Publication

• Title: Albumin Infusions in Hospitalized Patients with Cirrhosis
• Acronym: ATTIRE
• Year: 2021
• Journal published in: The New England Journal of Medicine
• Citation: China L, Freemantle N, Forrest E, Kallis Y, Ryder SD, Wright G, et al; ATTIRE Trial Investigators. Albumin Infusions in Hospitalized Patients with Cirrhosis. N Engl J Med. 2021;384(9):808-817.

Context & Rationale

• Background

  • Hospitalised decompensated cirrhosis carries high short-term risks of infection, renal dysfunction, ICU admission, and death.
  • Hypoalbuminaemia is common in cirrhosis and associates with immune dysfunction, circulatory derangements, and adverse outcomes.
  • Albumin has oncotic and “non-oncotic” functions (endotoxin and ligand binding, antioxidant capacity, immunomodulation), creating a plausible mechanistic link between low albumin and infection/organ failure risk.
  • Albumin is established for specific indications in cirrhosis (eg spontaneous bacterial peritonitis, large-volume paracentesis, hepatorenal syndrome), but the benefit of “targeted correction” of serum albumin in unselected hospitalised decompensation was uncertain.
  • Prior trials of longer-term albumin in cirrhosis (eg outpatient ascites programmes) suggested benefit in some contexts and no benefit in others, leaving a major evidence gap for inpatient practice.
• Research Question/Hypothesis

  • Does targeted daily 20% human albumin solution to raise/maintain serum albumin at ≥30 g/L in hospitalised patients with decompensated cirrhosis reduce infection, kidney dysfunction, or death compared with standard care?
• Why This Matters

  • Albumin is costly, widely used, and biologically active; clarifying whether routine inpatient “normalisation” improves clinically meaningful outcomes is central to value-based care.
  • If effective, this strategy could offer a scalable, ward-based intervention to reduce complications that drive ICU utilisation and mortality in cirrhosis.
  • If ineffective or harmful, the trial would prevent practice drift and refocus albumin use on proven indications while stimulating alternative mechanistic targets.

Design & Methods

• Research Question: In hospitalised adults with decompensated cirrhosis and serum albumin <30 g/L, does targeted daily 20% albumin infusion to achieve/maintain serum albumin ≥30 g/L reduce a composite of infection, kidney dysfunction, or death (days 3–15) versus standard care?
• Study Type: Multicentre, parallel-group, investigator-initiated, pragmatic randomised controlled trial in 35 NHS hospitals (England, Scotland, Wales); open-label; allocation via central system with minimisation.
• Population:
  • Adults (≥18 years) with diagnosed cirrhosis admitted with decompensation (eg ascites, encephalopathy, variceal bleeding, jaundice) requiring in-hospital treatment.
  • Serum albumin <30 g/L measured within 72 hours of admission.
  • Expected hospital stay ≥5 days (excluding social/administrative reasons).
  • Key exclusions included advanced hepatocellular carcinoma with life expectancy <8 weeks or receiving palliative care only; pregnancy or lactation; severe cardiac dysfunction; or other investigator-defined unsuitability.
• Intervention:
  • Targeted 20% human albumin solution (HAS) with daily dosing based on current serum albumin, infused at 100 mL/hour, continued until day 14 or hospital discharge.
  • Algorithm (daily): serum albumin <20 g/L → 400 mL; 20–25 g/L → 300 mL; 26–29 g/L → 200 mL; 30–34 g/L → 100 mL; ≥35 g/L → 0 mL.
  • Standard clinical indications for albumin (eg paracentesis, SBP, hepatorenal syndrome) could still be managed as clinically required.
• Comparison:
  • Standard care as delivered by the treating team.
  • Albumin permitted for usual clinical indications at clinician discretion, without protocolised “target-to-level” supplementation.
• Blinding: Unblinded treatment allocation; outcomes included objective components (creatinine thresholds, mortality) alongside clinician-diagnosed infections; selected safety events underwent structured review processes.
• Statistics: A total of 778 patients were required to detect a 30% relative reduction in the primary composite endpoint (assumed 30% in standard care reduced to 21% with targeted albumin) with 80% power at the 5% significance level; the protocol initially allowed for 10% withdrawal (target 866), but enrolment completed after 828 randomisations given lower-than-expected withdrawal. Primary analysis was intention-to-treat using a mixed-effects logistic regression framework adjusted for stratification variables, with prespecified sensitivity analyses including re-randomisations and alternative endpoint windows.
• Follow-Up Period: Primary endpoint assessed during days 3–15 after randomisation; mortality followed to 6 months.

Key Results

This trial was not stopped early. Recruitment ended after achieving the required information size with lower-than-expected withdrawal (828 randomisations; 777 unique patients in the primary analysis).

Outcome	Targeted albumin	Standard care	Effect	p value / 95% CI	Notes
Primary composite (infection, kidney dysfunction, or death; days 3–15)	113/380 (29.7%)	120/397 (30.2%)	Adjusted OR 0.98	95% CI 0.71 to 1.33; P=0.87	Primary endpoint window begins day 3 (to allow time to raise albumin); supportive analyses with alternative windows were consistent.
Infection component (days 3–15)	79/380 (20.8%)	71/397 (17.9%)	Adjusted OR 1.22	95% CI 0.85 to 1.75; P=0.28	Direction favoured standard care (numerically more infections with targeted albumin); not statistically significant.
Kidney dysfunction component (days 3–15)	40/380 (10.5%)	57/397 (14.4%)	Adjusted OR 0.68	95% CI 0.44 to 1.11; P=0.12	Numerically fewer kidney dysfunction events with targeted albumin; CI included no effect.
Death component (days 3–15)	30/380 (7.9%)	33/397 (8.3%)	Adjusted OR 0.95	95% CI 0.56 to 1.59; P=0.85	Deaths before day 3 were not counted as primary endpoint events by design.
Total albumin administered during treatment period	Median 200 g (IQR 140–280)	Median 20 g (IQR 0–120)	Adjusted mean difference 143 g	95% CI 127 to 158; P<0.001	Demonstrates strong biological separation between groups.
Death at 28 days	53/380 (14.0%)	62/397 (15.6%)	Adjusted OR 0.86	95% CI 0.57 to 1.30; P=0.47	No early survival benefit; direction favoured targeted albumin but imprecise.
Death at 3 months	92/380 (24.2%)	93/397 (23.4%)	Adjusted OR 1.05	95% CI 0.74 to 1.48; P=0.80	No medium-term survival signal.
Death at 6 months	132/380 (34.7%)	119/397 (30.0%)	Adjusted OR 1.27	95% CI 0.93 to 1.73; P=0.13	Numerically higher mortality with targeted albumin; CI included no effect.
Any pulmonary oedema or fluid overload (serious adverse events)	23 events	8 events	Not reported	Not reported	Safety signal for cardiopulmonary intolerance to high-dose albumin/volume load.

• Targeted albumin achieved large separation in administered albumin (median 200 g vs 20 g) but did not improve the primary composite endpoint (Adjusted OR 0.98; 95% CI 0.71 to 1.33).
• Component directions were discordant (numerically fewer kidney dysfunction events but numerically more infections), reducing plausibility of a unidirectional benefit through a single pathway.
• There was an important safety signal for pulmonary oedema/fluid overload serious adverse events (23 vs 8 events), emphasising the potential harm of “normalisation” strategies in cirrhosis.

Internal Validity

• Randomisation and allocation: Central allocation with minimisation (site and key baseline factors), supporting allocation concealment at enrolment and reducing selection bias.
• Drop out / exclusions: Of 829 randomisations, 1 withdrew permission for data usage; primary analysis was conducted in 777 unique patients (380 vs 397) with high completeness for key endpoints and mortality follow-up.
• Performance / detection bias: Open-label delivery increases risk of differential co-interventions and diagnostic suspicion (particularly for “infection”); objective components (creatinine thresholds, mortality) mitigate but do not eliminate this concern.
• Protocol adherence: A meaningful proportion did not receive allocated therapy (eg short admissions, early deterioration, discharge), but this reflects pragmatic inpatient practice and was handled via intention-to-treat.
• Baseline characteristics: Well balanced between arms; mean MELD 19.6 vs 19.5; mean baseline serum albumin 24.7±3.0 vs 24.7±2.9 g/L; baseline infection at randomisation 25.8% vs 28.5%.
• Timing: Serum albumin eligibility required measurement within 72 hours of admission; the primary endpoint window began on day 3 to allow time for albumin correction, which is coherent with the biological target but can de-emphasise very early events.
• Dose: High exposure in the intervention arm (median 200 g) with daily protocolised dosing by serum albumin category; plausibly sufficient to test the target but also plausibly excessive in those with reduced cardiac reserve.
• Separation of the variable of interest:
  • Total albumin: median 200 g (IQR 140–280) vs 20 g (IQR 0–120); adjusted mean difference 143 g (95% CI 127 to 158).
  • Daily median albumin per patient (selected days): day 2 = 60 g (IQR 40–60) vs 0 g (IQR 0–0); day 3 = 40 g (IQR 20–40) vs 0 g (IQR 0–0); days 4–12 = 20 g (IQR 0–20) vs 0 g (IQR 0–0).
• Heterogeneity: Multicentre delivery improves representativeness but introduces site-level practice variation; prespecified subgroup analyses showed no credible effect modification across common clinical strata (eg baseline albumin category, MELD strata, antibiotic exposure).
• Crossover: Albumin use in standard care occurred (median 20 g), consistent with permitted guideline indications; this could dilute effects but the between-group separation remained large.
• Outcome assessment: Composite endpoint incorporated infection (partly clinician-diagnosed) plus objective kidney dysfunction and death; clear prespecified definitions reduce ambiguity, but composite construction can mask diverging component effects.
• Statistical rigour: Prespecified information size achieved; primary effect estimate near null with tight CI excluding large benefit; sensitivity analyses (including alternative endpoint windows and inclusion of re-randomisations) were directionally consistent with the primary result.

Conclusion on Internal Validity: Overall, internal validity appears moderate-to-strong given robust randomisation, large between-group biological separation, and consistent sensitivity analyses; key limitations are the open-label design and the day 3–15 primary endpoint window, which may underweight very early events and introduces potential detection bias for infection outcomes.

External Validity

• Population representativeness: Hospitalised, acutely decompensated cirrhosis with hypoalbuminaemia is common internationally; the UK NHS setting and broad inclusion of decompensation syndromes support generalisability to similar inpatient services.
• Important exclusions: Patients with very short expected stays (<5 days), advanced HCC with limited life expectancy, and those with severe cardiac dysfunction were excluded; applicability to these groups is limited.
• Healthcare system applicability: Findings likely generalise to centres where albumin is available and already used for standard cirrhosis indications; in resource-limited settings where albumin is scarce or unaffordable, ATTIRE supports avoiding routine “target-to-level” albumin strategies.
• Intervention feasibility: Daily albumin measurements and protocolised dosing are feasible in high-resource ward environments; operational scalability may be lower in overstretched services.

Conclusion on External Validity: External validity is good for typical high-risk inpatient decompensated cirrhosis populations in systems with access to albumin, but generalisability is reduced for patients with brief admissions, advanced malignancy, or significant cardiac dysfunction, and for low-resource hospitals.

Strengths & Limitations

• Strengths: Large multicentre pragmatic design; clinically meaningful composite endpoint; substantial intervention–control biological separation (albumin dose); prespecified sensitivity analyses including alternative endpoint windows and inclusion of re-randomisations; robust mortality follow-up to 6 months.
• Limitations: Open-label treatment (risk of co-intervention and diagnostic bias for infection); primary endpoint window (days 3–15) excludes very early events by design; composite endpoint combines outcomes with different mechanistic pathways and directions; clinically relevant harm signal (pulmonary oedema/fluid overload) complicates risk–benefit; protocol intensity may limit feasibility in lower-resource settings.

Interpretation & Why It Matters

• Clinical practice

  • Routine inpatient “target-to-level” albumin infusions to raise serum albumin to ≥30 g/L should not be adopted as a default strategy for decompensated cirrhosis, given no improvement in infection/kidney dysfunction/death and evidence of harm.
  • Albumin use should remain focused on established indications (eg SBP, large-volume paracentesis, hepatorenal syndrome) rather than biochemical normalisation.
• Mechanistic inference

  • Correction of serum albumin concentration alone (despite large dose separation) did not translate into reduced infection or organ dysfunction, suggesting that hypoalbuminaemia in acute decompensation is not a sufficient causal lever in this context.
• Research implications

  • Future work should prioritise phenotyping (eg cardiac reserve, haemodynamic profile, inflammatory endotypes) to identify who is harmed by albumin loading and whether any subpopulation might benefit from different targets, preparations, or co-interventions.

Controversies & Subsequent Evidence

• Despite achieving the biochemical target, there was no improvement in clinically meaningful outcomes and there was increased pulmonary oedema/fluid overload; the accompanying editorial interpreted this as evidence that aggressive albumin correction in advanced cirrhosis can be “friend and foe”, and emphasised cardiopulmonary risk with high-volume oncotic therapy.¹
• Correspondence raised concern that cirrhotic cardiomyopathy (and related haemodynamic vulnerability) may predispose to pulmonary oedema with high-dose albumin, and queried whether variceal bleeding presentations and beta-blocker management could influence safety signals; the trialists replied that beta-blocker use was similar between groups and highlighted that a substantial proportion of gastrointestinal bleeding serious adverse events occurred in participants admitted with variceal bleeding in the albumin arm.²
• Methodological debate centres on the prespecified day 3–15 primary endpoint window (chosen to allow time for albumin correction): this can under-represent very early events, but supportive analyses using earlier windows were directionally consistent with the primary finding.

Summary

• ATTIRE tested whether protocolised 20% albumin infusions targeting serum albumin ≥30 g/L improve outcomes in hospitalised decompensated cirrhosis with hypoalbuminaemia.
• The intervention created large biological separation (median total albumin 200 g vs 20 g) but did not reduce infection, kidney dysfunction, or death between days 3–15 (Adjusted OR 0.98; 95% CI 0.71 to 1.33).
• Mortality at 28 days and 3 months was similar; 6-month mortality was numerically higher with targeted albumin (Adjusted OR 1.27; 95% CI 0.93 to 1.73).
• Targeted albumin increased serious cardiopulmonary complications (pulmonary oedema/fluid overload 23 vs 8 serious adverse events).
• Findings argue against routine inpatient “normalisation” of albumin and reinforce restricting albumin to proven indications in cirrhosis.

Further Reading

Other Trials

• 1999Sort P, Navasa M, Arroyo V, et al. Effect of intravenous albumin on renal impairment and mortality in patients with cirrhosis and spontaneous bacterial peritonitis. N Engl J Med. 1999;341(6):403-409.
• 2004Finfer S, Bellomo R, Boyce N, French J, Myburgh J, Norton R; 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.
• 2018China L, Freemantle N, Forrest E, et al. A randomized trial of albumin infusions in hospitalized patients with cirrhosis. Clin Gastroenterol Hepatol. 2018;16(5):753-761.e3.
• 2018Caraceni P, Riggio O, Angeli P, et al. Long-term albumin administration in decompensated cirrhosis (ANSWER): an open-label randomised trial. Lancet. 2018;391(10138):2417-2429.
• 2018Solà E, Solé C, Simón-Talero M, et al. Midodrine and albumin for prevention of complications in patients with cirrhosis awaiting liver transplantation: a randomised placebo-controlled trial. J Hepatol. 2018;69(6):1250-1259.

Systematic Review & Meta Analysis

• 2006D'Amico G, Garcia-Tsao G, Pagliaro L. Natural history and prognostic indicators of survival in cirrhosis: a systematic review of 118 studies. J Hepatol. 2006;44:217-231. (DOI not reported in the provided source files.)
• 2019Best LM, Freeman SC, Sutton AJ, et al. Treatment for hepatorenal syndrome in people with decompensated liver cirrhosis: a network meta-analysis. Cochrane Database Syst Rev. 2019;9:CD013103. (DOI not reported in the provided source files.)
• 2019Simonetti RG, Perricone G, Nikolova D, Bjelakovic G, Gluud C. Plasma expanders for people with cirrhosis and large ascites treated with abdominal paracentesis. Cochrane Database Syst Rev. 2019;6:CD004039. (DOI not reported in the provided source files.)
• 2019Rochwerg B, Alhazzani W, Gibson A, et al. Fluid type for resuscitation in sepsis: a systematic review. (Not reported in the provided source files.)

Observational Studies

• 2010Arvaniti V, D'Amico G, Fede G, et al. Infections in patients with cirrhosis increase mortality fourfold and should be used in determining prognosis. Gastroenterology. 2010;139:1246-1256. (DOI not reported in the provided source files.)
• 2011Tandon P, Garcia-Tsao G. Renal dysfunction is the most important independent predictor of mortality in cirrhotic patients with spontaneous bacterial peritonitis. Clin Gastroenterol Hepatol. 2011;9:260-265. (DOI not reported in the provided source files.)
• 2014Rahimi RS, Rockey DC. Complications of cirrhosis. Curr Opin Gastroenterol. 2014;30:223-229. (DOI not reported in the provided source files.)
• 2016Arroyo V, Moreau R, Kamath PS, et al. Acute-on-chronic liver failure in cirrhosis. Nat Rev Dis Primers. 2016;2:16041. (DOI not reported in the provided source files.)

Guidelines

• 2018European Association for the Study of the Liver. EASL clinical practice guidelines for the management of patients with decompensated cirrhosis. J Hepatol. 2018;69:406-460. (DOI not reported in the provided source files.)
• 2013Runyon BA. Introduction to the revised American Association for the Study of Liver Diseases guidelines on management of adult patients with ascites due to cirrhosis. Hepatology. 2013;57:1651-1653. (DOI not reported in the provided source files.)

Notes

• Several high-value systematic reviews, observational cohorts, and guidelines cited in the trial’s reference list did not include DOI information in the provided source PDFs; entries above are included for scholarly completeness without adding unverified DOI links.

Overall Takeaway

ATTIRE is landmark because it decisively tested a widely used, biologically plausible, resource-intensive inpatient strategy—protocolised albumin “normalisation”—and found no improvement in infection, kidney dysfunction, or death despite strong treatment separation. The trial also identified a clinically important harm signal (pulmonary oedema/fluid overload), shifting practice towards restricting albumin to evidence-based indications rather than targeting serum concentrations in unselected hospitalised decompensated cirrhosis.

Bibliography

• 1Garcia-Tsao G. Terlipressin and intravenous albumin in advanced cirrhosis—friend and foe. N Engl J Med. 2021;384(9):869-871.
• 2Elfeki MA, Khan AA, Alizadeh M, et al. Albumin Infusions in Hospitalized Patients with Cirrhosis. N Engl J Med. 2021;384(23):e90.