All You Really Need to Know to Interpret Arterial Blood Gases
Lawrence Martin, M.D.

Order from or Lippincott W&W


The following is adapted from Chapter 7 of All You Really Need to Know to Interpret Arterial Blood Gases, 2nd edition.

Electrolytes and the anion gap

While the focus of this book is on arterial blood gases, discussion of routine serum electrolytes is also important, for they are vital to blood gas interpretation in many situations. The previous discussion on venous CO2 highlights one interrelationship between electrolytes and blood gases.

An electrolyte abnormality is often the first laboratory sign of an acid-base disorder. As a minimum, electrolytes used to calculate the anion gap -- sodium, chloride and bicarbonate (measured as serum CO2) -- should always be measured in any patient with a blood gas or acid-base abnormality. Since potassium is often deranged in acid-base disorders, it should also be examined.

The anion gap (AG) calculation is the sum of routinely measured cations minus routinely measured anions: (Na++K+) - (Cl-+HCO3-). However, because K+ is a small value numerically, it is usually omitted from the AG equation so that, as most commonly used,

AG = Na+ - (Cl- + HCO3-).

Although this is the equation most often published in articles and textbooks, for reasons discussed above the equation incorporates not the calculated arterial HCO3- but the measured venous CO2. To add to the confusion, some labs report the measured venous value as "HCO3-" and other labs as "CO2". Whatever the label for the reported venous value, that is the one you should use in calculating AG, since normal values are based on the venous electrolyte measurements. Throughout this book the anion gap will be calculated as

AG = Na+ - (Cl- + CO2).

The normal AG calculated in this manner (without K+) is 12 4 mEq/L. The anion gap exists simply because not all electrolytes are routinely measured. Normally there is electrochemical balance, so that the sum of all negatively charged electrolytes (anions) equals the sum of all positively charged electrolytes (cations). However, several anions are not measured routinely, leading to the anion gap. The anion gap is thus an artifact of measurement, and not a physiologic reality.

Table 7-3 shows all the cations and anions with normal serum values. Note that if they were all measured there would be no gap, since positives equal negatives. However, because only Na+, K+, Cl - and CO2 are routinely measured, there is an anion gap; the gap exists because more anions are unmeasured than are cations (Oh 1977). Finally, because K+ is usually not used in the calculation the normal anion gap is about 12 mEq/L.

One important technical aspect should be noted about anion gap measurement before discussing its clinical utility. There can be variation in the normal AG depending on the technology used to measure electrolytes (Winter 1990, Sadjadi 1995). While technical aspects of measurement are beyond the scope of this book, it is important to realize that some clinical labs use a method that gives a lower normal range (e.g., 3-11 mEq/L; Winter 1990). Always use the normal AG for your lab, and recognize that it may well vary from the 12 4 mEq/L used in this book. As with any lab test, if you understand what is being measured, then even without knowing the technical aspects of measurement you can use the information effectively.

Table 7-3. Anions and Cations in Serum

Because of electrochemical balance the concentrations of serum cations and anions are the same. However, in the routine measurement of electrolytes, more anions are unmeasured than are cations; this leads to an expected "anion gap." As typically calculated, the anion gap is based on only three electrolytes: sodium, chloride and bicarbonate (or serum CO2).

All anions and cations
(Values in mEq/L)

Proteins 15 Calcium 5
Organic acids 5 Magnesium 1.5
Phosphates 2 Potassium 4.5
Bicarbonate 24 Sodium 140
Sulfates 1
Chloride 104

Anions and cation used to calculate anion gap
Cl- + HCO3- = 128 Na+ = 140

Difference (cations - anions) = 140 - 128 = 12

The anion gap can be normal, low or high, and each result has a different clinical implication.

  • NORMAL ANION GAP. In this case there is no lab evidence for anion gap acidosis. This result does not always rule out an anion gap acidosis, but certainly makes that diagnosis unlikely.

  • VERY LOW OR NEGATIVE ANION GAP. There are several reasons why this can occur, including:
  • halide ion measured as chloride, as seen in bromism (some cough medications contain dextromethorphan bromide);
  • excess unmeasured cation, as seen in lithium toxicity;
  • reduction in the unmeasured anions, as seen in hypo-proteinemia; a 1 gm/dl decrease in serum albumin causes a 2.5 mEq/L drop in the AG.
  • presence of abnormal, positively charged proteins (paraproteins), as may occur in multiple myeloma.

Except for hypoproteinemia, conditions that cause a reduced or negative anion gap are relatively rare compared to those associated with an elevated anion gap.

  • ELEVATED ANION GAP. The patient may have an anion gap metabolic acidosis (Table 8-1). The higher the gap above normal the more likely this will be the case.

All excess anions in the blood are buffered by bicarbonate, and this is why an elevated AG usually indicates a state of metabolic acidosis (Emmett 1977; Gabow 1980; Narins 1980; Gabow 1985; Oster 1988). This statement is true even if the actual measured venous CO2 is normal or above normal.

When AG is increased, one or more of the conditions listed in Chapter 8 (Table 8-1) should be considered. The most common causes are lactic acidosis, renal failure (build up of organic acids normally excreted by the kidney) and diabetic ketoacidosis. Less common causes include overdosing on acetyl-salicylic acid (aspirin), and breakdown products of some ingested poisons (ethylene glycol and methanol).

One problem with the anion gap is deciding what value is truly abnormal. In the majority of patients with anion gap between 16 and 20 mEq/L, no specific anion gap acidosis can be diagnosed. Above 20 mEq/L the probability of a true anion gap acidosis increases markedly (and is 100% if the AG is above 29 mEq/L). As a practical matter, you should consider an AG 20 mEq/L as reflecting an anion gap metabolic acidosis and search for the cause.


A 42-year-old man is admitted to the hospital with dehydration and hypotension. Electrolytes show Na+ 165 mEq/L, K+ 4.0 mEq/L, CO2 32 mEq/L, Cl- 112 mEq/L. No arterial blood gas is obtained. Does this patient have metabolic acidosis?

Yes: his anion gap is 165 - (32 + 112) = 21 mEq/L. Despite the fact that CO2 is elevated (reflecting a metabolic alkalosis from dehydration), there is also a slight metabolic acidosis; the acidosis is from lactic acidosis, a result of the hypotension and poor organ perfusion. The co-existence of metabolic acidosis and metabolic alkalosis are discussed further in the following section, as well as in Chapter 8.

Electrolytes and the bicarbonate gap

The subject of mixed acid-base disorders based on arterial blood gas interpretation is introduced in Chapter 8. However, mixed metabolic disorders alone can often be diagnosed from the serum electrolytes, as indicated in the previous example -- an increased anion gap in the setting of an elevated venous CO2.

In less obvious cases the co-existence of two metabolic acid-base disorders may be apparent by calculating the difference between the change in anion gap (delta AG) and the change in serum CO2 (delta CO2) (Wrenn 1990, Haber 1991). This calculation is called the bicarbonate gap. (NOTE: Terms for the difference between the change in anion gap and change in serum CO2 include "bicarbonate gap", "delta gap", and "deviation from the 1:1 correlation." In this book I use 'bicarbonate gap', since that term seems closest to describing the basic concept.)

Bicarbonate gap = delta AG - delta CO2


delta AG = patient's AG - 12 mEq/L

delta CO2 = 27 mEq/L - patient's CO2

If an anion gap acidosis is the only acid-base abnormality, there should be a 1-to-1 correlation between the rise in anion gap and the fall in bicarbonate (measured as serum CO2); that is, the normal difference between rise in AG and fall in serum CO2 should be zero. For example, if AG goes up by 10 mEq/L (to 24 mEq/L) then serum CO2 should go down by 10 mEq/L (to 17 mEq/L); in this case delta AG - delta CO2 = 10 - 10 = 0 bicarbonate gap.

Elevated AG with a significant variation of bicarbonate gap from zero, either + or -, suggests the patient has a mixed acid-base disorder: anion gap acidosis plus another disorder, such as metabolic alkalosis (+ bicarbonate gap) or hyperchloremic metabolic acidosis (- bicarbonate gap).

Although the concept is sound, one problem with bicarbonate gap is deciding the outer limits of normal. Since we don't know a given patient's baseline AG and serum CO2, deviations may be more or less significant than presumed. The problem is compounded by the fact that there is no accepted standard on how to calculate delta AG and delta CO2. For example, some authors calculate delta AG by subtracting the measured AG from the upper limit of the normal AG (e.g., 16 mEq/L), while others subtract it from the mean AG (e.g, 12 mEq/L).

For this reason, as well as the variations inherent in the underlying electrolyte values, there is no accepted normal value for bicarbonate gap. Some call the bicarbonate gap abnormal if it deviates more than 6 mEq/L (Wrenn 1990), whereas others propose a deviation of more than 8 mEq/L as abnormal (Paulson 1993).

More important than a precise abnormal value is the concept of how bicarbonate gap is used to diagnose mixed acid-base disorders. For didactic purposes I will call a bicarbonate gap of >+6 mEq/L or <-6 mEq/L as abnormal, meaning it should prompt a close search for the cause. The more abnormal the bicarbonate gap value, the more likely it will reflect one of the following acid-base disorders.


(delta AG - delta CO2) *

+ bicarbonate gap (> 6 mEq/L)

The serum CO2 is reduced less than predicted by the change in the anion gap, and suggests:

a) metabolic alkalosis


b) bicarbonate retention as compensation for respiratory acidosis

- bicarbonate gap (<-6 mEq/L)

The serum CO2 is reduced more than predicted by the change in the anion gap, and suggests:

a) hyperchloremic metabolic acidosis


b) bicarbonate excretion as compensation for respiratory alkalosis


The more positive or negative the bicarbonate gap, the more likely there is an acid-base disturbance as described; see text for discussion.

When presented with a set of electrolytes and the possibility of an acid-base disorder, you should make the following calculations. This process may appear cumbersome at first, but it can be done quickly and without paper and pencil. After you have learned this method I will show you a nice shortcut. The values below are from the case of the 42-year-old man previously presented.

1. Anion gap (AG) = Na+ - (Cl- + CO2)

= 165 - (112 + 32) = 21 mEq/L

2. delta AG = AG - normal anion gap

= 21 - 12 = 9 mEq/L

3. delta CO2 = normal CO2 - measured CO2

= 27 - 32 = -5 mEq/L

4. Bicarbonate gap = delta AG - delta CO2

= 9 - (-5) = 14 mEq/L

Note that the calculations use the average normal venous CO2 of 27 mEq/L (see Figure 7-1). It is called bicarbonate gap because the bicarbonate moiety is what is buffered by organic anions; however, the serum CO2 is used in the calculation because that is what the chemistry lab measures and what the anion gap is based on (as discussed earlier). Try not to become confused by this terminology. I purposely show the steps with both terms so you will understand that we are calculating the bicarbonate gap with a venous chemistry value that is usually called "CO2".

In this example the very high bicarbonate gap of 14 mEq/L indicates a metabolic alkalosis and/or compensation for respiratory acidosis. Of course either diagnosis is suggested even without all the calculations, since the venous CO2 is slightly elevated; indeed, the 'hidden' disturbance in this case is the metabolic acidosis, which is uncovered by calculating the AG. (Subsequent blood gas analysis showed normal PaCO2, so venous CO2 was elevated because of metabolic alkalosis.)

Shortcut to calculating the bicarbonate gap

The steps outlined above are important in understanding how the bicarbonate gap is derived and what it measures. Once you learn this method of calculation you can (and should) use a much simpler shortcut. The shortcut is derived from canceling out terms in delta AG and delta CO2. Thus:


= delta AG - delta CO2

= [AG - 12] - [27 - CO2]

= [(Na+ - Cl- - CO2) -12] - [27 - CO2]

= Na+ - Cl- - 39

Yes, (Na+ - Cl- - 39) is certainly simpler than the four separate calculations I first presented. However, without knowing how to do the four steps I don't believe one can appreciate what the bicarbonate gap represents. Also, of course, the constant '39' will vary with different average values for AG and CO2. In the following examples I will calculate bicarbonate gap using both the long and short methods.

Clinical use of the bicarbonate gap

As a diagnostic aid the bicarbonate gap is most useful when venous CO2 is not elevated, as shown in the following case.


What are the acid-base disorders in a 28-year-old man who presents to the ED with several days of vomiting, nausea and abdominal pain. His blood pressure is low and he has tenting of the skin. He has the following electrolytes:

Na+ 144 mEq/L

Cl - 95 mEq/L

K+ 4.2 mEq/L

CO2 14 mEq/L

1) Calculate the anion gap:

AG = Na+ - (Cl- + CO2)

= 144 - (95 + 14) = 35

2) Calculate delta AG. (Be sure to use normal AG in your lab.)

35 - 12 = 23

3) Calculate delta CO2

27 - 14 = 13

4) Calculate the bicarbonate gap: delta AG - delta CO2

23 - 13 = 10 mEq/L

SHORTCUT: Na+ - Cl- - 39

144 - 95 - 39 = 10 mEq/L

The bicarbonate gap is +10 mEq/L, indicating that the measured serum CO2 is 10 mEq/L higher than expected from the delta AG. Thus there is both an anion gap metabolic acidosis (from dehydration and poor perfusion) and a metabolic alkalosis (from vomiting and loss of stomach acid), but you might not appreciate the latter without calculating the bicarbonate gap. At first glance one might just note a low CO2 and miss the fact that it is too high for the anion gap.

Bicarbonate gap calculation can also uncover a co-existing non-anion gap metabolic acidosis, as shown in the following case.


What is (are) the acid-base disorder(s) evident in the following values, from a 27-year-old woman with acute renal failure?

Na+ 140 mEq/L
K+ 4 mEq/L
Cl - 115 mEq/L
CO2 5 mEq/L

pH 7.12
PaCO2 13 mm Hg
HCO3- 4 mEq/L

Clearly, the blood gases indicate a state of metabolic acidosis. But what type or types?

1) Calculate the anion gap:

AG = Na+ - (Cl- + CO2)

= 140 - (115 + 5) = 20

2) Calculate delta AG:

20 - 12 = 8

3) Calculate delta CO2:

27 - 5 = 22

4) Calculate the bicarbonate gap: delta AG - delta CO2

8 - 22 = - 14

SHORTCUT: Na+ - Cl- - 39

140 - 115 - 39 = -14 mEq/L

Her bicarbonate gap is significantly reduced at -14 mEq/L. Thus her measured CO2 is 14 mEq/L lower than we would expect from the excess anion gap alone. Stated another way, the acid or acids causing her anion gap should have lowered venous CO2 only to about 19 mEq/L; that her venous CO2 is actually 5 mEq/L indicates an additional reason for the acidosis, in this case hyperchloremic metabolic acidosis. Such a situation is fairly common in patients with renal failure, who have uremia (causing elevated AG metabolic) and interstitial nephritis (causing hyperchloremic metabolic acidosis, which doesn't elevate the AG).

It bears emphasis that an abnormal bicarbonate gap doesn't diagnose with certainty the type of acid-base disorder. The reasoning here is the same as when confronted with just an abnormal venous CO2 value. For example, an elevated venous CO2 and/or positive bicarbonate gap could arise from retention of bicarbonate as compensation for respiratory acidosis. Bicarbonate retention as compensation is not considered a true metabolic alkalosis.

Likewise, a reduced venous CO2 and/or negative bicarbonate gap could arise from bicarbonate excretion as compensation for respiratory alkalosis. Bicarbonate excretion as compensation is not considered a true metabolic acidosis. Definitions of metabolic acidosis and alkalosis are presented in Chapter 8.

Return to Table of Contents for All You Really Need to Know to Interpret Arterial Blood Gases
Alphabetical Index to all web sites / Subject Index to all web sites