Metabolic alkalosis and mixed acid–base disturbance in anaesthesia and critical care

Learning objectives.

By reading this article you should be able to:

• •Explain metabolic alkalosis using a traditional (bicarbonate) and physiochemical (Stewart) approach to acid–base disturbances.
• •Describe the main causes of metabolic alkalosis encountered in anaesthesia and ICU practice and discuss their treatment.
• •Identify the constituents of mixed acid–base disturbances.
• •Describe the acid–base effects of commonly used i.v. fluids.

Key points.

• •Metabolic alkalosis is very common in the critically ill and usually occurs in the context of a mixed acid-base disturbance.
• •Alkalosis refers to a process that reduces the plasma hydrogen ion concentration, whereas alkalaemia is a state of increased pH (>7.45).
• •Hypoalbuminaemia is very common in critically ill patients and results in a metabolic alkalosis. Hypoalbuminaemia can mask a metabolic acidosis arising from other causes.
• •With a traditional approach, metabolic alkalosis is caused by bicarbonate excess or hydrogen ion deficit. With a physiochemical approach, metabolic alkalosis is caused by a relative deficit of chloride, a relative excess of sodium or a reduction in the total weak acids in plasma.
• •Mixed acid–base disturbances can be evaluated using only a few measured variables (pH; Paco₂; sodium, chloride and albumin concentrations) and some simple calculated variables (apparent strong ion difference; strong ion gap; expected Paco₂).

Metabolic alkalosis is very common in critically ill patients and usually occurs in the context of a mixed acid-base disturbance after the first few days of admission to the ICU.^1,2 Thus, part of the challenge in understanding metabolic alkalosis is to understand mixed acid–base disturbances.

Alkalosis refers to a process that reduces the plasma hydrogen ion concentration, whereas alkalaemia is a state of increased pH. A primary acid–base disturbance refers to an underlying process. For instance, vomiting, with attendant loss of chloride and hydrogen ions, causes a primary metabolic alkalosis. Primary acid–base disturbances are associated with compensatory responses that tend to return pH towards normal. For instance, alveolar hypoventilation is a compensatory response to metabolic alkalosis. Mixed acid–base disturbance is defined by the presence of two or more primary disturbances. Mixed acid–base disturbances can drive pH in the same direction (e.g. metabolic alkalosis and respiratory alkalosis) or in opposite directions (e.g. metabolic alkalosis and metabolic acidosis). Consequently, with mixed disturbances it is possible to have metabolic alkalosis in the presence of acidaemia or alkalaemia.

This article is part of a series on acid–base disturbances in BJA Education and should be read in conjunction with ‘Acid–base balance: a review of normal physiology’ and ‘Acid–base quantification: a review of developing technology’.^3,4 Inevitably there is some overlap and differing emphasis between the material presented here and other articles in the series.

In this review, we provide an overview of metabolic alkalosis with particular reference to the causes and mechanisms encountered in critical care and anaesthesia. We discuss metabolic alkalosis using both traditional (bicarbonate) and physiochemical (Stewart) approaches; compensatory mechanisms and mixed disturbances; and provide a framework for diagnosing mixed disturbances. We use the term ‘unmeasured strong anions’ when referring to abnormally high concentrations of organic acids, such as lactate, ketones (e.g. β-hydroxybutyrate), and the metabolites of organic acids that accumulate in renal failure (e.g. sulphates). Although the term ‘unmeasured strong anions’ remains in common usage, several of these anions are now measured. Lactate is commonly reported as part of routine blood gas analysis and ketones are available as a point-of-care test.

Traditional vs physiochemical approaches to acid–base disturbance

Comparisons of the traditional and physiochemical approaches to acid–base disturbance are the subject of previous articles in this journal.^3,5

With the traditional approach, changes in hydrogen ion concentration are considered in relation to changes in bicarbonate concentration via the carbonic acid equilibrium:

$${CO}_2+H_{2}O\underset{\iff }{\text{carbonic}\text{anydrase}}{H}_2{CO}_3\iff H^{+}+{HCO}_3^{-}$$ (1)

The relationship between pH and bicarbonate is quantified by the Henderson–Hasselbalch equation, which may be written as:

$$\text{pH}={\text{pK}}_a+{\log }_{10}\frac{\left[{\text{HCO}}_3^{-}\right]}{\alpha \left[\mathit{P}{\text{CO}}_2\right]}$$ (2)

where pK_a is the negative logarithm of the carbonic acid dissociation constant (6.1) and α is the carbon dioxide solubility constant (0.03 for Paco₂ measured in mmHg). We can see from Eq (1) that changes in bicarbonate concentration occur in response to changes in Paco₂ and hydrogen ion concentration and, therefore, bicarbonate is not an independent determinant of pH.

In the late 1970s, Peter Stewart developed an alternative approach to acid–base quantification.⁶ Under Stewart's framework, the pH of any aqueous solution is determined by the dissociation of water into hydrogen and hydroxyl ions:

$$H_{2}O\iff {[H}^{+}]+[{OH}^{-}]$$ (3)

Based on physiochemical laws, Stewart proposed six equations describing the variables influencing plasma pH.³ Four of the six equations are presented here [Eqs (1), (3), (4), (12)]. Solving the equations analytically requires a calculator or a computer. Stewart showed that only three variables independently influence pH: (i) the Paco₂, (ii) the total weak acids in plasma (A_TOT) and (iii) the strong ion difference (SID). Under the Stewart model, metabolic alkalosis can only occur because of decreased A_TOT or increased SID.

A_TOT

Unlike strong acids, which are fully dissociated at physiological pH, weak acids exist in equilibrium:

$$\left[HA\right]\iff {[\mathrm{A}}^{-}]+{[\mathrm{H}}^{+}]$$ (4)

A_TOT is the total concentration of the non-volatile (i.e. non-CO₂) weak acids in plasma. That is to say:

$$A_{TOT}=\left[HA\right]+{[A}^{-}]$$ (5)

A_TOT comprises the plasma proteins, mainly albumin, with a small contribution from phosphate. Unmeasured strong anions all have a pK_a <4 and are therefore essentially fully dissociated at physiological pH and do not contribute to A_TOT.

Strong ion difference

The SID is the difference between the plasma concentration of all fully dissociated cations (e.g. sodium, potassium, magnesium, calcium) and anions (e.g. chloride, lactate, ketones, sulphate). If present, unmeasured strong anions contribute to (i.e. reduce) the SID. The SID comprises the plasma concentration of bicarbonate and the [A⁻] component of A_TOT.

In clinical practice, the SID may be estimated as the effective or the apparent SID.⁷ The effective SID (SID_e) is the sum of the bicarbonate concentration and the concentration of dissociated proteins:

$${SID}_e={[HCO}_3^{-}]+[A^{-}]$$ (6)

The apparent SID (SID_a) is the difference between the concentration of measured cations and anions. Changes in the plasma concentration of potassium, magnesium and calcium are typically small and are usually insufficient to influence the SID_a. Therefore, the SID_a may be simplified as the difference between the dominant cation (sodium) and the dominant anion (chloride):

$${SID}_a={[Na}^{+}]-{[Cl}^{-}]$$ (7)

Using this simplification, SID_a is normally ∼35 mEq L⁻¹. If present, unmeasured strong anions decrease the SID but this effect is not captured in the SID_a as defined in Eq (7).

Throughout the remainder of the article we use the abbreviation ‘SID’ for the strong ion difference unless specifically referring to the simplifications SID_a or SID_e.

Anion gap

The anion gap is the difference between the major cations and anions in plasma and is commonly calculated as:

$$\mathrm{A}\mathrm{G}={[Na}^{+}]+{[K}^{+}]-{[Cl}^{-}]-[{HCO}_3^{-}]$$ (8)

When calculated in this way, the normal range for the anion gap is 5–10 mEq L⁻¹. In contrast to the SID, both strong and weak anions contribute to the anion gap. In health, the anion gap comprises the [A⁻] component of A_TOT. The presence of unmeasured strong anions increases the anion gap and hypoalbuminaemia (low A_TOT) reduces the anion gap. Thus, if we wish to use the anion gap to quantify the presence of unmeasured strong anions we must account for the albumin concentration. The expected anion gap (i.e. in the absence of unmeasured strong anions) is roughly equal to the albumin concentration (in g L⁻¹) divided by 4.⁸ Eqs (7), (8) are affected by increased concentrations of lactate and ketones, which as noted above, are no longer ‘unmeasured’.

The strong ion gap

Any difference between SID_a and SID_e quantifies the presence of unmeasured strong anions, and is termed the strong ion gap (SIG)⁷:

$$SIG={SID}_a-{SID}_e$$ (9)

The SIG should be approximately zero (up to plus or minus 2). An SIG <−2 is very unusual and is caused by laboratory or calculation error, or the presence of unmeasured strong cations. Formulae exist for calculating the SIG based on Eq (9).^7,9 However, these formulae are rather cumbersome for routine clinical use. A simpler approach is to assume that the SIG is equivalent to the difference between the measured and expected anion gap:

$$\text{SIG}=\mathrm{A}\mathrm{G}-\frac{\left[\text{albumin}\right]}{4}$$ (10)

Eq (10) is a simple way of quantifying the presence of unmeasured strong anions, as we discuss below in the section on mixed acid–base disturbance.

In the next two sections, we consider the causes of metabolic alkalosis using traditional and physiochemical approaches (Table 1).

Table 1.

Causes of metabolic alkalosis using traditional and physiochemical approaches. RRT, renal replacement therapy.

Traditional (bicarbonate)	Physiochemical (Stewart)
Exogenous bicarbonate	Increased strong ion difference
Infusion of sodium bicarbonate solution Infusion of bicarbonate precursors: •Sodium acetate (Plasma-Lyte 148 i.v. fluid) •Sodium gluconate (Plasma-Lyte 148 i.v. fluid) •Sodium lactate (Hartmann's i.v. fluid) •Sodium citrate (citrate RRT fluid)	Chloride depletion relative to sodium: •Loop and thiazide diuretics •Vomiting and nasogastric fluid losses •Hypokalaemia •Compensation for hypercapnia Sodium excess relative to chloride: •Infusion of sodium bicarbonate or bicarbonate precursors •Activation of RAAS •Mineralocorticoid (aldosterone) excess
Endogenous bicarbonate	Reduction in ATOT
Diuretics: •Thiazides •Loop (furosemide) Compensation for hypercapnia	Hypoproteinaemia Hypoalbuminaemia
Hydrogen ion loss
Vomiting Nasogastric tube losses

Traditional approach to metabolic alkalosis

Under the traditional model, metabolic alkalosis arises from bicarbonate gain or hydrogen ion loss. Bicarbonate gain may be endogenous or exogenous. Exogenous gain arises from the use of bicarbonate solutions or solutions containing bicarbonate precursors (e.g. sodium citrate, sodium lactate, sodium acetate). Endogenous bicarbonate gain occurs with diuretics. Diuretics (thiazides, furosemide) increase sodium delivery to the distal nephron by inhibiting sodium and chloride reabsorption in the ascending limb of the loop of Henle. Enhanced sodium delivery to the distal nephron leads to increased bicarbonate generation within the cells of the collecting duct. Potassium and hydrogen ions are secreted into the tubular lumen and sodium and bicarbonate are reabsorbed.¹⁰ The net effect is that sodium and bicarbonate are retained and potassium and hydrogen ions are lost in the urine. Increased plasma bicarbonate concentration shifts the carbonic acid equilibrium [Eq (1)] to the left, reducing the hydrogen ion concentration and causing alkalaemia.

Direct hydrogen ion loss can occur from the gastrointestinal tract as a result of vomiting or nasogastric fluid losses.

Compensatory responses

Alkalaemia results in compensatory responses by the lungs and kidneys. Alkalaemia is detected by chemoreceptors and results in alveolar hypoventilation, with increased Paco₂. Carbon dioxide is soluble in plasma. Thus, any increase in Paco₂ including that caused by compensatory hypoventilation can cause intracellular acidosis. Increased Paco₂ shifts the carbonic acid equilibrium to the right [Eq (1)], returning pH towards normal. Respiratory compensation is rapid, occurring over minutes to hours. The expected compensatory response to metabolic alkalosis is given in the section below on mixed acid–base disturbances. Respiratory compensation is not possible in anaesthetised patients whose lungs are mechanically ventilated and may be impaired in patients with pulmonary disease.

In the kidneys, alkalaemia causes the upregulation of the bicarbonate–chloride exchanger pendrin in the collecting ducts of the distal nephron, leading to increased bicarbonate excretion and chloride reabsorption.^11,12 Pendrin is downregulated in acidaemia leading to increased bicarbonate reabsorption and chloride loss. In addition, acidaemia leads to upregulation of the sodium–hydrogen ion exchanger NHE3 in the proximal tubule, increasing sodium and bicarbonate reabsorption and hydrogen ion excretion. NHE3 is inhibited by acetazolamide, increasing sodium and bicarbonate excretion. The renal response to alkalaemia (and acidaemia) occurs over 3–5 days unless there is impaired renal function or reduced renal bicarbonate excretion secondary to volume contraction and electrolyte deficits.¹³

The volume contraction and electrolyte disturbance associated with diuretic use or gastrointestinal fluid loss leads to activation of the renin–angiotensin–aldosterone system (RAAS), which in turn leads to the retention of sodium and bicarbonate and the loss of potassium and hydrogen ions by the kidneys.¹⁴ Activation of the RAAS is the key mechanism acting to maintain metabolic alkalosis originating by other mechanisms. Conn's syndrome (mineralocorticoid excess) and Cushing's syndrome (glucocorticoid excess) have a similar effect to the physiologic activation of the RAAS and can also lead to metabolic alkalosis.

Physiochemical approach to metabolic alkalosis

Metabolic alkalosis caused by increased SID

In the physiochemical approach, the concentration difference between the dominant cation (sodium) and dominant anion (chloride) is the primary determinant of the plasma hydrogen ion concentration in metabolic alkalosis. Unlike the bicarbonate concentration, which changes in response to pH and Paco₂, changes in the sodium and chloride concentration reflect gain or loss of those ions. Loss of chloride (relative to sodium) or gain of sodium (relative to chloride) increases the SID leading to alkalosis. To see how this works, consider a solution containing only sodium, chloride, and a small amount of dissociated water. By the law of electrical neutrality:

$${[\text{Na}}^{+}]+{[\mathrm{H}}^{+}]={[\text{Cl}}^{-}]+{[\text{OH}}^{-}]$$ (11)

If the sodium concentration increases and chloride remains the same, then the hydrogen ion concentration must decrease via the water dissociation equilibrium [Eq (4)] to maintain electrical neutrality.⁵ The same is true if the chloride concentration decreases relative to sodium. In mammalian plasma, the law of electrical neutrality is more complicated³:

$$SID+\left[H^{+}\right]-\left[{HCO}_3^{-}\right]-\left[{CO}_3^{2-}\right]-{[A}^{-}]-{[OH}^{-}]=0$$ (12)

However, the same principle applies: a relative change in the sodium or chloride concentration influences the dissociation of water and thereby determines pH.

Metabolic alkalosis caused by decreased A_TOT

Decreased A_TOT leads to reduced concentrations of the components of Eq (4) (i.e. [HA], [A⁻], [H⁺]), thereby causing alkalosis. Bicarbonate concentration increases [Eq (1)] to maintain electrical neutrality. The renal response to hypoalbuminaemia is to retain chloride, thereby reducing the SID and restoring hydrogen ion concentration back towards normal. Thus, the SID is typically normal or slightly reduced and the anion gap is low.¹⁵ The chloride concentration increases by 1.4 mmol L⁻¹ for every 10 g L⁻¹ reduction in total protein concentration.¹⁶

Hypoalbuminaemia is common in critically ill patients and can mask coexisting metabolic acidosis. In this circumstance, the anion gap may be low, normal or increased depending on the concentration of unmeasured strong anions.

Relative chloride deficit

Chloride deficit relative to sodium is characteristic of metabolic alkalosis associated with diuretics, vomiting and nasogastric tube fluid losses. Relative chloride deficit is also a feature of the compensatory response to chronic hypercapnia.

Thiazide diuretics inhibit the NCC (sodium–chloride cotransporter) channel in the distal convoluted tubule, resulting in loss of volume with equal amounts of sodium and chloride, which increases the SID of plasma attributable to a larger proportional reduction in chloride concentration than sodium concentration, thereby causing metabolic alkalosis. This effect is the exact opposite of the acidifying effect associated with infusions of 0.9% saline or any other solution with an SID less than the plasma bicarbonate concentration. This concept is explained in more detail below.

Chloride loss relative to sodium is greater with loop diuretics (e.g. furosemide) than thiazide diuretics. Loop diuretics inhibit the NKCC2 channel in the thick ascending loop of Henle. The NKCC2 channel co-transports one sodium, one potassium and two chloride ions into the tubular lumen. Thus, inhibition of NKCC2 leads to volume loss that is chloride-rich relative to sodium. Although the fluid lost has an SID of zero, the plasma potassium deficit is replaced by intracellular potassium to help maintain potassium within its normal range. Therefore, the fluid ‘lost’ from the plasma contains more chloride than sodium, increasing the SID.

In chronic respiratory acidosis, downregulation of pendrin increases chloride loss, increasing the plasma SID to compensate for the acidosis.¹⁷ The metabolic alkalosis seen after normalisation of Paco₂ (e.g. with initiation of mechanical ventilation) can lead to severe alkalaemia, which can take the kidney 3–5 days to resolve. Alkalaemia resolves more rapidly by administering 0.9% saline (adding relatively more chloride) or acetazolamide (removing relatively more sodium). However, in a clinical trial of mechanically ventilated patients with hypercapnic respiratory failure, acetazolamide was not associated with a reduced length of mechanical ventilation.¹⁸

Relative sodium excess

Sodium excess relative to chloride usually results from infusion of sodium-rich, chloride-poor i.v. fluids such as sodium bicarbonate solution or—to a lesser extent—large volumes of balanced salt solutions (e.g. Plasma-Lyte 148).

After i.v. infusion of sodium bicarbonate, bicarbonate shifts the equilibrium of Eq (1) to the left, increasing carbon dioxide production that is expired by the lungs and leaving sodium to exert its influence on the SID. If carbon dioxide cannot be expired, then sodium bicarbonate has an acidifying effect equivalent to any solution with an SID of zero. The same effect occurs with bicarbonate precursors such as sodium citrate, which is used in renal replacement therapy. Under normal circumstances, citrate is metabolised to bicarbonate and the sodium exerts its effect on the SID. The remaining sodium reduces the dissociation of water to maintain electrical neutrality (see equation [11]), causing alkalosis.

However, if citrate cannot be metabolised to bicarbonate because of liver failure, then the fluids given have an SID of zero which will reduce the SID of plasma, causing acidosis. In this circumstance, citrate functions as an unmeasured strong anion.

Activation of the RAAS (e.g. as a result of intravascular volume depletion) or excess mineralocorticoid activity for other reasons (Conn's syndrome, Cushing's syndrome) leads to enhanced sodium reabsorption relative to chloride, increasing the SID and resulting in metabolic alkalosis.

The effects of metabolic alkalosis (and metabolic acidosis) on the SID, SID_a, anion gap and SIG are summarised in Table 2.

Table 2.

The effects of metabolic alkalosis and metabolic acidosis on the strong ion difference (SID), apparent SID (SID_a), anion gap and strong ion gap (SIG).

Variable	SID	SIDa Eq (7)	Anion gap Eq (8)	SIG Eq (10)
Metabolic alkalosis
Relative chloride deficit or relative sodium excess	Increased	Increased	Normal	Approx. zero
Reduced ATOT	Normal or slightly decreased	Normal or slightly decreased	Decreased	Approx. zero
Metabolic acidosis
Relative chloride excess or relative sodium deficit	Decreased	Decreased	Normal	Approx. zero
Unmeasured strong anions	Decreased	Normal	Increased	Positive

Consequences of metabolic alkalosis

Alkalaemia is associated with increased mortality, which is likely related to the underlying disease severity rather than the metabolic alkalosis per se. In one study, pH >7.6 was associated with an overall mortality of 50%, compared with <30% when pH was >7.44.² However, the harmful effects of alkalaemia are difficult to distinguish from the underlying disease process and the associated hypovolaemia and electrolyte disturbances, in particular hypokalaemia and ionised hypocalcaemia (alkalaemia increases the binding of calcium to albumin leading to ionised hypocalcaemia). Ionised hypocalcaemia mediates some of the adverse effects of alkalaemia, which are summarised in Table 3.

Table 3.

Adverse effects associated with alkalaemia.

Nervous system

•Cerebral vasoconstriction with reduced cerebral blood flow •Lethargy, confusion, delirium •Neuromuscular excitability (mediated by ionised hypocalcaemia) with numbness, tetany and seizures

Respiratory

•Leftward shift in the oxygen-haemoglobin dissociation curve (impaired unloading of oxygen in the tissues) •Compensatory alveolar hypoventilation (potential for hypoxaemia) •Inhibition of hypoxic pulmonary vasoconstriction (resulting in increased $\dot{V}$/ $\dot{Q}$ mismatch)

Cardiac

•Reduced myocardial contractility (mediated by ionised hypocalcaemia) •Arrhythmias (mediated by ionised hypocalcaemia, exacerbated by concomitant hypokalaemia)

Mixed acid–base disturbance

Metabolic alkalosis usually occurs in the context of a mixed acid–base disturbance. In particular, metabolic alkalosis (diuretics, hypoalbuminaemia) frequently coexists with respiratory acidosis (hypercapnic respiratory failure) or metabolic acidosis (sepsis, renal failure, diabetic ketoacidosis, intoxications).

Depending on the underlying processes, patients with mixed acid–base disturbance may have acidaemia, alkalaemia or normal pH. A stepwise approach, including elements of the traditional and physiochemical models, helps identify the constituents of mixed acid–base disturbances. A recent arterial blood gas (pH, Paco₂, bicarbonate, lactate) and biochemistry panel (sodium, chloride, albumin) are required. If clinically indicated, ketones should also be measured. For accuracy, we recommend using sodium and chloride measurements obtained from a biochemistry panel. In particular, chloride measurements obtained from automated blood gas analysers may be inaccurate.¹⁹

In addition to direct measurements, the following calculated values should be obtained:

• (i)The SID_a [Eq (7)]. The SID_a is a useful initial screening tool for metabolic acidosis or alkalosis caused by relative changes in sodium or chloride.
• (ii)The anion gap [Eq (8)]. Used to calculate the SIG.
• (iii)The SIG [Eq (10)] to allow identification of unmeasured ions. If SIG is >2, then lactate, ketones, salicylates should be measured as clinically indicated.
• (iv)The expected Paco₂. The relationship between pH and the bicarbonate concentration is quantified by Eq (2). Alkalaemia leads to compensatory hypoventilation with increased Paco₂ and acidaemia leads to compensatory hyperventilation with decreased Paco₂. The ‘expected Paco₂’ is the Paco₂ we would expect to observe if respiratory compensation was appropriate for the degree of metabolic alkalaemia or acidaemia. For mixed acid–base disturbance with normal pH (7.35–7.45), we expect the Paco₂ to be within the normal range (35–45 mmHg, 4.7–6 kPa).

For mixed acid–base disturbances with alkalaemia (pH>7.45), the expected Paco₂ (in mmHg) is²⁰:

$$\text{Expected}P\mathrm{a}{\mathrm{c}\mathrm{o}}_{\mathrm{2}}=0.7\left[{HCO}_3^{-}\right]+20\pm 2$$ (13)

For mixed acid–base disturbances with acidaemia the expected Paco₂ (in mmHg) is²¹:

$$\text{Expected}P\mathrm{a}{\mathrm{c}\mathrm{o}}_{\mathrm{2}}=1.5\left[{HCO}_3^{-}\right]+8\pm 2$$ (14)

We then use a stepwise approach to determine the underlying acid–base abnormalities:

Step 1

A_TOT effect: is there hypoalbuminaemia? If yes, there is an A_TOT metabolic alkalosis.

Step 2

Sodium–chloride effect: is the SID_a increased? An SID_a >35 mEq L⁻¹ suggests metabolic alkalosis (as a result of relative sodium excess or low chloride) and a SID_a <35 mEq L⁻¹ suggests metabolic acidosis (as a result of relative low sodium or excess chloride). However, an increased SID alkalosis may coexist with a metabolic acidosis because of unmeasured strong anions.

Step 3

Unmeasured strong ion effect: is the SIG normal (i.e. up to plus or minus 2)? If yes, metabolic acidosis attributable to unmeasured strong anions is excluded. If the SIG level is increased and corresponds (within 2 mEq L⁻¹) to the measured raised lactate level (in mmol L⁻¹), then the metabolic acidosis is only attributable to the increase in lactate. If SIG level is much greater than measured lactate, then there will likely be other unmeasured anions contributing towards the metabolic acidosis. (A large negative SIG may represent unmeasured cations, but this is very rare).

Step 4

Appropriate Paco₂?: calculate the expected Paco₂ using Eq (13) or (14). A measured Paco₂ that is higher than expected indicates respiratory acidosis. A measured Paco₂ that is lower than expected indicates respiratory alkalosis.

Some real data from a patient are shown in Table 4. The patient had undergone lung transplantation several days previously and had developed clinical evidence of sepsis. Before developing sepsis, the patient had been receiving i.v. furosemide. There was oliguria and an acute kidney injury. The patient was spontaneously breathing via a percutaneous tracheostomy with pressure support ventilation.

The pH indicates mild alkalaemia.

Step 1

there is hypoalbuminaemia indicating metabolic alkalosis.

Step 2

the SID_a is high, suggesting metabolic alkalosis.

Step 3

the SIG is also high, indicating concomitant metabolic acidosis. In this case the SIG was 15.1 mEq L⁻¹ but the lactate was only 7.1 mmol L⁻¹, indicating that there are other unmeasured anions also contributing to the metabolic acidosis.

Step 4

the measured Paco₂ is slightly lower than expected, suggesting mild respiratory alkalosis.

Thus, the patient has a concomitant metabolic alkalosis, metabolic acidosis and mild respiratory alkalosis. The likely causes of the metabolic alkalosis are hypoalbuminaemia and relative hypochloraemia from diuretic use and intravascular volume depletion (activation of the RAAS). Metabolic acidosis is attributable to increased lactate (sepsis) plus some other contributor (likely renal failure).

Table 4.

Case study: mixed acid–base disturbance case study.

Measured variable (normal range)		Calculated parameter
pH	7.46 (7.35–7.45)	SIDa [Eq (7)]	41 mEq L−1
Paco2	33 mmHg (35–45) 4.5 kPa (4.7–6)	Anion gap [Eq (8)]	22.1 mEq L−1
Bicarbonate	23 mmol L−1 (23–28)	Expected anion gap (28/4)	7 mEq L−1
Sodium	140 mmol L−1 (135–145)	SIG [Eq (10)]	15.1 mEq L−1
Potassium	4.1 mmol L−1 (3.6–5.2)	Expected Paco2 [Eq (13)]	34–38 mmHg (4.5–5.1 kPa)
Chloride	99 mmol L−1 (96–106)
Albumin	28 g L−1 (34–54)
Lactate	7.1 mmol L−1 (0.6–1.8)

Management of metabolic alkalosis

The mainstay of managing metabolic alkalosis is determining the underlying mechanism and stopping the precipitant cause(s). Management is usually split into chloride-responsive and chloride-resistant alkalosis.

Chloride-responsive alkalosis

Conditions that lead to volume contraction usually respond to a chloride-rich solution (i.e. 0.9% saline). Potassium chloride is equivalent to administering sodium-free chloride. Therefore, administering potassium chloride is useful if there is concomitant hypokalaemia. In a highly monitored environment, one approach is to add 40 mmol potassium chloride to a 1 L bag of 0.9% saline and administer via a volumetric pump over 2–4 h via a central venous catheter. The potassium concentration should be checked once during the infusion and once after, stopping if the potassium concentration >5 mmol L⁻¹. The ECG should be continuously monitored and the infusion stopped if there are signs of hyperkalaemia.

Chloride-resistant alkalosis

Metabolic alkalosis associated with volume expansion (excess mineralocorticoid activity) is less likely to respond to chloride-based fluids and more likely to respond to diuretics.

Acetazolamide and the potassium-sparing diuretics (e.g. spironolactone) are the agents of choice. Administering 0.9% saline in addition to acetazolamide or spironolactone corrects the alkalosis more rapidly than diuretics alone. The ultimate chloride-rich, sodium-poor fluid is hydrochloric acid, and although reported to have been given to patients to correct severe metabolic alkalosis, it is not recommended routinely!²²

Intravenous fluid effect on pH

Using the physiochemical approach, the effect of an i.v. fluid depends on the in vivo SID of the fluid. Although the in vitro SID of all i.v. fluids is zero, their effect in vivo depends on the metabolism of one or more of the anions present in the fluid. For example, a solution that contains sodium, chloride, (sodium) lactate and (sodium) acetate has a higher SID than 0.9% saline once given i.v. because of the metabolism of the lactate and acetate. Fluids with an in vivo SID <25 mEq L⁻¹ have an acidifying effect whereas fluids with an in vivo SID >25 mEq L⁻¹ have an alkalinising effect.

Five percent dextrose solution has an in vivo SID of zero and, therefore, like 0.9% saline, has an acidifying effect. However, in practice, only small volumes of 5% dextrose solution are typically administered and the effect is likely trivial.

In addition to the in vivo SID of the fluid, administering i.v. fluids also dilutes A_TOT, resulting in a metabolic alkalosis. Human albumin solutions vary as they can be made in both saline and balanced solutions but tend to have an acidifying effect because of an increase in A_TOT. To achieve a neutral effect on pH, a balanced salt solution must reduce extracellular SID at a rate that will counteract the A_TOT dilutional alkalosis. Experimentally, the ideal effective SID of the solution is 24 mEq L⁻¹.²³

Ringers' lactate solution (Hartmann's solution) contains 29 mEq L⁻¹ of lactate. However, this is as sodium lactate, which does not cause acidosis (in fact, once the lactate is metabolised it has an alkalinising effect). However, Ringers' lactate solution does cause a small increase in serum lactate concentration. In one study of healthy individuals, 30 ml kg⁻¹ Ringers' lactate solution resulted in a mean lactate increase of 0.93 mEq L⁻¹ compared with subjects receiving 0.9% saline.²⁴

Despite the physiological rationale that balanced salt solutions should be beneficial compared with 0.9% saline, to date the evidence is unconvincing.25, 26 In our view, any harm associated with 0.9% saline arises from misinterpreting the acidifying effect on plasma pH and subsequent over treatment (‘chasing the base deficit with further fluid’).

The in vivo SID of commonly used i.v. fluids is presented in Table 5.²⁷

Table 5.

The strong ion difference of commonly used i.v. fluids and the acid–base effect on plasma. ∗In vivo effect after metabolism of strong anions.

Fluid	0.9% Saline	5% Glucose	Plasma-Lyte 148	Hartmann's solution (Ringer's lactate)	8.4% Sodium bicarbonate solution
[Na+] (mmol L−1)	150	0	140	131	1000
Other cations (mmol L−1)			[K+] 5 [Mg2+] 1.5	[K+] 5.4 [Ca2+] 1.5
[Cl−] (mmol L−1)	150	0	98	112
[HCO3−] (mmol L−1)					1000
Organic anions (mEq L−1)			[Acetate] 27 [Gluconate] 23	[Lactate] 29
Net SID (mEq L−1)	0	0	50∗	27∗	1000
pH (in vitro)	4–7	3.5–6.5	7.4	5–7	7–8.5
Effect on plasma acid–base	Acidifying	Acidifying	Alkalinising	Near neutral	Alkalinising

Conclusions

Metabolic alkalosis is the most common acid–base disturbance encountered after the first few days of admission to the ICU and is associated with increased mortality. Increased mortality is likely to be attributable in part to the patient's underlying disease severity rather than alkalosis per se.

Metabolic alkalosis rarely occurs in isolation. Rather, metabolic alkalosis frequently coexists with respiratory acid–base disorders and metabolic acidosis. Thus, to diagnose metabolic alkalosis it is necessary to understand mixed acid–base disturbances. The physiochemical approach to acid–base disturbances is mathematically complicated, which is one of the reasons it has not been widely adopted. However, a simplified physiochemical approach, based on only a few measurements and simplified equations, provides an alternative framework for understanding metabolic alkalosis (and acidosis) and for diagnosing mixed acid–base disturbances.

Declaration of interests

DS is an editor and editorial board member of BJA Education. MP declares that he has no conflicts of interest.

Biographies

David Sidebotham FANZCA is a cardiac anaesthetist and intensivist in Auckland, New Zealand. He is an editorial board member and editor for BJA Education.

Michael Park FRCP Edin FCICM is an intensive care physician in Hastings, New Zealand. He is chair of the Central region critical care leadership group and a certified instructor for the BASIC courses.

Matrix codes: 1A01; 2C01; 3C00

MCQs

The associated MCQs (to support CME/CPD activity) will be accessible at www.bjaed.org/cme/home by subscribers to BJA Education.