Anaesthesia, 2004, 59, pages 11111115

.....................................................................................................................................................................................................................

Changes of serum chloride and metabolic acid-base state

in critical illness
G.-C. Funk,1 D. Doberer,1 G. Heinze,2 C. Madl,3 U. Holzinger3 and B. Schneeweiss4
1 Department of Pulmonary Medicine, Medical University of Vienna, 2 Department of Medical Computer Sciences,
Medical University of Vienna, 3 Intensive Care Unit 13H1, Medical University of Vienna, AKH, Waehringerguertel
1820, A-1090 Vienna, Austria
4 Medical Department, Landeskrankenhaus, Hausmanninger Str 8, A-4560 Kirchdorf a d Krems, Austria
Summary

Alterations of electrolytes and albumin cause metabolic acid-base disorders. It is unclear, however,
to what degree these plasma components affect the overall metabolic acid-base state in the course of
critical illness. We performed serial analyses of the metabolic acid-base state in 30 critically ill
patients over the course of 1 week. We applied a physicalchemical acid-base model and used a
linear regression model to determine the influence of sodium, chloride, unmeasured anions and
albumin on the net metabolic acid-base state. Progressive hypochloraemia was identified as the
main cause of developing metabolic alkalosis. Changes in serum chloride and unmeasured anions
were responsible for changes of 41% and 22% in the metabolic acid-base state, respectively. Sodium
and albumin played a minor role. In conclusion, chloride is the major determinant of metabolic
acid-base state in critical illness.
Keywords

Acid-base imbalance; chloride, critical illness.

. ......................................................................................................

Correspondence to: Georg-Christian Funk

E-mail: georg-christian.funk@meduniwien.ac.at
Accepted: 1 June 2004

Metabolic acid-base disorders can be life-threatening in

critical illness [1]. Understanding the pathogenesis of such
disorders is crucial to recognise underlying problems and
avoid further iatrogenic acid-base disorders.
The classic model of metabolic acid-base disorders
defining acidosis and alkalosis according to bicarbonate
levels has been expanded by the discovery that electrolytes and albumin play a key role in acid-base homeostasis [2]. Based on the physicalchemical approach, the
simultaneous presence of more than one metabolic acidbase disorder was found to be common in critical illness
[3]. Acid-base disorders caused by changes in chloride,
free water, hypoalbuminaemia or unmeasured tissue
anions can co-exist with varying effects on the resultant
metabolic acid-base state of the patient [3].
An analysis of the overall metabolic acid-base state in
critically ill patients revealed a mild metabolic alkalosis
due to unidentifiable causes [4]. However, it is clear that
acid-base balance does not remain static during the course
of critical illness. Therefore, the aim of this study was to

2004 Blackwell Publishing Ltd

determine which components of blood plasma alter the

metabolic acid-base state during the course of critical
illness.
Methods

Patients and sampling

Data were prospectively collected from records at a
medical intensive care unit of the Vienna Medical
University. All samples were routine samples taken from
arterial lines in patients requiring intensive care management. No additional sampling was required and
collected data did not identify patients. The study was
approved by the Ethics Committee of the Medical
University of Vienna and informed consent was
obtained.
Arterial blood gases were measured when clinically
indicated, and at least twice daily at 06:00 and 22:00 h.
Plasma concentrations of albumin, magnesium and
inorganic phosphate were measured once daily at 06:00 h.
1111

G.-C. Funk et al. Acid-base imbalance

Anaesthesia, 2004, 59, pages 11111115
. ....................................................................................................................................................................................................................

pH and PaCO2, ionised calcium and lactate were

measured with a blood gas analyser (ABL 725, Radiometer, Copenhagen, Denmark). Samples of separated
plasma were analysed for concentrations of sodium,
potassium, chloride, magnesium, inorganic phosphate
and albumin using a fully automated analyser (Hitachi
917, Roche Diagnostics GmbH, Mannheim, Germany).
Acid-base analysis
Acid-base state was assessed using a physicalchemical
acid-base model. In this model, PaCO2, non-volatile weak
acids (albumin and inorganic phosphate) and the difference between fully dissociated cations and anions determine plasma pH. Detailed descriptions of this acid-base
model and its application in medicine have been
published [2,3].
From the measured pH and PaCO2, we calculated
bicarbonate and standard base excess using the HendersonHasselbalch equation [5]. Standard base excess is
defined as the titrable base in blood and reflects metabolic
acid-base disorders [6]. Severe metabolic acidosis or
alkalosis were defined as standard base excess outside
the range )5 to +5 mmoll)1.
Apart from changes in the bicarbonate buffer system,
base excess is also influenced by changes of free water,
chloride, albumin and unmeasured anions. Thus, standard
base excess is composed of four subsets:
base excess caused by changes in free water (Base
ExcessSodium);
base excess caused by changes in chloride (Base
ExcessChloride);
base excess caused by changes in albumin (Base
ExcessAlbumin);
base excess caused by changes in unmeasured anions
(Base ExcessUnmeasured anions).
Severe hyperchloraemic acidosis or severe hypochloraemic alkalosis were defined as Base ExcessChloride =
)5 mmoll)1 and Base ExcessChloride = 5 mmoll)1, respectively. The calculation of the base excess subsets and a
detailed description of the physicalchemical acid-base
analysis are given in Appendix A. Reference values were
obtained from a sample of healthy volunteers (n = 10; five
males, five females; mean age 39 (standard deviation 12)
years).
Statistics
Data were analysed using linear regression of the standard
base excess on Base ExcessSodium, Base ExcessChloride, Base
ExcessAlbumin and Base ExcessUnmeasured anions. As the
samples were collected over time, autocorrelation of error
terms was required. We accounted for this autocorrelation by using the autoregressive error model implemented
in the statistical software (SAS Version 8.2, SAS Institute
1112

Inc., Cary, NC) assuming autocorrelation of error terms

of order three, the number of measurements taken per
patient and day. To account for interindividual variance,
separate intercept terms were estimated for each patient.
Alterations in standard base excess, Base ExcessSodium,
Base ExcessChloride, Base ExcessAlbumin and Base
ExcessUnmeasured anions were detected using a linear regression model adjusted for patients. To describe and
compare the importance of these variables for explaining
the variation in standard base excess, two different R2
measures, marginal and partial R2, were computed for
each variable. If R20 denotes the proportion of interindividual variance of standard base excess (i.e. the R2 of a
model only estimating patient-specific intercepts), then
the proportion of intra-individual variance of standard
base excess can be expressed by (1 ) R20 ). Furthermore, if
R2J denotes the R2 of the model additionally including
variable J, then marginal R2 of variable J was computed as
(R2J ) R20 ) (1 ) R20 ), which describes the amount of
intra-individual variance of standard base excess that can
be explained by one variable alone. If R2A and R2A
J
denote the R2 of the full model including all variables and
the R2 of the model including all variables but variable J,
then partial R2 of variable J was computed as (R2A
R2A
J ) (1 R20 ), which describes the amount of intraindividual variance of standard base excess that variable J is
able to explain additionally to all other independent
variables considered. Correlation of independent variables
was assessed by Pearsons correlation coefficients, partialised for the factor patient. Results are summarised as
means and standard deviations (SD) in brackets. A p-value
of less than 0.05 was considered statistically significant.
Results

Of 72 patients admitted during a 2-month period,

30 patients (21 males, nine females) stayed longer than
1 week and were included in the study. Reasons for
admission included respiratory and circulatory failure,
coma and sepsis. Altogether, 556 blood samples (mean 79
(12) per day, were analysed.
Standard base excess and base excess subsets at 06:00 h
on the seven consecutive days are given in Table 1. Mean
daily increases of standard base excess, Base ExcessChloride
and Base ExcessAlbumin were 0.95, 0.58 and 0.16 mmoll)1,
respectively, indicating the progressive development of a
hypochloraemic and hypoalbuminaemic alkalosis leading
to an overall metabolic alkalosis. The other metabolic
acid-base parameters remained unchanged. The increase
of standard base excess over 1 week is shown in Fig. 1.
Correlation of the base excess subsets with standard
base excess as well as marginal and partial R2 is shown in
Table 2. Base ExcessChloride had the greatest impact on

2004 Blackwell Publishing Ltd

Anaesthesia, 2004, 59, pages 11111115

G.-C. Funk et al. Acid-base imbalance
. ....................................................................................................................................................................................................................

Table 1 Changes of acid-base state in 30 critically ill patients over 1 week after admission to ICU.
Day
1
Standard Base Excess; mmoll)1
Base ExcessSodium; mmoll)1
Base ExcessChloride; mmoll)1
Base ExcessAlbumin; mmoll)1
Base ExcessUnmeasured anions; mmoll)1

)1.9
)1.2
)4.7
4.9
)1.0

2
)2.0
)1.0
)4.9
5.2
)1.1

(7.2)
(2.3)
(5.3)
(1.9)
(3.5)

3
(6.4)
(1.8)
(4.8)
(1.7)
(5.0)

0.2
)1.1
)3.7
5.7
)1.0

(6.1)
(1.4)
(5.1)
(1.7)
(4.9)

2.2
)1.0
)2.8
6.0
)0.1

5
(5.5)
(1.5)
(6.0)
(1.6)
(2.8)

2.3
)1.0
)2.0
5.8
)0.8

(5.7)
(1.5)
(5.6)
(1.4)
(2.9)

2.6
)1.1
)1.5
5.7
)0.8

7
(5.5)
(1.5)
(6.3)
(1.4)
(2.9)

3.2
)1.1
)2.1
6.0
0.2

(7.0)
(1.8)
(6.1)
(1.3)
(5.0)

Standard Base Excess (mmol.l1)

Standard base excess, Base ExcessChloride (base excess attributable to changes of serum chloride) and Base ExcessAlbumin (base excess attributable to
changes of serum albumin). Base ExcessSodium (base excess attributable to changes of free water). Base ExcessUnmeasured anions (base excess attributable to unmeasured anions).
Values are mean (standard deviation).

25

Table 2 Correlation of standard base excess with base excess

20

subsets.

15
rp with
standard
base excess

10
5

Marginal R2

Partial R2

0.77*

40%

41%

0.42*

39%

22%

0.05

4%

6%

8%

4%

0
5
10
15
20
25

Day
Figure 1 Increase of standard base excess in 30 critically ill

patients over the course of 1 week after admission to ICU.

Progressive metabolic alkalosis develops with a mean daily
increase of standard base excess by 0.95 mmol.l)1. The central
point and box displays the mean and standard deviation,
respectively, while the whiskers are the maximum and minimum values recorded.

standard base excess independent of changes of other base

excess subsets. Severe hyperchloraemic acidosis was
present in 56 out of 68 (82%) samples with severe
metabolic acidosis. Severe hypochloraemic alkalosis was
present in 44 out of 103 (43%) samples with severe
metabolic alkalosis. Pearson partial correlation coefficients
of the base excess subsets adjusted for patients are
presented in Table 3. A relevant negative correlation
is observed between Base ExcessUnmeasured anions and
Base ExcessChloride.
Discussion

The aim of this study was to determine to what extent

components of blood plasma influence metabolic acidbase state during critical illness. Mild metabolic alkalosis
developed within 7 days after ICU admission. Base

2004 Blackwell Publishing Ltd

Base ExcessChloride (base

excess attributable to
changes of serum chloride)
Base ExcessUnmeasured anions
(base excess attributable to
unmeasured anions)
Base ExcessSodium (base
excess attributable to
changes of free water)
Base ExcessAlbumin (base
excess attributable to
changes of albumin)

0.308*

rp = Pearsons correlation coefficient. Marginal R2 quantifies the

variation in standard base excess attributable to one base excess
subset alone. Partial R2 quantifies the variation in standard base
excess attributable to one base excess subset in addition to all other
subsets. Base ExcessChloride has the strongest correlation with standard
base excess and most influence on changes of standard base excess.
*p < 0.01.

ExcessChloride, and thus changes of chloride relative to

other plasma ions, was found to have greatest impact on
the overall metabolic acid-base state. Hypoalbuminaemia
also contributed to the metabolic alkalosis.
Hypoalbuminaemia is almost ubiquitous in critically ill
patients due to third space losses, plasma dilution or hepatic
failure [3,7]. In a previous report, hypoalbuminaemic
alkalosis was suspected to be the cause of a mild metabolic
alkalosis in 100 critically ill patients [4]. However, the
present longitudinal study suggests that hypochloraemic
alkalosis contributes more to the alkalosis.
Hyper- and hypochloraemia cause acidosis and alkalosis
by altering the strong ion difference [8]. However,
in hyperchloraemic acidosis, simultaneous elevation of
another base excess subset (e.g. Base ExcessAlbumin)
1113

G.-C. Funk et al. Acid-base imbalance

Anaesthesia, 2004, 59, pages 11111115
. ....................................................................................................................................................................................................................

Table 3 Pearsons correlation coefficient of base excess subsets.

Base
Base
Base
ExcessChloride ExcessAlbumin ExcessUnmeasured anions

Base ExcessSodium
0.07*
(base excess
attributable to
changes of free
water)
Base ExcessChloride
(base excess
attributable to
changes of serum
chloride)
Base ExcessAlbumin
(base excess
attributable to
changes of albumin)

0.10**

0.04

0.09**

0.24**

0.06**

*p < 0.05, **p < 0.01.

neutralises the acidifying effect of hyperchloraemia. The

opposite applies in hypochloraemia. Thus, not every
abnormal serum chloride results in a chloride-related
acid-base disorder [3].
Renal loss of chloride is a common cause of hypochloraemic alkalosis [9]. Diuretics are commonly used in
critical care but the incidence of diuretic-induced alkalosis
is unclear. Gastric and intestinal losses of chloride
following vomiting, gastric suction and diarrhoea are also
common causes of this acid-base disorder. Excessive
exogenous alkali and re-feeding alkalosis after a long fast
are less commonly observed [10].
Hypochloraemic alkalosis can also be observed as a
compensatory mechanism during chronic hypercapnia.
Rapid correction of hypercapnia by ventilation can cause
a metabolic alkalosis [11]. Recently, hypochloraemia has
been suggested to be a compensatory metabolic mechanism in lactic acidosis and diabetic ketoacidosis [12,13].
The negative correlation between Base ExcessChloride and
Base ExcessUnmeasured anions (i.e. the more unmeasured
anions, the less chloride) supports this concept.
Infusion of normal saline and other chloride-rich
crystalloids is the most common cause of hyperchloraemic
acidosis. The disorder is also observed in the early stages of
acute renal failure (due to tubular loss of bicarbonate), some
forms of chronic renal failure and in the rare congenital
forms of renal tubular acidosis. Bicarbonate loss via biliary
and pancreatic drainage, urinaryintestinal communications and pancreaticcystic anastomosis after pancreas
transplantation also cause decreased plasma bicarbonate,
with consequent renal retention of chloride and thus
hyperchloraemic acidosis [9,14]. The high prevalence of
severe chloride-mediated metabolic acid-base disorders in
the present material highlights the importance of chloridemediated acid-base disorders in critical illness.
1114

In the present data, multiple correlations between base

excess subsets and thus components of the metabolic acidbase state were observed. Most of the correlations were
weak statistically. Thus, their possible aetiologies and
clinical consequences remain speculative.
Our findings highlight the importance of chloriderelated acid-base changes as determinants of metabolic
acid-base state in critically ill patients. Severe hypochloraemic alkalosis and hyperchloraemic acidosis with an
appreciable effect on the overall metabolic acid-base state
were frequent in this population. This suggests that the
assessment of metabolic acid-base state in an ICU setting
should include measurement of chloride and Base
ExcessChloride, especially when severe deviations of standard base excess are observed. The development of
chloride-related metabolic acid-base disorders may be
avoided. Normal saline is the most common cause for
hyperchloraemic acidosis [3,15,16]. Normal saline has
been associated with a poor outcome in experimental
sepsis and is no longer recommended for volume resuscitation in diabetic ketoacidosis [17,18]. Hypochloraemic
alkalotic compensation in chronic hypercapnia and metabolic acidosis due to lactic acidosis and ketoacidosis should
probably not be regarded as a primary disorder and should
be left untreated. If hypochloraemic alkalosis occurs as a
primary disorder (e.g. following administration of diuretics) specific treatment should include administration of
normal saline or hydrochloric acid [19]. Hyperchloraemic
acidosis, on the other hand, can be treated with diuretics.
Loop diuretics and etacrynic acid are capable of reducing
serum chloride with little effect on serum sodium [9].
In conclusion, progressive hypochloraemic alkalosis
appears to be the main cause of developing metabolic
alkalosis in critical illness. Forty-one percent of the overall
metabolic acid-base changes can be assigned to changes of
serum chloride. Thus, assessment of chloride-related acidbase disorders might be helpful in diagnosis, prevention and
treatment of metabolic disorders in critically ill patients.
Acknowledgement

The authors would like to thank the nursing staff of 13H1

for collecting the printouts of the arterial blood gas
analyses.
References
1 Adrogue HJ, Madias NE. Management of life-threatening
acid-base disorders. First of two parts. New England Journal of
Medicine 1998; 338: 2634.
2 Stewart PA. Modern quantitative acid-base chemistry.
Canadian Journal of Physiology and Pharmacology 1983;
61: 144461.

2004 Blackwell Publishing Ltd

Anaesthesia, 2004, 59, pages 11111115

G.-C. Funk et al. Acid-base imbalance
. ....................................................................................................................................................................................................................

3 Fencl V, Jabor A, Kazda A, et al. Diagnosis of metabolic acidbase disturbances in critically ill patients. American Journal of
Respiratory and Critical Care Medicine 2000; 162: 224651.
4 Story DA, Poustie S, Bellomo R. Quantitative physical
chemistry analysis of acid-base disorders in critically ill
patients. Anaesthesia 2001; 56: 5303.
5 Siggaard-Andersen O, Wimberley PD, Fogh-Andersen N,
et al. Measured and derived quantities with modern pH
and blood gas equipment: calculation algorithms with 54
equations. Scandinavian Journal of Clinical and Laboratory
Investigation 1988; 48 (Suppl. 189): 715.
6 Siggaard-Andersen O, Fogh-Andersen N. Base excess or
buffer base (strong ion difference) as a measure of a nonrespiratory acid-base disturbance. Acta Anaesthesiologica
Scandinavica 1995; 39 (Suppl. 107): 1238.
7 Wilkes P. Hypoproteinaemia, strong-ion difference, and
acid-base status in critically ill patients. Journal of Applied
Physiology 1998; 84: 17408.
8 Gilfix BM, Bique M, Magder S. A physical chemical
approach to the analysis of acid-base balance in the clinical
setting. Journal of Critical Care 1993; 8: 18797.
9 Narins RG, Kupin W, Faber MD, Goodkin DA, Dunfee
TP. Pathophysiology, classification, and therapy of acid-base
disturbances. In: Arieff AI, DeFronzo RA, eds. Fluid,
Electrolyte, and Acid-Base Disorders, 2nd edn. New York:
Churchill Livingstone, 1995: 10598.
10 Galla JH. Metabolic alkalosis. In: Arieff AI, DeFronzo, RA,
eds. Fluid, Electrolyte, and Acid-Base Disorders, 2nd edn. New
York: Churchill Livingstone, 1995: 199221.
11 Madias NE, Adrogue HJ. Acid-base disturbances in pulmonary medicine. In: Arieff AI, DeFronzo, RA, eds. Fluid,
Electrolyte, and Acid-Base Disorders, 2nd edn. New York:
Churchill Livingstone, 1995: 22354.
12 Durward A, Skellett S, Mayer A, et al. The value of chloride: sodium ratio in differentiating the aetiology of metabolic
acidosis. Intensive Care Medicine 2001; 27: 82835.
13 Funk GC, Bauer E, Oschatz E, et al. Compensatory hypochloraemic alkalosis in diabetic ketoacidosis. Diabetologia
2003; 46: 8713.
14 Riley DJ, DeFronzo RS. Acute and chronic renal failure:
acid-base, electrolyte, and fluid disorders. In: Arieff AI,
DeFronzo, RA, eds. Fluid, Electrolyte, and Acid-Base Disorders,
2nd edn. New York: Churchill Livingstone, 1995: 64384.
15 Scheingraber S, Rehm M, Sehmisch C, et al. Rapid saline
infusion produces hyperchloraemic acidosis in patients
undergoing gynecologic surgery. Anesthesiology 1999; 90:
126570.
16 Waters JH, Miller LR, Clack S, et al. Cause of metabolic
acidosis in prolonged surgery. Critical Care Medicine 1999; 27:
21426.
17 Kellum JA. Fluid resuscitation and hyperchloraemic acidosis
in experimental sepsis: improved short term survival and
acid-base balance with Hextend compared with saline.
Critical Care Medicine 2002; 30: 3005.
18 Kitabchi AE, Umpierrez GE, Murphy MB, et al. Hyperglycaemic crisis in patients with diabetes mellitus. Diabetes
Care 2003; 26 (Suppl. 1): 10917.

2004 Blackwell Publishing Ltd

19 Adrogue HJ, Madias NE. Management of life-threatening

acid-base disorders. Second part. New England Journal of
Medicine 1998; 338: 10711.
20 Doberer D, Funk GC, Schneeweiss B. Dilutional acidosis:
an endless story of confusion. Crit Care Med 2003; 31:
3378.
Appendix A

Physicalchemical analysis of metabolic acid-base

disorders according to Gilfix et al. [8]
In normal blood the base excess is zero. Base excess can
be altered by the degree of plasma dilution, changes
of chloride, albumin and unmeasured anions. All
values should be expressed in mEq.l)1, if not otherwise
indicated.
Excess of free water causes acidosis by dilution of an
alkalotic solution, such as plasma by neutral water [20].
Sodium, as the regulated variable that controls the
extracellular fluid volume, is used to assess the effect of
dilution:
Base excess caused by free water effect:

Base ExcessSodium 0:3  Na

observed
Nanormal :

Hyper- and hypochloraemia alter acid-base state. The

effect of changes in chloride can be obtained by first
correcting chloride for changes in free water:

Cl

Nacorrected Cl  Nanormal =Naobserved :

Base excess caused by changes in chloride:

Base ExcessChloride Cl

normal
ClNacorrected :

Albumin is a weak non-volatile acid and so hypoalbuminaemia results in alkalosis. The base excess effect due to
albumin can be calculated as:
Base ExcessAlbumin 0:148  pH
0:818
 Albnormal
Albobserved :
Any change in base excess not caused by changes in
free water, chloride or albumin is attributed to unmeasured anions. Unmeasured anions include anions such as
lactate or ketone bodies. Their contribution to base excess
can be quantified as:
Base ExcessUnmeasured anions
Base Excess
Base ExcessSodium
Base ExcessChloride Base ExcessAlbumin :
Thus, the final sum of all four components is:
Base Excess Base ExcessSodium Base ExcessChloride
Base ExcessAlbumin
Base ExcessUnmeasured anions :

1115