Platelet Transfusion before CVC Placement in Patients with Thrombocytopenia

van Baarle. NEJM 2023. doi: 10.1056/NEJMoa2214322

Clinical Question

• In haematology or ICU patients with platelet counts between 10,000 and 50,000 per mm³, is receiving no prophylactic platelets non inferior compared to the administration of 1u platelets in order to reduce the risk of catheter-related bleeding?

Background

• CVC placement and thrombocytopenia are both common among haematology and ICU patients
• Retrospective studies have suggested that with USS guidance CVC insertion is safe even with a platelet count < 20,000 per mm³
• Transfusion guidelines regarding platelet-count thresholds before the placement of a CVC offer conflicting recommendations regarding transfusion threshold:
  • <20,000: British Society of Hematology 
  • <40 – 5000: American Society of Oncology
  • <50,000: AAGBI for “other major surgery or invasive procedures”,
• Platelet transfusion is not without associated morbidity, alongside being a scarce and expensive resource
  • Various adverse effects include TRALI, TACO, allergic reaction and alloimmunization

Design

• Randomised, open label, non-inferiority trial
• Non-inferiority margin set as 2.5% increase in the absolute risk of Grade 2-4 bleeding
  • i.e. if rate of bleeding was < 2.5% greater in no transfusion arm then this would be deemed non-inferior
• Occurrence of bleeding and any treatments recorded immediately post CVC placement and at 1- and 24-hours post insertion
• 1:1 randomization to transfusion vs no transfusion
  • Stratified according to the trial centre and catheter type (large-bore dialysis catheter or regular CVC)
• Unblinded for patient but if possible the operator was unaware of the trial-group assignment
• Ideally informed consent taken prior to randomisation, but if emergency (ie. in ICU) then deferred consent allowed
• Sample size calculation required 196 CVC insertions in each group:
  • Based on assumption that in transfusion group 1% will have Grade 2 bleeding and no patients will have Grade 3 or 4 bleeding
  • 80% power to detect the non-inferiority of the non-transfusion strategy with an alpha level of 0.05
  • No loss to follow-up accounted for

Setting

• 10 hospitals in the Netherlands (7 academic and 3 general)
  • Included 8 hematology wards and 7 ICUs (Figure S2a)
• February 2016 – March 2022

Population

• Inclusion:
  • Patients who had a platelet count of 10,000 to 50,000 per mm³
  • within 24 hours before the procedure
• Exclusion:
  • Therapeutic anticoagulation
  • A history of congenital or acquired coagulation factor deficiency or bleeding risk
  • INR>1.5
    • Changed to INR > 3.0 during study due to new evidence
  • Prior inclusion within 24 hours (the authors allowed multiple inclusions of patients who had been already included in the study, as long as each procedure was performed 24h apart.)
• Baseline Characteristics 
  • 411 CVC placements randomized –> 393 CVC placements (358 patients) were included in ITT analysis
    • 18 excluded due to withdrawal or no deferred consent
  • No major difference in baseline characteristics:
    • Comparing transfusion vs no transfusion:
    • Age: 58 vs 59
    • Female: 34 vs 38%
    • Median platelet count: 30,000 vs 30,000
    • Median INR: 1.1 vs 1.1
    • Median APTT: 29 vs 31
    • ICU: 43 vs 44%
    • Regular CVC (as opposed to dialysis catheter): 82 vs 84%
    • Tunneled: 11 vs 10%
    • IJ: 50 vs 50%
    • Subclavian 38 vs 38%
    • Prior platelet transfusion < 6 hrs before randomisation: 9 vs 10%

Intervention

• Transfusion 1u platelets prophylactically

Control

• No transfusion

Management common to both groups

• Ultrasound guided CVC
• Operator needed to have performed at least 50 CVC placements
• If possible, all catheters were placed approximately 1h after randomization
• Otherwise, CVCs were placed according to local clinical practice

Outcome

• Primary outcome:
  • Grade 2-4 catheter-related bleeding:
    • Transfusion 4.8% (transfusion) vs 11.9% (no transfusion)
    • Absolute risk reduction of 7.1% (90% CI: 1.3 to 17.8%), RR 2.45 (90% CI 1.27 – 4.70)
• Selected secondary outcomes:
  • Comparing transfusion vs no transfusion
  • Significant difference in transfusion group in:
    • Platelet count at 1h and 24h after CVC placement
      • 54,000 vs 26,000 after 1h; 36,000 vs 26,000 after 24h
    • Median ICU LOS
      • 9 vs 7 days
  • No significant difference in:
    • The risk of grade 3-4 catheter-related bleeding (2.1% vs 4.9%)
    • Hematoma occurrence (12.2 % vs 18.9%)
    • Rate of red-cell transfusion in ≤24 hr (0.48 vs 0.49)
    • Allergic transfusion reaction (1 vs 0.5%)
    • ICU mortality (57 vs 52%)
    • Hospital mortality (28 vs 32%)
• Subgroup and Other Outcomes
  • Exploratory subgroup analysis favoured transfusion group if patient had:
    • Non-tunneled catheter
    • Subclavian vein
    • Haematology ward
  • ICU patients has similar bleeding outcomes regardless of transfusion (4 vs 5%)
  • Non-transfusion group received more platelet transfusions in the 24h after CVC placement
    • Rate of platelet transfusion in 24 hours after CVC placement by pre-CVC platelet count (transfusion vs no-transfusion, per-protocol analysis)
      • Platelet count 10-19: 0.19 vs 0.67
      • Platelet count 20-29: 0.12 vs 0.58
      • Platelet count 30-39: 0.11 vs 0.35
      • Platelet count 40-50: 0.15 vs 0.16
  • Overall costs related to transfusion and bleeding events were higher in the transfusion group (total cost difference $410, 95% CI 285 – 545)

Authors’ Conclusions

• In patients with severe thrombocytopenia, withholding prophylactic platelet transfusion before CVC placement in those with a platelet count of 10,000 to 50,000 mm³ did not meet the predefined margin for non-inferiority, and resulted in more CVC-related bleeding than prophylactic platelet transfusion

Strengths

• Multi-centre RCT
• Trial protocol and statistical analysis plans were published prior to publication
• ITT analysis
• Population is generalizable to our daily ICU patients in terms of background comorbidities and acute physiology
• Pragmatic and included a wide range of CVC insertions:
  • Both tunneled and non-tunneled catheters
  • Location of insertion
  • Dialysis catheter vs standard catheters
• Inclusion of cost analysis
• Low rates of protocol violations

Weaknesses

• Single country may limit external validity
• Given operators aware that this trial was about rates of bleeding complications could lead to Hawthorne effect when comparing these bleeding rates to everyday practice
• Patient heterogeneity:
  • 1 unit of platelets may not have been sufficient in all patients
  • Is the bleeding risk the same in a general ICU patient compared to haematology patient?
  • Bleeding risk in this study appears to be higher in hematology subgroup compared to ICU
• Potential for confounders – Figure S2b suggests that a large proportion of bleeding in no transfusion group occurred in 2 haematology centres (14/22 of all Grade 2 – 4 bleeding episodes).
• Missing data:
  • 7% of the population did not have the primary outcome data available
• Not fully blinded:
  • ~25% of insertions were unblinded to operator (Figure S7); however, analysis showed no increased rates of bleeding in unblinded group

The Bottom Line

• Alongside paying meticulous attention to insertion technique and the high rates of post CVC platelet transfusion in the no transfusion group, I will prophylactically transfuse if the platelet count is < 50,000 per mm³

External Links

• article Platelet Transfusion before CVC Placement in Patients with Thrombocytopenia

Metadata

Summary author: Ryo Ueno
Summary date: 14th July 2023
Peer-review editor: George Walker

Picture by: Thavis3D /Unsplash