Can Stewart's Analysis Help Assess Acidosis in Kidney Disease?
ACVIM 2008
Helio Autran de Morais, DVM, PhD, DACVIM (Internal Medicine and Cardiology)
Madison, WI, USA

"All models are wrong, some are useful." George E.R. Box, 1979

Acid-base status is usually well preserved in chronic renal failure until GFR falls to 10 to 20% of normal. Classically, metabolic acidosis in chronic renal failure is considered to be hyperchloremic early in the course of the disease process (Widmer et al. 1979). Chloride initially increases proportionately to the increase in creatinine, but subsequently decreases in advanced renal failure, whereas the anion gap increases throughout the course of renal failure (Hakim et al, 1988). However, this pattern is not universal, and high anion gap was only observed in 20 % of human patients with end-stage renal disease in other studies (Wallia et al, 1986; Ray et al, 1990). In retrospective studies of small animal patients with chronic renal failure, plasma HCO3- concentrations were below 16 mEq/L in 40% of dogs with chronic renal failure caused by amyloidosis (DiBartola et al. 1989) and below 15 mEq/L in 63% of cats with chronic renal failure of various causes (DiBartola et al. 1987). A high anion gap was observed in 43% of affected dogs (>25 mEq/L) and in 19% of affected cats (>35 mEq/L) in these studies. In cats with naturally-occurring renal failure, acidosis was present in 15% of cats with moderate chronic renal failure, and in 50% of cats with severe chronic renal failure (Elliot et al 2003a). In cats with severe chronic renal failure, anion gap was elevated, whereas chloride was decreased. Deterioration of renal function was associated with worsening of acidosis, increase in anion gap, and decrease in chloride concentration (Elliot et al 2003b).

Stewart's Approach to Acid-Base Disorders

The two main goals of acid-base assessment are to identify and quantify the magnitude of an acid-base disturbance and to determine the mechanism for the acid-base disturbance. Peter Stewart attempted to answer the question on why the pH is changing, first in water and then in progressively more complex aqueous solutions (e.g., plasma). In the monograph "How to Understand Acid-Base", he reassessed physicochemical roots, especially electroneutrality, conservation of mass, and dissociation of electrolytes. Stewart's strong mathematical background helped him develop the new system before the era of microcomputers. Stewart's work stirred up a hornet's nest of controversy, mostly because it went against conventional thinking. That the book was heavily mathematical with almost no references did not help matters. Despite the passionate debate it stimulated, Stewart's strong ion approach is not incompatible with the traditional "bicarbonatocentric" approach based in the Henderson-Hasselbalch equation. They are just slightly different ways to look at the same physiological process. The Henderson-Hasselbalch equation can be viewed as a simpler model with all of the advantages and inherent disadvantages: it is more user-friendly, but it does not apply to all scenarios, and it does not provide as much information.

The Stewart's approach defines some variables as independent, variables that influence the system from the outside and cannot be affected by changes within the system or by changes in other independent variables. In contrast, dependent variables are influenced directly and predictably by changes in the independent variables (Stewart 1983). There are three independent variables: PCO2, strong ion difference (SID or the difference in charge between fully dissociated strong cations and anions in plasma), and the total plasma concentration of nonvolatile weak buffers (Atot).

It may be helpful to think of the similarities between Stewart's approach and the "bicarbonatocentric" approach before considering their differences. First, interpretation of respiratory disorders is similar because both systems consider PCO2 an independent variable. Second, in a theoretical plasma solution without proteins or phosphate, bicarbonate concentration will equal the difference between all strong cations and strong anions (i.e., Stewart's SID). In fact, in such solution a strong ion equation to estimate pH could be algebraically simplified to the Henderson-Hasselbalch equation. Thus, the first difference between the two approaches is that the effects of protein and phosphate concentration on pH are only accounted for in the strong ion model. (In other words, the Henderson-Hasselbalch and strong ion approaches are equivalent whenever plasma albumin, phosphate, and globulin concentrations are within their reference ranges). Albumin, globulins and inorganic phosphate are nonvolatile weak acids and collectively are the major contributors to what Stewart called Atot. Changes in Atot independently affect pH and bicarbonate concentration in humans (Wilkes, 1998). An increase in phosphate concentration will lead to acidosis, whereas a decrease in albumin concentration will cause alkalosis. In vitro, a 1g/dL decrease in albumin concentration is associated with an increase in pH of 0.093 in cats (McCullough et al, 2003) and 0.047 in dogs (Constable et al, 2005).

The second difference between the two approaches can be appreciated when Atot and PCO2 are kept constant. In this setting, changes in bicarbonate result only from, and are proportional to, changes in SID. Bicarbonate concentration estimates the severity of the acid-base disorder, but to understand why bicarbonate is changing, we need to look at the SID. A decrease in SID is associated with metabolic acidosis, whereas an increase in SID is associated with metabolic alkalosis. The strong ion approach integrates acid-base and electrolyte disorders because the most important strong ions in plasma are electrolytes (particularly sodium and chloride) and metabolic-generated strong anions such as lactate. There are three general mechanisms by which SID can change: (1) a change in the free water content of plasma; (2) a change in chloride concentration; and (3) an increase in the concentration of other strong anions.

Solely decreasing the content of water increases the plasma concentration of all strong cations and strong anions, and thus increases SID, whereas an increase in free water will decrease SID leading to metabolic acidosis. Changes in free water are recognized by changes in sodium concentration. Thus, hypernatremia is associated with SID acidosis (dilutional acidosis), whereas hyponatremia is associated with SID alkalosis (concentration alkalosis).

If there is no change in the water content of plasma, plasma sodium concentration will be normal. Other strong cations (e.g., magnesium, calcium, potassium) are regulated for purposes other than acid-base balance, and their concentrations never change sufficiently to substantially affect SID. Consequently, when water content is normal, SID changes only as a result of changes in strong anions. If sodium remains constant, decreases in chloride can increase SID (so-called hypochloremic alkalosis), whereas increases in chloride concentration lead to hyperchloremic acidosis. Accumulation of metabolically produced organic anions (e.g., L-lactate, acetoacetate, citrate, β-hydroxybutyrate) or addition of exogenous organic anions (e.g., salicylate, glycolate from ethylene glycol poisoning, formate from methanol poisoning) will cause metabolic acidosis because these strong anions decrease SID (so-called organic acidosis). Addition of organic anions can be identified by the anion gap, strong ion gap or by the base-excess algorithm (de Morais et al., 2006).

Stewart's Approach in Renal Failure

Metabolic acidosis in critically ill human patients with acute renal failure is multifactorial (Rocktaeschel et al., 2003). Compared to other critical care patients matched by APACHE II score, acute renal failure patients had a lower pH with increased strong ion gap (organic acidosis), decreased albumin (hypoalbuminemic alkalosis), and increased phosphate (hyperphosphatemic acidosis). The anions responsible for the increase in strong ion gap where not identified, but lactate was similar in both groups. Potential culprits include sulfate, urate, hydroxypropionate, hippurate, oxalate, and furanpropionate (Niwa, 1996).

In human patients with chronic renal failure (mean creatinine concentration 3.06 mg/dL) compared to normal subjects (Story et al, 2005), there was mild SID acidosis due to a combination of hyponatremia and hyperchloremia. Unmeasured strong ions had a minimal role in the genesis of metabolic acidosis in these patients. There was a significant difference in phosphate concentration, but it was small (mean 0.31 mmol/L). However, hyperphosphatemic acidosis is a contributor for the metabolic acidosis in chronic renal failure (Bellomo et al. 2005).

When 25 dogs with chronic renal failure and metabolic acidosis where compared to a group of healthy dogs, they showed an increase in anion gap, decrease in albumin, and increase and phosphate concentration (Kogika et al, 2006). Unfortunately, data for sodium and chloride were not provided. The mean difference in anion gap was 10 mmol/L. Correction of the anion gap for the decrease in albumin would raise the differences in anion gap between the groups to 13 mmol/L. Increase in phosphate concentration in renal failure patients, accounted for 55% (7.1 of 13 mmol/L) of the increase in the anion gap. This suggests that this population of renal failure dogs had a combination of hypoalbuminemic alkalosis with a more severe hyperphosphatemic acidosis. The anion gap suggests that other unmeasured strong anions are also increased in these chronic renal failure dogs.

In cats with severe chronic renal failure, deterioration of renal function was associated with metabolic acidosis (Elliot et al, 2003b). In those patients, there was a combination of hypochloremic alkalosis (decrease in corrected chloride of 7 mmol/L) and an increase in anion gap of 6.6 mmol/L. Phosphate and albumin concentration were not reported and their contribution to the anion gap and to changes in acid-base balance cannot be estimated. However, the anion gap would have been higher in the absence of hypochloremic alkalosis. Severe hyperphosphatemia (mean of 9 mmol/L) has been previously identified in cats with chronic renal failure prior to undergoing hemodialysis (Langston et al, 1997). The role of hyperphosphatemia in the genesis of metabolic acidosis, may help explain why uremic acidosis is one of the few high anion gap acidosis that respond to bicarbonate therapy (Gauthier et al, 2002). Sodium bicarbonate administered intravenously shifts phosphate inside cells and may be used as adjunctive therapy in patients with hyperphosphatemic acidosis (Barsotti et al, 1986).

The traditional approach for evaluation of acid-base status using pH, Pco2, and HCO3- has several clinically relevant limitations. It does not give a complete assessment of the sources of pathophysiologic changes in the metabolic component ([HCO3-]); it may lead to the conclusion that changes in electrolytes are only secondarily related to acid-base status; and it does not recognize changes in pH caused by changes in protein or inorganic phosphate concentrations. Using the strong ion model, the relationship between electrolytes and acid-base status becomes clear, and it becomes apparent that they should no longer be viewed as separate entities. The end result is a better understanding of how acid-base disorders develop and how they should be treated. It is hoped that improved patient care will follow enhanced understanding of the pathophysiologic principles underlying acid-base disturbances. Clearly, the strong ion approach is more time consuming than are conventional methods, and therefore it is less convenient in daily practice. This argument is particularly true in patients with normal protein and phosphate concentrations, in which the traditional Henderson-Hasselbalch approach in conjunction with an estimation of unmeasured anions works well as a first approximation of a more complex system and is therefore the preferred method.


1.  Barsotti G, et al. Miner Electrolyte Metab 12:103, 1986.

2.  Bellomo R, et al. Int J Artific Organs, 28:957, 2005.

3.  Constable PD, et al. J Vet Intern Med19:507, 2005.

4.  de Morais HAS, et al. Strong ion approach to acid-base disorders. In: DiBartola SP (ed). Fluid, Electrolyte, and Acid-Base Disorders. 3rd ed., Philadelphia, Elsevier, 2006, chap 13, p:310-321.

5.  DiBartola SP, et al. J Am Vet Med Assoc 190:1196, 1987.

6.  DiBartola SP, et al. J Am Vet Med Assoc 195:358, 1989.

7.  Elliot J, et al. J Small Anim Pract 44:261, 2003a.

8.  Elliot J, et al. J Small Anim Pract 44:65, 2003.

9.  Gauthier PM, et al. Crit Care Clin, 18:298, 2002.

10. Hakim RM, et al. Am J Kidney Dis 11:238, 1988.

11. Kogika MM, et al. Vet Clin Pathol 35 :441, 2006.

12. Langsto CE, et al. J Vet Intern Med, 11:348, 1997.

13. McCullough SM, et al. Am J Vet Res 64:1047-1051, 2003.

14. Niwa T. Semin Nephrol 16, 167, 1996.

15. Ray S, et al. Miner Electrolyte Metab, 16:355, 1990.

16. Rocktaeschel J, et al. Critic Care, 7:R60, 2003.

17. Stewart PA. How to understand acid-base. A Quantitative Acid-Base Primer for Biology and Medicine. New York, Elsevier-North Holland. 1981.

18. Story DA, et al. Int J Artific Organs, 28:961, 2005.

19. Wallia R, et al. Am J Kidney Dis 8:98, 1986.

20. Widmer B, et al. Arch Intern Med 139:1099, 1979.

21. Wilkes P. J Appl Physiol 84(5):1740-1748, 1998.

Speaker Information
(click the speaker's name to view other papers and abstracts submitted by this speaker)

Helio de Morais, DVM, PhD, DACVIM (Internal Medicine & Cardiology)
University of Wisconsin-Madison
Madison, WI

MAIN : SVN/U : Stewart’s Analysis
Powered By VIN