Atan

With apologies for the initial misspelling of Rafidah Atan’s name in our feature image.

A Double-Blind Randomised Controlled Trial of High Cutoff Versus Standard Haemofiltration in Critically Ill Patients With Acute Kidney Injury

R Atan. Critical Care Medicine. 2018;epublished August 1. doi:10.1097/CCM.0000000000003350

Clinical Question

  • In critically ill patients on noradrenaline for haemodynamic support, who required hemofiltration for AKI, does the use of a high cutoff hemofilter compared with a standard hemofilter impact the median hours of noradrenaline-free time at day 7?

Background

  • In patients receiving CRRT, high cutoff dialysis filters have been promoted as having enhanced middle molecule/inflammatory mediator clearance due to their higher permeability. It is not known if patients with vasoplegic states have a clinically meaningful benefit from such practice.
  • Several small trials have examined this question. An unblinded pilot study of 30 patients showed the group on the HCO filter had a significant decline in ‘adjusted’ noradrenaline dose over time.
  • Other studies have examined the optimal dialysis dose in patients with septic shock. For example, the IVOIRE trial looked at high-volume exchanges (70ml/kg/hr vs 35ml/kg/hr) and did not improve outcomes.
  • Removal of cytokines may be a potentially appealing concept in vasoplegic patients. However, with all appealing concepts the complexity is likely incompletely understood and it is vital that such therapies are tested in well conducted randomised trials before being implemented. The last thing we want to do is cause harm to our patients.

Design

  • 1:1 randomisation within 12 hours of a decision to start hemofiltration
  • Allocation concealment by computer generated program using permuted block sizes and opaque envelopes
  • Patients, healthcare workers and researchers were blinded to treatment allocation
    • The 2 different filters were indistinguishable in appearance
  • 76 patients in total carried an 80% power of detecting a 25% difference in noradrenaline-free time in the first week with an alpha of 0.05.

Setting

  • Single centre tertiary ICU in Australia

Population

  • Inclusion:
    • Patients on noradrenaline for haemodynamic support
    • Patients requiring CRRT for AKI (oliguria <100ml/6hr, unresponsive to fluid bolus, K+ >6.5mmol/l, acidemic  pH<7.2, urea > 25mmol/L, Cr>300umol/L or significant organ oedema).
  • Exclusion:
    • <18-years-old
    • death imminent
    • previous dialysis on this admission or chronic dialysis
    • pregnant/breast-feeding
  • 76 patients were randomised (38 per group)
    • 2 patients were subsequently excluded from analysis (one had an exclusion criteria and one died before treatment started)
    • This left 74 patients in the modified intention to treat analysis
  • Patients were well matched for age, sex, severity of illness (APACHE II, APACHE III and SOFA scores)
    • There were some imbalance in the variables lactate, INR and oliguria

Intervention

  • Continuous Veno-venous Haemofiltration with high cutoff filter (CVVH-HCO)
    • cut-off point of 100kDa

Control

  • Continuous Veno-venous Haemofiltration with standard filter (CVVH-std)
    • cut-off point 30kDa

Management common to both groups

  • Dialysis dose = 25ml/kg/h
  • Blood flow 200ml/min
  • Bicarbonate buffered replacement fluids
  • Anticoagulation at discretion of treating intensivist, but usually low dose pre-filter heparin
  • Maximum duration of therapy = 14 days
  • Dialysis continued until not needed, ICU discharge or death

Outcome

  • Primary outcome: median cumulative hours of noradrenaline-free time within the first week after randomisation
    • If the patient died whilst on noradrenaline, the days after death contributed to zero noradrenaline-free hours
    • CVVH-HCO: 32 hours noradrenaline-free
    • CVVH-std: 56 hours noradrenaline-free
    • P = 0.052
  • Secondary outcome: CVVH-HCO vs CVVH-std
    • Maximum noradrenaline rates: similar
    • Albumin levels: no difference
    • Time to cease dialysis: no difference
    • Unadjusted ICU mortality OR 2.17 (95% CI 0.84-5.58 p=0.109)
    • Unadjusted hospital mortality OR 2.4 (95% CI 0.94-6.15 p=0.067)
    • Adjusted for lactate, INR, serum albumin and APACHE 3
      • Adjusted ICU mortality OR 2.13 (95% CI 0.69-6.65 p=0.191)
      • Adjusted hospital mortality OR 2.49 (95% CI 0.81-7.66 p=0.112)

Authors’ Conclusions

  • In patients with severe acute kidney injury and vasoplegic shock, use of a high cutoff dialysis filter compared with a standard filter, did not result in more noradrenaline-free time and did not improve other important secondary outcomes

Strengths

  • Allocation concealment and permuted block sizes (minimising selection bias)
  • Modified intention to treat analysis, complete follow-up (minimising attrition bias)
  • Double-blinded (reducing detection bias)
  • Baseline imbalances were minor and were adjusted for in the secondary outcomes (reducing selection bias)
  • Other than the intervention applied, patients were managed in a pragmatic way at the clinicians discretion (minimising performance bias)
  • The primary outcome of “noradrenaline-free time within the first week after randomisation” is a robust surrogate and ameliorates the competing risk of mortality impacting on noradrenaline dose
  • The sample may be suitable for extrapolation to vasoplegic patients in general (with AKI and shock) as the sample was somewhat heterogenous (mainly septic shock, cardiogenic shock and liver failure)

Weaknesses

  • The primary outcome is a surrogate outcome and may not be clinically meaningful
  • Some of the secondary outcomes are clinically meaningful but it is difficult to draw too many conclusions from these as the study is not powered to detect a difference in these events, eg mortality
  • There were some baseline imbalances reflecting the small sample size
    • These were adjusted for but the analysis may not completely ameliorate this bias
  • There was no economic assessment
    • I know in the unit I work the price difference between HCO filters and standard filters is $275 vs $240, which is not a great difference in the scheme of things

The Bottom Line

  • This trial is very informative. There does not seem to be a benefit in using high cutoff filters. There is a signal toward harm.
  • Outside the setting of a larger clinical trial, I will not use high cutoff filters in my vasoplegic patients with AKI.

External Links

Metadata

Summary author: Celia Bradford
Summary date: 23 August 2018
Peer-review editor: Duncan Chambler

Leave a Reply

This site uses Akismet to reduce spam. Learn how your comment data is processed.