The Acid–Base Effects of Albumin in Sepsis: Reconciling the Stewart Physicochemical Approach with the Traditional Buffer–Base Paradigm

1. Introduction

Fluid resuscitation remains one of the cornerstones of sepsis management [1,2,3]. However, the physiological consequences of different resuscitation fluids remain incompletely understood, particularly regarding acid–base equilibrium. Although crystalloids have traditionally dominated initial resuscitation strategies, albumin continues to attract interest because of its unique physiological properties and its potential role in severe sepsis and septic shock [4,5,6].

Albumin differs fundamentally from crystalloid solutions because it simultaneously influences oncotic pressure, endothelial function, drug transport, antioxidant activity, intravascular volume, and acid–base physiology [7,8,9,10]. Unlike balanced crystalloids or saline, albumin administration modifies both strong ion balance and the circulating concentration of weak acids.

Traditionally, acid–base disorders have been interpreted through the Henderson–Hasselbalch equation and the bicarbonate-centered framework. Within this paradigm, albumin acts primarily as a non-bicarbonate plasma buffer affecting buffering capacity, base excess, and anion gap interpretation.

In contrast, Stewart proposed a physicochemical model in which pH is determined by three independent variables: partial pressure of carbon dioxide (pCO₂), strong ion difference (SID), and total concentration of weak acids (Atot). Within this framework, albumin is interpreted as a weak non-volatile acid directly contributing to Atot.

These two frameworks offer complementary explanations for albumin-related acid-base changes, particularly in septic patients with hypoalbuminemia, hyperchloremia, and mixed metabolic disorders.

This narrative review aims to reconcile traditional and physicochemical interpretations of albumin-related acid–base physiology in sepsis. We integrate mechanistic concepts, clinical trial context, conceptual simulations, and bedside implications to propose a practical framework for interpretation.

This article is intended as a mechanistic and clinically oriented narrative review, rather than as a systematic evidence synthesis.

2. Historical Evolution of Acid–Base Interpretation

Modern acid–base physiology emerged from more than a century of progressive conceptual refinement. Early physiological models focused primarily on the relationship between carbon dioxide and bicarbonate, whereas later approaches attempted to integrate electrolyte balance, buffering systems, and physicochemical principles into a unified interpretation of plasma acid–base equilibrium.

The Henderson equation, introduced in the early twentieth century and later reformulated by Hasselbalch into logarithmic form, established the conceptual basis of bicarbonate-centered acid–base analysis [10,11]. Within this framework, pH was interpreted as a function of bicarbonate concentration and dissolved carbon dioxide tension. This approach rapidly became clinically dominant because it was intuitive, mathematically accessible, and closely aligned with arterial blood gas measurements.

Subsequent developments by Van Slyke and Siggaard-Andersen expanded the clinical applicability of bicarbonate-centered interpretation through concepts such as buffer base and base excess [12,13,14]. These variables attempted to isolate the “metabolic” component of acid–base disorders independently of respiratory compensation. Base excess became deeply integrated into critical care and perioperative medicine because it provided a rapid bedside estimate of metabolic acid–base disturbances.

Despite its clinical usefulness, the traditional framework encountered increasing difficulties when confronted with complex ICU scenarios characterized by mixed electrolyte abnormalities, dilutional states, hypoalbuminemia, hyperchloremia, and large-volume fluid resuscitation [14,15]. In such situations, bicarbonate concentration often behaved more as a dependent variable than as a mechanistic explanation of acid–base change.

These limitations stimulated interest in alternative interpretative models. Stewart proposed a physicochemical approach in which hydrogen ion concentration was determined by three independent variables: pCO₂, strong ion difference (SID), and total concentration of weak acids (Atot) [16]. In this framework, bicarbonate was no longer considered an independent regulator of pH but rather a dependent consequence of physicochemical equilibrium.

Fencl and colleagues later translated Stewart’s mathematically complex model into a clinically accessible bedside approach by introducing simplified calculations for SID, albumin effects, and unmeasured anions [17,18,19]. This adaptation substantially increased the clinical applicability of physicochemical acid–base interpretation in critical care medicine.

The rise of Stewart physiology coincided with growing recognition of hyperchloremic metabolic acidosis during saline resuscitation and with increasing awareness that hypoalbuminemia itself exerted major acid–base effects [20,21]. Critically ill patients often exhibited apparently paradoxical combinations of normal pH, elevated lactate, low albumin, and severe electrolyte abnormalities that were incompletely explained by bicarbonate-centered interpretation alone.

Within this historical context, albumin occupies a particularly important conceptual position because it lies at the intersection of the two frameworks. In the traditional model, albumin is viewed primarily as a plasma buffer influencing base excess and anion gap interpretation. In the Stewart framework, albumin becomes one of the principal determinants of Atot and therefore a direct regulator of acid–base equilibrium.

The evolution from traditional to physicochemical interpretation should not be viewed as a replacement of one model by another. Rather, the Stewart approach expanded mechanistic understanding of acid–base physiology while traditional variables retained substantial practical and prognostic utility. Contemporary critical care practice increasingly relies on integration of both perspectives rather than exclusive adherence to either framework.

3. The Traditional Buffer–Base Interpretation

The traditional approach to acid–base physiology is centered on the Henderson–Hasselbalch equation:

$$pH=6.1+\mathrm{l}\mathrm{o}\mathrm{g}\left({\displaystyle \frac{\left[HC{O}_3^{-}\right]}{0.03\times PC{O}_2}}\right)$$

Within this framework, pH is determined by the relationship between bicarbonate concentration and pCO₂ [10,11].

Albumin functions primarily as a non-bicarbonate plasma buffer because of its multiple negatively charged residues capable of binding hydrogen ions. Consequently, hypoalbuminemia reduces total plasma buffering capacity.

This phenomenon has several clinically important consequences. First, hypoalbuminemia lowers the measured anion gap despite persistent accumulation of unmeasured anions [22,23]. Correction formulas for albumin-adjusted anion gap are therefore recommended in critically ill patients. Second, severe hypoalbuminemia may itself exert an alkalinizing effect because weak acidic buffering components are reduced [24]. Conversely, albumin administration may appear acidifying because restoration of weak acid buffering may lower bicarbonate concentration and base excess despite stable or improving tissue perfusion [25,26,27].

However, the traditional framework often struggles to distinguish albumin-specific effects from concomitant changes in chloride balance, sodium concentration, dilution, renal compensation, and tissue perfusion.

4. The Stewart Physicochemical Approach

Stewart proposed that hydrogen ion concentration is determined by three independent variables:

• Partial pressure of carbon dioxide (pCO₂)
• Strong ion difference (SID)
• Total concentration of weak acids (Atot)

This framework may be summarized conceptually as:

$$pH=f(PC{O}_2,SID,A_{tot})$$

Within plasma, albumin and phosphate constitute the principal weak acids contributing to Atot.

Strong ion difference may be simplified as:

$$SID=\left[N{a}^{+}\right]+\left[K^{+}\right]+\left[C{a}^{2+}\right]+\left[M{g}^{2+}\right]-\left[C{l}^{-}\right]-\left[Lactat{e}^{-}\right]$$

(1)

According to Stewart physiology, increases in Atot favor hydrogen ion dissociation and therefore lower pH [16]. Conversely, reductions in Atot exert an alkalinizing tendency. Fluids with SID substantially lower than plasma SID produce metabolic acidosis, whereas fluids with higher SID are relatively alkalinizing. Importantly, albumin administration may influence acid–base equilibrium through at least two simultaneous mechanisms:

• increased Atot (weak acid effect);
• altered SID due to accompanying chloride load.

This distinction explains why apparently similar acid–base changes may emerge through fundamentally different mechanisms.

5. Traditional Versus Stewart Interpretations

The traditional and Stewart approaches are often presented as alternative or competing models of acid–base physiology. However, in the context of albumin administration, they are better understood as complementary interpretative frameworks [28]. Both describe the same biological system, but they emphasize different explanatory levels. This conceptual contrast between the traditional buffer–base approach and the Stewart physicochemical approach is summarized in Figure 1.

The Stewart approach shifts the interpretative focus from bicarbonate to physicochemical determinants [18,19]. In this model, bicarbonate is not an independent determinant of pH but a dependent variable generated by the interaction among pCO₂, SID, and Atot. Albumin is therefore not simply a buffer but a major weak acid contributing to Atot. A fall in albumin concentration lowers Atot and tends to alkalinize plasma, whereas albumin administration increases Atot and tends to lower pH.

This difference is particularly relevant in septic shock, where hypoalbuminemia, hyperchloremia, hyperlactatemia, renal dysfunction, and large-volume fluid administration frequently coexist. In such settings, the traditional framework may correctly identify the presence of metabolic acidosis but may not fully explain why bicarbonate or base excess changes after albumin infusion. Conversely, the Stewart framework may clarify the mechanism but can be less immediately accessible at the bedside [29].

Neither interpretation necessarily implies clinical deterioration. A decrease in pH after albumin administration may occur even when hemodynamic variables improve, because the acid–base change may reflect restoration of weak acid concentration rather than worsening tissue hypoperfusion. This distinction is central to the clinical interpretation of albumin therapy in sepsis. The main conceptual differences between the two frameworks are also summarized in Table 1

6. Physicochemical Simulations of Albumin Administration

To illustrate the physiological mechanisms discussed in this review, we generated simplified conceptual simulations based on Stewart principles under fixed baseline assumptions. These simulations were not intended to provide calibrated patient-level predictions, but rather to illustrate the expected direction of acid–base changes when albumin concentration, chloride burden, SID, and lactate vary under controlled assumptions. The simulated directional effects of albumin 20%, albumin 4%, normal saline, and lactated Ringer solution are illustrated in Figure 2.

Three mechanisms were explored. First, increasing albumin concentration at stable SID and pCO₂ was used to illustrate the Atot-mediated acidifying effect of albumin. Second, saline and albumin-containing solutions were compared conceptually to distinguish SID-mediated acidosis from Atot-mediated acidification. Third, hypoalbuminemia was modeled as a condition capable of partially masking lactic or hyperchloremic acidosis, with albumin administration potentially “unmasking” the underlying metabolic disturbance.

These simulations are purely illustrative and should not be interpreted as validated quantitative physiological predictions or patient-level models. Their purpose is to clarify mechanisms, not to predict the exact pH response of individual patients.

7. Fluid Composition, Chloride Burden, and Strong Ion Difference

The relationship between chloride balance and acid–base physiology represents one of the most clinically relevant applications of Stewart’s physicochemical framework. Among critically ill patients, hyperchloremia is extremely common and frequently results from administration of chloride-rich intravenous fluids, particularly normal saline and some albumin preparations [27,29].

Within the traditional framework, saline-associated metabolic acidosis has historically been described as “dilutional” or “hyperchloremic” acidosis. However, Stewart physiology provides a mechanistic explanation by demonstrating that chloride excess reduces strong ion difference (SID), thereby favoring hydrogen ion dissociation and lowering pH [25].

Normal plasma SID is approximately 38–42 mEq/L, whereas the SID of normal saline is effectively zero because sodium and chloride concentrations are equal. Administration of large volumes of saline therefore lowers plasma SID and predictably produces metabolic acidosis. Balanced crystalloids differ substantially in this regard. Solutions such as lactated Ringer or Plasma-Lyte contain lower chloride concentrations and higher effective SID values. Consequently, they exert smaller acidifying effects compared with saline-based fluids.

This distinction has become increasingly relevant following major randomized trials comparing balanced crystalloids and saline. Trials such as SPLIT, SMART, BaSICS, and PLUS primarily evaluated clinical outcomes rather than acid–base physiology, yet they collectively reinforced the concept that fluid composition itself may influence renal function, electrolyte balance, and acid–base equilibrium [30,31,32,33,34].

Although the magnitude of clinical benefit remains debated, several mechanistic pathways have been proposed through which chloride-rich solutions may adversely affect organ function. Experimental studies suggest that hyperchloremia may induce renal vasoconstriction through tubuloglomerular feedback mechanisms, potentially reducing renal cortical perfusion and glomerular filtration [35]. Hyperchloremia may also influence inflammatory signaling, endothelial function, and microcirculatory regulation [29,35].

Albumin-containing solutions require additional interpretation because they simultaneously modify chloride balance and weak acid concentration. Albumin 4% solutions exert relatively modest Atot-related effects because of their lower weak acid concentration, whereas hyperoncotic albumin preparations may produce stronger Atot-mediated acidifying tendencies [5]. Furthermore, many commercially available albumin formulations are suspended in chloride-rich carrier solutions, thereby combining SID reduction with weak acid expansion.

From a clinical perspective, these differences are particularly relevant in septic shock, where hypoalbuminemia and cumulative chloride exposure frequently coexist. Under these conditions, isolated interpretation of bicarbonate concentration or base excess may incompletely capture the complexity of the underlying acid–base disorder. The expected directional acid–base effects of common resuscitation fluids are summarized in Table 2.

9. Masked Acidosis and Bedside Interpretation Pitfalls

Hypoalbuminemia is common in septic shock and critical illness. From a Stewart perspective, reduced albumin concentration lowers Atot and therefore exerts an alkalinizing tendency. This effect may partially offset hyperchloremia, lactic acidosis, or unmeasured anions, leading to apparently near-normal pH or base excess values despite relevant metabolic derangement [36]. This mechanism is illustrated schematically in Figure 3.

Albumin administration may reverse this masking effect. Restoration of weak acid concentration may transiently reduce pH or base excess even when hemodynamics and perfusion markers are stable or improving [25,37,38]. Therefore, isolated pH reductions after albumin infusion should be interpreted together with lactate, chloride, albumin concentration, renal function, and hemodynamic trends.

A second interpretative challenge is that saline and albumin may produce similar arterial blood gas patterns through different mechanisms. Saline-associated acidosis is predominantly mediated by chloride-induced SID reduction, whereas albumin-associated acidification mainly reflects expansion of weak acid concentration. Similar pH or base excess values may therefore correspond to distinct physiological states with different implications for fluid management.

Three practical pitfalls should be avoided at the bedside. First, near-normal pH or base excess should not be considered reassuring when severe hypoalbuminemia coexists with elevated lactate, hyperchloremia, renal dysfunction, or unmeasured anions. Second, a modest fall in pH after albumin administration should not be interpreted in isolation from hemodynamic and perfusion trends [39]. Third, chloride-associated acidosis should not be considered equivalent to lactate-associated acidosis, because the former is primarily SID-mediated whereas the latter may reflect altered metabolism, impaired clearance, or tissue hypoperfusion. Representative scenarios are summarized in Table 3.

10. Integrated Bedside Interpretation

In clinical practice, conventional arterial blood gas variables should be interpreted alongside albumin concentration, chloride balance, lactate, renal function, and hemodynamic response. This integrated approach helps distinguish SID-mediated acidosis from Atot-mediated acidification, identify masked metabolic acidosis, and avoid overinterpreting isolated changes in bicarbonate or base excess [25,26,39]. A proposed bedside framework is shown in Figure 4.

Traditional variables remain essential for rapid assessment of acidemia, respiratory compensation, and the metabolic component of acid–base disturbance. Stewart-derived variables add mechanistic resolution when hypoalbuminemia, chloride exposure, renal dysfunction, or mixed metabolic disorders coexist. In this context, albumin concentration, corrected anion gap, chloride, SID, and unmeasured ions may clarify apparently paradoxical arterial blood gas findings.

Importantly, acid–base interpretation should remain linked to organ-level physiology [40,41]. Lactate kinetics, vasopressor requirements, urine output, fluid balance, renal function, respiratory status, and overall perfusion should determine whether a change in pH reflects true clinical deterioration or a predictable physicochemical shift. A suggested integrated bedside interpretation framework is summarized in Table 4.

11. Cardiorespiratory Interactions During Albumin Administration and Acid–Base Resuscitation

Acid–base physiology during sepsis cannot be fully separated from cardiopulmonary interactions. In mechanically ventilated septic patients, alveolar ventilation, dead-space fraction, pulmonary vascular tone, right ventricular loading conditions, and tissue perfusion may all influence the clinical expression of albumin-related acid–base changes.

Carbon dioxide represents the respiratory determinant of pH within both traditional and physicochemical frameworks [42]. Therefore, ventilatory settings and CO₂ clearance may interact with SID- and Atot-mediated metabolic effects, particularly in patients managed with protective ventilation or permissive hypercapnia [43,44].

Albumin may also affect respiratory physiology indirectly through changes in intravascular oncotic pressure, endothelial permeability, and pulmonary fluid distribution. Under selected conditions, restoration of oncotic pressure may favor reabsorption of extravascular lung water, although these effects are context-dependent and should not be interpreted as evidence that albumin uniformly improves oxygenation or respiratory outcomes [45,46].

Correction of effective hypovolemia may improve preload, cardiac output, microcirculatory flow, and oxygen delivery, even when arterial pH modestly decreases because of increased Atot [47,48,49]. Conversely, hyperchloremic acidosis and severe acidemia may adversely affect pulmonary vascular resistance, right ventricular afterload, myocardial contractility, and vasomotor responsiveness [35].

Positive-pressure ventilation further modifies this balance by altering venous return, pulmonary perfusion, right ventricular afterload, and CO₂ elimination [45,46]. In septic patients with ARDS, the coexistence of permissive hypercapnia, chloride-rich fluid exposure, hypoalbuminemia, and lactate accumulation may generate mixed acid–base disorders in which apparently small pH changes conceal substantial physiological heterogeneity [48,49]. These interactions are summarized schematically in Figure 5.

12. Randomized Controlled Trials: Clinical Context for Albumin and Fluid Composition

Although this review focuses on acid–base physiology, the interpretation of albumin therapy in sepsis should be anchored to the major randomized trials evaluating albumin, colloids, saline, and balanced crystalloids in critically ill patients. These trials were not designed to test Stewart physiology, Atot, SID, or chloride-mediated acid–base trajectories, but they provide the clinical background within which physiological effects must be interpreted.

The SAFE trial compared 4% albumin with normal saline in 6997 critically ill patients requiring fluid resuscitation [4]. Overall, albumin and saline produced similar 28-day outcomes. For the present review, SAFE is important because it directly compares two fluids with distinct physicochemical profiles: albumin increases weak acid concentration, whereas saline has a SID of zero and may promote hyperchloremic acidosis. However, SAFE was not designed to evaluate whether acid–base differences between albumin and saline mediated clinical outcomes.

The ALBIOS trial is the most relevant randomized trial for albumin in sepsis [5]. It randomized 1818 patients with severe sepsis to receive 20% albumin plus crystalloids or crystalloids alone, with the aim of maintaining serum albumin concentrations of at least 30 g/L. Albumin did not improve 28-day or 90-day survival in the overall population, although hemodynamic variables and fluid balance differed between groups. From an acid–base perspective, ALBIOS is particularly relevant because it involved hyperoncotic albumin and systematic albumin replacement, but it did not specifically evaluate Atot-mediated acidification, chloride balance, SID, or albumin-corrected acid–base variables.

The CRISTAL trial compared colloids with crystalloids in ICU patients with hypovolemic shock [50]. Because albumin was included among several colloid options, the trial cannot be interpreted as albumin-specific evidence. Nevertheless, it provides useful clinical context by showing that the choice between colloids and crystalloids cannot be reduced to a simple survival advantage. For this review, CRISTAL reinforces the need to separate clinical outcome questions from mechanistic physiological questions. The major randomized trials evaluating albumin or colloids in critical illness are summarized in Table 5.

Trials comparing balanced crystalloids with saline provide complementary evidence relevant to SID and chloride physiology. SPLIT compared Plasma-Lyte 148 with saline in ICU patients and did not show a significant reduction in acute kidney injury [30]. SMART, a larger pragmatic ICU trial, found a lower rate of major adverse kidney events with balanced crystalloids compared with saline [31]. By contrast, BaSICS and PLUS did not demonstrate clear overall survival or kidney-outcome superiority of balanced solutions over saline in broad ICU populations [33,34]. These apparently heterogeneous results do not invalidate chloride/SID physiology; rather, they suggest that measurable physicochemical effects may not consistently translate into large outcome differences across unselected trial populations.

More recently, ARISS evaluated albumin replacement in septic shock [6]. The trial reported that albumin administration appeared safe but did not improve 90-day survival; premature termination limits the certainty of conclusions. ARISS is clinically important because it directly addressed albumin replacement in septic shock, but, like previous trials, it was not designed primarily around acid–base endpoints. The major randomized trials comparing balanced crystalloids and saline are summarized in Table 6.

Overall, these trials show that predictable physiological effects do not necessarily translate into consistent mortality benefit. Albumin may modify Atot, pH, bicarbonate, base excess, and anion-gap interpretation while remaining clinically neutral at the trial level. Similarly, balanced crystalloids may reduce chloride exposure and preserve SID without producing uniform outcome benefits.

13. Future Directions

Future research should move beyond simplistic classifications of fluids as “acidifying” or “alkalinizing.” The physiological effects of albumin likely depend on dynamic interactions among Atot, SID, chloride balance, dilution, renal compensation, and tissue perfusion.

Prospective physiological studies comparing albumin 4%, albumin 20%, balanced crystalloids, and saline-based strategies using both traditional and Stewart metrics may clarify these interactions.

Integration of physicochemical analyses into future sepsis trials may also improve interpretation of arterial blood gas trajectories during resuscitation.

14. Limitations

This review has some limitations. First, many physicochemical interpretations discussed herein derive from conceptual and mechanistic models rather than from randomized physiological trials specifically designed around Stewart variables. Second, the conceptual simulations presented are illustrative and are not intended as calibrated patient-level predictions. Third, acid–base responses during sepsis are highly dynamic and influenced by renal compensation, respiratory status, vasopressor exposure, and fluid balance, factors that cannot be fully captured by simplified frameworks alone. Finally, because this is a narrative review, no systematic literature search, formal study selection process, or risk-of-bias assessment was performed; therefore, the evidence discussed should be interpreted as mechanistic and contextual rather than as a comprehensive systematic synthesis.

15. Conclusions

Albumin affects acid-base equilibrium through weak acid expansion, chloride balance, dilution, and strong ion physiology. In septic patients, particularly those with hypoalbuminemia or chloride exposure, traditional and Stewart approaches should be integrated rather than opposed. This combined interpretation may help clinicians distinguish true metabolic deterioration from predictable physicochemical changes after albumin administration.