Effect of Early vs Delayed Initiation of Renal Replacement Therapy on Mortality in Critically Ill Patients With Acute Kidney Injury

Zarbock. JAMA 2016; 315(20):2190-2199. doi:10.1001/jama.2016.5828

Clinical Question

  • In critically unwell patients with acute kidney injury, does early initiation of renal replacement therapy (RRT) compared to delayed initiation reduce all cause mortality at 90 days?


  • Randomised, single centre, 2 group, parallel-group trial
  • 1:1 randomization between treatment arms
  • Block randomization in each stratum with block size of 10
  • Group sequential adaptive design with single interim analysis, performed after half the total number of death across both groups
  • 230 patients in total required to generate 80% power to detect treatment effect of 18% reduction in 90-day mortality from a baseline of 55% (based on previous literature)


  • University Hospital Munster, Germany
  • August 2013 to December 2015


  • Inclusion: must have met all five of the following
    1. Acute kidney Injury- KDIGO stage 2:
      • 2-fold increase in serum creatinine from baseline, or urine output <0.5ml/kg/hr for ≥12 hrs despite optimal resuscitation (PCWP>12, CI >2.6, MAP>65, IAP<15)
    2. NGAL >150ng/dl
    3. At least one of the following conditions:
      • Severe sepsis
      • Catecholamines (noradrenaline/adrenaline) >0.1mcg/kg/min
      • Refractory fluid overload (worsening pulmonary oedeoma, P/F <300mgHg, fluid balance >10% body weight)
      • Development or progression of non-renal organ dysfunction (SOFA organ system score ≥2)
    4. Age between 18 and 90
    5. Intention to provide full intensive care treatment for at least 3 days
  • Exclusion:
    • Patients with pre-existing kidney disease not requiring RRT
    • Patients who had received any previous RRT
    • AKI caused by permanent occlusion/surgical lesions of the renal artery
    • Obstructive AKI
    • HUS/TTP
    • Prior Kidney transplant
    • Hepatorenal syndrome
    • AIDS with CD4 count <0/05 X109/L
  • 604 screened, 231 randomised and included in primary analysis


  • Initiation of RRT within 8 hours of confirmation of KDIGO stage 2
  • Delivered to 100% of patients assigned to early group
    • Median time to RRT was 6.0 hours (IQR 4.0–7.0)


  • Inititation of RRT within 12 hours of either
    • KDIGO stage 3 criteria
      • Creatinine rise >3 fold increase from baseline, or oliguria <0.3ml/kg/hr for ≥24 hours or serum creatinine > 4mg/dl (353.6 μmol/l) with an acute increase of at least 0.5mg/dl (44.2μmol/l) within 24 hours
    • Absolute indication for RRT
      • Urea >100mg/dl
      • Potassium >6mmol/l and or ECG abnormalities
      • Magnesium >4mmol/l and/or anuria/absence of deep tendon reflexes
      • Blood pH <7.15
      • Urine production <200ml/12hr or anuria
      • Organ oedema in the presence of AKI resistant to diuretic treatment (defined as one trial of furosemide)
  • Delivered to 91% of patients assigned to delayed group
    • Median time to RRT was 25.5 hours (IQR 18.8–40.3)

Common to both groups

  • RRT standardized
    • CVVHDF
    • prediction 100%
    • dialysate:replacement fluid ratio 1:1
    • effluent flow 30ml/kg
    • minimum 110ml/min blood flow
    • citrate anticoagulation
  • RRT discontinued if renal recovery evident by both:
    • > 400ml urine  in 24hr without diuretics or > 2100ml in 24hr with diuretics
    • creatine clearance > 20ml/min occurred
  • RRT could be changed to SLED after 7 days if no recovery
  • Additional treatments (ventilation, nutrition) standardized
  • Follow up up to 1 year post randomization


Primary outcome:

  • 90 day mortality
    • Early group 39.3% vs Delayed group 54.7%
      • Hazard ratio 0.66 (95% CI 0.45 to 0.97; P=0.03)
      • Absolute risk reduction 15.34% (95% CI 2.62% to 28.06%; P= 0.025)
      • Number needed to treat (NNT) is 7
      • Fragility Index is 3

Secondary outcome: early vs delayed

  • Median duration of RRT: 9 days vs 25 days (P = 0.04, HR 0.69 [95% CI 0.48–1.00])
  • Enhanced recovery of renal function at day 90: 53.6% vs 38.7% (P = 0.02, OR 0.55 [95% CI: 0.32–0.93])
  • Median duration of mechanical ventilation: 125.5 hrs vs 181.0 hrs (P = 0.002)
    • Error noted in manuscript: should be 181.0 hours not days?
  • Length of hospital stay: 51 days vs 82 days (P < 0.001, HR 0.34 [95% CI 0.22–0.52])
  • No significant differences seen in
    • Requirement of RRT at day 90
    • Length of ICU stay
    • Adverse events, fluid balance, and RRT modality
    • Subgroup analysis of those who reached KDIGO stage 3 vs those who achieved absolute RRT requirement

Exploratory analysis:

  • Inflammatory mediators (comparison of MIF, IL-6, IL8, IL-10, IL-18):
    • No difference at randomisation
    • 24 hours post randomisation: significantly lower IL-6 and IL-8 in early group compared to delayed group (100% of early group and 21.8% of late group had received RRT but this time point)


Authors’ Conclusions

  • Amongst critically unwell patients with AKI, early RRT compared with delayed initiation of RRT reduced mortality over the first 90 days


  • Very strict definitions of AKI used and followed
  • No patients lost to follow up, and very little data lost
  • Clear separation achieved between groups
  • Pre-specified secondary outcome analysis
  • Exploratory analysis of inflammatory mediators provides a potential mechanism by which early RRT may improve outcomes


  • Fragility Index 3: a shift of 3 patients would render it non-significant (however, this is greater than the number lost-to-follow-up)
  • Single centre study: limits its external validation
  • Not all the treatments were standardised between groups. With the groups also being unblinded, this introduces bias, challenging its internal validity.
  • External validation further challenged by skewed patient population:
    • 216/231 (93.5%) were surgical of which 108 (46.75%) were cardiac surgery patients
    • 203 (88%) were mechanically ventilated at time of randomization

The Bottom Line

  • This single centre study demonstrates a significant reduction in 90 day mortality with early (stage 2 KDIGO) initiation of RRT, in a group of almost entirely surgical patients. If I worked in this unit or a similar highly surgical ICU, I would consider changing my practice.
  • However, other studies challenge these conclusions. A larger multi-centre trial including a more mixed ICU patient population is needed in order to validate these data on a larger scale.

External Links

[article] Effect of Early vs Delayed Initiation of Renal Replacement Therapy on Mortality in Critically Ill Patients With Acute Kidney Injury
[Editorial] Early to Dialyze: Healthy and Wise?
[further reading] Initiation Strategies for Renal Replacement Therapy in the Intensive Care Unit
[further reading] Nephrology journal club: Acute Kidney Injury, when to dialyse.


Summary author: Segun Olusanya
Summary date: 15 June 2016
Peer-review editor: Duncan Chambler


Leave a Reply

Your email address will not be published. Required fields are marked *

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