Traditional approach versus Stewart approach for acid–base disorders: Inconsistent evidence

Abstract

Purpose:

The traditional approach and the Stewart approach have been developed for evaluating acid–base phenomena. While some experts have suggested that the two approaches are essentially identical, clinical researches have still been conducted on the superiority of one approach over the other one. In this review, we summarize the concepts of each approach and investigate the reasons of the discrepancy, based on current evidence from the literature search.

Methods:

In the literature search, we completed a database search and reviewed articles comparing the Stewart approach with the traditional, bicarbonate-centered approach to November 2016.

Results:

Our literature review included 17 relevant articles, 5 of which compared their diagnostic abilities, 9 articles compared their prognostic performances, and 3 articles compared both diagnostic abilities and prognostic performances. These articles show a discrepancy over the abilities to detect acid–base disturbances and to predict patients’ outcomes. There are many limitations that could yield this discrepancy, including differences in calculation of the variables, technological differences or errors in measuring variables, incongruences of reference value, normal range of the variables, differences in studied populations, and confounders of prognostic strength such as lactate.

Conclusion:

In conclusion, despite the proposed equivalence between the traditional approach and the Stewart approach, our literature search shows inconsistent results on the comparison between the two approaches for diagnostic and prognostic performance. We found crucial limitations in those studies, which could lead to the reasons of the discrepancy.

Keywords

Henderson–Hasselbalch Stewart anion gap strong ion difference strong ion gap

Introduction

Originally, Henderson¹ recognized that carbon dioxide and bicarbonate were key elements of carbonate mass action. Hasselbalch² developed it into the negative logarithmic pH notation. Henderson–Hasselbalch equation considers bicarbonate one of the strongest buffers and determinants of pH in our physiologic system. In order to separate metabolic and respiratory components in acid–base disorders, the concept of base excess (BE) was first introduced by Siggaard-Andersen et al.³ and became the head of the Copenhagen school. On the other hand, exploiting the flaw of using in vitro concept of BE in a living organism, Schwartz and Relman⁴ developed the bicarbonate-centered approach setting out the relationship between partial pressure of carbon dioxide (pCO₂) and bicarbonate ion (HCO₃^–) in vivo, which became the center of the Boston school. The difference of the two approaches for metabolic components generated the “great trans-Atlantic acid-base debate” between the Boston school and the Copenhagen school.⁵

In the late 1900s, Peter Stewart questioned the bicarbonate-centered approach and the base excess method for acid-base phenomenon.^6–8 In his concept, each variable is classified as a dependent or independent factor in determining the H⁺ concentration of a solution, resulting in pH through the dissociation of water, in order to maintain electrical neutrality.^9,10 Although both the BE approach and the Stewart approach were developed in physio-chemical terms, the Stewart approach is sometimes called “physicochemical,” “modern,” or strong ion approach.^6,7 In contrast, the bicarbonate-centered approach and the base excess approach are called “traditional approach.”¹¹ Currently, most of the modern blood gas analyzers report both HCO₃⁻ and BE for many clinicians to use the traditional approach.

Since then, both the traditional and the Stewart approaches have been relevant subjects for clinical and research discussions. While some experts have suggested that the two approaches are essentially identical,^12,13 clinical researches have still been conducted and discussed which approach has a better performance as a diagnostic or prognostic tool. In this review, we summarize the concepts of each approach and investigate the reasons of this discrepancy, based on current evidence from the literature search, despite the proposed identity.

The traditional approach

Bicarbonate-centered (“Boston”) approach

In the early 1900s, an acid was defined as a substance that is capable of donating hydrogen to a solution, and a base was defined as a substance capable of accepting hydrogen from a solution. Henderson² first recognized that bicarbonate is a unique and important buffer, which has the ability to bind or release hydrogen ions in a solution to keep the pH relatively constant, in a physiologic system at constant pCO₂. Henderson–Hasselbalch equation provides a simple relationship among the respiratory parameter (pCO₂), the non-respiratory parameter bicarbonate (HCO₃^–), and the overall acidity parameter (pH).¹⁴

Based on the equation, Schwartz and Relman⁴ developed the CO₂/HCO₃^– approach predicting the nature of acid–base disorders. Although it is relatively easy to understand and to apply in clinical settings, there are some weaknesses we need to consider. Since there are non-bicarbonate buffers such as albumin and hemoglobin, a change in bicarbonate concentration does not always reflect the total amount of non-respiratory acids or bases.¹⁵ Furthermore, the equation listing pCO₂ and bicarbonate as determinants of pH can mislead their interdependence.

BE and standard BE (“Copenhagen”) approach

In 1948, Singer and Hastings¹⁶ introduced the concept of the buffer base, which is the sum of all plasma buffer anions and is composed of bicarbonate ion and non-volatile, weak acid buffers (mainly albumin and phosphate). It is shown that a change in a buffer base corresponds to a change in the metabolic component of acid–base balance and develops into the BE methodology.^17,18

In 1960, Siggaard-Andersen et al.^3,19 measured the plasma bicarbonate concentration at a fixed temperature and partial pCO₂ and compared the difference between their results and a reference value. When corrected by a constant, this difference yields the BE, which represents the amount of acid or alkali that must be added to 1 L of oxygenated blood, exposed in vitro to a pCO₂ of 40 mmHg to achieve the average normal pH of 7.40.^19,20

Blood BE measures the metabolic component that is independent from the respiratory component and incorporates the effect of hemoglobin as a buffer.^19,20 The most commonly used formula for calculating the BE is the Van Slyke equation, developed by Siggaard-Andersen¹⁹

\begin{array}{l} BE = {\begin{cases} [{HCO}_{3}^{-}] - 24.4 + (2.3 * [Hemoglobin (Hb)] + 7.7) \\ * (pH - 7.4) \end{cases}} \\ * (1 - 0.023 * [Hb]) \end{array}

The BE equation suffers from inaccuracy in vivo with changes in pCO₂, possibly due to equilibration across the entire extracellular fluid space, which is composed of whole blood and interstitial fluid.^7,21,22 Therefore, the equation was modified to “Standardize” the effect of hemoglobin on CO₂ titration in order to improve the accuracy in vivo⁷

Standard base excess (SBE) = 0.9287 * {\begin{cases} [{HCO}_{3}^{-}] \\ - 24.4 + 14.83 \\ * (pH - 7.4) \end{cases}}

However, the standard base excess (SBE) is still slightly subject to pCO₂ change.⁷ Furthermore, this equation assumes normal non-buffer ion levels; however, a decrease in albumin or phosphate, which is commonly encountered in intensive care unit (ICU), results in more unstable SBE.^7,8 In addition, the BE and SBE methods are unable to detect complicated acid–base disorders or identify different types of metabolic acidosis.

Anion gap

The anion gap (AG), the difference between unmeasured plasma anions and the unmeasured plasma cations,⁸ is an additional diagnostic tool to assess the metabolic components of the acid–base equilibrium. Albumin and phosphate, one of the circulatory proteins, mainly account for the AG under normal conditions. The rest of the possible candidates are composed of urate, lactate, ketone bodies, sulfate, salicylates, penicillins, citrate, pyruvate, and acetate.^23,24

This additional diagnostic tool provides new insight to the traditional approach, classifying metabolic acidosis into normal AG acidosis and high AG acidosis. However, severe pH disturbances and changes in the concentration of serum albumin, which behaves as an anion, have a significant impact on the AG.^25,26 Those disadvantages lower the sensitivity and specificity of this diagnostic tool to detect metabolic acidosis.

A noticeable attempt to improve the practical AG was the introduction of the corrected anion gap (AGc). The most popular AGc is “albumin-corrected” AG. For each 10 g/L decrement in the serum albumin concentration, the AG is expected to decrease by 2.5 mmol/L and needs to be corrected to compensate for abnormality of serum albumin concentration.⁸ However, this AGc attributes a fixed negative charge to albumin, taking no consideration for pH effects on the imidazole groups of albumin.⁷ In addition, this AGc ignores the phosphate contribution to all of the weak acids that might need to be considered.^27–29

Stewart approach

Concept of the Stewart approach

Stewart^9,10 questioned the traditional approach for acid–base equilibrium evaluation. He modeled a solution that contained a complex mixture of ions of constant charge over the physiological pH range (strong ions), non-volatile proton donor/acceptors which transfer H^⁺ within the physiological pH range (weak acid/base), and the volatile bicarbonate–CO₂ buffer system.⁸ Key aspect of Stewart’s concept was the classification of each variable as dependent or independent in determining the H⁺ concentration of the solution. In his theory, there are three responsible variables to independently determine the dissociation of water, and consequently the hydrogen ion concentration, in order to maintain electrical neutrality: (1) strong ion difference (SID), (2) total concentration of weak acids (A_TOT), and (3) partial pCO₂ of the solution.^8,30 Thus, in the Stewart’s approach, metabolic disorders are the results of changes in SID or A_TOT.^7,31

Apparent SID and effective SID

Apparent SID (SIDa) represents the difference between measured strong cations and strong anions.⁷ With the development of devices capable of detecting “unmeasured” ions (which we could not measure routinely), current calculation of the SIDa contains the following ions⁷

\begin{array}{l} SIDa = ([{Na}^{+}] + [K^{+}] + [{Ca}^{2 +}] + [{Mg}^{2 +}]) \\ - ([{Cl}^{-}] + [{L - lactate}^{-}]) \end{array}

where Na denotes sodium, K denotes potassium, Ca denotes calcium, Mg denotes magnesium, and Cl denotes chloride.

On the other hand, SID calculated to account for electrical neutrality is viewed as the effective SID (SIDe).⁷ The SIDe can be calculated as the sum of bicarbonate and weak acids ([A^–)), mainly albumin and phosphate⁸

SIDe = [{HCO}_{3}^{-}] + [{Alb}^{-}] + [{Pi}^{-}]

where Alb denotes albumin and Pi denotes inorganic phosphate

SIG

Although the law of electrical neutrality in the body requires SIDa and SIDe to be equal, failure to measure the concentration of all strong and weak ions in plasma yields a gap between the two. Thus, SIG, the difference of SIDa and SIDe, quantifies [unmeasured anions] – [unmeasured cations] of both strong and weak ions.⁷

One of the theoretical advantages of SIG over AG is the pure representation of unmeasured ions. Although both AG and SIG represent unmeasured ions, the “unmeasured” ions derived from AG are composed of [Mg²⁺], [Ca²⁺], [A^–] (mainly albumin and phosphate), [Lactate^–], and [other ions] clinicians do not routinely measure, whereas the unmeasured ions expressed by the SIG are composed of just [other ions]. While normal AG ranges from 7 to 17 mEq/L when using [K⁺] for the calculation, SIG is close to zero in normal situations.⁸ Although the albumin-corrected AG eliminates the effect of hypo/hyper albuminemia, the gap still persists.

A_TOT

Consideration of A_TOT alternations for acid–base disorders is another key aspect of this approach compared to the traditional one.^7,31 A_TOT, representing all non-bicarbonate buffers, is made up of mainly serum albumin and other minor charges such as phosphate and globulins.^7,31 In the Stewart approach, an increase in A_TOT would result in metabolic acidosis and a decrease would result in metabolic alkalosis.⁷

There is a controversy over the existence of A_TOT acidosis/alkalosis.^32,33 Although observations in vitro show that alterations in albumin concentration can affect acidity, there is no credible demonstration that the living organism, especially the liver, regulates albumin to maintain acid–base homeostasis.³⁰ One of the explanations is that the theoretical slight weak acid loss secondary to hypoproteinemia is compensated for by a decrease in SID (adjusted SID) without changes in pH, HCO₃^–, and BE as commonly seen in ICU, rather than a complex acid–base disorder such as a mixed metabolic acidosis/hypoalbuminemic alkalosis.^20,34,35

Although the traditional approach and the physicochemical approach originated from different concepts as mentioned above, their mathematical comparison showed very few differences once model coefficients are estimated in the consistent manner.¹² Representation of the bicarbonate buffers is almost identical, and representation of non-bicarbonate buffering in the van Slyke equation can be derived from the equations of Stewart. Representation of electrical neutrality comes from the preservation of charge equation described by Singer and Hastings.¹⁶ For both approaches, measurement of plasma protein concentration is essential if unmeasured anions are to be distinguished from protein buffers.¹² However, many clinical researches have still been conducted on which method is more informative and useful in clinical situations, and there has been no consistent conclusion. In order to find the reasons of the consistency, we conducted a literature search focusing on two main comparisons: diagnostic and prognostic performance of those approaches.

Literature search

The PubMed Database was initially searched from inception to 15 November 2016 to compare the physicochemical approach with the traditional approach. The search was performed with the relevant medical subject heading terms and strategies: ((SID) OR (strong ion gap)) AND ((AG) OR (BE)). References of selected publications were individually inspected for additional articles that might have been omitted or overlooked in the electronic database search.

The inclusion criteria for the review were (a) studies using both approaches for the same population and (b) studies comparing the diagnostic and/or predictive abilities directly or indirectly. Studies using the traditional methods with AG but without AGc were excluded because non-corrected AG lacks consideration of abnormal albumin concentration commonly seen in the ICU, and many studies already have shown that the simple AG cannot detect acid–base disorders that the Stewart method can identify.^36–39 Nonhuman studies, case reports, abstracts, and unpublished or any studies in which full text was not available were excluded.

Results

Our electrical literature search revealed 192 studies. One hundred and five nonhuman studies, case reports, abstracts, or otherwise irrelevant studies were excluded. Among 87 potentially relevant articles, we exclude 41 studies that did not compare the two approaches as for diagnostic and/or predictive performance and 29 studies that did not calculate corrected AG for the comparison. Thus, the remaining 17 articles were included in this review. Eight studies compared their diagnostic abilities and 12 articles compared their prognostic performances (Table 1).

Table 1.

Differences in studied population, measured ions, calculation of variables, and references among articles comparing the two approaches.

Author	Studied population	n	Measurements of electrolytes	What is AGc corrected for?	Reference of AGc (mmol/L)	Calculation of SIG	Reference of SIG	Main results and comments
Fencl et al.⁴⁰	ICU	152	Blood gas analyzer	Alb	>21 (based on healthy subjects	Mg²⁺, Ca²⁺, Alb^–, Pi^–	>14 (based on healthy subjects)	While unmeasured strong anions represented by SIG detected 35% of patients with normal BE, AGc found 59% of hidden metabolic acid–base disturbances
Cusack et al.⁴¹	ICU	100	Blood gas analyzer	Alb	>12	Mg²⁺, Ca²⁺, Alb^–, Pi^–	>0	SIG and SIDe in Stewart principle appear to offer no advantage in prediction of outcome
Rocktaeschel et al.⁴²	ICU	300	Central laboratory	Alb	N/A	Mg²⁺, Ca²⁺, Lac^–, Alb^–, Pi^–	N/A	AUROC curves of AGc, SIDe, and SIG for mortality prediction were relatively small
Hucker et al.⁴³	Accident and emergency department	1424	Central laboratory	Alb	N/A	Mg²⁺, Ca²⁺, Lac^–, Alb^–, Pi^–	N/A	All of each single variable in both approach have similar and unreliable predictive value
Martin et al.²⁴	Surgical ICU, trauma	2152	Central laboratory	Alb, Lac	N/A	Mg²⁺, Ca²⁺, Lac^–, Alb^–, Pi^–	>0	AUROC for mortality was strong for AGc with AUROC values of 0.68 compared with that for SIG (0.54)
Gunnerson et al.²⁷	ICU	9799	Central laboratory	Alb, Pi, Lac	N/A	Mg²⁺, Ca²⁺, Lac^–, Alb^–, Pi^–	>50% of SBE	AGc identified only 84% of patients classified as SIG acidosis SIG, not AGc, was an independent predictors of mortality (OR 1.065; 95% CI 1.03–1.10; p = 0.001)
Dubin et al.³³	ICU	935	Central laboratory	Alb	3 SD above or below the mean of 7 normal volunteers	Mg²⁺, Ca²⁺, Alb^–, Pi^–	3 SD above or below the mean of 7 normal volunteers	When AGc was included in acid–base analysis, the Stewart approach did not offer any diagnostic or prognostic advantages
Kaplan and Kellum²⁸	ICU, major trauma	78	Central laboratory	Alb, Pi, Lac	N/A	Mg²⁺, Ca²⁺, Lac^–, Alb^–, Pi^–	N/A	Although AGc had acceptable ROC curves (0.86) for 28-day mortality, it was significantly inferior to SIG (0.96) (p = 0.018)
Boniatti et al.²⁹	ICU (medical and surgical)	175	Central laboratory	Alb, Pi, Lac	>=17	Mg²⁺, Ca²⁺, Lac^–, Alb^–, Pi^–	>2	There was significant difference between survivors and non-survivors in SIG (p = 0.01), but not in AGc (p = 0.11)
Abdulraof Menesi et al.⁴⁴	Kidney transplant	83	Central laboratory	Alb	>16	Mg²⁺, Ca²⁺, Alb^–, Pi^–	>3	A greater percentage of patients presented with an increase in unexplained anions by SIG than by AGc (42 vs 32%, respectively) (p value; N/A)
Ratanarat et al.⁴⁵	Medical and surgical ICU	410	Blood gas analyzer	Alb	>12	Mg²⁺, Ca²⁺, Alb^–, Pi^–	>0	According to ROC curves, the predictive ability to discriminate between survivors and non-survivors of AGc and SIG were 0.72 and 0.67, respectively
Zheng et al.⁶	Nephrology ICU, metabolic acidosis	78	Central laboratory	Alb	N/A	Mg²⁺, Ca²⁺, Lac^–, Alb^–, Pi^–	N/A	SIG value was associated with mortality at 24 h, 72 h, 1 week, 1 month, and 3 months after acute kidney injury, whereas AGc was not associated with mortality at each follow-up
Antonogiannaki et al.⁴⁶	Emergency department	365	Central laboratory	Alb	>17 (based on healthy volunteers)	Mg²⁺, Ca²⁺, Alb^–, Pi^–	>6 (based on healthy volunteers)	Significantly fewer patients with unmeasured anions acidosis were identified with AGc than those with SIGc (p=.0001)
Ho et al.⁴⁷	ICU	6878	Blood gas analyzer	Alb	N/A	Mg²⁺, Ca²⁺, Alb^–, Pi^– (, Lac^–)	N/A	The abilities to predict hospital mortality in SIG (AUROC 0.52) and SIDe (0.63) are modest, whereas AGc (0.67) and BE (0.69) has stronger ability to differentiate between survivors and non-survivors
Morgan et al.⁴⁸	CPB	60	Blood gas analyzer	Alb	>20	Ca²⁺, Lac^–, Alb^–, Pi^–	>4	AUROC of SIG for detecting “unmeasured anions” was significantly higher than that of AGc (0.81 vs 0.79; p = 0.048)
Guérin et al.⁴⁹	Chronic respiratory failure	128	Central laboratory	Alb	8 healthy volunteers	Mg²⁺, Ca²⁺, Lac^–, Alb^–, Pi^–	8 healthy volunteers	The Stewart approach detected high SIDe in 13% of normal SBE and in 20% of normal AGc, and low SIDe in 22% of non-elevated HCO_3⁻, providing better diagnostic performance
Shen et al.⁵⁰	Acute pancreatitis	186	Central laboratory	Alb, Lac	13 health volunteers	Mg²⁺, Ca²⁺, Lac^–, Alb^–, Pi^–	13 health volunteers	SIG, but not AGc, had significant independent correlations with disease severity

ICU: intensive care unit; AGc: corrected anion gap; BE: base excess; SBE: standard base excess; SID: strong ion difference; SIDe: effective strong ion difference; SIG: strong ion gap; SIGc: corrected strong ion gap; Mg: magnesium; Ca: calcium; Alb: albumin; Pi: inorganic phosphate; Lac: lactate; HCO₃: bicarbonate; AUROC: area under receiver operating characteristic curve; ROC: receiver operating characteristic; N/A: not applicable; OR: odds ratio; CI: confidential interval; SD: standard deviation.

Inconsistent results on the superiority of one approach over the other approach

While 10 studies have shown the potential superiority of the Stewart approach,^{6,27–29,40,44,46,48–50} four articles failed to show the superiority of the physicochemical approach over the traditional one,^33,41–43 and three articles even showed greater strength of the traditional method than the modern one.^24,45,47

Discussion

Reasons for inconsistent results on diagnostic performance

Our literature search shows a discrepancy over the ability to detect acid–base disturbances on diagnostic performance of the two approaches. There are several possible explanations for the discordance. The first thing to be mentioned is the calculation of each variable in both approaches. Table 1 shows there are many differences in inclusive ions, especially lactate, phosphate, and magnesium ion, of each calculation of AGc and SIG among the studies. In addition, cumulative differences or errors in each variable should be considered. As each mathematical equation contains more measurement, there could be greater variability in the parameters, such as SIDa, SIDe, and SIG in Stewart approach, because the differences are exaggerated via complicated mathematical calculations.⁵¹ As shown by Matousek et al.,¹² there should be no difference between the approaches from a mathematical perspective. However, it is true only when the same ions are measured and taken into account and each measurement is accurate. Those differences of each calculation and potential cumulative errors could lead to the discordance about the usefulness as a diagnostic tool between the two approaches.

Another potential reason is technological differences or errors in measuring each variable. Morimatsu et al.⁵² showed that chloride measurements, made with point-of-care blood gas and electrolyte analyzers, differed significantly from those made using central laboratory biochemistry analyzers, resulting in different SID values and assessments of the acid–base status. Nguyen et al.⁵¹ compared two central laboratory analyzers for electrolyte measurement and reported that the biochemistry laboratory analyzers have large differences from each other. It should be noted that 12 of 17 articles in our review measured electrolytes using central laboratory analyzers, many of which are currently using diluted blood sample and indirect ion selective electrodes in order to measure the electrolytes, rather than blood gas analyzers (Table 1). Measurements by this method are affected by hypoalbuminemia and could be inaccurate compared with the ones measured by blood gas analyzers.⁵³ Studies that used indirect ion selective electrodes could lead to wrong calculation and acid–base interpretation, which could make an implausible conclusion. Thus, interpretation of the results in papers comparing these approaches needs attention on the analyzer that each study used. We found a wide variety of machines and technologies used to measure pH, pCO₂, and electrolytes in those articles, which could be one of the reasons for the inconsistent results on this topic.

Reference value of each parameter is another problem. The dependency on site recommends reference value should be determined in each institution.⁴⁶ However, our review showed that while only five studies collected healthy controls for the reference,^{33,40,46,49,50} other studies used pre-determined numbers,^{29,41,44,45,48} and the method for reference value selection in those studies was not specified.^{6,24,27,28,42,43} These incongruences of reference value due to arbitrary choice may cause a variety of discordant results. As there is no consensuses on the normal range of each variable, especially in the Stewart approach, we recommend that future researchers collect healthy controls for reference in each research institute.

We also need to pay attention to the differences in the normal result range between the two approaches in these studies, since more than one parameter in each method aims to represent the same concept. For example, Boniatti et al. defined normal SBE as −5 to +5 and normal SIDe as 38–42 mEq/L. Since changes in BE represent changes in SIDe if A_TOT is normal,⁸ the large difference of normal range (10 vs 4) could mislead the interpretation. This sort of “unfair” comparison might be one reason of the inconsistent results.

Studied populations need another consideration. Several studies showed that patients with renal failure,⁵⁴ liver diseases,⁵⁵ sepsis and trauma⁵⁶ often have accumulations of unmeasured anions. However, Dubin et al.³³ and Ho et al.⁴⁷ reported patient demographics in their studies; the percentage of patients with shock, acute renal failure, and hepatic failure was only 13%, 13%, and 4%, respectively, in one study, and liver diseases were only 2% in the other. A study by Cusack et al.⁴¹ included a high proportion of post-elective surgery patients, who generally have low severity of illness and low mortality. Hucker et al.⁴³ did not provide details about reasons for admission, patients’ illness severity, or the underlying medical conditions of patients in their accident and emergency department. For patients with severe illness, measuring more ions and involving them into variables such as AG and SIG could demonstrate their potential ability to detect unmeasured anions, revealing more detailed acid–base disturbances, no matter which approach is used. For populations with a small number of severe patients, measuring more ions would not be needed for detailed analysis of acid–base disorders. Thus, the combination of variety of the studied populations and the aforementioned differences of calculation in each variable among those studies could be one of the reasons of their inconsistent results.

Reasons for inconsistent results on prognostic performance

Those factors as the potential reasons for inconsistent results on diagnostic performance could also yield a controversy about the prognostic performance of the two approaches. Some authors have investigated the predictive value of the traditional approach and the physicochemical Stewart approach, mainly, AGc versus SIG. One of their questions is “Is there any association between AGc or SIG and outcomes?” or “Can AGc or SIG levels be used as a marker of poor outcomes?” Here the difference of measured and involved ions in each calculation could again mislead the conclusion. Simple comparison of AGc with SIG does not always answer these particular questions. Although both parameters represent unmeasured ions, consideration of lactate for AGc and SIG depends on each individual study. A bulk of evidence has shown that the level of lactate is associated with poor prognosis.^57,58 If we would like to simply compare the prognostic abilities of the two approaches, the contribution of lactate should be removed from their equations. Only five of all 17 articles remove the effect of lactate from their calculations of AGc and SIG.

It is not only lactate but also other ions, such as magnesium, that need to be considered when comparing the two methods. The changes in magnesium concentration are usually so small that they may usually be neglected, but this simplification is not applicable if the changes are significant. Theoretically, an increased level of magnesium reduces the AG, increases SID, and does not change SIG. There are no studies so far that compare these approaches for patients with abnormal serum magnesium concentrations. Thus, radical question of the comparison should not be “Is there any association between AGc or SIG and outcomes?” but “Is there any association between unmeasured anions that we does not measure in clinical practice and outcomes?” In order to answer this question directly, we need to exclude the contribution of lactate and other measured ion.

Finally, we cannot forget the effect of fluids used for resuscitation, which lead to iatrogenic acidosis. Hayhoe et al.⁵⁹ found 40% of acidosis were attributed to the use of polygeline, which acts as an acid resulting in increased unmeasured circulating anions. Similarly, gelatin-derived colloids have also been found to iatrogenically increase the SIG due to increased unmeasured anions.⁵⁹ None of the studies included in our review provided detailed information about the type and volume of administered resuscitation fluids. This iatrogenically fluid-induced increment of SIG and metabolic acidosis in less critical patients is not expected to have many adverse outcomes, and therefore, the prognostic value of these indices of the Stewart approach could be wrongly affected.

Conclusion

Although the traditional approach and the Stewart approach are seen as complementary giving the same information about the acid–base phenomena despite their different concepts, our literature search shows inconsistent results on the comparison between the traditional approach and the physicochemical approach for their diagnostic and prognostic performance. Many studies to date have crucial limitations in comparing these approaches. Those limitations are considered the reasons for the discrepancy in clinical researches.

Footnotes

Declaration of conflicting interests

The author(s) declared no potential conflicts of interest with respect to the research, authorship, and/or publication of this article.

Funding

The author(s) received no financial support for the research, authorship, and/or publication of this article.

ORCID iDs

Satoshi Kimura

Hiroshi Morimatsu

References

Henderson

LJ.

The theory of neutrality regulation in the animal organism. Am J Physiol 1907; 18: 427–448.

Hasselbalch

. Die Berechnung der Wasserstoffzahl des blutes auf der freien und gebundenen Kohlensaure desselben, und die Sauerstoffbind-ung des Blutes als Funktion der Wasserstoffzahl. Biochem Z 1916; 78: 112–144.

Siggaard-Andersen

Engel

Jorgensen

et al . A micro method for determination of pH, carbon dioxide tension, base excess and standard bicarbonate in capillary blood. Scand J Clin Lab Invest 1960; 12: 172–176.

Schwartz

Relman

AS.

A critique of the parameters used in the evaluation of acid-base disorders. “Whole-blood buffer base” and “standard bicarbonate” compared with blood pH and plasma bicarbonate concentration. N Engl J Med 1963; 268: 1382–1388.

Bunker

JP.

The great trans-atlantic acid-base debate. Anesthesiology 1965; 26: 591–594.

Stewart

. Independent and dependent variables of acid-base control. Respir Physiol 1978; 33(1): 9–26.

Stewart

. How to understand acid-base: a quantitative acid-base primer for biology and medicine. New York: Elsevier North Holland Inc, 1981.

Siggaard-Andersen

. The acid-base status of the blood. 4th ed. Copenhagen: Munksgaard, 1974, and Baltimore: William & Wilkins Company, 1974.

Kellum

Elbers

PWG

. Stewart’s textbook of acid-base Lulu Enterprises, UK Ltd, 2009.

10.

Todorović

Nešovic-Ostojić

Milovanović

et al . The assessment of acid-base analysis: comparison of the “traditional” and the “modern” approaches. Med Glas (Zenica) 2015; 12(1): 7–18.

11.

Kishen

Honoré

Jacobs

et al . Facing acid–base disorders in the third millennium—the Stewart approach revisited. Int J Nephrol Renovasc Dis 2014; 7: 209–217.

12.

Matousek

Handy

Rees

SE.

Acid-base chemistry of plasma: consolidation of the traditional and modern approaches from a mathematical and clinical perspective. J Clin Monit Comput 2011; 25(1): 57–70.

13.

Wooten

EW.

Calculation of physiological acid-base parameters in multicompartment systems with application to human blood. J Appl Physiol (1985) 2003; 95(6): 2333–2344.

14.

Van Slyke

. Studies of acidosis I. The bicarbonate concentration of the blood plasma; its significance and its determination as a measure of acidosis. J Biol Chem 1917; 30: 289–246.

15.

Hughes

Brain

MJ.

A simplified bedside approach to acid–base: fluid physiology utilizing classical and physicochemical approaches. Anaesth Intens Care Med 2013; 14(10): 445–452.

16.

Singer

Hastings

AB.

An improved clinical method for the estimation of disturbances of the acid-base balance of human blood. Medicine (Baltimore) 1948; 27(2): 223–242.

17.

Astrup

Jorgensen

Siggaard-Andersen

et al . The acid-base metabolism. A new approach. Lancet 1960; 1(7133): 1035–1039.

18.

Severinghaus

JW.

Acid-base balance nomogram: a Boston-Copenhagen detente. Anesthesiology 1976; 45(5): 539–541.

19.

Siggaard-Andersen

The van Slyke equation. Scand J Clin Lab Invest Suppl 1977; 146: 15–20.

20.

Kellum

JA.

Determinants of blood pH in health and disease. Crit Care 2000; 4(1): 6–14.

21.

Brackett

Cohen

Schwartz

WB.

Carbon dioxide titration curve of normal man. Effect of increasing degrees of acute hypercapnia on acid-base equilibrium. N Engl J Med 1965; 272: 6–12.

22.

Prys-Roberts

Kelman

Nunn

JF.

Determination of the in vivo carbon dioxide titration curve of anaesthetized man. Br J Anaesth 1966; 38(7): 500–509.

23.

Staempfli

Constable

PD.

Experimental determination of net protein charge and A(tot) and K(a) of nonvolatile buffers in human plasma. J Appl Physiol (1985) 2003; 95(2): 620–630.

24.

Martin

Murray

Berne

et al . Diagnosis of acid-base derangements and mortality prediction in the trauma intensive care unit: the physiochemical approach. J Trauma 2005; 58(2): 238–243.

25.

Morgan

Cowley

Weier

et al . Stability of the strong ion gap versus the anion gap over extremes of pCO₂ and pH. Anaesth Intensive Care 2007; 35(3): 370–373.

26.

Figge

Rossing

Fencl

The role of serum proteins in acid-base equilibria. J Lab Clin Med 1991; 117(6): 453–467.

27.

Gunnerson

Saul

et al . Lactate versus non-lactate metabolic acidosis: a retrospective outcome evaluation of critically ill patients. Crit Care 2006; 10(1): R22.

28.

Kaplan

Kellum

JA.

Comparison of acid-base models for prediction of hospital mortality after trauma. Shock 2008; 29(6): 662–666.

29.

Boniatti

Cardoso

Castilho

et al . Acid-base disorders evaluation in critically ill patients: we can improve our diagnostic ability. Intensive Care Med 2009; 35(8): 1377–1382.

30.

Adrogué

Madias

NE.

Assessing acid-base status: physiologic versus physicochemical approach. Am J Kidney Dis 2016; 68(5): 793–802.

31.

Rastegar

Clinical utility of Stewart’s method in diagnosis and management of acid-base disorders. Clin J Am Soc Nephrol 2009; 4(7): 1267–1274.

32.

Balasubramanyan

Havens

Hoffman

GM.

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. Crit Care Med 1999; 27(8): 1577–1581.

33.

Dubin

Menises

Masevicius

et al . Comparison of three different methods of evaluation of metabolic acid-base disorders. Crit Care Med 2007; 35(5): 1264–1270.

34.

Wilkes

Hypoproteinemia, strong-ion difference, and acid-base status in critically ill patients. J Appl Physiol (1985) 1998; 84(5): 1740–1748.

35.

Siggaard-Andersen

Fogh-Andersen

Base excess or buffer base (strong ion difference) as measure of a non-respiratory acid-base disturbance. Acta Anaesthesiol Scand Suppl 1995; 107: 123–128.

36.

Miller

Subramanian

Safi

et al . Influenza A 2009 (H1N1) virus in admitted and critically ill patients. J Intensive Care Med 2012; 27(1): 25–31.

37.

Kaplan

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(5): 1120–1124.

38.

Antonini

Piva

Paltenghi

et al . The early phase of critical illness is a progressive acidic state due to unmeasured anions. Eur J Anaesthesiol 2008; 25(7): 566–571.

39.

Kaplan

Cheung

Maerz

et al . A physicochemical approach to acid-base balance in critically ill trauma patients minimizes errors and reduces inappropriate plasma volume expansion. J Trauma 2009; 66(4): 1045–1051.

40.

Fencl

Jabor

Kazda

et al . Diagnosis of metabolic acid-base disturbances in critically ill patients. Am J Respir Crit Care Med 2000; 162(6): 2246–2251.

41.

Cusack

Rhodes

Lochhead

et al . The strong ion gap does not have prognostic value in critically ill patients in a mixed medical/surgical adult ICU. Intensive Care Med 2002; 28(7): 864–869.

42.

Rocktaeschel

Morimatsu

Uchino

et al . Unmeasured anions in critically ill patients: can they predict mortality? Crit Care Med 2003; 31(8): 2131–2136.

43.

Hucker

Mitchell

Blake

et al . Identifying the sick: can biochemical measurements be used to aid decision making on presentation to the accident and emergency department. Br J Anaesth 2005; 94(6): 735–741.

44.

Abdulraof Menesi

Verzola

Villaggio

et al . Evaluation of metabolic acidosis in patients with a kidney graft: comparison of the bicarbonate-based and strong ion-based methods. Transplant Proc 2011; 43(4): 1055–1062.

45.

Ratanarat

Sodapak

Poompichet

et al . Use of different approaches of acid-base derangement to predict mortality in critically ill patients. J Med Assoc Thai 2013; 96(Suppl. 2): S216–S223.

46.

Antonogiannaki

Mitrouska

Amargianitakis

et al . Evaluation of acid-base status in patients admitted to ED-physicochemical vs traditional approaches. Am J Emerg Med 2015; 33(3): 378–382.

47.

Lan

Williams

et al . A comparison of prognostic significance of strong ion gap (SIG) with other acid-base markers in the critically ill: a cohort study. J Intensive Care 2016; 4: 43

48.

Morgan

Anstey

Wolf

MB.

A head to head evaluation of 8 biochemical scanning tools for unmeasured ions. J Clin Monit Comput 2017; 31(2): 449–457.

49.

Guérin

Nesme

Leray

et al . Quantitative analysis of acid-base disorders in patients with chronic respiratory failure in stable or unstable respiratory condition. Respir Care 2010; 55(11): 1453–1463.

50.

Shen

Yang

et al . The prognostic value of the strong ion gap in acute pancreatitis. J Crit Care 2016; 36: 140–145.

51.

Nguyen

Vincent

Hamm

et al . The reproducibility of Stewart parameters for acid-base diagnosis using two central laboratory analyzers. Anesth Analg 2009; 109(5): 1517–1523.

52.

Morimatsu

Rocktäschel

Bellomo

et al . Comparison of point-of-care versus central laboratory measurement of electrolyte concentrations on calculations of the anion gap and the strong ion difference. Anesthesiology 2003; 98(5): 1077–1084.

53.

Van den Ancker

Haagen

van der Voort

. Direct sodium measurement prevents underestimation of hyponatremia in critically ill patients. Intensive Care Med 2015; 41(3): 553–554.

54.

Naka

Bellomo

Morimatsu

et al . Acid-base balance in combined severe hepatic and renal failure: a quantitative analysis. Int J Artif Organs 2008; 31(4): 288–294.

55.

Kellum

Kramer

Pinsky

MR.

Strong ion gap: a methodology for exploring unexplained anions. J Crit Care 1995; 10(2): 51–55.

56.

Clowes

George

Villee

et al . Muscle proteolysis induced by a circulating peptide in patients with sepsis or trauma. N Engl J Med 1983; 308(10): 545–552.

57.

Bakker

Coffernils

Leon

et al . Blood lactate levels are superior to oxygen-derived variables in predicting outcome in human septic shock. Chest 1991; 99(4): 956–962.

58.

Blow

Magliore

Claridge

et al . The golden hour and the silver day: detection and correction of occult hypoperfusion within 24 hours improves outcome from major trauma. J Trauma 1999; 47(5): 964–969.

59.

Hayhoe

Bellomo

Liu

et al . The aetiology and pathogenesis of cardiopulmonary bypass-associated metabolic acidosis using polygeline pump prime. Intensive Care Med 1999; 25(7): 680–685.

Traditional approach versus Stewart approach for acid–base disorders: Inconsistent evidence

Abstract

Purpose:

Methods:

Results:

Conclusion:

Keywords

Introduction

The traditional approach

Bicarbonate-centered (“Boston”) approach

BE and standard BE (“Copenhagen”) approach

Anion gap

Stewart approach

Concept of the Stewart approach

Apparent SID and effective SID

SIG

ATOT

Literature search

Results

Inconsistent results on the superiority of one approach over the other approach

Discussion

Reasons for inconsistent results on diagnostic performance

Reasons for inconsistent results on prognostic performance

Conclusion

Footnotes

Declaration of conflicting interests

Funding

ORCID iDs

References

A_TOT