STARRT - AKI trial. Harm and unnecessary Renal Replacement Therapy (RRT) when RRT is started early in the ICU.

Author: Peter Ryu Tofts MD

Editor: Martin M. Cearras, MD

Find the original article here!

Timing of Initiation of Renal-Replacement Therapy in Acute Kidney Injury | New England Journal of Medicine

Summary:

STARRT AKI is a large multicenter, international RCT, conducted in 168 hospitals in 15 countries, comparing early renal replacement therapy (RRT) to standard strategy in critically ill patients with acute kidney injury (AKI). Patients were randomized by computer in a 1:1 ratio. Randomized by variable block size (2 and 4) and site stratification via a centralized web-based platform. Primary outcome was mortality. There were pre-specified subgroup analysis. Early RRT has no improvement in mortality and causes harm as well as increases cost.  

PICOTT:

Population: Adults with AKI in the ICU. Defined as Cr > 1.13 (F) / 1.47 (M). Severe AKI Cr x2 or > 4.

Intervention: Early RRT < 12 hrs of meeting eligibility criteria

Comparison: Standard strategy discouraging renal replacement therapy unless conventional indications or Acute Kidney Injury (AKI) > 72hrs.

Outcomes: Primary: Mortality at 90 days, exploratory outcomes: Survivors with RRT dependency at 90 days. Composite of death, RRT dependence or major kidney event (GFR < 75%) at 90 days. Adverse events – hypotension and hypophosphatemia. ICU Length of stay.

Type of Question: Therapy

Type of Study: RCT

Interpretation of the Study:

There was no significant difference in mortality at 90 days between the Early RRT group and the standard management group. There were more harms associated with the early RRT group, mainly hypophosphatemia and hypotension. Early RRT lead to higher numbers of RRT dependent at 90 days. There was a significant increase in re-hospitalization in the early RRT group. The standard group remained in the ICU for longer, but not in their overall hospital stay. The composite outcome was unnecessary and potentially misleading. Death should not be in the same category as RRT and major kidney event. The last two are ok to analyze together.

• Mortality at 90 days: Not statistically significant. Early RRT 643/1465 (43.9%) v control 639/1462 (43.7%) Relative Risk, 1.00; 95%CI 0.93 to 1.09, & Risk Difference 0.2% (95% CI, -0.3 to 0.4; p=0.92)

• RRT initiation: Early RRT 1418/1465 (96.8%) vs Standard 903/1462 (61.8%)

  • Relative risk 1.565% or 156.7%. Relative risk Increase is 56.7%.

  • Absolute Risk Increase: 35%

  • Number needed to HARM (NNH): For every 2 patients on the early RRT, there will be 1 extra RRT initiation.

• Survivors RRT Dependence: Early RRT 85/814(10.4%) vs Control 49/815(6%)

  • Relative Risk 1.74 (95% CI, 1.24 to 2.43). RR Increase: 74%

  • Absolute Risk Increase: 4.4%

  • NNH: 22 to have an extra survivor with RRT dependence.

• Rehospitalization at 90 days; Early RRT 191/913(20.9%) vs Control 156/916(17%).

  • Relative Risk 1.23 95% CI, 1.02 to 1.49. RR Increase: 23%

  • Absolute Risk Increase: 3.9%

  • NNH: 25 to have an extra hospitalization.

Overall Risk of bias:  LOW

Allocation was concealed. Blinding of the treating physician was not possible due to the nature of the intervention. The treatment & control groups were prognostically similar in their assignments: demographics and in terms of other significant markers, ie:  CKD, surgery, sepsis and mean SAPS II and SOFA scores. The analysis was done as a modified Intention-to-treat (mITT). Since very few patients were excluded and probably not enough to change results, it has low risk of bias. Similar results when analyzed as-treated, further strengthen the results. There is a high number of events in all analyzed outcomes, making the study much stronger.

Context: 

It was unclear at the time whether early RRT initiation would improve outcomes. More-over there is still controversy regarding early RRT in cirrhotic patients. Expert recommendation and observation data seem to favor it. However, based on this data, it might be harmful and cause unnecessary vasopressor and RRT use.

Teaching points:

Composite outcomes

NNH

Modified Intention to treat vs per protocol/as- treated analysis

Verdict:

Somewhat settled - Might change with more data