Ovid: Unmeasured anions identified by the Fencl-Stewart method predic... http://ovidsp.tx.ovid.com/sp-3.26.1a/ovidweb.

cgi

Critical Care Medicine

Issue: Volume 27(8), August 1999, pp 1577-1581
Copyright: 1999 Lippincott Williams & Wilkins, Inc.
Publication Type: [Pediatric Critical Care]
ISSN: 0090-3493
Accession: 00003246-199908000-00030

[Pediatric Critical Care]

Unmeasured anions identified by the Fencl-Stewart method predict

mortality better than base excess, anion gap, and lactate in patients in
the pediatric intensive care unit
Balasubramanyan, Napa MD, MRCPI, MRCP (UK); Havens, Peter L. MD; Hoffman, George M. MD

Author Information
From the Department of Pediatrics (Drs. Balasubramanyan, Havens, and Hoffman) and Anesthesiology (Dr.
Hoffman), Children's Hospital of Wisconsin and Medical College of Wisconsin, Milwaukee, Wisconsin.
Supported, in part, by the Dr. Elaine Kohler Fund for Healing.
Presented, in part, at the annual meeting of the American Society of Anesthesiologists, New Orleans, LA,
October 1996, and the Society of Critical Care Medicine's 26th Educational and Scientific Symposium, San
Diego, CA, February 1997.
Address requests for reprints to: Napa Balasubramanyan, MD, Pediatric Intensive Care Unit, Mercy Hospital
Medical Center, 400 University Avenue, Des Moines, IA 50314.
Abstract

Objectives: This study was undertaken to compare three methods for the identification of unmeasured anions
in pediatric patients with critical illness. We compared the base excess (BE) and anion gap (AG) methods
with the less commonly used Fencl-Stewart strong ion method of calculating BE caused by unmeasured
anions (BEua). We measured the relationship of unmeasured anions identified by the three methods to
serum lactate concentrations and to mortality.

Design: Retrospective cohort study.

Setting: Tertiary care pediatric intensive care unit in an academic pediatric hospital.

Patients: The study population included 255 patients in the pediatric intensive care unit who had
simultaneous measurements of arterial blood gases, electrolytes, and albumin during the period of July 1995
to December 1996. Sixty-six of the 255 patients had a simultaneous measurement of serum lactate.

Measurements and Main Results: The BEua was calculated using the Fencl-Stewart method. The AG was
defined as (sodium plus potassium) - (chloride plus total carbon dioxide). BE was calculated from the
standard bicarbonate, which is derived from the Henderson-Hasselbalch Equation andreported on the blood
gas analysis. A BE or BEua value of <or=to-5 mEq/L or an AG >or=to17 mEq/L was defined as a clinically
significant presence of unmeasured anions. A lactate level of >or=to45 mg/dL was defined as being
abnormally elevated for this study. The presence of unmeasured anions identified by significantly abnormal
BEua was poorly identified by BE or AG. Of the 255 patients included in the study, 67 (26%) had a different
interpretation of acid base balance when the Fencl method was used compared with when BE and AG were

1 of 11 9/10/2017, 3:04 PM
Ovid: Unmeasured anions identified by the Fencl-Stewart method predic... http://ovidsp.tx.ovid.com/sp-3.26.1a/ovidweb.cgi

used. Plasma lactate concentration correlated better with BEua (r2 = .55; p = .0001) than with AG (r2 = .41; p
= .0005) or BE (r2 = .27; p = .025). Mortality was more strongly related to BEua <or=to-5 mEq/L (relative risk
of death = 10.25; p = .002) than to lactate >or=to45 mg/dL (relative risk of death = 2.35; p = .04). In logistic
regression analysis, mortality was more strongly associated with BEua (area under the receiver operating
characteristic curve = 0.79; p = .0002) than lactate (receiver operating characteristic curve area = 0.63; p =
.05), BE (receiver operating characteristic curve area = 0.53; p = .32), or AG (receiver operating
characteristic curve area = 0.64; p = .08) in this patient sample.

Conclusions: Critically ill patients with normal BE and normal AG frequently have elevated unmeasured
anions detectable by BEua. The Fencl-Stewart method is better than BE and similar to AG in identifying
patients with high lactate levels. Elevated unmeasured anions identified by the Fencl-Stewart method were
more strongly associated with mortality than with BE, AG, or lactate in this patient sample. (Crit Care Med
1999; 27:1577-1581)

Key Words: Base excess; anion gap; acid-base equilibrium; lactic acidosis; unmeasured anions; mortality;
outcome; hypoalbuminemia; hyperchloremia; strong ion method; Fencl equations; Stewart analysis; pediatrics;
critical care

Calculation of the base excess (BE) and the anion gap (AG) is commonly used to identify the presence and to
analyze the cause of metabolic acidosis in critically ill patients. The BE method requires measurement of the
standard bicarbonate (HCO (3)-) [1-3], which is defined as the concentration of bicarbonate in plasma after
equilibrating fully oxygenated whole blood to a PaCO2 of 40 torr (5.33 kPa) at 38[degree sign]C (100.4[degree
sign]F) to remove the respiratory component, and can be derived from measurements of pH and PCO2 by using
the Henderson-Hasselbalch equation: (Equation 1) BE is obtained by multiplying the deviation in standard
bicarbonate from a mean of 22.9 by a factor of 1.2 [1-3]. Patients with BE <or=to-5 mEq/L are considered to
have clinically significant metabolic acidosis.

Equation 1

The AG is calculated by the equation (sodium plus potassium) - (chloride plus tCO2), where the tCO2 is the total
CO2 content measured in the plasma [4-6]. An AG >17 defines the presence of unmeasured anions (UA);
critically ill patients with increased AG and metabolic acidosis are often assumed to have increased plasma
lactate, which is associated with a higher risk of death than normal lactate concentrations [7-9].

The calculation of BE assumes normal water content, electrolytes, and albumin, and changes in these values
will change the calculated BE independent of changes in lactate, bicarbonate, or UA. Calculation of the AG does
not control for changes in PCO2 or albumin. A decrease in albumin by 1 g/dL decreases the BE by 3.7 mEq/L
[10] and the AG by 2.75 mEq/L [4]. A high PCO2 may result in a falsely low AG, because the tCO2 used in the
AG calculation is not equivalent to the standard bicarbonate value used to identify metabolic acidosis by the BE
method [1].

2 of 11 9/10/2017, 3:04 PM
Ovid: Unmeasured anions identified by the Fencl-Stewart method predic... http://ovidsp.tx.ovid.com/sp-3.26.1a/ovidweb.cgi

The quantitative acid-base studies of Stewart [11] show that three independent variables determine the
hydrogen ion and the bicarbonate ion concentrations: the strong ion difference (i.e., the difference between
sodium and chloride), PCO (2), and albumin. Based on Stewart's studies, Fencl and Leith [12] and Gilfix et al.
[13] developed equations to correct the BE for changes in sodium, chloride, and albumin: Equation 2, Equation
3, Equation 4, Equation 5, Equation 6, and Equation 7

Equation 2

Equation 3

Equation 4

Equation 5

Equation 6

Equation 7

3 of 11 9/10/2017, 3:04 PM
Ovid: Unmeasured anions identified by the Fencl-Stewart method predic... http://ovidsp.tx.ovid.com/sp-3.26.1a/ovidweb.cgi

Thus, the BE caused by unmeasured anions (BEua) represents the corrected BE, taking into account changes
in free water, chloride, albumin, and PCO2, and theoretically should reflect metabolic changes better than the
less complete biochemical measurements, BE or AG.

This study was designed to compare the ability of BE and AG methods to identify patients with UA as
determined by the Fencl-Stewart method, to measure the association of BE, AG, and BEua with lactic acidosis,
and to compare the relationship of BE, AG, BEua, and lactate with mortality in patients in the pediatric intensive
care unit.

MATERIALS AND METHODS

After institutional review board approval, we reviewed the medical records of all the patients admitted to the 24-
bed, Medical College of Wisconsin, Milwaukee, affiliated pediatric intensive care unit at Children's Hospital of
Wisconsin from July 1995 through December 1996. Patients with arterial blood gas, electrolytes, and albumin
analyses performed from a single blood sample were included in the study. Other data collected were
simultaneous lactate measurements and patient survival at pediatric intensive care unit discharge. BEua was
calculated using the Fencl equations [12,13]. AG was calculated using the equation (sodium plus potassium) -
(chloride plus tCO2) [4].

Patients were grouped on the basis of the presence of metabolic acidosis or metabolic alkalosis (determined by
the value of BE from the arterial blood gas calculations) with increased or normal AG. For the purposes of this
analysis, the following definitions were used: BE <or=to-5 mEq/L defined metabolic acidosis; BE between -4.9
mEq/L and 4.9 mEq/L defined normal; BE >or=to5 mEq/L defined metabolic alkalosis; BEua <or=to-5 mEq/L
defined the presence of UA by the Fencl-Stewart method; BEua >-5 mEq/L defined the absence of UA by the
Fencl-Stewart method; AG >or=to17 mEq/L defined increased AG. Lactate >or=to45 mg/dL defined significant
elevation of plasma lactate. (Note: to convert lactate from mg/dL to mmol/L, divide the reported mg/dL value by
9).

BE and AG were compared with the identification of UA by the Fencl-Stewart method. We measured the
relationship between BE, AG, and BEua methods and plasma lactate concentration. The association of BE, AG,
and lactate with mortality was measured.

Correlation was measured using Spearman's correlation coefficient. Fisher's exact test was used to test the
statistical significance of relationships between categorical variables. Logistic regression analysis was used to
calculate area under the receiver operating characteristic curve and to explore the association of these variables
with mortality. All analyses were performed using STATA version 5.0 (State Corp., College Station, TX).

RESULTS
The BEua, BE, and AG were compared in 255 patient samples. Variables included in the study are given in
Table 1. The usual methods of identifying patients with metabolic acidosis (i.e., BE and AG) led to a different
clinical interpretation of acid-base status than the BEua method in 67 of 255 (26%) patients (Table 2). Of
patients with a normal BE (BE -4.9 to +4.9), 45 of 157 (15%) had UA by the BEua method. The BEua method
identified the presence of UA in seven of 29 (24%) patients with a metabolic alkalosis (BE >or=to5) and in 18 of
117 (15%) patients with normal AG (<17).

4 of 11 9/10/2017, 3:04 PM
Ovid: Unmeasured anions identified by the Fencl-Stewart method predic... http://ovidsp.tx.ovid.com/sp-3.26.1a/ovidweb.cgi

Table 1. Biochemical variables included in the study

Table 2. Identification of unmeasured anions by standard methods and comparison with the Fencl method

Abnormalities in fluid status, electrolytes, and albumin caused significant changes in base deficit and AG in
these patients. BE would be equal to BEua if there were no abnormalities in sodium, chloride, and albumin and if
UA were absent. However, BE and BEua were never the same. BE caused by free water effect, BE caused by
changes in chloride, and BE caused by changes in albumin were equal to zero in 9%, 0.8%, and 0% of samples,
respectively, showing their important contribution to BE.

Plasma lactate concentration, measured in 66 patients, was more strongly correlated with BEua (r2 = .55; p =
.0001) than with BE (r2 = .27; p = .025) or with AG (r2 = .41; p = .0005). Plasma lactate was increased (>or=to45
mg/dL) in 13 of the 66 (19.7%) patients. As a single variable, BEua or AG was more strongly associated with
increased plasma lactate concentration than BE (Table 3). The area under the receiver operating characteristic
curve was 0.83 for both BEua (p = .0001) and AG (p = .002), compared with 0.65 for BE (p = .2). Of 66 patients
in whom lactate was measured, 19 died. The presence of UA, suggested by BEua <or=to-5, was more strongly
related to mortality than high lactate, high AG, or metabolic acidosis suggested by BE <or=to-5 (Table 4).

5 of 11 9/10/2017, 3:04 PM
Ovid: Unmeasured anions identified by the Fencl-Stewart method predic... http://ovidsp.tx.ovid.com/sp-3.26.1a/ovidweb.cgi

Table 3. Identification of lactic acidosisa using logistic regression analysis (n = 66)

Table 4. Association of mortality with BE, AG, BEua, and lactate (n = 66)

Logistic regression analysis showed that as a single variable, BEua was most strongly associated with mortality
(Table 5). Using a combination of BE and BEua increased the model strength (likelihood ratio test, p = .0003),
but no further accuracy was gained by addition of either lactate or AG (likelihood ratio test, p = .83) to the BEua
model.

6 of 11 9/10/2017, 3:04 PM
Ovid: Unmeasured anions identified by the Fencl-Stewart method predic... http://ovidsp.tx.ovid.com/sp-3.26.1a/ovidweb.cgi

Table 5. Identification of mortality using logistic regression analysis (n = 66)

DISCUSSION
The BE and AG methods frequently fail to identify UA that can be found using the BEua method. The greater the
deviations in plasma sodium, chloride, and albumin from normal, as often seen in critically ill patients, the
greater are the differences between the BE and BEua methods. The BEua method is superior to the BE method
and equivalent to the AG method in identifying patients with increased plasma lactate concentrations. UA as
identified by the BEua are more strongly associated with mortality.

UA as suggested by increased base deficit and increased AG may be organic (e.g., lactate, keto acids,
albumin), inorganic (e.g., phosphate, sulfate), exogenous (e.g., salicylate, formate, nitrate, penicillin,
carbenicillin), and others (e.g., paraldehyde, acetate, ethylene glycol, methanol, salicylates, urea, glucose) [4].
Proteolysis associated with sepsis may release organic and inorganic acids, some of which are poorly defined
[14]. High concentrations of some of these acids are not present during health, and thus, the presence of UA
may serve as a marker for organ dysfunction.

The cause of increased AG in patients with critical illness remains obscure and may be different depending on
the underlying disorder. In patients with cholera and severe dehydration, an increased AG may be completely
explained by an increase in plasma concentrations of lactate, phosphate, and proteins [6]. In patients with
sepsis, increases in plasma concentrations of lactate, phosphate, total serum proteins, and urate did not
completely explain the increased AG [15]. In an unselected group of critically ill patients, 62% of the increase in
AG was because of lactate and keto anions, 15% was from changes in proteins, phosphate, potassium, and
calcium, and 23% was unaccounted for by these measurements [5].

An increase in the plasma lactate concentration accounted for only half of the reduction in bicarbonate in a
canine endotoxic shock model and 15% of the increase in AG in a murine septic model [16]. Changes in plasma
concentrations of pyruvate, citrate, beta-hydroxybutyrate, acetoacetate, albumin, and anionic amino acids,
including aspartic acid and glutamic acid, were insufficient to completely account for the increased AG in this
septic model [16]. In critically ill horses, increased plasma concentrations of lactate, pyruvate, ketoacids,
phosphate, and albumin were unable to completely account for the increased AG [17].

7 of 11 9/10/2017, 3:04 PM
Ovid: Unmeasured anions identified by the Fencl-Stewart method predic... http://ovidsp.tx.ovid.com/sp-3.26.1a/ovidweb.cgi

Our data did not find a strong link between BE and the presence of lactic acidosis but did identify that BEua was
superior to BE in identifying patients with lactic acidosis. The poor relationship of BE or AG with lactic acidosis
has also been shown in other studies [18-25]. An elevated plasma lactate concentration was associated with
normal AG in 80% of unselected intensive care unit patients [19]. AG was found to correlate poorly with lactate
concentrations in trauma victims [20]. It was not sensitive as a marker of mild elevations of lactate [18,21,22]
and was a poor predictor of lactate concentrations in critically ill children [21]. In contrast, there was a fair
correlation between AG and lactate, with an r (2) of .61 in one study of septic patients, probably related to higher
mean lactate concentrations [15]. The sensitivity of AG in identifying high lactate improves with higher lactate
concentrations [18,21,22]. BE also correlated poorly with lactate concentrations in trauma victims [20] and
critically ill adults [23,24]. BE had a better correlation than AG, with mild to moderate elevations of lactate in
postcardiopulmonary bypass patients [18]. In a canine model of hemorrhagic shock, the mean BE was always
nearly twice the lactate level [25]. These studies suggest that other UA or electrolyte disturbances may coexist
with lactic acidosis in critically ill patients.

This is the first study that we know of to measure the association of BEua and mortality. We found that mortality
is more strongly associated with BEua than plasma lactate concentration, BE, or AG. Plasma lactate
concentration is strongly associated with mortality in acute shock states associated with trauma [20],
postcardiopulmonary bypass [7], and adult surgical patients [22] but may not be a good predictor of survival in
patients who were not in acute shock [26]. In patients with acute shock, a lactate concentration between 36 and
45 mg/dL has been shown to discriminate between survivors and nonsurvivors [7-9,25]. BE was associated with
mortality in one study of patients with trauma and shock [27], but another study found no association [20].
Moderate elevations of BE were not associated with mortality in a canine model of hemorrhagic shock [25]. AG
was not a good predictor of mortality in children after open-heart surgery and in trauma victims [8,20].

In our sample, patients with high plasma lactate concentrations had relatively high mortality, as did patients with
BEua <or=to-5 mEq/L. However, the case fatality rate in patients with normal lactate concentrations (<45 mg/dL)
was much higher than the rate in patients with normal BEua (Table 4). We speculate that BEua may be better
than plasma lactate concentration as an indicator of splanchnic hypoperfusion and multiorgan dysfunction.

In summary, the equations by Fencl and Leith [12] based on the quantitative acid-base analysis of Stewart [11]
provide a relatively simple method to identify UA. UA identified by the Fencl-Stewart method were not reliably
identified by BE or AG methods and actually seemed to identify patients at risk for mortality better than BE, AG,
or lactate. We suggest that further prospective studies in critically ill patients should include BEua as a predictor
of outcome.

REFERENCES
1. Astrup P, Jorgensen K, Siggaard Andersen O, et al: The acid-base metabolism-a new approach. The Lancet
1960; i:1035-1039 [Context Link]

2. Siggaard-Andersen O: Acid-base terminology. Lancet 1965; ii:1010-1012 [Context Link]

3. Siggaard-Andersen O: The Acid-Base Status of the Blood. Copenhagen, Munksgaard, 1963, p 134 [Context
Link]

4. Oh MS, Carroll HJ: The anion gap. N Engl J Med 1977; 297:814-817 SFX [Context Link]

5. Gabow PA, Kaehny WD, Fennessey PV, et al: Diagnostic importance of an increased serum anion gap. N
Engl J Med 1980; 303:854-858 SFX [Context Link]

8 of 11 9/10/2017, 3:04 PM
Ovid: Unmeasured anions identified by the Fencl-Stewart method predic... http://ovidsp.tx.ovid.com/sp-3.26.1a/ovidweb.cgi

6. Wang F, Butler T, Rabbani GH, et al: The acidosis of cholera-contributions of hyperproteinemia, lactic
acidemia and hyperphosphatemia to an increased serum anion gap. N Engl J Med 1986; 315:1591-1595 SFX
[Context Link]

7. Peretz DI, Scott HM, Duff J: The significance of lactic acidemia in the shock syndrome. Ann NY Acad Sci
1965; 119:1133-1141 [Context Link]

8. Siegel LB, Dalton HJ, Hertzog JH, et al: Initial postoperative serum lactate levels predict survival in children
after open heart surgery. Intensive Care Med 1996; 22:1418-1423 SFX [Context Link]

9. Broder G, Weil MH: Excess lactate: An index of reversibility of shock in human patients. Science 1964;
143:1457-1459 SFX [Context Link]

10. McAuliffe JJ, Lind LJ, Leith DE, et al: Hypoproteinemic alkalosis. Am J Med 1986; 81:86-90 SFX [Context
Link]

11. Stewart PA: Modern quantitative acid-base chemistry. Can J Physiol Pharmacol 1983; 61:1444-1461 SFX
[Context Link]

12. Fencl V, Leith DE: Stewart's quantitative acid-base chemistry: Application in biology and medicine (review).
Respir Physiol 1993; 91:1-16 SFX [Context Link]

13. Gilfix BM, Bique M, Magder S: A physical chemical approach to the analysis of acid-base balance in the
clinical setting. J Crit Care 1993; 8:187-197 SFX [Context Link]

14. Clowes GHA Jr, George BC, Villee CA Jr, et al: Muscle proteolysis induced by a circulating peptide in
patients with sepsis or trauma. N Engl J Med 1983; 308:545-552 SFX [Context Link]

15. Mecher C, Rackow EC, Astiz ME, et al: Unaccounted for anion in metabolic acidosis during severe sepsis in
humans. Crit Care Med 1991; 19:705-711 SFX [Context Link]

16. Rackow EC, Mecher C, Astiz ME, et al: Unmeasured anion during severe sepsis with metabolic acidosis.
Circ Shock 1990; 30:107-115 SFX [Context Link]

17. Gossett KA, Cleghorn B, Adams R, et al: Contribution of whole blood L-lactate, pyruvate, D-lactate,
acetoacetate and 3-hydroxybutyrate concentrations to the plasma anion gap in horses with intestinal disorders.
Am J Vet Res 1987; 48:72-75 SFX [Context Link]

18. Ernest D, Herekes RG, Raper RF: Alterations in anion gap following cardiopulmonary bypass. Crit Care Med
1992; 20:52-56 SFX [Context Link]

19. Mehta K, Kruse JA, Carlson RW: The relationship between anion gap and elevated lactate. Crit Care Med
1986; 14:405 SFX [Context Link]

20. Mikulaschek A, Henry SM, Donovan R, et al: Serum lactate is not predicted by anion gap or base excess
after trauma resuscitation. J Trauma 1996; 40:218-224 Ovid Full Text SFX [Context Link]

9 of 11 9/10/2017, 3:04 PM
Ovid: Unmeasured anions identified by the Fencl-Stewart method predic... http://ovidsp.tx.ovid.com/sp-3.26.1a/ovidweb.cgi

21. Aufricht C, Ties M, Hartl I, et al: Anion gap-a screen for hyperlactatemia in critically ill infants and children?
Klin Paediatr 1992; 204:378-381 [Context Link]

22. Iberti T, Leibowitz A, Papadakos P, et al: Low sensitivity of the anion gap as a screen to detect
hyperlactatemia in critically ill patients. Crit Care Med 1990; 18:275-277 SFX [Context Link]

23. Aduen J, Bernstein WK, Miller J, et al: Relationship between blood lactate concentrations and ionized
calcium, glucose, and acid-base status in critically ill and non critically ill patients. Crit Care Med 1995;
23:246-252 [Context Link]

24. Nimmo GR, Grant IS, Mackenzie SJ: Lactate and acid base changes in the critically ill. Postgrad Med J
1991; 67:S56-S61 SFX [Context Link]

25. Dunham CM, Siegel JH, Weireter L, et al: Oxygen debt and metabolic acidemia as quantitative predictors of
mortality and the severity of the ischemic insult in hemorrhagic shock. Crit Care Med 1991; 19:231-243 SFX
[Context Link]

26. Anderson CT, Westgard JO, Schlimgen K, et al: Contribution of arterial blood lactate measurement to the
care of critically ill patients. Am J Clin Pathol 1977; 68:63-67 SFX [Context Link]

27. Siegel JH, Rivkind AI, Dalal SA, et al: Early physiologic predictors of injury severity and death in blunt
multiple trauma. Arch Surg 1990; 125:498-508 SFX [Context Link]

IMAGE GALLERY

Select All Export Selected to PowerPoint

Equation 2 Equation 3
Equation 1

Equation 4 Equation 5 Equation 6

Equation 7

Table 2

Table 1

10 of 11 9/10/2017, 3:04 PM
Ovid: Unmeasured anions identified by the Fencl-Stewart method predic... http://ovidsp.tx.ovid.com/sp-3.26.1a/ovidweb.cgi

Table 3
Table 4
Table 5

Back to Top

2017 Ovid Technologies, Inc. All rights reserved. About Us Contact Us Terms of Use

OvidSP_UI03.26.01.073, SourceID 108757

11 of 11 9/10/2017, 3:04 PM