Email updates

Keep up to date with the latest news and content from Critical Care and BioMed Central.

Open Access Highly Accessed Open Badges Research

Timing of dialysis initiation in AKI in ICU: international survey

Charuhas V Thakar*, James Rousseau and Anthony C Leonard

Author Affiliations

Department of Internal Medicine, University of Cincinnati, 231 Albert Sabin Way, Cincinnati, OH 45267, USA

For all author emails, please log on.

Critical Care 2012, 16:R237  doi:10.1186/cc11906

See related commentary by Joannidis et al.,

Published: 19 December 2012



Initiating dialysis in acute kidney injury (AKI) in an intensive care unit (ICU) remains a subjective clinical decision. We examined factors and practice patterns that influence early initiation of dialysis in ICU patients with acute kidney injury.


An online survey presented nephrologists (international) with three case scenarios with unstated predicted mortality rates of < 10%, 10 - 30% and > 30%. For each case the respondents were asked 4 questions about influences on the decision whether or not to initiate dialysis within 24 hours: Q1, likelihood of initiating dialysis; Q2, threshold of BUN levels (< 50, 50 - 75, 76 - 100, > 100 mg/dl) considered relevant to this decision; Q3, magnitude of creatinine elevation (two to three-fold increase; greater than threefold increase; absolute level > 5 mg/dl regardless of change) considered relevant; Q4, a rank order of the influence of five parameters (BUN level, change of creatinine from baseline, oxygen saturation, potassium level, and urine output), 1 being the most influential and 5 being the least influential.


One hundred seventy-two nephrologists (73% in practice for > 5 years; 70% from the U.S.A.) responded to the survey. The proportion of subjects likely to initiate early dialysis increased (76% to 94%), as did the predicted mortality (p < 0.001). The proportion of subjects considering early dialysis at a BUN level ≤ 75 increased from 17% to 30 to 40% as the predicted mortality of the cases increased (p < 0.0001). The proportion of subjects choosing absolute creatinine level to be more influential than relative increment, went from 60% to 54% to 43% as predicted mortality increased (p < 0.0001). Rank-order analysis indicated that influence of oxygen saturation and potassium level on dialysis decision showed a significant change with severity of illness, but BUN level and creatinine elevation remained less influential, and did not change with severity.


Severely ill patients were more likely to be subjected to early dialysis initiation, but its utility is not clear. Rank-order analysis indicates dialysis initiation is still influenced by "imminent" indications rather than a "proactive" decision based on the severity of AKI or azotemia.