Q J Med 2011; 104:237–243

doi:10.1093/qjmed/hcq185 Advance Access Publication 8 October 2010

Challenges of defining acute kidney injury

M. OSTERMANN1 and R.W.S. CHANG2
From the 1Department of Critical Care, Guy’s & St Thomas’ Foundation Hospital, London SE1 7EH
and 2Department of Nephrology & Transplantation, St George’s Hospital, London SW17 0QT, UK

Address correspondence to M. Ostermann. Department of Critical Care, Guy’s and St Thomas’ Foundation
Hospital, London, UK. email: marlies.ostermann@gstt.nhs.uk

Received 3 August 2010 and in revised form 17 September 2010

Summary
Background: Until recently, there was a lack of a
uniform definition for acute kidney injury (AKI).
The ‘acute renal injury/acute renal failure syn-
drome/severe acute renal failure syndrome’ criteria,
the Risk - Injury - Failure - Loss of kidney function -
End stage renal disease (RIFLE) criteria and the
Acute Kidney Injury Network (AKIN) classification
were the most recent proposals.
Aim: To compare the performance of the different
AKI definitions.
Design and Methods: Application of the three most
recent AKI definitions to 41 972 critically ill ICU pa-
tients and comparison of their performance.
Results: Incidence and outcome of AKI varied de-
pending on the criteria. The RIFLE and AKIN classi-
fication led to similar total incidences of AKI (35.9
vs. 35.4%) but different incidences and outcomes of
the individual AKI stages. Multivariate analysis
showed that the different stages of AKI were
independently associated with mortality. The worst
stage of AKI was associated with an increased odds
ratio for mortality of 1.59–2.27. Non-surgical admis-
sion, maximum number of associated failed organ
systems, emergency surgery and mechanical venti-
lation were consistently associated with the highest
risk of hospital mortality.
The proposed AKI definitions differ in the cut-off
values of serum creatinine, the suggested time
frame, the approach towards patients with missing
baseline values and the method of classifying pa-
tients on renal replacement therapy. All classifica-
tions can miss patients with definite AKI.
Conclusions: The three most recent definitions of
AKI confirmed a correlation between severity of
AKI and outcome but have limitations and the po-
tential to miss patients with definite AKI. These limi-
tations need to be considered when using the
criteria in clinical practice.

Introduction In 2001, Bellomo and colleagues proposed cri-

The lack of a uniform definition for acute kidney
injury (AKI) is considered to be one of several rea-
sons for conflicting epidemiological data and ab-
sence of therapeutic progress in the field of AKI. In
the past 10 years, the renal and critical care com-
munity have experienced a journey from proposed
AKI criteria in 2001 to the Risk - Injury - Failure -
Loss of kidney function - End stage renal disease
(RIFLE) classification in 2004 and the Acute Kidney
Injury network (AKIN) classification in 2007.1–3
teria which distinguished between three different
stages of AKI [acute renal injury (ARI), acute renal
failure syndrome (ARFS) and severe acute renal fail-
ure syndrome (SARFS)] based on absolute values of
serum creatinine, serum urea, urine output and/or
treatment with renal replacement therapy (RRT)
(Table 1).1 The criteria defined AKI as well as
acute-on-chronic renal disease.
In 2004, the Acute Dialysis Quality Initiative
(ADQI) working party published the RIFLE criteria,

! The Author 2010. Published by Oxford University Press on behalf of the Association of Physicians.
All rights reserved. For Permissions, please email: journals.permissions@oxfordjournals.org
238 M. Ostermann and R.W.S. Chang

Table 1 Criteria for ARI, ARFS and severe ARFS

ARI ARFS SARFS

[Creatinine] >120 mmol/l and [Urea] [Creatinine] >240 mmol/l and [Urea] Need for RRT in the presence
>8 mmol/l and/or urine output >16 mmol/l and/or urine output of either ARI or ARFS
<800 ml/24 h or <200 ml/6 h <400 ml/24 h or <100 ml/6 h

Acute on chronic ARI Acute on chronic ARFS Acute on chronic severe ARFS

[Creatinine] rise by 60 mmol/l or [urea] [Creatinine] rise by 120 mmol/l or [urea] Need for RRT and acute on
rise by 4 mmol/l and/or urine output rise by 8 mmol/l and/or urine output chronic criteria for ARI or
<800 ml/24 h or <200 ml/6 h <400 ml/24 h or <100 ml/6 h ARFS

Table 2 RIFLE classification

RIFLE category GFR criteria Urine output criteria

Risk Increased serum creatinine  1.5 or decrease of Urine output <0.5 ml/kg/h for 6 h
GFR > 25%
Injury Increased serum creatinine  2 or decrease of Urine output <0.5 ml/kg/h for 12 h
GFR > 50%
Failure Increased serum creatinine  3 or decrease of Urine output <0.3 ml/kg/h for 12 h
GFR > 75% or serum creatinine  4 mg/dl or anuria for 12 h
Loss Complete loss of renal function for >4 weeks
End-stage kidney disease Need for RRT for >3 months

which differentiate between three severity classes
and two outcome stages of AKI (Table 2).2 The clas-
sification is based on relative changes of serum cre-
atinine or estimated glomerular filtration rate (GFR)
and absolute urine volumes. If previous creatinine
results are not available, an assumption can be
made that baseline renal function was normal.
Clinical studies with total enrolment of >70 000
patients confirmed a close correlation between the
RIFLE criteria and mortality with a stepwise increase
in relative risk of death from 2.4 in patients with
RIFLE-Risk to 6.37 for RIFLE-Failure.4
In 2005, Chertow and colleagues demonstrated
that even smaller changes in serum creatinine than
suggested in the RIFLE classification, were indicative
of significant renal dysfunction.5 They found that a
rise in serum creatinine by 0.3 mg/dl (26.4 mmol/l)
during hospital admission was independently asso-
ciated with a 4.1 increased odds ratio (OR) for
in-hospital death. The AKI Network revised the
RIFLE criteria and incorporated Chertow’s important
finding into the AKIN classification (Table 3).3
The creation of these definitions was hailed as an
important step forward to facilitate high quality
epidemiological, preventative and therapeutic trials
in AKI. We previously applied the different
classifications to a large population of >40 000 in-
tensive care unit (ICU) patients and confirmed that
all three correlated with outcome.6–8 The aim of this
article is to compare the performance of the different
classifications directly and to illustrate some
strengths and limitations which should be kept in
mind when using the criteria.
Materials and methods
Study population
We analysed the data of 41 972 adult patients
admitted to 22 ICUs in the UK and Germany be-
tween June 1989 and October 1999. Eight hundred
patients were excluded from the analysis (797 pa-
tients with dialysis dependent end-stage renal failure
and three patients with missing data).
Data analysis
The exact methods of data extraction were
described in detail in the original papers.6–8 AKI
was defined according to serum creatinine values
obtained during stay in ICU. We did not have
access to any blood results before admission to
ICU and also did not have 6-hourly urine results.
 Challenges of defining acute kidney injury 239

Table 3 AKIN classification

Diagnostic criteria for AKI

An abrupt (within 48 h) reduction in kidney function defined as an absolute increase in serum creatinine of either
0.3 mg/dl (26.4 mmol/l) or a percentage increase of 50% (1.5-fold from baseline) or a reduction in urine output (after
exclusion of hypovolaemia and obstruction)

AKI stage Creatinine criteria Urine output criteria

Staging system
AKI stage I Increase of serum creatinine by 0.3 mg/dl (26.4 mmol/l) or Urine output <0.5 ml/kg/h for
increase to 150–200% from baseline >6 h
AKI stage II Increase of serum creatinine to >200–300% from baseline Urine output <0.5 ml/kg/h for
>12 h
AKI stage III Increase of serum creatinine to >300% from baseline or Urine output <0.3 ml/kg/h for
serum creatinine 4.0 mg/dl (354 mmol/l) after a rise of >24 h or anuria for 12 h
44 mmol/l or treatment with RRT

The final AKI category was based on the maximum
AKI stage during ICU admission.
Ethics approval
The local research and ethics committee confirmed
that informed consent was not required because the
study included neither an intervention nor breach of
privacy or anonymity.
Statistical analysis
The statistical analyses are outlined in detail in the
original papers.6–8 Using multivariate logistic regres-
sion analysis, independent predictors of all-cause
hospital mortality were identified. Calibration was
assessed by the Hosmer–Lemeshow ‘Goodness-of-
fit’ statistic for significance (P > 0.05) and discrimin-
ation was evaluated by determination of the area
under the receiver operating characteristics (ROC)
curve. The statistical package SPSS (Version 14.0,
Woking, UK) was used for all analyses.
Results
The incidence and outcome of AKI depended on the
criteria used (Table 4). The RIFLE criteria and the
AKIN classification led to similar total incidences
of AKI but different incidences and outcomes of
the individual AKI stages. Similarly, mortality rates
of the individual stages differed depending on the
criteria used.
Only the AKIN classification demonstrated a step-
wise increase in risk of mortality. Multivariate ana-
lyses showed that different stages of AKI were
independently associated with outcome. The worst
stage of AKI was associated with an OR for hospital
mortality of 1.59–2.27. Non-surgical admission,
emergency surgery, maximum number of failed
organ systems and mechanical ventilation were con-
sistently associated with the highest risk of hospital
mortality. The area under the ROC was consistently
>80%.
All classifications had the risk of not identifying
patients with definite progressive creatinine rises as
AKI. Reasons were the use of absolute creatinine
values (i.e. the ARI/ARFS/SARFS criteria) or a
narrow time period (i.e. the AKIN classification).
Discussion
Designing criteria for a dynamic condition like AKI
is a major undertaking. The validity of any classifi-
cation for AKI depends on whether it is able to clear-
ly differentiate between normal renal function and
different stages of AKI, has objective cut-off criteria
which are easy to ascertain and not ambiguous, and
has prognostic properties. Application of three dif-
ferent classifications for AKI to the same database
showed that incidences of AKI and associated out-
comes varied depending on the definition used.
However, all classifications confirmed a correlation
between severity of AKI and prognosis. Even when
controlling for confounding factors in a multiple
variable regression analysis, severe AKI was an in-
dependent risk factor for mortality.
This analysis also shows that all three AKI classi-
fications have some shortcomings. Using the ARI/
ARFS/SARFS criteria with absolute cut-off values
for serum creatinine, patients with definite AKI (for
instance, patients with a rise in serum creatinine
Table 4 Comparison of three different classifications for AKI
240

Factor ARI/ARFS/SARFS criteria RIFLE classification AKIN classification

Incidence ARI 17.9% Risk 17.2% AKI I 19.1%

ARFS 6.3% Injury 10.99% AKI II 3.8%
SARFS 4.2% Failure 7.6% AKI III 12.5%
Total AKI 28.5% Total AKI 35.9% Total AKI 35.4%
ICU mortality ARI 22.8% Risk 14.7% AKI I 20.1%
ARFS 44.3% Injury 36.5% AKI II 25.9%
SARFS 54.6% Failure 47.6% AKI III 49.6%
Hospital mortality ARI 29.5% Risk 20.9% AKI I 29.9%
ARFS 49.2% Injury 45.6% AKI II 35.8%
SARFS 63.0% Failure 56.8% AKI III 57.9%
Independent risk of hospital mortality ARI 1.78 (1.61–1.98) Risk 1.40 (1.28–1.53) AKI I 0.98 (0.90–1.08)
related to AKI OR (95% CI)
ARFS 1.65 (1.47–1.84) Injury 1.96 (1.80–2.14) AKI II 1.11 (0.94–1.31)
SARFS 1.31 (1.15–1.49) Failure 1.59 (1.43–1.76) AKI III 2.01 (1.72–2.36)
Independent risk factors for hospital
mortality, OR (95% CI)
Cardiac surgery – 0.5 (0.44–0.57) 0.42 (0.35–0.49)
Male gender 0.90 (0.85–0.97) 0.93 (0.87–0.995) 0.89 (0.83–0.96)
Age 1.02 (1.02–1.03) 1.02 (1.02–1.03) 1.03 (1.03–1.04)
APACHE II score on admission to ICU 1.14 (1.13–1.14) 1.097 (1.09–1.104) –
SOFA score on admission to ICU – – 1.10 (1.09–1.12)
Pre-existing chronic diseases – 1.17 (1.08–1.26) 1.75 (1.61–1.91)
Maximum number of failed organs 1.64 (1.57–1.73) 2.13 (2.03– 2.23) 2.29 (2.19–2.39)
M. Ostermann and R.W.S. Chang

Ventilation 1.83 (1.69–1.98) 1.52 (1.41–1.65) 1.62 (1.47–1.79)

Emergency surgery 3.95 (3.56–4.39) 3.08 (2.77–3.42) 2.26 (1.99–2.56)
Non-surgical admission 5.97 (5.49–6.49) 3.92 (3.58–4.30) 3.06 (2.75–3.40)
Area under the ROC curve 0.90 0.897 0.84
Hosmer–Lemeshow 2 47.3; P < 0.001 48.32; P < 0.001 40.987; P < 0.0001
Specific limitations of criteria 1. Risk of missing patients who start RRT 1. Use of MDRD formula to estimate GFR 1. Inclusion of ‘need for RRT’ as a
before the criteria for ARI or ARFS are 2. Assumption that patients whose baseline criterion without clear indication
fulfilled renal function is not known, have when to initiate it
2. No clear time window normal renal function 2. Tight 48-h window; patients with
3. No criteria for patients on RRT; all RIFLE more gradual decline in renal
categories may contain patients on RRT function may be missed
4. Patients who start RRT before they fulfil 3. The criterion ‘Need to exclude
the RIFLE criteria for ‘Risk’ will not be hypovolaemia and obstruction’ is
identified as having AKI and be missed not specific

SOFA, sequential organ failure assessment; APACHE, Acute Physiology and Chronic Health Evaluation.
 Challenges of defining acute kidney injury
from a baseline value of 50 mmol/l to 115 mmol/l
indicating a fall in GFR by >50%) will not be iden-
tified unless they also have a relevant reduction in
urine output. Similarly, patients receiving RRT are
only classified as having AKI if they also fulfil the
criteria for ARI or ARFS, i.e. have a serum creatinine
>120 mmol/l and/or oliguria. We identified 100 pa-
tients in our database who were treated with RRT
early (i.e. serum creatinine was still <120 mmol/l)
but were subsequently classified as having ‘no
AKI’ because the serum creatinine remained consist-
ently <120 mmol/l as a result of RRT.
A major criticism of the RIFLE classification is the
recommendation to assume normal baseline renal
function if previous creatinine results are not avail-
able. In this case, the premorbid serum creatinine
can be calculated according to an assumed esti-
mated GFR of 75 ml/min/1.72 m2 derived from the
‘modification of diet in renal disease (MDRD)’ for-
mula.9,10 First, this approach ignores the fact that a
proportion of AKI patients may have pre-existing
chronic kidney disease (CKD). Therefore, the inci-
dence of AKI may be overestimated. Secondly, the
MDRD formula was originally derived from patients
with stable CKD with an average GFR of 40 ml/min/
1.73 m2 but has not been validated in patients with
AKI and rapidly changing creatinine levels.
The classifications vary in the method of classify-
ing patients on RRT. Clinical practice of RRT is
known to be very variable with no consensus re-
garding optimal timing.11–13 The RIFLE classification
does not include RRT as a criterion, an approach
which eliminates the subjective element. As a
result, all RIFLE categories include patients on
RRT: patients may be classified as RIFLE-Risk or
even ‘no AKI’ if RRT is started early (so that serum
creatinine levels are maintained near-normal) or as
RIFLE-Failure if RRT is started later after serum cre-
atinine has risen by >300%. The AKI Network
decided to classify patients on RRT automatically
as having the worst degree of AKI, independent of
serum creatinine.
Apart from the inclusion of RRT as a specific cri-
terion, the AKIN classification differs from the RIFLE
criteria in five other aspects: (i) the inclusion of a
48-h window for the diagnosis of AKI, (ii) a smaller
change in creatinine to qualify for the diagnosis of
AKI, (iii) the elimination of the MDRD formula, (iv)
differentiation between three stages of severity but
no outcome categories and (v) no allowance for
missing baseline creatinine levels.
A time frame is clearly necessary when deciding
whether kidney injury is acute or chronic. CKD is
traditionally defined as the presence of renal dys-
function for 3 months. It would follow that renal
dysfunction during a <3 months period should be
241
called ‘acute renal disease’. The AKI Network chose
a 48-h window to ensure that AKI was definitely
‘acute’. In contrast, the RIFLE classification uses a
7-day window. The danger of these relatively
narrow time windows is that patients with slowly
progressive acute renal dysfunction are not identi-
fied correctly. In our analysis we noticed that 2014
patients who were classified as having ‘no AKI’ (as
per AKIN classification), had serum creatinine levels
>140 mmol/l of whom 316 patients even had values
>270 mmol/l. Although some patients had a degree
of pre-existing CKD, others had progressive rises in
serum creatinine but not within the required time
period. Using the AKIN classification with a 7-day
period (similar to the RIFLE criteria) instead of a 48-h
window, we observed an increase in the total inci-
dence of AKI: 39.5% instead of 35.4%. It is likely
that a longer time frame, i.e. 2–4 weeks would have
identified even more patients.
The AKIN classification emphasizes that AKI
should only be diagnosed after exclusion of hypo-
volaemia and obstruction. Although it can be diffi-
cult in retrospective analyses to be certain that these
criteria were fulfilled, accurate fluid assessment re-
mains a challenge in clinical practice, too. A small
retrospective single-centre study attempted to evalu-
ate the AKIN classification strictly and concluded
that the association of AKI with hospital mortality
was still significant even when fluid depletion was
not strictly excluded.14 During a 1-year period,
213 patients admitted to a medical ICU fulfilled
the serum  urine creatinine criteria for AKI. Only
123 patients had complete data on fluid challenges,
including the administration of a minimum of
500 ml of volume expander during <1 h in response
to oliguria or raised serum creatinine. The authors
illustrated that the association of AKI with hospital
mortality was still significant even when the appro-
priate fluid challenge requirement was discarded.
A major problem of all proposed AKI classifica-
tions is the fact that they aim to be diagnostic as well
as prognostic, include subjective criteria, like ‘treat-
ment with RRT’ and rely on serial measurements of
serum creatinine as the determining parameter.
Serum creatinine not only depends on renal
function but is also affected by non-renal factors
like age, muscle bulk and volume of distribution.
Furthermore, in the early phase of AKI, significant
decreases in GFR only lead to small increases in
serum creatinine due to the exponential relationship
of serum creatinine and GFR.15
Waikar et al. emphasized that the generation rate
and dynamic changes of creatinine varied depend-
ing on underlying renal function.16 Twenty-four
hours after a 90% reduction in creatinine clearance,
the rise in serum creatinine was 246% in patients
242 M. Ostermann and R.W.S. Chang
with normal renal function, 174% in CKD stage II,
92% in CKD stage III and only 47% in patients with
CKD stage IV. In patients with normal renal func-
tion, CKD stages I or II, a 30% reduction in creatin-
ine clearance never led to a 50% increase in serum
creatinine (the earliest stage of the RIFLE
classification).
Macedo et al. highlighted the impact of fluid over-
load on serum creatinine.17 In a group of 253 crit-
ically ill patients with AKI, the median cumulative
fluid balance increased from 2.7 l on Day 2 to 6.5 l
on Day 7. As a result of dilution, the measured cre-
atinine values were significantly lower compared
with creatinine values adjusted for fluid accumula-
tion (8–57 mmol/l over time). The authors warned
that this underestimation of serum creatinine
values in fluid overloaded patients may delay the
diagnosis of AKI. They suggested that serum creatin-
ine values should be corrected for fluid accumula-
tion to allow a more accurate determination of AKI
in critically ill patients. Finally, serum creatinine
levels are also affected by laboratory technologies
used for measurement as well as interference by
hyperbilirubinaemia and drugs (i.e. Cimetidine,
Trimethoprim).18,19
Despite the problems associated with the use of
creatinine, it is currently the most widely used par-
ameter to describe renal function, especially since
none of the other newer biomarkers of renal func-
tion have been fully established in clinical practice.
The usefulness of urine criteria for the definition of
AKI has been debated previously.9,20,21 Proponents
argue that urine output often portends renal dysfunc-
tion in critical care patients before changes in serum
creatinine.21 In contrast, critics point out that urine
output is affected by volume status, intrinsic levels of
antidiuretic hormone, presence of obstruction and
use of diuretics. Also, urine output criteria can
only be accurately assessed in patients with a urin-
ary catheter. In a review of 10 studies using the
RIFLE classification, patients with RIFLE-Risk based
on the creatinine criteria had a worse prognosis than
those in the same class defined by the urine output
criteria alone.22 This observation raises the question
whether the urine criteria indeed add value and sec-
ondly, whether they are appropriately matched with
serum creatinine criteria.
The AKI classifications in their current format do
not specify the direction of change of creatinine
values and do not differentiate between patients
with a rising creatinine and patients with an initially
high serum creatinine, which gradually falls. We
found that within each AKI stage, ICU mortality
was significantly higher in AKI patients with a
rising creatinine compared with patients with AKI
and falling creatinine levels.8
The RIFLE and AKIN classification have been dir-
ectly compared before. A previous study in 662 ICU
patients showed that the AKIN classification identi-
fied more patients as having AKI compared with the
RIFLE criteria.23 Mortality was significantly higher
for AKI defined by the RIFLE criteria. Bagshaw and
colleagues analysed the Australian New Zealand
Intensive Care Society Adult Patient Database of
120 123 adult patients admitted to 57 ICUs between
2000 and 2005.24 They showed that the number of
patients classified as having a degree of AKI in the
first 24 h in ICU using either the AKIN or RIFLE clas-
sification, was similar with no major difference in
sensitivity, robustness and predictive ability. Our
analysis also found no difference in the total inci-
dence of AKI between the AKIN classification and
the RIFLE criteria using data from the total stay in
ICU, but the incidences and outcomes of individual
stages of AKI differed (Table 4).
Most clinicians and researchers welcome the
existing attempts and efforts to agree on consensus
criteria for the diagnosis of AKI. However, it is im-
portant to emphasize that the identification of a par-
ticular AKI stage does not provide any information
on aetiology and clinical management. AKI stage 1
as a result of cardiogenic shock is very different from
AKI stage 1 due to crescentic glomerulonephritis
with regards to necessary therapeutic intervention
as well as prognosis. The efforts to agree on a def-
inition for AKI should occur in parallel with ongoing
education in the area of Critical Care Nephrology,
including emphasis on diagnosing the aetiology
of AKI.
There is also some concern that the existing cri-
teria for AKI do not capture the full extent of acute
kidney disease. CKD is traditionally defined as the
presence of impaired renal function for 3 months.
The RIFLE classification uses a 1-week window to
define AKI which leaves patients with progressive
but slow rises in creatinine over a longer period un-
diagnosed. As mentioned earlier, we found that all
three AKI classifications failed to identify some pa-
tients with definite rises in serum creatinine consist-
ent with AKI. Clearly more work is needed in the
area of ‘defining AKI’ during the 3-month period
outside the 48 h or 1-week period for AKI but
before the 3 months cut-off when they can be
labelled as having CKD.
In conclusion, the three most recent classifica-
tions for AKI confirm a correlation between severity
of AKI and outcome. All three classifications have
strengths and weaknesses and potential for not iden-
tifying patients with definite acute renal dysfunction.
Future revisions of the criteria should include guid-
ance on the direction of change, corrections for
underlying CKD and fluid overload and guidance
 Challenges of defining acute kidney injury
on how to define patients with slowly progressive
renal dysfunction. Until new biomarkers for AKI
are fully established in clinical practice, definitions
of AKI are likely to rely on changes in serum creatin-
ine  urine output. The generation of AKI criteria
and their validation should not distract from the
need to explore the pathomechanism of different
types of AKI and to search for therapies.
Funding
Departmental funds.
Conflict of interest: None declared.
References
1. Bellomo R, Kellum J, Ronco C. Acute renal failure: time for
consensus. Intensive Care Med 2001; 27:1685–8.
2. Bellomo R, Ronco C, Kellum JA, Mehta RL, Palevsky P, and
the ADQI workgroup. Acute renal failure – definition, out-
come measures, animal models, fluid therapy and informa-
tion technology needs: the Second International Consensus
Conference of the Acute Dialysis Quality Initiative (ADQI)
Group. Crit Care 2004; 8:R204–12.
3. Mehta RL, Kellum JA, Shah SV, Molitoris BA, Ronco C,
Warnock DG, et al. Acute Kidney Injury Network (AKIN):
report of an initiative to improve outcomes in acute kidney
injury. Crit Care 2007; 11:R31.
4. Ricci Z, Cruz D, Ronco C. The RIFLE criteria and mortality in
acute kidney injury: a systematic review. Kidney Int 2008;
73:538–46.
5. Chertow GM, Burdick E, Honour M, Bonventre JV,
Bates DW. Acute kidney injury, mortality, length of stay,
and costs in hospitalized patients. J Am Soc Nephrol 2005;
16:3365–70.
6. Ostermann ME, Chang RW. Prognosis of acute renal failure:
An evaluation of proposed consensus criteria. Intensive Care
Med 2005; 31:250–6.
7. Ostermann M, Chang RW. Acute kidney injury in the inten-
sive care unit according to RIFLE. Crit Care Med 2007;
35:1837–43.
8. Ostermann M, Chang R, and the Riyadh ICU Program Users
Group. Correlation between the AKI classification and out-
come. Crit Care 2008; 12:R144.
9. Joannidis M. Classification of acute kidney injury: are we
there yet? Intensive Care Med 2007; 33:572–4.
10. Delanaye P, Krzesinski JM, Cavalier E, Lambermont B. The
RIFLE criteria: are the foundations robust? Crit Care Med
2007; 35:2669.
243
11. Ricci Z, Ronco C, D’Amico G, De Felice R, Rossi S, Bolgan I,
et al. Practice patterns in the management of acute renal
failure in the critically ill patient: an international survey.
Nephrol Dial Transplant 2006; 21:690–6.
12. Overberger P, Pesacreta M, Palevsky PM for the VA/NIH
Acute Renal Failure Trial Network. Management of renal re-
placement therapy in acute kidney injury: a survey of prac-
tioner prescribing practices. Clin J Am Soc Nephrol 2007;
2:623–30.
13. Pannu N, Klarenbach S, Wiebe N, Manns B, Tonelli M, and
Alberta Kidney Disease Network. Renal replacement therapy
in patients with acute renal failure: a systematic review.
JAMA 2008; 299:793–805.
14. Barrantes F, Tian T, Vazquez R, Amoateng-Adjepong Y,
Manthous CA. Acute kidney injury criteria predicts outcomes
of critically ill patients. Crit Care Med 2008; 36:1397–403.
15. Lameire N, Hoste E. Reflections on the definition, classifica-
tion, and diagnostic evaluation of acute renal failure. Curr
Opin Crit Care 2004; 10:468–75.
16. Waikar SS, Bonventre JV. Creatinine kinetics and the defin-
ition of acute kidney injury. J Am Soc Nephrol 2009;
20:672–9.
17. Macedo E, Bouchard J, Soroko SH, Chertow GM,
Himmelfarb J, Ikizler TA, et al. Fluid accumulation, recogni-
tion and staging of acute kidney injury in critically ill pa-
tients. Crit Care 2010; 14:R82.
18. Joffe M, Hsu CY, Feldman HI, Weir M, Landis JR, Hamm LL.
Chronic Renal Insufficiency Cohort (CRIC) Study Group.
Variability of creatinine measurements in clinical labora-
tories: results from the CRIC study. Am J Nephrol 2010;
31:426–34.
19. Slack AJ, Wendon J. The liver and kidney in critically ill pa-
tients. Blood Purif 2009; 28:124–34.
20. Van Biesen W, Vanholder R, Lameire N. Defining Acute
Renal Failure: RIFLE and Beyond. Clin J Am Soc Nephrol
2006; 1:1314–19.
21. Molitoris BA, Levin A, Warnock DG, Joannidis M, Mehta RL,
Kellum JA, et al. Improving outcomes from Acute Kidney
Injury. J Am Soc Nephrol 2007; 18:1992–3.
22. Hoste EAJ, Kellum JA. Acute kidney injury: epidemiology and
diagnostic criteria. Curr Opin Crit Care 2006; 12:531–7.
23. Lopes JA, Fernandes P, Jorge S, Gonçalves S, Alvarez A,
Costa e Silva Z, et al. Acute kidney injury in intensive care
unit patients: a comparison between the RIFLE and the Acute
Kidney Injury Network classification. Crit Care 2008;
12:R110.
24. Bagshaw SM, George C, Bellomo R, and ANZICS Database
Management Committee. A comparison of the RIFLE and
AKIN criteria for acute kidney injury in critically ill patients.
Nephrol Dial Transplant 2008; 23:1569–74.