Renal failure in the ICU: comparison of the impact of acute renal failure and end-stage renal disease on ICU outcomes

Kidney Int. 2002 Sep;62(3):986-96. doi: 10.1046/j.1523-1755.2002.00509.x.


Background: Acute renal failure (ARF) is associated with a persistent high mortality in critically ill patients in intensive care units (ICUs). Most studies to date have focused on patients with established, intrinsic ARF or relatively severe ARF due to multiple factors. None have examined outcomes of dialysis-dependent chronic renal failure [end-stage renal disease (ESRD)] patients in the ICU. We examined the incidence and outcomes of ARF in the ICU using a standard definition and compared these to outcomes of ICU patients with either ESRD or no renal failure. We sought to determine the impact of renal dysfunction and/or loss of organ function on outcome.

Methods: We prospectively scored 1530 admissions to eight ICUs over a 10-month period for illness severity at ICU admission using the Acute Physiological and Chronic Health Evaluation (APACHE III) evaluation tool. Patients were defined as having ARF based on the definition of Hou et al (Am J Med 74:243-248,1983) designed to detect significant measurable declines in renal function based on serum creatinine. ESRD patients were identified as being chronically dialysis-dependent prior to ICU admission and the remainder had no renal failure. Clinical characteristics at ICU admission and ICU and hospital outcomes were compared between the three groups.

Results: We identified 254 cases of ARF, 57 cases of ESRD and 1219 cases of no renal failure for an incidence of ARF of 17%. Roughly half the ARF patients had ARF at ICU admission and the remainder developed ARF during their ICU stay. Only 11% of ARF patients required dialysis support. ARF patients had significantly higher acute illness severity scores than those with no renal failure, whereas patients with ESRD had intermediate severity scores. ICU mortality was 23% for patients with ARF, 11% for those with ESRD, and 5% for those with no renal failure. There was no difference in outcome between patients who had ARF at ICU admission and those who developed ARF in the ICU. Patients with ARF severe enough to require dialysis had a mortality of 57%. APACHE III predicted outcome very well in patients with no renal failure and patients with ARF at the time of scoring but underpredicted mortality in those who developed ARF after ICU admission and overestimated mortality in patients with ESRD.

Conclusions: ARF is common in ICU patients and has a persistent negative impact on outcomes, although the majority of ARF is not severe enough to require dialysis support. The mortality of patients with ARF from all causes is almost exactly similar to that noted using the same criteria two decades ago. More profound ARF requiring dialysis continues to have an even greater mortality. Nevertheless, acute declines in renal function are associated with a mortality that is not well explained simply by loss of organ function. The majority of ARF patients who did not require dialysis still had a considerably higher mortality than the ESRD patients, all of whom required dialysis; while ARF patients who did require dialysis had a much higher morality than ESRD patients. APACHE III performs well and captures the mortality of patients with ARF at the time of scoring. Development of ARF after scoring has a profound effect on standardized mortality. We were unable to identify a unique mortality associated with ARF, but the presence of measurable renal insufficiency continues to be a sensitive marker for poor outcome.

Publication types

  • Comparative Study

MeSH terms

  • Acute Kidney Injury / mortality*
  • Acute Kidney Injury / therapy
  • Cohort Studies
  • Critical Care / statistics & numerical data*
  • Humans
  • Incidence
  • Kidney Failure, Chronic / mortality*
  • Kidney Failure, Chronic / therapy
  • Middle Aged
  • Predictive Value of Tests
  • Renal Dialysis / mortality
  • Survival Analysis
  • Transplants / statistics & numerical data
  • Treatment Outcome