Critical care Topic

Hang the albumin?

Albumin is one of the most commonly given blood products, and one of the most commonly given outside the evidence. Six calls, then 5% versus 25%, sepsis, cirrhosis, long-term use, where it harms, and plasma exchange and burns.

Reviewed October 2026 · verify against current guidelines

Albumin · 1 of 10

Hang the albumin?

Make your call on each — give or don't — before reading on. The answers are on the last card.

Your call

8-liter paracentesis
Septic shock, 4 L of crystalloid in
Serum albumin 1.9 g/dL
SBP, day 1
Severe traumatic brain injury
Furosemide not working

Albumin · 2 of 10

Twenty-five grams, two ways to give it.

5% and 25% deliver the same protein and expand plasma by about the same amount. What differs is where the water comes from: 25 g as 5% (500 mL bag) gains its plasma all from the bag; 25 g as 25% (100 mL bag) recruits most of it from the interstitium.

Side by side

	5% — iso-oncotic	25% — hyperoncotic
Volume for 25 g	500 mL	100 mL
Oncotic pull	About that of plasma — stays where it is put	Several times plasma — draws water in
Sodium	130 to 160 mmol/L — same concentration, but 500 mL delivers five times the total sodium load	Same concentration; one-fifth the total sodium for the same 25 g
Rate ceiling	Up to 5 mL/min	1 to 2 mL/min
Needs	A plasma deficit to fill	Interstitial water to recruit

Borrowed water — 25% moves water, it does not add it. The gain over the infused volume comes out of the patient's own interstitium; in the edematous patient that is the point, in the dry patient there is nothing to borrow. Trials mix 4%, 5%, 20% and 25%: 4–5% behave alike, as do 20–25%. Doses are written in grams, so they transfer across products; volumes do not.

Never dilute 25% with sterile water — the result is hypotonic and has caused fatal hemolysis. Dilute with 0.9% saline or 5% dextrose if a lower concentration is needed.

Albumin · 3 of 10

Where the albumin goes.

Albumin is not confined to the vessels. It circulates through the interstitium and back, and inflammation speeds the outward leg.

Plasma albumin leaving the circulation each hour

Healthy	About 5%/h — returns by lymph; circulation, not loss.
Septic shock	Up to ~15%/h — up to 3× normal.

The numbers that set the effect

Share of oncotic pressure	About 75 to 80% of plasma colloid pressure, roughly 25 mmHg.
Body pool	About 40% intravascular, 60% interstitial.
Return path	Lymph, a full circuit about once a day.
Half-life	About 19 days — it is recycled, not consumed.

The revised Starling principle — J_v = K_f [ (P_c − P_i) − σ (π_c − π_sg) ]. π_sg replaces interstitial π_i: the fluid just beneath the glycocalyx is nearly protein-free, so plasma albumin opposes filtration more than classic Starling predicted — and interstitial albumin barely matters. Capillaries filter; they do not reabsorb. Lymph returns the fluid. Absorption occurs only briefly, after a fall in capillary pressure or a 25% infusion. Colloid holds best when capillary pressure is low.

The trial ratio

Textbook physiology predicts 1 volume of albumin replaces 3 to 5 of crystalloid. In SAFE the ratio was about 1 to 1.4.

Low serum albumin is mostly redistribution — escape rises and synthesis falls in acute illness. Infused albumin joins the same leaking pool, which is why correcting the number has not changed outcomes.

Albumin · 4 of 10

More than an oncotic agent.

Albumin carries, buffers and scavenges. Those roles change how you read labs and dose drugs — they have not yet shown that giving albumin improves outcomes.

What it does, and where you see it

	Function	At the bedside
Carrier	Binds bilirubin, fatty acids, calcium and acidic drugs	Low albumin raises the free fraction — total phenytoin and calcium mislead
Drug clearance	Unbound drug is filtered and distributed faster	Highly bound antibiotics, such as ceftriaxone and ertapenem, may run below target
Weak acid	The largest unmeasured anion in plasma	Hides a raised anion gap — see below
Antioxidant	Scavenges oxidants through a free thiol at cysteine-34	Oxidized in cirrhosis and sepsis — the lab value overstates working albumin

The anion gap it hides — measured gap of 12, plus the albumin correction, mEq/L

Albumin	Measured	Correction	Corrected
4 g/dL	12	—	12
3 g/dL	12	+2.5	14.5
2 g/dL	12	+5	17

Mechanism is not indication — add about 2.5 to the gap per 1 g/dL albumin below 4. The correction is approximate but catches lactic and ketoacidosis a normal-looking gap would miss. The pleiotropic case has been tested: trials that raised albumin to a target in sepsis and cirrhosis — ALBIOS, ARISS, ATTIRE — did not improve survival.

Measure free levels, do not infuse to fix them — when albumin is low, order free phenytoin or ionized calcium. Albumin given to normalize a total level treats the assay, not the patient.

Albumin · 5 of 10

Sepsis: safe, not superior.

Three large trials compared albumin with crystalloid in the critically ill. Each found albumin safe, and none found it saved lives.

Mortality, albumin against control

Trial	Design	Result
SAFE · 2004	4%, all ICU, 28 d	No difference
ALBIOS · 2014	20% to ≥3 g/dL, 90 d	No survival benefit
ARISS · 2026	Shock, to >3 g/dL, 90 d	No survival benefit

What the guidelines say

SSC 2026	Suggests crystalloid alone over adding albumin (moderate certainty). Albumin may be appropriate after large crystalloid volumes or in cirrhosis; avoid in TBI.
ICTMG 2024	Not suggested for first-line volume replacement or to raise serum albumin in critically ill adults.

Why the shock signal is still open — ALBIOS hinted at septic shock: in a post hoc subgroup of 1121, 90-day mortality was lower with albumin, and ARISS was built to confirm it. ARISS stopped at 440 of a planned 1412. More than half the albumin arm never reached the target. The result is neutral but underpowered, not a clean negative.

SSC and ICTMG now agree — the 2026 update moved from suggesting albumin after large volumes to suggesting against routine use. Neither supports targeting a serum level.

Albumin · 6 of 10

Cirrhosis: where it earns its place.

Three acute events share one mechanism — a vasodilated circulation losing effective volume. All three are dosed in grams of 20–25%.

Paracentesis > 5 L: 6 to 8 g per liter, whole tap

Removed	Rule range
5 L	30–40 g
6 L	36–48 g
7 L	42–56 g
8 L	48–64 g
9 L	54–72 g
10 L	60–80 g

One approach: round to the nearest 12.5 or 25 g vial to limit waste — a teaching habit, not a guideline. Evidence: prevents post-paracentesis circulatory dysfunction; ICTMG suggests (conditional, very low certainty).

Other indications

	Dose	Evidence
SBP	1.5 g/kg at diagnosis, 1 g/kg on day 3, with antibiotics	Less kidney failure and death; ICTMG suggests
AKI > stage 1A	1 g/kg a day for 2 days, max 100 g a day, diuretics stopped	EASL and AASLD; ADQI–ICA differs — below
HRS-AKI	20 to 40 g a day with terlipressin	Strong in ADQI–ICA; ICTMG made no recommendation

The open conflict: the 48-hour challenge — ADQI–ICA 2024 drops the mandatory 48-hour albumin trial before diagnosing HRS-AKI: crystalloid if volume is clearly low; albumin if status is unclear; terlipressin at diagnosis. EASL validation data push back: about 40% of albumin responders had not responded by 24 hours. A shorter challenge may label them HRS and expose them to terlipressin.

Terlipressin plus albumin fills the lungs — respiratory failure drove the excess deaths in CONFIRM. Reassess volume daily and stop albumin at the first sign of overload.

Albumin · 7 of 10

Long-term albumin: who, not whether.

Scheduled albumin in decompensated cirrhosis has one positive trial, one neutral and one negative. The populations and regimens explain more than the drug.

Primary outcome, albumin against standard care

Trial	Outcome	Effect
ANSWER · 2018	Death, 18 months	0.62
PRECIOSA · 2025	Transplant-free, 1 year	0.80
ATTIRE · 2021	Composite, in hospital	0.98

Three designs

	Who	Regimen
ANSWER	431 outpatients, uncomplicated ascites on diuretics	40 g twice weekly for 2 weeks, then weekly, up to 18 months
PRECIOSA	410 after acute decompensation, high ACLF risk	1.5 g/kg of 20% every 10 days for 12 months
ATTIRE	777 inpatients, serum albumin < 3 g/dL	Daily 20% to a target of 3.5 g/dL, up to 14 days

Guidelines split on the outpatient question — ICTMG 2024 does not suggest routine outpatient albumin, or targeting > 3 g/dL in hospital; ANSWER was open-label and has not been replicated. In parts of Europe it is standard care: PRECIOSA cut SBP and HRS and showed a 3-month survival signal, which keeps the question open for selected patients.

Targeting a level in hospital harms — ATTIRE gave far more albumin to the target arm and saw more pulmonary edema and fluid overload, with no fewer infections, kidney injuries or deaths.

Albumin · 8 of 10

Where it harms, and where it does nothing.

Brain injury carries a clear mortality signal. Cardiac surgery trials found no benefit, with signals of bleeding and kidney injury. Several common uses have no evidence of benefit.

Harm signals in randomized trials

	Albumin	Control
SAFE-TBI · 2007 — death, 24 months	33.2%	20.4%
ALBICS · 2022 — major adverse events (difference not significant)	37.1%	33.8%
ALBICS-AKI · 2025 — AKI after surgery	48.9%	43.4%

Common uses without benefit

Low serum albumin	A marker of illness; raising it has not changed outcomes.
With furosemide	Small, inconsistent trials; low albumin itself has not been shown to blunt diuresis.
Dialysis hypotension	Not suggested for prevention, treatment or ultrafiltration.
Non-SBP infection	In cirrhosis, no effect on death or kidney failure.

Proposed mechanisms — brain: 4% albumin is mildly hypotonic; across a broken blood–brain barrier, the protein and the water both follow into tissue, and ICP ran higher with albumin in the first week. Bypass: bleeding and kidneys; ALBICS saw more bleeding, resternotomy and infection as secondary outcomes, and ALBICS-AKI saw more AKI and transfusion with 20%. Albumin dilutes clotting factors and may impair platelet function.

Avoid albumin in brain injury — the harm in SAFE-TBI was concentrated in severe injury, GCS 3 to 8. Whether 20–25% behaves differently has not been tested.

Albumin · 9 of 10

Plasma exchange and burns.

Two settings outside cirrhosis where guidelines support albumin. In one it replaces plasma removed; in the other it limits crystalloid. Original plasma left after exchange: about 37% after 1 plasma volume, 22% after 1.5 (1 volume ≈ 40 mL/kg).

Side by side

	Plasma exchange	Burns ≥ 20% TBSA
Product	5%, alone or with saline	5% most often
Dose	1 to 1.5 plasma volumes, about 40 mL/kg each	Adjunct in the first 24 hours; rescue when crystalloid keeps rising
Gain	Iso-oncotic, no infection or TRALI risk of plasma	Lower total volume, better urine output
Source	ASFA practice standard	ABA 2024; ICTMG excluded burns

What albumin does not replace — exchange strips clotting factors and IgG. Daily albumin exchanges lower fibrinogen; check it before procedures, and use plasma instead in TTP or active bleeding. In burns the evidence is volume, not survival: ABA found albumin lowers resuscitation volume and raises urine output; its effect on edema complications and death is uncertain.

Smaller signals — ovarian hyperstimulation, refractory nephrotic edema and dilutional hyponatremia in cirrhosis have small or observational support only. Treat them as case-by-case, not indications.

Albumin · 10 of 10

Choosing 5% or 25%.

Settle the indication first. The product follows from whether you need to fill the plasma or pull water into it.

Supported uses, by product

5% — fill the plasma	Plasma exchange replacement. Burns ≥ 20% TBSA, first 24 hours or rescue. Sepsis: SSC 2026 suggests against routine use; may fit after large volumes or in cirrhosis.
25% — grams, little water	Paracentesis > 5 L — 6 to 8 g per liter. SBP — 1.5 g/kg, then 1 g/kg on day 3. Cirrhosis AKI — 1 g/kg a day for 2 days, max 100 g a day. HRS-AKI — 20 to 40 g a day with terlipressin.

Before you hang either

Is it on the list?	If not, crystalloid first. A low serum level is not an indication.
Brain injury?	Avoid — SAFE-TBI found higher mortality.
Lungs full?	Either product can cause pulmonary edema; 25% acts faster.
Write it in grams	25 g is 500 mL of 5% or 100 mL of 25%.

Card 1 answers

Give	8-liter paracentesis
Consider	Septic shock, 4 L in
Don't	Severe traumatic brain injury
Give	SBP, day 1
Don't	Serum albumin 1.9 g/dL
Don't	Furosemide not working

The deciding question — is the plasma short, with little tissue water to draw on? Give 5%: it adds what it contains and borrows nothing. Is the tissue waterlogged but the plasma short? Give 25%: it draws three to four times its own volume in from the interstitium.

Rate limits — up to 5 mL/min for 5%, 1 to 2 mL/min for 25%. Dilute 25% only with saline or 5% dextrose, never sterile water.

Educational reference only — not medical advice. Verify against local protocol.

Sources

Verify against current guidelines and local protocol before acting.

International Collaboration for Transfusion Medicine Guidelines (ICTMG) 2024 — albumin use.
Surviving Sepsis Campaign guidelines 2026.
SAFE (2004) — 4% albumin vs saline in the ICU.
SAFE-TBI (2007) — albumin in traumatic brain injury.
ALBIOS (2014) — 20% albumin to ≥3 g/dL in severe sepsis.
ARISS (2026) — albumin in septic shock.
ANSWER (2018) — long-term albumin in cirrhosis with ascites.
PRECIOSA (2025) — albumin after acute decompensation of cirrhosis.
ATTIRE (2021) — targeted albumin in hospitalized cirrhosis.
CONFIRM — terlipressin in hepatorenal syndrome.
ALBICS (2022); ALBICS-AKI (2025) — albumin in cardiac surgery.
ADQI–ICA 2024 consensus on AKI in cirrhosis; EASL and AASLD guidance.
ASFA practice standard for therapeutic plasma exchange.
American Burn Association (ABA) 2024 guideline.

Downloads

Every card behind this topic, as published — view full size or download.

Carousel10 slides