormal metabolism ongoing.

The ECMO survivors also had an average of their cumulative averages of 11 +/-2.7 mEq/L.  The nonsurvivors had an average of their cumulative averages of 15 +/-5.4 mEq/L.  This is a statistically significant value compared to the survivors’ average (p<0.03).  These numbers validate a study by Shackelton in 1987.

Dr. Shackelton looked at over 30 preoperative indices in over 100 of his abdominal aortic aneurysm patients, including a variety of lab tests, blood gases, hemodynamics and clinical observations.  He found that only 3 things were predictive of  outcome; level of consciousness prior to surgery, a history of congestive heart failure and the preoperative anion gap.  Patients who were unconscious prior to surgery with a history of congestive heart failure and an elevated anion gap all died.

The anion gap was the most telling single index of survival. 

Survivors had an average AG of 12 +/- 2 and nonsurvivors had an AG of 15 +/-7.  These figures from adults compare very favorably with our figures from children.  This means that the anion gap is a useful, predictive index over a wide range of ages and diagnoses.  It also means that once the AG rises above  11 or 12 mEq/L for a sustained period, the chance for survival begins to diminish. In our study, the survival of patients with an average AG of 11 mEq/L or less was 88%.  Patients with an average AG of greater than 11 mEq/L had a survival of only 57% (p<0.0005).  Another way of looking at this is by population distribution.  About 15% of the study population had average AG values of 9 or less.  All these patients survived.  About 70% of the study population had an average AG value of 10 to 14 .  83% of these patients survived.  About 15% of the study population had an average AG value of greater than 14.  Only 48% of these patients survived (p<0.03).  The higher the average AG value rises the less the chance for survival.  Only 1 patient with a cumulative average AG of 20 mEq/L or greater survived compared to 8 nonsurvivors (11% survival).

The AG also predicts morbidity.

The group with 100% survival and the lowest AG had an average ECMO time of only 121 +/-69 hours.  In the group with only 48% survival and the highest AG, the survivors were on ECMO for an average of 252 +/-115 hours (p<0.003).  The longer a patient is on ECMO the greater the risk of complications and the greater the morbidity.   Since the average AG correlates to survival and morbidity, the logical conclusion is for the perfusionist to look for interventions which tend the lower the AG or prevent the AG from increasing.  This process must start prior to intervention with extracorporeal support.

For example, all the patients with a pre-ECMO AG average of about 30 mEq/L or greater died despite the extracorporeal intervention and correction of traditional indices to normal limits. Too much damage was incurred prior to the ECMO intervention and this prevented the ECMO treatment from being successful. So, watch potential patients closely.  If the usual interventions do not reverse the increasing AG, extracorporeal support should be instituted before the AG hits the 30 mEq/L plateau and remains there for more than 24 hours.

In summary

  • The normal anion gap is 8 +/-3 mEq/L via venous sample and ion selective electrode.
  • When monitoring the anion gap of a patient on extracorporeal support try to obtain venous samples.
  • If well ventilated arterial samples are used, remember that they may be falsely elevated by 1-3 mEq/L.  Obtain multiple samples daily and average them.  The more samples you can get, the more accurate the anion gap average will be.  Samples drawn every 6-8 hours is adequate.
  • If you are evaluating a potential patient for possible extracorporeal  support, draw an AG every time you draw a blood gas.

The anion gap will give you a better indication if the patient is improving or getting sicker, especially if bicarbonate therapy is ongoing.

  • Try to institute extracorporeal support before the patient reaches the AG plateau of 30 mEq/L (the patient may decompensate long before reaching 30 mEq/L).
  • Chances are poor for survival if an AG of 30 mEq/L is sustained for more than 12-24 hours.
  • Once on extracorporeal support, use as much “pump” as is necessary to stabilize and reduce the anion gap.
  • Adjusting the pump based on a 70% venous saturation in the presence of an increasing anion gap is a death sentence for the patient.
  •  Use interventions that lower the anion gap, avoid interventions that increase the anion gap.  Remember, sometimes you and the patient are going to lose, no matter what you do, but you have a better chance of success with an understanding of  Oxygen Pressure Field Theory and the anion gap.