Strong ions and the analysis of acid-base disturbances (Stewart approach)

BACKGROUND — Most internists use an approach to acid-base disturbances that was popularized by Relman and Schwartz in the 1960s [1,2]. This "traditional" approach accepts the Brønsted-Lowry definitions of acids and bases. The hydrogen ion concentration is expressed in blood as a function of the ratio between the partial pressure of carbon dioxide (PCO₂) and the serum bicarbonate (HCO₃^-) concentration as defined by the following equation:

where K is the dissociation constant for the reaction and PCO₂ is equal to the partial pressure of carbon dioxide.

An analysis of metabolic acidosis using the traditional approach is presented in detail separately. (See "Approach to the adult with metabolic acidosis".)

By comparison, many surgeons, critical care specialists, and anesthesiologists have embraced an approach introduced by Peter Stewart in 1981 [3-8]. This alternative approach, termed "strong ion difference" (SID) or the Stewart approach, will be reviewed briefly in this topic.

STRONG ION DIFFERENCE — The cardinal tenet of the Stewart approach is that the serum bicarbonate (HCO₃^-) concentration does not alter blood pH. Stewart defined acids as ions that shift the dissociation equilibrium of water to a higher concentration of hydrogen ions (H⁺) and a lower concentration of hydroxide (OH^-). This is in contrast to the Brønsted-Lowry approach, which defines acids as proton donors and bases as proton acceptors.

The Stewart approach is based upon mathematics and simple physicochemical principles. Stewart identified six simultaneous equations that define the hydrogen ion concentration in terms of its dissociation equilibrium with various constituents present in plasma [9]. A polynomial solution for the six simultaneous equations solved for the hydrogen ion concentration led to the conclusion that the serum bicarbonate is a dependent and not an independent variable. As a result, the serum bicarbonate concentration is not a determinant of blood pH.

According to this theory as modified by Figge and Watson, the independent variables responsible for changes in acid-base balance are [10,11]:

●The "strong ion difference" (SID), the difference between the completely dissociated cations and anions in plasma. Because not all strong ions can be measured, the "apparent SID" (SIDa) is calculated from the net charge of the major strong cations, sodium (Na⁺), potassium (K⁺), calcium (Ca²⁺), and magnesium (Mg²⁺), and the major strong anions, chloride (Cl^-) and lactate.

●The plasma concentration of nonvolatile weak acids (ATot). As opposed to strong ions, weak acids can exist at physiologic pH as dissociated (A-) or associated with a proton (AH). ATot = A- + AH. Weak acids are often referred to as buffers. In health, ATot is comprised primarily of phosphate and the ion equivalence of the albumin concentration; in the absence of unmeasured ions, A- is equal to the anion gap (AG; see below).

●The arterial carbon dioxide tension (PaCO₂).

Stewart's approach, like the traditional approach, defines disorders resulting from changes in PaCO₂ as respiratory acidosis and alkalosis. However, in contrast to the traditional approach, metabolic disorders are considered to result from changes in SID and ATot rather than from changes in bicarbonate. Thus, in the Stewart approach, metabolic acidosis can be caused by alterations in water or electrolyte balance that diminish the difference between the serum sodium and chloride concentrations (strong ion acidosis), either by increasing the serum chloride concentration by more than the sodium concentration or by decreasing the serum sodium concentration by more than the chloride concentration; metabolic alkalosis (strong ion alkalosis) can be caused by alterations in water or electrolyte balance that widen the difference between the serum sodium and chloride concentrations, either by increasing the serum sodium concentration by more than the chloride concentration or by decreasing the serum chloride concentration by more than the sodium concentration. In the Stewart approach, metabolic acidosis can also be caused by an increase in ATot (nonvolatile buffer ion acidosis), due to hyperalbuminemia or hyperphosphatemia; metabolic alkalosis (nonvolatile buffer ion alkalosis) can be caused by hypoalbuminemia or hypophosphatemia. The traditional approach to these issues is discussed separately in the appropriate topic reviews on metabolic and respiratory acidosis and alkalosis. (See "Simple and mixed acid-base disorders".)

Apparent strong ion difference — Clinically, the SIDa is defined as the difference between the sum of the strong cations (sodium [Na⁺], potassium [K⁺], calcium [Ca²⁺], magnesium [Mg²⁺]) and the sum of the net charge of major strong anions present in health (chloride [Cl^-] and lactate):

Alternatively, SIDa can be estimated using a simplified formulation:

Under normal conditions, the SIDa is approximately 40 mEq/L; processes that increase the strong ion difference increase blood pH, whereas processes that reduce it decrease pH. As an example, the loss of gastric fluid, which has a higher concentration of chloride than cation, increases the SIDa, causing alkalosis; by comparison, infusion of sodium chloride, a solution with an SID of zero, reduces the SIDa and is therefore acidifying. The Stewart approach attributes gastric alkalosis to excessive loss of chloride (and its effect on SID) rather than to loss of hydrogen ion, and it attributes acidosis caused by large volume saline infusions to the excessive gain of chloride rather than to dilution of the serum bicarbonate concentration.

Effective strong ion difference — The "effective SID" (SIDe) is the sum of the bicarbonate concentration ([HCO₃^-]) and the anion equivalency of albumin ([Alb^-]) and inorganic phosphate ([P_i^-]):

The anionic equivalency of albumin is a function of the serum albumin concentration ([Alb]) in g/L and blood pH; it can be estimated by a formula:

The anionic equivalency of phosphate ([P_i^-]) is a function of the serum inorganic phosphate concentration ([P_i]) in mmol/L and blood pH; it can be estimated by a formula:

The bicarbonate concentration is a function of the partial pressure of carbon dioxide (PCO₂) and pH; it can be estimated by a formula:

Strong ion gap — Cation and anion concentrations must be equal. Thus, under normal conditions:

By definition, then, under normal conditions:

If there is an accumulation of unmeasured anions, due to endogenous acids or the intake of exogenous acids or acid precursors (eg, methanol, ethylene glycol, or salicylates), SIDa and SIDe will not be equal; the difference between them is called the strong ion gap (SIG):

Under normal conditions the SIG should be zero. A positive value suggests a possible organic acidosis. Lactic acidosis will increase SIG if SIDa is calculated using its simplified formulation ([Na⁺] + [K⁺] - [Cl^-]).

Given the complexity of the determination of SIDe, preprogrammed calculators are available that permit clinicians to enter the variables that determine the SIG.

STRONG ION VERSUS TRADITIONAL APPROACH — The Stewart or strong ion approach to acid-base disturbances has gained acceptance among many intensivists because of perceived weaknesses of the traditional approach in the evaluation of patients with critical illnesses [12,13]. The major practical difference between the two approaches is the inclusion of the serum albumin and phosphate concentrations in the Stewart approach, which provides some increase in accuracy in certain clinical settings [14].

However, if serum albumin concentration is considered in the traditional analysis of acid-base disturbances, we feel that the Stewart approach offers no advantage over the traditional Schwartz-Bartter approach to acid-base disturbances. In addition, the Stewart approach is much more complex and may be more prone to error, given its reliance upon multiple measurements of several variables. When measurements obtained from automated blood chemistry devices or direct ion-specific electrodes are used to calculate the strong ion gap (SIG), the results can also vary substantially depending upon which device is used [15,16]. We therefore prefer the traditional approach in the analysis of all acid-base disturbances.

Relative diagnostic efficacy — Some evidence has suggested that the Stewart approach provides more accurate analysis of acid-base disturbances than the traditional approach. One study of 255 patients in a pediatric intensive care unit who had simultaneous measurements of arterial blood gases, electrolytes, and serum albumin concentration found that 26 percent of patients had a different interpretation of acid-base balance when the Stewart approach was used compared to when the conventional approach was used [13]. The Stewart method was superior to the traditional method in identifying patients with high lactate levels.

However, this benefit vanishes when hypoalbuminemia, a common problem in hospitalized patients, is accurately considered when using the conventional approach. To best understand this, a brief review of traditional analysis using the anion gap (AG) is required.

In the evaluation of metabolic acidosis, the AG serves to differentiate a hyperchloremic acidosis (normal gap) from high AG acidosis associated with increased concentrations of nonvolatile anions other than chloride [17]. Negative charges on plasma albumin are a large component of the unmeasured anions that contribute to the calculated AG in normal subjects. Thus, the presence of hypoalbuminemia, if not considered, will mask the presence of an organic acidosis when the clinician relies upon the AG for analysis.

Albumin-corrected AG — It is possible to adjust for the effect of low serum levels of albumin [18,19]. Each 1 g/dL decrease or increase in serum albumin below or above 4.4 g/dL respectively lowers or raises the actual concentration of unmeasured anions by approximately 2.3 to 2.5 mEq/L [14]. (See "Approach to the adult with metabolic acidosis".)

The albumin-corrected AG can be calculated from the measured AG and the serum albumin concentration in g/dL:

The formula for the albumin-corrected AG is not precise because it attributes a fixed charge to albumin without considering pH effects on the imidazole groups of albumin, and because it ignores negative charges contributed by the serum phosphate concentration [16].

Accumulation of unmeasured anions due to metabolic acidosis (eg, ketoacidosis, lactic acidosis, or acidosis from toxic alcohols or salicylates) increases the calculated AG above its normal value. In the traditional approach to acid-base disturbances, the difference between the measured AG and its normal value is called the “delta gap,” and the presence of a delta gap defines a high AG acidosis. In the Stewart approach, accumulation of unmeasured anions due to metabolic acidosis results in an increase in the SIG. Thus, the SIG and the albumin-corrected delta gap should be similar.

With appropriate adjustments for hypoalbuminemia, the traditional and Stewart approach are similarly accurate [20,21]. This was shown in a study of 50 consecutive patients with metabolic acidosis who were admitted to a critical care unit [20]. There was no advantage of the Stewart approach over the traditional approach to acid-base disturbances augmented by the albumin-corrected AG. Although the uncorrected AG failed to detect increased blood lactate levels, there was a strong correlation (r2 = 0.932) between the SIG and the albumin-corrected delta gap.

Acid-base analyses based upon the strong ion difference (SID) require multiple measurements of several variables, each subject to error. This may be a particular problem when point of care measurements are used for these measurements because of the larger errors associated with these devices [22].

Prognostic value — There are conflicting data concerning the relative prognostic value of the Stewart versus the traditional approach; however, on average, the AG corrected for albumin and SIG perform equally well in predicting mortality [7]:

●In a review of 255 patients in a pediatric intensive care unit who had simultaneous measurements of arterial blood gases, electrolytes, and serum albumin concentration, 26 percent of patients had a different interpretation of acid-base balance when the Stewart approach was used compared to when the conventional approach was used [13]. The Stewart approach was more strongly associated with mortality than were the AG or plasma lactate level.

●By comparison, a study of 100 consecutive adults admitted to an intensive care unit found that traditional acid-base indices (lactate, AG, and pH) accurately discriminated between survivors and nonsurvivors [23]. The calculation of SIG, the effective strong ion difference (SIDe), and the apparent SID (SIDa) offered no advantage as prognostic markers.

●A study of 78 consecutive adults admitted to an intensive care unit for hemorrhagic shock caused by major trauma found that indicators of unmeasured anions (AG or SIG) were strongly associated with hospital mortality whereas pH and lactate were of limited value as predictors; the source of the unmeasured anions that are generated by critical illness has not been defined [24]. SIG performed slightly better than either the AG or the AG corrected for albumin in predicting mortality [24]. However, statistical comparisons between these measures are difficult because the variables used to calculate them are interconnected [7,25].

●In a larger study of 6878 patients admitted to an intensive care unit, the SIG, SIDa, SIDe, AG, and albumin-corrected delta AG did not differ as predictors of mortality, and all were inferior to the arterial lactate concentration [26].

SUMMARY AND RECOMMENDATIONS

●Most internists use the traditional approach to acid-base analysis, which relies upon the definition of the hydrogen ion concentration in blood as a function of the ratio between the PCO₂ and the serum bicarbonate concentration. (See 'Background' above.)

●The cardinal tenet of the strong ion difference or Stewart approach to acid-base analysis is that the serum bicarbonate concentration does not alter blood pH. Instead, the independent variables responsible for changes in acid-base balance are the strong ion difference (SID), plasma concentration of nonvolatile weak acids or buffers (ATot), and arterial carbon dioxide tension (PaCO₂). (See 'Strong ion difference' above.)

●Because clinically important acid-base derangements are thought to result from changes in PaCO₂, SID, and ATot, the strong ion approach distinguishes six primary acid-base disturbances. These are respiratory acidosis, respiratory alkalosis, strong ion acidosis, strong ion alkalosis, nonvolatile buffer ion acidosis, and nonvolatile buffer ion alkalosis. (See 'Strong ion difference' above.)

●The apparent SID (SIDa) is defined as the difference in net charge between the sum of the strong cations and the sum of the major strong anions present in health.

●The calculated SID value, which is a function of pH and ATot, is called the "effective strong ion difference" or "SIDe." It can be calculated from measured values of pH, serum albumin, and phosphate. (See 'Strong ion gap' above.)

●The difference between SIDa and SIDe is the strong ion gap (SIG), which should be zero. A positive value suggests that an organic acidosis may be present. (See 'Strong ion gap' above.)

●The major practical difference between the Stewart approach and the conventional approach to acid-base disturbances is the inclusion of the serum albumin concentration in the Stewart approach. However, if changes in serum albumin concentration are accounted for in measurement of the anion gap (AG), the more complex Stewart approach does not appear to offer a clinically significant advantage over the traditional Schwartz-Bartter approach to acid-base disturbances. Each 1 g/dL decrease or increase in serum albumin below or above 4.4 g/dL respectively lowers or raises the actual concentration of unmeasured anions by approximately 2.3 to 2.5 mEq/L. (See 'Relative diagnostic efficacy' above.)

Based upon these considerations, we prefer the traditional approach in the analysis of acid-base disturbances.