Liberal or Restrictive Postoperative Transfusion in Patients at High Cardiac Risk: The TOP Randomized Clinical Trial

Kougias P. JAMA 2025;334:2197-2207. doi:10.1001/jama.2025.20841

Clinical Question

• In high cardiac risk adults who develop postoperative anaemia following vascular or general surgery, does a liberal transfusion threshold (Hb <10 g/dL) compared to a restrictive transfusion threshold (Hb <7 g/dL) reduce 90‑day incidence of all‑cause death, major ischemic events or acute renal failure? 

Background

• Postoperative anaemia is common following major vascular or general surgical procedures and has been associated with increased morbidity and mortality in patients at elevated cardiac risk- likely due to increased susceptibility to the associated decline in oxygen delivery
• Current clinical evidence in stable hospitalised patient advocates for restrictive transfusion practices
• However, evidence for when to transfuse high cardiac risk postoperative patients remains unclear
• Postoperative anaemia is a common complication that may contribute to the incidence of postoperative ischemic events (MI, stroke, acute kidney injury) and their associated morbidity/mortality
• As high-cardiac risk surgical patients are at greater risk of anaemia associated complications, clarifying safe transfusion thresholds in this population is important 
• Multiple clinical trials in general medical/surgical populations have demonstrated evidence in support of restrictive strategies, however, recent trials in acute MI have suggested harm with this approach (e.g. MINT)  

Design

• Parallel, single‑blind, randomised clinical superiority trial
• Randomised at a 1:1 ratio to either liberal vs restrictive transfusion thresholds via central, telephone based system
• Primary endpoint: 90‑day composite of death or major ischemic events/renal failure. 
• Major ischaemic events consist of myocardial, coronary revascularisation or ischaemic stroke
• Enrolled participants were randomised if a postoperative inpatient haemoglobin level was detected at less the 10g/dL within 15 days of the index operation
• Patients were followed up at both 30 and 90 days after randomisation
• Single‑blind (patients blinded) 
• Subgroup analyses: age and relative cardiac risk index (RCRI) 
• Power calculation: 
  • Sample size estimation n= 1520 to detect 25% difference in primary outcome 
  • Based on 22.5% event rate in liberal group and 30% in restrictive
  • 90% power using 2-sided test with an α of 0.05
• Recruitment terminated early due to funding limitations

Setting

• 16 Veterans Affairs Medical Centers across the United States
• Enrollment dates: February 2018 to March 2023

Population

• Inclusion:
  • Adult veterans (>18 years old) 
  • Elevated cardiac risk 
  • Prior ischaemic heart disease, myocardial infarction, peripheral artery disease, stroke or transient ischaemic attack
  • Developed a post-operative anaemia with a haemoglobin of <10 g/dL
  • Had undergone a major vascular or general surgery
• Exclusion:
  • Unwilling to receive blood transfusions
  • Active involvement in another interventional trial
  • Active infection with or incomplete recovery from COVID-19
  • Declined consent
  • Active pregnancy
  • Haemodynamically unstable/in cardiogenic shock for ≥48hrs prior to randomisation
  • Prior history of hereditary anaemias, bleeding disorders or previous adverse reactions to blood products
•  Participant numbers:
  • 44,183 assessed for eligibility, 3022 enrolled
  • 1,428 randomised (largely due to not having a Hb < 10 g/dL)
  • 1,424 analysed (712 per protocol).
• Comparing baseline characteristics of intervention vs. control group:
  • Age: 69.7 vs 70.2   
  • Male sex: 96.9% vs 98.7% 
  • BMI≥30: 26.5% vs 29.9% 
  • Procedure type (vascular/general): 91.4%/8.6% vs 90.7%/9.3% 
  • RCRI: 
  • Class I: 17.8% vs 17.0% 
  • Class II: 30.3% vs 29.6% 
  • Class III: 27.4% vs 28.7% 
  • Class IV: 24.4% vs 24.7% 
  • Medical History: 
  • Peripheral artery disease: 70.8% vs 73.9% 
  • Coronary artery disease: 58.4% vs 60.1% 
  • Ischaemic stroke or TIA: 21.6% vs 23.5% 
  • Diabetes: 46.9% vs 51.1% 
  • ESKD: 7.7% vs 7.0% 
  • Medications: 
  • Aspirin or clopidogrel: 76.3% vs 75.8% 
  • Anticoagulants: 26.4% vs 28.0% 
  • Baseline Hb (on randomisation day): 9.3 vs 9.0 g/dL

Intervention

• Liberal transfusion strategy — transfuse PRBC’s when postoperative Hb <10 g/dL.
• Randomised participants assigned to liberal strategy received transfusions to maintain Hb ≥10 g/dL per protocol until 30 days after randomisation or discharge.
• Transfusion limits: 
  • Hb≥8.5g/dL, transfuse 1u PRBC
  • Hb 7.5g-8.4dL, transfuse 2u PRBC
  • Hb<7.5dL, transfuse 3u PRBC

Control

• Restrictive transfusion strategy — transfuse PRBC’s when postoperative Hb <7 g/dL. 
• Randomised participants in restrictive arm received transfusions only when Hb <7 g/dL per protocol.  
• Transfusion limits: 
  • Hb 5.5-6.9g/dL, transfuse 1u PRBC 
  • Hb<5.5g/dL, transfuse 2u PRBC  

Management common to both groups and protocol adherence

• Both patient groups received the same standard postoperative care as protocolised by their treating institution
• Hb was maintained at 10g/dL or higher in the liberal group and 7g/dL in the restrictive group until discharge or 30 days after randomisation
• Transfusion data for liberal vs restrictive groups: 
  • Median units before index surgery: 0 vs 0 (4.9 vs 7.6% received at least one)
  • Median units during index surgery: 0 vs 0 (27.3 vs 30.2% received at least one)
  • Median units after index surgery, before randomisation: 0 vs 0 (7.6 vs 4.6% received at least one)
  • Median units (IQR) per participant after randomisation : 2 (1-3) vs 0 (0-0)
    • 1 unit after randomisation: 27.8 vs 12.8%
    • 2 units after randomisation: 24.4 vs 6.2%
    • >= 3 units after randomisation: 41.3 vs 4.1%
• 114 (not transfused when Hb < 10) vs 39 protocol violations (transfused when Hb > 7) reported in liberal and restrictive groups

Outcome

• Primary outcome:
  • Primary outcome: 90-day composite of all‑cause death, myocardial infarction, coronary revascularisation, acute kidney failure, or ischemic stroke. 
  • Primary outcome: 9.1% (61/670) vs 10.1% (71/700); RR 0.90 (95% CI 0.65–1.24)
  • No difference in each individual component
  • No difference in subgroups (RCRI, Age, eFigure 2)
• Secondary outcomes:
• 90-day composite of non-MI cardiac complications (arrythmias, heart failure, non-fatal cardiac arrest): 
•  5.9% (38/647) vs 9.9% (67/678); RR 0.59 (95% CI 0.36–0.98)  
•  No differences in:
  • 90d infectious complications
  • 30d composite of all‑cause death, myocardial infarction, coronary revascularisation, acute kidney failure, or ischemic stroke
  • 1y all-cause mortality  (14.1 vs 15.0%)

Authors’ Conclusions

• After major vascular or general surgery operations among patients at high risk of a cardiac event, a liberal transfusion strategy did not reduce 90-day death or major ischemic outcome rates compared with a restrictive strategy

Strengths

• Large multi-center trial (n=1424) addressing a common and important clinical challenge  
• Clearly identified definitions that were easily measurable for primary outcome
• Robust length of patient follow-up (90-days for primary composite)
• Large between group difference in anaemia/Hb levels (mean 2.0g/dL) and blood products received when compared to previous transfusion trials
• Minimal loss to follow up
• Relatively low rates of protocol violations 

Weaknesses

• The study population was almost entirely male veterans
  • This limits the generalisability of the study to female and non-veteran populations  
• Similarly, the study only involved individuals who had undergone general and predominantly vascular surgery (90%)
•  Treating clinicians were aware of patient thresholds and may have resulted in performance bias/introduction of additional interventions
• Event rate was significantly lower than authors’ anticipated  
• No economic analysis (yet!) – an important consideration given the resource limited nature of blood products

The Bottom Line

• Compared to traditional restrictive transfusion thresholds, applying liberal transfusion threshold did not significantly reduce the 90‑day composite of death or major ischemic events 
• Incidence of non-MI cardiac complications (arrythmia’s, heart failure) were lower in the liberal transfusion strategy group, however, this study was not designed or power to draw conclusions on this outcome
• Transfusion decisions should always factor in patient symptoms, haemoglobin trend and haemodynamic status rather than strict adherence to protocolised Hb limits

External Links

• article Liberal or Restrictive Postoperative Transfusion in Patients at High Cardiac Risk
• editorial Postoperative Transfusion in Patients at High Cardiac Risk Evidence, Uncertainty, and Nuance

Metadata

Summary author: Kieran McKenna
Summary date: 1 May 2026
Peer-review editor: George Walker

Picture by: Duks / Pexels