When I was a brand-new ICU trainee, around 20 years ago, I was given a photocopied article about Stewart’s approach to acid–base physiology.
It was a long article. I understood some of it, recognised that there was something important in the parts I could not follow, and made it my aim to understand it properly one day.
That article was a tricky place to begin. Acid–base teaching often starts several steps into the explanation, with equations whose ingredients we have never quite understood. So I want to start from the beginning. No chemistry knowledge assumed.
What is an acid? What is a buffer? What does pH actually measure? Why do bicarbonate and carbon dioxide appear together? What is base excess telling us?
By the end of this first episode, we should be able to look at a blood gas and identify the main acid–base processes using the chemistry we have built along the way. Part 2 will take us further: why do those numbers have the values they do, and what can they reveal about the patient? Basic to advanced, step by step.
What an acid actually does
An ion is an atom, or a group of atoms, carrying an electrical charge. Sodium in blood is positively charged and written Na⁺. Chloride is negatively charged and written Cl⁻. Positive ions are called cations; negative ions are anions.
The ion at the centre of acid–base physiology is H⁺, the hydrogen ion. You will also see it called a proton. Molecules can pick up and release H⁺. That exchange is the starting point for understanding acids and bases.
An acid can donate H⁺. A base can accept it.
Imagine a simple acid consisting of a hydrogen attached to the rest of a molecule. We can call the whole molecule HA. When it releases H⁺, the remaining part, A⁻, carries a negative charge:
HA ⇌ H⁺ + A⁻
The double arrow means that the reaction can run in either direction. A⁻ can take the H⁺ back and become HA again. It is therefore a base: specifically, the conjugate base of HA. The terminology sounds more complicated than the idea. It is simply the same chemical pair, with or without the H⁺ attached.
Blood contains many substances that can take up H⁺. Bicarbonate can. So can groups on proteins such as haemoglobin and albumin. That means adding acid to blood does not leave every added H⁺ floating freely in the plasma.
Buffers
Suppose more H⁺ enters a solution containing HA and A⁻. Some A⁻ takes it up and becomes HA. If H⁺ is removed, some HA releases it again. The mixture therefore limits the change in free H⁺. This is what a buffer does.
The reactions do not stop when the system reaches equilibrium. Molecules continue picking up and releasing H⁺ in both directions. At equilibrium, those opposing reactions balance so that the overall proportions remain stable.
This also explains something that can initially seem odd. If an acid splits into H⁺ and its conjugate base:
HA ⇌ H⁺ + A⁻
why does the A⁻ not simply take up all the H⁺ again?
Some of it does. The forward reaction continues too, releasing H⁺ as other molecules take it back up. The system settles at a balance in which some HA remains intact and some H⁺ and A⁻ remain separate.
The amount of free H⁺ at that balance determines the pH.
Buffers also have limits. As more of the available base takes up H⁺, less remains available to buffer the next addition.
And binding a proton does not remove it from the body. Buffers contain the disturbance while other processes deal with the source and disposal of acid.
Some acids release H⁺ almost completely when dissolved in water. These are called strong acids. Hydrochloric acid is an example. Under comparable conditions, weak acids retain a larger proportion of their H⁺.
“Strong” does not mean concentrated. Strength describes how completely an acid dissociates; concentration describes how much is present.
For now, that is enough chemistry.
What pH tells us
pH tells us about the amount of freely available H⁺. Strictly, pH reflects hydrogen-ion activity, which accounts for how ions behave in their surroundings. Clinically, it is reasonable to think of it as reflecting free H⁺ concentration.
The scale runs backwards:
more H⁺ → lower pH
It is also logarithmic. A fall of one whole pH unit represents a tenfold increase in hydrogen-ion activity.
At pH 7.40, the H⁺ concentration is approximately 40 nanomoles per litre. At pH 7.10, it is about 80. So a fall of only 0.30 on the blood-gas report represents roughly a doubling of free H⁺.
That also shows how tiny the free H⁺ concentration is compared with the millimole-per-litre concentrations of electrolytes and buffers. Most H⁺ added to the body does not remain freely dissolved. It is taken up by buffers.
The pH therefore tells us the current concentration of free H⁺. It does not tell us how much acid has entered the body.
For an arterial sample measured at 37°C, the normal reference interval is approximately 7.35–7.45. Below this is acidaemia. Above it is alkalaemia. These terms describe the state of the blood at the moment it was sampled.
An acidosis is a process tending to lower pH. An alkalosis tends to raise it.
A patient can have more than one process at the same time. If an acidosis and an alkalosis oppose each other closely enough, the measured pH may even fall within the normal range.
A normal pH therefore does not necessarily mean normal acid–base physiology.
How carbon dioxide changes pH
Cells continuously produce carbon dioxide. CO₂ does not contain hydrogen, so it cannot simply donate H⁺ itself. Instead, it reacts with water to form carbonic acid:
CO₂ + H₂O ⇌ H₂CO₃
Carbonic acid can then release H⁺:
H₂CO₃ ⇌ H⁺ + HCO₃⁻
This maps directly onto the simple acid reaction we used earlier:
HA ⇌ H⁺ + A⁻
Here, HA = H₂CO₃ and A⁻ = HCO₃⁻.
Bicarbonate is therefore the conjugate base of carbonic acid.
Because carbonic acid exists only in a relatively small amount and exchanges rapidly with dissolved CO₂, we usually hide the carbonic-acid step in the middle and shorten the whole sequence to:
CO₂ + H₂O ⇌ H⁺ + HCO₃⁻
Now suppose CO₂ rises. More CO₂ enters this system, so the balance shifts towards more carbonic acid and then towards more H⁺ and bicarbonate.
That gives us an initially surprising combination:
CO₂ ↑
H⁺ ↑
pH ↓
HCO₃⁻ ↑
If bicarbonate is a base, why does it not simply mop up all the extra H⁺?
Some of it does. The reaction is reversible:
H⁺ + HCO₃⁻ ⇌ H₂CO₃ ⇌ CO₂ + H₂O
But, just as with our simple HA example, the reaction does not run completely in either direction. It settles at a new equilibrium.
With more CO₂ present, that new equilibrium contains more free H⁺ than before, so the pH is lower. It also contains more bicarbonate.
At a pH of 7.40, the concentration of free H⁺ is only about 40 nanomoles/L, or 0.00004 mmol/L. Bicarbonate, by comparison, is present at around 24 mmol/L.
When CO₂ rises, the additional H⁺ generated by the carbonic-acid system does not all remain free. Much is taken up by other buffers, particularly haemoglobin and other proteins, allowing additional bicarbonate to remain.
But buffering does not have to leave zero additional free H⁺. A rise from 40 to 60 nanomoles/L would represent only another 0.00002 mmol/L of free H⁺. That is almost nothing compared with the quantities of bicarbonate and other buffers present, but it is a 50% increase in free H⁺ and therefore produces a substantial fall in pH.
This is why a respiratory acidosis can produce a raised bicarbonate. A high bicarbonate does not automatically mean metabolic alkalosis. Its meaning depends on the CO₂ and the rest of the system.
How the body handles an acid–base disturbance
Buffers act immediately. The lungs and kidneys then alter what the body retains or removes.
Ventilation controls CO₂. Retaining more CO₂ tends to increase free H⁺ and lower pH. Removing more CO₂ tends to reduce free H⁺ and raise pH.
The kidneys adapt more slowly. They reclaim filtered bicarbonate and excrete acid, much of it carried in the urine as ammonium or bound to urinary buffers such as phosphate.
The details are complicated, but the principle is simple: the kidneys alter acid excretion and bicarbonate balance over hours to days.
This gives us several very different ways to disturb acid–base physiology.
A patient may retain CO₂ because ventilation is inadequate. They may generate ketoacids. They may lose bicarbonate through diarrhoea. They may lose gastric acid through vomiting. Renal dysfunction may limit acid excretion. Intravenous fluids may change the chemical composition of the plasma.
The measured pH is the final result of all those processes acting together.
Henderson–Hasselbalch earns its place
We have seen how a rise in CO₂ can increase both free H⁺ and bicarbonate. The next step is to put numbers to that relationship.
We can start with the reaction:
H₂CO₃ ⇌ H⁺ + HCO₃⁻
At a given temperature and comparable solution conditions, chemical equilibrium fixes a relationship between these three substances. Carbonic acid is also linked to dissolved CO₂. We can express the combined relationship as:
[H⁺] ∝ dissolved CO₂ / [HCO₃⁻]
The square brackets mean concentration; ∝ means “is proportional to”. In words:
more CO₂ relative to bicarbonate → more free H⁺
more bicarbonate relative to CO₂ → less free H⁺
Henderson developed the underlying relationship early in the twentieth century. Hasselbalch subsequently expressed it using the logarithmic pH scale.
For blood at approximately 37°C:
pH ≈ 6.1 + log₁₀[HCO₃⁻ ÷ (0.225 × PCO₂ in kPa)]
or, using mmHg:
pH ≈ 6.1 + log₁₀[HCO₃⁻ ÷ (0.03 × PCO₂ in mmHg)]
Here bicarbonate is in mmol/L, numerically the same as mEq/L for this ion. PCO₂ is the partial pressure of CO₂; the factor 0.225 or 0.03 converts it to an approximate dissolved concentration in the corresponding pressure units.
The ratio does not replace H⁺ as the thing determining pH. It describes the free H⁺ concentration when the system reaches equilibrium.
You do not need to calculate the equation routinely at the bedside. Its value is in understanding what the numbers mean.
Take a bicarbonate of 24 mmol/L and a PCO₂ of about 5.3 kPa (40 mmHg). Dissolved CO₂ is approximately 1.2 mmol/L. The bicarbonate-to-CO₂ ratio is about 20:1, giving a pH close to 7.40.
The two parts of the ratio are chemically linked, but they are not forced to change in the same proportion.
Suppose PCO₂ acutely doubles to about 10.7 kPa (80 mmHg). The amount of dissolved CO₂ approximately doubles. Bicarbonate rises too, but much less, typically only by a few mmol/L in an acute respiratory acidosis.
For example:
HCO₃⁻ 24 mmol/L / dissolved CO₂ 1.2 mmol/L ≈ 20
might become approximately:
HCO₃⁻ 28 mmol/L / dissolved CO₂ 2.4 mmol/L ≈ 12
The ratio has fallen. That lower ratio corresponds to a higher free H⁺ concentration and therefore a lower pH.
Now consider a different change: halve both of our starting values.
Bicarbonate is 12 mmol/L and PCO₂ is about 2.7 kPa (20 mmHg). The ratio remains about 20:1, so pH is still about 7.40. But that gas is not normal. We will come back to it.
Henderson–Hasselbalch is fully quantitative: given bicarbonate and PCO₂, it lets us calculate the corresponding pH. What it does not establish, by itself, is why bicarbonate has the value it does.
Stewart’s framework approaches the same chemistry using different inputs. With consistent assumptions, the two approaches give compatible answers. We will explore what that different view adds in Part 2.
What the blood-gas machine actually measures
The blood-gas analyser directly measures pH and PCO₂. Bicarbonate is usually calculated from those measurements using the known relationship between them. Base excess is calculated as well.
That means pH, PCO₂, bicarbonate and base excess are not four independent measurements confirming the same conclusion. Some are derived from others.
The chemistry remains useful. We just need to understand what each number represents.
Why I like base excess
Base excess tries to isolate the metabolic component of the disturbance.
Imagine taking a blood sample and bringing its PCO₂ to the standard value of 5.33 kPa, or 40 mmHg, at 37°C.
How much strong acid or base per litre would need to be added to bring the pH to 7.40?
That amount defines the base excess for the blood sample.
If acid would be required, the base excess is positive. If base would be required, it is negative.
The version most useful clinically is standard base excess, or SBE. It estimates the metabolic disturbance across the whole extracellular fluid (ECF) volume, using a haemoglobin concentration of 50 g/L (5 g/dL)—approximately what we would get if the blood’s haemoglobin were distributed throughout that volume.
So:
SBE +6 mmol/L (+6 mEq/L) suggests a net metabolic alkalinising effect.
SBE −8 mmol/L (−8 mEq/L) suggests a net metabolic acidifying effect.
Zero is the reference point, and that is the main reason I find BE easier to work with than bicarbonate. Both have normal ranges, but BE already expresses the metabolic result as a positive or negative deviation from zero. There is no need to subtract a reference value of around 24 mmol/L (24 mEq/L) before starting the bedside arithmetic. Positive and negative contributions can be placed alongside one another and added. We will use that advantage in Part 2.
SBE is a net result. A strongly acidifying process and a strongly alkalinising process may partly cancel one another, so a nearly normal SBE does not necessarily mean nothing is happening. An abnormal SBE can also reflect renal compensation for a sustained respiratory disturbance, rather than a separate primary metabolic disorder.
And a base deficit is not automatically a prescription for bicarbonate. We still need to understand what produced it.
The first reading of a blood gas
Before interpreting the numbers, establish the context. Is the sample arterial or venous, and when was it taken? What ventilation, oxygen therapy and treatment were being given? If the results do not fit the patient, ask how the sample was taken and whether it needs repeating.
Then read pH, PCO₂ and the metabolic component together. Traditionally, acid–base disturbances are divided into respiratory and metabolic processes. A primary change in PCO₂ is respiratory. Bicarbonate cannot be interpreted on its own: it also rises during respiratory acidosis, with further renal adaptation over time.
The four basic patterns are:
Respiratory acidosis: PCO₂ rises. Bicarbonate rises as part of the buffering response, with a further increase as the kidneys compensate over time.
Respiratory alkalosis: PCO₂ falls. Bicarbonate falls as part of the buffering response, with a further decrease as the kidneys compensate over time.
Metabolic acidosis: Bicarbonate and SBE fall. The expected respiratory compensation is increased ventilation, lowering PCO₂.
Metabolic alkalosis: Bicarbonate and SBE rise. The expected respiratory compensation is reduced ventilation, raising PCO₂.
Compensation means that the body responds in a direction that tends to return pH towards normal. For example, metabolic acidosis stimulates ventilation. CO₂ falls.
That fall in CO₂ is expected. It does not automatically mean that the patient has a second respiratory disorder. The question is whether the response is of the expected size.
Is the compensation appropriate?
Empirical compensation rules tell us roughly what happens when there is a single primary disturbance. For metabolic acidosis, the best-known is Winter’s formula.
Using bicarbonate in mmol/L:
Expected PCO₂ in kPa ≈ 0.2 × bicarbonate + 1.07, ±0.27 kPa
In mmHg:
Expected PCO₂ ≈ 1.5 × bicarbonate + 8, ±2 mmHg
These are clinical approximations, not laws of physics.
If the measured PCO₂ is higher than expected, there is likely to be an additional respiratory acidosis. If it is lower, there is likely to be an additional respiratory alkalosis.
Now return to our earlier example:
bicarbonate 12 mmol/L
PCO₂ about 2.7 kPa, or 20 mmHg
pH about 7.40
Winter’s formula predicts a PCO₂ of about 3.5 kPa, or 26 mmHg, with an expected range of roughly 3.2–3.7 kPa (24–28 mmHg).
The actual PCO₂ is considerably lower. This patient has metabolic acidosis with an additional respiratory alkalosis. The two processes oppose one another closely enough to leave the pH near 7.40.
Calling the blood gas normal would miss both processes. Calling it simply a compensated metabolic acidosis would miss the additional respiratory disorder.
What we can now say — and what we can’t
We can now read pH, CO₂ and the metabolic component together, assess compensation and recognise important disturbances that a normal pH can conceal. Standard base excess gives us a useful estimate of the net metabolic result.
But suppose the SBE is −10 mmol/L (−10 mEq/L). What accounts for that −10? One process may dominate, or a larger acidifying effect may be partly concealed by an opposing alkalinising one.
And why can a chloride-rich fluid produce acidosis when chloride contains no H⁺?
In Part 2, for paid subscribers, we will follow what happens when the composition of blood changes, and see how Stewart’s framework helps us understand the result. Along the way, we will examine two explanations common in the literature: that sodium bicarbonate works because of its sodium, while bicarbonate itself is irrelevant; and that changes in strong ions cause acidosis by making water dissociate and release H⁺. But do they explain what actually happens in the blood?
We will take those explanations apart carefully, then bring the chemistry to the bedside. Using the anion gap and Story’s simple arithmetic, we will investigate the metabolic result, uncover effects that partly cancel one another, and decide what needs measuring next.