The 2012 Kidney Disease: Improving Global Outcomes (KDIGO) clinical practice guideline classifies acute kidney injury (AKI) into 3 stages defined by serum creatinine elevation or urine output decline. We evaluated the potential impact of further categorizing AKI stage 1 into two stages based on serum creatinine criteria, with a focus on how the resulting 4-stage classification would affect the association of AKI stages with clinical outcomes. We defined AKI stage 1a as an absolute increase in serum creatinine of 0.3 mg/dl within 48 hours and stage 1b as a 50% relative increase in serum creatinine within 7 days. We screened all admissions to 5 hospitals from 2012 to 2014 using standardized inclusion and exclusion criteria and included 81,651 admissions in this retrospective cohort study. The incidence of in-hospital AKI was 7.5% for stage 1a, 4.9% for stage 1b, 1.5% for stage 2, and 0.9% for stage 3. Length of stay following the first incidence of AKI was 3.9 days for stage 1a, 6.2 days for stage 1b, 8.8 days for stage 2, and 12.0 days for stage 3. Compared to patients with no AKI, the odds of in-hospital mortality were progressively higher for patients with higher AKI stages (odds ratio 4.3 for patients with stage 1a, 10.9 for stage 1b, 40.6 for stage 2, and 60.0 for stage 3 AKI). Patients with AKI stages 1a and 1b experienced clinically meaningful and statistically significant differences in length of stay and mortality. This study suggests that a modified 4-stage version of the KDIGO AKI classification may provide additional prognostic information.

Original languageEnglish (US)
Pages (from-to)905-913
Number of pages9
JournalKidney international
Issue number4
StatePublished - Apr 2019


  • acute kidney injury
  • clinical epidemiology
  • hospital mortality
  • length of stay
  • serum creatinine

ASJC Scopus subject areas

  • Nephrology


Dive into the research topics of 'Disparate outcomes observed within Kidney Disease: Improving Global Outcomes (KDIGO) acute kidney injury stage 1'. Together they form a unique fingerprint.

Cite this