American Journal of Emergency Medicine (2012) 30, 391-398

Original Contribution

Defining metabolic acidosis in patients with septic shock using Stewart approach

Jihad Mallat MD?, Damien Michel MD, Pascale Salaun MD, Didier Thevenin MD, Laurent Tronchon MD

Service de Reanimation Polyvalente, Centre Hospitalier Dr Schaffner de Lens, 62307 Lens Cedex, France

Received 3 November 2010; revised 23 November 2010; accepted 24 November 2010

Abstract

Purpose: The aim of this study was to define the nature of metabolic acidosis in patients with septic shock on admission to intensive care unit (ICU) using Stewart method. We also aimed to compare the ability of standard base excess (SBE), anion gap , and corrected AG for albumin and lactate (AGcorr) to accurately predict the presence of unmeasured anions (UA).

Patients and Methods: Thirty consecutive patients with septic shock were prospectively included on ICU admission. Stewart equations modified by Figge were used to calculate the strong ion difference and the strong ion gap .

Results: Most patients had multiple underlying mechanisms explaining the metabolic acidosis. Unmeasured anions and hyperchloremia were present in 70% of the patients. Increased UA were present in 23% of patients with normal values of SBE and [HCO^–]. In these patients, plasma [Cl^–] was significantly lower compared with patients with low SBE and increased UA (103 [102-106.6] vs 108 [106-111] mmol/L; P = .01, respectively). Corrected AG for albumin and lactate had the best correlation with SIG (r2 = 0.94; P b .0001) with good agreement (bias, 0, and precision, 1.22) and highest area under the receiver operating characteristic curve (0.995; 95% confidence interval, 0.87-1) to discriminate SIG acidosis.

3

Conclusions: Patients with septic shock exhibit a complex metabolic acidosis at ICU admission. High UA may be present with normal values of SBE and [HCO^–] as a result of associated “relative” hypochloremic alkalosis. Corrected AG for albumin and lactate offers the most accurate bedside alternative to Stewart calculation of UA.

3

(C) 2012

Introduction

Metabolic acidosis is one of the most frequent acid-base disturbances occurring in patients with septic shock [1]. It is generally associated with greater morbidity and mortality [2]. Thus, understanding the nature of this derangement and

* Corresponding author.

E-mail address: [email protected] (J. Mallat).

consequently identifying approaches that can avoid/or correct it are of utmost importance.

Traditional methods used to interpret acid-base data, which includes the determination of the standard base excess (SBE), bicarbonate [HCO^–], and anion gap (AG) [3], give an insight into the derangement, but they provide little information as to the source of the problem [4]. The SBE is a calculated figure, derived from PaCO2 and arterial pH. Its calculation assumes normal plasma protein and electrolyte contents [5,6]. The observed AG also ignores the role of the main nonbicarbonate

3

0735-6757/$ – see front matter (C) 2012 doi:10.1016/j.ajem.2010.11.039

buffers in blood plasma such as plasma proteins and inorganic phosphate. Thus, when electrolyte and protein abnormalities are present, simplistic Clinical interpretation of acid-base distur- bances may be misleading. The common finding of hypoalbu- minemia among critically ill patients, notably in septic shock, requires adjustment of AG for abnormal albumin concentra- tions to improve its usefulness [7].

An alternative evaluation to this conventional model is the Mathematical model based on physiochemical principles described by Stewart [8] and modified by Figge et al [5,9]. The model proposes that 3 independent variables determine pH in plasma by primarily changing the degree of water dissociation into hydrogen and hydroxide ions: the strong ion difference (SID, difference between fully dissociated anions and cations), the PaCO2, and the total weak acid concentration (consisting mainly of albumin and phosphate). This method allows the clinician to quantify individual components of acid- base abnormalities and provides insight into their pathogenesis [10]. Using physicochemical evaluation, a few studies showed that traditional approach often fails to identify acid-base disorders in the population of critically ill patients [11].

Although Stewart model theory has been used previously to better understand metabolic derangements in intensive care patients, no studies have been published evaluating its use at the early phase of septic shock. Moreover, the precise composition of metabolic acidosis in patients with septic shock is not well known. Therefore, using the quantitative physicochemical methodology, the aims of our study were, first, to compare the ability of SBE, AG, and corrected AG for albumin and lactate (AGcorr) to identify the presence of unmeasured anions (UA) in patients with septic shock on admission to intensive care unit (ICU) and, second, to define the nature of metabolic acidosis in this population.

Methods

This prospective observational study took place in a single, mixed medical and surgical adult ICU with 15 beds at a general hospital in France from September 2007 to July 2008. Patients were considered eligible for the study if they had a diagnosis of septic shock according to the current definition

[12] with less than 24 hours of organ dysfunction. Patients were excluded if they did not have all the laboratory variables needed for the proposed acid-base evaluation. This study was approved by the institutional ethics committee. Because the blood tests and data collected in this study were all standard clinical practice, informed consent was waived. Written informed consent was obtained from healthy volunteers.

On admission, demographic data, volume and type of fluid administered before ICU admission, renal function, and the use of mechanical ventilation were recorded. Acute Physiology and Chronic Health Evaluation II score [13] and Sepsis-related Organ Failure Assessment score [14] were calculated. Arterial blood samples were taken from the Arterial line at the time of admission for blood gases to measure pH, PaCO2, and ionized

calcium (GEM PREMIER 3000; Instrumentation Laboratory Co, Paris, France). At the same time, further arterial samples were also analyzed for plasma sodium, potassium, chloride (ion selective electrode, Roche Diagnostics, Meylan, France), plasma magnesium (chlorophosphonazo 3 colorimetric technique, Roche Diagnostics), phosphate (ammonium molybdate complex colorimetric technique, Roche Diagnostics), albumin (immuno- turbidimetry technique, Roche Diagnostics), and lactate (enzy- matic and colorimetric method, Roche Diagnostics).

Calculated variables

Bicarbonate and SBE were calculated using the Henderson- Hasselbach and Van Slyke equations, respectively [15,16]. The AG was calculated as follows [3]:

AG = [Na⁺] + [K⁺] – [HCO^–] – [Cl^–]. Anion gap was

3

corrected for the effect of abnormal albumin concentration

[7] and lactate using the following formula: AGcorr = AG +

0.25 x (45 – [albumin in gram per liter]) – [lactate in millimolar]. Based on the SBE, metabolic acidosis was defined as SBE of less than -2 mEq/L [17].

Physicochemical analysis was performed using the Stewart equations [8] modified by Figge et al [5,9] to consider the effects of plasma proteins. The effective SID (SIDe) was calculated as follows [9]: SIDe = 2.46 x 10^pH-8 x PaCO2 + [albumin^–] + [Pi^–]. This equation takes into account the role of weak acids (albumin and phosphate) and CO₂ in the balance of electrical charges in plasma water. In this equation, PaCO2 is measured in millimeters of mercury. [Albumin^–] and inorganic phosphate [Pi^–] (millimolar) were calculated from the measured albumin (gram per liter), phosphate [Pi] (millimolar), and pH [11]: [albumin^–] = [albumin] x (0.123 x pH – 0.631) and [Pi^–] = [Pi] x (0.309 x

pH – 0.469). The apparent SID (SIDa) was calculated: SIDa = [Na⁺] + [K⁺] + [Ca2⁺] + [Mg2⁺] – [Cl^–] – [lactate^–] (all concentrations in milliequivalent per liter). The strong ion gap (SIG) was calculated by subtracting the SIDe from the SIDa: SIG = SIDa – SIDe as previously described [8,9]. A positive value was defined as representing the presence of UA, which must be included to account for the measured pH. The concentration of plasma nonvolatile buffers charges [A^–] was calculated as follows [5,9]: [A^–] milliequivalent per liter = [albumin^–]+ [Pi^–].

Reference values for the SIG, AG, and electrolytes were obtained from arterial blood samples of 13 healthy volunteers. The normal values considered were those between the 2.5 and 97.5 percentiles of values from the healthy volunteers. Chloride was adjusted to water excess/ deficit as follows: [Cl^–]corrected = [Cl^–]observed x ([Na⁺] normal/[Na⁺]observed) [11]. Hyperlactatemia was defined as lactate levels greater than 2 mmol/L.

Statistical analysis

Data are presented as median [interquartile range, 25-75]. Data were assumed to be nonparametric, and comparisons

between independent groups were analyzed using the Mann- Whitney U test. The strength of the relationship between SIG and SBE, AG, and AGcorr was assessed using Pearson correlation coefficient. The ability of these variables to discriminate a SIG acidosis (SIG, N8 mEq/L) was quantified by the area under the receiver operating characteristic (ROC) curve (AUC) with a 95% confidence interval (CI). Their predictive values were compared by calculating sensitivity, specificity, and positive and negative predictive values. We performed Bland-Altman plot to analyze the agreement between SIG and AGcorr. We calculated the bias (the mean difference between the 2 variables [d]) after subtracting d from AGcorr and the precision as the SD of the bias [18]. The limits of agreement were defined by +-1.96 SDs. Statistical analysis was performed using SPSS (SPSS for Windows release 17.0, Chicago, IL). P b .05 was considered statistically significant.

Results

Thirty patients with septic shock were included consec- utively in the study. Their demographic, clinical, and outcome variables are shown in Table 1. Acid-base and electrolyte data for the study population are presented in Table 2. The ranges of normal values extracted from arterial blood samples of 13 normal volunteers (10 males; 27.5 [22-35] years) expressed as percentiles 2.5th and

97.5th were as follows: [99-104] mEq/L for chloride, [37-42] mEq/L for SIDe, [12-19] mEq/L for AG, and [0-8] mEq/L for SIG. Thus, a SIG greater than 8 mEq/L was

Table 1 Demographic, clinical, and outcome variables

Variables	Median [interquartile range, 25-75]
pH	7.33 [7.27-7.38]
Paco2	37 [33-45]
Sodium (mmol/L)	136 [132-138]
Potassium (mmol/L)	3.9 [3.5-4.6]
Phosphate (mmol/L)	1.34 [0.93-1.74]
Chloride (mmol/L)	102 [101-108]
Chloride corrected (mmol/L)	107 [104-110.5]
[HCO–] (mmol/L)	20 [15.3-24.9]
Magnesium (mmol/L)	0.86 [0.74-1]
Calcium (mmol/L)	1.08 [1.04-1.78]
SBE (mmol/L)	-5.2 [-9.5-0.13]
Lactate (mmol/L)	2.5 [1.3-4]
Albumin (g/L)	21.4 [17.8-26.4]
AG (mmol/L)	15.3 [12-19.2]
AGcorr (mmol/L)	18.5 [14.6-21.3]
SIDa (mEq/L)	39 [34-42]
SIDe (mEq/L)	28.5 [24-33]
SIG (mEq/L)	10.7 [6.4-12]
SBE standard base excess, AG anion gap, AGcorr anion gap corrected for albumin and lactate, SIDa apparent strong ion difference, SIDe effective strong ion difference, SIG strong ion gap.

considered elevated. The 3 female volunteers were con- firmed as definitely nonpregnant.

The presence of UA defined by elevated SIG occurred in 21 (70%) patients. Hyperlactatemia was present in 18 (60%) patients. Hyperlactatemia was severe (lactate, N5 mEq/L) in only 5 (17%) patients. Hypoalbuminemia was found in all patients. Hyperchloremia ([Cl^–]corrected, N104 mmol/L) occurred in 21 (70%) patients, and in only 2 of 21 patients, this was an isolated hyperchloremia with no increased SIG or lactate levels. The contributions of the 3 main causes of metabolic acidosis (hyperlactatemia, hyperchloremia, and increased levels of UA) are presented in Table 3. Metabolic acidosis based on SBE was found in 19 (63%) patients, whereas metabolic alkalosis (SBE, N+2 mEq/L) was present in only 3 (10%) patients.

Table 2 Acid-base and electrolyte data

Table 3 Distribution of the 3 main underlying mechanisms of metabolic acidosis

N 30

Age (y) 76.5 [57.5-78.5] Sex (male/female) 19/12 Weight, kg 81 [72-94] APACHE II score 26 [20-35] Admission SOFA score 11 [8.5-13] Time to ICU admission (h) 2 [1.5-3] Mechanical ventilation, n (%) 28 (93) Mortality in the ICU, n (%) 14 (47) Serum creatinine (mg/dL) 2.3 [1.3-3.2] Fluids before ICU (mL) 1300 [500-2850] Normal saline (mL) 1000 [500-2625] Lactate ringer (mL) 250 [0-500] Infection source Lung, n (%) 13 (43) Postoperative, n (%) 10 (34) Urinary tract, n (%) 1 (3) Meningitis, n (%) 2 (7) Catheter/bloodstream, n (%) 1 (3) Other, n (%) 3 (10)

APACHE II indicates Acute Physiology and Chronic Health Evaluation II; SOFA, Sepsis-related Organ Failure Assessment.

Underlying mechanism n, (%)

Increased SIG 21 (70) Increased lactate 18 (60) Increased chloride corrected 21 (70) Increased SIG + lactate 10 (33) Increased SIG + chloride corrected 14 (47) Increased lactate + chloride corrected 12 (40) Increased SIG + lactate + chloride corrected 7 (23)

Increased SIG is defined as more than 8 mEq/L; increased chloride corrected, as more than 104 mmol/L; increased lactate, as more than 2 mmol/L.

Nonacidosis SBE/elevated SIG (n = 7)		Acidosis SBE/elevated SIG (n = 14)		P
[Cl–]corr (mmol/L)	103 [102-106.6]	108	[106-111]	.01
SIDe mEq/L	32.7 [30.5-33.8]	24.4	[19.3-28]	.0001
Normal saline (mL)	0 [0-250]	1500	[500-3500]	.007
Lactate ringer (mL)	500 [250-500]	0	[0-300]	.15
Serum creatinine	2.45 [1.8-3.1]	2.45	[1.3-2.8]	.8
(mg/dL)
[HCO–] (mmol/L)	24.9 [24-25.7]	15.9	[13-18.8]	.0001
pH	7.37 [7.35-7.39]	7.27	[7.23-7.34]	.02
Albumine (g/L)	20.2 [17.8-21.8]	19	[17-25]	.9
[A–] (mEq/L)	7.7 [7.1-8]	8	[7.1-9.4]	.4
Nonacidosis SBE/elevated SIG is defined as SBE more than -2 mEq/L and SIG more than 8 mEq/L; acidosis SBE/elevated SIG, as SBE less than -2 mEq/L and SIG more than 8 mEq/L. [Cl–]corr indicates chloride corrected; [A–], plasma nonvolatile buffers charges. Data are expressed as median and interquartile range.

Of the 8 patients with normal SBE, 6 (75%) had a low SIDe (SIDe, b37 mEq/L), and 2 of the 3 patients with metabolic alkalosis (based on SBE) had also a low SIDe. Conversely, of the 19 patients with metabolic acidosis according to the traditional approach based on SBE, all also had a low SIDe. Consequently, the use of Stewart method allowed the additional diagnosis of metabolic acidosis in 8 (27%) patients. We could only identify 1 patient in whom hypoalbuminemia was the unique cause of metabolic alkalosis identified by high SBE and elevated [HCO^–] with normal SIDe and normal SIG.

Table 4 Clinical and Biochemical parameters for the subgroup of raised SIG samples with or without metabolic acidosis based on SBE

3

Standard base excess missed the presence of UA based on elevated SIG in 7 (23%) patients. In these patients with nonacidosis SBE (SBE, N-2 mEq/L) and increased SIG, the median of [Cl^–]corrected was significantly lower and the SIDe, significantly higher compared with the group of patients with low SBE (SBE, b-2 mEq/L) and increased SIG (103 [102-106.6] vs 108 [106-111] mmol/L; P = .01

and 32.7 [30.5-33.8] vs 24.4 [19.3-28] mEq/L; P b .0001,

respectively). There was no difference between these groups regarding [A^–] (7.7 [7.1-8] vs 8 [7.1-9.4] mEq/L; P = .4) (Table 4). Thus, the failure of SBE to detect the presence of UA in these patients could be explained by the alkalinizing effect of this “relative” hypochloremia. The patients with low SBE/increased SIG received more saline solution before ICU admission compared with patients with nonacidosis SBE/increased SIG (1500 [500-3500] vs 0 [0-250] mL; P = .007, respectively). There were no differences in the volume of Ringer-lactate solution received (Table 4). On the other hand, the Stewart method identified the presence of UA in 14 (64%) of 22 patients with normal AG and in only 1 (10%) of 10 patients with normal AGcorr.

Regression analysis confirmed that SBE and SIG were poorly correlated (r2 = 0.18; P = .02; Fig. 1). Strong ion gap and AGcorr showed a very good correlation (r2 = 0.94; P b

.0001; Fig. 1) and good agreement by Bland-Altman analysis. The bias was 0, and the precision was 1.22. The limits of agreement were therefore -2.4 and 2.4 (Fig. 1).

The ability of SBE, AG, and AGcorr to predict the presence of UA was compared by calculating the ROC curve analysis for each variable. Overall, AGcorr had the best discriminatory ability, with an area under the ROC curve (AUC) of 0.995 (95% CI, 0.874 – 1; Table 5).

Discussion

The main findings of this study are, first, that 23% of the patients with an increase of UA showed normal SBE and [HCO^–] levels because of the simultaneous presence of relative hypochloremic metabolic alkalosis. Second, we found a strong relationship between AGcorr and SIG with excellent agreement in our patients with septic shock. Third, UA, excluding lactate, were found to be the main underlying mechanisms of metabolic acidosis, as was hyperchloremia.

3

serum bicarbonate, SBE, and AG are commonly used to assess acid-base disorders [3]. However, it is recognized that this method can fail to identify the complex metabolic disturbances seen in critically ill patients, and so is generally inadequate in explaining them [11]. Assessment of UA, using the principles described by Stewart, may overcome these problems. Although this method has been validated in critically ill patients, there is a few data regarding its usefulness in patients with septic shock in whom profound hypoalbumi- nemia and complex electrolytes disorders are frequent.

Fencl et al [11] had already demonstrated that SBE fails as a measure of metabolic acidosis in critically ill patients. In this study, low SID was unnoticed by changes in SBE because the low SID acidosis was masked by the alkalinizing effect of hypoalbuminemia, present in all patients. McAuliffe et al [6] described the “primary hypoproteinemic alkalosis” in hypoalbuminemic ICU patients with positive SBE and elevated [HCO^–], discarding conditions associated with UA, SID, and electrolytes disorders. Nevertheless, the actual role of hypoalbuminemia to produce metabolic alkalosis has been recently challenged [19]. According to Dubin et al [19], we could only detect 1 patient fulfilling the criteria of primary hypoproteinemic alkalosis among profound septic shock patients with hypoalbuminemia.

3

In our study, the confounding factor in the interpretation of acid-base disorders was not hypoalbuminemia; the alkalinizing factor was relative hypochloremia. Indeed, there was no difference regarding the concentrations of plasma nonvolatile buffers charges [A^–] between the group of patients with low SBE/elevated SIG and the other group with nonacidosis SBE/elevated SIG. Therefore, it seems that the alkalinizing effect related to plasma nonvolatile buffers is identical between these groups. However, the corrected

Fig. 1 Regression and Bland-Altman analysis between metabolic variables. A, Linear regression between AG and the SIG. B, Linear regression between AGcorr and SIG. C, Linear regression between SBE and SIG. D, Agreement between AGcorr and SIG (bias, 0, and precision, 1.22).

chloride concentration was significantly lower in the group of patients with nonacidosis SBE/elevated SIG leading to an alkalinizing effect by increasing SIDe compared with the

Table 5 Area under the curve of calculated ROC curves for each of the variable examining their ability to predict an increase of UA (SIG, N8 mEq/L)

Sens indicates sensitivity; Spec, specificity; PPV, positive predictive value; NPV, negative predictive value; AG, anion gap.

Values given for SBE less than -2 mEq/L, AG more than 19 mEq/L, and

AGcorr more than 17 mEq/L.

Variable	AUC	95% CI	Sens	Spec	PPV	NPV
SBE	0.715	0.520-86	70	50	74	45.5
AG	0.865	0.69	35	90	87.5	41
AGcorr	0.995	0.87	95	100	100	92

other group. Although the value of corrected chloride concentration was in the reference range of our laboratory, in fact, it was abnormally low (relative hypochloremia), preventing a decrease of SID to compensate the effect of hypoalbuminemia. Thus, this relative decrease of corrected chloride concentration, and thereby, this relative hypo- chloremic alkalosis offsets the effect of increased UA on SBE. In fact, base excess is claimed to be equal to the deviation of SID from its normal value [20]. However, this is true only if the plasma concentrations of the nonbicarbonate buffers (albumin and phosphate) are normal. When this condition is not met, the reference state for “normal SID” must be adjusted. Wilkes [21] showed that the loss of weak acid secondary to hypoproteinemia is compensated by a renal-mediated increase in [Cl^–], so SID decreases without change in pH. In accordance with our results, Tuhay et al

[22] found that 20% of their critically ill patients had severe hyperlactatemia (lactate level, N4 mmol/L) with normal pH, [HCO^–], and base excess because of a concomitant presence of hypochloremic alkalosis. The presence of relative hypochloremic alkalosis in our patients with normal SBE might be related to the fact that these patients received less saline solution (0 [0-250] vs 1500 [500-3500] mL; P = .007) as Fluid loading.

3

Accordingly with previous reports, SIG and AGcorr were tightly correlated and showed excellent agreement in our patients with septic shock [19,23]. Moviat et al [23] studied 50 patients with metabolic acidosis and found an excellent relationship between AGcorr and SIG (r2 = 0.94) with good agreement. In their study, 22 (44%) patients were in septic shock. A recent study in critically ill patients found AGcorr to be a good discriminator to disclose increased UA (AUC, 0.89) [17]. We have confirmed this finding in patients with septic shock and have demonstrated the AGcorr to be more reliable and a simple alternative to the calculation of SIG. The clinical usefulness of AGcorr is demonstrated in Table 5. The high value for AUC indicates that within a population of patients, AGcorr is the best screening test for UA.

The metabolic acidosis seen in our patients with septic shock is a complex disorder, primarily resulting from hyperchloremia and UA (Table 3). In agreement with our results, Moviat et al [23] found a 98% increase of SIG and 80% of hyperchloremia in an unselected group of critically ill patients with metabolic acidosis. Recently, Noritomi et al

[24] also found that patients with severe sepsis and septic shock exhibit a complex metabolic acidosis at ICU admission, caused predominantly by hyperchloremic and UA acidosis. However, the source and biochemical nature of the UA constituting the SIG were unclear. These anions may be generated in peripheral tissues during global hypoxic states [25]. The liver has been suggested to be a source of UA in experimental sepsis [26]. In addition, it has been shown that the concentrations of anions normally associated with the Krebs acid cycle are elevated in the plasma of the patients with high AG metabolic acidosis [27]. On the whole, the nature of the UA in patients with septic shock is largely unknown.

There has been controversy over what constitutes a “normal” range of SIG. Indeed, this is reflected in what one could term a transatlantic acid-base debate because American studies describe abnormal values at around 5 mEq/L in critically ill patients [28,29], whereas studies from Europe and Australasia found far higher values [4]. The use of resuscitation fluids that contain UA (eg, gelatins) could be the explanation, but this has not been established. However, any biochemical tool requires local calibration against a reference population. Otherwise, results can be misinter- preted, and diagnostic conclusions, skewed [30]. The SIG is no exception. Any laboratory reporting SIG values needs to establish local CIs, incorporating measurement and popula- tion variability. Some authors have defined reference values for the Stewart approach from the blood samples of healthy

subjects taken as control [17,19,31]. In their studies, the cutoff values of an elevated SIG varied from a SIG greater than 6 mEq/L to a SIG greater than 8.9 mEq/L. Our results for the reference range values of Stewart variables are consistent with those observations. Furthermore, a contro- versial point is the importance of a raised SIG in clinical practice. Several studies found a significant association between SIG and mortality, whereas others did not. However, these discrepancies between studies might be related to different populations and to the methodologies used. Although in this study, we did not evaluate the relationship between SIG and outcome, we think that the presence of UA in the circulation especially early in the course of illness or injury before resuscitation portends a poor prognosis.

The occurrence of hyperchloremic acidosis in this study might be attributable, partially, to pre-ICU admission resuscitation with isotonic saline 1000 [500-2625] mL in patients already exhibiting an impairment of renal function at admission (serum creatinine 2.3 [1.3-3.2] mg/dL). This can be deduced from the frequent occurrence of hyperchloremia in our patients in relation to the plasma sodium concentra- tion. Isotonic saline solution has equal concentrations of sodium and chloride (154 mmol/L). This results in a reduction of SID, which, in turn, produces an increase in the number of hydrogen ions to preserve electrical neutrality. Hyperchloremic acidosis is a well-known phenomenon in the ICU [32]. Another reason for hyperchloremia is a possible shift of chloride from the interstitial or intracellular compartments as suggested by Kellum et al [33] who demonstrated that some degree of endogenous hyperchlor- emia occurs in endotoxemic animals, which was not related to intravenous infusion or diminished renal excretion. Although studies of ICU patients have failed to detect a significant effect on survival attributable to hyperchloremic acidosis [34], hyperchloremia has been shown to cause hypotension, renal dysfunction, and increment in plasma cytokine levels [35]. Therefore, it would seem prudent to avoid chloride-rich fluids in sepsis despite controversy on whether acidosis results in physiologic injury or is just a side effect of illness. At present, the best evidence for acidosis- induced organ injury is mainly from animal studies, thus making any specific recommendation difficult.

Hyperlactatemia was the third cause of metabolic acidosis in the present study. The importance of hyperlactatemia as a marker of shock and its prognostic significance are well known. Nevertheless, pure hyperlactatemia was found in only 7 (23%) patients. A raised SIG with a normal lactate was more commonly seen (37% of patients) than a raised lactate with a normal SIG (23% of patients). This phenomenon has been reported by others and may be related to the fact that the production and clearance of UA and lactate are not directly coupled [21,26]. Kellum et al [26,36] have demonstrated net uptake of UA via skeletal muscle and gastrointestinal tract, in contrast to lactate, which is hepatically metabolized.

The advantage of using a different acid-base approach as in this study would be a greater sensitivity in diagnosis and a better understanding of the causes of acid-base disorders. Thus, in clinical practice, a metabolic acidosis with high SIG or high AGcorr in patients with septic shock despite normal SBE might be an early indicator of hypoperfusion and oxygen debt. Optimization of Tissue oxygenation and perfusion, therefore, become the primary therapeutic goals. Conversely, a metabolic acidosis based only on SBE (SBE, b-2 mEq/L, without increased SIG), if not identified as being hyperchloremic acidosis, may be misinterpreted as a sign of inadequate tissue oxygenation because of low cardiac output. Chloride is important in acid-base dis- turbances and should always be seen in relation to sodium. Hyperchloremia with hypernatremia or hypochloremia with hyponatremia will not change the SID and, thus, will not affect the acid-base balance. In this study, we did not evaluate possible changes in treatment based on change in diagnostic evaluation, but we believe that this is worthy of further evaluation.

The present study has some limitations. First, the number

of patients included is small. Second, we were not able to describe the patients with sepsis on admission to the emergency department before they were well resuscitated. This would have provided an important answer about the origin of early hyperchloremia. Nevertheless, we believe that the characterization of the patients at ICU admission is an important marker for therapeutic and prognostic follow-up. Third, serial measurements might have allowed a more comprehensive understanding and better insight into the mechanisms of metabolic acidosis.

Conclusion

The present study suggests that patients with septic shock exhibit a complex metabolic acidosis at ICU admission. Stewart approach to acid-base disorders has important Therapeutic implications and can be used to differentiate between UA acidosis and hyperchloremic acidosis. Increased UA were present in 23% of patients with septic shock with normal values of [HCO^–] and SBE because of a concomitant presence of relative hypochloremic alkalosis. Corrected AG offers the most accurate bedside alternative to Stewart calculation of UA.

3

References

1. Mecher C, Rackow EC, Astiz ME. Unaccounted for anion in metabolic acidosis during severe sepsis in human. Crit Care Med 1991;19: 705-11.
2. Gunnerson KJ, Kellum JA. Acid-base and electrolyte analysis in critically ill patients: are we ready for the new millennium? Curr Opin Crit Care 2003;9:468-73.
3. Oh MS, Carroll HJ. The anion gap. N Engl J Med 1977;297:814-7.
4. Cusack RJ, Rhodes A, Lochhead P, et al. The strong ion gap does not have prognostic value in critically ill patients in a mixed medical/sur- gical adult ICU. Intensive Care Med 2002;28:864-9.
5. Figge J, Rossing TH, Fencl V. The role of serum proteins in acid-base equilibria. J Lab Clin Med 1991;117:453-67.
6. McAuliffe JJ, Lind LF, Leith DE, et al. Hypoproteinemic alkalosis. Am J Med 1986;81:86-90.
7. Figge J, Jabor A, Kadza A, et al. Anion gap and hypoalbuminemia. Crit Care Med 1998;26:1807-10.
8. Stewart PA. Modern quantitative acid-base chemistry. Can J Physiol Pharmacol 1983;61:1444-61.
9. Figge J, Mydosh T, Fencl V. Serum proteins and acid-base equilibria: a follow-up. J Lab Clin Med 1992;120:713-9.
10. Murray DM, Olhsson V, Fraser JI. Defining acidosis in postoperative cardiac patients using Stewart’s method of strong ion difference. Pediatr Crit Care Med 2004;5:240-5.
11. Fencl V, Jabor A, Kazda A, et al. Diagnosis of metabolic acid-base disturbances in critically ill patients. Am J Resp Crit Care Med 2000; 162:2246-51.
12. Bone RC, Sibbald WJ, Sprung CL. The ACCP-SCCM consensus conference on sepsis and organ failure. Chest 1992;101:1481-3.
13. Knaus WA, Draper EA, Wagner DP, et al. APACHE II: a Severity of disease classification system. Crit Care Med 1985;13:818-29.
14. Vincent JL, Moreno R, Takala J, et al. The SOFA (Sepsis-related Organ Failure Assessment) score to describe organ dysfunction/failure. On behalf of the Working Group on Sepsis-Related Problems of the European Society of Intensive Care Medicine. Intensive Care Med 1996;22:707-10.
15. Henderson LJ. The theory of neutrality regulation in the animal organism. Am J Physiol 1908;21:427-8.
16. Siggaard-Andersen O. The Van Slyke equation. Scand J Clin Lab Invest 1977;37(Suppl 146):15-20.
17. Park M, Taniguchi LU, Noritomi DT, et al. Clinical utility of standard base excess in the diagnosis and interpretation of metabolic acidosis in critically ill patients. Braz J Med Biol Res 2008;41:241-9.
18. Bland JM, Altman DG. Statistical methods for assessing agreement between two methods of clinical measurement. Lancet 1986;i: 307-10.
19. Dubin A, Menises MM, Masevicius FD, et al. Comparison of three different methods of evaluation of metabolic acid-base disorders. Crit Care Med 2007;35:1264-70.
20. Siggaard-Andersen O, Fogh-Andersen N. Base excess or buffer base (strong ion difference) as a measure of non-respiratory acid- base disturbance. Acta Anesthesiol Scand 1995;39(Suppl 107): 123-8.
21. Wilkes P. Hypoproteinemia, strong ion difference, and acid-base status in critically ill patients. J Appl Physiol 1998;84:1740-8.
22. Tuhay G, Pein MC, Masevicius FD, et al. Severe hyperlactatemia with normal base excess: a quantitative analysis using conventional and Stewart approaches. Crit Care 2008;12:R66.
23. Moviat M, Van Haren F, van der Hoeven H. Conventional or physicochemical approach in intensive care unit patients with metabolic acidosis. Crit Care 2003;7:R41-5.
24. Noritomi DT, Soriano FG, Kellum JA, et al. Metabolic acidosis in patients with severe sepsis and septic shock: a longitudinal quantitative study. Crit Care Med 2009;37:2733-9.
25. Kaplan LJ, Kellum JA. Initial pH, Base deficit, lactate, anion gap, strong ion difference, and strong ion gap predict outcome from major Vascular injury. Crit Care Med 2004;32:1120-4.
26. Kellum JA, Bellomo R, Kramer DJ, et al. Hepatic anion flux during acute endotoxemia. J Appl Physiol 1995;78:2212-7.
27. Forni LG, McKinnon W, Lord GA, et al. Circulating anions usually associated with the Krebs cycle in patients with metabolic acidosis. Crit Care 2005;9:R591-5.
28. Gunnerson KJ, Roberts G, Kellum JA. What is normal strong ion gap (SIG) in healthy subjects and critically ill patients without acid-base abnormalities. Crit Care Med 2003;31:A111.
29. Kellum JA. Acid-base disorders and strong ion gap. Contrib Nephrol 2007;156:158-66.
30. Morimatsu H, Rocktaschel J, Bellomo R, et al. Comparison of point- of-care versus central laboratory measurement of electrolytes con- centrations on calculations of the anion gap and the strong ion difference. Anesthesiology 2003;98:459.
31. Funk GC, Doberer D, Richling N, et al. The strong ion gap and outcome after cardiac arrest in patients treated with therapeutic hypothermia: a retrospective study. Intensive Care Med 2008;35: 232-9.
32. Constable PD. Hyperchloremic acidosis: the classic example of strong ion acidosis. Anesth Analg 2003;96:919-22.
33. Kellum JA, Bellomo R, Kramer DJ, et al. Etiology of metabolic acidosis during saline resuscitation in endotoxemia. Shock 1998;9: 364-8.
34. Gunnerson KJ, Saul M, He S, et al. Lactate versus non-lactate metabolic acidosis: a retrospective outcome evaluation of critically ill patients. Crit Care 2006;10:R22.
35. Kellum JA, Song M, Almasri E. Hyperchloremic acidosis increases circulating inflammatory molecules in experimental sepsis. Chest 2006;130:962-7.
36. Kellum JA, Bellomo R, Kramer DJ, Pinsky MR. Fixed acid uptake by visceral organs during early endotoxemia. Adv Exp Med Biol 1997; 411:275-9.