j o u r n a l o f s u r g i c a l r e s e a r c h  a p r i l 2 0 1 8 ( 2 2 4 ) 5 e1 7

Available online at www.sciencedirect.com

ScienceDirect

journal homepage: www.JournalofSurgicalResearch.com

Research review

Metabolic acidosis and the role of unmeasured

anions in critical illness and injury

Tobias Zingg, MD,*,1 Bishwajit Bhattacharya, MD, and Linda L. Maerz, MD

Department of Surgery, Section of General Surgery, Trauma & Surgical Critical Care, Yale University School of
Medicine, New Haven, Connecticut

article info abstract

Article history: Acidebase disorders are frequently present in critically ill patients. Metabolic acidosis is
Received 3 July 2017 associated with increased mortality, but it is unclear whether as a marker of the severity of
Received in revised form the disease process or as a direct effector. The understanding of the metabolic component
4 September 2017 of acidebase derangements has evolved over time, and several theories and models for
Accepted 3 November 2017 precise quantification and interpretation have been postulated during the last century.
Available online xxx Unmeasured anions are the footprints of dissociated fixed acids and may be responsible for
a significant component of metabolic acidosis. Their nature, origin, and prognostic value
Keywords: are incompletely understood. This review provides a historical overview of how the un-
Metabolic acidosis derstanding of the metabolic component of acidebase disorders has evolved over time and
Unmeasured anions describes the theoretical models and their corresponding tools applicable to clinical
Strong ion difference practice, with an emphasis on the role of unmeasured anions in general and several spe-
Strong ion gap cific settings.
Anion gap ª 2017 Elsevier Inc. All rights reserved.
Base excess

Introduction treatment of the underlying disease process and minimize

iatrogenic morbidity.5,6
The first clinical reports describing acidebase derangements Maintenance of the pH within physiological limits is essen-
stem from observations made during the cholera outbreaks tial for normal protein structure and enzymatic function.
in the nineteenth and early twentieth centuries.1 Acidebase Several regulatory mechanisms maintain the pH within the
disorders are frequently identified in critically ill patients normal range.7 The underlying mechanism causing the resul-
and may be associated with increased mortality.2-4 Injury, tant disorder can be respiratory, metabolic, or a combination
acute or chronic disease, and therapeutic measuresesuch as thereof. When the underlying disorder is respiratory, the
fluid resuscitation or mechanical ventilationecan influence resultant shift in pH is caused by a change in pCO2. The meta-
acidebase status. Timely recognition, quantification, and bolic influence on pH is more complex.8 To better characterize
correct interpretation of acidebase disorders can aid in nonrespiratory disorders, the traditional bicarbonate-based

* Corresponding author. Department of Visceral Surgery, Lausanne University Hospital e CHUV, Rue du Bugnon 46, 1011 Lausanne,
Switzerland. Tel.: þ1141 79 556 79 43; fax: þ1141 21 314 24 11.
E-mail address: tobias.zingg@chuv.ch (T. Zingg).
1
Present address: Department of Visceral Surgery, Lausanne University Hospital-CHUV, Rue du Bugnon 46, 1011 Lausanne,
Switzerland.
0022-4804/$ e see front matter ª 2017 Elsevier Inc. All rights reserved.
https://doi.org/10.1016/j.jss.2017.11.013
6 j o u r n a l o f s u r g i c a l r e s e a r c h  a p r i l 2 0 1 8 ( 2 2 4 ) 5 e1 7

approach,9 aided by the anion gap (AG) method,10,11 as well as The Henderson-Hasselbalch equation. SCO2 is the solubility
the more recent standard base excess (SBE)12-14 and the modern coefficient for carbon dioxide (0.03) in blood, and pK is the acid
physicalechemical strong ion-based models15 have proposed dissociation constant.
surrogates for unmeasured nonvolatile acids or bases. In 1915, Hasselbalch and Svend Aage Gammeltoft demon-
strated the influence of alveolar ventilation on pH and
demonstrated that a decrease in bicarbonate was accompa-
History nied by an increased respiration and decreased CO2 tension.1
Donald Dexter Van Slyke, at the Rockefeller University Hos-
Svante August Arrhenius earned the Nobel Prize in Chemistry pital in New York, developed the Van Slyke apparatus, measuring
in 1903 for his work on dissociation and the theory of ioniza- the total carbon dioxide content in plasma, which is only slightly
tion. He postulated that acids are potential electrolytes, higher than bicarbonate.23 It became the standard method for
forming cations (Hþ) and anions (A) when dissolved in water. quantification of metabolic acidebase disorders for the next
Simply stated, acids are hydrogen salts.16 Approximately 40 years. Based on his clinical and experimental observations,
20 years later, Joannes Brønsted17 and Thomas Lowry18 inde- Van Slyke developed the first diagram relating pH to bicarbon-
pendently defined an acid as a substance that could donate a ate.24 Bicarbonate became the central element of acidebase
proton and a base as a substance that could bind a proton. interpretation and was not only viewed as an indicator but as a
This extended the applicability of the definition of acids determinant of the pH. In 1948, Richard Singer and Albert Baird
beyond water. Because body fluids are aqueous solutions, Hastings coined the term buffer base for the sum of all non-
both the Arrhenius and the BrønstedeLowry theories apply to volatile weak acids (mainly bicarbonate and proteins) in plasma.25
human physiology. During the poliomyelitis epidemic in Denmark from 1952 to
1953, Poul Astrup and Ole Siggaard-Andersen refined the buffer
HA4Hþ þ A base method.26,27 Measurements of blood pH by glass electrodes
and total carbon dioxide content with the Van Slyke method
HA is the acid dissociating into a hydrogen ion (Hþ) and its
were performed on a large scale. It was recognized that total CO2
conjugate base (A). The equilibrium constant (K) indicates
alone was a poor indicator of the metabolic component of the
the grade of dissociation or strength of an acid:
acidebase status. The high total of CO2 content measured in
½Hþ   ½A   þ  K  ½HA patients with poliomyelitis was erroneously interpreted as
K¼ / H ¼
½HA ½A  metabolic alkalosis instead of respiratory acidosis with a
concomitant increase in bicarbonate.28 Inspired by the clinical
A large K value indicates a strong acid, a small K value a weak
importance of the correct identification of acidebase disorders,
acid. A strong acid fully dissociates, and a weak acid only
Astrup, Siggaard-Andersen, and Knud Engel, who embraced the
partially.
“modern” BrønstedeLowry definition of acids, developed the
Lawrence Henderson, a biochemist at Harvard, investi-
base excess (BE) method,27 the first technique to quantify the
gated the relationship between bicarbonate and carbon diox-
CO2-invariant metabolic component of an acidebase disorder
ide gas and its role as a buffer of fixed acids. In 1909, he
and was based on the pH, pCO2 and the hemoglobin (Hb) con-
rewrote the law of mass action for weak acids and their salts
centration. Electrodes capable of directly measuring pCO2 had
and applied it to the equilibrium reaction for carbonate
become available in 1957.29 Base excess was defined as the
species.19
quantity of base or acid added to blood (at 37 and pCO2 adjusted
CO2 þ H2 O4H2 CO3 4Hþ þ HCO to 40 mmHg, to eliminate the respiratory component of the
3
change in pH) in vitro to bring its pH to 7.40. First with a nomo-
Law of mass action for carbonic acid (H2CO3), which forms gram,30,31 then with a mathematical formula, the Van Slyke
from carbon dioxide (CO2) reacting with water (H2O). HCO
3 ¼ equation,32 allowed blood gas machines to calculate the BE from
bicarbonate. whole blood by measuring pH, pCO2 and Hb.
 
½Hþ   HCO   K  ½CO2  BE ¼HCO
K¼ / Hþ ¼   3  24:4 þ ð2:3  Hb þ 7:7Þ  ðpH  7:4Þ
3
½H2 CO3  HCO
3  ð1  ð0:023  HbÞÞ
The Henderson equation. The K value for this reaction is The Van Slyke equation. BE is expressed in mEq/L, HCO 3 and
small, hence in water, H2CO3 only partially dissociates into Hþ Hb concentrations in mmol/L.
and HCO 3. The opinions regarding the new tool BE differed between
The same year, Søren Sørensen introduced the pH as the the schools of Copenhagen and Boston. The so-called great
dimensionless representation of a solution’s Hþ concentra- transatlantic debate began.33 The main critique from Boston
tion, defined as the negative decimal logarithm of the latter.20 was that the BE was an in vitro method that could not repro-
Karl Albert Hasselbalch performed the first blood pH mea- duce the in vivo carbon dioxide titration curve34 because only
surement using a platinum electrode in 1912.21 He later rear- the intravascular compartment is assessed, excluding the
ranged Henderson’s equation into a logarithmic form and interstitial space, which has a much weaker buffering capac-
replaced the [CO2] with pCO2 by applying Henry’s law.22 ity than blood due to the absence of erythrocytes and a lower
    concentration of albumin.35,36 Instead of the BE, the American
HCO 3
pH ¼ pK þ log contingent proposed the “six Bostonian rules of thumb”9 to
SCO2  pCO2
recognize the CO2-invariant changes in bicarbonate. Many
 zingg et al  metabolic acidosis and unmeasured anions 7

clinicians thus continued to focus on the Hender-

Table 1 e The six Bostonian Rules to identify pCO2-
soneHasselbalch (HH) equation, with the concept that not invariant changes in HCO3- and mixed disorders.
only pCO2 but also HCO 3 influenced the pH. In 1977, the AG
Primary disorder Expected Expected
was proposed to characterize the metabolic component and
HCO
3 ¼ 24 þ . pCO2 ¼ .
distinguish acidosis due to accumulation of organic acids
from the loss of alkali.37 James Figge later demonstrated that Acute respiratory ([pCO2]40)/10
acidosis
the AG is imprecise if albumin is not considered.11,38,39
In 1978, Peter Arthur Stewart, a Canadian physiologist who Chronic respiratory ([pCO2]40)/3
acidosis
was at that time professor at Brown University, published a
physicalechemical approach to quantify the metabolic com- Acute respiratory (40[pCO2])/5
alkalosis
ponents of the acidebase status of plasma focusing on the
dissociation equilibrium of water. The difference between the Chronic respiratory (40[pCO2])/2
alkalosis
charges of strong cations and strong anions, the strong ion
difference (SID), is the major force driving the dissociation of Metabolic acidosis 1.5  [HCO3] þ 8

water and thus the Hþ concentration.15,40 Stewart embraced Metabolic alkalosis 0.7  [HCO3] þ 21
the Arrhenius definition of acid. By integrating six formulas pCO2 ¼ partial carbon dioxide pressure (mmHg).
and applying the principles of electroneutrality, conservation
of mass, and dissociation of electrolytes, he found three in-
dependent effectors on the plasma pH. These are the SID, the hemoglobin compared to whole blood, and thus more
sum of weak acids (ATOT), and the pCO2. According to Stewart, accurately reproduce in vivo conditions, Siggaard-Andersen
the bicarbonate and BE methods are useful for determining modified the original Van Slyke equation32 by using a
the extent of a disorder rather than its mechanism. Norman lower than normal blood hemoglobin concentration of 50 g/
Jones and John Kellum coined the term strong ion gap (SIG), L.50 SBE is a three compartment (interstitium, plasma, and
which by the Stewart method represents unmeasured ions erythrocytes) model and is thus also called extracellular BE.
(UIs) (alkali or fixed acids other than lactate, which is routinely  

measured).41,42 The original Stewart approach was later SBE ¼ 0:9287  HCO
3  24:4 þ 14:83  ½pH  7:4

refined,38,39,43-46 with the most recent version integrating all SBE: The Van Slyke equation standardized for a blood Hb
relevant body fluid compartments.47,48 concentration of 50 g/L. The SBE reproduces the in vivo CO2
titration curve better than the original BE,47,51 but still not
perfectly. As for the HH approach, rules of thumb (Table 2)
Principal methods of acidebase interpretation have been derived to calculate the expected BE or pCO2 for a
given measured pCO2 or BE.13 Any superimposed respiratory
There are four major descriptors of metabolic acidebase de- disorder or an expected BE for a chronic respiratory disorder
rangements. They are based on plasma HCO 3 , whole blood can be identified. Edward Wooten recently refined the SBE by
SBE, plasma AG, the plasma SID, and the plasma ATOT. integrating the plasma weak acids in a formula that more
closely reproduces the data obtained experimentally.43,52 In
The HH method clinical practice, SBE correlates significantly with the
sequential organ failure assessment score,53 and thus clinical
Merely serving as a description of the acidebase status, this outcome.54
traditional approach allows for the classification of an
acidebase disorder as primarily respiratory or metabolic. The AG method
The only values that are measured with this method are the
pCO2, and the pH, HCO 3 being calculated from the first two
The AG is defined as the difference between unmeasured
using the HH formula: anions (UAs) and unmeasured cations.10,37 Given the law of
electroneutrality, it must be identical to the difference be-

HCO
3 ðmmol=LÞ ¼ 2:46  10
8
 pCO2 ðmmHgÞ 10pH tween the measured cations [primarily sodium (Naþ) and po-
tassium (Kþ)] and anions [primarily chloride (Cl) and HCO3 ].
Based on experimental and clinical measurements of pH,
If an organic acid accumulates, it will dissociate into a
pCO2, and HCO 3 in vivo, empirical rules (Table 1) were
derived.9 Analyzing the patterns of pH, pCO2, and HCO3 with
these rules can determine the true metabolic (CO2-invariant)
component of HCO3, the magnitude of any mixed disorders,
Table 2 e Compensation rules for BE in relation to pCO2.
and differentiate acute from chronic respiratory changes.9,49
Primary disorder Expected DBE Expected DpCO2

The standard base excess method Acute respiratory 0

Chronic respiratory 0.4  DpCO2
The original BE method was developed in an attempt to Metabolic acidosis DBE
isolate and objectively quantify the metabolic component of Metabolic alkalosis 0.6  DBE
acidebase disorders.26 To account for the weaker buffering
pCO2 ¼ partial carbon dioxide pressure (mmHg).
capacity of the interstitial space due to the lack of
8 j o u r n a l o f s u r g i c a l r e s e a r c h  a p r i l 2 0 1 8 ( 2 2 4 ) 5 e1 7

conjugate base (A) and a hydrogen ion (Hþ). The latter is then regulated by alveolar ventilation, the metabolic component
neutralized by a buffer (primarily HCO3 ). An increase in an- of the acidebase status only depends on two variables, the
ions in the form of conjugate bases (footprints of organic SID and ATOT. Because of the law of electrical neutrality, the
acids) therefore increases the AG. When using Naþ, Kþ, Cl electrical sum of strong ions, the apparent SID (SIDa), must
and HCO 3 , the normal AG is 12  4 mmol/L.
55
mirror the electrical sum of weak ions, the effective SID

(SIDe), which includes the negative charges of HCO 3 and


AG ¼ Naþ þ Kþ  HCO
3 þCl ¼ 12  4 mEq=L ATOT.
or, if Kþ is not accounted for:


SIDa ¼ Naþ þ Kþ þMg2þ þCa2þ  Cl þLactate ¼ 42



AG ¼ Naþ  HCO
3 þCl ¼ 8  4 mEq=L
SIDe ¼ HCO
3 þ ATOT ¼ 42
Proteins and phosphate are weak acids. Their levels can be
significantly altered in critical illness and not taking them into
ATOT ¼ 2:7  ½Albðg=dLÞþ0:6  ½Phosðmg=dLÞ
account leads to misinterpretation of the AG.56 Hypo-
albuminemia in critically ill patients is very common.57-59 The two independent non-respiratory determinants of pH
Figge demonstrated that the plasma protein effect on the (SID and ATOT), allow for classification of metabolic acidebase
acidebase status was accurately represented by albumin disorders into four categories: high SID alkalosis from
alone and developed formulas to calculate the AG corrected hypochloremia61 or hypernatremia (free water deficiency),62
for albumin (AGc),11,38,39 the simplest of which is: low SID acidosis from hyperchloremia63 or hyponatremia
  (free water excess),64 high ATOT acidosis from hyper-
Albuminðg=LÞReference Albuminðg=LÞMeasured phosphatemia65 or hyperalbuminemia66 and low ATOT alka-
AGc ¼ AG þ
4 losis from hypoalbuminemia.59
or, as lactate is a routinely measured anion: The influence of chloride on the SID explains its role in
compensatory mechanisms for acidebase disorders. For
 
Albuminðg=LÞReference Albuminðg=LÞMeasured example, in accumulation of fixed acids, renal chloride
AGc ¼ AG þ excretion counteracts the acidosis by increasing the SID. In
4
one study, 23% of patients with a metabolic acidosis due to
Lactateðmmol=LÞ accumulated fixed acids had a normal SBE, the reason being
The AGc has been demonstrated to have a strong correlation the presence of a concomitant hypochloremia.61 Another
with the mathematically more complicated SIG as a surrogate frequent situation encountered in the critically ill patient is
for UA.3,54 Still, it does not consider the phosphate (Phos) level hypoalbuminemic alkalosis59 being neutralized by a low SID
and assumes a fixed negative charge for albumin (Alb), which “acidosis”, thus normalizing the net SBE. Wilkes et al.67 have
depends on the pH. Figge’s more complex formula considers demonstrated that the renal compensatory mechanism for
both of these factors39: the loss of negative charges (as in hypoalbuminemia) is


increased chloride reabsorption. In an in-vivo study of pigs,
AGc ¼ Naþ þ Kþ  HCO
3 þCl  ð½Alb
Langer et al.68 have demonstrated that the kidney responds to
 ðð0:123  pHÞ0:631Þ þ ½Phos  ðð0:309  pHÞ0:469ÞÞ
changes in plasma SID with opposite changes in the urinary
SID. A clinical study in critically ill patients with impaired
Stewart’s physicalechemical approach renal function has demonstrated that the blood pH was
inversely related to the urinary SID, illustrating the impor-
For Stewart, the water dissociation equilibrium plays the tance of renal chloride handling.69 Exploiting this regulatory
central role in his model, in which the physicalechemical mechanism, the plasma chloride-to-sodium ratio has been
properties of a solution act as the forces dictating the dissoci- proposed as a simple method to detect UAs.70,71 Given its
ation of water and thus Hþ concentration. Cations are conju- important role in metabolic compensatory mechanisms for
gate acids and anions are conjugate bases. Strong ions are fully acidebase disturbances,67 it is questionable whether hyper- or
dissociated at physiological pH and thus do not participate in hypochloremia should be considered as causing acidosis or
proton-transfer reactions.43 This applies to anions with pK alkalosis or simply as normal compensation for the underly-
values of 4 or less (e.g., sulfate, lactate, b-hydroxybutyrate). ing primary disorder.50
Applying the principles of electroneutrality, conservation of Any difference between the SIDa and SIDe indicates the
mass, and mass action (electrolyte dissociation), and presence of UIs (Fig. 1) and was termed SIG,41,42 which is some-
combining six equations, the following formula is obtained.60 what a misnomer, because any UI, strong or weak, can influence
its value. In a healthy person, the SIG should equal zero.72
½SIDþ   Ka ½ATOT =Ka þ10pH
pH ¼ pK01 þlog
S  PCO2 SIG ¼ SIDa SIDe ¼ 0  2 mEq=L

The three variables independently influencing the pH are the A positive SIG indicates the presence of UAs, a negative SIG
SID, ATOT and the pCO2. If ATOT (albumin and inorganic the presence of unmeasured cations. As lactate is a routinely
phosphate) is set to zero, the formula is identical to the HH measured strong anion, it is usually included in the SID.
equation. This demonstrates that considering the weak acids In analogy to the AGc, the original Stewart formula has
albumin and phosphate is the major difference between the been revised, and the method for evaluation of ATOT has been
traditional and physicalechemical theories. Since pCO2 is further refined.38,39,44,45 Wooten recently developed a multi-
 zingg et al  metabolic acidosis and unmeasured anions 9

compartment model that in addition corrects for hemoglobin The HH method

and thus brings it to the same quantitative level as the SBE
method.43 Used in isolation, this method fails to identify and quantify
An approach that combines Stewart’s with Siggaard- the underlying contributors of the metabolic component of
Andersen’s BE method has been described by Vladimir the acidebase disorder. The bicarbonate level varies with the
Fencl14 and Brian Gilfix.73 The contribution of the SID (which pCO2, mediated by chemical equilibrium, not compensatory
can be further separated into free water and chloride disor- adaptation.72 Chloride, not bicarbonate, may play the central
ders) and ATOT to the net BE allows for determination of the compensatory role in regulating the metabolic contribution to
amount of BE that is caused by UAs (BEUA). David Story pro- the pH.67,78 Even when applying the Bostonian rules of
posed a simplification of the method by partitioning the BE thumb,9 the observed discrepancies in bicarbonate do not
into three parts: the BESID, BEAlb, and BEUA.74 The BESID is truly reflect the quantity of accumulated acid or alkali because
calculated using only the two dominant ions sodium and nonbicarbonate buffers, such as albumin and hemoglobin, as
chloride. well as electrolytes are not considered.14 Both electrolyte and

albumin abnormalities are very common in critically ill pa-
BESID ðmEq=LÞ¼ Naþ Cl 38
tients.79 In addition, the rules for calculation of the expected
bicarbonate level have their limitations and become unreli-
BEAlb ðmEq=LÞ¼ 0:25  ð42  ½Albðg=LÞÞ able for pCO2 values above 80-100 mm Hg.80 The HH calcula-
tions have the advantage of being easily performed at the
BEUA ¼ BEmeasured BESID BEAlb bedside. Combining this method with the AGc method may
allow for a more accurate characterization of the acidebase
disorder.
Strengths and weaknesses of the methods
The SBE method
Recent studies have demonstrated a good correlation of
HCO 3 , SBE, and SID in diagnosing metabolic acidosis in crit- SBE levels have been demonstrated to predict intensive care
ically ill and healthy patients.57,75 Controversy exists unit (ICU) mortality at admission.3,54,81-83 The SBE method and
regarding which method is best used for further character- its refinements43 provide accurate quantitative information
izing acidebase disorders, and no single model has proven to regarding the metabolic acidosis but are not helpful to deter-
be superior to another.33,50,76 Combining the traditional mine the mechanism of the derangement.57 As described by
bicarbonate-based and modern electrolyte-based ap- Kaplan et al.,5 this may have important clinical implications. A
proaches could allow for optimal understanding of acid- decreased SBE in the setting of hyperchloremia, a frequent
ebase disorders and their relationship to electrolyte occurrence,84 may be misinterpreted as acidosis from inade-
homeostasis.77 quate tissue perfusion and, therefore, lead to inappropriate

Fig. 1 e Graphical representation of the Stewart model. ATOT [ sum of weak acids (albuminL D inorganic phosphateL). SIDa
(apparent strong ion difference) [ electrical net sum of strong ions (NaD D KD D MgDD D CaDD D ClL D lactateL),
normal [ 40-42 mEq/L. SIDe (effective strong ion difference) [ electrical net sum of weak ions (ATOT D HCOL
3 ),
normal [ negative 40-42 mEq/L. SIG (strong ion gap) [ SIDa-SIDe, normal [ 0-2 mEq/L.
10 j o u r n a l o f s u r g i c a l r e s e a r c h  a p r i l 2 0 1 8 ( 2 2 4 ) 5 e1 7

volume loading, potentially worsening the acidosis, especially In vivo physiology includes four compartments: plasma,
if chloride-rich fluids are used.85 Experimental and clinical erythrocytes, interstitium, and intracellular space. Stewart’s
data suggest potential harm from hyperchloremic original model is based on a single compartment (plasma).
acidosis.84,86-88 Erythrocytes, containing the potent buffer hemoglobin, were
As described earlier, the SBE is an in vitro measure and not included. One can argue that the electrolyte distribution is
represents the net sum of all metabolic acidebase disorders even in the extracellular space, but this does not hold true for
present. Thus, the SBE can be normal in the presence of albumin. It has, therefore, been questioned whether the SID,
complex metabolic acidebase disorders, as observed in pCO2, and ATOT can be considered as independent from each
several clinical studies.5,57,61,79,89-92 Partitioning the SBE into other.100 Wooten refined the original Stewart equations to
its three physicalechemical contributors with the Fencl- account for plasma, the interstitial space, and erythrocytes
Stewart approach allows for a better qualitative description (Fig. 2).43,47
of metabolic acidebase disorders.73 Conditions like hypo-
chloremia in the setting of a SIG acidosis or hyperchloremia in
the context of hypoalbuminemia might actually not represent
complex metabolic acidebase disorders per se, but just Clinical relevance of metabolic acidosis
physiological compensatory mechanisms.72 A normal SBE in
this situation may thus mask something irrelevant, and this Both metabolic alkalosis and acidosis, of which the latter is
raises the question about the clinical significance of complex present in the majority of critically ill patients,3 are associated
acidebase disorders in the setting of a normal SBE. Part of SBE with increased morbidity and mortality,4,58,81,101 with acid-
can be physiological compensation for a chronic respiratory ebase changes being more profound in nonsurvivors.58,101
disorder and thus, in order not to overlook mixed acidebase Whether the acidebase derangement is simply a marker of
disturbances, rules of thumb, based on clinical and experi- the severity of disease or directly contributes to mortality is
mental observations, have also been developed for SBE.13 controversial.102 Acidosis has many adverse effects on the
respiratory, hemodynamic, cerebral, and immunological sys-
tems.103 Conversely, it has been observed in animal models
The AG method
that acidosis can protect myocardium and liver tissue during
ischemia.104,105 Respiratory acidosis, as in the setting of
The uncorrected AG potentially misses the presence of sig-
permissive hypercapnia for acute respiratory distress syn-
nificant acidebase disorders.56 The corrected AG (AGc) can
drome management,106 is usually well tolerated over longer
unmask the presence of an organic acidosis previously
periods of time. At similar pH, different etiologies of acidosis
masked by hypoalbuminemia.57 The AGc has been demon-
have been demonstrated to have different effects on the
strated to be an adequate tool to quantify UAs54,57,63 and is
myocardium of rabbits.107 The question arises whether it is
mathematically simpler to calculate than the SIG. The AGc
the Hþ concentration or the type of accumulated anion that is
does not account for the measured cations magnesium (Mgþþ)
deleterious. The importance of lactate and lactate clearance
and calcium (Caþþ), but these are tightly regulated and fluc-
as markers for shock and prognosis in critically ill patients is
tuate less than phosphate and albumin levels.
well established.3,84,101,108 Lactate levels correlate with

Stewart’s physicalechemical approach

The claimed advantage of this approach resides in its capacity

to detect complex metabolic acidebase disorders. In up to 64%
of critically ill patients with a normal SBE, simultaneous
mixed metabolic acidebase disorders can be diagnosed using
the physicalechemical approach.91,93 Kaplan et al. demon-
strated that trauma patients in whom the etiologies of meta-
bolic acidosis were identified with the strong ion method
normalized their acidebase status more rapidly and spent less
days on mechanical ventilation, possibly from avoiding
inappropriate fluid resuscitation. The same study also
demonstrated that the method identified acidebase disorders
not identified by the traditional methods.5 The choice of
resuscitation fluid, by virtue of its physicalechemical prop-
erties, can influence the acidebase balance in hemorrhagic
shock.94 The SID of the resuscitation fluid has been demon- Fig. 2 e IPE model for body compartments relevant for
strated to influence the SBE95 and the plasma pH.68,96 Some acidebase metabolism. Arrows indicate the movements
investigators and clinicians consider the SIG to be the gold and distribution of each determinant of acidebase status
standard for quantification of UIs.97,98 Studies on surrogates (NaD, KD, MgDD, CaDD, ClL, lactateL, phosphateL,
for UA among kidney transplant patients74 and 365 patients in albuminL, CO2, UAsLand HbL) across the compartments.
an emergency department99 demonstrated better sensitivity I [ interstitial space; P [ plasma; E [ erythrocytes. (Color
of the SIG compared to the AGc for detecting UA. version of figure is available online.)
 zingg et al  metabolic acidosis and unmeasured anions 11

mortality in shock,109 sepsis,110 and trauma.111 Gunnerson for the SIG or BEUA indicated in the literature range from 0 to
et al.3 found a significant association between metabolic 13 mEq/L.41,63,79,82,83,89,90,112 Most clinical studies defined a
acidosis and mortality for lactate and UA, but not for hyper- cut-off level of 5 mEq/L, if the SIG was used as the surrogate
chloremia. For hyperchloremia, although a common cause of for UA, with values ranging from 3-8.9 mEq/L.82,122,123 In more
metabolic acidosis,58,61,63,84 most studies have not demon- recent studies, defining reference SIG values in blood drawn
strated any association with mortality.3,58,85,101,112,113 Howev- from healthy volunteers, cut-off values varied between 6 and
er, in severe sepsis and septic shock, acidosis of 8.9 mEq/L.54,57,61,84,120 Using a Monte Carlo methodology,
hyperchloremic origin may be associated with unfavorable Anstey derived a reference range with a 95% confidence in-
outcomes. Noritomi et al.84 found the SID, primarily attributed terval of 3.9  6.4 mEq/L.124 Laboratories should establish a
to hyperchloremia, to not only contribute significantly more to local reference range for the SIG, considering population and
the negative BE in nonsurvivors than in survivors (8.94  7.06 measurement variability.125 The risk of errors from accu-
versus 5.64  4.96 mEq/L, P ¼ 0.039), but after multivariate mulating imprecision from each of the variables required for
analysis also to be independently associated with mortality the physicalechemical approach is a factor limiting the
(P ¼ 0.004), along with the creatinine level (P ¼ 0.020) and the definition of a reference range.126
APACHE II score (P ¼ 0.042). Similarly, in a study of 22,851
patients undergoing noncardiac surgery, postoperative
hyperchloremia was an independent predictor of mortality Sources of unmeasured anions
(P < 0.01).114
The nature and origin of the largest part of UAs is un-
known.61,127 Based on observations from clinical and experi-
Contribution of unmeasured anions to metabolic mental studies, several hypotheses pertaining to the nature of
acidosis UAs have been proposed. In general, UAs result from protein
dissociation and from intermediate products of energy
The only ion routinely measured as a surrogate for a dissoci- metabolism, accumulated because of a disequilibrium be-
ated acid is lactate, of which the plasma levels correlate with tween UA production and clearance in the setting of critical
mortality in several clinical scenarios.109-111 However, it has illness or trauma.
been demonstrated in animal experiments that during Elevated levels of UA are observed during global hypox-
metabolic acidosis in sepsis, lactate comprises less than 50% emic states.82,91,117,119-121,128 Besides lactate, intermediate
of fixed acids.115 This is in agreement with a clinical study metabolites of the Krebs Cycle, particularly acetate and cit-
among patients with severe sepsis, where lactate comprised rate,121 have been observed with UA metabolic acidosis, sug-
only 50% of the AGc.116 Several of the clinical studies that gesting mitochondrial dysfunction129 as an etiology. Diverse
quantified the acidosis by cause reported nonlactate UAs91,117 metabolites such as urate, amino acids (aspartate, isoleucine,
and hyperchloremia61,63,84,118 as the predominant causes of and ornithine), and organic acids (acetate, citrate, succinate,
metabolic acidosis. One study of patients after out-of-hospital pyroglutamate, and p-hydroxyphenyl-lactate) accounted for
cardiac arrest demonstrated that UAs and hyper- only 7.9% of all UAs. Urate had the largest contribution to the
phosphatemia combined made up for 46% of the (predomi- UA, with 2.2%.127 Another potential source of UA during
nantly lactic) metabolic acidosis.119 hypoperfusion states is shedding of the endothelial glyco-
calyx130 rich in negatively charged heparan sulfate.128
Increased levels of UAs have also been observed in
Unmeasured anions: surrogates and reference renal63,65,69,131 and hepatic41,132,133 dysfunction. A correlation
values between both bilirubin and creatinine values with the SIG as
demonstrated in patients with severe malaria also suggests
UIs are defined as the net sum of electrical charges originating hepatic and renal contributions to the generation or non-
from cations and anions that are not routinely measured. clearance of UAs.117 Anions accumulating during chronic
Given the law of electrical neutrality, it must be equal to the renal failure include primarily sulfate134 and phosphate,69 but
negative net sum of measurable ions.15,40 Because metabolic also urate, hydroxypropionate, hippurate, oxalate, glutamate,
acidosis is predominant in critically ill patients, this sum is aspartate, and furanpropionate.135 Bellomo et al.136 demon-
most often negative, caused by the presence of UAs. strated that high-intensity continuous venovenous hemo-
Depending on the method used for analysis, the surrogates diafiltration significantly reduced the SIG, increased the mean
allowing for quantification of UIs are the SIG,41 AGc,39 and arterial pressure, and decreased the vasopressor requirement
BEUA.14,73 All three surrogates perform well and have a strong compared to baseline in patients with metabolic acidosis and
correlation.3,57,61,63,120,121 acute kidney injury. As 25%-30% of lactate is metabolized by
Healthy subjects and laboratory animals have few, if any, the kidneys, it can also accumulate during renal failure.137 The
circulating UIs. Exercising healthy humans exhibited UI acute phase proteins C-reactive protein and fibrinogen do not
levels of 0.3  0.6 mEq/L in one study.41 A recent study has contribute significantly to UAs.138 A more recent study by
demonstrated a mean SIG of 1.4  1.8 mEq/L in healthy vol- Zampieri et al.139 demonstrated an association between an
unteers compared to a significantly higher value of 5.1  2.1 increased SIG and the presence of inflammatory cytokines in
mEq/L in stable ICU patients with a normal SBE (defined as critically ill patients.
0  2 mEq/L). Two-thirds of the latter had occult acidebase Exogenous sources such as intravenous fluids like polyge-
disorders revealed by the Stewart approach.89 Normal values lines or tromethamine,140-142 toxins (ethanol, methanol,
12 j o u r n a l o f s u r g i c a l r e s e a r c h  a p r i l 2 0 1 8 ( 2 2 4 ) 5 e1 7

ethylene glycol, propylene glycol, formaldehyde, and lactate and observed a significant association between major
formate), and various drugs, such as lithium (strong cation), adverse event and lactate levels, but not UAs. No information
salicylates, antibiotics (polymyxin B, disodium carbenicillin), on the nature of the CPB priming solution was provided in this
infusions of nimodipine (dissolved in 20% ethanol), lorazepam study. This could play a role in the study results because there
and etomidate (in propylene glycol solvent), as well as paral- are data to demonstrate that certain fluids used for priming of
dehyde and methylene blue administration can contribute to the CPB circuit, such as polygelines, contain a significant UA
the UI load.102,143 Pyroglutamate (5-oxoproline) is observed load.140,141,153 Thus, in such a setting, elevated levels of UAs
with the administration of acetaminophen, flucloxacillin, and may not necessarily be associated with unfavorable
vigabatrin and is generated when hepatic glutathione syn- outcomes.
thase is reduced.144,145
Critically ill adults (general medical or mixed medical-
surgical ICU)
Prognostic value of unmeasured anions
Three studies found an elevated UA level (either expressed as
Elevated levels of UAs were associated with increased SIG or AGc) to be an independent predictor of mortality, with
mortality or unfavorable outcome in diverse set- lactate performing equally well.3,91,146 Rocktaeschel et al.147
tings,3,57,79,82,83,91,112,117,120,122,123,146-149 whereas some studies found that UAs could predict mortality similarly to lactate,
did not demonstrate any association of UAs with but all acidebase variables performed worse than the APACHE
outcome.58,63,79,84,101,118,147,150 The interpretation of the II score. Similarly, in a prospective observational study by
currently available data on the prognostic value of UA is Dubin et al.,57 although all associated with mortality, UAs (SIG
rendered difficult by the fact that therapeutic agents, such as and AGc), lactate, SBE, and bicarbonate levels performed
certain antibiotics or resuscitation fluids can by themselves worse than the sequential organ failure assessment score. In a
contribute UAs.79,121,140-143,151 study by Cusack et al.,79 only one of the surrogates for UA (AGC)
was predictive of 28-day mortality, whereas the SIG was not.
Critically ill children (general pediatric & pediatric cardiac In this study, the APACHE II score performed best, followed by
surgery ICU) the pH and SBE. Two other studies from the same group
demonstrated no correlation between UAs and survival in this
Balasubramanyan et al.83 observed a stronger correlation of patient population.58,118 In their first study, Maciel et al.118
mortality with UA (BEUA) levels >5 mEq/L than for any other found significant prognostic value for APACHE II, pH, bicar-
acidebase variable, including SBE, AG, and lactate bonate, SBE, and lactate, but not for UAs. In their second
(>5 mmol/L), in critically ill pediatric patients. Mann et al. study, survivors and nonsurvivors didn’t differ in type, but in
compared the ability of the SIG, SBE, and lactate to predict severity (expressed as SBE) of metabolic acidosis. They found
unfavorable neurological outcomes of neonates with increased levels of UAs in nonsurvivors, both on their day of
hypoxic-ischemic encephalopathy and found better out- admission and on day of death, without reaching statistical
comes associated with lower UA levels with an overall significance.58 UAs have also predicted mortality in adult burn
similar prognostic value as lactate. Both performed signifi- patients,154 general emergency department patients,155 pa-
cantly better than SBE.148 On the other hand, Hatherill tients with renal failure,131 acute pancreatitis,156 and hypoxic
et al.,150 in a prospective observational study among patients hepatitis.133
in a general pediatric intensive care unit, demonstrated a
significant association between lactate levels (>2 mmol/L), Adult ICU patients with sepsis
but not UAs (SIG) > 3 mEq/L, and unfavorable outcomes.
Durward et al. demonstrated superiority of the SIG In a study by Noritomi et al., adult ICU patients with severe
(>3 mEq/L) to predict mortality after cardiopulmonary bypass sepsis or septic shock had variables measured at pre-
(CPB) in children, measured on admission to the pediatric determined time points during the first five days after
intensive care unit and 24 hours later, when compared to admission. Utilizing the quantitative physicochemical meth-
lactate (>2 mmol/L) levels. The fluids used for priming the CPB odology, they extrapolated that although patients requiring
circuit in this study were packed red blood cells, hydroxyethyl ICU admission had severe metabolic derangements, these
starch, Ringer’s lactate solution, and sodium bicarbonate. derangements were attributable to elevated lactate and chlo-
Levels of UAs and lactate correlated with the complexity of the ride levels, and a decrease of these strong ions over time was
surgery but not with the duration of CPB.112 Murray et al.122 associated with increased survival. They found that for prog-
demonstrated that postcardiac surgery admission UA levels nosis, SBE, mostly because of a hyperchloremic decrease of
(>3 mEq/L), but not lactate levels (>2 mmol/L), were predictive the SID, was the only acidebase variable associated with
of major adverse events. As in the study by Durward et al.,112 mortality at ICU admission, in addition to the APACHE II score
there was no relation between CPB duration and UA level. and serum creatinine level. Admission UAs and lactate levels
The typical priming solution used in this study was composed were similar in survivors and nonsurvivors. Interestingly,
of packed red blood cells, PlasmaLyte (balanced crystalloid clearance of UAs and lactate was associated with survival,
solution), 20% albumin solution, and sodium bicarbonate.122 whereas clearance of hyperchloremia was not.84 Part of the
The results of these two studies are in contrast to a study by large contribution of hyperchloremia to acidosis could be
Park et al.,152 who compared intraoperative (after weaning explained by fluid resuscitation with normal saline, although
from CPB and again after chest closure) levels of UAs (SIG) and chloride shifts during endotoxemia have been described in an
 zingg et al  metabolic acidosis and unmeasured anions 13

animal model.157 Experimental data demonstrate decreased on outcome than the Hþ concentration itself. Further studies
renal blood flow158 and a proinflammatory effect of hyper- and clinical investigations are required to better delineate the
chloremic acidosis in sepsis.159 role and origin of UAs in severely ill patients, for better prog-
nostication and to identify potential targets for therapy.
Adult ICU patients aftercardiac arrest

Cardiac arrest leads to severe systemic metabolic acidosis and

death if not reversed rapidly. Funk et al. examined the
Acknowledgment
contribution of UAs on the acidebase derangement in
Authors’ contributions: T.Z. performed an extensive literature
patients after cardiac arrest with return of spontaneous
search, critically evaluated the pertinent literature, drafted
circulation after more than 20 minutes and whether acidosis
the manuscript, and edited the manuscript based on
had an impact on patient survival. Patients were placed into
consensus review by the authorship. All authors critically
hypothermic conditions with standard resuscitative efforts
reviewed the manuscript for important intellectual content
being continued. Multivariate analysis demonstrated that UAs
and approved the submitted version.
(SIG >8.9 mEq/L), measured 12 hours after return of sponta-
This research did not receive any specific grant from
neous circulation, were correlated with an unfavorable
funding agencies in the public, commercial, or not-for-profit
neurological outcome and an increased mortality rate over the
sectors.
6 months immediately following cardiac arrest (P ¼ 0.0021),
whereas lactate levels were not (P ¼ 0.8035).120
Disclosure
Adult trauma patients
The authors reported no proprietary or commercial interest in
Multiple analyses have been performed on trauma patients any product mentioned or concept discussed in this article.
with severe injuries requiring ICU admission. Kaplan et al.82,123
demonstrated that when comparing survivors to non-
survivors, UAs (SIG >5 mEq/L) had the strongest correlation references
with mortality among all acidebase parameters. Martin
et al.149 found UAs (BEUA) to have the strongest overall corre-
lation with mortality among acidebase parameters in a 1. Van Slyke DD. Some points of acid-base history in
physiology and medicine. Ann N Y Acad Sci. 1966;133:5e14.
trauma ICU, particularly in patients with normal lactate levels.
2. Gunnerson KJ, Kellum JA. Acid-base and electrolyte analysis
in critically ill patients: are we ready for the new
Adult ICU patients with severe malaria millennium? Curr Opin Crit Care. 2003;9:468e473.
3. Gunnerson KJ, Saul M, He S, Kellum JA. Lactate versus non-
Falciparum malaria is commonly accompanied by a high AG lactate metabolic acidosis: a retrospective outcome
metabolic acidosis from anaerobic glycolysis in the setting of evaluation of critically ill patients. Crit Care. 2006;10:R22.
tissue hypoxia. Severe anemia and hemodynamic shock are 4. Webster NR, Kulkarni V. Metabolic alkalosis in the critically
ill. Crit Rev Clin Lab Sci. 1999;36:497e510.
the usual contributors to tissue hypoxia. The metabolic
5. Kaplan LJ, Cheung NH, Maerz L, et al. A physicochemical
acidosis associated with severe malaria can only be partially approach to acid-base balance in critically ill trauma
explained by lactate, phosphate, or acute renal failure. Acids patients minimizes errors and reduces inappropriate plasma
associated with currently unidentified anions are quantita- volume expansion. J Trauma. 2009;66:1045e1051.
tively the most important contributors to acidosis in this 6. O’Dell E, Tibby SM, Durward A, Murdoch IA. Hyperchloremia
disease. The presence of high concentrations of UAs, is the dominant cause of metabolic acidosis in the
postresuscitation phase of pediatric meningococcal sepsis.
measured as the SIG, had major prognostic significance in
Crit Care Med. 2007;35:2390e2394.
patients with severe malaria, independent of lactate levels.117
7. Kellum JA. Determinants of plasma acid-base balance. Crit
Care Clin. 2005;21:329e346.
8. Kurtz I, Kraut J, Ornekian V, Nguyen MK. Acid-base analysis:
Conclusions a critique of the Stewart and bicarbonate-centered
approaches. Am J Physiol Renal Physiol. 2008;294:F1009eF1031.
There is no consensus regarding preferred methodology for 9. Narins RG, Emmett M. Simple and mixed acid-base
disorders: a practical approach. Medicine. 1980;59:161e187.
the evaluation of acidebase derangements in critically ill pa-
10. Emmett M, Narins RG. Clinical use of the anion gap. Medicine.
tients. The physicalechemical approach by Stewart does not 1977;56:38e54.
have a clear advantage over the traditional bicarbonate-based 11. Figge J, Jabor A, Kazda A, Fencl V. Anion gap and
method. Both should be regarded as complementary for hypoalbuminemia. Crit Care Med. 1998;26:1807e1810.
optimal understanding of acidebase disorders. 12. Siggaard-Andersen O. Titratable acid or base of body fluids.
UAs pose an as yet undetermined effect on mortality in Ann N Y Acad Sci. 1966;133:41e58.
13. Schlichtig R, Grogono AW, Severinghaus JW. Human PaCO2
patients who have significant physiological derangements
and standard base excess compensation for acid-base
and who are unable to regain appropriate homeostasis by
imbalance. Crit Care Med. 1998;26:1173e1179.
eliminating excess anions. Based on observations from clin- 14. Fencl V, Leith DE. Stewart’s quantitative acid-base
ical and experimental studies, it seems that at similar pH chemistry: applications in biology and medicine. Respir
values, the type of accumulated anions exerts a greater effect Physiol. 1993;91:1e16.
14 j o u r n a l o f s u r g i c a l r e s e a r c h  a p r i l 2 0 1 8 ( 2 2 4 ) 5 e1 7

15. Stewart PA. Independent and dependent variables of acid- 41. Kellum JA, Kramer DJ, Pinsky MR. Strong ion gap: a
base control. Respir Physiol. 1978;33:9e26. methodology for exploring unexplained anions. J Crit Care.
16. Arrhenius S. Theories of solutions. New Haven, CT: Yale 1995;10:51e55.
University Press; etc; 1912. 42. Jones NL. A quantitative physicochemical approach to acid-
17. Bronsted JN. Einige Bemerkungen über den Begriff der base physiology. Clin Biochem. 1990;23:189e195.
Säuren und Basen. Recl Trav Chim Pays Bas. 1923;42:718e728. 43. Wooten EW. Calculation of physiological acid-base
18. Lowry TM. The electronic theory of valency. IV. The origin of parameters in multicompartment systems with application
acidity. Trans Faraday Soc. 1924;20:13e15. to human blood. J Appl Physiol. 2003;95:2333e2344.
19. Henderson LJ. Das Gleichgewicht der Säuren und Basen im 44. Watson PD. Modeling the effects of proteins on pH in
tierischen Organismus. Ergeb Physiol Biol Chem Exp Pharmakol. plasma. J Appl Physiol. 1999;86:1421e1427.
1909;8:254e325. 45. Staempfli HR, Constable PD. Experimental determination of
20. Sörensen SPL. Enzymstudien II. Mitteilung. Über die net protein charge and A(tot) and K(a) of nonvolatile buffers
Messung und die Bedeutung der in human plasma. J Appl Physiol. 2003;95:620e630.
Wasserstoffionenkonzentration bei enzymatischen 46. Lloyd P, Freebairn R. Using quantitative acid-base analysis in
Prozessen. Biochem Z. 1909;21:131e304. the ICU. Critical care and resuscitation. J Australas Acad Crit
21. Hasselbalch KA, Lundsgaard C. Elektrometrische Care Med. 2006;8:19e30.
Reaktionsbestimmung des Blutes bei Körpertemperatur. 47. Wooten EW. The standard strong ion difference, standard
Biochem Z. 1912;38:77. total titratable base, and their relationship to the Boston
22. Hasselbalch KA. Die Berechnung der Wasserstoffzahl des compensation rules and the Van Slyke equation for
Blutes aus der freien und gebundenen Kohlensäure extracellular fluid. J Clin Monit Comput. 2010;24:177e188.
desselben und die Sauerstoffbindung des Blutes als Funktion 48. Wolf MB, Deland EC. A comprehensive, computer-model-
der Wasserstoffzahl. Biochem Z. 1916;78:112e144. based approach for diagnosis and treatment of complex
23. Van Slyke DD, Neill JM. The determination of gases in blood acid-base disorders in critically-ill patients. J Clin Monit
and other solutions by vacuum extraction and manometric Comput. 2011;25:353e364.
measurement. J Biol Chem. 1924;61:523. 49. Albert MS, Dell RB, Winters RW. Quantitative displacement
24. Van Slyke DD. Studies of acidosis. J Biol Chem. 1921;48:153e175. of acid-base equilibrium in metabolic acidosis. Ann Intern
25. Singer RB, Hastings AB. An improved clinical method for the Med. 1967;66:312e322.
estimation of disturbances of the acid-base balance of 50. Siggaard-Andersen O, Fogh-Andersen N. Base excess or
human blood. Medicine. 1948;27:223e242. buffer base (strong ion difference) as measure of a non-
26. Astrup P, Jorgensen K, Andersen OS, Engel K. The acid-base respiratory acid-base disturbance. Acta Anaesthesiol Scand
metabolism. A new approach. Lancet. 1960;1:1035e1039. Suppl. 1995;107:123e128.
27. Siggaard-Andersen O, Engel K, Jorgensen K, Astrup P. A 51. Park M, Maciel AT, Noritomi DT, Pontes de Azevedo LC,
Micro method for determination of pH, carbon dioxide Taniguchi LU, da Cruz Neto LM. Effect of PaCO2 variation on
tension, base excess and standard bicarbonate in capillary standard base excess value in critically ill patients. J Crit
blood. Scand J Clin Lab Invest. 1960;12:172e176. Care. 2009;24:484e491.
28. Astrup P, Severinghaus JW. The history of blood gases, acids and 52. Wooten EW. Science review: quantitative acid-base
bases. Copenhagen: Munksgaard; 1986:1e322. physiology using the Stewart model. Crit Care.
29. Stow W, Baer RF, Randall B. Rapid measurement of the 2004;8:448e452.
tension of carbon dioxide in blood. Arch Phys Med Rehabil. 53. Vincent JL, Moreno R, Takala J, et al. The SOFA (Sepsis-
1957;38:646e650. related organ failure assessment) score to describe organ
30. Andersen OS. The pH-log pCO2 blood acid-base nomogram dysfunction/failure. On behalf of the working group on
revised. Scand J Clin Lab Invest. 1962;14:598e604. sepsis-related Problems of the European Society of intensive
31. Andersen OS. A graphic representation of changes of the care medicine. Intensive Care Med. 1996;22:707e710.
acid-base status. Scand J Clin Lab Invest. 1960;12:311e314. 54. Park M, Taniguchi LU, Noritomi DT, Liborio AB,
32. Siggaard-Andersen O. The van Slyke equation. Scand J Clin Maciel AT, Cruz-Neto LM. Clinical utility of standard
Lab Invest Suppl. 1977;146:15e20. base excess in the diagnosis and interpretation of
33. Severinghaus JW. Siggaard-Andersen and the “Great Trans- metabolic acidosis in critically ill patients. Braz J Med Biol
Atlantic Acid-Base Debate”. Scand J Clin Lab Invest Suppl. Res. 2008;41:241e249.
1993;214:99e104. 55. Gabow PA. Disorders associated with an altered anion gap.
34. Brackett Jr NC, Cohen JJ, Schwartz WB. Carbon dioxide Kidney Int. 1985;27:472e483.
titration curve of normal man. Effect of increasing degrees of 56. Salem MM, Mujais SK. Gaps in the anion gap. Arch Intern Med.
acute hypercapnia on acid-base equilibrium. N Engl J Med. 1992;152:1625e1629.
1965;272:6e12. 57. Dubin A, Menises MM, Masevicius FD, et al. Comparison of
35. Schwartz WB, Relman AS. A critique of the parameters used three different methods of evaluation of metabolic acid-base
in the evaluation of acid-base disorders. “Whole-blood disorders. Crit Care Med. 2007;35:1264e1270.
buffer base” and “standard bicarbonate” compared with 58. Maciel AT, Park M. Differences in acid-base behavior
blood pH and plasma bicarbonate concentration. N Engl J between intensive care unit survivors and nonsurvivors
Med. 1963;268:1382e1388. using both a physicochemical and a standard base excess
36. Bunker JP. The great trans-atlantic acid-base debate. approach: a prospective, observational study. J Crit Care.
Anesthesiology. 1965;26:591e594. 2009;24:477e483.
37. Oh MS, Carroll HJ. The anion gap. N Engl J Med. 59. McAuliffe JJ, Lind LJ, Leith DE, Fencl V. Hypoproteinemic
1977;297:814e817. alkalosis. Am J Med. 1986;81:86e90.
38. Figge J, Rossing TH, Fencl V. The role of serum proteins in 60. Constable PD. A simplified strong ion model for acid-base
acid-base equilibria. J Lab Clin Med. 1991;117:453e467. equilibria: application to horse plasma. J Appl Physiol.
39. Figge J, Mydosh T, Fencl V. Serum proteins and acid-base 1997;83:297e311.
equilibria: a follow-up. J Lab Clin Med. 1992;120:713e719. 61. Mallat J, Michel D, Salaun P, Thevenin D, Tronchon L.
40. Stewart PA. Modern quantitative acid-base chemistry. Can J Defining metabolic acidosis in patients with septic shock
Physiol Pharmacol. 1983;61:1444e1461. using Stewart approach. Am J Emerg Med. 2012;30:391e398.
 zingg et al  metabolic acidosis and unmeasured anions 15

62. Lindner G, Schwarz C, Grussing H, Kneidinger N, outcome from major vascular injury. Crit Care Med.
Fazekas A, Funk GC. Rising serum sodium levels are 2004;32:1120e1124.
associated with a concurrent development of metabolic 83. Balasubramanyan N, Havens PL, Hoffman GM. Unmeasured
alkalosis in critically ill patients. Intensive Care Med. anions identified by the Fencl-Stewart method predict
2013;39:399e405. mortality better than base excess, anion gap, and lactate in
63. Moviat M, van Haren F, van der Hoeven H. Conventional or patients in the pediatric intensive care unit. Crit Care Med.
physicochemical approach in intensive care unit patients 1999;27:1577e1581.
with metabolic acidosis. Crit Care. 2003;7:R41eR45. 84. Noritomi DT, Soriano FG, Kellum JA, et al. Metabolic acidosis
64. Haskins SC, Hopper K, Rezende ML. The acid-base impact of in patients with severe sepsis and septic shock: a
free water removal from, and addition to, plasma. J Lab Clin longitudinal quantitative study. Crit Care Med.
Med. 2006;147:114e120. 2009;37:2733e2739.
65. Rocktaeschel J, Morimatsu H, Poustie S, et al. Acid-base 85. Brill SA, Stewart TR, Brundage SI, Schreiber MA. Base deficit
status of critically ill patients with acute renal failure: does not predict mortality when secondary to
analysis based on Stewart-Figge methodology. Crit Care. hyperchloremic acidosis. Shock. 2002;17:459e462.
2003;7:R60. 86. Kellum JA. Fluid resuscitation and hyperchloremic acidosis
66. Rossing TH, Maffeo N, Fencl V. Acid-base effects of altering in experimental sepsis: improved short-term survival and
plasma protein concentration in human blood in vitro. J Appl acid-base balance with Hextend compared with saline. Crit
Physiol. 1986;61:2260e2265. Care Med. 2002;30:300e305.
67. Wilkes P. Hypoproteinemia, strong-ion difference, and acid- 87. Kellum JA, Song M, Venkataraman R. Effects of
base status in critically ill patients. J Appl Physiol. hyperchloremic acidosis on arterial pressure and circulating
1998;84:1740e1748. inflammatory molecules in experimental sepsis. Chest.
68. Langer T, Carlesso E, Protti A, et al. In vivo conditioning of 2004;125:243e248.
acid-base equilibrium by crystalloid solutions: an 88. Kellum JA, Song M, Li J. Lactic and hydrochloric acids induce
experimental study on pigs. Intensive Care Med. different patterns of inflammatory response in LPS-
2012;38:686e693. stimulated RAW 264.7 cells. Am J Physiol Regul Integr Comp
69. Moviat M, Terpstra AM, van der Hoeven JG, Pickkers P. Physiol. 2004;286:R686eR692.
Impaired renal function is associated with greater urinary 89. Gunnerson KJ, Srisawat N, Kellum JA. Is there a difference
strong ion differences in critically ill patients with metabolic between strong ion gap in healthy volunteers and intensive
acidosis. J Crit Care. 2012;27:255e260. care unit patients? J Crit Care. 2010;25:520e524.
70. Durward A, Skellett S, Mayer A, Taylor D, Tibby SM, 90. Fencl V, Jabor A, Kazda A, Figge J. Diagnosis of metabolic
Murdoch IA. The value of the chloride: sodium ratio in acid-base disturbances in critically ill patients. Am J Respir
differentiating the aetiology of metabolic acidosis. Intensive Crit Care Med. 2000;162:2246e2251.
Care Med. 2001;27:828e835. 91. Antonini B, Piva S, Paltenghi M, Candiani A, Latronico N. The
71. Kurt A, Ecevit A, Ozkiraz S, Ince DA, Akcan AB, Tarcan A. The early phase of critical illness is a progressive acidic state due
use of chloride-sodium ratio in the evaluation of metabolic to unmeasured anions. Eur J Anaesthesiol. 2008;25:566e571.
acidosis in critically ill neonates. Eur J Pediatr. 92. Boniatti MM, Cardoso PR, Castilho RK, Vieira SR. Acid-base
2012;171:963e969. disorders evaluation in critically ill patients: we can
72. Kellum JA. Determinants of blood pH in health and disease. improve our diagnostic ability. Intensive Care Med.
Crit Care. 2000;4:6e14. 2009;35:1377e1382.
73. Gilfix BM, Bique M, Magder S. A physical chemical approach 93. Moviat M, van den Boogaard M, Intven F, van der Voort P,
to the analysis of acid-base balance in the clinical setting. J van der Hoeven H, Pickkers P. Stewart analysis of apparently
Crit Care. 1993;8:187e197. normal acid-base state in the critically ill. J Crit Care.
74. Story DA, Morimatsu H, Bellomo R. Strong ions, weak acids 2013;28:1048e1054.
and base excess: a simplified Fencl-Stewart approach to 94. Kaplan LJ, Philbin N, Arnaud F, Rice J, Dong F, Freilich D.
clinical acid-base disorders. Br J Anaesth. 2004;92:54e60. Resuscitation from hemorrhagic shock: fluid selection and
75. Abdulraof Menesi F, Verzola D, Villaggio B, et al. Evaluation infusion strategy drives unmeasured ion genesis. J Trauma.
of metabolic acidosis in patients with a kidney graft: 2006;61:90e97. discussion 97-98.
comparison of the bicarbonate-based and strong ion-based 95. Smith RJ, Reid DA, Delaney EF, Santamaria JD. Fluid therapy
methods. Transplant Proc. 2011;43:1055e1062. using a balanced crystalloid solution and acid-base stability
76. Fidkowski C, Helstrom J. Diagnosing metabolic acidosis in after cardiac surgery. Crit Care Resusc. 2010;12:235e241.
the critically ill: bridging the anion gap, Stewart, and base 96. Carlesso E, Maiocchi G, Tallarini F, et al. The rule regulating
excess methods. Can J Anaesth. 2009;56:247e256. pH changes during crystalloid infusion. Intensive Care Med.
77. Seifter JL. Integration of acid-base and electrolyte disorders. 2011;37:461e468.
N Engl J Med. 2014;371:1821e1831. 97. Gunnerson KJ. Clinical review: the meaning of acid-base
78. Berend K, van Hulsteijn LH, Gans RO. Chloride: the queen of abnormalities in the intensive care unit part I -
electrolytes? Eur J Intern Med. 2012;23:203e211. epidemiology. Crit Care. 2005;9:508e516.
79. Cusack RJ, Rhodes A, Lochhead P, et al. The strong ion gap 98. Agrafiotis M, Sileli M, Ampatzidou F, Keklikoglou I,
does not have prognostic value in critically ill patients in a Panousis P. The base excess gap is not a valid tool for the
mixed medical/surgical adult ICU. Intensive Care Med. quantification of unmeasured ions in cardiac surgical
2002;28:864e869. patients: a retrospective observational study. Eur J
80. Martinu T, Menzies D, Dial S. Re-evaluation of acid-base Anaesthesiol. 2013;30:678e684.
prediction rules in patients with chronic respiratory 99. Antonogiannaki EM, Mitrouska I, Amargianitakis V,
acidosis. Can Respir J. 2003;10:311e315. Georgopoulos D. Evaluation of acid-base status in patients
81. Smith I, Kumar P, Molloy S, et al. Base excess and lactate as admitted to ED-physicochemical vs traditional approaches.
prognostic indicators for patients admitted to intensive care. Am J Emerg Med. 2015;33:378e382.
Intensive Care Med. 2001;27:74e83. 100. Masevicius FD, Dubin A. Has Stewart approach improved
82. Kaplan LJ, Kellum JA. Initial pH, base deficit, lactate, anion our ability to diagnose acid-base disorders in critically ill
gap, strong ion difference, and strong ion gap predict patients? World J Crit Care Med. 2015;4:62e70.
16 j o u r n a l o f s u r g i c a l r e s e a r c h  a p r i l 2 0 1 8 ( 2 2 4 ) 5 e1 7

101. Ho KM, Lan NS, Williams TA, et al. A comparison of 120. Funk GC, Doberer D, Sterz F, et al. The strong ion gap and
prognostic significance of strong ion gap (SIG) with other outcome after cardiac arrest in patients treated with
acid-base markers in the critically ill: a cohort study. J therapeutic hypothermia: a retrospective study. Intensive
Intensive Care. 2016;4:43. Care Med. 2009;35:232e239.
102. Morris CG, Low J. Metabolic acidosis in the critically ill: part 121. Bruegger D, Kemming GI, Jacob M, et al. Causes of metabolic
2. Causes and treatment. Anaesthesia. 2008;63:396e411. acidosis in canine hemorrhagic shock: role of unmeasured
103. Adrogue HJ, Madias NE. Management of life-threatening ions. Crit Care. 2007;11:R130.
acid-base disorders. First of two parts. N Engl J Med. 122. Murray D, Grant D, Murali N, Butt W. Unmeasured anions in
1998;338:26e34. children after cardiac surgery. J Thorac Cardiovasc Surg.
104. Bonventre JV, Cheung JY. Effects of metabolic acidosis on 2007;133:235e240.
viability of cells exposed to anoxia. Am J Physiol. 123. Kaplan LJ, Kellum JA. Comparison of acid-base models for
1985;249:C149eC159. prediction of hospital mortality after trauma. Shock.
105. Wright G, Kingston MA, Ross IS. Role of metabolic acidosis 2008;29:662e666.
on cardiac mechanical performance during severe acute 124. Anstey CM. An assessment of the population variance of the
hypoxia and reoxygenation is small and transient. strong ion gap using Monte Carlo simulation. Anaesth
Cardiovasc Res. 1995;29:611e615. Intensive Care. 2009;37:983e991.
106. Villar J, Kacmarek RM, Perez-Mendez L, Aguirre-Jaime A. A 125. Morimatsu H, Rocktaschel J, Bellomo R, Uchino S,
high positive end-expiratory pressure, low tidal volume Goldsmith D, Gutteridge G. Comparison of point-of-care
ventilatory strategy improves outcome in persistent acute versus central laboratory measurement of electrolyte
respiratory distress syndrome: a randomized, controlled concentrations on calculations of the anion gap and
trial. Crit Care Med. 2006;34:1311e1318. the strong ion difference. Anesthesiology.
107. Berger DS, Fellner SK, Robinson KA, Vlasica K, Godoy IE, 2003;98:1077e1084.
Shroff SG. Disparate effects of three types of extracellular 126. Story DA, Poustie S, Bellomo R. Comparison of three
acidosis on left ventricular function. Am J Physiol. methods to estimate plasma bicarbonate in critically ill
1999;276:H582eH594. patients: Henderson-Hasselbalch, enzymatic, and strong-
108. Levraut J, Ichai C, Petit I, Ciebiera JP, Perus O, Grimaud D. ion-gap. Anaesth Intensive Care. 2001;29:585e590.
Low exogenous lactate clearance as an early predictor of 127. Moviat M, Terpstra AM, Ruitenbeek W, Kluijtmans LA,
mortality in normolactatemic critically ill septic patients. Pickkers P, van der Hoeven JG. Contribution of various
Crit Care Med. 2003;31:705e710. metabolites to the “unmeasured” anions in critically ill
109. Weil MH, Afifi AA. Experimental and clinical studies on patients with metabolic acidosis. Crit Care Med.
lactate and pyruvate as indicators of the severity of acute 2008;36:752e758.
circulatory failure (shock). Circulation. 1970;41:989e1001. 128. Annecke T, Rehm M, Bruegger D, et al. Ischemia-
110. Bakker J, Coffernils M, Leon M, Gris P, Vincent JL. Blood reperfusion-induced unmeasured anion generation and
lactate levels are superior to oxygen-derived variables in glycocalyx shedding: sevoflurane versus propofol
predicting outcome in human septic shock. Chest. anesthesia. J Invest Surg. 2012;25:162e168.
1991;99:956e962. 129. Forni LG, McKinnon W, Lord GA, Treacher DF, Peron JM,
111. Moomey Jr CB, Melton SM, Croce MA, Fabian TC, Proctor KG. Hilton PJ. Circulating anions usually associated with the
Prognostic value of blood lactate, base deficit, and oxygen- Krebs cycle in patients with metabolic acidosis. Crit Care.
derived variables in an LD50 model of penetrating trauma. 2005;9:R591e595.
Crit Care Med. 1999;27:154e161. 130. Chappell D, Jacob M, Becker BF, Hofmann-Kiefer K,
112. Durward A, Tibby SM, Skellett S, Austin C, Anderson D, Conzen P, Rehm M. Expedition glycocalyx. A newly
Murdoch IA. The strong ion gap predicts mortality in discovered “Great Barrier Reef”. Anaesthesist.
children following cardiopulmonary bypass surgery. Pediatr 2008;57:959e969.
Crit Care Med. 2005;6:281e285. 131. Abramowitz MK, Hostetter TH, Melamed ML. The serum
113. Van Regenmortel N, Verbrugghe W, Van den Wyngaert T, anion gap is altered in early kidney disease and associates
Jorens PG. Impact of chloride and strong ion difference on with mortality. Kidney Int. 2012;82:701e709.
ICU and hospital mortality in a mixed intensive care 132. Kellum JA, Bellomo R, Kramer DJ, Pinsky MR. Hepatic anion
population. Ann Intensive Care. 2016;6:91. flux during acute endotoxemia. J Appl Physiol.
114. McCluskey SA, Karkouti K, Wijeysundera D, Minkovich L, 1995;78:2212e2217.
Tait G, Beattie WS. Hyperchloremia after noncardiac surgery 133. Kneidinger N, Funk GC, Lindner G, Drolz A, Schenk P,
is independently associated with increased morbidity and Fuhrmann V. Unmeasured anions are associated with short-
mortality: a propensity-matched cohort study. Anesth Analg. term mortality in patients with hypoxic hepatitis. Wien Klin
2013;117:412e421. Wochenschr. 2013;125:474e480.
115. Kellum JA, Bellomo R, Kramer DJ, Pinsky MR. Splanchnic 134. Kirschbaum B. Effect of hemodialysis on the
buffering of metabolic acid during early endotoxemia. J Crit hypersulfatemia of chronic renal failure. ASAIO J.
Care. 1997;12:7e12. 1998;44:314e318.
116. Mecher C, Rackow EC, Astiz ME, Weil MH. Unaccounted for 135. Niwa T. Organic acids and the uremic syndrome: protein
anion in metabolic acidosis during severe sepsis in humans. metabolite hypothesis in the progression of chronic renal
Crit Care Med. 1991;19:705e711. failure. Semin Nephrol. 1996;16:167e182.
117. Dondorp AM, Chau TT, Phu NH, et al. Unidentified acids of 136. Bellomo R, Lipcsey M, Calzavacca P, et al. Early acid-base
strong prognostic significance in severe malaria. Crit Care and blood pressure effects of continuous renal
Med. 2004;32:1683e1688. replacement therapy intensity in patients with metabolic
118. Maciel AT, Park M. Unmeasured anions account for most of acidosis. Intensive Care Med. 2013;39:429e436.
the metabolic acidosis in patients with hyperlactatemia. 137. Bellomo R. Bench-to-bedside review: lactate and the kidney.
Clinics (Sao Paulo). 2007;62:55e62. Crit Care. 2002;6:322e326.
119. Makino J, Uchino S, Morimatsu H, Bellomo R. A quantitative 138. Kneidinger N, Lindner G, Fuhrmann V, et al. Acute phase
analysis of the acidosis of cardiac arrest: a prospective proteins do not account for unmeasured anions in critical
observational study. Crit Care. 2005;9:R357eR362. illness. Eur J Clin Invest. 2007;37:820e825.
 zingg et al  metabolic acidosis and unmeasured anions 17

139. Zampieri FG, Kellum JA, Park M, et al. Relationship between 149. Martin M, Murray J, Berne T, Demetriades D, Belzberg H.
acid-base status and inflammation in the critically ill. Crit Diagnosis of acid-base derangements and mortality
Care. 2014;18:R154. prediction in the trauma intensive care unit: the
140. Liskaser F, Story DA, Hayhoe M, Poustie SJ, Bailey MJ, physiochemical approach. J Trauma. 2005;58:238e243.
Bellomo R. Effect of pump prime on acidosis, strong-ion- 150. Hatherill M, Waggie Z, Purves L, Reynolds L, Argent A.
difference and unmeasured ions during cardiopulmonary Mortality and the nature of metabolic acidosis in children
bypass. Anaesth Intensive Care. 2009;37:767e772. with shock. Intensive Care Med. 2003;29:286e291.
141. Hayhoe M, Bellomo R, Liu G, McNicol L, Buxton B. The 151. Morgan TJ, Vellaichamy M, Cowley DM, Weier SL,
aetiology and pathogenesis of cardiopulmonary bypass- Venkatesh B, Jones MA. Equivalent metabolic acidosis with
associated metabolic acidosis using polygeline pump prime. four colloids and saline on ex vivo haemodilution. Anaesth
Intensive Care Med. 1999;25:680e685. Intensive Care. 2009;37:407e414.
142. Hoste EA, Colpaert K, Vanholder RC, et al. Sodium 152. Park SJ, Kim HS, Byon HJ, Kim CS, Cheong IY, Kim JT.
bicarbonate versus THAM in ICU patients with mild Intraoperative plasma lactate as an early indicator of major
metabolic acidosis. J Nephrol. 2005;18:303e307. postoperative events in pediatric cardiac patients. Tohoku J
143. Lipner HI, Ruzany F, Dasgupta M, Lief PD, Bank N. The Exp Med. 2012;228:239e245.
behavior of carbenicillin as a nonreabsorbable anion. J Lab 153. Liskaser FJ, Bellomo R, Hayhoe M, et al. Role of pump prime
Clin Med. 1975;86:183e194. in the etiology and pathogenesis of cardiopulmonary
144. Dempsey GA, Lyall HJ, Corke CF, Scheinkestel CD. bypass-associated acidosis. Anesthesiology.
Pyroglutamic acidemia: a cause of high anion gap metabolic 2000;93:1170e1173.
acidosis. Crit Care Med. 2000;28:1803e1807. 154. Berndtson AE, Palmieri TL, Greenhalgh DG, Sen S. Strong ion
145. Armenian P, Gerona RR, Blanc PD, Wu AH, Mookherjee S. 5- difference and gap predict outcomes after adult burn injury. J
oxoprolinemia causing elevated anion gap metabolic Trauma Acute Care Surg. 2013;75:555e560. discussion 560-561.
acidosis in the setting of acetaminophen use. J Emerg Med. 155. Hucker TR, Mitchell GP, Blake LD, et al. Identifying the sick:
2012;43:54e57. can biochemical measurements be used to aid decision
146. Lipnick MS, Braun AB, Cheung JT, Gibbons FK, making on presentation to the accident and emergency
Christopher KB. Difference between critical care initiation department. Br J Anaesth. 2005;94:735e741.
anion gap and prehospital admission anion gap is 156. Shen X, Ke L, Yang D, et al. The prognostic value of the
predictive of mortality in critical illness. Crit Care Med. strong ion gap in acute pancreatitis. J Crit Care.
2013;41:49e59. 2016;36:140e145.
147. Rocktaeschel J, Morimatsu H, Uchino S, Bellomo R. 157. Kellum JA, Bellomo R, Kramer DJ, Pinsky MR. Etiology of
Unmeasured anions in critically ill patients: can they predict metabolic acidosis during saline resuscitation in
mortality? Crit Care Med. 2003;31:2131e2136. endotoxemia. Shock. 1998;9:364e368.
148. Mann C, Latal B, Padden B, Scheer I, Goebel G, Bernet V. Acid- 158. Wilcox CS. Regulation of renal blood flow by plasma
base parameters for predicting magnetic resonance imaging chloride. J Clin Invest. 1983;71:726e735.
measures of neurologic outcome after perinatal hypoxia- 159. Kellum JA, Song M, Almasri E. Hyperchloremic acidosis
ischemia: is the strong ion gap superior to base excess and increases circulating inflammatory molecules in
lactate? Am J Perinatol. 2012;29:361e368. experimental sepsis. Chest. 2006;130:962e967.