13. Albumin vs. Crystalloid in Critical Illness (SAFE and ALBIOS)

Quick Recap

Cross-cutting protocol — companion to Balanced Crystalloids vs. Saline in Critical Illness protocol, addressing the separate, historically older question of colloid (albumin) vs. crystalloid resuscitation. Grounded in the SAFE trial, one of the field's genuinely decisive, practice-clarifying results — unusual in this library for actually resolving decades of conflicting smaller-trial and meta-analytic evidence rather than adding another layer of discordance — alongside a persistent, still-unresolved sepsis-specific subgroup signal that three separate trials have now failed to definitively confirm or refute.

1. Definition

Albumin: a naturally occurring plasma protein, produced by the liver, responsible for approximately 80% of intravascular oncotic pressure, with additional physiological roles in acid-base homeostasis, drug/molecule transport, and scavenging of reactive oxygen species — available as a resuscitation/replacement fluid in 4–5% (iso-oncotic) or 20–25% (hyperoncotic) concentrations.

The pre-SAFE evidence landscape: prior to 2004, the albumin-vs-crystalloid question had been addressed only by smaller trials and conflicting meta-analyses — a 1998 Cochrane Injuries Group systematic review had raised concern that albumin might be harmful, while other, contemporaneous small-trial data suggested organ-function benefit in hypoalbuminemic patients — genuine, substantive disagreement in the literature that motivated the large, definitive trial discussed below.

Hypoalbuminemia in sepsis: recognized as a poor prognostic marker, thought to reflect disease severity via endothelial dysfunction and decreased serum albumin half-life — the biological plausibility for albumin replacement (as distinct from albumin as a resuscitation fluid choice) in sepsis specifically, a conceptually distinct question from the general ICU resuscitation-fluid-choice question SAFE addressed (Section 11).

2. Pathophysiology

Albumin's oncotic properties theoretically provide more effective, sustained intravascular volume expansion per unit volume infused compared to crystalloids, which redistribute more readily into the extravascular space — a mechanistically coherent rationale for colloid resuscitation that nonetheless, as the evidence below demonstrates, has not translated into a clinically detectable outcome advantage in the general ICU population, a pattern structurally similar to several other topics in this library where plausible mechanism did not confer confirmed clinical benefit.

Why traumatic brain injury specifically emerged as a population of concern: albumin's oncotic effect, combined with its potential to alter cerebral perfusion pressure dynamics in the setting of disrupted blood-brain-barrier integrity after TBI, has been proposed as the mechanistic basis for the population-specific harm signal identified in SAFE (Section 11) — though the precise mechanism was not definitively established by the trial itself, this represents the most consistently replicated population-specific finding in this entire evidence base.

3. Immediate Stabilization (ABCDE) — Not Applicable in the Traditional Sense

This protocol addresses fluid-type selection rather than acute stabilization; cross-reference Fluid Responsiveness & Hemodynamic Assessment and Balanced Crystalloids vs. Saline protocols for the broader resuscitation framework.

Checklist:

4. Focused History

• Underlying reason for fluid resuscitation requirement
• Traumatic brain injury status specifically (the clearest, most consistently replicated contraindication)
• Sepsis/septic shock status and severity (relevant to the persistent, unresolved subgroup question)
• Baseline serum albumin level (relevant to the distinct question of albumin replacement for hypoalbuminemia, as tested in ALBIOS, rather than albumin as a general resuscitation fluid choice)
• Cirrhosis/liver disease (a population with a biologically distinct rationale for albumin use, addressed in dedicated hepatology-specific literature and protocols, distinct from this protocol's general critical-illness focus)

5. Comprehensive System-wise Examination — Not a Primary Component

Not a primary examination-driven protocol; cross-reference Fluid Responsiveness & Hemodynamic Assessment protocol for the broader volume-status assessment this fluid-type decision sits within.

6. Syndrome Identification — Reframed as Population-Specific Applicability Classification

• General, heterogeneous critically ill ICU population requiring fluid resuscitation: SAFE established clear clinical equivalence between 4% albumin and saline for 28-day mortality — crystalloid remains the appropriate, cost-effective default (cross-reference Balanced Crystalloids vs. Saline protocol for the further, more granular crystalloid-type question)
• Traumatic brain injury: albumin was associated with significantly worse outcomes in SAFE's own subgroup analysis — a clear, population-specific contraindication
• Severe sepsis/septic shock: a genuinely persistent, unresolved question — SAFE's own severe sepsis subgroup showed a non-significant trend toward benefit with albumin (p=0.09), and this signal has been pursued through multiple subsequent, larger dedicated trials (ALBIOS) without full resolution (Section 11)
• Hypoalbuminemic critically ill patients, albumin as a replacement/correction strategy rather than resuscitation fluid choice: a conceptually distinct question specifically tested in ALBIOS, addressed separately in Section 11

7. Differential Diagnosis — Not Applicable

8. Severity/Risk Assessment

Chance-finding caveat, explicitly stated by SAFE's own investigators: regarding the sepsis subgroup signal, the trial's own published discussion specifically noted that "such differences between subgroups frequently occur by chance and that only specifically designed and appropriately powered studies can determine whether any such treatment effects are real" — a direct, appropriately cautious statement from the original investigators themselves, worth taking as seriously as the numerical finding itself.

Disease severity as a specific effect modifier in ALBIOS: a striking, disease-severity-dependent split emerged in ALBIOS's own results (Section 11) — patients with septic shock specifically showed a significant benefit signal, while patients with severe sepsis without shock showed a non-significant trend toward harm, illustrating that even within the broader "sepsis" population, effect direction may plausibly differ by severity/shock status rather than being uniform.

9. Investigations

Routine labs: serum albumin level, relevant specifically to the distinct hypoalbuminemia-correction question (ALBIOS) rather than the general resuscitation fluid-choice question

10. Point-of-Care Ultrasound — Not Applicable

11. Evidence-Based Management — A Genuinely Decisive General Finding Alongside a Persistent Sepsis-Specific Puzzle

SAFE (2004) — A Rare, Genuinely Decisive Trial in This Literature

• SAFE trial (Finfer et al., NEJM 2004): a large, blinded, multicenter RCT (ANZICS Clinical Trials Group), n=6,997, randomizing a heterogeneous general ICU population requiring IV fluid resuscitation to 4% albumin or saline
• Primary outcome (28-day all-cause mortality): no significant difference — the trial's own conclusion stated directly: "albumin and saline should be considered clinically equivalent treatments for intravascular volume resuscitation in a heterogeneous population of patients in the ICU"
• This is explicitly noted, in retrospective independent commentary, as one of the field's more decisive results: SAFE is credited with "decisively moving the field beyond conflicting small-study meta-analyses" that had previously left the albumin-vs-crystalloid question genuinely unsettled — a notable contrast to several other topics in this library where even large, rigorous trials have left genuine, persistent ambiguity; SAFE's overall, general-population finding is treated by this protocol as a rare example of a definitively resolved question
• A clear, population-specific harm signal: traumatic brain injury patients receiving albumin had significantly worse outcomes — this specific subgroup finding has been consistently influential, directly shaping subsequent clinical practice and guideline recommendations against albumin use in TBI, and represents one of the more confidently established population-specific contraindications in this general area of critical care fluid management
• A non-significant, hypothesis-generating trend in the opposite direction: the severe sepsis subgroup showed a non-significant trend toward lower mortality with albumin (p=0.09) — explicitly NOT statistically significant, but influential enough to motivate a substantial, dedicated subsequent trial program (Section 11 below)

ALBIOS (2014) — A Dedicated, Larger Sepsis-Specific Trial — Still Not Fully Resolving the Question

• ALBIOS trial (Caironi et al., NEJM 2014): a different, more targeted question than SAFE — rather than testing albumin as a general resuscitation fluid, ALBIOS specifically tested albumin replacement to correct hypoalbuminemia (targeting serum albumin ≥30 g/L) in patients with severe sepsis or septic shock, n=1,818 across 100 Italian ICUs, open-label, with crystalloid given as the resuscitation fluid to both arms — a genuinely distinct clinical question from SAFE's fluid-choice comparison
• Primary outcome (28- and 90-day mortality): no significant overall difference
• A striking, disease-severity-dependent split within the trial's own results: among the ~1,100 patients with septic shock specifically, those receiving albumin had a significant relative risk for death of 0.87 (favoring albumin); since the overall trial was negative, this necessarily implied that patients with severe sepsis without shock had a corresponding, non-statistically-significant increased risk (RR 1.13) with albumin — a genuine, internally consistent severity-dependent effect-modification pattern, not simply statistical noise, though the shock-specific finding was itself a subgroup analysis within an overall negative trial
• Genuine, acknowledged methodological limitations: the trial was open-label (treatment teams knew randomization assignments, introducing potential unmeasured confounding in care decisions), and did not test albumin as a resuscitation fluid per se — patients had already undergone goal-directed therapy before receiving the albumin/no-albumin intervention, meaning ALBIOS answers a narrower, more specific question (hypoalbuminemia correction in already-resuscitated sepsis patients) than the broader "should albumin be used in sepsis" framing sometimes applied to it
• A subsequent secondary analysis: specifically in patients with severe shock and raised lactate, identified a survival benefit reaching statistical significance — a further, more granular subgroup signal reinforcing the shock-severity-dependent pattern, though again a secondary, hypothesis-generating analysis rather than the trial's primary, pre-specified confirmatory finding

Meta-Analytic Synthesis — A Modest, Shock-Specific Signal That Has Not Yet Been Definitively Confirmed

• A 2014 meta-analysis (Patel et al., BMJ, incorporating SAFE, ALBIOS, and three other trials, with trial sequential analysis methodology) found a significant improvement in mortality specifically for patients with severe septic shock, with trends toward benefit (not full statistical confirmation) for the broader severe sepsis population — broadly consistent with, and reinforcing, the shock-severity-dependent pattern observed within ALBIOS's own internal subgroup split
• A 2013 Cochrane review (broader population, not sepsis-specific), by contrast, found no benefit of albumin over crystalloids for general fluid resuscitation (all-cause mortality RR 1.01, 95% CI 0.93–1.10), explicitly concluding that "the use of colloids is unjustified given their increased cost and lack of benefit" for the general ICU population — a strong, direct statement reinforcing SAFE's own general-population equivalence finding

Practical Synthesis

For the general, heterogeneous critically ill ICU population, SAFE provides a rare, genuinely decisive answer: crystalloid and albumin are clinically equivalent for mortality, and given albumin's approximately 50-fold higher cost, crystalloid remains the appropriate default (cross-reference Balanced Crystalloids vs. Saline protocol for the further crystalloid-type-specific question). Traumatic brain injury represents a clear, consistently replicated contraindication to albumin use. The sepsis/septic shock-specific question remains genuinely, persistently unresolved despite three decades and multiple dedicated trials: SAFE's own subgroup trend, ALBIOS's internal shock-vs-no-shock split, and the subsequent meta-analytic reinforcement of a shock-specific signal together suggest a plausible, disease-severity-dependent benefit concentrated specifically in septic shock — but this remains built substantially on subgroup and secondary analyses across multiple trials rather than a single, adequately-powered, primary confirmatory finding, and should be presented with corresponding epistemic caution rather than as an established indication. Current Surviving Sepsis Campaign guidance (cross-reference Sepsis Bundles & Source Control and Septic Shock protocols) reflects this genuine uncertainty with a suggestion (not a strong recommendation) to consider albumin in patients requiring substantial crystalloid volumes, rather than a definitive, universal directive.

12. Organ Support

Interacts directly with Fluid Responsiveness & Hemodynamic Assessment and Balanced Crystalloids vs. Saline protocols for the broader fluid resuscitation strategy this specific fluid-type question sits within.

13. Disease-Specific Therapy

• Crystalloid: the appropriate default resuscitation fluid for the general ICU population, per SAFE's equivalence finding and the substantially lower cost
• Albumin (4–5%): reasonable, cost-justified consideration specifically in septic shock requiring substantial crystalloid volumes, per the persistent (though not definitively confirmed) shock-specific signal across SAFE, ALBIOS, and subsequent meta-analysis; contraindicated in traumatic brain injury per SAFE's clear subgroup harm finding
• Albumin replacement targeting a specific serum albumin threshold (e.g., ≥30 g/L per ALBIOS protocol): a distinct clinical question from resuscitation fluid choice — current evidence does not support this as a routine, independently beneficial strategy in unselected severe sepsis, though the shock-specific subgroup signal noted above applies here as well

14. Consultation Matrix

15. Monitoring Framework

• Clinical: standard volume status assessment per Fluid Responsiveness & Hemodynamic Assessment protocol
• Laboratory: serum albumin trend if albumin replacement strategy is specifically being pursued

16. ICU Bundle Checklist

17. Complications

Albumin-specific:

• Significantly worse outcomes when used in traumatic brain injury — a clear, established harm signal
• Cost burden without demonstrated benefit in the general ICU population

Prevention: appropriate population-specific fluid selection per Section 6/13; explicit TBI screening before any albumin use

18. Escalation & De-escalation — Not Applicable

19. ICU Discharge Criteria — Not Applicable

20. Documentation & Medicolegal Checklist

• Fluid type selected and rationale documented, particularly for any albumin use and the specific population/indication supporting it
• TBI status explicitly documented as excluded before any albumin administration

21. Key Guidelines

• Surviving Sepsis Campaign: suggests (not strongly recommends) albumin use in patients requiring substantial crystalloid volumes, reflecting the genuine, persistent uncertainty in the sepsis-specific evidence base
• 2013 Cochrane review: general ICU population — no benefit of colloids over crystalloids; use of colloids "unjustified given their increased cost and lack of benefit"

22. Landmark Trials

23. Controversies

• SAFE's genuine decisiveness for the general population stands in useful contrast to this library's more common pattern of persistent, unresolved discordance: this protocol explicitly notes that SAFE represents a comparatively rare example of a large trial that actually settled a previously genuinely contested question (the pre-2004 conflicting meta-analyses) rather than adding another layer of ambiguity — worth recognizing as a positive counterexample to the more common pattern (single-center enthusiasm reversed by larger trials, or genuinely persistent equipoise) seen throughout most of this library's other protocols.
• The sepsis-specific subgroup signal remains genuinely, persistently unresolved despite three decades of accumulating evidence: unlike TBI (where SAFE's finding has been consistently treated as definitive), the sepsis/shock-specific benefit signal has now appeared across SAFE's own subgroup analysis, ALBIOS's internal shock-vs-no-shock split, a subsequent lactate-specific secondary analysis, and pooled meta-analysis — a genuinely consistent pattern across multiple independent analyses, yet no single trial has been specifically, adequately powered with this shock-specific benefit as its primary, pre-specified hypothesis. This protocol treats this as a case of substantial, converging circumstantial evidence that has nonetheless not yet crossed into a definitively confirmed indication — a distinct epistemic category from either a settled question (like the general-population equivalence finding) or genuine, unresolved equipoise with no consistent signal in either direction.
• ALBIOS's open-label design and its narrower research question (replacement of hypoalbuminemia in already-resuscitated patients, not fluid choice during active resuscitation) are genuine, distinct limitations that should temper how directly its findings are extrapolated to the broader "should albumin be used as a resuscitation fluid in septic shock" question that popular discussion of the trial sometimes implies it answered — the trial's own design was more specific than this broader framing suggests.
• The severity-dependent effect-modification pattern (benefit in shock, possible harm without shock) deserves particular attention as a genuinely interesting, biologically plausible finding rather than dismissal as noise — it suggests that if a true albumin benefit exists in sepsis, it may be specifically tied to the degree of capillary leak/endothelial dysfunction present in overt shock, a mechanistically coherent hypothesis that has not yet been definitively tested as a primary research question in its own right.

24. References

1. Finfer S, Bellomo R, Boyce N, et al; SAFE Study Investigators. A comparison of albumin and saline for fluid resuscitation in the intensive care unit. N Engl J Med. 2004;350(22):2247-2256.
2. Cook DJ. Is albumin safe? [editorial]. N Engl J Med. 2004;350(22):2294-2296.
3. SAFE Study Investigators; Finfer S, McEvoy S, Bellomo R, et al. Impact of albumin compared to saline on organ function and mortality of patients with severe sepsis. Intensive Care Med. 2011;37(1):86-96.
4. Cochrane Injuries Group Albumin Reviewers. Human albumin administration in critically ill patients: systematic review of randomised controlled trials. BMJ. 1998;317(7153):235-240.
5. Roberts I, Blackhall K, Alderson P, Bunn F, Schierhout G. Human albumin solution for resuscitation and volume expansion in critically ill patients (Cochrane review). Cochrane Database Syst Rev. 2013.
6. Caironi P, Tognoni G, Masson S, et al; ALBIOS Study Investigators. Albumin replacement in patients with severe sepsis or septic shock. N Engl J Med. 2014;370(15):1412-1421.
7. Patel A, Laffan MA, Waheed U, Brett SJ. Randomised trials of human albumin for adults with sepsis: systematic review and meta-analysis with trial sequential analysis of all-cause mortality. BMJ. 2014;349:g4561.
8. Dubois MJ, Orellana-Jimenez C, Melot C, et al. Albumin administration improves organ function in critically ill hypoalbuminemic patients: a prospective, randomized, controlled, pilot study. Crit Care Med. 2006;34(10):2536-2540.
9. The Washington Manual of Critical Care, 4th ed. 2025 — fluid therapy chapter.
10. ICU Protocols: A Step-wise Approach, 2nd ed. — fluid management content.