Image may be NSFW.
Clik here to view.
Ms. G is a 42-year-old woman admitted to your ICU with septic shock in the setting of a urinary tract infection. She was started on a norepinephrine infusion in the emergency room. Her labs reveal a creatinine of 3.0 mg/dL, increased from a prior baseline of 0.6 mg/dL; pH is 7.3 and potassium is 4.9 mEq/L. Her urine output is dropping, despite the fact that she received crystalloids for fluid resuscitation. In other words, she has clear signs of acute kidney injury (AKI). You remember the “AEIOUs,” the mnemonic for the indications for urgent dialysis – acidosis, electrolyte derangement, intoxications, overload, uremia. Ms. G doesn’t meet any of these criteria yet, but you are worried about her. Will delaying dialysis cause her harm?
A study published online in May and appearing in this week’s NEJM aims to answer this clinical question. Prior studies have suggested a possible mortality benefit from early renal-replacement therapy (RRT), which can facilitate tighter control of volume status and potential prevention of uremic complications such as coagulopathy and encephalopathy. This present trial, which took place in 31 ICUs in France, enrolled adult patients with severe acute kidney injury (AKI) from acute tubular necrosis and who required mechanical ventilation, vasopressors or both and randomized them either an early or delayed strategy for renal-replacement therapy. Dialysis was initiated as soon as possible after randomization in the early strategy, while the patients in the delayed-strategy group were dialyzed only if they met one of several specific indications, such as hyperkalemia, acidemia, uremia, severe oliguria or anuria. Patients in both groups could be treated with either intermittent hemodialysis (HD) or continuous veno-venous hemofiltration (CVVH).
Of 620 patients enrolled, 312 were randomized to early strategy and 308 to delayed strategy. 98% of the patients in the early-strategy group underwent dialysis, as compared with only 51% patients in the delayed-strategy group. The primary outcome was mortality at 60 days, which did not differ between the two groups (48.5% in early vs. 49.7% in delayed group). There was also no difference between the two groups in a number of pre-specified secondary outcomes, including number of ventilator-free days, vasopressor-free days, length of stay in the ICU, and need for RRT at day 28 and 60. The delayed-strategy group had fewer days with dialysis catheters, and notably also fewer bloodstream infections.
Almost 50% of patients in the delayed-strategy group never received dialysis. The patients in the delayed-strategy group also were observed to have more rapid recovery of renal function.
The authors acknowledge several limitations to their study. Notably, 50% of the patients received HD initially, and only 30% received solely CVVH during their treatment. While prior meta-analyses have not supported a benefit of one method of RRT over another in the critical care setting, there are possible downsides to HD that could influence outcomes. The authors are careful not to overreach the claims based on the trial results; they recommend careful surveillance when deciding to delay RRT and rapid initiation of dialysis if and when acute indications arise.
In an accompanying editorial, Dr. Ravinda Mehta, Professor of Medicine in the Division of Nephrology at the University of California, San Diego, notes that the trial has good methodology and clear outcomes, but highlights the need for “dynamic risk-stratification tools to identify patients who will not need renal replacement therapy.”
So, based on the results of this trial, you opt for a strategy of careful active surveillance for Ms. G. You continue to monitor her urine output and electrolytes closely, not initiating dialysis for now. You are relieved when she never meets any indications for dialysis. Her renal function improves, and she makes a full recovery.
The authors of this paper are available to answer your questions and discuss their research from July 6-16 on the NEJM Group Open Forum. Join the discussion now!
