The Acidity of Normal Saline and the Stewart Approach
Jon-Emile S. Kenny [@heart_lung]
How might you answer this oral board examination question? Describe the systemic effect on pH following 2 litres infusion of each of the following solutions:
A. normal saline with a pH of 7.0
B. normal saline with a pH of 5.6
C. normal saline with a pH of 4.6.
Pick up a bag of normal saline [NS] and look at the reported pH [or half-normal saline for that matter]. Yes, normal saline is acidic, but firstly, how can pure water with admixed sodium cations and chloride anions have anything but a pH of 7.0? Secondly, how can there be a range of pH? Thirdly, should this make us pause when infusing it into a patient?
What’s in normal saline?
First and foremost, saline contains water, which – when pure – has a pH of 7.0 at 25 degrees centigrade. When water is exposed to atmosphere, however, there is a pressure gradient for atmospheric carbon dioxide [CO2] to dissolve into solution. Using Henry’s Law [and assuming that atmospheric carbon dioxide is 0.036%], the pH of water exposed to the atmosphere turns out to be 5.7.
But NS also contains sodium and chloride in equal amounts. Do these ion species alter the pH at all? In theory, yes. Charged ions in water alter the physiochemical properties of pure H2O such that carbon dioxide is less soluble, but also increases the dissociation of carbonic acid. There are a number of other plausible effects including the interesting ‘Grotthuss’ mechanism [describing the passing of protons between water molecules like a pail of water passed between firefighters]. Nevertheless, in totality these ion-specific processes are estimated to lower pH by only 0.01.
Finally, and most-surprisingly, is the effect of the bag containing NS. The plastic pouch containing NS is typically composed of polyvinylchloride [PVC] which may lower the pH of NS via a multitude of means including the liberation of: DEHP, formic and acetic acid or even hypochloric acid! Saline prepared in PVC has a median pH of 4.6 compared to saline packaged in polypropylene with a pH of 5.7.
Thus, the three fluids at the outset of this discussion are all plausible and include ‘virgin’ saline unexposed to the atmosphere, saline packaged in polypropylene and saline in PVC, respectively. The question which flows naturally then is: will solution C result in a more acidemic patient than infusion of solution A? Simply, the answer is no. Each solution above will have an identical effect on a patient’s systemic pH; but why?
The Stewart Approach to Acid-Base
Well-over 3 decades ago, Canadian physical chemist Peter Stewart proposed a quantitative approach to acid-base in the context of human biology. His theory posits that the concentration of weak ions [e.g. H+ and HCO3-] are dependent variables and determined by solving a system of equations that includes the equilibrium constants of all involved chemical species, the laws of electroneutrality and conservation of mass. There are 3 independent variables, he argues, which define the physiochemical boundaries of the weak ions.
Firstly, strong ions [e.g. Na+, K+, Cl-] completely dissociate and define the limits of electroneutrality, within which the weak ions fall. Thus the difference between the concentration of strong cations [e.g. Na+] and strong anions [e.g. Cl-] within the plasma fix the strong ion difference – SID. Consequently, the SID is an independent variable affecting the concentration of the weak ions. The other two independent variables which define the concentration of weak ions are the PaCO2 and the total concentration of other weak acid ions [Atot].
As [Atot] is determined primarily by albumin in plasma, and PaCO2 is constant under normal ventilatory conditions, the degree of the strong ion difference [SID] determines the relative concentrations of the weak ions [e.g. H+ and HCO3-] at equilibrium. The SID in plasma is reduced to [Na+] less [Cl-], and is typically 42 mEq/L. As the SID increases, [H+] – and all other positively charged weak cations – will decrease while [HCO3-] – and all other negatively charged weak anions – will increase by mass action, simply to maintain electroneutrality. By this reasoning, metabolic acid-base disturbances may be explained simply by changes in the SID, [Atot] or both. The following rules can be extracted from Stewart’s equations with respect to metabolic acid-base disturbances:
Increased [Atot] favours metabolic acidosis while diminished [Atot] favours metabolic alkalosis. By contrast, an increase in SID promotes metabolic alkalosis while a diminished SID results in metabolic acidosis. While no crystalloid contains Atot, they do have varying degrees of SID which nicely predicts their effects on systemic acid-base.
Infusion of a crystalloid will have competing acid-base effects per the Stewart model. Firstly, as above, because the [Atot] of a crystalloid is zero, [Atot] will fall following crystalloid administration – resulting in a metabolic alkalosis. Because SID of plasma is normally 42 mEq/L, the degree to which the infusate SID is below 42 mEq/L will determine how low the serum SID falls; again, a diminished SID forces the [H+] to rise and [HCO3-] to fall by mass action. Experimentally, a ‘balanced’ crystalloid SID is 24 mEq/L. Therefore, a SID of 24 mEq/L will shrink serum SID below 42 mEq/L, driving up the [H+]. This inverse relationship between SID and [H+] then offsets the infusion-induced fall in [Atot]. Balanced crystalloids contain weak anion ‘surrogates’ such as L-lactate, gluconate, citrate and acetate as these species are metabolized away following infusion, which is why their SID remains positive. This can be thought of as infusing 'only' [Na+], because its anion partner [e.g. lactate] is removed by the liver. The positive SID of 'balanced' crystalloids cushions the fall of the serum SID following infusion; the less the serum SID falls, the less [H+] will be elaborated. This is why 'balanced' solutions mitigate the acidemia associated with zero-SID solutions [e.g. NS] as described below.
By reasoning somewhat opposite to ‘balanced’ crystalloids above, the metabolic acidosis caused by KCl infusion may also be explained using the SID. In contrast to administering sodium lactate which raises the SID as lactate disappears from the extra-cellular compartment, infusion of KCl causes the SID to fall as [K+] disappears into the intra-cellular compartment. Often, in diuretic-associated hypokalemia there is a concomitant increase in SID and, therefore, metabolic alkalosis. Provision of IV potassium chloride is similar to infusing ‘only’ [Cl-] as potassium is predominantly an intra-cellular species. By adding ‘only’ chloride to the extra-cellular compartment, the SID falls and [H+] rises.
The SID of NS is zero; in fact, the SID of solutions A, B and C listed at the outset of this post is also zero. Interestingly, the SID of half-normal saline and D5W – also zero. Thus, despite widely differing in vitro pH, the in vivo effects on pH are fully explained by how the strong ion difference and [Atot] of the plasma are altered. The administration of a zero SID fluid will invariably pull the plasma SID of 42 mEq/L downwards and raise [H+]. Fluids A-C will have an identical effect on systemic pH.
The Stewart Model is not without its critics and the in vivo effects of crystalloid can be explained via more traditional models of buffering. Regardless, the important point is that whether thought of in terms of change in SID, or buffer-base dilution, in vitro crystalloid pH is - arguably - clinically meaningless.
*For an update on acid-base, please see this post. For more recent clinical data on balanced crystalloids, please see a review of the SMART trial here.